Biophysical studies of membrane proteins/peptides. Interaction with ...

UNIVERSIDADE TÉCNICA DE LISBOA
INSTITUTO SUPERIOR TÉCNICO
BIOPHYSICAL STUDIES OF MEMBRANE
PROTEINS/PEPTIDES. INTERACTION WITH
LIPIDS AND DRUGS.
Fábio Monteiro Fernandes
(Licenciado)
Dissertação para obtenção do Grau de Doutor em Química
Orientador:
Co-orientador:
Doutor Manuel José Estevez Prieto
Doutor Luís Miguel Santos Loura
Presidente:
Vogais:
Júri
Reitor da Universidade Técnica de Lisboa
Doutor Horst Vogel
Doutor José Manuel Gaspar Martinho
Doutor Miguel Augusto Rico Botas Castanho
Doutor Manuel José Estevez Prieto
Doutor Mário Nuno de Matos Sequeira Berberan e Santos
Doutora Maria Raquel Murias dos Santos Aires de Barros
Doutor Luís Miguel Santos Loura
3 de Setembro de 2007

PREFÁCIO
PREFÁCIO
Biomembranas são estruturas de complexa composição química. As diversas classes
de moléculas presentes, em especial lípidos e proteínas, apresentam propriedades e
afinidades muito diferenciadas. Esta diferenciação resulta na ocorrência de
heterogeneidades na distribuição lateral dos componentes na membrana. De facto, crêse
que a célula controla certas actividades celulares operadas em biomembranas, através
da mediação da distribuição destes componentes. Um maior conhecimento acerca das
interacções estabelecidas entre os componentes de membranas biológicas é portanto de
importância crucial para a compreensão das funções celulares baseadas nestas
estruturas. Por outro lado as drogas são frequentemente dirigidas à interacção com
proteínas inseridas em biomembranas. Assim, a caracterização das interacções destas
moléculas com a bicamada lipídica e com os seus alvos proteicos é de grande relevância
no processo de desenvolvimento de fármacos.
Os estudos que compõem esta tese resultaram da aplicação de técnicas físicas,
nomeadamente metodologias de fluorescência, à caracterização de interacções entre
diversos componentes de membranas biológicas e de interacções proteína-fármacos.
Sistemas modelo mimetizadores de membranas biológicas (lipossomas) foram
utilizados de forma a trabalhar sobre uma base de estudo simplificada, mantendo ainda
as propriedades fundamentais das biomembranas.
Esta tese foi objecto dos seguintes artigos já publicados ou submetidos para
publicação:
Fernandes, F., Loura, L. M. S., Prieto, M., Koehorst, R., Spruijt, R., and Hemminga, M. (2003).
Dependence of M13 major coat protein oligomerization and lateral segregation on bilayer
composition. Biophys. J. 85: 2430-2431.
Fernandes, F., Loura, L. M. S., Koehorst, R., Spruijt, R., Hemminga, M., Fedorov, A., and
Prieto, M. (2004). Quantification of protein-lipid selectivity using FRET. Application to the
M13 major coat protein. Biophys. J. 87: 344-352.
Fernandes, F., Loura, L. M. S., Koehorst, R., Dixon, N., Kee, T. P., Hemminga, M., and Prieto,
M. (2006). Interaction of the indole class of V-ATPase inhibitors with lipid bilayers.
Biochemistry 45: 5271-5279.
Fernandes, F., Loura, L. M. S., Koehorst, R., Dixon, N., Kee, T. P., Prieto, M., and Hemminga,
M. (2006). Binding assays of inhibitors towards selected V-ATPase domains. Biochim Biophys
Acta. 1758: 1777-1786.
i

Fernandes, F., Loura, L. M. S., Fedorov, A., and Prieto, M. (2006). Absence of clustering of
phosphatidylinositol-(4,5)-bisphosphate in fluid phosphatidylcholine. J. Lipid Res. 47: 1521-
1525.
Fernandes, F., Neves, P., Gameiro, P., Loura, L. M. S., and Prieto, M. Ciprofloxacin
Interactions With Bacterial Protein OmpF: Modelling of FRET from a Multi-Tryptophan
Protein Trimer (aceite para publicação).
Fernandes, F., Loura, L. M. S., Matos, A. P., Fedorov, A., and Prieto, M. Role of Helix-0 of the
N-BAR domain in membrane curvature generation. (submetido).
Uma vez que a maior parte do trabalho já foi aceite para publicação em revistas
internacionais, optou-se pela sua apresentação na forma de artigo científico e pela
redacção desta tese em língua inglesa.
Este manuscrito está dividido em nove partes. No primeiro Capítulo é feita uma
introdução às biomembranas e às interacções proteína-lípido. Na segunda parte, os
trabalhos respeitantes ao estudo da interacção da proteína principal da cápside do
baacteriófago M13 são apresentados. O Capítulo III é dedicado aos estudos de
interacção do potencial fármaco SB242784 com sistemas modelo de membranas e da
ligação de inibidores da V-ATPase a um péptido contendo um putativo local de ligação
para estes inibidores. No Capítulo IV é apresentado o trabalho respeitante aos estudos
de ligação do antibiótico ciprofloxacina à proteína membranar OmpF. Seguidamente, no
Capítulo V, os estudos da interacção do péptido N-terminal do domínio N-BAR com
sistemas modelo de membranas são introduzidos e o Capítulo VI é dedicado aos estudos
de caracterização da distribuição lateral de fosfoinositol-(4,5)-bisfosfato em liposomas
de fosfocolinas. Os Capítulos II a VI, são sempre iniciados com uma breve introdução
aos sistemas estudados.
No Capítulo VII são apresentadas as conclusões globais referentes aos estudos de
cada um dos sistemas investigados. Finalmente, o Capítulo VIII é dedicado à
apresentação de considerações finais e perspectivas futuras e o Capítulo IX compreende
a bibliografia.
Gostaria ainda de finalizar agradecendo sinceramente a todos que de uma maneira
ou outra contribuiram para a realização dos trabalhos aqui descritos:
Agradeço ao Centro de Química-Física Molecular, na pessoa do Professor José
Manuel Gaspar Martinho, pelas facilidades concedidas para a realização deste trabalho.

PREFÁCIO
Aos Professores Manuel Prieto e Luís Loura, meus orientadores dos trabalhos de
doutoramento, pela dedicação, apoio incansável, pela oportunidade de partilhar da sua
paixão pela ciência, pelo muito que me ensinaram, mas principalmente pela amizade.
À Professora Ana Coutinho pelas muito úteis discussões científicas, pela
disponibilidade constante e pela amizade.
Aos meus colegas do grupo de Biofísica Molecular: Liana Silva (companheira de
doutoramento), Rodrigo de Almeida, Bruno Castro e Sandra Pinto pela amizade,
ambiente de camaradagem e apoio no dia-a-dia do laboratório.
Aos antigos membros do grupo de Biofísica Molecular (Miguel Castanho, Sílvia
Lopes, Ana Silva e Renske Hesselink) pela sua amizade. A Sílvia Lopes merece um
agradecimento especial e com muito carinho, pela paciência, ajuda inestimável na
realização de medidas de dicroísmo linear, mas principalmente pela amizade e pelo
prazer da sua companhia! Um agradecimento especial também ao Prof. Miguel
Castanho pelo encorajamento e disponibilidade que sempre demonstrou.
Aos restantes membros do Laboratório de Biofísica Molecular da Faculdade de
Ciências da Universidade de Lisboa pela amizade e pelos momentos de diversão.
Ao Doutor Alexander Fedorov, pela contribuição essencial para este trabalho
através da realização das medidas de fluorescência resolvidas no tempo.
À todos os colegas do Centro de Química-Física Molecular pelo companheirismo.
Ao Prof. Marcus Hemminga, Doutor Ruud Sprüijt e Rob Koehorst por todo o apoio
concedido durante as minhas visitas ao seu laboratório. Um agradecimento especial ao
Rob Koehorst pela sua amizade, pela preocupação e pelo seu entusiasmo contagiante
pela ciência.
Ao Doutor António Pedro Matos pela realização das medidas de microscopia
electrónica de transmissão.
Ao Professor Benedito Cabral pela realização dos cálculos sobre o momento de
transição do fármaco SB24784.
Ao Professor João Pessoa por me ter facultado a utilização do aparelho de dicroísmo
circular.
À Fundação para a Ciência e Tecnologia, pelo apoio financeiro concedido (bolsa
SFRH/BD/14282/2003 e vários projectos no âmbito do POCTI e POCI).
À Comissão Europeia pelo apoio concedido através do contracto número QLG-CT-
2000-01801 (MIVASE consortium).
iii

Ao Pedro e Hugo, amigos de longa data, por me conseguirem arrancar de casa e
sempre ajudarem a manter a vida em perspectiva.
E finalmente gostava de agradecer aos meus irmãos e meus pais, pelo exemplo,
encorajamento e amor, sem os quais esta tese não seria certamente produzida.

CONTENTS
CONTENTS
Prefácio
Contents
Abbreviations and Symbol List
Resumo
Palavras-Chave
Abstract
Keywords
Sinopse
Outline
I - INTRODUCTION
I
V
IX
XI
XI
XIII
XIII
XV
XXI
1
1.- Biomembranes
1.1. – Function and architecture of biomembranes
1.2. – Molecular composition of biomembranes
1.2.1. – Lipid composition
1.2.1.1. – Glycerolipids
1.2.1.2. – Sphingolipids
1.2.1.3. – Glycolipids
1.2.1.4. – Sterols
1.3. – Lipid asymmetry across the bilayer
1.4. – Lipid structure and curvature
1.5. – Lamellar phase transitions in lipid bilayers
1.6. – Membrane thickness
1.7. – Lateral heterogeneity in lipid bilayer
1.8. – Membrane proteins
1.9. – Lateral dynamics in biomembranes
1.10. – Membrane model systems
1
1
3
3
3
5
6
6
7
8
10
12
14
17
20
21
2. – Lipid-Protein Interactions
2.1. – Membrane protein reconstitution
23
23
v

2.2. – Peptides as models
2.3. – Amphipatic helix
2.4. – Peptide partitioning to the membrane
2.5. – Anchoring of transmembrane domains
2.6. – Lipid-protein hydrophobic mismatch
2.7. – Transmembrane protein-lipid interface
2.8. – Lipid selectivity at protein interfaces
2.9. – Lipid phase preferential partition of membrane proteins
2.10. – Lipid sorting by proteins and formation of lipid domains
2.11. – Lipid-mediation of protein-protein interactions
25
26
28
30
31
35
37
41
43
44
II – PROTEIN-PROTEIN AND PROTEIN-LIPID INTERACTIONS
OF M13 MAJOR COAT PROTEIN
1. – Introduction
2. – Dependence of M13 Major Coat Protein Oligomerization and Lateral
Segregation on Bilayer Composition
3. – Quantification of Protein-Lipid Selectivity Using FRET: Application to
the M13 MCP
47
47
53
67
III – BINDING OF INHIBITORS TO A PUTATIVE BINDING
DOMAIN OF V-ATPase
1. – Introduction
2. – Interaction of the Indole Class of Vacuolar H + -ATPase Inhibitors with
Lipid Bilayers
3. – Binding Assays of Inhibitors Towards Selected V-ATPase Domains
79
79
83
95
IV – BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
1. – Introduction
2. – Ciprofloxacin Interactions With Bacterial Protein OmpF: Modelling of
FRET From a Multi-Tryptophan Protein Trimer
107
107
109

CONTENTS
V – INTERACTION OF HELIX-0 OF THE N-BAR DOMAIN WITH
LIPID MEMBRANES
1. – Introduction
2. – Role of Helix-0 of the N-BAR Domain in Membrane Curvature
Generation
137
137
141
VI – CLUSTERING OF PI(4,5)P 2 IN FLUID PC BILAYERS
1. – Introduction
2. – Absence of Clustering of Phosphatidylinositol-(4,5)-bisphosphate in
Fluid Phosphatidylcholine
169
169
173
VII – CONCLUSIONS
181
VIII – FINAL CONSIDERATIONS AND PROSPECTS
189
IX – BIBLIOGRAPHY
191
vii

ABBREVIATIONS AND SYMBOL LIST
ABBREVIATIONS AND SYMBOL LIST
C
cmc
C p
D
DLPC
DMPC
DMPE
DMPA
DOPC
DOPG
DPPC
DSC
DSPC
∆G 0
ESR
F
GUV
H0-NBAR
H I
H II
k
K p
K b
L
L i
L α
L β
L β’
L c
Specific heat
Critical micelle concentration
Heat capacity at constant pressure
Lateral diffusion coefficient
1,2-Dilauroyl-sn-glycero-3-phosphatidylcholine
1,2-Dimyristoyl-sn-glycero-3-phosphatidylcholine
1,2-Dimyristoyl-sn-glycero-3-phosphatidylethanolamine
1,2-Dimyristoyl-sn-glycero-3-phosphatidic acid
1,2-Dioleoyl-sn-glycero-3-phosphatidylcholine
1,2-Dioleoyl-sn-glycero-3-phosphatidylglycerol
1,2-Dipalmitoyl-sn-glycero-3-phosphatidylcholine
Differential scanning calorimetry
1,2-Distearoyl-sn-glycero-3- phosphatidylcholine
Free energy of binding
Electron spin resonance
Lipid perturbation energy
Giant unilamellar vesicle
N-terminal amphipatic helix of N-BAR domain
Hexagonal type I phase
Hexagonal type II (or inverted) phase
Boltzmann constant
Partition coefficient
Binding constant
Lipid phase
Lipid species i
Fluid liquid-disordered phase
Gel solid-ordered phase
Tilted gel phase
Crystalline phase
ix

L o
L/P
LUV
MARCKS
mcp
MLV
n S,i
PI(4)P
PI(5)P
PI(4,5)P
2
PI(3,4,5)P
3
PL i
P β’
PA
PC
PE
PG
PI
PS
R
SDS
sn
s o
SUV
V i
T
TFE
T m
TM
T p
W
Liquid-ordered phase
Lipid to protein/peptide ratio
Large unilamellar vesicle
Myristoylated alanine-rich C kinase substrate
Major coat protein
Multilamellar vesicle
Moles of solute in phase i
Phosphatidylinositol-(4)-phosphate
Phosphatidylinositol-(5)-phosphate
Phosphatidylinositol-(4,5)-bisphosphate
Phosphatidylinositol-(3,4,5)-trisphosphate
Complex of protein and lipid species i
Rippled Gel or Periodic Gel Phase
Phosphatidic acid
Phosphatidylcholine
Phosphatidylethanolamine
Phosphatidylglycerol
Phosphatidylinositol
Phosphatidylserine
Ideal gas constant
Sodium dodecyl sulfate
Stereospecific number
Gel or solid-ordered phase
Small unilamellar vesicle
Volume of phase i
Temperature
Trifluoroethanol
Main transition temperature
Transmembrane
Pre-transition temperature
Aqueous phase

RESUMO
RESUMO
As biomembranas são responsáveis pela operação de diversas funções celulares e a
multiplicidade destas interacções requer uma composição química complexa. Apesar da
estrutura básica das biomembranas ser a bicamada lipídica, outros componentes estão
presentes, em particular proteínas de membrana. Estes diferentes componentes
apresentam energias de interacção distintas, induzindo frequentemente heterogenidades
na distribuição lateral de biomembranas.
Neste trabalho foram estudadas as interacções entre componentes de biomembranas
em diferentes sistemas. É demonstrado que a distribuição lateral da proteína principal da
cápside do bacteriófago M13 (mcp) é dependente da composição lipídica da membrana.
Uma nova metodologia para quantificação de selectividades proteína-lípido foi
desenvolvida e aplicada com sucesso à mcp.
Estudos de ligação entre drogas e proteínas de membrana foram conduzidos
seguindo diferentes aproximações. No estudo da ligação de inibidores de V-ATPase a
um domínio transmembranar do enzima, uma aproximação reducionista foi escolhida,
enquanto que no estudo da ligação do ciprofloxacin para a OmpF, a proteína foi
utilizada na forma intacta.
A interacção entre um segmento N-terminal anfipático de um domínio BAR com
membranas foi investigada e conclusões foram obtidas relativamente ao papel deste
segmento na remodelação de membranas por domínios BAR.
Finalmente, a hipótese de agregação espontânea de fosfoinositol-(4,5)-bisfosfato foi
testada e excluída.
PALAVRAS‐CHAVE
Interacções Proteína‐Lípido
Proteína Principal da Cápside do M13
Inibição do V‐ATPase
PI(4,5)P 2
Domínios BAR
FRET
xi

ABSTRACT
ABSTRACT
Biomembranes are responsible for the performance of multiple cellular functions
and the multiplicity of these tasks requires a complex chemical composition. While the
structural framework of biomembranes is the lipid bilayer, other components, notably
proteins, are present. These distinct components present different energies of
interaction, leading to heterogeneous lateral distribution in the membrane.
In this work, the interactions between the components of biomembranes in several
different systems were studied. The lateral distribution of major coat protein (mcp) of
the M13 bacteriophage was shown to be highly dependent on the lipid composition of
the membrane. A new methodology for quantification of protein-lipid selectivity was
also developed and applied with success to the mcp.
Protein-drug binding studies were performed following different approaches. To the
study of binding of a V-ATPase inhibitor to a selected transmembrane domain of the
enzyme, a reductionist approach was considered, while in the case of binding of
ciprofloxacin to OmpF, the intact protein was used.
The interaction of an N-terminal amphipatic segment of a BAR domain with
membranes was also investigated and insight was achieved on the possible role of this
segment in membrane remodelling.
Finally, the hypothesis of spontaneous clustering of phosphatidylinositol-(4,5)-
bisphophate was tested and ruled out.
KEYWORDS
Protein‐Lipid Interactions
M13 Major Coat Protein
V‐ATPase Inhibition
PI(4,5)P 2
BAR domains
FRET
xiii

SINOPSE
SINOPSE
Nas últimas duas décadas, um esforço colectivo massivo permitiu um conhecimento
mais detalhado acerca da natureza das interacções entre os componentes de
biomembranas e as importantes consequências biológicas dessas mesmas interacções. O
modelo do mosaico fluído de Singer-Nicolson para biomembranas (1972) descrevia um
sistema em que os lípidos se apresentavam como pouco mais do que uma matriz onde
proteínas se insiriam. A visão actualmente aceite é muito mais complexa. Não só as
interacções proteína-proteína podem ser responsáveis pela mediação de funções
celulares, como as interacções proteína-lípido são consideradas muito mais complexas e
relevantes do que anteriormente. Estudos biofísicos detalhados como os aqui descritos,
permitem a recolha de informação valiosa relativamente às características destas
interacções.
As biomembranas traçam os limites para a vida celular. Elas permitem a
diferenciação de ambientes moleculares através do controle da permeabilidade, criando
as condições para a especialização no interior da célula. Esta tarefa requer o consumo de
quantidades de energia massivas, enfatizando a importância dramática destas estruturas.
A base estrutural das biomembranas é a bicamada lipídica. No entanto, os lípidos
estão longe de corresponder a uma classe homogénea de moléculas. Quando as energias
de interacção entre moléculas lipídicas são significativamente diferentes, separação de
fases em grande escala pode ocorrer e mesmo para pequenas diferenças entre energias
de interacção lípido-lípido, heterogenidades em pequena escala são expectáveis. Assim,
a distribuição lateral de uma bicamada lipídica é frequentemente heterogénea.
Esta representação das biomembranas torna-se ainda mais complexa tendo em conta
que os lípidos não são os únicos componentes presentes. De facto,as biomembranas são
uma mistura de diferentes classes de moléculas com diferentes funções. Além dos
lípidos, os componentes mais importantes são as proteínas, que se encontram altamente
concentradas nas biomembranas e são largamente responsáveis pelas funções celulares
operadas nestas estruturas.
As interacções entre estas diversas espécies moleculares são bastante complexas.
Ligações de hidrogénio, interacções electrostáticas, e de van der Waals, operam
simultaneamente na ordenação do empacotamento de equilíbrio de proteínas e lípidos.
Proteínas e lípidos organizam-se de forma a proteger os seus domínios hidrófobos do
xv

ambiente aquoso e assim prevenir as perdas entrópicas resultantes da interacção com as
moléculas de água. Como consequência, proteínas de membrana e lípidos interactuam
de forma diferenciada, dependendo da distribuição dos resíduos polares e hidrófobos na
proteína e do comprimento do domínio hidrófobo no lípido.
Estas alterações nas selectividades de associação podem levar a alterações na
distribuição lateral dos componentes de membrana, de uma forma similar à observada
para misturas lipídicas. Desta forma, a distribuição heterógenea de lípidos pode induzir
uma distribuição heterogénea de proteínas e vice-versa. De facto, é conhecida
actualmente a tendência de proteínas e lípidos para se distribuir de forma heterogénea
em biomembranas e este fenómeno é essencial para garantir o carácter localizado de
determinados eventos celulares.
As drogas são agentes que frequentemente também demonstram afinidade por
membranas lipídicas. Os alvos de diversas drogas são proteínas de membrana e a sua
acção é frequentemente realizada no ambiente membranar. Assim, as interacções de
drogas com componentes das biomembranas são de grande relevância biológica. Da
mesma forma, para drogas dirigidas a proteínas no interior da célula, a partição para a
membrana plasmática é o primeiro passo no processo de entrada no ambiente celular e
os seus coeficientes de partição água/lípido são deste modo parâmetros biologicamente
relevantes.
As interacções entre os componentes das biomembranas são utilizadas pelas células
para mediar e localizar diversas funções nas membranas. Uma melhor compreensão
destes mecanismos é o primeiro passo na compreensão das biomembranas e das suas
funções. No entanto, as membranas celulares são muitas vezes sistemas demasiado
complexos para estudar em detalhe as características das interacções estabelecidas pelos
seus componentes. Uma alternativa viável é o estudo destas interacções em sistemas
modelo de membranas. No presente trabalho, a maioria dos estudos foram efectuados
em vesículas unilamelares com diâmetro aproximado de 100 nm. Este sistema modelo
permite uma base de estudo simplificada e a manutenção das propriedades fundamentais
das biomembranas.
As técnicas de fluorescência são ferramentas ideais no estudo da distribuição lateral
em membranas. Através da escolha de um aminoácido fluorescente intrínseco (Trp ou
Tyr), ou através da marcação de proteínas e/ou lípidos, é possível monitorar as
alterações de fluorescência (intensidade, tempos de vida e anisotropia) resultantes da
interacção com outros componentes membranares. Estudos de extinção de fluorescência

SINOPSE
podem fornecer informação acerca das eficiências de colisão, e esta informação pode
dar indicações sobre: i) a localização do grupo fluorescente; ii) a concentração do
agente responsável pela extinção de fluorescência; iii) as propriedades difusivas do
agente responsável pela extinção de fluorescência e da molécula fluorescente. Uma
técnica que é particularmente útil na avaliação das distribuições laterais na membrana é
a transferência de energia electrónica por ressonância (FRET). A eficicência de FRET
para um único par dador-aceitante depende fortemente da distância entre eles. No caso
de membranas lipídicas com uma distribuição de dadores e aceitantes, FRET é sensível
às alterações na distribuição de aceitantes próximos ao dador, numa escala espacial
comparável à distância de Förster (até 10 nm dependendo do par dador-aceitante).
Vários sistemas foram aqui alvo de estudo. No Capítulo II, uma proteína integral
também pertencente à cápside do bacteriófago M13 (mcp), foi estudada em sistemas
modelo de membranas. O estado de oligomerização da proteína sofre alterações
repetidas durante o ciclo reproductivo do bacteriófago M13. No entanto, crê-se que o
estado da proteína quando incorporada na membrana do hospedeiro é essencialmente
monomérico, pois tal é necessário para a correcta incorporação da proteína numa
partícula de bacteriófago em formação. Ainda assim, esta questão é alvo de debate,
uma vez que estados oligomerizados da proteína foram observados em certas condições.
Aqui é demonstrado que mcp é um monómero quando incorporado em membranas
compostas por lípidos com espessura hidrófoba idêntica à da proteína (cadeias
monoinsaturadas com 18 carbonos – conformidade hidrófoba). Por outro lado, quando
foram utilizadas bicamadas apresentando uma espessura hidrófoba maior que a da
proteína (cadeias monoinsaturadas com 22 carbonos), foi detectada agregação. Quando
foram utilizadas misturas de lípidos com espessuras hidrófobas em conformidade e
descomformidade com a espessura hidrófoba da proteína, foi observada uma
segregação de mcp para domínios enriquecidos em lípidos com conformidade
hidrófoba. A mcp nestes domínios manteve um estado monomérico. Um aumento de 4
carbonos nas cadeias acilo da matriz lipídica pode induzir agregação proteica, e no caso
da presença de uma fracção significativa de lípido em conformidade hidrófoba,
segregação lípido-lípido pode ser induzida pela presença da proteína. Este resultado é
demonstrativo da enorme relevância da conformidade hidrófoba para a organização de
proteínas e lípidos.
Uma metodologia de FRET foi desenvolvida com o intuito de quantificar o grau de
selectividade exibida pela mcp relativamente a diferentes lípidos. Os resultados obtidos
xvii

para diferentes grupos polares de lípidos, são practicamente idênticos aos obtidos com
espectroscopia de ressonância paramagnética eletrónica (ESR) para a mesma proteína,
ainda que num estado agregado. Esta concordância estabelece esta metodologia para
quantificação da selectividade de proteínas para lípidos como uma alternativa ao uso de
ESR. Esta metodologia de FRET foi também aplicada ao estudo da selectividade da
mcp para diferentes comprimentos de cadeias acilo. Surpreendentemente, os resultados
revelaram constantes de selectividade baixas para um lípido de conformidade hidrófoba,
em bicamadas mais finas e mais espessas que o domínio hidrófobo da proteína.
Duas estratégias diferentes foram utilizadas no estudo da ligação de drogas a
proteínas de membrana. No Capítulo III, uma aproximação reducionista foi escolhida.
Um péptido correspondente ao quarto domínio transmembranar da subunidade c do V-
ATPase foi utilizado em estudos de ligação aos inibidores do V-ATPase, bafilomicina e
SB242784. Era esperado que este domínio transmembranar compreendesse o local de
ligação a estes inibidores. SB242784 foi desenvolvido para utilização terapêutica em
pacientes com osteoporose. Este inibidor apresenta grande selectividade para a forma
osteoclática do V-ATPase que é parcialmente responsável pela degradação óssea. A
bafilomicina apresenta níveis de toxicidade muito elevados devido à ausência de
selectividade para uma forma particular do enzima. FRET foi novamente utilizada para
determinar ligação entre inibidor e péptido (a tirosina do péptido foi o dador e os
inibidores os aceitantes) e os resultados indiciam uma interacção de baixa afinidade
entre a bafilomicina e o péptido. Já a ligação de SB242784 ao péptido não foi detectada.
Globalmente, os resultados indicam que os requesitos moleculares para ligação aos
inibidores na subunidade c incluem provavelmente contribuições dos restantes domínios
transmembranares, reflectindo uma visão mais complexa do mecanismo de inibição do
V-ATPase do que inicialmente considerado.
No Capítulo IV, descreve-se um estudo de ligação entre o antibiótico ciprofloxacin
(CP) e um trímero purificado de uma porina da membrane externa, OmpF. O
mecanismo de entrada do CP na célula, requer a interacção do CP com a OmpF. Neste
caso, a estratégia escolhida envolveu o uso da estrutura nativa da proteína. FRET foi
novamente aplicada, desta vez fazendo uso da presença de dois tipos de Trp no
monómero de OmpF (como dadores) e das propriedades de absorção no ultravioleta do
CP (como aceitante). A análise da extinção de fluorescência de aminoácidos intrínsecos
de proteínas de grandes dimensões por FRET pode revelar-se difícil, uma vez que cada
uma das populações de dadores pode estar sujeita a diferentes distribuições de

SINOPSE
aceitantes. Dadores mais próximos da periferia da proteína podem estar mais próximos
de aceitantes do que dadores no centro da proteína. Tendo em conta esta preocupação,
um modelo de FRET foi desenvolvido para aplicação em sistemas contendo proteínas
com múltiplos resíduos de Trp, considerando as restrições geométricas da OmpF. Esta
metodologia permitiu a recuperação de um intervalo possível para a constante de
ligação entre a OmpF e o CP. Em virtude das grandes dimensões do trímero de OmpF,
dependendo do local de ligação ao antibiótico assumido no modelo utilizado na análise
dos dados de FRET, a eficiência de ligação pode ser diferente. Dois locais de ligação ao
CP correspondendo a situações limite (ligação no centro do trímero e ligação à periferia
do trímero) foram considerados. Comparando os resultados obtidos segundo estes
modelos e os resultados obtidos através de um método independente, foi possível
concluir que o local de ligação do CP à OmpF está provavelmente distante do centro do
trímero e mais próximo da periferia.
No Capítulo V, a interacção de um péptido anfipático N-terminal de um domínio N-
BAR com sistemas modelo de membranas foi estudada. Diversos autores consideraram
a hipótese de que este segmento contribui para as propriedades remodeladoras de
membranas do domínio N-BAR (tubulação de liposomas esféricos), através da inserção
profunda no interior da bicamada. Isto teria como consequência a indução de assimetria
entre as monocamadas dos lipossomas, que por fim levaria à alterações da curvatura na
membrana. Esta hipótese foi aqui testada através do estudo das interacções de um
péptido correspondendo a este segmento com sistemas modelo de membranas.
Novamente, metodologias de fluorescência foram utilizadas e complementadas com
microscopia electrónica. Conclui-se que a ligação do péptido às membranas é
essencialmente electrostática, afastando a hipótese frequentemente levantada de este
segmento dos domínios N-BAR agir através da estabilização da conformação do
domínio associado com membranas, mediante fortes interacções hidrófobas com o
interior da bicamada lipídica. De qualquer forma, os resultados de um estudo de FRET
apontam para um papel alternativo do segmento N-terminal do domínio N-BAR. Os
resultados demonstram claramente que o péptido se encontra na forma dimerizada
depois da ligação à membrana. Estes resultados estão em concordância com uma
proposta recente de que o segmento anfipático N-terminal dos domínios N-BAR poderia
agir através da mediação da associação entre dímeros de domínios BAR. A microscopia
electrónica não permitiu identificar alterações morfológicas significativas dos
lipossomas após interacção com o péptido.
xix

Finalmente, a agregação de fosfoinositol-4,5-bisfosfato (PI(4,5)P 2 ) em bicamadas de
fosfocolinas foi estudada (Capítulo VI), de forma a testar a hipótese de agregação
espontânea deste lípido numa matriz de fosfocolinas. Os resultados de extinção de
fluorescência colisional obtidos com um derivado de PI(4,5)P 2 marcado com uma sonda
fluorescente e de um estudo de FRET utilizando este lípido derivatizado como dador
para uma sonda distribuida homogeneamente na membrana, são inequívocos na
confirmação de uma distribuição homogénea de PI(4,5)P 2 na bicamada lipídica. No
mesmo estudo, é demonstrado que a diferentes estados de protonação de PI(4,5)P 2
correspondem diferentes eficiências de partição para o ambiente lipídico, um fenómeno
que pode apresentar relevância biológica significativa.

OUTLINE
OUTLINE
The last two decades, through a massive collective effort, allowed some insight into
the nature of the interactions between the components of biomembranes and also the
highly relevant biological consequences of these interactions. The Singer-Nicolson fluid
mosaic model for biomembranes (1972), described a system where lipids were little else
than a matrix where proteins were dispersed. The picture currently accepted is much
more complex. Not only protein-protein interactions can be responsible for mediation of
cellular functions, but also protein-lipid interactions are much more complex and
important than originally thought. Detailed biophysical studies such as the ones
described here, allow gathering of valuable information regarding the character of such
interactions.
Biomembranes delineate the boundaries of cellular life. They allow the
differentiation of molecular environments through the control of permeability, creating
the conditions for specialization within the cell. This task demands the expenditure of a
massive amount of energy, emphasizing the dramatic importance of biomembranes.
The foremost structural framework of biomembranes is the lipid bilayer, but lipids
are far from being a homogeneous class of molecules. When the energies of interaction
between lipid molecules are significantly different, large scale phase separation of lipid
molecules in the membrane can occur, and even for small differences in lipid-lipid
interaction energies, short-range heterogeneities are expected. Therefore, the lateral
distribution of a multicomponent lipid bilayer is often heterogeneous.
This portrait of biomembranes is made even more complex by the notion that lipids
are not the only components found there. In fact, biomembranes are a mixture of
different classes of molecules with different functions. Apart from lipids, the most
notable component are proteins, which are highly concentrated in biomembranes and
are largely responsible for the cellular functions operated there.
Interactions between these quite different molecular species are very complex.
Hydrogen bonds, electrostatic, and van der Waals forces, all operate simultaneously in
dictating the equilibrium packing of proteins and lipids in bilayers. Notably, proteins
and lipids will arrange in such a manner that their hydrophobic domains are shielded
from the aqueous environment in order to prevent the entropy loss resulting from
interaction with water molecules. As a consequence, membrane proteins and lipids will
xxi

interact differentially, depending on the distribution of polar and hydrophobic residues
on the protein and on the size of the hydrophobic lipid domain.
These changes in selectivities of association can lead to changes in the lateral
distribution of the membrane components in a similar manner as observed for lipid
mixtures. In this way, a heterogeneous distribution of lipids can drive membrane protein
heterogeneity and vice-versa. In fact, it is known that proteins and lipids are often
distributed in a heterogeneous manner in the membrane, and this is essential in ensuring
the localized character of some events in cellular membranes.
Drugs are another type of agents which often display affinity for lipid membranes.
The targets of several drugs are membrane proteins, and their action is frequently
expected to be exerted in the membrane environment. Therefore, interaction of drugs
with biomembrane components is of great biological relevance. In addition, for drugs
targeting proteins inside the cell, partition to the plasma membrane is the first step to
cell entry, and water/lipid partition coefficients are in this way, relevant parameters in
drug research.
It is clear that interactions between biomembrane components are used by cells to
mediate and localize diverse functions in the cell membrane. A better understanding of
these mechanisms is the first step in rationalizing biomembranes and their functions.
However, cellular membranes are often too intricate for the properties of some
membrane component interactions to be resolved, and a valuable alternative is the use
of membrane model systems. Here, the majority of the studies were carried out in large
unilamellar vesicles, with a diameter around 100 nm. This model system granted us a
simplified basis for the study of the interactions between membrane components, at the
same time maintaining all the fundamental properties of the biomembrane.
Fluorescence techniques are ideal tools to study lateral distribution of membranes.
By choosing an intrinsic fluorescent amino acid (Trp or Tyr), or fluorescent labelling of
proteins and lipids, we can monitor fluorescence changes (intensity, lifetime,
anisotropy) upon interaction with other membrane components. Fluorescence quenching
studies can give information on the collision efficiency, and this can give us indications
on: i) the location of the fluorescent group; ii) on the concentration of quencher; and iii)
on the diffusion properties of both quencher and fluorescent molecule. One technique
that was particularly useful in evaluating lateral distributions in the membrane was
Förster resonance energy transfer (FRET). FRET efficiency for a single donor-acceptor
pair is strongly dependent on the distance between the two. In the case of lipid

OUTLINE
membranes with a distribution of donors and acceptors, FRET will be sensitive to
changes in the distribution of acceptors around the donor population in the space range
of the Förster radius (up to ~10 nm depending on the donor-acceptor pair).
Several different systems were investigated. In Chapter II, an integral protein from
the coat of the bacteriophage M13 (mcp) was studied in model membranes. The
oligomerization state of the protein changes repeatedly during the reproductive cycle of
bacteriophage M13, but the state of the protein in the membrane was thought to be
monomeric as this is apparently required for correct insertion in nascent bacteriophages.
However, this was still matter of debate, as in the past, oligomeric mcp was shown to
exist in some conditions. Here, it is shown that mcp is a monomer when incorporated in
membranes composed by lipids with hydrophobic thickness that matched the
hydrophobic thickness of the protein (monounsatured chains C 18 chains). On the other
hand, when bilayers with thicker hydrophobic thickness (monounsatured chains C 22
chains) were used, aggregation was observed. Surprisingly, when mixtures of both
matching and mismatching lipids were used, segregation of mcp to domains enriched in
matching lipid was apparently induced, and while in these domains, mcp remained
monomeric. This result is a dramatic demonstration of the relevance of hydrophobic
matching for protein-lipid organization. Only a small increase of 4 carbons in the acylchain
of the matrix lipid can drive protein aggregation, and in case that a significant
fraction of matching lipid is present, lipid-lipid segregation can be induced by the
presence of the protein.
A FRET methodology was developed to quantify the degree of selectivity exhibited
by mcp to different lipids. The results obtained for lipids with different headgroups are
almost perfectly matched to previous results obtained with electron spin resonance
(ESR) for the same protein, even though in an aggregated state. This agreement
establishes the FRET methodology for quantification of protein-lipid selectivity
described here as an alternative to the use of ESR. This FRET methodology was also
applied to the study of selectivity for acyl-chain length, which is as already mentioned,
essential in dictating the lateral distribution of the protein in liposomes. Also
surprisingly, the results revealed weak selectivity constants for a hydrophobically
matching lipid, both in thinner as in thicker bilayers.
Drug binding to membrane proteins was studied following two different strategies.
In chapter III, a reductionist approach was used. A peptide expected to correspond to the
putative 4 th transmembrane domain of the membrane bound c-subunit of V-ATPase,
xxiii

was used in binding studies for the V-ATPase inhibitors bafilomycin and SB242784.
This transmembrane domain is expected to comprise the binding site for the studied
inhibitors. SB242784 was developed to be used as a potential drug for treatment of
osteoporosis. This inhibitor presents selectivity for the osteoclastic form of the enzyme,
while bafilomycin is extremely toxic due to its lack of selectivity. FRET was again used
(a tyrosine residue of the peptide was used as a donor and the inhibitors as acceptors) to
determine binding, and the results indicate a weak binding of the chosen peptide with
bafilomycin whereas binding of the peptide to SB242784 was not detected. Overall, the
results indicate that the V-ATPase inhibitor binding site is likely not formed only by the
4 th transmembrane segment of the c-subunit of V-ATPase, but is the result of
contributions from other transmembrane domains, reflecting a more complex view of
the inhibitory mechanism of V-ATPase than originally proposed.
In Chapter IV, a binding study was conducted between ciprofloxacin (CP), a
quinolone antibiotic, and a purified trimer of the outer membrane porin, OmpF. CP
requires interactions with OmpF for efficient entry into the cell. In this case, the native
protein structure was used instead of the minimalist approach described in Chapter III.
FRET was again applied, this time making use of the presence of two types of
tryptophans of OmpF (as donors) and the UV absorbing properties of CP (acceptor).
Fluorescence from intrinsic amino acids of large proteins in FRET can be difficult to
analyze, since each donor (fluorescent amino acid) population is expected to sense a
different population of acceptors. Donors closer to the protein periphery, can e.g., be in
closer proximity to acceptors than donors in the core of the protein. With this in mind, a
FRET methodology suitable for application to FRET studies with multi-donor proteins
was developed according to the geometric restrictions of the OmpF system. This model
allowed the recovery of a range for the binding constants of the OmpF-CP association
process. Due to the large dimensions of the OmpF trimer, depending on the site of
inhibitor binding assumed in the model for FRET data analysis, the retrieved binding
efficiency can be different. Two limiting binding sites were considered, and comparing
the results from our analysis with the results obtained with an independent method, it
was possible to conclude that the binding site for CP in OmpF is likely to be displaced
from the center of the trimer and closer to the periphery.
In Chapter V, the interaction of the N-terminal amphipatic peptide of a N-BAR
domain with model membranes was studied. Several authors hypothesized that this
segment of N-BAR domain contributed to the membrane remodelling properties of N-

OUTLINE
BAR domains (tubulation of spherical liposomes) by inserting deeply into the bilayer
core, driving monolayer asymmetry and thereby forcing a change in the normal
curvature of the bilayer. We tested this hypothesis by following the interaction of the
corresponding peptide with membrane model systems. Again, fluorescence
spectroscopy methodologies were used and complemented with the application of
electron microscopy to investigate changes in liposome morphology upon binding of the
peptide. It is concluded that binding to liposomes is essentially electrostatic, ruling out
the frequently hypothesized role of this protein segment to operate through the
stabilization of the membrane bound conformation of BAR domains via hydrophobic
short-range interactions with the bilayer core. Nevertheless, the results of a FRET study
clearly indicate the tendency of the peptide to oligomerize in the lipid membrane
environment, supporting a recently proposed function of this segment as a mediator of
aggregation between BAR domain dimers. Electron microscopy measurements were not
able to detect significant morphology changes in the liposomes upon addition of
peptides.
Finally, clustering behaviour of phosphatidylinositol-4,5-bisphophate (PI(4,5)P 2 ) in
phosphatidylcholine bilayers was studied (Chapter VI) in order to test the hypothesis of
spontaneous aggregation of this lipid in a PC matrix. Both fluorescence self-quenching
data obtained with a fluorescently labelled PI(4,5)P 2 , and a FRET study using this lipid
as a donor to a homogeneously distributed membrane probe, unequivocally confirmed a
homogeneous distribution of PI(4,5)P 2 in the lipid bilayer. In the same study, it is
demonstrated that different protonation states of PI(4,5)P 2 correspond to different
water/lipid partition coefficients, a phenomenon that can be of significant biological
relevance.
xxv

INTRODUCTION: BIOMEMBRANES
I
INTRODUCTION
1.BIOMEMBRANES
1.1. Function and architecture of biomembranes
Biomembranes delineate the boundaries of cells as plasma membranes, and in the
eukaryotic cell, enclose organelles in the form of intracellular membranes, allowing the
subdivision of cellular activities and the more diverse and specialized functions found in
eukaryotes. The biomembrane is the passive permeability barrier that allows the
maintenance of different molecular environments in the inside and outside of the cell or
organelle. The importance of this task is emphasized by the fact that a significant
fraction of the energy required for life is expended in the preservation of the differences
in molecular environments across biomembranes.
The foremost structural framework of the biomembrane is the lipid bilayer. The
lipid bilayer is formed by spontaneous self-assembly of lipid molecules. During this
process, the decrease in entropy resulting from hydrocarbon-water interaction, which is
also called the hydrophobic effect (Tanford, 1980), acts to organize lipid molecules so
that acyl-chains are screened from the water environment. This can also lead to the
formation of micelles or other types of organization depending on the structure of the
lipid, as will be discussed in chapter 1.4.
Due to the amphipatic nature of most lipids, the lipid bilayer presents a hydrocarbon
environment in the interior core screened from water molecules by the polar groups of
lipids (see Fig. I.1). The hydrophobic acyl-chains are in a fluid state whereas the polar
groups of the lipids are assembled in an orderly array as in a liquid crystal. This
hydrophobic core of the lipid bilayer is the most important structural feature in the role
of biomembranes as barriers to passive molecular diffusion. It allows the passage of
water and other small uncharged molecules but presents great impediment to passive
1

ion diffusion, as the energy required to move a hydrated ion through it, is extremely
high (for a monovalent ion with 2 Å radius, it takes about 40 Kcal/mol to transfer it into
the membrane core – Gennis, 1989).
Figure I.1 – a) Electron micrograph of a section from the plasma membrane of a erythrocyte membrane.
b) Schematic depiction of a lamellar arrangement of a lipid bilayer. Polar headgroups face outwards and
shield the interior hydrophobic tails (from Lodish et al., 2000).
The bilayer can be regarded as a two-dimensional solvent that provides the
hydrophobic anchor to membrane proteins in biomembranes. This two-dimensional
fluid character of the lipid bilayer is central to lipid-lipid, lipid-protein and proteinprotein
interactions. It is responsible for an increase of the efficiency of membranebound
enzymes whose substrates are located in the plane of the bilayer. The average
time for a reactant to reach a target site depends drastically on the dimensionality.
Reactions that require collisions of three bodies are very rare in three dimensions, but
can be very effective in two dimensions (Hardt, 1979; Sackmann, 1995).
The biomembrane is not only a physical barrier between the inside and outside of
the cell or between intracellular compartments and the cytosol. As the cell needs to get
nutrients in and out, the membrane must accommodate this. Some membrane-bound
2

INTRODUCTION: BIOMEMBRANES
proteins are responsible by the regulation of the transport of ions and molecules across
the membrane. Thus, the biomembrane is actually a selective barrier. To this effect, the
distributions of both lipids and proteins exhibit asymmetry in each side of the bilayer.
Another consequence of the fluid character of the lipid bilayer is its high flexibility
that allows for an easy adaptation to exterior perturbations. In biomembranes, the lipid
bilayer also interacts with the cytoskeleton. This interaction is responsible for the
mechanical stability of cells that combined with the flexibility of the lipid bilayer
confers the biomembrane with unique mechanical properties.
1.2. Molecular composition of biomembranes
1.2.1. Lipid composition
Biomembranes are typically composed by four classes of lipids: glycerolipids,
sphingolipids, glycolipids and sterols. Their common characteristic is the amphipatic
character which determines the lipid bilayer structure.
1.2.1.1. Glycerolipids
The hydrophobic section of glycerolipids is composed of linear hydrocarbon chains
esterified to sn-1 and sn-2 (stereospecific numbers) positions of a glycerol moiety.
These hydrocarbon chains exhibit great diversity of length (commonly from 14 to 24
carbon atoms) and unsaturation (from 0 to 4) in biomembranes. A list of some of the
most common acyl-chains and their nomenclature is shown in Table I.1. A short
notation normally used for describing them is based on the number of carbons and
double bonds. Thus 18:1 stands for an acyl-chain with 18 carbons in the hydrocarbon
chain and one double bond. In nature, virtually all double bonds in fatty-acids present a
cis configuration and generally the longer the hydrocarbon chain, the more double
bounds are present. The cis configuration has drastic consequences in the packing of
lipids (Berg et al., 2002) as will be discussed later. The number of carbons in the
hydrophobic chain is nearly always even. In biomembranes, lipids generally contain two
different acyl-chains and frequently one of them is unsaturated (Mouritsen, 2005).
The predominant class of lipids in biomembranes are the phosphoglycerolipids.
These molecules are glycerolipids in which the sn-3 position is esterified to phosphoric
3

acid. If no more groups are linked to the molecule, the resulting compound is
phospatidic acid (PA), the simplest phosphoglycerolipid. In general, the phosphate
group is linked to an alcohol, and together, the phosphate group and the bound alcohol
shape the polar headgroup section of the molecule. Phosphoglycerolipids are classified
according to the nature of the alcohol moiety in the headgroup. The most common
alcohols found in phospholipids are serine, ethanolamine, choline, glycerol, and inositol
(Figure I.2). The corresponding phosphoglycerolipids are named phosphatidylserine
(PS), phosphatidylethanolamine (PE), phosphatidylcholine (PC), phosphatidylglycerol
(PG), and phosphatidylinositol (PI). The latter can also be phosphorylated at three
different positions in the inositol ring resulting in seven different combinations that
correspond to seven different species as will be discussed in chapter VI.
Table I.1 - Nomenclature of relevant acyl-chains
Short notation
Name
14:0 Myristoyl
14:1 (9-cis) Myristoleoyl
16:0 Palmitoyl
16:1 (9-cis) Palmitoleoyl
18:0 Stearoyl
18:1 (9-cis) Oleoyl
18:2 (9,12 cis) Linoleoyl
18:3 (6,9,12–cis) γ-Linolenoyl
18:3 (9,12,15–cis) α-Linolenoyl
20:0 Arachidoyl
20:4 (5,8,11,14–cis) Arachidonoyl
22:0 Behenoyl
22:1 (13-cis) Erucoyl
The phosphate group in phosphoglycerolipids is always negatively charged, and the
net charge of the molecule is dictated by the charge of the alcohol moiety. Thus PA, PG,
PS, PI, and phosphorylated PI are negatively charged, while PC and PE present a net
neutral charge due to the positively charged amine in choline and ethanolamine. PC is
the most abundant phosphoglycerolipid in the plasma membrane, whereas PE is in
general the major component in bacterial membranes.
4

INTRODUCTION: BIOMEMBRANES
Figure I.2 – a) Schematic representation of the phosphoglycerolipid structure. b) Most common alcohols
found in the headgroups of phosphoglycerolipids. From Berg et al., 2002). c) Structure of a
phosphatidylcholine - 1,2-dimyristoyl-sn-glycero-3-phosphatidylcholine (DMPC).
The acyl-chains found in each class of phosphoglycerolipids is also different.
PC’s are mainly composed of short saturated chains (16 to 18 carbons) or 18:1 and 18:2
unsaturated chains. PE’s have a large fraction of polyunsaturated chains, in particular
20:4 (arachidonoyl). Charged phosphoglycerolipids also present a significant content of
unsaturated acyl-chains (Sackmann, 1995).
Other types of glycerolipids are found in biomembranes, namely the
plasmalogens, in which one of the acyl-chains is bound to glycerol by a vinyl ether
linkage. Plasmalogens constitute about 20 % of the total content of phosphoglycerides
in humans, although their abundance varies dramatically among tissues and species.
Human brain and heart tissues are particularly enriched in this class of lipids (Lodish et
al., 2000). Cardiolipin is a dimer lipid since it contains four acyl chains. It constitutes
about 20 % of the inner mitochondrial membrane and is also found in the membrane of
plant chloroplasts and of certain bacterias.
1.2.1.2. Sphingolipids
Sphingolipids are based on sphingosine instead of glycerol. Sphingosine is an
aminoalcohol with a long unsaturated hydrocarbon chain. A long fatty-acid is attached
to sphingosine by an amide bound. This basic structure is a ceramide, which is involved
in cell death and is an essential component of skin (Mouritsen, 2005). The linkage of a
choline to the terminal hydroxyl group of sphingosine in ceramide leads to
5

sphingomyelin, the most abundant sphingolipid, which is a significant component of
animal plasma membranes.
1.2.1.3. Glycolipids
Both glycerolipids and sphingolipids can be linked to a carbohydrate in their
headgroups, giving rise to a glycolipid. Glycolipids play important roles in the
interactions of the cell with its surroundings. Glycoglycerolipids are very abundant in
chloroplast membranes, as well as in blue algae and bacteria. They are however rarely
found in animals (Gennis, 1989). In glycosphingolipids, one or more carbohydrates are
linked to the terminal hydroxyl group of the sphingosine backbone. The simplest form
of glycosphingolipid is a cerebroside that contains a single carbohydrate residue, either
glucose or galactose. Complex glycosphingolipids called gangliosides present one or
two branched carbohydrate chains containing sialic acid groups. Glycosphingolipids are
as a rule located in the outer surface of the plasma membrane with varying
concentrations (2-10 %), and are particularly abundant in the nervous system.
1.2.1.4. Sterols
The basic structure of sterols is a four ring hydrocarbon. This structure confers
the molecule much more rigidity than other lipids. The most important sterol in animal
membranes is cholesterol (Fig. I.3). In cholesterol, a hydrocarbon chain is linked to one
end of the ring system and a hydroxyl group is attached to the other end of the structure,
this being the polar headgroup of the molecule. In the lipid bilayer the molecule is
oriented parallel to the fatty-acid chains of other lipids while the hydroxyl group
interacts with nearby headgroups. Cholesterol is only found in eukaryotes, particularly
in animal plasma membranes, lysosomes, endosomes and Golgi apparatus. The
concentration of cholesterol is especially high in the animal plasma membrane (20 - 30
%). In plants cholesterol exists in low amounts, and other sterols are present, namely
sitoesterol and stigmasterol. Ergosterol is found in yeast and other eukaryotic
microorganisms.
6

INTRODUCTION: BIOMEMBRANES
Figure 1.3 – Structure of cholesterol (From Berg et al., 2002).
1.3. Lipid asymmetry across the bilayer
Lipid composition in the cytoplasmatic and exoplasmatic leaflets of the bilayer is
asymmetric. In plasma membranes, the cytoplasmatic leaflet is enriched in PE, PS, and
PI, while in the exoplasmatic leaflet lipids like PC, glycolipids, sphingomyelin and
cholesterol exist in higher concentrations.
One of the factors contributing to this lipid asymmetry is the site of synthesis of the
lipids in the membranes of the endoplasmic reticulum and Golgi, as it dictates the
plasma membrane leaflet where the lipid is to be inserted. In this way, addition of
carbohydrates to glycolipids is done through the luminal side of the Golgi apparatus that
is topologically equivalent to the exterior of the cell. This however does not account for
all asymmetry observed and the action of ATP-powered transport proteins called
translocases is essential (Lodish et al., 2000).
However, asymmetry would be lost if lipids easily crossed from one leaflet to the
other. In fact, in protein free liposomes, the kinetics of this process is extremely slow,
on the order of hours/days. This is due to the extremely energetically unfavourable
process of inserting the polar headgroup of a lipid in the hydrophobic membrane
interior.
Lipid asymmetry is of functional importance, and many cytosolic proteins bind to
specific lipid headgroups (like PS or PI) in the cytosolic leaflet of the bilayer. Animals
also use phospholipid asymmetry in the plasma membrane of cells as a control to
distinguish between dead and healthy cells, as when animal cells undergo programmed
cell death, or apoptosis. In this case, PS lipids that are normally found enriched in the
cytosolic leaflet, distribute between both leaflets of the bilayer. This functions as a
7

signal for neighbouring cells to phagocytose the dead cell and digest it (Alberts et al.,
2002).
1.4. Lipid structure and curvature
Lipids in hydrated conditions exhibit polymorphic behaviour as they can assemble
in different structures. The structure adopted by a lipid aggregate can be influenced by
the molecular structure of the lipid itself and a myriad of environmental conditions,
such as water content, pH, ionic strength, temperature and pressure. Lipid molecules can
assemble in an aqueous environment either in a lamellar structure (as in a lipid bilayer)
or in non-lamellar phases (micelle, hexagonal, inverted hexagonal phases, and the cubic
phase (see Fig. I.4)). Because phase transitions between these different aggregates can
be activated by changes in water content, these phases are called lyotropic.
The different possibilities for organizations of lipids are the result of the intrinsic
shape of the lipid molecule. When the headgroup of a lipid occupies the same area than
the area occupied by its hydrophobic section, the leaflets or monolayers formed by
association of molecules of this lipid will present zero spontaneous curvature, forming a
planar membrane for which both leaflets (or monolayers) present null curvature. A lipid
bilayer composed of this lipid with the same number of molecules in each monolayer
should be flat. However, in order to eliminate the aqueous exposure of the hydrophobic
edges of the bilayer, the bilayer can unite the edges and form a curved vesicle. In this
aggregate the exterior monolayer must present a concave (or positive) curvature and the
interior monolayer must present a curvature with opposite direction. In this way,
spontaneous curvature of a lipid monolayer refers to the curvature observed in the
absence of edge conditions (Zimmerberg, 2000).
For monolayers composed of lipids presenting headgroups with a cross-section
different to that of the hydrophobic tails, a spontaneous curvature will be present, and
the packing of these lipids in a lipid bilayer with a lamellar structure will result in
curvature stress, that can be supported by the lamellar structure only up to a certain
extent. In case the lipid bilayer cannot sustain this curvature stress, the lamellar
structure (L) will be broken and non-lamellar phases will arise. The basic structural
phase of biological membranes is nevertheless a lamellar phospholipid bilayer matrix,
and deviations from a lamellar arrangement are generally not desirable in the plasma
membrane. The inclusion of lipids with propensity to non-lamellar structure leads to a
8

INTRODUCTION: BIOMEMBRANES
curvature stress in the membrane, and local modulation of this stress can be achieved
either by enzymatic action in the lipids, as removal of headgroups (Nieva et al., 1993) or
a acyl-chain, or by protein binding, allowing local formation of non-lamellar phases.
Micellar
phase
Hexagonal
phase
Lamellar
phase
Cubic
phase
Inverted
hexagonal
phase
Figure I.4 - Representation of the most important forms of organization of lipid aggregates in an aqueous
environment. Different lipid structures correspond to different lipid curvatures and are arranged in
accordance to the value of the packing parameter P. Adapted from (Mouritsen, 2005) and (Seddon and
Templer, 1995).
The ability of a lipid to fit into a particular type of aggregate can be described by a
packing parameter P, which is defined in Figure I.4. A deviation of P from 1, either
positive or negative, corresponds to a deviation of the lipid shape when in the lipid
aggregate from a cylindrical shape, which is the shape that fits a planar bilayer structure.
Negative deviations (P < 1) correspond to hexagonal (H I ) structures and in the limit P <
1/3 to micelles. Highly phosphorylated PI can induce micelles as will be discussed in
chapter IV. Positive deviations (P > 1) correspond to cubic and inverted hexagonal (H II )
aggregates. These two structures are of significant biological relevance, playing a
9

functional role in process such as endo- and exocytosis, membrane recycling, protein
trafficking, fat digestion, membrane budding and fusion. PE lipids show propensity to
arrange into this structures (Mouritsen, 2005). The cubic phase, which consists of short
tubes connected in a hexagonal array, is sometimes found on mixtures of lipids that are
in transition from a lamellar phase to an inverted hexagonal phase. The inverted
hexagonal phase consists of hexagonally packed water cylinders with an outer lining of
lipid molecules oriented with their acyl-chains away from the aqueous cylinders
(Yeagle, 1993).
1.5. Lamellar phase transitions in lipid bilayers
Apart from transitions between different lipid morphologies, lipids also experience
phase transitions without drastic changes in morphology. In a lamellar symmetry, lipids
can experience several packing conditions that correspond to different lipid phases.
These different structures are classified as distinct phases because upon crossing the
transition temperature (or other thermodynamic variable), several physical properties of
the lipid bilayer change abruptly, including the heat capacity, lateral diffusion,
permeability, thickness, area, vesicle shape, etc.
In Figure I.5, differential scanning calorimetry (DSC) profiles of three
phospholipids are presented. In these, the specific heat of the phospholipids is shown as
a function of temperature. For dimyristoyl-sn-glycero-3-phosphatidylcholine (DMPC)
two peaks are clearly visible, and each peak corresponds to a phase transition. From
this, it is clear that DMPC can exist as three different phases between 10 and 30 ºC. All
other PC lipids present identical behaviour. For the PE and PA lipids the first transition
is absent, the main transition (T m ) is however generally common to all phospholipids.
This transition is considered to have first order characteristics (Gennis, 1989).
Below the main transition, lipids are tightly packed, the acyl chains are ordered and
extended in an all-trans chain configuration, while the molecules are arranged in a
regular lattice as in a crystalline solid (Figure I.5). This phase is for that reason called
solid-ordered or gel phase (s 0 ). In the gel phase, individual molecules diffuse slowly in
the plane of the membrane, and the lateral diffusion coefficient (D) of a phospholipid
molecule is on the order of 10 -10 cm 2 s -1 . Lipids with large headgroups as PC must also
tilt in the plane of the bilayer while in the gel state (Figure I.5). The cross-sectional area
of the headgroup of these lipids is larger than the cross-sectional area of the acyl-chains,
10

INTRODUCTION: BIOMEMBRANES
and the only way to achieve effective packing is through a tilt of the hydrocarbon
chains.
Above the T m of the lipid, the acyl chain ordering characteristic of the gel phase is
lost (gauche conformations are favoured), and in this new phase, the lateral diffusion of
the lipid molecules is substantially higher (D is on the order of 10 -8 cm 2 s -1 ). This phase
is called fluid or liquid-crystalline phase (L α ). As a result of the smaller number of trans
conformations in the acyl-chains, bilayers in the fluid phase are considerably thinner
than in the gel phase (up to 6 Å in dipalmitoyl-sn-glycero-3-phosphatidylcholine-DPPC
bilayers), but the cross-sectional area increases dramatically (from 52 Å to 71 Å in
DPPC individual molecules) (Nagle and Tristram-Nagle, 2000). This increase in area is
a consequence of the weaker van der Waals attractive interactions between acyl-chains
in the fluid phase (Gennis, 1989). As a rule, a strong coupling exists between the lipid
phases in each monolayer (Bagatolli and Gratton, 2000), exceptions however have
already been observed (Devaux and Morris, 2004).
The acyl-chains of a phospholipid dictate to a large extent the stability of each of the
gel and fluid phases, and as a result they define the T m of the lipid. This is due to the
importance of the van der Waals interactions in stabilizing the gel phase. Long acylchains
permit stronger van der Waals attractive forces, stabilizing the gel phase and
increasing the lipid T m . Unsaturated chains prevent effective ordered packing into the
gel state, inducing a decrease in lipid T m . As most lipids in biomembranes present
unsaturations, lipid membranes are highly fluid. Headgroup structure can also influence
the transition temperature of the lipid. PE lipids present higher T m than PC due to
stabilizing hydrogen bonding in the gel phase. pH, ionic strength and pressure can also
influence T m .
PC lipids, as seen in Figure I.5, experience an additional phase transition. This
transition is called pretransition and occurs between two different gel states, L β’ , and a
ripple or periodic gel phase (P β’ ), at a characteristic temperature (T P ). Due to the
presence of the planar ring structure, cholesterol disrupts the packing of lipids when
mixed with lipids in the gel phase. For lipids in the fluid phase, cholesterol has an
ordering influence (Ipsen et al., 1987). The mixture of cholesterol and saturated
phospholipids can give rise to an additional phase called liquid ordered phase (L o ). In
the L o phase, the acyl-chains present a higher degree of ordering as compared to the one
11

in the L α phase, while lateral diffusion in the plane of the bilayer and freedom of
rotation remains high.
Figure I.5 – A – Differential scanning calorimetry profile of DMPC, 1,2-dimyristoyl-sn-glycero-3-
phosphatidylethanolamine (DMPE), and 1,2-dimyristoyl-sn-glycero-3-phosphatidic acid (DMPA). B –
Depiction of most common lamellar lipid phases: the fluid L α phase, the gel L β and the ripple P β’ phase
(taken from Gennis, 1989).
1.6. Membrane thickness
Due to the complex chemical structure of lipids, the chemical environment along the
normal axis of the bilayer is highly stratified. In hydrated bilayers, fluctuations add
increased complexity to the task of the structural description of the lipid membrane. A
description for the position of the atoms inside the bilayer using a broad statistical
distribution function can be used. Figure I.6 is a representation of the liquidcrystallographic
structure of L α -phase dioleoyl-sn-glycero-3-phosphatidylcholine
(DOPC) bilayers along the axis of the bilayer normal.
In the presence of this dynamical concept for the lipid bilayer structure it becomes
difficult to define certain bilayer properties such as membrane thickness. It is reasonable
nevertheless to assume the bilayer thickness as equal to the distance between the peaks
12

INTRODUCTION: BIOMEMBRANES
of the distribution functions of phosphate groups in each monolayer. In this way, the
bilayer thickness of fluid DOPC is 37 Å. Another important concept is the hydrophobic
thickness of a bilayer, i.e. the length across the bilayer normal which is occupied by the
hydrocarbon core of lipids. Wiener and White (1992) showed that the transbilayer
carbonyl-to-carbonyl distance accurately reported the thickness of the hydrocarbon core.
For fluid DOPC, this value is of 27-28 Å (Lewis and Engelman, 1983; Nagle and
Tristram-Nagle, 2000). Interestingly, the presence of the double bond in DOPC, which
carries 18 carbons in the acyl-chains, decreases the hydrophobic thickness up to almost
the same value of DMPC, which carries only 14 carbons but presents no unsaturation.
As a general rule, the hydrophobic thickness of a phospholipid bilayer increases 1,75 Å
per additional carbon in the phospholipid acyl-chain (Lewis and Engelman, 1983).
Figure 1.6 - Crystallographic structure of fluid DOPC bilayers along the bilayer normal axis (Wiener and
White, 1992).
Apart from the phospholipid structure, other factors can influence significantly the
thickness of a bilayer. Increases in temperature and hydration lead to decreases in the
thickness of the bilayer. Cholesterol also has an important impact as it acts by
thickening the membrane (Mouritsen, 2005).
13

1.7. Lateral heterogeneity in lipid bilayer
As already pointed out, the lipid composition of biomembranes is highly complex.
In the case of multicomponent lipid bilayers, no longer a single T m applies to the whole
population of lipid molecules. Now the transition takes place over a range of
temperatures (between the T m of the lipid with the lower transition temperature and the
T m of the lipid with the higher transition temperature), in which the lipids phase separate
in gel and fluid phases (Mouritsen, 2005). This phase separation is a direct consequence
of the highly cooperative character of lipid phase. Immiscibility between lipids can
result from differences in acyl-chain length, unsaturation or headgroup composition
(Shimshick and McConnell,1973; Arnold et al., 1981; Lentz et al., 1976).
Phase diagrams provide an efficient tool for the description of phase behaviour of
multicomponent lipid mixtures. In the phase diagrams, the lines correspond to the
temperatures at which a transition starts or is finished. Three phase diagrams for three
different binary lipid mixtures are presented in Figure I.7. It is clear from Figure I.7 that
as the divergence in acyl-chain length increases, the size of the region of gel-fluid phase
coexistence also increases due to a more accentuated immiscibility between the two
lipids.
Figure I.7 – Phase diagrams of three different binary mixtures of saturated PC lipids presenting different
acyl-chain lengths – dilauroyl-sn-glycerol-3-phosphatidylcholine-DLPC (12 carbons); DMPC (14
carbons); DPPC (16 carbons) and distearoyl-sn-glycerol-3- phosphatidylcholine -DSPC (18 carbons). The
fluid phase is represented by an f and the gel phase by g (taken from Mouritsen, 2005).
14

INTRODUCTION: BIOMEMBRANES
Lipid immiscibility gives rise to lateral structuring of the lipid bilayer and the
resulting lateral organizations are called domains. Lipid domains are not static
structures. They exhibit dynamics both in space and time, as they can exist as short
(fluctuations) or long-lived structures in the lipid bilayer. Large size domains (in the
order of micrometers) have already been detected through imaging techniques (Korlach
et al., 1999; Bagatolli and Gratton, 2000). These techniques make use of fluorescent
probes that show preferential partition in one of the lipid phases or differential
fluorescent properties in each phase (Klausner and Wolf, 1980; Bagatolli and Gratton,
2000). They are however restricted by the diffraction limit to detect domains larger than
~ 200 nm. An example is shown in Figure I.8.
Figure I.8 - Gel-Fluid coexistence in a giant unilamellar vesicle (GUV) (see 1.10) detected by
confocal microscopy. The system is a DLPC/DSPC (0.4/0.6 mol/mol) mixture. Red areas correspond to
the fluorescence arising from a probe with preferential partition to the gel phase (enriched in DSPC) and
green areas correspond to the fluorescence of a probe with preferential partition to the fluid phase
(enriched in DLPC) (taken from Korlach et al., 1999).
For a situation of thermodynamical equilibrium, one should expect complete phase
separation, i.e., only two very large domains inside the vesicle, one in the gel phase, and
the other in the fluid phase. This is due to the interfacial tension between lipid domains.
In the interface, the phases show differences in hydrophobic thickness, and the
consequent exposure of the hydrocarbon chains to water molecules create a packing
stress in the interface of domains. This tension is the driving force for the fusion of
small domains into larger lateral structures after phase separation, as larger domains
correspond to a smaller interface area. At infinite time, the complete macroscopic phase
separation should be achieved. However, several domains of limited size are observed
(Figure I.8). This is possibly due to the coupling of phase separation to the curvature of
15

the vesicle through bilayer deformation in the interface and to the insertion of
phospholipids in the interface region with intermediate conformations (between gel and
fluid like conformations) in order to substantially decrease the interface stress between
lipid domains (Jorgensen and Mouritsen, 1995). Another possibility is that the finite
size domains correspond to non-equilibrium structures, detected due to the very long
lifetime of the process of phase separation (de Almeida et al., 2002).
Even when at the same lamellar phase, lipids do not mix ideally. This is particularly
true for the gel phase, as the strict packing restrictions force dissimilar lipids to
immiscibility. As a result, clustering or domain formation is observed and the bilayer
gains lateral heterogeneity. This type of behaviour is observed e.g. for the highly
nonideal mixture DLPC-DSPC (Fig. I.7), as two different gel phases (each enriched in
one of the lipids) are present below the phase coexistence region.
Deviations from ideal lipid mixing and domain formation also occur in the fluid
phase, and PC lipids with a significantly different thickness were already shown to
segregate into domains due to hydrophobic mismatch stress. Also for this situation, the
discrepancy of hydrophobic thickness of the two PC lipids can create a packing stress in
the bilayer due to exposure of hydrocarbons to water, and stimulate lipid segregation
(Lehtonen et al., 1996). The lateral structures created by this type of phase separation
are expected to be on the nanometer scale in opposition to gel-fluid phase separation
which can be detected in the micrometer scale. Therefore, these domains are elusive to
most of the imaging techniques, and other spectroscopic techniques must be applied,
namely macroscopic fluorescence methodologies.
The ideality degree of a lipid mixture dictates not only the phase separation of the
mixture but also the size and shape of the phase-separated domains. Apart from
temperature, other factors influencing the lateral structuring of a multicomponent lipid
bilayer are dehydration and the presence of divalent ions. Dehydration can cause
lamellar to inverted hexagonal phase transition and induce phase separation (Webb et
al., 1993). Mixtures containing an anionic phospholipid can experience severe phase
separation in the presence of divalent ions like Ca 2+ . Calcium ions induce clustering and
phase separation of anionic phospholipids due to intermolecular cross bridges (Silvius,
1990).
A phase coexistence of particular biological relevance is that of L α and L o phases. In
1997 Simons and Ikonen proposed that small lipid domains in the L o phase (lipid rafts)
composed of saturated lipids and cholesterol coexisted with a matrix of lipids in the
16

INTRODUCTION: BIOMEMBRANES
fluid phase. Since then, several cellular mechanisms have been associated with domains
of cholesterol and saturated lipids, and several evidences for the existence of these
structures have been gathered. There is indication that rafts are involved in cell surface
signalling, intracellular trafficking and sorting of lipids and proteins (Mouritsen, 2005)
Lateral structuring of the lipid bilayer, either through L α -L β , L α -L o or fluid-fluid
domains, implies that membrane functions do not require the dependence on random
collisions for interactions between reagents (Mouritsen, 2005). It provides a structuring
principle for lipid bilayer organization, that can permit not only a higher efficiency of
several membrane processes by compartmentalization, but also a mechanism of
modulating these processes through control of the membrane lateral structure.
1.8. Membrane proteins
The lipid bilayer provides the basic architecture of the biomembranes. However,
membrane proteins are responsible for the majority of cell functions taking place in the
membrane. The distinctive properties and functions of different lipid membranes inside
the cell are mainly dictated by their protein content. The concentration of proteins inside
biomembranes is drastically different among the several cellular membranes. In the
mitochondrial membrane the protein content reaches about 76% (w/w), while in the
myelin membrane proteins are only 18% of total membrane content. The mapping of
the yeast genome showed that most of the genome code is expected to code for
membrane proteins, in demonstration of their key role for life (Mouritsen, 2005).
Membrane proteins are defined by their degree of insertion in the lipid bilayer
(Figure I.9). Thus, membrane proteins deeply buried in the lipid bilayer are called
integral, frequently spanning the membrane one or several times, whereas proteins
bound to the exoplasmic or cytoplasmic periphery of the lipid bilayer are entitled
peripheral. Peripheral membrane proteins are bound to lipid bilayers by interaction
either with lipid headgroups or with other membrane bound proteins. Another class of
proteins are bound to the membrane through covalent linkage to a lipid molecule. These
membrane proteins are classified as lipid-anchored.
17

Figure I.9 – Depiction of the several possibilities for association of proteins with biomembranes. Protein
domains in the exoplasmic side of the membrane are frequently glycosilated and can interact with the
extracellular matrix. Protein domains on the cytoplasmic side on the other hand are responsible for the
coupling of the cytoskeleton network with the lipid membrane (From Lodish et al., 2000).
Membrane proteins on the extracellular side of the plasma membrane generally bind
other molecules, including external signalling proteins, ions, small metabolites, and
adhesion molecules in other cells. Protein domains within the plasma membrane, mainly
those that are part of channels and pores, are generally responsible for transport of
molecules across the bilayer, while peripheral protein membranes in the cytosolic side
of the plasma membrane are responsible for many different functions (Lodish et al.,
2000).
The possibilities of secondary structure available for membrane spanning protein
domains are very limited. Only two structures are available, the alpha-helix and the
beta-barrel (Figure I.10). These structures have a common characteristic. In both, the
peptide bonds of the aminoacids are hydrogen bonded. In fact, if this were not the case,
insertion in the highly apolar environment would be energetically unfavourable, even
for the most hydrophobic amino acids. The energetic cost of partitioning a peptide bond
into a highly apolar phase, is significantly larger than the free energy reduction
associated with insertion of the Trp side chain in the same environment (White and
Wimley, 1999). Therefore, only a polypeptide segment with complete backbonebackbone
hydrogen bonding can insert in the highly hydrophobic core of the bilayer,
and the alpha-helix and the beta-barrel are the two structures that satisfy this condition.
18

INTRODUCTION: BIOMEMBRANES
A
B
Figure I.10 – Two secondary structures available for membrane spanning protein domains, the alphahelix
(A) and the beta-barrel (B).
The amino acids in an alpha-helix are arranged in a right-handed helical structure.
The dimensions of the alpha helix are 5,4 Å wide (if accounting for aminoacid side
chains the alpha helix is about 10 Å wide). Each aminoacid correspond to a 100º turn,
and consequently for each turn, 3.6 aminoacids are required. Each aminoacid also
corresponds to a translation across the helical axis of 1,5 Å, and each amine group of the
peptide bound forms a hydrogen bond with the carbonyl group of the pepide bound four
aminoacids earlier. Beta barrels are large beta sheet structures for which the last strand
is hydrogen bonded with the first, creating in this way a closed structure. These
structures are commonly found in porins (see chapter IV). The beta-sheet structure is
based on the side-by-side hydrogen bonded arrangement of stretchs of amino acids
linearly extended.
Hydrogen bonding between peptide bonds is essential for formation of TM protein
segments, however it is not sufficient. The sum of the free energy reduction from
insertion of the side chains of the polypeptide must surpass the energetic cost of burying
the hydrogen bonded peptide bonds. This is only possible in the presence of highly
hydrophobic aminoacids, and in the absence of to many hydrophilic ones, particularly
charged residues. Octanol is frequently regarded as a suitable model for studying the
insertion of molecules in a hydrophobic medium such as the hydrocarbon core of the
bilayer. Figure I.11 reproduces the results obtained for the free energy of transfer of
whole aminoacids (including the peptide bond), from water to the octanol phase, and to
the interface region (polar environment) of POPC bilayers.
19

Figure I.11 – Experimentally obtained hydrophobicity scales for whole residues (including the peptide
bond). Results obtained for the transfer from water to the polar interface of POPC bilayers and from water
to octanol are presented (Wimley and White, 1996; Wimley et al., 1996).
It is clear from Figure 1.11 that residues like Trp (W), Phe (F), Tyr (Y), Leu (L), Ile
(I), Met (M) or Val (V) favour incorporation in hydrocarbon environments, while
charged residues are particularly averse to this. From this information and from the
primary sequence of a protein it is possible to make previsions for possible alpha helical
TM segments inside the protein sequence. A calculation is made through summation of
the free energy required to insert (from water into the bilayer hydrocarbon core)
successive segments of the polypeptide chain. Segments tested have a finite size, around
20 aminoacids, as this is just about sufficient to spam the hydrocarbon core of a typical
bilayer (this aminoacid length corresponds to 30 Å if the helix is oriented along the
bilayer normal). This procedure allows the creation of hydropathy plots that signal the
polypeptide segments which are the most likely candidates for a TM configuration
inside the protein.
Aminoacids also exhibit preferential localization in certain positions along the
bilayer normal. This fact is of utmost importance to the final position of an alpha-helix
in membranes and will be discussed in more detail in section 2 of this chapter.
1.9. Lateral dynamics in biomembranes
The Singer-Nicolson fluid mosaic model for biomembranes (1972) (Figure I.12)
described the lipid membrane as a two dimensional liquid where both lipids and
membrane proteins moved freely. In fact, in pure lipid bilayers in the fluid phase, lipids
experience fast lateral diffusion, as well as rotational diffusion, wobbling, and
movements in the axis of the bilayer. Crossing from one monolayer to the other is
20

INTRODUCTION: BIOMEMBRANES
however much more restricted as discussed in 1.3. Lateral dynamics in the gel phase is
approximately 100 times slower (Mouritsen, 2005). The speed of integral membrane
proteins is typically 100 times smaller than the speed of lipid molecules.
Figure I.12 – The Singer-Nicolson model of fluid mosaic (Taken from Singer and Nicolson, 1972)
However, in biomembranes, diffusion of lipids and proteins is hindered by the
presence of compartmentalisations of the membrane in the form of domains.
Additionally a fraction of the membrane proteins is completely immobile due to
interaction with the cytoskeleton network. Generally the average diffusion coefficient of
a protein in the plasma membrane of intact cells is about 10-30 times smaller than the
one observed in pure lipid liposomes (Mouritsen, 2005). The Singer-Nicolson model is
therefore a simplistic description of the complex interactions and dynamics existent in
biomembranes.
1.10. Membrane Model Systems
Cellular membranes are often too intricate for their properties to be fully resolved as
too many variables are present. A valuable alternative is the use of membrane model
systems. Model systems grant us a simplified basis for the study of bilayer structure and
properties, while at the same time maintaining all the fundamental properties of the
biomembrane.
Liposomes are the most popular of the membrane model systems. Preparation is
often very simple, requiring solubilization of lipids in organic solvents, followed by
21

drying into a film and ressuspension in an aqueous environment. Suspensions of
liposomes prepared in this manner are composed by concentric bilayers or multilamellar
vesicles (MLV’s). Due to this arrangement of the lipid vesicles, most of the lipid bilayer
surface is buried inside the liposome, as only about 10 % is found in the outside surface
(Yeagle, 1993). The application of MLV´s in photophysical studies can result in several
difficulties. Due to the size of the MLV’s particles, scattering is very significant and for
these reason, MLV´s are seldom used.
There are different strategies for obtaining unilamellar liposomes. By applying
ultrasonic power to suspensions of liposomes, unilamellar vesicles of small diameter
(~30 Å), also called small unilamellar vesicle (SUV) are obtained. Due to the smaller
size of the particles, scattering in suspensions of SUV’s is drastically reduced, however
the curvature of SUV’s is much higher than the curvature of cell membranes, resulting
in poor mimics of the properties of biomembranes.
Larger unilamellar vesicles (LUV’s) can be produced either by dialysis of
detergents, reverse phase evaporation, or fast extrusion through polycarbonate filters.
The latter method is particularly useful due to the shorter times required when compared
to the other ones, e.g., dialysis of detergents can take several days for efficient removal
of detergent from liposomes (for a review, see Gennis, 1989).
Much larger unilamellar vesicles (up to 300 µm) can also be prepared by gentle
hydration and electroformation (Rodriguez et al., 2005). These vesicles are also known
as giant unilamellar vesicles or GUV’s and are particularly suitable for microscopy
applications as their size enables visualization and micromanipulation.
Still, other types of membrane model systems exist. Lipid monolayers in water-air
interface allow the study of the effect of the lateral surface pressure on membrane
components and on the interactions between them. Black Lipid Membranes have also
proven valuable in the study of electrical properties (Yeagle, 1993).
22

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
2.LIPID-PROTEIN INTERACTIONS
2.1. Membrane protein reconstitution
Even for the simplest organisms, membrane proteins in protein-lipid extracts are
present in a very complex and heterogeneous environment, which can impair the
possibility of conducting biophysical studies that require a simpler matrix for the
protein (with absence of possible contaminants), and more controlled conditions.
However, integral membrane proteins generally cannot be studied in homogeneous
systems such as aqueous or organic solutions due to the complex solubility problems
derived from their bitopic nature. In addition, the study of their properties in a isotropic
medium, would not be biologically relevant. Systems are required that satisfy both the
hydrophobicity of the membrane embedded sections of the protein as well as their
hydrophilic domains. Liposomes are frequently used for such studies, since, as
discussed in the previous section, they present good mimetic conditions of
biomembranes. Liposomes containing reconstituted proteins are called proteoliposomes.
In order to reconstitute membrane proteins in liposomes it is first necessary to purify
and solubilize them. Neither of these procedures is trivial and detergents have proven
invaluable tools in both. They meet the requirements of amphipatic structure necessary
to solubilize the two different environments present in membrane proteins and these are
frequently soluble in micellar structures. Detergents are classified according to the
charge of the headgroup as ionic (cationic or anionic), non-ionic or zwitterionic.
Examples of ionic detergents often used in membrane solubilization are sodium
dodecyl sulphate (SDS) and bile acid salts. SDS is a linear chain detergent and a
extremely powerful solubilizing agent for membrane proteins, however, it is also
frequently denaturing. Protein renaturation is eventually possible under certain
conditions after transfer to other medium (Dong et al., 1997). Bile acid salts are ionic
detergents with steroidal groups for backbones. Due to the planar characteristics of the
steroidal structure, instead of a proper headgroup they present a polar and an apolar
face. These detergents are much weaker than SDS and more efficient in maintaining
protein activity. Examples of bile acid salts are sodium cholate and sodium
deoxycholate. Non-ionic detergents are also mild and relatively non-denaturating/non-
23

deactivating agents. Zwitterionic detergents combine the properties of ionic and nonionic
detergents (Seddon et al., 2004).
Several methods exist for insertion of integral membrane proteins in liposomes. The
most popular makes use of mediation by detergents. The choice of the best detergent
varies greatly between proteins, confirming the complexity of the interactions involved
in the reconstitution process. The simplest of the detergent-mediated reconstitution
method is a dilution approach. Liposomes are added to detergent micelles with protein
and the detergent is removed. Removal of detergent evolves until the concentration of
the detergent is below the critical micelle concentration (cmc) and micelles become
unstable inducing transfer of the protein to the liposomes. Removal of detergent can be
achieved by: i) dialysis, making use of a membrane with a pore size smaller than the
size of the proteoliposomes; ii) column chromatography; iii) incubation with detergent
adsorbing beads. However, pure liposomes are frequently impermeable to proteins and
often a destabilization of the bilayer is required for correct insertion of the protein
(Rigaud et al., 1995). Destabilization of the bilayer can be achieved by using mixtures
of lipids and detergents instead of pure liposomes. The resulting structures are more
receptive to protein insertion, and after this step the detergent can be removed by one of
the methods described above.
The required degree of detergent-destabilization of the liposomes depends on the
nature of the detergents. Several proteins were reconstituted through mediation of nonionic
detergents by destabilization of liposomes with the concentration of the detergent
set at saturation levels for the liposomes. This method frequently leads to asymmetrical
orientation of the proteins in the liposomes (a large fraction of the proteins face the
same direction after reconstitution) (Knol et al, 1998). On the other hand, reconstitution
of some proteins with sodium cholate required a concentration of detergent above the
saturation point for liposomes, i.e proteins in detergent micelles had to be mixed to
detergent micelles containing lipids (Seddon et al., 2004). This latter method usually
results in symmetrical orientations for the protein inside the proteoliposomes (Yeagle,
1993). In the case of bile acid salts, dialysis is a good detergent removal procedure.
These detergents present high cmc (high solubility as monomers) so dialysis can be
efficient at reasonable time scales.
Other reconstitution strategies include protein/peptide solubilization in organic
solvents. This procedure must be applied with caution as it often results in aggregation
and denaturation. Some protocols for proteins have been developed that make use of the
24

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
ability of some organic solvents to stimulate hydrogen bonding and alpha-helical
structure (Killian et al., 1994). This is particularly important when working with TM
alpha-helical peptides, since this treatment can prevent aggregation and/or transition to
different secondary structures that can be irreversible in some cases. Examples of such
solvents are trifluoroethanol (TFE) and hexafluoroisopropanol.
Sonication of mixed solutions of proteins in buffer and lipids can also affect the
transfer of proteins to liposomes. The downfall of this technique is that it requires
solubility of the protein in water, it can induce protein denaturation and always results
in SUV’s, which can be problematic since some proteins reconstituted in high curvature
liposomes have been shown to underwent changes in their activities. Freeze-thawing of
mixtures of sonicated liposomes and proteins has also been used in cases where the
protein was sensitive to both detergents and sonication (Seddon et al., 2004).
For a protein to be considered as correctly reconstituted, some requirements must be
fulfilled. The characteristic function or activity of the protein in vivo must be regained
(at least partially) after reconstitution. Ideally the process should result in a defined lipid
environment for the protein with an also defined lipid to protein ratio (Yeagle, 1993).
2.2. Peptides as models
Membrane proteins are extremely complex entities. In addition to protein-lipid
interactions, protein-protein interactions are crucial in dictating final properties as
protein domains interact between themselves in the final structure. In order to study
protein-lipid interactions an alternative is the use of peptide models, either TM or
peripherically associated with the membrane. An additional advantage of this
minimalist approach is that synthetic peptides can be obtained in large quantities, which
is not feasible for many membrane proteins. The possibility of synthesis also allows an
easy control over the primary sequence and wide mutation possibilities, and this is
essential in understating the role of particular amino acids (Wimley and White, 1996)
and peptide segments on the interactions with the lipid environment, and on the function
of the protein itself. For large membrane proteins the range of possible mutations is
quite limited due to problems related to cell viability of mutants.
Putative TM alpha-helices can be estimated from hydropathy plots as described in
Section 1.8, and hydrophobic moment estimations can assist on the search for
25

amphipatic helices (see Section 2.3). Predictions derived from these methods allow the
design of synthetic peptides for detailed biophysical studies.
A number of experiments showed that synthetic peptides corresponding to the
alpha-helical TM segments of some proteins can assemble and substitute the native
proteins as well as their functions after independent folding, validating the approach of
studying isolated peptide segments (Marsh, 1996). Popot and Engelman (1990; 1993)
rationalized these results with the proposal of a two-stage model for membrane insertion
of the TM segments from a intrinsic membrane protein. In the first stage, alpha-helices
are independently folded and inserted in the lipid bilayer, while in the second stage,
alpha-helices interact and assemble in the final protein structure.
2.3. Amphipatic helix
Amphipatic helices are protein alpha-helices for which one face along the helical
axis is hydrophilic whereas the opposite face is hydrophobic. The amphipatic helix is a
common feature of biologically active polypeptides and proteins, such as hormones,
antibiotics, and venoms. In this way, this structure can play a large number of roles. A
function of great structural importance is the shielding of the hydrophobic interior of
proteins exposed to aqueous environments, while for membrane proteins the roles
played by amphipatic helices are diverse. In the case of peripheric proteins they can act
as anchors to the membrane by tight binding to the hydrocarbon/water interface of the
bilayer as the structure of this boundary is highly complementary to the amphipatic
helix. The amphipatic structure is also able to promote protein-protein interactions
between hydrophobic segments of helices. In some cases this allows amphipatic
peptides to create channels and pores in the membrane via aggregation into oligomeric
bundles for which the hydrophobic side faces the hydrocarbon chains of the lipids and
the hydrophilic faces of the helices are exposed to the inside. Thus delineating a pore in
the bilayer (Epand, 1993). Recent studies also suggest a role of amphipatic helices in
the modulation of membrane curvature (see chapter V).
Overall, the amphiphilic character is a much more important dictator of peptide
conformation than the specific characteristic of its constituent amino acids (Taylor,
1990). Parameters like the ratio of hydrophobic to hydrophilic amino acids as well as
26

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
the angle of the hydrophobic face are crucially important in dictating the location of the
peptide in the membrane and the type of lipid interactions established (Kitamura et al.,
1999; Brasseur, 1991).
Methods exist to predict possible amphipatic helices from primary sequences of
proteins. From the knowledge of the periodicity of alpha-helices, Shiffer-Edmundson’s
(1967) helical wheels and helical net representations (Dunnil, 1968) can be drawn like
the ones in Figure I.13. In Shiffer-Edmundson’s representation, residues are sketched
around a circle with each residue being placed 100º from the preceding one. If the
segment corresponds to an amphipatic helix, then one face of the circle should be
enriched in hydrophilic residues while the opposite face should present clustering of
hydrophobic amino acids. The net helical representation illustrates a hollow cylinder cut
open along a parallel line to its axis and laid flat. The residues are placed in such a way
so that if the cylinder would be closed, the amino acids are positioned as in a alphahelix
(Auger, 1993). These methods are very easy to apply and provide great tools to
visualize the disposition of the amino acids in a putative amphipatic helix but are not
quantitative, i.e. they do not provide a quantification of the amphipatic character of the
protein domain being evaluated.
A
B
Figure I.13 - A- Shiffer-Edmundson’s helical wheel representation for amphipatic domains. B-
Helical net representation for amphipatic domains (Taken from Anantharamaiah et al., 1993).
Some of the available quantitative methods work by comparing a sequence of
hydrophobicities (of each amino acid) with a sinusoid, reporting the evolution of the
amino acid positions in the alpha-helical polypeptide chain (in the axis perpendicular to
the helical axis). The most common quantitative description of the amphipatic degree of
a protein domain is given by the hydrophobic moment (Eisenberg et al., 1982). In this
analysis, the hydrophobic moment of a helix is equal to the summation of the vectors of
27

size corresponding to the hydrophobicity of each amino acid drawn from the centre of
the putative helix. Through calculation of the hydrophobic moment of an appropriate
residue window (typically 11 – three turns in the alpha-helix) for each position in the
protein sequence, it is possible to draw an “amphipatic profile” of the protein (Fig.
I.14), and by comparison with values obtained from known amphipatic helices it is
possible to predict which protein domains are likely to correspond to amphipatic helices
(Auger, 1993).
Figure I.14 – Amphipatic profile of gp41 from envelope protein of HIV variant bh10 (taken from
Auger, 1993).
2.4. Peptide partitioning to the membrane
Several approaches have been used to study peptide binding to lipid membranes.
The most important are fluorescence spectroscopy (Santos et al., 2003), circular
dichroism, surface plasmon resonance and isothermal titration calorimetry. The binding
of a peptide to the lipid bilayer is not a conventional reaction with a defined
stoichiometry. In reality, the process is better described as a water/bilayer partition
problem. Although stable complex formation between proteins and specific lipids have
been proved and bound lipid molecules have been observed in the crystal structures of
some proteins (see Section 2.8) (Lee, 2003), this type of interaction is not generally
found for most membrane proteins.
28

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
Partitioning of TM peptides to lipid bilayers is often a very difficult problem to
study. A TM configuration for a peptide implies a large number of strongly
hydrophobic residues in the sequence, as the size of the hydrophobic segment must be
large enough to span the hydrocarbon (apolar) region of the bilayer (see Section 2.8). As
a result their solubilities in water are frequently very low or null, and the partition
equilibrium of these peptides is completely shifted to the lipid phase with almost no
monomeric peptide in the aqueous environment (the available strategy to shield
hydrophobic residues is non-specific aggregation). For these reasons the study of
lipid/water partition of TM peptides is frequently inaccessible apart from the
considerations in Section 1.8. As for peripheral peptides, i.e. peptides that bind to the
surface of lipid bilayers, the hydrophobicity requirements are not so severe and
solubility in aqueous environments is often observed.
The interaction of the peptide with the membrane is mainly dictated by the
following effects:
1) Electrostatic contributions are of key importance when basic residues are present in
the peptide and the lipid surface is rich in negatively charged lipids. However, if the
lipids are zwitterionic, the electrostatic influence can be negligible, and if both lipids
and peptides present negative charges, the electrostatic contribution becomes negative
(repulsion). In case of effective partition due to electrostatic attraction, this contribution
is no longer independent of peptide concentration at very small lipid to peptide ratios
(L/P). In this range of concentrations, the potential of the bilayer surface becomes
dependent on the peptide concentration as a consequence of screening of lipid negative
charges by the peptide (Seelig, 2004).
2) Classical hydrophobic contributions are entropically driven, and result from the
energetic requirements for insertion of apolar residues in the aqueous environment
(rotational restriction of water molecules in the hydration shell). They are also
dependent on the final positioning of the peptide in the lipid bilayer (bilayer surface or
core).
3) Perturbation of the lipid structure and packing.
4) Hydrophobic matching of TM peptides (see Section 2.6).
5) Hydrogen bonding properties of the lipid bilayer interface are highly complex, since
both hydrogen bond donors and acceptors are present. As an example, PC bilayers
establish stronger hydrogen bonds with phenolic compounds than PE bilayers due to the
presence of the choline group. Additionally, the dielectric constant of the lipid bilayer
29

changes abruptly in the interfacial region, varying from the value in water (ε = 78) to
the value in the hydrocarbon core (ε = 2) (McIntosh, 2002).
5) After adsorption of the peptide to the bilayer surface, conformational changes
generally take place that entail changes in the thermodynamic properties of the binding
process. One common scenario is that peptides are unstructured (random coil) while in
the aqueous environment and after binding to the bilayer take on an amphipatic
structure. The bound structure can also be dependent on the L/P in the membrane
(transition from an alpha-helical to beta-sheet structure). The change to an amphipatic
structure leads to multiple intermolecular hydrogen bonds and results in a largely
favourable contribution to partition. Isothermal titration calorimetry studies with some
amphipatic peptides revealed that the favourable enthalpic contributions for peptide
binding arising from alpha-helix formation amounted to half of the free energy of
binding (McIntosh, 2002).
The thermodynamic parameter that relates directly to the partition of the peptide
between the lipid and water phases is the partition coefficient (K p ):
K =
p
n
n
S,L
S,W
/ V
/ V
L
W
2.1
where V i are the volumes of the phases, and n S,i are the moles of solute in each phase (i
=W, aqueous phase; i=L, lipid phase) (Mateo et al., 2006). The partition coefficient is
converted in free energy of binding to the membrane (∆G 0 ) by:
o
∆G = -RT ln
( Kp)
2.2
After binding, amphipatic peptides are expect to reside in the lipid/water interface of
the membrane. The exact position is nevertheless dependent on both the membrane and
the peptide sequence. From X-ray diffraction measurements on DOPC bilayers, the axis
of an ideally amphipatic peptide was located between the mean positions of the glycerol
and carbonyl groups (Fig. I.6), exactly at the interface between the polar headgroups
and the hydrocarbon core (Mishra et al., 1994).
30

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
2.5. Anchoring of transmembrane domains
TM proteins are enriched in hydrophobic amino acids such as Leu, Ile, Val, Trp,
Tyr, and Phe. However, the occurrence of the aromatic amino acids Trp and Tyr are
strongly dependent on the position inside the TM segment. These aromatic amino acids
are particularly enriched at the end of TM sequences, close to the headgroup region of
the lipid bilayer (Wallin et al., 1997). Apparently, these aromatic residues exhibit strong
preferential interactions for the complex electrostatic carbonyl-glycerol environment of
the bilayer (see Fig. I.6). In this region the aromatic rings are stabilized, and,
simultaneously, penetration in the hydrocarbon core of the bilayer is somewhat
unfavourable due to exclusion of the flat aromatic rings (Yau et al., 1988, Braun and
von Heijne, 1999). Interestingly, another aromatic residue, Phe, does not exhibit
significant preferential localization in the membrane environment (Braun and von
Heijne, 1999).
Another important characteristic of TM domains is that they are generally flanked
by polar regions, rich in charged residues. Positively charged amino acids are found
close to the hydrophobic domain of the protein and in some cases they can even be
deeply inserted in this region (Caputo and London, 2003). The energetic cost of burying
of a charge in the hydrophobic environment of the lipid bilayer is very large, and in that
situation the long charged side chains of these residues protrude towards the headgroup
region in a phenomenon called “snorkling”. For acidic residues (Asp and Glu), the
absence of long side chains present a greater barrier for insertion in the hydrophobic
core of the bilayer. This is however observed for TM sequences with uninterrupted
hydrophobic domains of at least 12 residues (Caputo and London, 2004), and in these
conditions it was already proposed that the buried acidic residues can become
protonated (Monné et al., 1998).
2.6. Lipid-protein hydrophobic matching
Hydrophobic matching is not solely relevant for lipid-lipid interactions (Section
1.7), and the problem exists also for protein-lipid interactions. In this case, the
hydrophobic length of the TM domain of the protein should match to the hydrophobic
31

thickness of the bilayer (Section 1.8). This principle was first introduced by Bloom and
Mouritsen in 1984.
Lipid-protein hydrophobic mismatching can have two origins (Figure I.15). For
positive hydrophobic mismatch, the protein hydrophobic thickness (d P ) is longer than
the hydrophobic thickness of adjacent lipid molecules (d L ) and exposure of hydrophobic
residues of the protein to the aqueous environment results in an energetically
unfavourable term to the protein-lipid interaction. In the case of opposite situation, i.e.
negative mismatch, when lipid chains are longer than the hydrophobic domain of the
protein, the energetically unfavourable term to protein-lipid interaction originates from
the exposure of the hydrocarbon chains to water.
Both situations of hydrophobic mismatch create an energetic stress in the proteinlipid
interface that must elicit a response from the system. Most models for protein-lipid
mismatch assume a response in the form of stretching or compression of lipid acylchains
(Mall et al., 2002).
Figure I.15 – Schematic representation of positive and negative hydrophobic mismatch in proteinlipid
interfaces (From Dumas et al. 1999).
Several theoretical models have been developed to estimate the energetic penalties
in protein-lipid hydrophobic mismatch. Fattal and Ben-Shaul (1993) considered that the
total free energy of the protein-lipid interaction was a summation of the contributions
from the chain and headgroup terms. The presence of a rigid protein body (as it was
assumed in the model) in contact with lipid acyl-chains, implied an energetic penalty for
the interaction. The stretching or compression of lipid-chains to fit the hydrophobic
thickness of the membrane protein also translated into a positive contribution to the
free-energy of interaction. The contributions from interactions in the headgroup region
included an attractive term related to the exposure of the hydrocarbon core to the
aqueous environment as well as a repulsive term resulting from electrostatic and
32

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
excluded-volume interactions between headgroups. The resulting free-energy profile for
lipid-protein mismatch was nearly symmetrical about the point of hydrophobic
matching, and the lipid perturbation energy F (in units of kT per Å 2 of protein
circumference) is described by (Ben-Shaul, 1995):
F
( ) 2
= 0,37
+ 0,005 ⋅ d P
− d L
2.3
This model assumed concentration of the hydrophobic mismatch perturbation on the
first shell of lipids around the protein (see 2.7). For a hydrophobic mismatch of 7 Å and
assuming that each lipid adjacent to the protein should occupy at least 4 Å 2 of the
protein perimeter, the free-energy for lipid-protein interaction would increase by 3.6 kJ
mol -1 . This change in lipid-protein interaction energy corresponds to a decrease factor in
lipid-protein binding constant of 4.3. If propagation of perturbation spreaded to the
more external shells, the effects of hydrophobic mismatch on the adjacent lipid
molecules would be decreased. For example, if effects were averaged over three shells
of lipids, the change in binding constant would decrease by a factor of 1.6 (Mall et al.,
2002). Alternative models for treating hydrophobic mismatch came to very similar
conclusions regarding interaction energies (Nielsen et al., 1998; Mouritsen and Bloom,
1993).
Energies of this magnitude are sufficient to induce changes not only in the
interacting lipid molecules but in the protein itself. Experimental studies allowed to
conclude that β-barrel proteins are not easily susceptible to structural/conformational
changes due to hydrophobic mismatch and the energetic penalty for protein-lipid
interaction was only affecting lipid structure (O´Keeffe et al, 2000). However, in the
cases of alpha-helical TM proteins such as the potassium channel KcsA, and the
mechanosensitive channel MscL, the experimentally obtained dependence of binding
constants on the extent of hydrophobic mismatch was smaller than expected
(Williamson et al., 2002; Powl et al., 2003). This was rationalized as due to the smaller
rigidity of the α-helical proteins that makes them susceptible to distortion when
hydrophobic mismatch was present and therefore create an additional degree of freedom
for relaxation of lipid-protein hydrophobic mismatch. Distortion of α-helical membrane
proteins by hydrophobic mismatch is supported by the observation that the activities of
33

some α-helical membrane proteins are dependent on the membrane thickness (East and
Lee, 1982).
Most likely, the most common distortion available for α-helical membrane proteins
as a response to positive hydrophobic mismatch is a change of tilt of the TM helices
relative to the bilayer normal. This mechanism allows for an easy modulation of the
length along the bilayer normal occupied by the hydrophobic domains of the protein.
Indeed, tilting has been observed and measured for various TM proteins (Harzer and
Bechinger, 2000).
Rotation of amino acid residues at the end of TM α-helices was also proposed as
another distortion mechanism available for membrane proteins as this results in changes
in the effective hydrophobic thickness of the protein (Lee, 2003).
However, the ability of proteins and lipids to adapt to situations of hydrophobic
mismatch is limited, and drastic changes in the membrane distribution of these elements
is possible. In cases of severe hydrophobic mismatch, the TM orientation of TM helices
might not be maintained anymore and transitions to non-TM orientations are possible
(Ren et al., 1997; Ren et al., 1999) as well as decrease of partition to bilayers (Figure
I.16). Aggregation of TM domains is another possibility in hydrophobic mismatch as
this reduces the lipid-protein interface, minimizing stress. Protein aggregation will be
discussed in greater detail in section 2.11.
A
B
Figure I.16 – Possible adaptations to hydrophobic mismatch between a TM peptide and the lipid bilayer.
A – Response to positive hydrophobic mismatch (a): acyl-chain ordering (b), peptide backbone
deformation (c), peptide oligomerization (d), peptide tilt (e), non-TM orientation or absence of binding to
the bilayer (f). B – Response to negative hydrophobic mismatch (a): acyl-chain disordering (b), peptide
backbone deformation (c), peptide oligomerization (d), non-lamellar phase formation (e), non-TM
orientation or absence of binding to the bilayer (f). (Adapted from de Planque and Killian, 2003).
34

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
In the presence of lipids with tendency to establish nonlamellar structures, cubic and
nonlamellar lipid phases can be formed in the interface with the protein as a response to
hydrophobic mismatch. Nonlamellar arrangements of lipids change the curvature of the
bilayer in the vicinity of the protein and allow for a better fitting of the hydrophobic
sections of both proteins and lipids (Killian, 2003).
Peptide backbone deformation, such as transition from α-helix to π-helix (helices
with a wider helical pitch), could be a possible mechanism for adjustment to situations
of hydrophobic mismatch. However, several studies have confirmed that the α-helical
structure is insensitive to hydrophobic mismatch and this strategy to minimize
hydrophobic mismatch is unlikely (de Planque and Killian, 2003). Nevertheless a recent
study suggests that TM α-helices may flex in the lipid bilayer in order to submerge most
of the hydrophobic domain inside the bilayer core (Yeagle et al., 2007).
The hydrophobic matching principle might be a mechanism by which the activity of
some membrane proteins can be modulated. Thickness variations induced either
internally or externally can trigger some proteins in or out of an active status. This has
already been observed for some ion pumps such as Ca 2+ -ATPase and Na + ,K + -ATPase
that exhibit maximum activity in bilayers with a specific number of carbon atoms (Lee,
2003).
2.7. Transmembrane protein-lipid interface
As already stated, the lipids on the vicinity of the protein can either stretch or
compress their acyl-chains in response to hydrophobic mismatch. In pure lipids this
often results in the observation of two different lipid populations in protein-lipid
systems. One of these is very similar to the lipid in the absence of the protein (but not
necessarily identical), and its contribution to the total lipid population is inversely
dependent on protein concentration. The other population is more significant at higher
protein/peptide concentrations, frequently presenting a different gel-fluid phase
transition temperature (smaller T m for bilayers thinner than the protein and higher T m for
thicker bilayers) as well as a smaller cooperativity for this transition (Liu et al., 2002;
Morein et al., 2002). These two components are expected to correspond to free and
protein-associated lipids, respectively.
35

The formation of a lipid population associated with proteins or peptides with
properties different from the bulk or free lipid was predicted from results of different
techniques, namely differential scanning calorimetry and electron spin resonance (ESR).
From ESR experiments, a fraction of the spin labelled lipids was found to be
dynamically restricted by interaction with membrane proteins or peptides. From the
number of lipid molecules affected by the presence of a single TM domain it was
possible to retrieve the stoichiometry for the interaction of lipids and TM peptides. The
value recovered was 12, meaning that only the first shell of lipids around the TM
domain is significantly affected by its presence (a hexagonal-type arrangement of lipids
around a TM peptide is expected, six in each monolayer – see Figure I.17) (Marsh et al.,
2002). This shell around the TM domain is entitled the annulus and the lipids within are
called annular lipids.
Figure I.17 – Hexagonal lipid packing of annular lipids around a TM protein domain. Annular lipids
are shown in dark grey and bulk lipids are depicted in light grey.
The fact that only the first shell of lipids around the TM domain is significantly
affected explains why single TM peptides are generally not able to induce considerable
physical changes in the bulk membrane (except when the lipid/peptide ratio is large
enough so that a significant fraction of the lipids are annular lipids). However, multipass
TM proteins (containing multiple TM domains) were already shown to be able to affect
the bulk lipids and change the hydrophobic thickness of the bilayer (Harroun et al.,
1999). These results seem to indicate that the packing of lipid chains around a single α-
helix (which has a diameter ~ 10 Å, comparable to a lipid molecule) is significantly
different from the way lipids pack against a large protein surface (Weiss et al., 2003).
Still, when multipass TM proteins were studied by ESR with spin labelled-lipids, the
ratio of immobilized lipids/TM domains is often no longer 12 but smaller. This is
obviously due to interactions and contacts between the TM domains that decrease the
area available for contact with lipids (Marsh et al., 2002). However, if lipids outside the
36

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
annulus were considerably restricted by the presence of protein, the immobilized
lipids/TM domains ratio should come larger that for a single α-helix. Likely, the
influence of the TM proteins on the annular lipids is much stronger than on lipids on the
exterior shells, and the effects on the latter are not detectable by ESR.
The lipid exchange rate at the interface of the protein body was determined to be in
the region of 10 7 s -1 , which is on the same order but significantly lower than the value
obtained for lipid diffusion in the absence of protein (Marsh, 1990). This means that
although temporarily restricted to the protein surface, the lipid molecules interacting
with the protein body are not immobile and are still in exchange with other lipids.
Of relevance is the fact that lipids display different properties when at the surface of
proteins. This indicates that optimal hydrophobic matching might not necessarily occur
when the hydrophobic thickness of the protein matches the hydrophobic thickness of the
unperturbed bilayer (Dumas et al., 1999).
2.8. Lipid selectivity at protein interfaces
For membranes containing multiple lipid species, the hydrophobic matching
principle can result in even more complex behaviour at the lipid-protein interface.
Different lipid species are expected to present different free energies for interaction with
the presented TM domain and as a result, the lipid composition around the protein can
be different from the bulk lipid composition (Figure I.18).
Selectivity of membrane proteins for lipids in specific phases was observed in
various cases. In the presence of a mixture of DLPC and DSPC, a combination of
experimental and theoretical approaches indicated that bacteriorhodopsin exhibited
preferential interaction with DLPC at low temperatures, when both lipids where in the
gel phase, and at the fluid phase, bacteriorhodopsin had a preference for long chain
DSPC (Dumas et al., 1997). DLPC and DSPC present a very large difference in
hydrophobic length of acyl-chains and strongly nonideal mixing. Consequently, it is not
surprising that the affinity of the protein for the lipids differs drastically, and that the
protein is preferentially solvated by the lipids that are the best hydrophobic match in
each phase. Still, selectivity of proteins for lipids in binary mixtures with a much more
ideal mixing is also experimentally verified (Lee, 2003). Nevertheless, deviation of a
37

homogeneous distribution of lipids in the membrane does have an energetic entropic
cost and the degree of protein-lipid selectivity is a balance between the need to satisfy
hydrophobic matching conditions in the protein-lipid interface and the intrinsic
tendency of the system for mixing.
Figure I.18 – Illustration of a lipid bilayer with an embedded protein and two different lipid species.
The hydrophobic matching principle implies an accumulation of the lipid species that is hydrophobically
best matched to the protein (taken from Jensen and Mouritsen, 2004).
The hydrophobic surface of a membrane protein is not smooth. The interface
between the protein and the lipids surrounding it is likely to be heterogeneous, and the
interactions taking place there complex (Lee, 2003). Still the results described above,
concerning the presence of a fixed stoichiometry for annular lipids, denote some degree
of ordering in the TM alpha helix-lipid interface and in this way, processes of lipid
binding to simpler systems such as single TM domains are suitable to be described in
terms of a uniform surface for which several (12) identical binding sites are available
(Marsh et al., 2002). With total coverage of the protein surface by lipids, one lipid
molecule must leave the surface before another can enter. In this sense, the process can
be depicted as competitive binding of lipids at the binding sites on the protein surface
(Lee, 2003) (Figure I.19). This method of analysis has been applied with success to
even more complex membrane proteins (O’ Keefe et al., 2000; Williamson et al., 2002;
Powl et al., 2003).
38

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
Figure I.19 – View from above of lipid binding sites on a TM domain surface. Two lipid species (L 1
and L 2 ) are shown exchanging at one site.
At each site an equilibrium exists:
PL 1 + L 2 PL 2 + L 1
where PL 1 and PL 2 are the protein-lipid complexes with lipid species 1 and 2. This
equilibrium can be described by a binding constant (K b ):
b
[ PL
2
] ⋅[ L1]
[ PL ] ⋅[ L ]
K =
2.4
1
2
If L 2 is present at small amounts in the system (L 2 > [PL 2 ], then the probability (µ) of a lipid
annular site to be occupied by L 2 is:
[ PL
2
]
[ PL ] + [ PL ]
2
1
[ PL
2
]
[ PL ]
1
b
[ L
2
]
[ L ]
µ = = = K ⋅
2.5
1
Through knowledge of [PL 2 ] ([PL 1 ]=[P]-[PL 2 ]), and the concentration of the lipid
species 1 and 2 it is then possible to recover a binding constant for protein-lipid
selectivity.
The most popular methodologies for quantifying protein-lipid selectivity have been
the already described ESR measurements with spin labelled lipids, and measurements of
protein fluorescence quenching by brominated lipids (East and Lee, 1982; O’ Keeffe et
al., 2000; Williamson et al., 2002; Powl et al., 2003). From these studies, binding
constants were retrieved as function of fatty-acid chain for a series of proteins (Fig.
I.20).
39

Figure I.20 – Relative binding constants for OmpF (ο), KscA (∆) and Ca 2+ -ATPase (□) for
monounsaturated phosphatidylcholines with different acyl-chains relative to that for DOPC (Taken from
Lee, 2003).
As already discussed, the amplitude of changes in the relative binding constants of
the α-helical proteins, KscA and Ca 2+ -ATPase are not as high as one should expect
from equation 2.3 due to adaptation of proteins to more hydrophobically matching
configurations, likely by tilting and changes in the packing of α-helices (proteins are not
rigid bodies) whereas KscA exhibits preference for lipids with 22 carbons in the acylchain,
lipid binding constants for Ca 2+ -ATPase are hardly dependent on acyl-chain, and
OmpF is preferentially bound to shorter lipids (14 carbons).
Membrane proteins also exhibit selectivity for lipid headgroups. Strong interactions
are often found between positively charged TM domains (rich in Lys and Arg) and
negatively charged phospholipids. These selectivities can be the result of charge
interactions, and in these cases, after binding of the anionic lipid to the surface of the
protein, the positive potential in the vicinity of the protein would be slightly reduced. If
enough anionic lipid molecules bind to the vicinity of the protein, the positive potential
will be neutralized and remaining interactions with anionic phospholipids will be
nonspecifical. However, membrane proteins often present different selectivities for
different anionic phospholipids. This is likely to result from different balances in
charges inside each class of phospholipids. In this way, PS lipids due to the presence of
the positively charged ammonium group as well as the negatively charged carboxyl
group in the headgroup region may present reduced affinity to basic protein domains
when compared to phosphatidic acid. Also, PG, with a single charge in the phosphate
40

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
group interacts differently with charged membrane proteins, and proteins often present
smaller affinities for PG than for PS and PA (Mall et al., 2002).
The complexity to the protein-charged lipid interaction problem is raised by the
position of amino acid residues in the protein surface. Basic helices with different
hydrophobic thickness were shown to present different selectivities to negatively
charged lipids. This was rationalized as the result of tilting of helices in the membrane,
that located basic residues away from the headgroup region where strong interactions
between proteins and lipid headgroups are more likely (Mall et al., 2002). Additionally,
experiments with model peptides showed that insertion of Tyr decreased binding
affinities to anionic phospholipids, suggesting that the presence of Tyr residues
prevented close association of anionic phospholipids and cationic residues (Mall et al.,
2000).
These results suggest that the effects of charge on the interactions between lipids
and membrane proteins, though important are not determinant, and will be strongly
dependent on the detailed structure of the peptide and its orientation in the membrane
(Mall et al., 2002). Binding affinities of proteins for lipids are generally presented
relative to the binding affinity of PC lipids. PC and PE lipids frequently present the
lowest affinities for membrane proteins.
Binding sites for hydrophobic molecules have been shown to exist for several
proteins distinct from the sites for interaction with annular lipids. These are called nonannular
sites. They are generally located between TM α-helices or at protein-protein
interfaces. For some cases, these sites are available for interaction with some lipids but
not for others (Lee, 2003). Some lipid molecules were shown in crystal structures of
proteins to be bound to such sites. These observations suggest that the binding affinities
for these sites are much higher than the ones observed for annular lipids (Lee et al.,
2003).
Notably, several protein domains bind specifically to one or more forms of
phosphorylated PIP molecules through interactions that are very specific. Such domains
are often found on proteins involved in signal transduction. Other proteins bind
specifically to some components of raft domains, establishing a mechanism for
targeting the protein to specific structures in the membrane (Epand, 2005; Pérez-Gil et
al., 2005).
41

2.9. Lipid phase preferential partition of
membrane proteins
Many membrane proteins and peptides show a preference for bilayers in the fluid
phase over the gel phase. This results in a difference in partition between domains when
both phases coexist (Dumas et al., 1997). With even greater biological relevance,
proteins exhibit also differential partition between liquid ordered and liquid disordered
phases. In this way, the lateral structure of the lipid membrane is able to impose major
restrictions to the lateral distribution of membrane proteins. Preferential partition of a
membrane protein to a given region or phase of the membrane is likely to define
important structural features, such as protein densities or the probability of protein
encountering homologous or heterologous counterparts (see Section 2.11) (Pérez-Gil et
al., 2005).
Acylated proteins or lipid-anchored proteins will also preferentially partition to
certain domains. Acylation with palmitic acid or myristic acid promotes preferential
partition to more ordered domains such as membrane rafts. On the other hand,
prenylated proteins are commonly excluded from liquid ordered domains (Epand,
2005).
Changes of the lateral lipid membrane structure would thus have the ability to
induce profound rearrangements of the protein lateral distribution. This would have
major implications on the establishment of specific protein-protein interaction networks.
In fact, it has been proposed that the natural membrane composition is optimized to
maintain membranes on the threshold of a phase transition. Membranes in this situation
would present a highly dynamical behaviour able to respond through redistribution of
the protein network to environmental stimulus (Pérez-Gil et al., 2005).
Hydrophobic matching might also be the mechanism by which proteins are targeted
to their final destinations in the membrane. Sorting of proteins along the secretory
pathway might be performed by means of a gradient of hydrophobic thickness of the
membrane systems that the proteins have to pass on their way to their target. The
concentrations of cholesterol and sphingomyelin, which have a thickening effect on the
membrane, increase going from the endoplasmic reticulum, via the Golgi to the plasma
membrane. Also, the hydrophobic domains of proteins that reside in the Golgi are
shorter by about 5 amino acids when compared to those of the plasma membrane.
42

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
However, this proposed mechanism for sorting of membrane proteins is still
controversial (Mouritsen, 2005).
Concerning membrane protein partition to specific regions of the bilayer, of
particular interest is the possibility that certain membrane proteins prefer insertion in the
boundaries between lipid phases in the membranes. These boundaries present a higher
probability of lipid packing defects that can be used by certain proteins to access deeper
regions in the membrane (Pérez-Gil et al., 2005).
2.10. Lipid sorting by proteins and formation of
lipid domains
The presence of very stable lipid domains in the membrane is only expected if the
free energies for interaction between phase separated lipids is very high. In
biomembranes this is rarely the case, and therefore lateral structures that arise must have
a reasonably dynamical nature. In model membranes, the differences in free-energies
for lipid-lipid interactions are typically on the order of a few calories per mole. In
contrast, the free-energies involved in protein-lipid interactions are much larger
(Hinderliter et al., 2001). Therefore membrane proteins when incorporated in lipid
membranes can contribute decisively to the final lipid phase equilibria.
In the presence of a multicomponent lipid mixture in the fluid phase, insertion of a
membrane protein can lead, as already described, to enrichment of one of the lipid
species around the protein due to more efficient packing in the protein-lipid interface.
This preferential association can stabilize a lipid phase around the protein, even under
conditions where it is metastable in the protein-free system. This preferred phase is said
to wet the protein (Gil et al., 1998). The wetting layer of lipids only extends over a
finite (nonmacroscopic) distance. Theoretical studies have predicted that effects on
lipids by inclusion of proteins may extend up to 20 lipid shells around the protein (Fattal
and Ben-Shaul, 1993; Sperotto and Mouritsen, 1991), and results from some
experimental studies were rationalized on the basis of wetting of membrane proteins and
formation of lipid phases induced by protein-lipid interactions (Lehtonen and Kinnunen,
1997; Dumas et al., 1997; Hinderliter et al, 2004).
43

In Figure I.21, theoretical simulations are shown that predict the wetting
phenomenon. For Figure I.21c), sharing of a wetting layer by two or more proteins can
give rise to colocalization and aggregation phenomena as described in detail on Section
2.11.
A B B
Figure I.21 – Computer simulation snapshots of a large protein embedded in a binary lipid matrix in
which one of the lipids is preferentially bound to the protein. (a) – Selective binding of one of the lipid
species at the protein-lipid interface. (b) – Wetting of one of the lipids around the protein. (c) – Wetting
and formation of a capillary condensate between two adjacent proteins (taken from Gil et al., 1998).
These types of lateral structures initiated by proteins can function in biomembranes
as the signal for recruitment of other molecules to those domains. This has been
proposed as the molecular mechanism behind some signalling platforms (Pérez-Gil et
al., 2005).
2.11. Lipid-mediation of protein-protein
interactions
The principle of hydrophobic matching, and the possibility of membrane proteins to
change the lipid environment around them, offer membranes a mechanism for lipidmediation
of protein-protein interactions.
When proteins change the order parameter of lipids around them, overlap of
disturbed lipids can again lead to colocalization of proteins in the same area of the
membrane or aggregation (Gil et al., 1998) (Figure I.22). The system, when moving to
equilibrium, will have the tendency to decrease the extent of presented interface, and
this is achievable by clustering of proteins and ultimately by protein aggregation. Of
course, protein aggregation can have its energetic costs that may rise from helix-dipole
moment and charge repulsion or entropic factors.
44

INTRODUCTION: LIPID-PROTEIN INTERACTIONS
Figure I.22 - Schematic representation of protein-protein interactions mediated by hydrophobic
mismatch. (a) – Hydrophobically matched membrane. (b, c) - Lipid-mediated protein-protein attraction by
same type of hydrophobic mismatch. (d) - Lipid-mediated protein-protein repulsion by opposite type of
hydrophobic mismatch (Taken from Gil et al., 1998).
When more than one lipid species is present, wetting of membrane proteins also
implies formation of an interface between lipids in the wetting phase and lipids in the
protein-poor phase. Again the tendency to minimize the interface between these two
phases can lead to coalescence or capillary condensation of wetting domains and
stimulation of protein-protein interactions (Fig. I.21.c)). In this way, wetting occurs in
addition to direct protein-protein (Van der Waals and electrostatic) and the above
described lipid-mediated interactions, by selecting the environment in which they take
place, i.e., wetting provides the mechanism for modulation of membrane protein
organization (clustering and aggregation) on larger length scales (Gil et al., 1998).
45

PROTEIN-PROTEIN AND PROTEIN-LIPID INTERACTIONS
OF M13 MCP
II
1. Introduction
PROTEIN-PROTEIN AND
PROTEIN-LIPID
INTERACTIONS OF M13
MAJOR COAT PROTEIN
The M(unich)13 bacteriophage was discovered in 1963 by P. H. Hofschneider in an
incubation of waste water samples with Escherichia coli bacteria. It belongs to the large
family of filamentous phages (Inoviridae), which comprise some of the smallest
bacteriophages known (Spruijt et al., 1999).
These viruses are able to infect bacteria without great disturbance of the host cells.
In this way, reproduction of the virus inside the host goes on as cells continue to grow
and divide even if at a reduced rate. For efficient infection, filamentous phages require
expression of pili on the surface of the host bacteria. The pili acts as a receptor site, and
depending on the specificity for pili interaction, filamentous bacteriophages are divided
in different classes. The M13 bacteriophage belongs to the Ff group, which is only able
to infect E. coli cells carrying sex pili (Spruijt et al., 1999). Genetic manipulation of
plasmids derived from the M13 bacteriophage proved to be relatively easy, and this
allowed the M13 genome to be widely used as a cloning vector (Spruijt et al., 1999).
One of the factors that contributed to the enormous popularity of the M13
bacteriophage in biotechnology was its relatively simple architecture. The phage
47

particle is 900 nm in length and 6.5 nm wide. It is formed by a single strand of DNA,
about 6400 nucleotides long, encapsulated by a cylindrical protein coat. The protein
coat presents about 3000 copies of the major coat protein (mcp or gp8), the product of
gene VIII. In addition to mcp, four other coat proteins are present in few copies, capping
the ends of the bacteriophage.
During the infection process, parental coat proteins are embedded in the membrane
while the viral DNA is released in the cytoplasm (see Figure II-1). New copies of mcp
are synthesized in the host and inserted in the membrane, where they are processed by a
leader peptidase (Kuhn et al., 1986). After peptidase action, the C-terminus of mcp is
oriented to the cytoplasm and the N-terminus is located in the periplasm. In the mature
form, mcp is a polypeptide chain 50 aminoacids long and presents a single
transmembrane domain. During the reproductive cycle of the bacteriophage, coat
proteins are stored at very high local levels and finally assembled around viral DNA
into new phage particles (Hemminga et al., 1992).
Figure II-1: Representation of the reproductive cycle of bacteriophage M13 (taken from Spruijt et
al., 1999).
mcp presents three distinct domains which are expected to be related to the distinct
type of interactions that this protein is involved in during the reproductive cycle of the
bacteriophage. In this way, the negatively charged region in the N-terminal domain of
the protein (residues 1-20), rich in glutamate and aspartate, forms the outer hydrophilic
surface of the virus particle. The hydrophobic central domain (residues 21-39) is
responsible for the tight membrane interaction of the membrane form of mcp and for the
48

PROTEIN-PROTEIN AND PROTEIN-LIPID INTERACTIONS
OF M13 MCP
protein-protein interactions between mcp units inside the phage coat, which are essential
in providing stability to the particle. Finally, the C-terminus of mcp (residues 40-50) is
enriched in positively charged lysines. This domain lines the inside of the phage particle
and is responsible by non-specific interactions with DNA. The amino acid sequence of
mcp and the structure of the membrane bound form is shown in Figure II-2.
The N-terminus of the membrane bound form of mcp is a flexible α-helix with
amphipatic character. A loop connects this amphipatic domain with the transmembrane
helix (Figure II-2). The transmembrane domain of mcp is tilted by 20 ± 10º with respect
to the lipid bilayer normal (Glaubitz et al., 2000; Koehorst et al., 2004). ESR and
fluorescence studies making use of site-directed labelling of mcp (Spruijt et al., 1996;
Stopar et al., 1997a) allowed to conclude that Thr 36 is located in the center of the bilayer
in DOPC and 1,2-dioleoyl-sn-glycerol-3-phosphatidylglycerol (DOPG) bilayers, while
aminoacids 25 and 46 delineate the edges of the transmembrane domain of mcp. DOPC
is expected to present ideal hydrophobic matching conditions to mcp.
A
B
Figure II-2: A - Amino acid sequence of mcp (taken from Hemminga et al., 1992). B – Structure of
the membrane bound form of mcp (taken from Stopar et al., 2003).
mcp was also found to be strongly anchored to the lipid bilayer through its C-
terminus, namely by the two phenylalanines found there and by three lysine residues.
When mcp was incorporated in bilayers with slightly thinner or thicker hydrophobic
49

thickness than mcp, the N-terminus of the protein shifted its position in order to
accommodate the differences in hydrophobic lengths of protein and lipid, but the C-
terminus remains at approximately the same position in the bilayer interface due to
strong anchoring (Meijer et al., 2001). Another adaptation of mcp to hydrophobic
mismatch is changing the tilt angle of the transmembrane domain, as well as the rotation
of the helix at its long axis so as to optimize the hydrophobic and electrostatic
interactions at the C-terminus (Koehorst et al., 2004).
Several questions concerning lipid-protein and protein-protein interactions of mcp
are still subject of debate. During the life-cycle of bacteriophage M13, the aggregation
state of mcp changes repeatedly. In the phage particle it is aggregated in the coat, and in
the membrane it is believed to be monomeric (Russel, 1991; Spruijt and Hemminga,
1991), although a structural dimer might be formed as an intermediate species during
phage disruption (Henry and Sykes, 1992; Stopar et al., 1997b; Stopar et al., 1998). The
mcp structure is predominantly α-helical when inserted in the membrane, but depending
on the protein purification and membrane reconstitution procedure a irreversibly
aggregated β-sheet conformation of the protein appeared (Hemminga et al., 1992). Also,
factors such as L/P ratio, phospholipid headgroup type, length and degree of
unsaturation of the acyl-chains of lipids used, and the ionic strength of the buffer were
shown to be able to influence the α-helical to β-sheet transition. It is found that saturated
lipid chains and very high protein concentration resulted inevitably in high levels of
irreversible β-sheet aggregation (Spruijt and Hemminga, 1991). Nevertheless, the
irreversible β-sheet conformation of mcp is expected to be an artefact resulting from
inappropriate conditions of protein purification and reconstitution, as α-helical structure
is necessary for insertion in the phage particle (Hemminga et al., 1992).
The lipid composition of the inner membrane of non infected E. coli is about 70% of
PE, 25% of PG and 5% cardiolipin. During the infection of E. coli by the M13
bacteriophage, the levels of anionic lipids in the cell membrane are slightly increased
and blocking of phage assembly resulted in significant increases in anionic lipid content
in the inner membrane of bacteria (Pluschke et al., 1978). This phenomenon might
indicate that anionic lipids are required to maintain a functional state of the membranebound
form of the protein (Hemminga et al., 1992).
The in situ aggregational state of α-helical mcp is however still a matter of
discussion, as reversible oligomeric alpha-helical mcp was shown to exist in some
conditions (Hemminga et al., 1992). In vitro, aggregation behaviour is even more likely
50

PROTEIN-PROTEIN AND PROTEIN-LIPID INTERACTIONS
OF M13 MCP
than in vivo, as the reconstitution procedures normally result in randomized orientations
for mcp instead of the single orientation observed in vivo. mcp in the membrane with
identical orientations are expected to present smaller protein-protein electrostatic
attractive forces due to the stronger repulsion between the heavily charged C-terminus,
and consequently a higher free energy is expected for the oligomeric condition. In order
to achieve high local concentrations of mcp in the host membrane prior to phage
assembly, a dynamic protein-lipid network should be present, characterized by absence
of preferential association between coat proteins and/or lipids (Hemminga et al., 1992).
Concerning the protein-lipid interactions established by mcp, it is interesting to note
that the monomeric form of mcp was not able to induce sufficient immobilization of a
fraction of lipid molecules so that it could be detected by ESR with spin labelled lipids
(Sanders et al., 1992). On the other hand, oligomeric mcp immobilized a population of
lipids, especially at very high protein concentrations (Peelen et al., 1992). Thus, it can
be concluded that monomeric M13 mcp is not able to induce a long living lipid
boundary shell around it. Possible explanations are that the hydrophobic surface of mcp
exposed to the lipids is too small or that the diffusion rate of the protein is not
sufficiently lower than that of the lipids.
51

DEPENDENCE OF M13 MCP OLIGOMERIZATION AND LATERAL
SEGREGATION ON BILAYER COMPOSITION
2. DEPENDENCE OF M13 MAJOR COAT
PROTEIN OLIGOMERIZATION AND LATERAL
SEGREGATION ON BILAYER COMPOSITION
53

2430 Biophysical Journal Volume 85 October 2003 2430–2441
Dependence of M13 Major Coat Protein Oligomerization and Lateral
Segregation on Bilayer Composition
Fábio Fernandes,* Luís M.S.Loura,* y Manuel Prieto,* Rob Koehorst, z
Ruud B. Spruijt, z and Marcus A. Hemminga z
*Centro de Química-Física Molecular, Instituto Superior Técnico, Lisboa, Portugal; y Departamento de Química, Universidade de Évora,
Évora, Portugal; and z Laboratory of Biophysics, Wageningen University, Wageningen, The Netherlands
ABSTRACT M13 major coat protein was derivatized with BODIPY (n-(4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-sindacene-3-yl)methyl
iodoacetamide), and its aggregation was studied in 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC)
and DOPC/1,2-dioleoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (DOPG) or 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine
(DOPE)/DOPG (model systems of membranes with hydrophobic thickness matching that of the protein) using photophysical
methodologies (time-resolved and steady-state self-quenching, absorption, and emission spectra). It was concluded that the
protein is essentially monomeric, even in the absence of anionic phospholipids. The protein was also incorporated in pure
bilayers of lipids with a strong mismatch with the protein transmembrane length, 1,2-dierucoyl-sn-glycero-3-phosphocholine
(DEuPC, longer lipid) and 1,2-dimyristoleoyl-sn-glycero-3-phosphocholine (DMoPC, shorter lipid), and in lipidic mixtures
containing DOPC and one of these lipids. The protein was aggregated in the pure vesicles of mismatching lipid but remained
essentially monomeric in the mixtures as detected from BODIPY fluorescence emission self-quenching. From fluorescence
resonance energy transfer (FRET) measurements (donor-n-(iodoacetyl)aminoethyl-1-sulfonaphthylamine (IAEDANS)-labeled
protein; acceptor-BODIPY labeled protein), it was concluded that in the DEuPC/DOPC and DMoPC/DOPC lipid mixtures,
domains enriched in the protein and the matching lipid (DOPC) are formed.
INTRODUCTION
M13 coat protein has been shown to exist in many aggregation
states, depending on factors like isolation, reconstitution
procedure, pH, ionic strength, and amphiphile composition
(Hemminga et al., 1993; Stopar et al., 1997). The mechanism
of phage assembly in the Escherichia coli membrane is not
yet completely understood, but the assembly site has been
proposed to be composed of a dynamic protein-lipid network,
characterized by absence of preferential association between
M13 coat protein and/or lipids (Hemminga et al., 1993),
which allows storage of monomeric coat protein at very high
local concentrations, as the insertion of the protein in the
assembling phage particle is only possible on the monomeric
form (Russel, 1991). The type of interactions between lipids
and coat protein which allows for formation of this structure is
largely unknown and in this study it is intended to obtain
information on the influence of lipid bilayer composition on
the lateral distribution and oligomerization properties of M13
coat protein in membrane model systems, to better understand
the phage particle assembly mechanism.
It has been proposed that self-association behavior of
transmembrane proteins while incorporated in lipid membranes
is influenced by hydrophobic matching conditions on
the protein-lipid interface (Mouritsen and Bloom, 1984;
Submitted January 21, 2003, and accepted for publication June 24, 2003.
Address reprint requests to Manuel Prieto, Centro de Química-Física
Molecular, Complexo I, Instituto Superior Técnico, Av. Rovisco Pais,
1049-001 Portugal. Tel.: 35-121-841-9219; Fax: 35-121-846-4455; E-mail:
prieto@alfa.ist.utl.pt.
Ó 2003 by the Biophysical Society
0006-3495/03/10/2430/12 $2.00
Killian, 1998). The monomeric protein is expected to be
stable under perfect matching conditions with the phospholipids
surrounding it. In case of hydrophobic mismatch at the
protein-lipid interface, it is possible that the boundary lipids
reorganize themselves, to lower the tension created by exposure
of hydrophobic acyl chains or amino-acid residues,
which can be achieved by ordering/disordering of the phospholipids
(Nezil and Bloom, 1992). Moreover, if the hydrophobic
mismatch is too high for correction with small
adjustments of bilayer hydrophobic thickness, this might
result in protein aggregation to obtain minimization of the
protein-lipid contacts (Ren et al., 1999; Lewis and Engelman,
1983, Mobashery et al., 1997; Mall et al., 2001). Other
responses to hydrophobic mismatch might be transmembrane
helix tilt (for long transmembrane hydrophobic protein
domains), change in conformation or decrease of bilayer
partitioning of the protein, transition of phospholipids to
nonlamellar phases, and molecular sorting of proteins and
lipids in the presence of a binary lipid system (for reviews
see Killian, 1998; Dumas et al., 1999).
Some studies have already related the oligomerization
state of proteins to the hydrophobic mismatch extent in the
bilayer. Aggregation was shown to depend on hydrophobic
mismatch for bacteriorhodopsin at extreme hydrophobic
mismatch conditions (Lewis and Engelman, 1983) and
gramicidin (Mobashery et al., 1997). Mall and co-workers
(Mall et al., 2001) concluded that, for a synthetic peptide, the
free energy of dimerization increased linearly with each
additional carbon of the phospholipid fatty acyl chain, with
a slope of 0.5 kJ mol ÿ1 . Apparently, this effect is much more
significant upon negative hydrophobic thickness (thicker
hydrophobic section for the bilayer than for the protein; Ren

M13 Coat Protein Lateral Distribution 2431
et al., 1999; Lewis and Engelman, 1983; Mall et al., 2001).
Meijer et al. (2001) found by electron spin resonance that the
M13 major coat protein incorporated in di(22:1)PC (1,2-
dierucoylphosphatidylcholine) aggregated or existed in several
orientations/conformations, whereas in di(14:1)PC (1,2-
dimyristoleoylphosphatidylcholine) no indication was found
toward aggregation.
In addition, for proteins incorporated in binary phospholipid
mixtures, strong selectivity to one lipid component, and
phase separation or lipid sorting effects (depending on their
miscibility), was predicted to occur as a result of hydrophobic
mismatch (Sperotto and Mouritsen, 1993). This phenomenon
has also been observed experimentally in a mixture of
di(12:0)PC (dilauroylphosphatidylcholine)/di(18:0)PC(distearoylphosphatidylcholine),
in which bacteriorhodopsin was
shown to preferentially associate with the short chain lipid in
the gel-gel and gel-fluid coexistence regions (Dumas et al.,
1997). Although the preferential association of bacteriorhodopsin
with short chain lipid in the gel-fluid coexistence
region could be understood by an eventual preference for the
disordered phase, these results were rationalized as being a
consequence of lipid-protein hydrophobic matching interactions.
A similar effect was observed for E. coli lactose
permease (Lehtonen and Kinnunen, 1997), and for gramicidin
(Fahsel et al., 2002). In the case of bacteriorhodopsin, Dumas
and co-workers (Dumas et al., 1997), using a theoretical
model, concluded that the protein was preferentially associated
with the longer chain lipid on the mixed fluid lipid phase.
In this mixture no macroscopic phase separation was occurring,
but only preference of protein association with the phospholipid
which allowed for more favorable energetic
interactions, resulting in bacteriorhodopsin being surrounded
by di(18:0)PC. This phenomenon is also denominated by
wetting, and may extend to multiple layers of phospholipids
(Gil et al., 1998). The interfacial stress, which is likely to
occur between the wetting phase and the lipid matrix, can
lead to sharing of these microdomains by many proteins,
to minimize the interface area. As a result, formation of protein-enriched
domains would occur in the bilayer. This
process has been proposed as a mechanism for protein
aggregation inducement (Gil et al., 1998).
Some studies have also dealt with transmembrane protein/
peptide interaction selectivity toward anionic phospholipids,
based on electrostatic interactions of the lipid headgroup and
basic residues on the protein (Horvàth et al., 1995a,b). A
glucosyltransferase from Acholeplasma laidlawii was found
to exhibit preference for localization on PG domains formed
in a PC matrix (Karlsson et al., 1996).
The purpose of this work is to study the influence of the
bilayer composition on the aggregation state of the M13 coat
protein. In addition, the protein was also incorporated in
binary lipidic systems, where there is strong hydrophobic
mismatch of the protein with one of the components. The
possibility of protein segregation to lipidic domains in this
situation was also investigated.
For these purposes, several fluorescence methodologies
(fluorescence self-quenching, absorption and emission spectra,
and energy transfer) were applied, using the protein
derivatized with n-(4,4-difluoro-5,7-dimethyl-4-bora-3a,4adiaza-s-indacene-3-yl)methyl
iodoacetamide (BODIPY FL
C 1 -IA) or n-(iodoacetyl)aminoethyl-1-sulfonaphthylamine
(IAEDANS), as described in more detail below.
Through the use of the self-quenching of the BODIPY
fluorescence, it is expected to obtain information about the
aggregation/oligomerization state of protein. With the same
objective, the presence of BODIPY ground-state dimers is
investigated. These techniques allow us to check for molecular
contacts between BODIPY groups labeled on the transmembrane
section of the mutant protein.
To solve the question of whether the presence of M13
major coat protein is capable of inducing lipid domain
formation through electrostatic interactions or hydrophobic
mismatch, FRET measurements are carried out on different
lipid mixtures with M13 major coat protein incorporated. As
the C-terminal of the M13 major coat protein is heavily
basic, it is intended to know if the presence of protein would
lead to formation of domains enriched in anionic phospholipids
and protein. Additionally, by using mixtures of
phospholipids with different acyl-chain lengths, formation
of protein-enriched domains due to preferential binding to
hydrophobically matching lipids is checked.
In reconstituted systems we can have two different
orientations for the M13 coat protein in the bilayer (with
the N-terminus sticking to the inside or the outside of the
lipid vesicle), and in this way, interactions between proteins
might involve parallel or antiparallel protein orientations.
For this reason, the mutants chosen for the present work were
T36C and A35C, because for the coat protein in vesicles of
DOPC, the Thr36 and Ala35 residues are located near the
center of the bilayer, as was shown by AEDANS wavelength
of maximum emission (Spruijt et al., 2000). This increases
the possibility of self-quenching due to the formation of
complexes or from collisions between fluorophores, and
allowed us to ignore situations with complex oligomers
containing proteins with parallel and antiparallel orientations
as well as to simplify the energy transfer analysis to the twodimensional
case, as, for both orientations, both residues
should be located at approximately the same position.
MATERIALS AND METHODS
1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1,2-dioleoyl-sn-glycero-3-[phospho-rac-(1-glycerol)]
(DOPG), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine
(DOPE), 1,2-dierucoyl-sn-glycero-3-phosphocholine
(DEuPC) and 1,2-dimyristoleoyl-sn-glycero-3-phosphocholine (DMoPC),
were obtained from Avanti Polar Lipids (Birmingham, AL). N-(iodoacetyl)aminoethyl-1-sulfonaphthylamine
(1,5-IAEDANS) and N-(4,4-difluoro-
5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-yl)methyl) iodoacetamide
(BODIPY FL C 1 -IA) were obtained from Molecular Probes (Eugene,
OR). Fine chemicals were obtained from Merck (Darmstadt, Germany). All
materials were used without further purification.
Biophysical Journal 85(4) 2430–2441

2432 Fernandes et al.
Coat protein isolation and labeling
The wild-type M13 major coat protein and the T36C mutant were grown,
purified from the phage (Spruijt et al., 1996) and the A35C mutant was
obtained from transformed cells of E. coli B21 (DE3) (Spruijt et al., 2000).
Both mutants were labeled with BODIPY and IAEDANS as described
previously (Spruijt et al., 1996). The remaining impurities were extracted
using size exclusion chromatography and the protein was eluted with 50 mM
sodium cholate, 150 mM NaCl, and 10 mM Tris-HCl pH 8.
Coat protein reconstitution in lipid vesicles
The labeled protein mutant was reconstituted in the DOPC/DOPG (80/20
mol/mol), DOPC, DOPE/DOPG (70/30 mol/mol), DMoPC/DOPC (60/40
mol/mol), DEuPC/DOPC (60/40 mol/mol), DMoPC, and DEuPC vesicles
using the cholate-dialysis method (Spruijt et al., 1989). The phospholipid
vesicles were produced as follows: the chloroform from solutions containing
the desired phospholipid amount was evaporated under a stream of dry N 2
and last traces removed by a further evaporation under vacuum. The lipids
were then solubilized in 50 mM sodium cholate buffer (150 mM NaCl, 10
mM Tris-HCl, and 1 mM EDTA) at pH 8 by brief sonication (Branson 250
cell disruptor) until a clear opalescent solution was obtained, and then mixed
with the wild-type and labeled protein. Samples had a phospholipid
concentration of 1 mM and the lipid-to-protein ratio (L/P) was always kept at
50, with addition of wild-type protein when necessary, except for the studies
with pure DMoPC and DEuPC bilayers in which the L/P was always \50
(due to a smaller labeling efficiency obtained in the preparation of A35C
mutant) and no wild-type protein was added. Dialysis was carried at room
temperature and in the dark (to prevent IAEDANS degradation), with a 100-
fold excess buffer containing 150 mM NaCl, 10 mM Tris-HCl, and 1 mM
EDTA at pH 8. The buffer was replaced 53 every 12 h.
Fluorescence spectroscopy
Absorption spectroscopy was carried out with a Jasco V-560 spectrophotometer
(Tokyo, Japan). The absorption of the samples was kept \0.1 at the
wavelength used for excitation.
CD spectroscopy was performed on a Jasco J-720 spectropolarimeter
with a 450 W Xe lamp (Easton, MD).
Steady-state fluorescence measurements were carried out with an SLM-
Aminco 8100 Series 2 spectrofluorimeter (Rochester, NY; with double
excitation and emission monochromators, MC400) in a right-angle
geometry. The light source was a 450-W Xe arc lamp and for reference
a Rhodamine B quantum counter solution was used. Correction of emission
spectra was performed using the correction software of the apparatus. 5 3 5
mm quartz cuvettes were used. All measurements were performed at room
temperature.
In the fluorescence quenching measurements, the BODIPY emission
spectra were recorded with an excitation wavelength of 460 nm and
a bandwidth of 4 nm for both excitation and emission.
AEDANS-labeled protein quantum yield was determined using quinine
sulfate (f ¼ 0.55) (Eaton, 1988) as reference.
The excitation wavelength for the energy transfer measurements was 340
nm and the emission spectra were recorded with an excitation and emission
bandwidth of 4 nm.
Fluorescence decay measurements of IAEDANS were carried out with
a time-correlated single-photon counting system, which is described elsewhere
(Loura et al., 2000). Measurements were performed at room
temperature. Excitation and emission wavelengths were 340 and 440 nm,
respectively. The timescales used were between 12 and 67 ps/ch, depending
on the amount of BODIPY-labeled protein present in the sample. Data
analysis was carried out using a nonlinear, least-squares iterative convolution
method based on the Marquardt algorithm (Marquardt, 1963). The goodness
of the fit was judged from the experimental x 2 -value, weighted residuals, and
autocorrelation plot.
In all cases, the probes florescence decay were complex and described by
a sum of exponentials,
where a i are the normalized amplitudes.
The average lifetimes are defined by
IðtÞ ¼+ a i expðÿt=t i Þ; (1)
i
hti ¼
+ a i t 2 i
i
+
i
a i t i
; (2)
and the amplitude average or lifetime-weighted quantum yield (Lakowicz,
1999), is obtained as
t ¼ + a i t i : (3)
i
THEORETICAL BACKGROUND
Fluorescence self-quenching
Collisional quenching is responsible for a decrease in both
the quantum yield and lifetime of a fluorophore, whereas
static quenching only affects the fluorescence intensity due to
the dark (‘‘nonfluorescent’’) character of the self-quenching
complex. The effects of the collisional contribution of selfquenching
on the fluorescence lifetime are described by the
Stern-Volmer equation,
t 0 =t ¼ 1 1 k q 3 t 0 3 ½FŠ; (4)
where t 0 is the lifetime of the fluorophore in the absence of
quenching, t is the lifetime of the fluorophore in presence of
quenching, k q is the bimolecular diffusion rate constant, and
[F] is the concentration of the fluorophore.
The diffusion coefficient of the fluorophore (D) can be
calculated from k q using the Smoluchowski equation
(Lakowicz, 1999), taking into account transient effects
(Umberger and Lamer, 1945),
k q ¼ 4 3 p 3 N a 3 ð2 3 R c Þ 3 ð2 3 DÞ
3 ½1 1 2 3 R c =ð2 3 t 0 3 DÞ 1=2 Š; (5)
where N a is the Avogadro number and R c is the collisional
radius. Diffusion in membranes is here considered to take
place in an isotropic three-dimensional medium. If the
membrane were strictly bidimensional, different boundary
conditions for the Smoluchowski formalism should be
applied (Razi-Naqvi, 1974). The best approach to the
specific situation of probe diffusion in a membrane is the
one used by Owen (1975), in which the finite bilayer width is
considered (cylindrical geometry). Owen introduced the
parameter t s , which defines the crossover from the spherical
(three-dimensional) to the cylindrical geometry, its value
being t s ffi 42 ns when considering the bilayer and the
peptide. This value is much longer than the longest
fluorescence lifetime of the probes in the peptide (6.1 ns),
Biophysical Journal 85(4) 2430–2441

M13 Coat Protein Lateral Distribution 2433
and longer than the experimental time-window (28 ns ¼ 28
ps/channel 3 1000 channels) so the three-dimensional
framework approximation is essentially correct. Almgren
(1991), in a comparative study of quenching in restricted
dimensionality, also states that deviations from the threedimensional
occur only for very long fluorescence lifetimes.
The static quenching component can be described through
a sphere of action, which accounts for statistical contact pairs
formed at the moment of excitation. These contact pairs are
nonfluorescent, although they do not form a complex, i.e.,
the interaction energy is \k 3 T. The combined contributions
of the collisional and the sphere-of-action effects on the
fluorescence intensity are given by Loura et al. (2000) as
I F ¼
C 3 ½FŠ 3 expðÿV
1
s 3 N A 3 ½FŠÞ: (6)
1 k q 3 ½FŠ
t 0
Here, I F is the fluorescence intensity, C is a constant, and V s
is the sphere-of-action volume. The sphere-of-action radius
is obtained by
R s ¼ V s = 4 1=3
3 3 p ; (7)
and for a collisional quenching mechanism it should be close
to the sum of the Van der Waals radii.
In case that a complex is formed, the model to describe
static quenching effects should take into account its equilibrium
constant. For a monomer/dimer equilibrium of only
one molecular species, the fluorescence intensity is given by
I F ¼
C 3 ½FŠ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ÿ1 1 1 1 8 3 K a 3 ½FŠ
3 ; (8)
1
4 3 K
1 k q 3 ½FŠ
a
t 0
where K a is the oligomerization constant.
However, for our system, in which there are two different
protein species (labeled and unlabeled, where the unlabeled
class includes both nonlabeled mutant and wild-type protein),
there will be several combinations of protein species
within an aggregate. For a dimer, there will be three different
combinations available—labeled protein/labeled protein,
labeled protein/unlabeled protein, and unlabeled protein/unlabeled
protein—but only formation of the first one induces
self-quenching of BODIPY. A complexation model describing
fluorescence static quenching in our system will have to
account for this fact. From the knowledge of the concentration
of each species, the fraction of labeled protein participating
in oligomers (dimers/trimers) containing more than
one BODIPY labeled protein (and as a result nonfluorescent),
can be obtained for a given K a . The resulting set of
nonlinear equations was solved (for a given total protein
concentration, labeling efficiency, labeled protein concentration,
aggregation number, and K a ) using Maple V (Waterloo,
Ontario).
Fluorescence resonance energy transfer
The energy transfer between fluorophores can be used to
characterize the lateral distribution of labeled coat protein
mutant in the bilayer. The degree of fluorescence emission
quenching of the donor caused by the presence of acceptors
is used to calculate the experimental energy transfer efficiency
(E):
E ¼ 1 ÿ I DA =I D : (9)
Here, I DA is the donor fluorescence emission intensity in the
presence of acceptor, and I D is the donor fluorescence
emission intensity in the absence of acceptor.
Wolber and Hudson (1979) obtained the analytical solution
for energy transfer efficiencies in a random distribution of
acceptors in a bidimensional space,
‘
E ¼ 1 ÿ + ÿp 3 G 2 j G j
3 R 2 0
j¼0 3
3 n 3 1 1
2 3 ;
j!
(10)
where R 0 is the Förster radius, G is the complete gamma
function, and n 2 is the acceptor numerical density (number of
acceptors per unit area).
The Förster radius is given by
R 0 ¼ 0:2108 3 ðJ 3 k 2 3 n ÿ4 3 f D Þ 1=6 ; (11)
where J is the spectral overlap integral, k 2 is the orientation
factor, n is the refractive index of the medium, and f D is the
donor quantum yield. J is calculated from
ð
J ¼ f ðlÞ 3 eðlÞ 3 l 4 3 dl; (12)
where f(l) is the normalized emission spectra of the donor
and e(l) is the absorption spectra of the acceptor. If the
l-units in Eq. 12 are nm, the calculated R 0 in Eq. 11 has
Å-units (Berberan-Santos and Prieto, 1987).
RESULTS
The M13 coat protein is known to form large irreversible
aggregates under specific conditions. These aggregates are
not found in vivo, and therefore are regarded as an artifact
(Hemminga et al., 1993). Due to their large percentage of
b-sheet conformation, they can be detected by CD spectroscopy.
Both the wild-type and labeled mutant protein
conformation were checked by CD spectroscopy in all lipid
systems used, and all spectra obtained were typical for
a-helices (data not shown), indicating the absence of significant
b-sheet protein conformation.
The fluorescence decay of BODIPY in the labeled mutant
proteins incorporated in DOPC, DMoPC/DOPC, DEuPC/
DOPC, DMoPC, and DEuPC bilayers was described by two
components t 1 ¼ 6.23 ns (a 1 ¼ 0.9) and t 2 ¼ 3.27 ns, which
Biophysical Journal 85(4) 2430–2441

2434 Fernandes et al.
leads to an average lifetime of ht 0 i¼6.1 ns (Eq. 2), as measured
in samples with [BD-M13 coat protein] eff \10 ÿ3 M
(BODIPY-labeled protein effective membrane concentration).
For the determination of protein effective concentration
in the membranes, the lipid molar volumes were calculated
from the reported lipid areas (72 Å 2 ) and membrane thicknesses
(Tristam-Nagle et al., 1998; Lewis and Engelman,
1983).
Considering a random protein distribution in the bilayers,
and fitting Eq. 4 to the experimental average lifetimes, linear
Stern-Volmer plots are obtained (Fig. 1) and k q values are
recovered (Table 1).
The fluorescence steady-state quenching profiles for BD-
M13 coat protein in DOPC, DMoPC/DOPC, DEuPC/DOPC,
DMoPC, and DEuPC are presented in Fig. 2. This data can
be fitted to Eq. 6 using the already-obtained k q -values to
retrieve the fluorescence quenching contribution of the
sphere-of-action effect (Table 1).
Many studies on protein and peptide aggregation make
use of fluorophore spectral changes induced by ground or
excited state dimer formation (Otoda et al., 1993; Liu et al.,
1998; Melnyk et al., 2002; Shigematsu et al., 2002). Recently,
under restricted geometry, BODIPY was shown to
form ground-state dimers in two different conformations (D j
and D k ), with distinct spectroscopic properties from the
monomer. The D j BODIPY dimer (parallel transition dipoles)
is characterized by a strong excitation peak at 477 nm
and a null quantum yield. The D k dimer (planar transition
dipoles) exhibits an absorption peak at 570 nm and a broad
fluorescence emission band centered at 630 nm (Bergström
et al., 2001). In our BODIPY-derivatized protein none of the
spectral properties assigned to the D j and D k dimers are
observed, and both the excitation and fluorescence emission
spectra obtained for the samples with higher concentration of
TABLE 1
Systems k q /(10 9 M ÿ1 s ÿ1 ) D/(10 ÿ7 cm 2 s ÿ1 ) R s (Å)
T36C in DOPC 2.3 0.7 14
A35C in DOPC 1.7 0.4 14
A35C in DMoPC 2.6 0.8 23
A35C in DEuPC 5.0 2.6 27
T36C in DMoPC/DOPC 6.6 4.2 14
T36C in DEuPC/DOPC 20 22 14
Bimolecular diffusion rate constants (k q ), diffusion coefficients (D), and
apparent sphere-of-action radii (R s ) recovered from BODIPY fluorescence
emission self-quenching from BODIPY-labeled T36C and A35C mutants.
Values of R s [ 14 Å are evidence of molecular aggregation (see text for
details).
BODIPY-labeled protein were that expected for BODIPY
monomers and identical in all lipid systems (Fig. 3).
The donor-acceptor pair chosen for the energy transfer
study shows a wide overlap of donor (AEDANS) fluorescence
emission and acceptor (BODIPY) absorption (Fig. 3).
Energy transfer studies were performed for the labeled
protein incorporated in DOPC, DOPC/DOPG (80/20 mol/
mol), DOPE/DOPG (70/30 mol/mol), DEuPC/DOPC (60/40
mol/mol), and DMoPC/DOPC (60/40 mol/mol). It was intended
to study the influence of electrostatic interactions,
hydrophobic mismatch, and the presence of nonlamellar
lipids in the aggregational and compartmentalization properties
of the M13 coat protein. Due to the nonlamellar
character of phosphatidylethanolamines, it was necessary to
include a fraction of lamellar lipids (PG) in the lipid mixture
for bilayer stabilization.
The fluorescence emission spectra of the T36C mutant
labeled with IAEDANS and reconstituted in these lipid
systems were identical in the DOPC, DOPC/DOPG, DOPE/
DOPG, DEuPC/DOPC, and DMoPC/DOPC lipid systems
FIGURE 1 Transient state Stern-Volmer plots, describing BODIPY fluorescence self-quenching at different labeled protein concentrations. The lines are fits
of Eq. 4 to the data. (A) Labeled protein incorporated in DOPC (m); DMoPC/DOPC (60/40 mol/mol) (); and DEuPC/DOPC (60/40 mol/mol) (d). (B) Labeled
protein incorporated in DMoPC (n); DEuPC (n); and Stern-Volmer fit to DOPC data (see A) (- - -). The BODIPY-labeled mutant used in the DMoPC and
DEuPC lipid systems was A35C, and for the other three lipid compositions the mutant used was T36C.
Biophysical Journal 85(4) 2430–2441

M13 Coat Protein Lateral Distribution 2435
FIGURE 2 Fluorescence steady-state data for BODIPY fluorescence self-quenching at different labeled protein concentrations. (A) Protein incorporated in
DOPC (m); DMoPC/DOPC (60/40 mol/mol) (); and DEuPC/DOPC (60/40 mol/mol) (d). Eq. 6 is fitted to the data on the basis of dynamical quenching and
a sphere-of-action quenching model (14.4 Å of radius) (—) for the protein in all lipid systems. (B) Protein incorporated in DOPC (m); DMoPC (); and DEuPC
(). Eq. 6 fit of data from DOPC bilayers using a sphere-of-action radii of 14 Å (–), from DEuPC bilayers with a sphere-of-action radii of 27 Å (- - -), and from
DMoPC bilayers using a sphere-of-action of 23 Å (–). These higher values are evidence of aggregation. See text for details.
(results not shown), their wavelength of maximum fluorescence
emission (477 nm) being characteristic of a very apolar
environment (Hudson and Weber, 1973) and very similar to
the one obtained by Spruijt and co-workers for the same
mutant (478 nm; Spruijt et al., 2000). The wavelengths of
maximum emission for the IAEDANS-labeled protein in
DMoPC and DEuPC were slightly different, 478 nm and 475
nm, respectively.
Donor fluorescence intensities (I D and I DA in Eq. 9)
obtained by steady-state measurements and by integrated
donor decays were identical. The IAEDANS-labeled protein
quantum yield determined by us was f ¼ 0.64. Using Eqs.
11 and 12, assuming k 2 ¼ 2/3 (the isotropic dynamic limit)
and n ¼ 1.4 (Davenport et al., 1985), we obtain R 0 ¼ 48.8 Å
FIGURE 3 Corrected excitation spectra of IAEDANS-labeled protein
(—), and of BD-M13 coat protein (- - -). Corrected emission spectra of
IAEDANS-labeled protein (–), and of BD-M13 coat protein (- --).
for this FRET pair. The value k 2 ¼ 2/3 was considered
because for fluorophores in the center of a fluid bilayer, the
rotational freedom should be sufficiently high to randomize
orientations. This is supported by the reasonably low steadystate
anisotropy values obtained for the IAEDANS and
BODIPY probes labeled on the T36C M13 coat protein
mutant (hri AEDANS ¼ 0.14, hri BODIPY ¼ 0.23; for a detailed
discussion, see Loura et al., 1996).
The results for BD-M13 coat protein in DOPC, DOPC/
DOPG, DOPE/DOPG, DMoPC/DOPC, and DEuPC/DOPC
bilayers are presented in Fig. 4.
DISCUSSION
For membrane proteins incorporated in lipid bilayers, the
net tendency for protein aggregation should be weaker
under good lipid-protein hydrophobic matching conditions
(Mouritsen and Bloom, 1984). As the hydrophobic a-helix
of the M13 coat protein is composed by 20 amino-acid
residues, its length is ;30 Å. The lipids used in this work,
with the exception of DMoPC (21 Å) and DEuPC (35 Å),
form bilayers with almost the same hydrophobic thickness
(28 Å) (Lewis and Engelman, 1983). Therefore, the oligomerization
properties of the M13 major coat protein in
DMoPC and DEuPC should differ from those in the hydrophobically
matching phospholipid DOPC.
The conditions of hydrophobic mismatch considered in
this study do not seem to be able to induce any protein
conformational change, as checked by CD spectroscopy in
both lipid systems (no spectral change found). The absence
of b-sheet conformation for the M13 coat protein implies no
irreversible aggregation in the bilayer, and consequently any
self-association observed should be due to reversible interactions
between the a-helices.
Biophysical Journal 85(4) 2430–2441

2436 Fernandes et al.
FIGURE 4 (A) Donor (AEDANS) fluorescence quenching by energy transfer acceptor (BODIPY). Experimental data for DOPC (m); DOPC/DOPG (80/20
mol/mol) (d); and DOPE/DOPG bilayers (70/30 mol/mol) (). Theoretical expectation (Eq. 10) for energy transfer in a random distribution of acceptors (—).
Energy transfer simulation for a total co-localization of M13 coat protein in 20% (- - -), and 30% (— - —) of the surface area available. (B) Donor (AEDANS)
fluorescence quenching by energy transfer acceptor (BODIPY). Theoretical expectation for energy transfer in a random distribution of acceptors (—). Energy
transfer simulation for a segregation of M13 coat protein (mcp) to 60% of the surface area available (- - -). Experimental data for DEuPC/DOPC (60/40 mol/
mol) (d); experimental data for DMoPC/DOPC (60/40 mol/mol) (). I DA and I D were obtained by integration of donor decays.
The monitoring of IAEDANS maximum fluorescence
emission (l max ) of the labeled mutant also rules out any
change in conformation or exclusion from the bilayer of the
M13 coat protein when incorporated in hydrophobically
mismatching phospholipid, for the IAEDANS l max in these
samples (475–478 nm) are almost identical to that observed
with DOPC and the other hydrophobic matching phospholipids,
and are typical for the fluorophore located near the
center of the bilayer (Spruijt et al., 2000).
Although BODIPY fluorescence decay is essentially
monoexponential, the complex decay we obtained (dominated
by a component of 6.3 ns) is also reported for derivatized
proteins (Karolin et al., 1994).
The BODIPY fluorescence emission self-quenching
studies were performed with two different mutants. For the
lipid mixtures T36C was used, and the A35C mutant was
employed in the studies with pure mismatching lipid. As a
control, the experiments were performed for both mutants in
pure DOPC bilayers and the results were identical (Table 1).
The bimolecular quenching constants calculated from the
self-quenching results for BD-M13 coat protein incorporated
in DOPC, DMoPC/DOPC (60/40 mol/mol) DEuPC/DOPC
(60/40 mol/mol), DEuPC, and DMoPC allow the estimation
of the labeled protein molecular diffusion coefficient through
Eq. 5. Considering for BODIPY a collisional radius of 6 Å,
we obtain for D BD-M13 coat protein in DOPC bilayers a value of
7.0 3 10 ÿ8 cm 2 s ÿ1 , which is the same order of magnitude of
the values of D for the M13 coat protein incorporated in fluid
bilayers reported in the literature (Smith et al., 1979, 1980).
However, for the lipid mixtures in which the predominant
lipid does not hydrophobically match with the hydrophobic
core of the M13 coat protein, the values obtained for D are
unreasonably high. For the DMoPC/DOPC bilayers, a value
of 4.2 3 10 ÿ7 cm 2 s ÿ1 is obtained, whereas in DEuPC/
DOPC it is even higher (2.2 3 10 ÿ6 cm 2 s ÿ1 ). If D BD-M13 coat
protein in pure vesicles of DOPC is considered to report a
random distribution in the bilayer, the values in these mixtures
are likely to be reporting protein segregation effects in
the bilayer. We believe that this is caused by the hydrophobic
mismatch constraints the protein finds when incorporated in
bilayers that have too-long, or too-short phospholipids in
their composition, probably leading to formation of localized
areas with increased content of DOPC and protein. In this
way, the effective apparent concentration in Eq. 4, should be
higher than the one assumed on the basis of a random distribution,
leading to an overestimation of k q , and so of D.
However, the increase in local protein concentration arising
from this effect would still be insufficient to explain the one
order-of-magnitude increase of D from pure DOPC bilayers
to the studied DEuPC/DOPC mixture (see below for further
discussion).
This rationalization is supported by D values obtained
from BODIPY labeled protein in the pure mismatching lipid
DMoPC and DEuPC (Table 1). These values are smaller than
the ones obtained from the mixtures, and the diffusion
coefficient in pure DMoPC is almost identical to the value in
pure DOPC. For pure vesicles of DEuPC, D BD-M13 coat protein
is larger than in DOPC, but it is still much smaller than the
value obtained from the DEuPC/DOPC mixture. The results
from dynamical self-quenching indicate therefore that although
in pure vesicles of DEuPC there are already more
collisions between BODIPY groups than what could be
expected from a random distribution of labeled protein in the
bilayer (probably due to aggregation), when DOPC is added
the probability of collision greatly increases, and this can in
part be explained in terms of protein segregation to DOPC-
Biophysical Journal 85(4) 2430–2441

M13 Coat Protein Lateral Distribution 2437
FIGURE 5 Fluorescence steady-state data for BODIPY fluorescence selfquenching
at different labeled protein concentrations. (A) Protein incorporated
in DOPC (m) and simulations considering protein oligomerization
including static quenching: due to trimerization of the protein with a K a
¼ 1000 ( 10% total mcp oligomerization) (—); due to dimerization of the
protein with a K a ¼ 10 ( 25% total mcp oligomerization) (- - -). (B) Protein
incorporated in DMoPC () and simulations for dimerization of protein with
a K a ¼ 20 ( 13% total mcp oligomerization for the most concentrated data
point) (- - -) and trimerization of the protein with a K a ¼ 1300 ( 13% total
mcp oligomerization for the most concentrated data point) (—). (C) Protein
incorporated in DEuPC (d) and simulations for dimerization of protein with
a K a ¼ 30 ( 17% total mcp oligomerization for the most concentrated data
enriched microdomains. In DMoPC/DOPC the effect is
similar, but the bimolecular quenching constant is smaller
than in DEuPC/DOPC.
In Fig. 2, the obtained steady-state quenching profiles are
presented together with the theoretical expectation for a
sphere-of-action quenching model (Eq. 6). For the BODIPYlabeled
protein in the DOPC-containing lipid systems
(DOPC, DMoPC/DOPC, and DEuPC/DOPC) the results
are well described using a sphere-of-action radius of 14 Å.
For the pure mismatching lipids it is necessary to use larger
radii to describe the results using Eq. 6.
In Fig. 5, a simulation was included for a small degree of
protein aggregation in DOPC, DMoPC, and DEuPC
bilayers. In these simulations it was considered that, due to
the small degree of self-association considered, there was no
change in M13 coat protein distribution and dynamics, and
that in an oligomer, the fluorescence intensity of a BODIPY
group is reduced to zero by the presence of another BODIPY
group in the same aggregate. For DOPC bilayers, the
prediction using a low fraction of aggregation (25% for
dimerization and 10% for trimerization) clearly overestimates
the extent of aggregation at the high labeled protein
concentration, the range where this methodology is more
sensitive. In agreement, from Fig. 2, it is clear that the data
for the three DOPC-containing lipid systems are rationalized
on the basis of dynamic quenching and a sphere of action,
without need for assumption of aggregation. The recovered
radius (R s ¼ 14 Å) is close to the sum of the Van der Waals
radii. These results indicate that BD-M13 coat protein in the
studied DOPC-containing bilayers does not oligomerize.
This conclusion is further supported by the absence of
BODIPY dimers in our samples, which would be revealed in
the absorption/emission spectra.
It could be possible that in a labeled protein oligomer,
there is no contact between the BODIPY groups, and
therefore the formation of oligomers would not necessarily
induce fluorescence self-quenching. Some mutational studies
actually include the Thr36 residue in a lipid-interactive
face of the transmembrane segment of the M13 coat protein.
The M13 coat protein amino acids in this lipid-interactive
face would only have contact with the phospholipid acyl
chains inside the bilayer, and would be excluded from the
transmembrane domain face involved in protein-protein interactions,
where the amino acids responsible for the affinity
of identical transmembrane segments are located (Deber
et al., 1993; Webster and Haigh, 1998). That positioning of
the mutated Cys36 residue would make the contact between
BODIPY groups from labeled proteins participating in an
oligomer unlikely, and therefore exclude the formation of
‘‘dark’’ (nonfluorescent) dimers. Localization of the BOD-
IPY group on the lipid-interactive face of the transmembrane
point) (- - -) and trimerization of the protein with a K a ¼ 7500 ( 25% total
mcp oligomerization for the most concentrated data point) (—). See text for
details on the simulations.
Biophysical Journal 85(4) 2430–2441

2438 Fernandes et al.
section of the protein can be seen as an advantage, inasmuch
as the mutation of an amino-acid residue to cysteine after the
labeling with fluorophore, at the protein-protein interactive
face, could cause dramatic changes in the oligomerization
behavior of the M13 coat protein. The T36C mutant labeled
with BODIPY could be seen, for that reason, as a more ideal
model for wild-type protein aggregation studies than the
M13 coat protein mutant A35C, in which the mutated aminoacid
residue is located in the protein-interactive face of the
hydrophobic core domain. However, as can be concluded
from the results obtained for both mutants labeled with
BODIPY in DOPC (Table 1), the positioning of the
BODIPY group is not critical. Probably the fluorophore
has sufficient mobility to extend from the surface of the
transmembrane helix and probe a large extent of the area
surrounding the protein.
The BODIPY self-quenching results obtained in the present
study clearly indicate that there is no protein aggregation
in presence of hydrophobic matching phospholipids for all
hydrophobic headgroup compositions used, and points to
a large monomer stability under those conditions, which had
already been suggested in other studies (Stopar et al., 1997;
Spruijt et al., 1989; Sanders et al., 1991).
Still, the results from self-quenching on pure bilayers
of hydrophobic mismatching lipids (DEuPC and DMoPC)
point to some aggregation, as the sphere-of-action radii
recovered from the data fit to Eq. 6 were very unrealistic
(27 and 23 Å for DEuPC and DMoPC, respectively).
Simulations for BODIPY emission self-quenching due to
aggregation are compared with the experimental data in these
lipid systems in Fig. 5. DMoPC data could be reasonably
described using a K a ¼ 20 for dimerization and a K a ¼ 1300
for trimerization (13% of aggregated protein at the protein
concentration of the most concentrated data point in both
simulations).
For the data obtained in DEuPC, the degree of aggregation
is higher than in DMoPC (recovered sphere-of-action radius
on fit to Eq. 6 was larger), and the fitting of the aggregation
models to the data points proved more difficult. Probably the
M13 major coat protein in DEuPC bilayers forms somewhat
larger aggregates than in DMoPC and that could be an explanation
for the increased BODIPY dynamical self-quenching
observed in the longer lipid. This result is in agreement
with the observations of Meijer et al. (2001), who, from electron
spin resonance studies, reported that the protein appeared
to exist in several orientations/conformations or in an aggregated
form while incorporated in DEuPC bilayers. The larger
extent of coat protein aggregation observed in the longer
lipid bilayers can be explained by the fact that negative hydrophobic
mismatch is considered to be energetically less
favorable than positive mismatch (Killian, 1998; Mall et al.,
2001).
As mentioned above, for the DEuPC/DOPC and DMoPC/
DOPC lipid mixtures used in the present study, no change in
conformation or orientation was found (CD spectroscopy/
AEDANS fluorescence emission spectra), and even for the
protein in DEuPC/DOPC (60/40 mol/mol) no aggregation
was detected (BODIPY self-quenching). These results point
to stabilization of the protein by the hydrophobic matching
phospholipid (DOPC), which was probably achieved by
protein segregation to domains enriched in that phospholipid,
partly explaining the high M13 coat protein apparent
molecular diffusion coefficients obtained for the protein
incorporated in DEuPC/DOPC and DMoPC/DOPC bilayers.
Although an increase in BODIPY emission dynamical selfquenching
was already visible for pure DEuPC bilayers
(probably due to the large dimensions of the aggregates,
which will have a similar effect as the segregation of protein
to microdomains on inducing co-localization of protein in
the bilayer), the bimolecular diffusion rate constant value in
DEuPC/DOPC bilayers is much higher, and still, this can
only be explained by segregation to DOPC-enriched microdomains.
Being that M13 coat protein preferably locates in DOPCenriched
domains, an interesting question is whether these
domains are induced by the protein or exist even in the
absence of protein (and the latter merely distributes differently
among the pre-existing DOPC-rich and DOPC-poor
regions). In this regard we obtained preliminary results of
1,6-diphenylhexatriene fluorescence anisotropy upon varying
compositions for both DMoPC/DOPC and DEuPC/
DOPC mixtures in the absence of protein (data not shown),
pointing to the existence of ‘‘phase coexistence regions’’
containing the compositions studied in this work. This
experiment is not informative regarding the extent of phase
separation, but it is compatible with the existence of small
lipid clusters enriched in one of the components, because
fluorescence anisotropy only senses the immediate vicinity
of the fluorophore. Regarding the DEuPC/DOPC mixtures,
these measurements imply that for 60:40 (mol/mol) DEuPC/
DOPC, DEuPC-rich domains should coexist with DOPCrich
domains, and the latter should account for approximately
one-third of the mixture (not shown). From this
result, we could expect at most a threefold increase in protein
local concentration for this mixture, and an identical factor
for the increase of the recovered k q value relative to that in
DOPC. As seen in Table 1, this factor is significantly higher.
This suggests that whereas protein segregation into DOPCrich
domains should be occurring, it is not the sole cause for
the increase in the apparent diffusion coefficient in 60:40
(mol/mol) DEuPC/DOPC. In this regard the other contributing
factors are not clear. It should be pointed out, however,
that the estimate presented above for the fraction of DOPCrich
domains was made from experiments performed in
the absence of protein, whereas the quenching results refer to
L/P ¼ 50.
The centered position of BODIPY in the bilayer allows for
a simplification of the energy transfer analysis for a twodimensional
situation, as described by Eq. 10, i.e., there is no
need to consider bilayer FRET geometry (Loura et al., 2001).
Biophysical Journal 85(4) 2430–2441

M13 Coat Protein Lateral Distribution 2439
Simulations of energy transfer for random distribution of
acceptors using Eq. 10 can therefore be compared with our
experimental results (Fig. 4). The energy transfer efficiencies
obtained for BD-M13 coat protein in the DMoPC/DOPC and
DEuPC/DOPC bilayers support the other results discussed
above for these mixtures, as they can only be explained by
protein segregation in the bilayer or severe aggregation (Fig.
4 B). However, as discussed above, the data obtained by
fluorescence emission self-quenching indicate that segregation
into DOPC-enriched domains (rather than aggregation)
is the major phenomenon in these lipid mixtures.
Although the results can be reasonably explained on the
basis of protein segregation to 60% of the total bilayer area
(Fig. 5 B), this rationalization should be considered an oversimplification,
and is presented as an illustration. Indeed, the
measured efficiencies are only reporting the average BD-
M13 coat protein surface density that each IAEDANSlabeled
protein is sensing. Probably there will be M13 coat
protein interacting with the hydrophobically mismatching
phospholipids, but the majority of the proteins will be
preferentially surrounded by DOPC, and microdomains
enriched in DOPC and M13 coat protein should be formed.
The lipid mixtures used in this work for hydrophobic
matching studies (DMoPC/DOPC and DEuPC/DOPC) are
thought to be considerably closer to ideality than the ones
used in the studies of Dumas and co-workers (gel/fluid
coexistence; Dumas et al., 1997) and Lehtonen and Kinunnen
(natural and pyrene-derivatized lipids; Lehtonen and
Kununnen, 1997). However, protein segregation was observed
in our study for both DMoPC/DOPC and DEuPC/
DOPC, whereas in DOPC random distribution of protein
in the bilayer was confirmed. Also interestingly, the degree
of segregation appears to be similar for both mixtures (Fig.
5 B), which was not expected due to the observed larger
aggregation degree of coat protein in DEuPC, and therefore
to an apparent larger packing difficulty with the longer
lipid.
The formation of local structure within the thermodynamic
fluid phase for a mixture of two PCs with a 4-carbon
difference in acyl-chain lengths (DMPC/DSPC) was theoretically
predicted by Mouritsen and Jørgensen (1994). The
microdomains size in the Monte Carlo configurations
obtained by these authors appears to be very small (10–20
molecules at most). However, significant alterations in FRET
efficiency as measured by Lehtonen et al. (1996) in mixtures
of unsaturated PCs, and indeed our study of FRET between
IAEDANS-labeled protein and BODIPY-labeled protein,
require that the domain size should be of the order of magnitude
of R 0 , that is ffi5 nm. In our system, this large domain
size might be a consequence of protein-induced phase separation.
In this work we also studied whether similar heterogeneities
could be induced by the presence of positively
charged M13 coat protein in bilayers composed of mixtures
of anionic and neutral phospholipids. Due to the basic character
of M13 coat protein C-terminal, it is reasonable to
consider the possibility of anionic phospholipid-enriched
domain induction by M13 coat protein incorporation in the
bilayer. The formation of these domains could actually help
explain some of the mechanisms involved in the creation of
the phage assembly site.
Assuming that the hypothetical domains were composed
by all the protein and DOPG content in the sample, and
that the protein would be randomly distributed inside them,
we can, using Eq. 10, obtain theoretical curves describing
the energy transfer within these domains. These plots are
compared with the experimental data points for M13 coat
protein incorporated in DOPC/DOPG (80/20 mol/mol) and
DOPE/DOPG (70:30 mol/mol) in Fig. 5 A.
It is concluded that the segregation of M13 coat protein to
a PG-rich phase in the mixed systems, induced by electrostatic
interactions between the positively charged protein and
the negatively charged phospholipid, is ruled out on the basis
of the obtained data, for this process would lead locally to
greater surface densities of acceptor, and therefore, a very
significant increase of energy transfer efficiencies should be
expected in these conditions.
In addition to being the predominant phospholipids of the
E. coli inner membrane (Woolford et al., 1974; Burnell et al.,
1980), nonlamellar phospholipids (such as DOPE) are
known to interact distinctly from lamellar lipids with proteins,
and in some cases to influence their conformation and
activity (Hunter et al., 1999; Ahn and Kim, 1998). Our
results suggest that these phospholipids have apparently no
effect on the lateral distribution (Fig. 4 A) of the protein,
which is kept random at the total protein concentration used.
Higher protein concentrations than L/P 50 were not tested
due to the risk of inducing nonlamellar phases, but it is likely
that at smaller L/P ratios than those used in this work
attractive interactions between monomers of M13 coat
protein could take place, especially as we have two protein
orientations in these reconstituted systems (parallel/antiparallel),
and the repulsive electrostatic forces between the
heavily basic C-terminal of the protein would be eliminated.
In vivo, as there is only one orientation, the achievement of
high protein concentrations in the monomeric state at the
assembly site would be more favorable (Hemminga et al.,
1993). The presence of anionic phospholipids, as shown, is
not essential for monomer stabilization, and the increase
in phosphatidylglycerol and cardiolipin production during
virus infection (Chamberlain et al., 1978) should only play
a stabilizing role at the very high M13 coat protein concentrations
that are expected to exist at the phage assembly site.
CONCLUSIONS
From this study, it is concluded that the M13 coat protein
monomeric state is highly stable when incorporated in
bilayers containing hydrophobic matching phospholipids.
The lack of anionic phospholipids has no effect on the pro-
Biophysical Journal 85(4) 2430–2441

2440 Fernandes et al.
tein oligomerization properties at the protein concentrations
used in this study. When the protein is incorporated in pure
vesicles of mismatching lipid there is evidence for protein
aggregation but for mixtures of lipids containing
both hydrophobically matching (DOPC) and mismatching
(DEuPC or DMoPC) phospholipids, the protein probably
segregates to domains enriched in DOPC, which can explain
the stability of the monomeric species of the protein in these
lipid systems. This segregation effect is only observed when
hydrophobic mismatching phospholipids are present, suggesting
that the hydrophobic matching conditions on the
protein-lipid interface are more important than electrostatic
interactions between the M13 coat protein and the
phospholipids, for the protein lateral distribution on the
bilayer.
The authors thank Dr. A. Fedorov for assistance in the time-resolved
measurements.
F.F. acknowledges financial support from Fundação para a Ciência e
Tecnologia (FCT), project POCTI/36458/QUI/2000 and COST Action
D:22; L.M.S.L. and M.P. acknowledge financial support from FCT,
projects POCTI/36458/QUI/2000 and POCTI/36389/FCB/2000.
REFERENCES
Ahn, T., and H. Kim. 1998. Effects of non-lamellar-prone lipids on the
ATPase activity of SecA bound to model membranes. J. Biol. Chem.
273:21692–21698.
Almgren, M. 1991. Kinetics of excited states processes in micellar media.
In Kinetics and Catalysis in Microheterogeneous Systems. M. Gratzel,
and K. Kalyanasundaram, editors. Marcel Dekker, New York. 63–113.
Berberan-Santos, M. N., and M. J. E. Prieto. 1987. Energy transfer in
spherical geometry. Application to micelles. J. Chem. Soc. Faraday
Trans. 283:1391–1409.
Bergström, F., I. Mikhalyov, P. Hägglöf, R. Wortmann, T. Ny, and L. B.-Å.
Johansson. 2001. Dimers of dipyrrometheneboron difluoride (BODIPY)
with light spectroscopic applications in chemistry and biology. J. Am.
Chem. Soc. 124:196–204.
Burnell, E., L. Van Alphen, A. Verkleij, and B. De Kruijff. 1980. 31P
Nuclear magnetic resonance and freeze-fracture electron microscopy
studies on Escherichia coli. I. Cytoplasmic membrane and total
phospholipids. Biochim. Biophys. Acta. 597:492–501.
Chamberlain, B. K., Y. Nozaki, C. Tanford, and R. E. Webster. 1978.
Association of the major coat protein of FD bacteriophage with
phospholipid vesicles. Biochim. Biophys. Acta. 510:18–37.
Davenport, L., R. E. Dale, R. H. Bisby, and R. B. Cundall. 1985.
Transverse location of the fluorescent probe 1,6-diphenyl-1,3,5-hexatriene
in model lipid bilayer membrane systems by resonance energy
transfer. Biochemistry. 24:4097–4108.
Deber, C. M., A. R. Khan, Z. Li, C. Joensson, and M. Glibowicka. 1993.
Val ! Ala mutations selectively alter helix-helix packing in the
transmembrane segment of phage M13 coat protein. Proc. Natl. Acad.
Sci. USA. 90:11648–11652.
Dumas, F., M. M. Sperotto, M.-C. Lebrun, J.-F. Tocanne, and O. G.
Mouritsen. 1997. Molecular sorting of lipids by bacteriorhodopsin in
dilauroylphosphatidylcholine/distearoylphosphatidylcholine lipid bilayers.
Biophys. J. 73:1940–1953.
Dumas, F., M. C. Lebrun, and J.-F. Tocanne. 1999. Is the protein/lipid
hydrophobic matching principle relevant to membrane organization and
functions FEBS Lett. 458:271–277.
Eaton, D. F. 1988. Reference materials for fluorescence measurement. Pure
Appl. Chem. 60:1107–1114.
Fahsel, S., E. M. Pospiech, M. Zein, T. L. Hazlet, E. Gratton, and
R. Winter. 2002. Modulation of concentration fluctuations in phaseseparated
lipid membranes by polypeptide insertion. Biophys. J. 83:
334–344.
Gil, T., J. H. Ipsen, O. G. Mouritsen, M. C. Sabra, M. M. Sperotto, and
M. J. Zuckermann. 1998. Theoretical analysis of protein organization in
lipid bilayers. Biochim. Biophys. Acta. 1376:245–266.
Hemminga, M. A., J. C. Sanders, C. J. A. M. Wolfs, and R. B. Spruijt.
1993. Lipid-protein interactions involved in bacteriophage M13 infection.
In Protein-Lipid Interactions, New Comprehensive Biochemistry,
Vol. 25. A. Watts, editor. Elsevier, Amsterdam. 191–212.
Horvàth, L. I., T. Heimburg, P. Kovachev, J. B. Findlay, K. Hideg, and
D. Marsh. 1995a. Integration of a K 1 channel-associated peptide in a
lipid bilayer: conformation, lipid-protein interactions, and rotational
diffusion. Biochemistry. 34:3893–3898.
Horvàth, L. I., P. F. Knowles, P. Kovachev, J. B. Findlay, and D. Marsh.
1995b. A single-residue deletion alters the lipid selectivity of a K 1
channel-associated peptide in the b-conformation; spin-label electron
spin resonance studies. Biophys. J. 73:2588–2594.
Hudson, E. N., and G. Weber. 1973. Synthesis and characterization of two
fluorescent sulfhydryl reagents. Biochemistry. 12:4154–4161.
Hunter, G. W., S. Negash, and T. C. Squier. 1999. Phosphatidylethanolamine
modulates Ca-ATPase function and dynamics. Biochemistry.
38:1356–1364.
Karlsson, O. P., M. Rytomaa, A. Dahqvist, P. K. Kinnunnen, and
A. Wieslander. 1996. Correlation between bilayer dynamics and activity
of the diglucosyldiacylglycerol synthase from Acholeplasma laidlawii
membranes. Biochemistry. 35:10094–10102.
Karolin, J., J. B.-Å. Johansson, L. Strandberg, and T. Ny. 1994.
Fluorescence and absorption spectroscopic properties of dipyrrometheneboron
difluoride (BODIPY) derivatives in lipids, lipid membranes, and
proteins. J. Am. Chem. Soc. 116:7801–7806.
Killian, J. A. 1998. Hydrophobic mismatch between proteins and lipids in
membranes. Biochim. Biophys. Acta. 1376:401–416.
Lakowicz, J. R. 1999. Principles of Fluorescence Spectroscopy, 2nd Ed.
Kluwer Academic/Plenum Press, New York. Chap. 8.
Lewis, B. A., and D. M. Engelman. 1983. Bacteriorhodopsin remains
dispersed in fluid phospholipid bilayers over a wide range of bilayer
thickness. J. Mol. Biol. 166:203–210.
Lehtonen, J. Y. A., J. M. Holopainen, and P. K. J. Kinnunen. 1996.
Evidence for the formation of microdomains in liquid crystalline large
unilamellar vesicles caused by hydrophobic mismatch of the constituent
phospholipids. Biophys. J. 70:1753–1760.
Lehtonen, J. Y. A., and P. K. J. Kinnunen. 1997. Evidence for phospholipid
microdomain formation in liquid crystalline liposomes reconstituted with
Escherichia coli lactose permease. Biophys. J. 72:1247–1257.
Liu, W., Y. Chen, H. Watrob, S. G. Bartlett, L. Jen-Jacobson, and M. D.
Barkley. 1998. N-termini of EcoRI restriction endonuclease dimers are in
close proximity on the protein surface. Biochemistry. 37:15457–15465.
Loura, L. M. S., A. Fedorov, and M. Prieto. 1996. Resonance energy
transfer in a model system of membranes: application to gel and liquid
crystalline phases. Biophys. J. 71:1823–1836.
Loura, L. M. S., A. Fedorov, and M. Prieto. 2000. Membrane probe
distribution heterogeneity: a resonance energy transfer study. J. Phys.
Chem. B. 104:6920–6931.
Loura, L. M. S., R. F. M. de Almeida, and M. Prieto. 2001. Detection and
characterization of membrane microheterogeneity by resonance energy
transfer. J. Fluoresc. 11:197–209.
Mall, S., R. Broadbridge, R. P. Sharma, J. M. East, and A. G. Lee. 2001.
Self-association of model transmembrane helix is modulated by lipid
structure. Biochemistry. 40:12379–12386.
Marquardt, D. W. 1963. An algorithm for least-squares estimation of nonlinear
parameters. J. Soc. Ind. Appl. Math. 11:431–441.
Meijer, A. B., R. B. Spruijt, C. J. A. M. Wolfs, and M. A. Hemminga. 2001.
Membrane-anchoring interactions of M13 major coat protein. Biochemistry.
40:8815–8820.
Biophysical Journal 85(4) 2430–2441

M13 Coat Protein Lateral Distribution 2441
Melnyk, R., A. Partridge, and C. Deber. 2002. Transmembrane domain
mediated self-assembly of major coat protein subunits from FF
bacteriophage. J. Mol. Biol. 315:63–72.
Mobashery, N., C. Nielsen, and O. S. Andersen. 1997. The conformational
preference of gramicidin channels is a function of lipid bilayer thickness.
FEBS Lett. 412:15–20.
Mouritsen, O. G., and M. Bloom. 1984. Mattress model of lipid-protein
interactions in membranes. Biophys. J. 46:141–153.
Mouritsen, O. G., and K. Jørgensen. 1994. Dynamical order and disorder in
lipid bilayers. Chem. Phys. Lipids. 73:3–25.
Nezil, F. A., and M. Bloom. 1992. Combined influence of cholesterol and
synthetic amphiphilic peptides upon bilayer thickness in model
membranes. Biophys. J. 61:1176–1183.
Otoda, K., S. Kimura, and Y. Imanishi. 1993. Orientation and aggregation
of hydrophobic helical peptides in phospholipid bilayer membrane.
Biochim. Biophys. Acta. 1150:1–8.
Owen, C. S. 1975. Two-dimensional diffusion theory: cylindrical diffusion
model applied to fluorescence quenching. J. Chem. Phys. 62:3204–3207.
Razi-Naqvi, K. 1974. Diffusion-controlled reactions in two-dimensional
fluids: discussion of measurements of lateral diffusion of lipids in
biological membranes. Chem. Phys. Lett. 28:280–284.
Ren, J., S. Lew, J. Wang, and E. London. 1999. Control of the
transmembrane orientation and interhelical interactions within membranes
by hydrophobic helix length. Biochemistry. 38:5905–5912.
Russel, M. 1991. Filamentous phage assembly. Mol. Microbiol. 5:1607–
1613.
Sanders, J. C., N. A. J. van Nuland, O. Edholm, and M. Hemminga. 1991.
Conformational and aggregation of M13 coat protein studied by
molecular dynamics. Biophys. Chem. 41:193–202.
Shigematsu, D., M. Matsutani, T. Furuya, T. Kiyota, S. Lee, G. Sugihara,
and S. Yamashita. 2002. Roles of peptide-peptide charge interaction and
lipid phase separation in helix-helix association in lipid bilayer. Biochim.
Biophys. Acta. 1564:271–280.
Smith, L. M., B. A. Smith, and H. M. McConnell. 1979. Lateral diffusion of
M-13 coat protein in model membranes. Biochemistry. 18:2256–2259.
Smith, L. M., J. L. R. Rubenstein, J. W. Parce, and H. M. McConnel. 1980.
Lateral diffusion of M-13 coat protein in mixtures of phosphatidylcholine
and cholesterol. Biochemistry. 19:5907–5911.
Sperotto, M. M., and O. G. Mouritsen. 1993. Lipid enrichment and
selectivity of integral membrane proteins in two-component lipid
bilayers. Eur. Biophys. J. 22:323–328.
Spruijt, R. B., C. J. A. M. Wolfs, and M. A. Hemminga. 1989. Aggregation
related conformational change of the membrane-associated coat protein
of bacteriophage M13. Biochemistry. 28:9158–9165.
Spruijt, R. B., C. J. A. M. Wolfs, J. W. G. Verver, and M. A. Hemminga.
1996. Accessibility and environment probing using cysteine residues
introduced along the putative transmembrane domain of the major coat
protein of bacteriophage M13. Biochemistry. 35:10383–10391.
Spruijt, R. B., A. B. Meijer, C. J. A. M. Wolfs, and M. A. Hemminga. 2000.
Localization and rearrangement modulation of the N-terminal arm of the
membrane-bound major coat protein of the bacteriophage M13. Biochim.
Biophys. Acta. 1509:311–323.
Stopar, D., R. B. Spruijt, C. J. A. M. Wolfs, and M. A. Hemminga. 1997. In
situ aggregational state of M13 bacteriophage major coat protein in
sodium cholate and lipid bilayers. Biochemistry. 36:12268–12275.
Tristam-Nagle, S., H. I. Petrache, and J. F. Nagle. 1998. Structure and
interactions of fully hydrated dioleoylphosphatidylcholine bilayers.
Biophys. J. 75:917–925.
Umberger, J. Q., and V. K. Lamer. 1945. The kinetics of diffusioncontrolled
molecular and ionic reactions in solution as determined by
measurements of the quenching of fluorescence. J. Am. Chem. Soc. 67:
1099–1109.
Webster, R. E., and N. G. Haigh. 1998. The major coat protein of
filamentous bacteriophage F1 specifically pairs in the bacterial
cytoplasmic membrane. J. Mol. Biol. 279:19–29.
Wolber, P. K., and B. S. Hudson. 1979. An analytical solution to the Förster
energy transfer problem in two dimensions. Biophys. J. 28:197–210.
Woolford, J. L., Jr., J. S. Cashman, and R. E. Webster. 1974. F1 Coat
protein synthesis and altered phospholipid metabolism in F1-infected
Escherichia coli. Virology. 58:544–560.
Biophysical Journal 85(4) 2430–2441

QUANTIFICATION OF PROTEIN-LIPID SELECTIVITY USING FRET:
APPLICATION TO THE M13 MCP
3. QUANTIFICATION OF PROTEIN-LIPID
SELECTIVITY USING FRET: APPLICATION
TO THE M13 MAJOR COAT PROTEIN.
67

344 Biophysical Journal Volume 87 July 2004 344–352
Quantification of Protein-Lipid Selectivity using FRET: Application
to the M13 Major Coat Protein
Fábio Fernandes,* Luís M.S.Loura,* y Rob Koehorst, z Ruud B. Spruijt, z Marcus A. Hemminga, z
Alexander Fedorov,* and Manuel Prieto*
*Centro de Química-Física Molecular, Instituto Superior Técnico, Lisbon, Portugal; y Departamento de Química,
Universidade de Évora, Évora, Portugal; and z Laboratory of Biophysics, Wageningen University, Wageningen, The Netherlands
ABSTRACT Quantification of lipid selectivity by membrane proteins has been previously addressed mainly from electron spin
resonance studies. We present here a new methodology for quantification of protein-lipid selectivity based on fluorescence
resonance energy transfer. A mutant of M13 major coat protein was labeled with 7-diethylamino-3((4#iodoacetyl)amino)phenyl-
4-methylcoumarin to be used as the donor in energy transfer studies. Phospholipids labeled with N-(7-nitro-2-1,3-
benzoxadiazol-4-yl) were selected as the acceptors. The dependence of protein-lipid selectivity on both hydrophobic mismatch
and headgroup family was determined. M13 major coat protein exhibited larger selectivity toward phospholipids which allow for
a better hydrophobic matching. Increased selectivity was also observed for anionic phospholipids and the relative association
constants agreed with the ones already presented in the literature and obtained through electron spin resonance studies. This
result led us to conclude that fluorescence resonance energy transfer is a promising methodology in protein-lipid selectivity
studies.
INTRODUCTION
For integral membrane proteins, the interaction with lipids is
dependent, among other factors, on the hydrophobic segments
and interface properties of both lipids and protein.
Minimal bilayer perturbation is achieved when the hydrophobic
length of the protein matches that of the surrounding
lipid (Mouritsen and Bloom, 1984). Deviations from perfect
hydrophobic matching at the protein-lipid interface create
a tension arising from the exposure of hydrophobic residues
or acyl-chains to the hydrophilic medium. It is considered
that the protein-lipid system adapts to hydrophobic mismatch
conditions through several alternative and nonexclusive
strategies such as ordering or disordering of perturbed
phospholipids (change in bilayer thickness), lipid phase
transition to nonlamellar phases, protein oligomerization/
aggregation (minimization of interface area), helices tilting
or side-chain rotation of a helical terminal residue (reduction
in effective hydrophobic length), change in protein orientation,
and decrease in the bilayer partitioning of the protein
(for reviews see Killian, 1998; Dumas et al., 1999).
In the case of proteins incorporated in lipid systems
containing lipids with different electrostatic properties or
hydrophobic lengths, selectivity to one lipid component at
the protein-lipid interface or preferential phase partitioning
(depending on the lipid miscibility) may occur (Dumas et al.,
1997, Lehtonen and Kinnunen, 1997; Fahsel et al., 2002).
In this study, we focused on the process of lipid selectivity
at the protein-lipid interface. The problem of protein-lipid
selectivity quantification has been addressed mainly from
electron spin resonance (ESR) studies (see Marsh and
Horváth, 1998, for a review) or other techniques which focus
only on the protein-lipid interface, like tryptophan fluorescence
quenching by brominated phospholipids (Everett et al.,
1986; Williamson et al., 2002; O’Keeffe et al., 2000). The
results obtained from ESR studies agree well with an annular
model for protein-lipid selectivity, in which only the first
shell of lipids around the integral protein, and in direct
contact with it, is significantly disturbed by the protein
incorporation in the bilayer (Lee, 2003; Marsh and Horváth,
1998).
ESR results report the fraction of motionally restricted
lipids, whereas fluorescence collisional quenching depends
on molecular contact. On the other hand, fluorescence
resonance energy transfer (FRET) only depends on donoracceptor
distances and is an alternative technique to quantify
lipid selectivity. Gutierrez-Merino derived approximate
analytical expressions for the average rate of FRET (hk T i)
in membranes undergoing phase separation or protein
aggregation (Gutierrez-Merino, 1981a,b) and extended this
formalism to the study of protein-lipid selectivity (Gutierrez-
Merino et al., 1987). His model has proved to be useful to the
study of the lipid annulus around the oligomeric acetylcholine
receptor (Bonini et al., 2002; Antollini et al., 1996).
However, there are some limitations to the model, namely,
the simplification that underlies the formalism, which
consists of considering resonance energy transfer (RET)
only to neighboring acceptor molecules. On the other hand,
with the experimental observable being the average RET
efficiency given by
Submitted January 21, 2004, and accepted for publication March 5, 2004.
Address reprint requests to Luís M.S. Loura, Centro de Química-Física
Molecular, Complexo I, Instituto Superior Técnico, Av. Rovisco Pais,
1049-001 Lisbon, Portugal. Tel.: 35-121-841-9219; Fax: 35-121-846-4455;
E-mail: pclloura@alfa.ist.utl.pt.
Ó 2004 by the Biophysical Society
0006-3495/04/07/344/09 $2.00 doi: 10.1529/biophysj.104.040337

FRET Study of Protein-Lipid Selectivity 345
k T
hEi ¼ ; (1)
k T 1 k D
where k D is the donor intrinsic decay rate coefficient, the
relation with hk T i is not straightforward. It is proposed that if
the setting of experimental conditions is such that hEi is low
(namely, hk T i is much smaller than k D ), then hEi ffihk T i/k D
(Gutierrez-Merino, 1981a). However, accurate low RET
efficiencies are difficult to measure experimentally.
In the present work a new FRET formalism for an annular
model of protein-lipid selectivity is proposed, and used in the
quantification of M13 major coat protein selectivity toward
different phospholipids. M13 major coat protein is the main
protein component of the filamentous bacteriophage M13
with ;2800 copies. It contains a single hydrophobic transmembrane
segment of ;20 amino-acid residues, apart from
an amphipathic N-terminal arm and a heavily basic
C-terminus with a high density of lysines (for reviews see
Stopar et al., 2003; Hemminga et al., 1993).
The present study is separated in two sections. The first
section focuses on the effect of hydrophobic length, and
the selectivity of M13 toward 1,2-dioleoyl-sn-glycero-3-
phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl)
((18:1) 2 -PE-NBD) was determined in unsaturated phosphatidylcholine
bilayers of different acyl chain lengths (14:1,
18:1, and 22:1). Whereas for 18:1 chains the chain length
matches the hydrophobic length of the protein, there is
significant hydrophobic mismatch for the other lipids used.
The second part deals with specificity of M13 major coat
protein to different phospholipid headgroups, some zwitterionic
and other negatively charged. The results are compared
to the results from the other methodologies for quantification
of protein-lipid selectivity. Conclusions on the validity of the
annular model for the M13 coat protein interaction with
lipids are obtained.
MATERIALS AND METHODS
1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC; (18:1) 2 -PC), 1,2-dierucoyl-sn-glycero-3-phosphocholine
(DEuPC; (22:1) 2 -PC), 1,2-dimyristoleoyl-snglycero-3-phosphocholine
(DMoPC; (14:1) 2 -PC), 1,2-dioleoyl-sn-glycero-
3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) ((18:1) 2 -PE-
NBD), 1-Oleoyl-2-[12-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl]-sn-Glycero-Phosphocholine
(18:1-(12:0-NBD)-PC), 1-Oleoyl-2-[12-[(7-
nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl]-sn-Glycero-Phosphoethanolamine
(18:1-(12:0-NBD)-PE), 1-Oleoyl-2-[12-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl]-sn-Glycero-Phosphoserine
(18:1-(12:0-NBD)-
PS) (Sodium salt), 1-Oleoyl-2-[12-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl]-sn-Glycero-Phosphate
(18:1-(12:0-NBD)-PA) (Monosodium
salt), and 1-Oleoyl-2-[12-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl]-sn-Glycero-3-[Phospho-rac-(1-glycerol)]
(18:1-(12:0-NBD)-PG) (Sodium
salt), were obtained from Avanti Polar Lipids (Birmingham, AL).
7-diethylamino-3((4#iodoacetyl)amino)phenyl-4-methylcoumarin (DCIA)
was from Molecular Probes (Eugene, OR). Fine chemicals were obtained
from Merck (Darmstadt, Germany). All materials were used without further
purification.
Coat protein isolation and labeling
The T36C mutant of the M13 major coat protein was grown, purified from
the phage and labeled with DCIA as described previously (Spruijt et al.,
1996). For the removal of free label, DNA and other coat proteins, the
mixture was applied to a Superdex 75 prep-grade HR 16/50 column
(Pharmacia, Amersham Biosciences, Piscataway, NJ) and eluted with 50
mM sodium cholate, 150 mM NaCl, and 10 mM Tris-HCl pH 8. Fractions
with an A 280 /A 260 absorption ratio .1.5 were collected and concentrated by
Amicon filtration (Amicon, Millipore, Bedford, MA).
Coat protein reconstitution in lipid vesicles
The labeled protein mutant was reconstituted in DOPC ((18:1) 2 -PC),
DMoPC ((14:1) 2 -PC), and DEuPC ((22:1) 2 -PC) vesicles using the cholatedialysis
method (Spruijt et al., 1989). The phospholipid vesicles were
produced as follows: the chloroform from solutions containing the desired
NBD labeled and unlabeled phospholipid amount was evaporated under
a stream of dry N 2 and last traces were removed by further evaporation under
vacuum. The lipids were then solubilized in 50 mM sodium cholate buffer
(150 mM NaCl, 10 mM Tris-HCl, 1 mM EDTA) at pH 8 by brief sonication
(Branson 250 cell disruptor, Branson Ultrasonics, Danbury, CT) until a clear
opalescent solution was obtained, and then mixed with the wild-type and
labeled protein. Samples had a phospholipid concentration between 0.5 and
1 mM (phospholipid concentration was determined through the analysis of
inorganic phosphate according to McClare, 1971) and the lipid to protein
ratio (L/P) was always kept at 700. Dialysis was carried out at room
temperature and in the dark, with a 100-fold excess buffer containing
150 mM NaCl, 10 mM Tris-HCl, 1 mM EDTA at pH 8. The buffer was
replaced five times every 12 h.
Fluorescence spectroscopy
Absorption spectroscopy was carried out with a Jasco V-560 spectrophotometer
(Tokyo, Japan). The absorption of the samples was kept ,0.1 at the
wavelength used for excitation.
Steady-state fluorescence measurements were obtained with an SLM-
Aminco 8100 Series 2 spectrofluorimeter (Rochester, NY; with double
excitation and emission monochromators, MC400) in a right-angle
geometry. The light source was a 450-W Xe arc lamp and for reference
a Rhodamine B quantum counter solution was used. 5 3 5 mm quartz
cuvettes were used. All measurements were performed at room temperature.
The quantum yield of DCIA-labeled protein was determined using
quinine bisulfate dissolved in 1 N H 2 SO 4 (f ¼ 0.55; Eaton, 1988) as
a reference.
Fluorescence decay measurements of DCIA were carried out with a timecorrelated
single-photon timing system, which is described elsewhere
(Loura et al., 2000). Measurements were performed at room temperature.
Excitation and emission wavelengths were 340 and 450 nm, respectively.
The timescales used were between 3 and 12 ps/ch, depending on the amount
of NBD-labeled phospholipid present in the sample. Data analysis was
carried out using a nonlinear, least-squares iterative convolution method
based on the Marquardt algorithm (Marquardt, 1963). The goodness of the
fit was judged from the experimental x 2 value, weighted residuals, and
autocorrelation plot.
In all cases, the probe florescence decay was complex and described by
a sum of exponentials,
IðtÞ ¼+ a i expðÿt=t i Þ; (2)
i
where a i are the normalized amplitudes and t i are the fluorescence lifetimes.
Biophysical Journal 87(1) 344–352

346 Fernandes et al.
THEORETICAL BACKGROUND
Fluorescence resonance energy transfer
Fluorescence resonance energy transfer (FRET) can be used
to characterize the lateral distribution of labeled coat protein
mutants in the bilayer. In the case of energy heterotransfer,
the degree of fluorescence emission quenching of the donor
caused by the presence of acceptors is used to calculate the
experimental energy transfer efficiency (E):
E ¼ 1 ÿ t DA =t D : (3)
Here t DA is the donor lifetime-weighted quantum yield in the
presence of acceptor and t D is the donor lifetime-weighted
quantum yield in the absence of acceptor. In turn, lifetimeweighted
quantum yields are defined by Lakowicz (1999) as
The Förster radius is given by
t ¼ + a i t i : (4)
i
R 0 ¼ 0:2108ðJk 2 n ÿ4 f D Þ 1=6 ; (5)
where J is the spectral overlap integral, k 2 is the orientation
factor, n is the refractive index of the medium, and f D is the
donor quantum yield. J is calculated as
Z
J ¼ f ðlÞeðlÞl 4 dl; (6)
where f(l) is the normalized emission spectrum of the donor
and e(l) is the absorption spectrum of the acceptor. If the
l-units in Eq. 6 are nm, the calculated R 0 in Eq. 5 has Å units
(Berberan-Santos and Prieto, 1987).
The acceptors in the annular shell (Fig. 1) are at a constant
distance (d) to the coumarin fluorophore located in the center
of the transmembrane domain, and therefore we can assume
that the energy transfer to each of these acceptors is described
by the rate constant
k r ¼ 1 6
R 0
; (8)
t D d
where t D is the donor lifetime (in the absence of acceptor).
The NBD fluorophores in the acceptor probes used in this
study (phospholipids labeled with NBD in the headgroup or
in the acyl-chain) are assumed to be located in the bilayer
surface. For the chain-labeled lipids, this is justified because
the NBD group ‘‘loops up’’ to the surface when attached to
the end of the phospholipids acyl-chain (Chattopadhyay,
1990). The donor fluorophore is labeled in the M13 major
coat protein 36th residue, located near the center of the
bilayer (Spruijt et al., 1996). Therefore, to calculate d it is
necessary to estimate the average distance (l) between the
donor plane (center of bilayer) and the acceptors planes (both
leaflets), as well as the lateral separation between both probes
inside the transmembrane protein-annular shell lipids
complex. For DOPC bilayers the position of the NBD
fluorophore (l) in the derivatized phospholipids has been
calculated through the parallax method (Abrams and
London, 1993), and it was 18.9 and 19.8 Å from the bilayer
center for the phospholipids labeled at the headgroup and at
the acyl-chain, respectively. These values agree with other
studies which employed different techniques to obtain the
fluorophore position (Wolf et al., 1992; Màzeres et al.,
1996). The reason for a position of NBD closer to the surface
of the membrane while labeled at the acyl-chain is probably
the increase in flexibility that the C12 chain allows. The
Annular model for M13 coat protein selectivity
toward phospholipids
To analyze the FRET results, a model for transmembrane
protein selectivity toward phospholipids was derived. The
model assumes two populations of energy transfer acceptors,
one located in the annular shell around the protein and the
other outside it. The donor fluorescence decay curve will
have energy transfer contributions from both populations,
i DA ðtÞ ¼i D ðtÞr annular ðtÞr random ðtÞ: (7)
Here i D and i DA are the donor fluorescence decay in the
absence and presence of acceptors respectively, and r annular
and r random are the FRET contributions arising from energy
transfer to annular labeled lipids and to randomly distributed
labeled lipids outside the annular shell, respectively.
FIGURE 1 Molecular model for the FRET analysis ((A) side view; (B) top
view). Protein-lipid organization presents a hexagonal geometry. Donor
fluorophore from the mutant protein is located in the center of the bilayer,
whereas the acceptors are distributed in the bilayer surface. Two different
environments are available for the labeled lipids (acceptors), the annular
shell surrounding the protein and the bulk lipid. Energy transfer to acceptors
in direct contact with the protein has a rate coefficient dependent on the
distance between donor and annular acceptor (Eq. 8). Energy transfer toward
acceptors in the bulk lipid is given by Eq. 11 (see text for details).
Biophysical Journal 87(1) 344–352

FRET Study of Protein-Lipid Selectivity 347
lateral separation between the probes was assumed to be 8 Å.
The estimated d was then 20.5 and 21.4 Å for NBD
derivatized in the headgroup and acyl-chain, respectively.
For the DMoPC and DEuPC bilayers the values used for l
were 15.4 and 22.4 Å, respectively, as for each additional
carbon in the phospholipid chain in the liquid crystalline
phase the bilayer thickness increases 1.75 Å (Lewis and
Engelman, 1983).
Considering a hexagonal-type geometry for the proteinlipid
arrangement (Fig. 1 b), each protein will be surrounded
by 12 annular lipids. In bilayers composed by both labeled
and unlabeled phospholipids, these 12 sites will be available
for both of them. The probability (m) of one of these sites to
be occupied by labeled phospholipid is given by
m ¼ K S
n NBD
n NBD 1 n lipid
: (9)
Here, n NBD is the concentration of labeled lipid, and n lipid is
the concentration of unlabeled lipid. K S is the relative
association constant, which reports the relative affinity of
the labeled and unlabeled phospholipid. Using a binomial
distribution we can calculate the probability of each occupation
number (0–12 sites occupied simultaneously by
labeled lipid), and finally the FRET contribution arising from
energy transfer to annular lipids,
n¼12
r annular ðtÞ ¼ + e ÿnk Tt 12
m n ð1 ÿ mÞ 12ÿn : (10)
n¼0
n
The FRET contribution from energy transfer to acceptors
randomly distributed outside the annular region in two
different planes at the same distance to the donor plane (from
the center of the bilayer to both leaflets) is given by
Davenport et al. (1985) as
RESULTS
M13 coat protein selectivity toward phospholipids
with different hydrophobic thickness
The DCIA-labeled protein quantum yield was determined
(f ¼ 0.41). Using Eqs. 5 and 6, and assuming k 2 ¼ 2/3
(the isotropic dynamic limit) and n ¼ 1.4 (Davenport et al.,
1985), R 0 ¼ 39.3 Å is obtained for the DCIA-NBD FRET
pair (Fig. 2). The value k 2 ¼ 2/3 was used, because for
fluorophores in the center of a liquid crystalline bilayer, the
rotational freedom should be sufficiently high to randomize
orientations (for a detailed discussion see Loura et al., 1996).
FRET selectivity studies were performed in bilayers of
one lipid component (DOPC, DMoPC, or DEuPC) using
T36C M13 major coat protein mutant labeled with DCIA as
the donor and (18:1) 2 -PE-NBD (1,2-dioleoyl-sn-glycero-
3-phosphoethanolamine derivatized with NBD at the headgroup)
as the acceptor.
The donor fluorescence intensities ratio (t DA =t D ), which
is related to the energy transfer efficiency, decreases upon
increasing the acceptor (Eq. 3). The results are presented in
Fig. 3. The results of fitting the derived formalism to the data
are also shown in this figure, and the corresponding K S
values are summarized in Table 1.
M13 coat protein selectivity toward phospholipids
with different headgroups
Energy transfer studies were also performed to determine the
selectivity properties of M13 coat protein toward phospholipids
with different headgroups. Again, the donor was T36C
coat protein mutant labeled with coumarin (DCIA), but
various probes were used as acceptors, all studies being
made in DOPC vesicles. The probes used as acceptors were
r random
( Z pffiffiffiffiffiffiffiffi
)
l
¼ exp ÿ4n 2 pl 2 l 2 1 R 2 1 ÿ expðÿtb 3 a 6 Þ
e
a 3 da ;
0
(11)
where b ¼ðR 2 0 =lÞ2 t ÿ1=3
D ; n 2 is the acceptor density in each
leaflet, l is the distance between the plane of the donors and
the planes of acceptors, and R e is the distance between the
protein axis and the second lipid shell (exclusion distance for
bulk-located acceptors). In the present system, l is the
unlabeled lipid bilayer thickness, and the exclusion distance
is 16 Å assuming a radii of 5 Å and 4.5 Å for the protein and
the phospholipid, respectively; see Fig. 1 b). The value n 2
must be corrected for the presence of labeled lipid in the
annular region, which therefore is not part of the randomly
distributed acceptors pool.
FIGURE 2 Corrected emission spectrum of DCIA-labeled M13 major
coat protein (—), and corrected excitation spectrum of NBD-derivatized
phospholipid (---).
Biophysical Journal 87(1) 344–352

348 Fernandes et al.
FIGURE 3 Donor (DCIA-labeled protein) fluorescence quenching by energy transfer acceptor ((18:1) 2 -PE-NBD) in pure phosphatidylcholine bilayers with
different hydrophobic thickness. (d), Experimental energy transfer efficiencies; (—), theoretical simulations obtained from the annular model for protein-lipid
interaction using the fitted K S ; and (---), simulations for random distribution of acceptors (K S ¼ 1.0). (A) Labeled protein incorporated in DOPC (fitted K S ¼
1.4); (B) labeled protein incorporated in DMoPC (fitted K S ¼ 2.9); and (C) labeled protein incorporated in DEuPC (fitted K S ¼ 2.1).
phospholipids of identical acyl-chains (18:1 and 12:0) and
different headgroups (PC, PE, PS, PG, and PA) classes,
derivatized with NBD at the 12:0 chain. The results are
shown in Fig. 4, together with the model fits. Table 1
summarizes the recovered K S values.
M13 coat protein could be submitted to irreversible
aggregation. This hypothesis has been ruled out for both
lipids in recent studies (Meijer et al., 2001; Fernandes et al.,
2003). Nevertheless, M13 major coat protein was shown
recently by us to aggregate reversibly while in these con-
DISCUSSION
M13 coat protein selectivity toward phospholipids
with different hydrophobic thickness
The M13 coat protein is known to form large irreversible
aggregates under specific conditions. These aggregates are
not found in vivo, and therefore are regarded as an artifact
(Hemminga et al., 1993). Due to their b-sheet conformation,
they are detected by CD spectroscopy, and because of the
hydrophobic mismatch between the protein and DEuPC
(longer) or DMoPC (shorter) lipids, it was possible that
while incorporated in pure bilayers of these components, the
TABLE 1 Labeled phospholipids’ relative association
constants toward M13 major coat protein
Labeled phospholipid Bilayer composition K S K S /K S (PC)*
((18:1) 2 -PE-NBD) DOPC (18:1) 2 PC 1.4 —
((18:1) 2 -PE-NBD) DEuPC (22:1) 2 PC 2.1 —
((18:1) 2 -PE-NBD) DMoPC (14:1) 2 PC 2.9 —
(18:1-(12:0-NBD)-PE) DOPC 2.0 1.0
(18:1-(12:0-NBD)-PC) DOPC 2.0 1.0
(18:1-(12:0-NBD)-PG) DOPC 2.3 1.1
(18:1-(12:0-NBD)-PS) DOPC 2.7 1.3
(18:1-(12:0-NBD)-PA) DOPC 3.0 1.5
*K S (PC) is the relative association constant of (18:1-(12:0-NBD)-PC).
Biophysical Journal 87(1) 344–352

FRET Study of Protein-Lipid Selectivity 349
FIGURE 4 Donor (DCIA-labeled protein) fluorescence quenching by energy transfer acceptor (18:1-(12:0-NBD)-PX), where X stands for the different
headgroup structures, in pure bilayers of DOPC. (d), Experimental energy transfer efficiencies; (—), theoretical simulations obtained from the annular model
for protein-lipid interaction using the fitted K S ; and (---), simulations for random distribution of acceptors (K S ¼ 1.0). (A) PC-labeled phospholipid (fitted K S ¼
2.0); (B) PE-labeled phospholipid (fitted K S ¼ 2.0); (C) PG-labeled phospholipid (fitted K S ¼ 2.3); (D) PS-labeled phospholipid (fitted K S ¼ 2.7); and (E) PAlabeled
phospholipid (fitted K S ¼ 3).
ditions from a study of BODIPY labeled protein steady-state
fluorescence emission (Fernandes et al., 2003). However, in
that study much higher concentrations of protein were used
when compared with the present one, and assuming the aggregation
constants obtained in that work, only up to 5% at
the very most of protein could be aggregated at the protein
concentration used throughout this study. Therefore we can
consider that our results report the phospholipid selectivity
properties of the monomeric M13 major coat protein.
The fitting of the annular model for protein-lipid
interactions to the FRET data (Figs. 3 and 4), converged
always to K S values above 1 (Table 1), as the energy transfer
Biophysical Journal 87(1) 344–352

350 Fernandes et al.
efficiencies (1 ÿ t DA =t D ) are above the expected value for
random distribution of the labeled phospholipids. As our
annular model assumes a random distribution outside the
protein-lipid interface (which should be true for onecomponent
bilayers in the liquid crystalline phase; Loura
et al., 1996), this result is rationalized as an increase in local
concentration of probe in the lipid annular shell around the
protein. For (18:1) 2 -PE-NBD probe in DOPC ((18:1) 2 -PC)
bilayers (Fig. 3 A), the value of K S was 1.4, pointing to
almost complete randomization of the probe distribution in
the bilayer, and therefore identical selectivity to the DOPC
lipid. This was expected, because the probe acyl-chains are
identical to the unlabeled lipid and allow a perfect hydrophobic
matching of the protein.
The results from Fig. 3, A–C, all report energy transfer
efficiencies to the (18:1) 2 -PE-NBD probe, but in different
bilayers of one lipid component. In DOPC bilayers the value
of K S was 1.4 as discussed above, but in DMoPC ((14:1) 2 -
PC) and DEuPC ((22:1) 2 -PC) bilayers the relative association
constant values were 2.9 and 2.1, respectively,
confirming a greater selectivity toward the hydrophobic
matching unlabeled phospholipid (DOPC).
M13 coat protein selectivity toward phospholipids
with different headgroups
Results from analysis of the data on M13 coat protein
selectivity toward phospholipids headgroups are presented in
Fig. 4. Clearly the anionic-labeled phospholipids exhibit
larger selectivity to the lipid annular region around the
protein, especially the 18:1-(12:0-NBD)-PA and 18:1-(12:
0-NBD)-PS probes (K S ¼ 3.0 and 2.7, respectively). The
18:1-(12:0-NBD)-PG probe presents an intermediate selectivity
(K S ¼ 2.3), whereas 18:1-(12:0-NBD)-PC and 18:
1-(12:0-NBD)-PE have identical relative association constants
(K S ¼ 2.0). The selectivity for anionic phospholipids
must be a consequence of electrostatic interaction of these
with the highly basic C-terminal domain of the protein,
which contains four lysines.
Overall the selectivity of annular lipid-M13 coat protein is
not large, which is common for intrinsic membrane proteins
(Lee, 2003). Additionally, it has been shown that selectivity
of some proteins toward anionic lipids is significantly decreased
in the presence of increasing ionic strength (Marsh
and Horváth, 1998). In our case the ionic strength was kept
high, because it is necessary to keep the protein in the
monomeric state (Spruijt and Hemminga, 1991), and this
further explains our results.
Phospholipid selectivity ESR studies have already been
performed with aggregated forms of M13 major coat protein
(Peelen et al., 1992; Wolfs et al., 1989, Datema et al., 1987),
resulting in similar selectivity patterns of M13 coat protein
to phospholipid headgroups. Peelen et al. (1992), using
a identical buffer type, ionic strength, and pH to that used in
the present study, obtained the following relative association
constants ratios (K S (PX)/K S (PC)) of M13 coat protein incorporated
in 1,2-dimiristoyl-sn-glycero-3-phosphocholine
(DMPC): K S (PA)/K S (PC) ¼ 1.6, K S (PS)/K S (PC) ¼ 1.2,
K S (PG)/K S (PC) ¼ 1.1, and K S (PE)/K S (PC) ¼ 1. Overall the
selectivity pattern is the same, and the relative association
constants ratios are almost identical. The M13 coat protein
was aggregated in that study, and according to the authors
the number of first shell sites was five, that is, for each
protein only a maximum of five lipids could be motionally
restricted due to contact with the protein surface. For
a monomeric helix, however, a value of 12 should be
expected (Marsh and Horváth, 1998), and that was the
number used in our model for the data analysis. Therefore it
is particularly interesting that the ratios of the relative
association constants remain almost identical. Apparently
protein aggregation lowers the selectivity degree of each
protein for phospholipids only through a decrease of
available area for protein-lipid contacts but the relative
association ratio with phospholipids of different headgroups
remains the same. Even though the protein presents higher
selectivity for the NBD-labeled phospholipids than for
DOPC (K S (PC) ¼ 2.0) (possibly due to electrostatic
interactions with the NBD at the bilayer interface), the result
presented above clearly shows that the presence of NBD at
the bilayer interface does not change significantly the relative
association ratios of the phospholipids.
Sanders et al. (1992) were not able to determine the
selectivity of the M13 coat protein monomeric species
toward phospholipids using ESR, because the monomer was
not able to produce a sufficiently long-living boundary shell
of lipids that could be detected by ESR spectroscopy. The
fact that it was possible to clearly quantify relative association
constants using FRET in the present study presents
this technique as an alternative to ESR in protein-lipid
studies.
One important difference between the ESR and FRET
techniques is that the latter is not restricted to the lipids
adjacent to a given protein molecule. Not only labeled lipids
in the first shell of lipids will be potential acceptors to
a donor-labeled integral protein, but also the acceptors in the
other lipid shells surrounding the protein will contribute to
the final result. For that reason, this study also seems to
confirm the hypothesis of selectivity to anionic phospholipids
by the protein to largely confine itself to an annular shell
of lipids in direct contact with the protein, in the case of the
M13 major coat protein.
In case that the annular region would extend beyond this
first shell, our FRET analysis methodology (based in transfer
to a single annular shell and also to the bulk) would recover
substantially larger values for the relative association
constants. Moreover, as commented above, our recovered
K S /K S (PC) match those obtained from ESR measurements,
which only detects immobilization of annular lipids upon
incorporation of protein. The existence of a single annular
Biophysical Journal 87(1) 344–352

FRET Study of Protein-Lipid Selectivity 351
lipid layer for this protein might be related to the fact that it
has a sole transmembrane segment.
The FRET methodology has three interesting features.
First, by choosing donor-acceptor pairs with different Förster
radii it is possible to specifically study mainly the first-shell
of lipids or also the outside shells, as was the case in the
present study. The joint analysis of results coming from these
different donor-acceptor pairs could allow for an even more
detailed description of the protein-lipid arrangement in more
complex systems. In our study, the relatively large R 0 value
for the used donor-acceptor pair meant that the experimental
quenching curves shown in both Figs. 3 and 4 look similar at
first sight. Nevertheless it is impressive that the analysis
methodology is able to retrieve significant K s values. Of
course, this methodology could still be improved by the use
of a donor-acceptor pair with a smaller R 0 value, closer to the
distances under measurement. Second, the more economic
character of fluorescence studies, which requires much
smaller amounts of material than ESR, should be stressed.
And third, although this model leads to a somewhat complex
decay law (Eqs. 7, 10, and 11), it is actually not necessary to
analyze the decay curves with this law to recover the relevant
parameters, unlike in other FRET studies (e.g., of lipid phase
separation; see Loura et al., 2001). The theoretical curves are
conveniently simulated and integrated in a worksheet to
calculate the theoretical FRET efficiencies. These can be
matched to experimental values by varying the K S value (the
sole unknown parameter). The experimental FRET efficiencies
could also be obtained from steady-state data. In our
case, we obtained them from integration of donor decay
curves because these are less prone to artifacts (e.g., light
scattering, inner filter effects, measurement of absolute
intensities), which in any case, could in principle be corrected
for in a steady-state experiment.
CONCLUSIONS
In the present study FRET has been applied with success in the
characterization of the M13 major coat protein selectivity
toward phospholipids. As expected, the protein has no
significant selectivity for the lipid probe containing two
oleoyl acyl-chains while in DOPC bilayers, but exhibits larger
selectivity for the same probe while in bilayers with different
hydrophobic thickness due to hydrophobic mismatch stress.
The protein also presents larger selectivity for anionic lipids,
particularly for phosphatidic acid and phosphatidylserine
phospholipids. FRET was shown here to be a promising
methodology in protein-lipid selectivity studies.
F.F. acknowledges financial support from Fundacxão para a Ciência e
Tecnologia (FCT), project POCTI/36458/QUI/2000, and COST-European
Cooperation in the Field of Scientific and Technical Research Action D:22.
L.M.S.L. and M.P. acknowledge financial support from FCT, projects
POCTI/36458/QUI/2000, and POCTI/36389/FCB/2000. A.F. acknowledges
a research grant (BPD/11488/2002) from Programa Operacional
‘‘Ciência, Tecnologia, Inovacxão’’ (POCTI)/FCT.
REFERENCES
Abrams, F. S., and E. London. 1993. Extension of the parallax analysis of
membrane penetration depth to the polar region of model membranes:
use of fluorescence quenching by a spin-label attached to the
phospholipids polar headgroup. Biochemistry. 32:10826–10831.
Antollini, S. S., M. A. Soto, I. B. de Romanelli, C. Gutiérrez-Merino, P.
Sotomayor, and F. J. Barrantes. 1996. Biophys. J. 70:1275–1284.
Berberan-Santos, M. N., and M. J. E. Prieto. 1987. Energy transfer in
spherical geometry. Application to micelles. J. Chem. Soc. Faraday
Trans. 283:1391–1409.
Bonini, I. C., S. S. Antollini, C. Gutiérrez-Merino, and F. J. Barrantes.
2002. Sphingomyelin composition and physical asymmetries in native
acetylcholine receptor-rich membranes. Eur. Biophys. J. 31:417–427.
Chattopadhyay, A. 1990. Chemistry and biology of N-(7-nitrobenz-2-oxa-
1,3-diazol-4-yl)-labeled lipids: fluorescent probes of biological and
model membranes. Chem. Phys. Lipids. 53:1–15.
Datema, K. P., C. J. A. M. Wolfs, D. Marsh, A. Watts, and M. A.
Hemminga. 1987. Spin label electron spin resonance study of
bacteriophage M13 coat protein incorporation into mixed lipid bilayers.
Biochemistry. 26:7571–7574.
Davenport, L., R. E. Dale, R. H. Bisby, and R. B. Cundall. 1985.
Transverse location of the fluorescent probe 1,6-diphenyl-1,3,5-hexatriene
in model lipid bilayer membrane systems by resonance energy
transfer. Biochemistry. 24:4097–4108.
Dumas, F., M. C. Lebrun, and J.-F. Tocanne. 1999. Is the protein/lipid
hydrophobic matching principle relevant to membrane organization and
functions FEBS Lett. 458:271–277.
Dumas, F., M. M. Sperotto, M.-C. Lebrun, J.-F. Tocanne, and O. G.
Mouritsen. 1997. Molecular sorting of lipids by bacteriorhodopsin
in dilauroylphosphatidylcholine/distearoylphosphatidylcholine lipid bilayers.
Biophys. J. 73:1940–1953.
Eaton, D. F. 1988. Reference materials for fluorescence measurement. Pure
Appl. Chem. 60:1107–1114.
Everett, J., A. Zlotnick, J. Tennyson, and P. W. Holloway. 1986.
Fluorescence quenching of cytochrome b5 in vesicles with an
asymmetric transbilayer distribution of brominated phosphatidylcholine.
J. Biol. Chem. 261:6725–6729.
Fahsel, S., E. M. Pospiech, M. Zein, T. L. Hazlet, E. Gratton, and R.
Winter. 2002. Modulation of concentration fluctuations in phaseseparated
lipid membranes by polypeptide insertion. Biophys. J. 83:
334–344.
Fernandes, F., L. M. S. Loura, M. Prieto, R. Koehorst, R. Spruijt, and M. A.
Hemminga. 2003. Dependence of M13 major coat protein oligomerization
and lateral segregation on bilayer composition. Biophys. J. 85:2430–
2441.
Gutierrez-Merino, C. 1981a. Quantitation of the Förster energy transfer for
two-dimensional systems. I. Lateral phase separation in unilamellar
vesicles formed by binary phospholipids mixtures. Biophys. Chem.
14:247–257.
Gutierrez-Merino, C. 1981b. Quantitation of the Förster energy transfer for
two-dimensional systems. II. Protein distribution and aggregation state in
biological membranes. Biophys. Chem. 14:259–266.
Gutierrez-Merino, C., F. Munkonge, A. M. Mata, J. M. East, B. L.
Levinson, R. M. Napier, and A. G. Lee. 1987. The position of the ATP
binding site on the (Ca 21 1 Mg 21 )-ATPase. Biochim. Biophys. Acta.
897:207–216.
Hemminga, M. A., J. C. Sanders, C. J. A. M. Wolfs, and R. B. Spruijt.
1993. Lipid-protein interactions involved in bacteriophage M13 infection.
Protein Lipid Inter. New Compr. Biochem. 25:191–212.
Killian, J. A. 1998. Hydrophobic mismatch between proteins and lipids in
membranes. Biochim. Biophys. Acta. 1376:401–416.
Lakowicz, J. R. 1999. Principals of Fluorescence Spectroscopy, 2nd Ed.
Kluwer Academic/Plenum Press, New York. Chap. 8.
Lee, A. G. 2003. Lipid-protein interactions in biological membranes:
a structural perspective. Biochim. Biophys. Acta. 1612:1–40.
Biophysical Journal 87(1) 344–352

352 Fernandes et al.
Lehtonen, J. Y. A., and P. K. J. Kinnunen. 1997. Evidence for phospholipid
microdomain formation in liquid crystalline liposomes reconstituted with
Escherichia coli lactose permease. Biophys. J. 72:1247–1257.
Lewis, B. A., and D. M. Engelman. 1983. Lipid bilayer thickness varies
linearly with acyl chain length in fluid phosphatidylcholine vesicles.
J. Mol. Biol. 166:211–217.
Loura, L. M. S., A. Fedorov, and M. Prieto. 1996. Resonance energy
transfer in a model system of membranes: application to gel and liquid
crystalline phases. Biophys. J. 71:1823–1836.
Loura, L. M. S., R. F. M. de Almeida, and M. Prieto. 2001. Detection and
characterization of membrane microheterogeneity by resonance energy
transfer. J. Fluorescence. 11:197–209.
Loura, L. M. S., A. Fedorov, and M. Prieto. 2000. Membrane probe
distribution heterogeneity: a resonance energy transfer study. J. Phys.
Chem. B. 104:6920–6931.
Marquardt, D. W. 1963. An algorithm for least-squares estimation of nonlinear
parameters. J. Soc. Ind. Appl. Math. (SIAM J.). 11:431–441.
Marsh, D., and L. I. Horváth. 1998. Structure, dynamics and composition of
the lipid-protein interface. Perspectives from spin-labelling. Biochim.
Biophys. Acta. 1376:267–296.
Màzeres, S., V. Schram, J.-F. Tocanne, and A. Lopez. 1996. 7-Nitrobenz-2-
oxa-1,3-diazole-4-yl-labeled phospholipids in lipid membranes: differences
in fluorescence behavior. Biophys. J. 71:327–335.
McClare, C. 1971. An accurate and convenient organic phosphorus assay.
Anal. Biochem. 39:527–530.
Meijer, A. B., R. B. Spruijt, C. J. A. M. Wolfs, and M. A. Hemminga. 2001.
Membrane-anchoring interactions of M13 major coat protein. Biochemistry.
40:8815–8820.
Mouritsen, O. G., and M. Bloom. 1984. Mattress model of lipid-protein
interactions in membranes. Biophys. J. 46:141–153.
O’Keeffe, A. H., J. M. East, and A. G. Lee. 2000. Selectivity in lipid
binding to the bacterial outer membrane protein OmpF. Biophys. J.
79:2066–2074.
Peelen, S. J. C. J., J. C. Sanders, M. A. Hemminga, and D. Marsh. 1992.
Biochemistry. Stoichiometry, selectivity and exchange dynamics of lipidprotein
interaction with bacteriophage M13 coat protein studied by spin
label electron spin resonance. Effects of protein secondary structure.
Biochemistry. 31:2670–2677.
Sanders, J. C., M. F. Ottaviani, A. van Hoek, and A. J. W. G. Visser. 1992.
A small protein in model membranes: a time-resolved fluorescence and
ESR study on the interaction of M13 coat protein with lipid bilayers. Eur.
Biophys. J. 21:305–311.
Spruijt, R. B., and M. A. Hemminga. 1991. The in situ aggregational and
conformational state of the major coat protein of bacteriophage M13 in
phospholipid bilayers mimicking the inner membrane of host Escherichia
coli. Biochemistry. 30:11147–11154.
Spruijt, R. B., C. J. A. M. Wolfs, and M. A. Hemminga. 1989. Aggregation
related conformational change of the membrane-associated coat protein
of bacteriophage M13. Biochemistry. 28:9158–9165.
Spruijt, R. B., C. J. A. M. Wolfs, J. W. G. Verver, and M. A. Hemminga.
1996. Accessibility and environment probing using cysteine residues
introduced along the putative transmembrane domain of the major coat
protein of bacteriophage M13. Biochemistry. 35:10383–10391.
Stopar, D., R. B. Spruijt, C. J. A. M. Wolfs, and M. A. Hemminga. 2003.
Biochim. Biophys. Acta. 1611:5–15.
Williamson, I. M., S. J. Alvis, J. M. East, and A. G. Lee. 2002. Interactions
of phospholipids with the potassium channel KcsA. Biophys. J. 83:2026–
2038.
Wolf, D. E., A. P. Winiski, A. E. Ting, K. M. Bocian, and R. E. Pagano.
1992. Determination of the transbilayer distribution of fluorescent lipid
analogues by nonradiative fluorescence resonance energy transfer.
Biochemistry. 31:2865–2873.
Wolfs, C. L. A. M., L. I. Horváth, D. Marsh, A. Watts, and M. A.
Hemminga. 1989. Spin label ESR of bacteriophage M13 coat protein in
mixed lipid bilayers. Characterization of molecular selectivity of charged
phospholipids for the bacteriophage M13 coat protein in lipid bilayers.
Biochemistry. 28:995–1001.
Biophysical Journal 87(1) 344–352

BINDING OF INHIBITORS TO A PUTATIVE BINDING
DOMAIN OF V-ATPase
III
BINDING OF INHIBITORS TO A
PUTATIVE BINDING DOMAIN
OF V-ATPase
1. Introduction
Osteoporosis, a disease endemic in Western society, is characterized by a net loss of
bone mass which results from an imbalance of the bone-remodelling process, with bone
resorption exceeding bone formation. Bone demineralization involves acidification of
the isolated extracellular microenvironment. The lowering of pH is necessary both to
dissolve the bone material and to provide the acidic environment required by the
collagenases to degrade the bone matrix (Teitelbaum, 2000). The acidification is due to
proton translocation by the vacuolar-type H + -ATPases (V-ATPases) that are present in
large numbers on the ruffled border of the osteoclast while the counterions diffuse
passively through a chloride channel (Fig. III-1). Although there are a number of drugs
directed to osteoporosis treatment through indirect inhibition of osteoclasts (Lark and
James, 2002; Rodan and Martin, 2000), they allow only for a limited therapy as bone
loss is only temporarily reduced due to weak inhibition of these cells. The osteoclast V-
ATPase is, therefore, an attractive alternative target for the development of better
therapeutic agents that effectively and safely reduce osteoclast activity.
V-ATPases are large, multisubunit complexes organized in two domains (Fig. III.2)
(Nishi and Forgac, 2002; Kawasaki-Nishi et al., 2003). The V 1 domain, which is
responsible for ATP hydrolysis, comprises eight different subunits (A-H) that can be
removed from the membrane in soluble form. The V o domain, which is tightly
associated with the membrane and composed of five different subunits (a, c, c’, c’’, and
d), is responsible for proton translocation. The enzyme functions as a rotary motor
(Yokoyama et al., 2003; Harrison et al., 1997). By analogy with the mechanism
proposed for the F-ATPase, ATP hydrolysis at the catalytic sites drives rotation of the
79

central D-subunit, which in turn drives rotation of the hexameric ring of c subunits
immersed in the membrane. Subunit a is held fix relative to the A3B3 hexamer by the
peripherical stalk, which functions as a stator. Proton transport is hypothesized to occur
at the interface between the subunit a and the subunits c, c’ and c’’: rotation of the ring
of c subunits relative to subunit a, brings the buried acidic residues into sequential
contact with two hemichannels in subunit a, that provide access to each side of the
membrane, driving unidirectional proton transport (Fig. III-2) (Grabe et al., 2000).
Figure III-1: Representation of the resorbing osteoclast and acidification of the resorption lacunae
(taken from Väänanen, 2005).
Figure III-2: Structural Model of V-ATPase and proposed rotary mechanism of ATP-driven proton
translocation. The most important amino acid residues for proton translocation are marked (adapted from
Nishi and Forgac, 2002)
80

BINDING OF INHIBITORS TO A PUTATIVE BINDING
DOMAIN OF V-ATPase
In eukaryotic cells, V-ATPase, apart from being present in the plasma membrane of
osteoclasts and kidney acid secreting cells, is also found in many organelles such as
vacuoles, lysossomes, coated vesicles, Golgi, and secretory vesicles. Numerous
physiological processes depend on the activity of V-ATPases, including intracellular
targeting, protein processing, transport of metabolites, receptor-mediated endocytosis,
neurotransmitter uptake, and apoptosis (Finbow and Harrison, 1997; Nishi and Forgac,
2002).
The macrolide antibiotics bafilomycins and concanamycins are specific, highly
potent inhibitors of all eukaryotic V-ATPases in vitro and in vivo (Fig. III-3). However,
their unselective action towards V-ATPases turns them to be extremely toxic agents and
unsuitable for therapeutical use in osteoporosis treatment (Dröse and Altendorf, 1997).
Recently, novel and selective inhibitors of osteoclasts V-ATPase have been synthesized
according to structure-function relationships of bafilomycins derivatives (Gagliardi et
al., 1998a; Gagliardi et al. 1998b), suggesting that selective modulation of different V-
ATPases might be possible by this new indole class of inhibitors, namely by its most
promising representative SB 242784 (INH-3) (see Fig. III.3) (Nadler et al., 1998). SB
242784 studies with animal models of bone resorption showed a remarkable efficiency
in the prevention of ovariectomy-induced bone loss on rats, and no adverse effects of its
administration on the animals was observed. In situ studies revealed that SB 242784
inhibited osteoclastic V-ATPase activity at 1000 times smaller concentrations than
enzymes from any other tissue evaluated (Visentin et al., 2000).
Figure III-3: V-ATPase inhibitors.
81

Our understanding of the mechanism of V-ATPase inhibition is rudimentary.
Nevertheless, site-directed mutagenic studies strongly supported the hypothesis of the
location of the bafilomycin A 1 binding site in the c-subunit of V-ATPase (Bowman and
Bowman, 2002; Bowman et al., 2004). These studies indicated several amino acid
residues in the c-subunit of V-ATPase for which mutations induced resistance to
bafilomycin A 1 . The majority of these residues were located at the 4 th putative TM
helix, and four of them (L132, F136, L140, and Y143) are positioned in the same helix
side, which is expected to be exposed to the lipid environment (Harrison et al., 2000).
Some residues found in other domains of the c-subunit are apparently also relevant for
bafilomycin binding. However, it is not clear, if they are part of the bafilomycin binding
site, as a structure formed by these residues would imply a helix organization for the c-
subunit which is in disagreement with the one suggested by cysteine cross-linking
experiments (Harrison et al., 1999). Another study detected immobilization of spinlabeled
residues of c-subunit in the presence of concanamycin A, in agreement with a
binding site location on this region of V-ATPase (Páli et al., 2004).
Whyteside et al. (2005) found that SB242784 bound competitively with
concanamycin to the c-subunit of V-ATPase. However, a mutation on Y143 from the c-
subunit of V-ATPase on yeast conferred different degrees of resistance for
concanamycin and SB242784, suggesting that the binding sites for these inhibitors are
overlapping but not identical (Páli et al., 2004).
Subunit c and c´ present very similar sequences. However, mutations in subunit c´
homologous to the ones in subunit c that conferred resistance to bafilomycin have no
effect (Bowman et al., 2004). This fact, led to the proposal that either subunit c´ does
not present bafilomycin binding sites or that the affinity of the macrolide antibiotic
binding site was much smaller in subunit c´.
One strategy to resolve the problem of the binding site location for the above
mentioned inhibitors is the use of a reductionist approach. Reductionist approaches to
protein-ligand problems through use of peptides have been used in the past with success
and peptides comprising entire binding sites were already shown to present ligand
binding affinities identical to the ones observed with the intact protein (Lentz et al.,
1998; Wilson et al., 1985; Neumann et al., 1986; Taylor et al., 1995; Ried et al., 1996;
Scatigno et al., 2004).
82

INTERACTION OF THE INDOLE CLASS OF V-ATPase
INHIBITORS WITH LIPID BILAYERS
2. INTERACTION OF THE INDOLE CLASS OF
VACUOLAR H + -ATPase INHIBITORS WITH
LIPID BILAYERS
83

Biochemistry 2006, 45, 5271-5279
5271
Interaction of the Indole Class of Vacuolar H + -ATPase Inhibitors with Lipid
Bilayers †
F. Fernandes, ‡ L. Loura, ‡,§ R. B. M. Koehorst, | N. Dixon, ⊥ T. P. Kee, ⊥ M. A. Hemminga, | and M. Prieto* ,‡
Centro de Química-Física Molecular, Instituto Superior Técnico, Lisbon, Portugal, Departamento de Química, UniVersidade de
EÄVora, EÄVora, Portugal, Laboratory of Biophysics, Wageningen UniVersity, Wageningen, The Netherlands, and
Department of Chemistry, UniVersity of Leeds, Leeds LS2 9JT, United Kingdom
ReceiVed NoVember 7, 2005; ReVised Manuscript ReceiVed March 8, 2006
ABSTRACT: The selective inhibitor of osteoclastic V-ATPase (2Z,4E)-5-(5,6-dichloro-2-indolyl)-2-methoxy-
N-(1,2,2,6,6-pentamethylpiperidin-4-yl)-2,4-pentadienamide (SB 242784), member of the indole class of
V-ATPase inhibitors, is expected to target the membrane-bound domain of the enzyme. A structural study
of the interaction of this inhibitor with the lipidic environment is an essential step in the understanding
of the mechanism of inhibition. In this work, a comprehensive study of the relevant features of this
interaction was performed. Inhibitor partition coefficients to lipid vesicles as well as its transverse location,
orientation (order parameters), and dynamics while bound to bilayers were determined through
photophysical techniques, taking advantage of the intrinsic fluorescence of the molecule. To better evaluate
the functionally relevant features of SB 242784, a second inhibitor, INH-1, from the same class and
having a reduced activity was also examined. It is shown that regarding membrane interaction their
properties remain very similar for both molecules, suggesting that the differences in inhibition efficiencies
are solely a consequence of the molecular recognition processes within the inhibition site in the V-ATPase.
Bone degradation requires a lowering of pH in the
resorption lacuna (extracellular space between the osteoclast
cell and bone surface) (1). This is achieved through pumping
of protons by V-ATPases 1 located in the ruffled border of
osteoclasts (2). Inhibitors of osteoclast V-ATPase activity
could therefore be used in new therapeutics for treatment of
bone diseases related to excess bone resorption, namely,
osteoporosis.
The macrolide antibiotic bafilomycin is a powerful inhibitor
of V-ATPases (3) and was shown to prevent bone
resorption (4). Nevertheless, due to the lack of specificity
of bafilomycin toward osteoclastic V-ATPase, several other
important V-ATPases are inhibited, and this is responsible
for the extremely high toxicity and, consequently, for the
clinical inadequacy of this compound. Recently, some
†
This work was supported by Contract QLG-CT-2000-01801 of the
European Commission (MIVase Consortium) and by FCT (Portugal)
under the Program POCTI. M.A.H. and M.P. are members of the COST
D22 Action of the European Union. F.F. acknowledges Grant SFRH/
BD/14282/2003 from FCT (Portugal).
* Corresponding author. E-mail: prieto@alfa.ist.utl.pt. Telephone:
(+351) 218413248. Fax: (+351) 218464455.
‡
Instituto Superior Técnico.
§
Universidade de EÄ vora.
|
Wageningen University.
⊥
University of Leeds.
1
Abbreviations: V-ATPase, vacuolar H + -ATPase; DOPC, 1,2-
dioleoyl-sn-glycero-3-phosphocholine; DOPG, 1,2-dioleoyl-sn-glycero-
3-[phospho-rac-(1-glycerol)]; DMPC, 1,2-dimyristoyl-sn-glycero-3-
phosphocholine; 5-DOX-PC, 1-palmitoyl-2-stearoyl(5-DOXYL)-snglycero-3-phosphocholine;
12-DOX-PC, 1-palmitoyl-2-stearoyl(12-
DOXYL)-sn-glycero-3-phosphocholine; SB 242784, (2Z,4E)-5-(5,6-
dichloro-2-indolyl)-2-methoxy-N-(1,2,2,6,6-pentamethylpiperidin-4-yl)-
2,4-pentadienamide, INH-1, methyl (2Z,4E)-5-(5,6-dichloro-2-indolyl)-
2-methoxy-2,4-pentadienoate.
FIGURE 1: Chemical structures of SB 242784 and INH-1.
derivatives of bafilomycin have been synthesized which
maintained high inhibitory activity toward V-ATPases and
exhibited high selectivity for the osteoclast form of the
enzyme (5, 6). The most promising of these compounds,
(2Z,4E)-5-(5,6-dichloro-2-indolyl)-2-methoxy-N-(1,2,2,6,6-
pentamethylpiperidin-4-yl)-2,4-pentadienamide (SB 242784)
(Figure 1), was shown to have more than 1000-fold selectivity
for the osteoclast V-ATPase compared with the enzyme
measured in kidney, liver, spleen, stomach, brain, or endothelial
cells and to have no effect on other cellular ATPases.
SB 242784 was also extremely effective in preventing bone
loss in ovariectomized rats (7). This inhibitor is therefore a
promising candidate for osteoporosis treatment.
The site of action of bafilomycin in the enzyme was shown
to be located in the membrane-bound domain (8-11).
Therefore, the type and extent of the interaction of the
bafilomycin derivative SB 242784 with the lipid environment
and the properties of the inhibitor molecule while in this
medium are relevant, since it is expected that the first step
prior to the interaction with the enzyme is the membrane
10.1021/bi0522753 CCC: $33.50 © 2006 American Chemical Society
Published on Web 04/05/2006

5272 Biochemistry, Vol. 45, No. 16, 2006 Fernandes et al.
incorporation of the inhibitor with the concomitant increase
of its effective concentration.
Recently, spin-labeled derivatives of SB 242784 have been
studied by electron spin resonance (ESR) (12). It was then
possible to acquire information regarding the location in
the membrane of the piperidine ring where the nitroxide label
was inserted. However, a photophysical approach would
enable a report on the whole molecule, since the indole
ring is conjugated with the double bond system, and this is
largely expected to dictate the inhibitor properties in the lipid
phase.
In this work we report a thorough study on the interaction
of SB 242784 with lipid vesicles together with a detailed
study of the compound behavior in aqueous solution, using
for this effect its intrinsic photophysical properties (absorption,
fluorescence, fluorescence anisotropy, linear dichroism),
thus avoiding the derivatization of the molecule. The waterlipid
partition coefficient of SB 242784 was determined, and
its aggregation was studied. The position and orientation of
the molecule when interacting with lipid bilayers were also
obtained. The same studies were performed with another less
powerful inhibitor of the same class (INH-1) to compare the
results and retrieve information on the relevance of specific
structural features of SB 242784 on these properties.
MATERIALS AND METHODS
DOPC, DOPG, and DMPC were obtained from Avanti
Polar Lipids (Birmingham, AL). 5-DOX-PC and 12-DOX-
PC were from Avanti Polar Lipids (Birmingham, AL). Fine
chemicals were obtained from Merck (Darmstadt, Germany).
All materials were used without further purification.
SB 242784 and INH-1 were synthesized according to
Nadler et al. (6) and Gagliardi et al. (5).
Inhibitor Incorporation in Lipid Vesicles. The V-ATPase
inhibitors were incorporated in lipid vesicles by two different
methods: In the cosolubilization method the phospholipid
and inhibitor solutions were mixed in chloroform and dried
under a flow of dry N 2 . The last traces of solvent were
removed under vacuum. Multilamellar vesicles were then
obtained through solubilization in Tris buffer, and large
unilamellar vesicles (LUVs) were produced by extrusion
through polycarbonate filters with a pore size of 100 nm
(13). In an alternative method of incorporation, a very small
volume (

V-ATPase Indole Inhibitor Interaction with Bilayers Biochemistry, Vol. 45, No. 16, 2006 5273
The order parameters (〈P 2 〉 and 〈P 4 〉) are obtained from
the equations:
sin(ω)A ω
Α ω)(π/2)
) 1 +
3〈P 2 〉
(1 - 〈P 2 〉)n 2 cos2 ω (3)
I VH /I VV ) a sin 2 R+b (4)
A ω is the absorbance at angle ω, n is the refraction index [n
) 1.5 (21)], R is the angle at which the fluorescence is
measured, and the a and b parameters depend on both 〈P 2 〉
and 〈P 4 〉. Complete formalisms and further details are
described in Lopes et al. (22).
Dichroic ratios were determined using Glan-Thompson
polarizers. In fluorescence emission measurements for determination
of 〈P 4 〉, excitation and emission wavelengths were
kept the same as in the remaining steady-state measurements.
From the knowledge of 〈P 2 〉 and 〈P 4 〉, a combination of
the maximum entropy method together with the formalism
of the Lagrange multipliers is used to describe the single
particle distribution function f(Ψ), where Ψ is the angle
between the transition moment of the molecule and the
director of the system (normal to the bilayer plane). This
distribution is the broadest one that is compatible with the
experimental order parameters, and to obtain the population
density probability function, which is the probability of
finding the transition moment between Ψ and Ψ + dΨ, it
should be multiplied by sin Ψ (22).
The determination of the transition moment of SB 242784
was performed for an optimized geometry of the molecule
based on the density functional theory (DFT) (23), which
recovered the 2Z,4E isomer that is the preponderant species
(6). The theoretical level applied to the calculations was
Becke3LYP/6-31G(d) (24-26).
RESULTS
Indole Inhibitor BehaVior in Aqueous EnVironment. Since
the interaction with the membrane proceeds from the
inhibitors in aqueous solution, their characterization in an
aqueous environment was first carried out. Preliminary results
showed both SB 242784 and INH-1 (Figure 1) to be unstable
under aqueous condition (27). The fluorescence quantum
yield of the two molecules in aqueous buffer was extremely
low (0.01 and 0.004 for SB 242784 and INH-1, respectively)
when compared with the quantum yield observed in other
environments (e.g., 0.06 and 0.05 for SB 242784 and INH-1
in ethanol, respectively). The large decrease in fluorescence
intensity of the inhibitors is the result of formation of
nonfluorescent or “dark” aggregates, as the quantum yield
values determined from transient state fluorescence measurements
were larger than the ones obtained from steady-state
data. This aggregation is due to the hydrophobic character
of the inhibitors, and the fluorescence quenching effect is
enhanced by the presence of chlorine atoms in the molecule.
In the present work we measured the extent and kinetics of
aggregation of both inhibitors in an aqueous environment.
For this purpose, fluorescence intensity measurements were
started immediately after addition of 10 µL of a ethanol stock
solution to a cuvette containing 3 mL of buffer under
permanent stirring, and the final concentration of inhibitor
in buffer was 4 µM. The buffer used was 10 mM Tris and
FIGURE 2: SB 242784 fluorescence emission self-quenching
induced by aggregation in an aqueous environment. The buffer used
was 10 mM Tris and 150 mM NaCl, pH ) 7.4.
150 mM NaCl, pH 7.4. The result for SB 242784 is presented
in Figure 2. The kinetics for the aggregation in water of
INH-1 was very fast, and the phenomenon could not be
resolved as well as it was for SB 242784. In addition, the
extent of aggregation, as taken from the remaining fluorescence
intensity after the process reaches equilibrium, was
larger for INH-1 than for SB 242784. For both inhibitors
the fluorescence intensity increased dramatically upon addition
of sodium dodecyl sulfate to the samples (results not
shown), this suggesting disruption of aggregates and emission
of monomeric inhibitors dispersed in the micellar medium
and ruling out a significant contribution of photobleaching.
The anisotropies of INH-1 and SB 242784 in water were
0.20 and 0.22, respectively, and did not change during the
process of aggregation. This indicates that the aggregates
being formed are nonfluorescent, and the residual emission
of monomeric species is the one contributing to anisotropy.
In an attempt to determine the existence of a critical
micellar concentration (cmc) for the inhibitors in an aqueous
environment, we measured fluorescence emission intensities
for inhibitor solutions with concentrations up to 4 µM. A
cmc should appear as a change in linearity of fluorescence
intensity vs inhibitor concentration. However, a completely
linear plot was obtained up to 10 nM, the limit of detection
of the fluorometer (results not shown). Therefore, the
aggregation offset, if existing at all, should occur for both
inhibitors in water at concentrations below 10 nM, and
therefore no evidence was obtained for micellar-type aggregates
of these inhibitors.
Inhibitor Partition to Lipid Vesicles. The fluorescence
emission spectra of SB 242784 and INH-1 in the presence
of DOPC vesicles are presented in Figure 3. DOPC and
DOPG lipids were chosen for the reconstitution of the
inhibitors because SB242784 was used by us in binding
assays to selected transmembrane peptide fragments of
V-ATPase that required DOPC (and in some cases a small
fraction of DOPG) for correct peptide conformation while
inserted in liposomes (unpublished experiments). Using the
cosolubilization method for incorporation of the inhibitors
(see Materials and Methods section), the fluorescence
emission maximum for SB 242784 is 436 nm whereas for
INH-1 it is 461 nm. On the other hand, in DOPC-DOPG
(80:20 mol/mol) bilayers, the SB 242784 fluorescence

5274 Biochemistry, Vol. 45, No. 16, 2006 Fernandes et al.
FIGURE 3: Absorption and fluorescence emission spectra of the
indole class of V-ATPase inhibitors. Fluorescence emission spectra
were obtained using excitation wavelengths corresponding to the
maxima of each inhibitor: (thin dashed line) absorption spectrum
of INH-1; (thin solid line) absorption spectrum of SB 242784; (thick
dashed line) fluorescence spectrum of INH-1; (thick solid line)
fluorescence spectrum of SB 242784.
emission maximum was 432 nm. For the alternative incorporation
method of adding the inhibitor to the preformed
vesicles, the fluorescence emission maxima were slightly blue
shifted (≈3 nm) when comparing with the previous values,
but after a small period of time (≈5 min) they became
identical to the ones obtained through the cosolubilization
method, thus suggesting that the initially formed aggregates
are disrupted upon membrane interaction. The quantum
yields of SB 242784 and INH-1 determined in the presence
of 2 mM DOPC LUVs were 0.15 and 0.10, respectively.
These values are not biased by the fraction of inhibitors in
the aqueous phase, because at this lipidic concentration it
can be concluded from the determined partition coefficients
(see below) that the inhibitors are almost completely
incorporated in the membrane (>95%), and in addition, the
quantum yield in water is negligible.
Although an extremely large blue shift in the inhibitor
fluorescence is observed upon incorporation in lipid vesicles
(e.g., the maximum emission wavelength of SB 242784 in
water is 505 nm), this is not per se evidence for a large extent
of partitioning of the two molecules to the lipid environment,
because, as mentioned above, the quantum yield in the
aqueous medium is very low due to aggregation. To obtain
a quantitative description, both the extent and kinetics of
the partition of both inhibitors to DOPC vesicles were
studied. Partition coefficients were determined through
monitoring of the change in fluorescence emission intensity
(I) vs lipid concentration [L] (Figure 4) and applying the
equation (28):
I ) I w + K p γ L [L]I L
1 + K p γ L [L]
(5)
where I w is the fluorescence of the fluorophore in water,
I L is the fluorescence of the fluorophore incorporated in the
lipid vesicles, K p is the partition coefficient defined in terms
of solute concentrations in each phase, and γ L is the lipid
molar volume [0.78 dm -3 mol -1 for DOPC (29)]. The results
FIGURE 4: Increase in steady-state fluorescence emission intensity
of indole V-ATPase inhibitors with lipid concentration: (b) SB
242784 incorporated in DOPC vesicles by the cosolubilization
method; (O) INH-1 incorporated in DOPC vesicles by the method
of addition after lipid solubilization. The buffer used was 10 mM
Tris and 150 mM NaCl, pH ) 7.4. The lines represent the fits of
eq 5 to the data, and the partition coefficients are shown in Table
1.
Table 1: V-ATPase Inhibitor Partition Coefficients to Lipid
Vesicles
system
inhibitor
cosolubilization
with DOPC
cosolubilization
with DOPC-DOPG
addition to
liposomes
with DOPC
SB 242784 (1.20 ( 0.08) × 10 4 (6.64 ( 1.76) × 10 3 (1.29 ( 0.23) × 10 4
INH-1 (1.49 ( 0.26) × 10 4 (2.84 ( 0.99) × 10 4
from a nonlinear fitting of eq 5 to the experimental data are
shown in Table 1.
For SB 242784, the anisotropy obtained while adding the
inhibitor to preformed vesicles was 〈r〉 ) 0.33, the same
value being obtained through the cosolubilization method.
In vesicles composed of DOPC-DOPG (4:1 mol/mol), SB
242784 anisotropy was virtually identical (〈r〉 ) 0.32). The
INH-1 anisotropy in samples with preformed vesicles (〈r〉
) 0.24) was lower than observed for the same molecule
incorporated by cosolubilization (〈r〉 ) 0.31). Again, these
experiments were carried out at a lipidic concentration for
which the fraction of light emitted from inhibitor molecules
in water is negligible.
Inhibitor TransVerse Location in Lipid Vesicles. The
effects on the inhibitors’ fluorescence emission spectral shape
and intensity caused by the presence of nitroxide-labeled lipid
in DOPC vesicles were investigated. A 10% ratio of labeled
(5- or 12-DOX-PC) to nonlabeled lipids was used. For SB
242784, there was a very slight shift on the fluorescence
emission maximum, depending on the nitroxide-labeled lipid
present. In the absence of quenchers the fluorescence
emission maximum was at 436 nm, whereas in the presence
of 5-DOX-PC this wavelength was 435 nm. Using 12-DOX-
PC a red shift was obtained, the maximum being now 438
nm (Figure 5A). For INH-1, there was no difference in the
shape of the emission spectrum in the presence of 12-DOX-
PC (maxima were at 461 nm), but using 5-DOX-PC a very
small blue shift (2 nm) was obtained (Figure 5B). These shifts
are evidence that the two nitroxide labels, which are localized
at different depths in the membrane, are selectively quenching
inhibitor populations which lie at different transverse
positions.

V-ATPase Indole Inhibitor Interaction with Bilayers Biochemistry, Vol. 45, No. 16, 2006 5275
FIGURE 6: Orientation of the transition moment of SB 242784,
obtained from TD-DFT calculations (see text for details).
Table 2: Order Parameters and Lagrange Coefficients of the Two
Inhibitors in DOPC Multilayers Obtained from Linear Dichroism
Studies a 〈P 2〉 〈P 4〉 λ 2 λ 4
SB 242784 0.15 -0.21 0.99 -2.31
INH-1 0.18 -0.31 1.27 -4.67
a
See text for details.
FIGURE 5: Fluorescence emission spectra of indole V-ATPase
inhibitors in the absence (s) and presence of different nitroxidelabeled
lipids: (- - -) 5-DOX-PC and (‚‚‚) 12-DOX-PC). (A) SB
242784. (B) INH-1.
Quantitative information on the inhibitor location in the
membrane can be obtained via the so-called parallax method
(30) using the equation:
z cf ) L cl + (-ln(F 1 /F 2 )/πC - L 21 2 )/2L 21 (6)
where z cf is the distance of the fluorophore from the center
of the bilayer, F 1 is the fluorescence intensity in the presence
of the surface-located quencher (5-DOX-PC), F 2 is the
fluorescence intensity in the presence of the quencher located
deeper in the membrane (12-DOX-PC), L c1 is the distance
of the shallow quencher from the center of the bilayer, L c2
is the distance of the deep quencher from the center of the
bilayer, L 21 is the distance between the shallow and deep
quenchers (L c1 - L c2 ), and C is the concentration of the
quencher in molecules/Å 2 . Using the distances of the
nitroxide quenchers from the center of the bilayer given by
Abrams and London (31), the obtained position for INH-1
was of 11.9 Å from the center of the bilayer, whereas for
SB 242784 this value was 12.8 Å.
To gain further information regarding the inhibitors’
location in the membrane, acrylamide quenching assays were
also performed. Acrylamide is a hydrophilic quencher, and
therefore differences in the extent of quenching by acrylamide
allow a direct comparison of the bilayer penetration
by the fluorophores. Acrylamide quenched both inhibitors
with very similar efficiencies (results not shown), and a very
slightly increased quenching observed for INH-1 is due to
the also slight difference in average lifetimes (16) for the
two inhibitors while incorporated in DOPC vesicles (〈τ〉 INH-1
) 0.60 ns and 〈τ〉 SB 242784 ) 0.55 ns), because the Stern-
Volmer rate constant is the product between the effective
bimolecular rate constant (which can be decreased due to
fluorophore shielding from the quencher) and the intrinsic
lifetime of the fluorophore.
Inhibitor Orientation in Lipid Vesicles. From the linear
dichroism methodology, information regarding the orientation
of the transition moment is obtained. In this way, to obtain
information about the orientation of the molecule regarding
the director of the system (normal to the membrane
interface), it is necessary to know the orientation of the
transition moment relative to the molecular axes. To this
effect, a quantum chemical calculation was carried out.
Applying time-dependent density functional theory (TD-
DFT) to an energy-optimized geometry of SB 242784, the
orientation of the transition dipole moment relative to the
molecule structure was obtained. As shown in Figure 6, the
transition dipole moment is almost parallel to the molecular
axis defined by the double bond conjugated system, and an
approximately identical transition dipole moment orientation
can be considered for INH-1, as both molecules have very
similar fluorophores and the molecular differences are not
expected to largely influence this property.
The orientation parameters 〈P 2 〉 and 〈P 4 〉 and Lagrange
coefficients obtained for INH-1 and SB 242784 from the
linear dichroism measurements are presented in Table 2. The
corresponding orientational density probability functions are
shown in Figure 7.
DISCUSSION
Inhibitor Aggregation and Partition to Lipid Vesicles. It
is surprising that INH-1 appears to aggregate in aqueous
solution more readily and to a larger extent than SB 242784.
The presence of the piperidine ring in the latter was expected

5276 Biochemistry, Vol. 45, No. 16, 2006 Fernandes et al.
FIGURE 7: Orientation distribution function in DMPC of SB 242784
(- - -) and INH-1 (s).
FIGURE 8: Proposed topography and orientation of SB 242784 in
a DOPC bilayer. The molecule is depicted from the estimated most
probable angle and depth (see Discussion).
to result in a smaller hydrophilicity, but this does not seem
to be the case. Regarding the extent of aggregation, one
possibility is that the piperidine ring induces a different type
of packing in SB 242784 aggregates which quenches the
molecule fluorescence less efficiently than in INH-1 aggregates,
resulting in partially fluorescent aggregates. For
INH-1 aggregates it is reasonable to assume that these would
consist of stacked molecules with π-π interactions, but for
SB 242784, this type of π-π stacking would not be able to
exclude the piperidine ring from the aqueous environment
and another aggregation geometry would have to come into
play. The larger self-quenching of INH-1 in water would
then not be solely dictated by the extent of aggregation but
also related to a more effective geometry of packing
regarding the quenching of fluorescence. The absence of a
micellar type of aggregate for these inhibitors is most
probably related to the absence of significative amphiphilic
character.
Both inhibitors partition with a high efficiency to lipid
vesicles, and for SB 242784 the partition coefficient is
independent of the method of incorporation (Table 1).
Nevertheless, the partition to the lipid phase is only complete
at moderately high lipid concentrations (for a K p ≈ 12000,
95% of the total inhibitor population is incorporated only at
a lipid concentration of 2 mM).
For INH-1, the molecule partitions slightly more effectively
when added to the preformed liposomes, but the
differences in the partition coefficients are close to the
uncertainty of the measurements. On thermodynamic grounds
the two partition coefficients should in fact be identical upon
reaching equilibrium. However, in most situations, the one
from cosolubilization is higher than adding the solute from
the aqueous solution, this being the case among other
situations when stable aggregates (e.g., micelles) are formed
in water. The identical values observed for the two methodologies
of incorporation are consistent with the absence
of a cmc previously described. The lower value for the
partition coefficient of SB 242784 in DOPC-DOPG bilayers
indicates a decreased partition to the bilayer upon the change
in headgroup composition. This effect will be addressed in
detail below when the fluorescence emission shifts of the
inhibitors are discussed.
INH-1 anisotropy was smaller when it was added to
preformed vesicles (〈r〉 ) 0.24) when compared with the
one observed for the same molecule incorporated via
cosolubilization (〈r〉 )0.31), and this is likely to be related
to energy migration inside aggregates in the membrane. As
there is an overlap between the absorption and fluorescence
emission spectra of the inhibitors (Figure 3), energy migration
(energy homotransfer) is operative for two molecules
of inhibitor lying in close proximity as in the case of an
aggregate. The INH-1 Förster radius (R 0 )(16) for energy
migration while in lipid membranes is 11 Å (result not
shown), and therefore for INH-1 in close proximity fluorescence
depolarization is expected. The effects of energy
migration can only be observed via the fluorescence anisotropy
which is expected to decrease, since the fluorescence
lifetime remains constant in this type of photophysical
interaction. At variance with the aggregates in water, these
aggregates must be fluorescent; i.e., there is no selfquenching,
as in order to change the steady-state anisotropy
they have to contribute to a significant fraction of the total
fluorescence. This feature is probably related to the specific
geometry of the aggregate in the membrane. In addition, in
the case that quantitative dimer formation is invoked, the
intermolecular distance would be 12.7 Å as concluded from
the theoretical formalism (32); this is slightly larger than the
distance for a collisional complex, so this would mean that
no contact (static) fluorescence quenching would happen.
Possibly, INH-1, when added to preformed vesicles, either
(i) segregates to a membrane-perturbed region, and what is
being observed is not strictly aggregate formation but
increased probability of insertion in a membrane-perturbed
area, leading to a higher local monomer concentration, such
as observed for micelles (33) and membranes (34), or (ii) is
only partially aggregated (with intermolecular distance

V-ATPase Indole Inhibitor Interaction with Bilayers Biochemistry, Vol. 45, No. 16, 2006 5277
structural differences of the two molecules (presence of the
piperididine ring in SB 242784).
Inhibitor TransVerse Location and Orientation in Lipid
Vesicles. The large blue shifts observed upon the incorporation
of the inhibitors in lipid vesicles are an indication that
the inhibitors are not adsorbed to the surface of the bilayer
but are somehow buried, having some contact with the acyl
chain region. The wavelengths of maximum fluorescence
emission for both inhibitors (436 and 461 nm) are almost
the same as the ones observed for the same molecules in
acetone (27). Taking as a reference the dielectric constant
of acetone (ɛ ) 20.7), the corresponding region of PC
bilayers with the same polarity properties lies in the
headgroup region, outside the hydrocarbon core of the bilayer
(35, 36), which for DOPC can be roughly located at >14 Å
from the center of the bilayer (29). On the other hand, SB
242784 in solution only exhibited such a blue-shifted
emission when solubilized in nonprotic solvents, and as such
this inhibitor is likely not to be in contact with a high density
of water molecules when in lipid bilayers, excluding the
possibility of a strictly superficial position, and therefore a
positioning in the polar/apolar interface of the bilayer is more
likely. Although other factors such as hydrogen bonding
might affect the emission properties of the fluorophores, it
is worthwhile noting that in addition to both inhibitors having
fluorescence properties almost identical to those observed
in acetone, they are located almost at the same distance to
the center of the bilayer, as observed by acrylamide quenching
and the parallax method, the latter positioning SB 242784
and INH-1 at 12.8 and 11.9 Å from the center of the bilayer
(corresponding approximately to the position of the fourth
to third carbon of the DOPC acyl chain in the fluid state),
respectively. The difference in the position between the two
inhibitors is

5278 Biochemistry, Vol. 45, No. 16, 2006 Fernandes et al.
wavelengths. At variance, the fraction of molecules deeper
inside the membrane is emitting more in the blue. This would
explain the observed shifts in fluorescence emission maximum
in the presence of quenchers at different depths in the
bilayer for SB 242784. The tilted configuration of SB
242784, and the resulting poor packing with the bilayer
lipids, can be the explanation for the perturbations on lipid
mobility induced by the presence of this V-ATPase inhibitor
in the bilayer as detected by ESR (40).
The orientational probability density function (Figure 7)
of SB 242784 has a small contribution near 90°. This
contribution is very likely to be spurious and probably results
from the data analysis method (15), as it seems unlikely that
a population of the inhibitors lies parallel to the bilayer
surface in view of the other results presented in this study,
and bilayer defects that could induce a closer to random
orientation of the molecule were absent in the INH-1 samples
in the same lipid system. On the other hand, the contribution
to the orientational probability density function of SB 242784
for smaller angles relative to the bilayer normal could
originate from a fraction of SB 242784 presenting orientations
closer to the bilayer normal, due to the presence of the
piperidine ring, which, as already mentioned, is expected to
prefer the hydrocarbon core of the bilayer and therefore to
dislocate the inhibitor toward the bilayer center.
CONCLUSIONS
Summarizing our results, both inhibitors readily aggregate
in an aqueous environment and present relatively large
partition coefficients to lipid bilayers. It is reasonable to place
the fluorophore section of the inhibitors close to the polar/
apolar interface of the bilayer, in contact with the headgroup
region but still protected from the aqueous environment. The
conjugated double bonds are likely to be in contact with the
first carbons of the lipid acyl chains, and for the SB 242784
V-ATPase inhibitor the piperidine ring should insert deeper
among the acyl chains. The localization of the molecules in
bilayers seems to be almost completely dictated by the
fluorophore moiety, shared to a large extent by the two
molecules. In this way, and assuming that the first step is
the inhibitor incorporation in the membrane followed by
diffusion to the protein, it would be likely that the site for
interaction in the protein would be at a similar position inside
the membrane to maximize the diffusional encounters.
The two inhibitors studied here are samples of the entire
indole class of V-ATPase inhibitors. SB 242784 exhibits very
high inhibitory activities against osteoclastic V-ATPase, with
efficient inhibition in nanomolar concentrations, whereas
INH-1 is capable of inhibition of the enzyme only in
micromolar concentrations (5-7). According to our results,
it can be inferred that this high difference in inhibitory
efficiency inside the class of V-ATPase indole inhibitors is
essentially due to specific molecular recognition processes
with the enzyme inhibition site. The behavior in membranes
of the two molecules is very similar, namely, its effective
concentration in the membrane, since the partition coefficients
do not vary significantly. This strong discrimination
by the enzyme reflected on their IC 50 values (30 µM and 26
nM for INH-1 and SB 242784, respectively) is interesting,
considering the close similarity of the two molecules.
Experiments are under way to study the interactions of
V-ATPase inhibitors with selected transmembrane segments
of the enzyme to gain further insight into the proteininhibitor
recognition process.
ACKNOWLEDGMENT
We thank Benedito Cabral for the TD-DFT calculations
with SB 242784.
REFERENCES
1. Baron, R., Neff, L., Louvard, D., and Courtoy, P. J. (1986) Cellmediated
extracellular acidification and bone resorption: evidence
for a low pH in resorbing lacunae and localization of a 100-kD
lysosomal membrane protein at the osteoclast ruffled border, J.
Cell Biol. 101, 2210-2222.
2. Blair, H. C., Teitelbaum, S. L., Ghiselli, R., and Gluck, S. (1989)
Osteoclastic bone resorption by a polarized vacuolar proton pump,
Science 245, 855-857.
3. Bowman, E. J., Siebers, A., and Altendorf, K. (1988) Bafilomycins:
a class of inhibitors of membrane ATPases from microorganisms,
animal cells, and plant cells, Proc. Natl. Acad. Sci. U.S.A.
85, 7972-7976.
4. Sundquist, K., Lakkakorpi, P., Wallmark, B., and Väänänen, K.
(1990) Inhibition of osteoclast proton transport by bafilomycin
A1 abolishes bone resorption, Biochem. Biophys. Res. Commun.
168, 309-313.
5. Gagliardi, S., Nadler, G., Consolandi, E., Parini, C., Morvan, M.,
Legave, M. N., Belfiore, P., Zocchetti A., Clarke, G. D., James,
I., Nambi, P., Gowen, M., and Farina, C. (1998) 5-(5,6-Dichloro-
2-indolyl)-2-methoxy-2,4-pentadienamides: Novel and selective
inhibitors of the vacuolar H + -ATPase of osteoclasts with bone
antiresorptive activity, J. Med. Chem. 41, 1568-1573.
6. Nadler, G., Morvan, M., Delimoge, I., Belfiore, P., Zocchetti, A.,
James, I., Zembryki, D., Lee-Rycakzewski, E., Parini, C., Consolandi,
E., Gagliardi, S., and Farina, C. (1998) (2Z,4E)-5-(5,6-
Dichoro-2-indolyl)-2-methoxy-N-(1,2,2,6,6-pentamethylpiperidin-
4-yl)-2,4-pentadienamide, a novel, potent and selective inhibitor
of the osteoclast V-ATPase, Bioorg. Med. Chem. Lett. 8, 3621-
3626.
7. Visentin, L., Dodds, R. A., Valente, M., Misiano, P., Bradbeer, J.
N., Oneta, S., Liang, X., Gowen, M., and Farina, C. (2000) A
selective inhibitor of the osteoclastic V-H + -ATPase prevents bone
loss in both thyroparathyroidectomized and ovariectomized rats,
J. Clin. InVest. 106, 309-318.
8. Mattsson, J. P., and Keeling, D. J. (1996) [ 3 H]Bafilomycin as a
probe for the transmembrane proton channel of the osteoclast
vacuolar H(+)-ATPase, Biochim. Biophys. Acta 1280, 98-106.
9. Zhang, J., Feng, Y., and Forgac, M. (1994) Proton conduction
and bafilomycin binding by the V o domain of the coated vesicle
V-ATPase, J. Biol. Chem. 269, 23518.
10. Crider, B. P., Xie, X. S., and Stone, D. K. (1994) Bafilomycin
inhibits proton flow through the H + channel of vacuolar proton
pumps, J. Biol. Chem. 269, 17379-17381.
11. Wang, Y., Inoue, T., and Forgac, M. (2005) Subunit a of the yeast
V-ATPase participates in binding of bafilomycin, J. Biol. Chem.
280, 40481-40488.
12. Dixon, N., Páli, T., Kee, T. P., and Marsh, D. (2004) Spin labelled
V-ATPase inhibitors in lipid membranes, Biochim. Biophys. Acta
1665, 177-183.
13. Mayer, L. D., Hope, M. J., and Cullis, P. R. (1986) Vesicles of
variable sizes produced by a rapid extrusion procedure, Biochim.
Biophys. Acta 858, 161-168.
14. McClare, C. (1971) An accurate and convenient organic phosphorus
assay, Anal. Biochem. 39, 527-530.
15. Castanho, M. A. R. B., Lopes, S., and Fernandes, M. (2003) Using
UV-Vis. linear dichroism to study the orientation of molecular
probes and biomolecules in lipidic membranes, Spectroscopy 17,
377-398.
16. Lakowicz, J. R. (1999) Principles of Fluorescence Spectroscopy,
2nd ed., Kluwer Academic/Plenum Publishers, New York.
17. Chen, R., and Bowman, R. L. (1965) Fluorescence polarization:
measurements with ultraviolet-polarizing filters in a spectrophotofluorometer,
Science 147, 729-732.
18. Eaton, D. F. (1988) Reference materials for fluorescence measurement,
Pure Appl. Chem. 60, 1107-1114.

V-ATPase Indole Inhibitor Interaction with Bilayers Biochemistry, Vol. 45, No. 16, 2006 5279
19. Loura, L. M. S., Fedorov, A., and Prieto, M. (2000) Membrane
probe distribution heterogeneity: a resonance energy transfer
study, J. Phys. Chem. B 104, 6920-6931.
20. Marquardt, D. W. (1963) An algorithm for least squares estimation
of non linear parameters, J. Soc. Ind. Appl. Math. 11, 431-441.
21. Kooyman, R. P. H., Levine, Y. K., and Van der Meer, B. W. (1981)
Measurement of second and fourth rank order parameters by
fluorescence polarization experiments in a lipid membrane system,
Chem. Phys. 60, 317-326.
22. Lopes, S., and Castanho, M. A. R. B. (2002) Revealing the
orientation of nystatin and amphotericin B in lipidic multibilayers
by UV-Vis linear dichroism, J. Phys. Chem. B 106, 7278-7282.
23. Parr, R. G., and Yang, W. (1989) Density Functional Theory of
Atoms and Molecules, Oxford University Press, Oxford.
24. Becke, A. D. (1993) Density-functional thermochemistry. 3. The
role of exact exchange, J. Chem. Phys. 98, 5648.
25. Lee, C., Yang, W., and Parr, R. G. (1988) Development of the
Colle-Salvetti correlation-energy formula into a functional of the
electron density, Phys. ReV. 37, 785-789.
26. Hariharan P. C., and Pople, J. A. (1974) Accuracy of AHn,
equilibrium geometries by single determinant molecular orbital
theory, Mol. Phys. 27, 209.
27. Fernandes, F., Loura, L. M. S., Koehorst, R., Hemminga, M., and
Prieto, M. (2004) Interaction of the indole class of V-ATPase
inhibitors with lipid bilayers, Biophys. J. 88, 240a.
28. Santos, N. C., Prieto, M., and Castanho, M. A. R. B. (2003)
Quantifying molecular partition into model systems of biomembranes:
an emphasis on optical spectroscopic methods, Biochim.
Biophys. Acta 1612, 123-135.
29. Tristram-Nagle, S., Petrache, H. I., and Nagle, J. F. (1998)
Structure and interactions of fully hydrated dioleoylphosphatidylcholine
bilayers, 75, 917-925.
30. Chattopadhyay, A., and London, E. (1987) Parallax method for
direct measurement of membrane penetration depth utilizing
fluorescence quenching by spin-labeled phospholipids, Biochemistry
26, 39-45.
31. Abrams, F. S., and London, E. (1992) Calibration of the parallax
fluorescence quenching method for determination of membrane
penetration depth: refinement and comparison of quenching by
spin-labeled and brominated lipids, Biochemistry 31, 5312-5322.
32. Runnels, L. W., and Scarlata, S. F. (1995) Theory and application
of fluorescence homotransfer to melittin oligomerization, Biophys.
J. 69, 1569-1583.
33. Berberan-Santos, M. N., and Prieto, M. J. E. (1987) Energy transfer
in spherical geometry. Application to micelles, J. Chem. Soc.,
Faraday Trans. 83, 1391-1409.
34. de Almeida, R. F. M., Loura, L. M. S., Prieto, M., Watts, A.,
Fedorov, A., and Barrantes, F. J. (2004) Cholesterol modulates
the organization of the M4 transmembrane domain of the muscle
nicotinic acetylcholine receptor, Biophys. J. 86, 2261-2272.
35. Koehorst, R. B. M., Spruijt, R. B., Vergeldt, F. J., and Hemminga,
M. A. (2004) Lipid bilayer topology of the transmembrane alfahelix
of M13 major coat protein and bilayer polarity profile by
site-directed fluorescence spectroscopy, Biophys. J. 87, 1445-
1455.
36. White, S. H., and Wimley, W. C. (1998) Hydrophobic interactions
of peptides with membrane interfaces, Biochim. Biophys. Acta
1376, 339-352.
37. Contreras, L. M., de Almeida, R. F. M., Villalaín, J., Fedorov,
A., and Prieto, M. (2001) Interaction of -melanocyte stimulating
hormone with binary phospholipid membranes: structural changes
and relevance of phase behaviour, Biophys. J. 80, 2273-2283.
38. Yau, W., Wimley, W. C., Gawrisch, K., and White, S. H. (1998)
The Preference of tryptophan for membrane interfaces, Biochemistry
37, 14713-14718.
39. de Planque, M. R. R., Kruijtzer, J. A. W., Liskamp, R. M. J.,
Marsh, D., Greathouse, D. V., Koeppe II, de Kruijff, B., and
Killian, J. A. (1999) Different membrane anchoring positions of
tryptophans and lysine in synthetic transmembrane alpha-helical
peptides, J. Biol. Chem. 274, 20839-20846.
40. Páli, T., Dixon, N., Kee, T. P., and Marsh, D. (2004) Incorporation
of the V-ATPase inhibitors concanamycin and indole pentadiene
in lipid membranes. Spin-label EPR studies, Biochim. Biophys.
Acta 1663, 14-18.
BI0522753

BINDING ASSAYS OF INHIBITORS TOWARDS SELECTED
V-ATPase DOMAINS
3. BINDING ASSAYS OF INHIBITORS TOWARDS
SELECTED V-ATPase DOMAINS
95

Biochimica et Biophysica Acta 1758 (2006) 1777–1786
www.elsevier.com/locate/bbamem
Binding assays of inhibitors towards selected V-ATPase domains
F. Fernandes a , L.M.S. Loura a,b, ⁎ , A. Fedorov a , N. Dixon c , T.P. Kee c ,
M. Prieto a , M.A. Hemminga d
a Centro de Química-Física Molecular, Instituto Superior Técnico, Lisbon, Portugal
b Departamento de Química, Universidade de Évora, Évora, Portugal
c Department of Chemistry, University of Leeds, Leeds LS2 9JT, UK
d Laboratory of Biophysics, Wageningen University, Wageningen, The Netherlands
Received 14 March 2006; received in revised form 4 May 2006; accepted 13 July 2006
Available online 21 July 2006
Abstract
The macrolide antibiotic bafilomycin and the related synthetic compound SB 242784 are potent inhibitors of the vacuolar H + -ATPases (V-
ATPase). It is currently believed that the site of action of these inhibitors is located on the membrane bound c-subunits of V-ATPases. To address
the identification of the critical inhibitors binding domain, their specific binding to a synthetic peptide corresponding to the putative 4th
transmembrane segment of the c-subunit was investigated using fluorescence resonance energy transfer (FRET), and for this purpose a specific
formalism was derived. Another peptide of the corresponding domain of the c′ isoform, was checked for binding of bafilomycin, since it is not
clear if V-ATPase inhibition can also be achieved by interaction of the inhibitor with the c′-subunit. It was concluded that bafilomycin binds to the
selected peptides, whereas SB 242784 was unable to interact, and in addition for bafilomycin, its interaction with the peptides either corresponding
to the c- or the c′-subunit isoforms is identical. Since the observed interactions are however much weaker as compared to the very efficient binding
of both bafilomycin and SB 242784 to the whole protein, it can be concluded that assembly of all V-ATPase transmembrane segments is required
for an efficient interaction.
© 2006 Elsevier B.V. All rights reserved.
Keywords: V-ATPase; Bafilomycin A 1 ; Fluorescence; FRET
1. Introduction
Vacuolar ATPases (V-ATPases) are responsible for proton
pumping in acidic organelles and plasma membranes of
eukaryotic cells and their activity is essential in a large variety
of cellular processes [1,2]. V-ATPases are composed of two
domains (Fig. 1A), a soluble domain (V 1 ), where the catalytic
center for ATP hydrolysis is located, and a membrane bound
domain (V 0 ), responsible for proton pumping. V 0 contains
multiple subunits including the 16-kDa proteolipids (subunits c)
(4–5 copies) and their isoforms c′ and c″ (one copy each) [3],
expected to play a major role in proton translocation. The
⁎ Corresponding author. Centro de Química-Física Molecular, Complexo I,
Instituto Superior Técnico, Lisbon, Portugal.
E-mail address: pclloura@alfa.ist.utl.pt (L.M.S. Loura).
c-subunit contains 4 putative transmembrane domains, and an
essential glutamate residue located in the 4th putative
transmembrane sequence is expected to be the carrier of
protons across the membrane through protonation/deprotonation
events [4].
Through site-directed mutagenic studies it was reported that
mutations in three residues of the c-subunit of the V-ATPase
from Neurospora crassa conferred resistance to bafilomycin A 1
(Fig. 1B) [5], a macrolide antibiotic inhibitor of V-ATPases
produced by Streptomyces griseus [6]. Two of these residues
reside in the same side of the putative 4th transmembrane helix
(F136 and Y143), and this helix side is exposed to the lipid
environment [7]. Immobilization of spin-labeled residues of c-
subunit by concanamycin A, a macrolide antibiotic very similar
in structure to bafilomycin A 1 , has also been observed [8]. Also
recently, six other sites where mutations induced resistance to
bafilomycin A 1 were reported, and three of these are located in
0005-2736/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.bbamem.2006.07.006

1778 F. Fernandes et al. / Biochimica et Biophysica Acta 1758 (2006) 1777–1786
Fig. 1. (A) V-ATPase structure (adapted from [2]). (B) V-ATPase inhibitors SB 242784 and bafilomycin A 1 .
the putative 4th transmembrane segment of c-subunit [9]. Of the
latter, two residues are located in the same helix side as F136
and Y143, pointing to the existence of an inhibitor binding site
on this protein surface. It is not clear yet if residues from other
helices (1 and 2) could also be part of this binding site and form
a binding pocket [9], as this structure would imply a different
helix organization than the one suggested from cysteine crosslinking
experiments [10].
The exact location and structure of the inhibitor binding site
remains a matter of discussion, and in the present study we aim
to detect the molecular requirements for bafilomycin A 1 binding
through the use of a reductionist approach, using a simpler
model system, a synthetic c-subunit derived peptide expected to
mimic the putative 4th transmembrane segment of the V-
ATPase c-subunit (Fig. 2). Synthetic peptides have proven to be
useful tools in the characterization of ligand binding-sites on
proteins and in some cases these isolated peptides were capable
of ligand binding efficiencies identical to the intact protein [11–
16]. Additionally, peptide fragments corresponding to transmembrane
sequences in intact membrane proteins were shown
to specifically disrupt the intact protein activity (likely by
competing with native transmembrane domains) [17], and
functional membrane proteins have already been reformed from
reconstituted isolated transmembrane segments [18–20], proving
that isolated peptides corresponding to transmembrane
fragments are able to maintain specific information even when
incorporated in the absence of its neighboring transmembrane
Fig. 2. Primary sequence of the model peptides. The sequences correspond to the
putative 4th transmembrane segment of the V-ATPase c-(H4) and c′-(H4 c′ )
subunit from Saccharomyces cerevisiae. Peptide H4 A ← E corresponds to a
modification of peptide H4 in which the glutamate residue was replaced by an
alanine. This mutation is underlined. The peptides were flanked by lysines to
increase the membrane anchoring.
segments. Similar to a previous study on synthetic peptides that
mimic the 7th putative helix of the Vph1p subunit of yeast V-
ATPase [21], flanking lysines were included in the design of the
model peptides used in the present study to stabilize the
transmembrane orientation due to the anchoring effect of the
lysine residue.
Another V-ATPase inhibitor, (2Z,4E)-5-(5,6-dichloro-2-
indolyl)-2-methoxy-N-(1,2,2,6,6-pentamethylpiperidin-4-yl)-
2,4-pentadienamide (SB 242784) (Fig. 1B), a synthetic
molecule developed by Farina and coworkers [22] which is
based on the presumptive pharmacophore of bafilomycin A 1
[23], and whose mechanism of inhibition is therefore expected
to be similar, was proposed as an agent for a new therapeutical
strategy in osteoporosis disease due to its specificity for the
osteoclastic form of the enzyme [22,24]. Bafilomycines are
powerful toxins, since they do not present specificity for any
particular type of V-ATPases, inhibiting indiscriminately all
enzymes, and thus preventing their therapeutical use. The
reason for the difference in selectivity between the two
inhibitors is still unclear, especially when noting that the
inhibitor acts on the c-subunit that is extremely conserved
between c-subunits from enzymes of different origins. A
possible explanation would be the participation of the a-subunit
(part of the enzyme membrane bound domain—V 0 )(Fig. 1A) in
the inhibitory mechanism, since the isoform a3 is expressed
primarily in osteoclasts [25], and recent mutational studies
support this hypothesis [26]. However, in a recent study using
fluorescence resonance energy transfer (FRET), it was reported
that SB 242784 and concanamycin bound competitively to
isolated c-subunits, and SB 242784 was shown to bind to these
as strongly as to the entire V 0 domain, apparently excluding a
contribution of the a-subunit to the inhibitor binding mechanism
[27]. The same authors also detected binding efficiencies of c-
subunit for SB 242784 consistent with the inhibitor′s IC 50
values, whereas for concanamycin these were slightly lower
than expected [27]. Mutation of the tyrosine residue (Y143)
from the 4th putative transmembrane segment had different
effects on the conferred degree of resistance of yeast to
concanamycin and to SB 242784, even though both were

F. Fernandes et al. / Biochimica et Biophysica Acta 1758 (2006) 1777–1786
1779
increased [8]. The binding sites for both inhibitors seem, for this
reason, to be overlapping but not identical.
Subunits c and c′ present a very large degree of sequence
similarity, but mutations in subunit c′ homologous to those in
subunit c producing bafilomycin A 1 insensitive strains have no
effect in bafilomycin resistance [9]. On the basis of these results,
Bowman and coworkers suggested that if c-subunit isoforms
possess bafilomycin-binding sites, then these should have lower
inhibitor affinities than in the c-subunit.
Using the UV absorption properties of bafilomycin A 1
(acceptor), we applied FRET in an assay (Tyr is the donor) to
check for the existence of binding between the inhibitor and
synthetic peptides corresponding to the putative 4th transmembrane
segment of the c and c′-subunits from the V-
ATPase of Saccharomyces cerevisiae in a lipidic environment.
The same methodology was applied in binding assays with
SB 242784. We detected binding of bafilomycin A 1 to the
selected peptides, while for SB 242784 no interaction took
place.
2. Materials
1,2-Dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-
nitro-2-1,3-benzoxadiazol-4-yl) (NBD-DOPE), 1,2-Dioleoylsn-glycero-3-phosphocholine
(DOPC), 1,2-Dioleoyl-sn-glycero-3-[Phospho-rac-(1-glycerol)]
(DOPG), 1,2-Dierucoyl-snglycero-3-phosphocholine
(DEuPC), 2-Lauroyl-sn-glycero-3-
phosphocholine (DLPC) and 1,2-Miristoyl-sn-glycero-3-phosphocholine
(DMPC) were obtained from Avanti Polar Lipids
(Birmingham, AL). (2Z,4E)-5-(5,6-dichloro-2-indolyl)-2-methoxy-N-(1,2,2,6,6-pentamethylpiperidin-4-yl)-2,4-pentadienamide
(SB 242784) was synthesized as described elsewhere [22].
Bafilomycin A 1 was obtained from LC Laboratories (Woburn,
MA). Peptides H4, H4 A ← E , and H4c′ (see Fig. 2) were
synthesized by Pepceuticals (Nottingham, UK). Trifluoroacetic
acid (TFA), 16-DOXYL-stearic acid (2-(14-Carboxytetradecyl)-2-ethyl-4,4-dimethyl-3-oxazolidinyloxy)
and 5-DOXYLstearic
acid (2-(3-Carboxypropyl)-4,4-dimethyl-2-tridecyl-3-
oxazolidinyloxy) were obtained from Sigma-Aldrich (St.
Louis, USA). Trifluoroethanol (TFE) was obtained from
Acrōs Organics (Geel, Belgium). Other fine chemicals were
obtained from Merck (Darmstadt, Germany).
3. Experimental procedures
3.1. Sample preparation
The peptides (Fig. 2) were solubilized in 100 μl of TFA and immediately
dried under a N 2(g) flow. Following that, the peptide was suspended in TFE.
When the TFA solubilization step was not introduced, the solubility in TFE was
greatly reduced, and the peptide aggregation levels after reconstitution in lipid
bilayers were enhanced.
For peptide reconstitution in lipid bilayers, the desired amount of
phospholipids and solubilized peptide (and of the inhibitors in the inhibitor
binding studies) were mixed in chloroform and dried under a N 2(g) flow. The
sample was then kept in vacuum overnight. Liposomes were prepared with
buffer Tris 10 mM pH 7.4. The hydration step was performed with gentle
addition of buffer at a temperature above the phospholipid main transition
temperature.
For the fluorescence quenching experiments with the N-DOXYL-stearic
acids, the nitroxide labeled fatty acids were 10% of the total lipid (molar
fraction).
3.2. CD measurements
CD spectroscopy was performed on a Jasco J-720 spectropolarimeter with a
450 W Xe lamp. Samples for CD spectroscopy were extruded 8 times on a
homemade extruder using polycarbonate filters of 0.1 μm. Peptide concentration
was always 40 μM.
3.3. Fluorescence spectroscopy
Steady-state fluorescence measurements were carried out with an SLM-
Aminco 8100 Series 2 spectrofluorimeter (with double excitation and emission
monochromators, MC400) in a right angle geometry. The light source was a
450 W Xe arc lamp and for reference a Rhodamine B quantum counter solution
was used. Correction of emission spectra was performed using the correction
software of the apparatus. 5 ×5 mm quartz cuvettes were used. All
measurements were performed at room temperature.
The emission spectrum from the tyrosine of peptide H4 was recorded
using an excitation wavelength of 270 nm. The tyrosine quantum yield of
peptide H4 in lipid bilayers was determined using quinine sulfate (ϕ=0.55)
as a reference [28].
4. Results
4.1. Characterization of reconstituted peptides
When using a peptide as a model for a protein section, it is important that it retains the structural properties of the latter [21].
Therefore, it was important for the present study to accurately know the behavior of the peptides when incorporated in lipid
bilayers. The characterization focused on secondary structure, aggregation and position in the bilayer. The 4th transmembrane
segment from which the peptide is derived is expected to assume a α-helical structure inside the V-ATPase subunit c, and
therefore conditions that favor this type of structure should be determined and used throughout the study.
CD spectra of peptide H4 incorporated in DOPC at a lipid to protein ratio (L/P-molar ratio) of 50 revealed the presence of
different types of secondary structures (Fig. 3). Deconvolution of the spectrum with the CDNN CD Spectra Deconvolution v.
2.1 software recovered a fraction of about 40% of α-helical structure, 29% was reported to be random coil, while the
remaining was divided between the different types of β-sheet structure. When a L/P ratio of 25 was used, the fraction of α-
helical structure present decreased to 30.5%. The presence of increasing amounts of β-sheet structure at lower L/P ratios is an
indication of peptide aggregation [29]. It was not possible to obtain a reasonable quality CD spectrum with samples of larger
L/P ratios as those used in the inhibitor binding studies, as the lipid concentration became too high and large light scattering
problems arose.

1780 F. Fernandes et al. / Biochimica et Biophysica Acta 1758 (2006) 1777–1786
Fig. 3. CD spectrum of peptide H4 incorporated in DOPC at an L/P of 50.
Another method to study aggregation of peptides is to follow the quantum yield changes of the aromatic residues present.
Fluorescence from tyrosine residues is known to be quenched by several mechanisms besides FRET, namely interactions with other
residues side chains such as glutamate and aspartate, and uncharged forms of basic residues such as arginine, lysine and histidine, as
well as electron transfer to peptide α-carbonyl groups [30,31]. Considering this, it can be expected that peptides forming aggregates
will have a lower quantum yield than their monomeric counterparts. The quantum yield dependence of Tyr15 (corresponding to
Tyr142 in the native protein [32]) from reconstituted peptides on the lipid composition is shown in Fig. 4. The presence of anionic
phospholipids clearly increases the quantum yield of peptide H4. In liposome of pure zwitterionic phospholipids, the Tyr15 quantum
yield trend is: DEuPC< DOPC

F. Fernandes et al. / Biochimica et Biophysica Acta 1758 (2006) 1777–1786
1781
Fig. 5. (A): Model used for FRET simulations. The donor (Tyr15) is located 6 Å from the center of the bilayer while the acceptors can be located at different planes,
assuming a superficial location such as for NBD-DOPE (A 1 ) or more buried positions such as for the V-ATPase inhibitors (A 2 ). In case of aggregation the number of
protein–protein contacts increases while protein–lipid contacts decrease, as a result, R e (exclusion distance) increases. (B) H4 fluorescence quenching of Tyr15 of
peptide H4 due to FRET to NBD-DOPE in DOPC bilayers using a lipid to protein ratio of 50 (●) and 100 (○). Curve corresponds to an exclusion radius fit to the energy
transfer data using equations 1–5 (—). A value of 9 Å was recovered for R e .(–––) FRET simulation for a random distribution of acceptors around a donor with an
exclusion radius of 20 Å. n 2 is the superficial concentration of labeled phospholipids (molecules per Å 2 ). The range of NBD labeled phospholipid to total phospholipid
ratios in this experiment was 0.4% to 2%. Inset: Overlap between tyrosine fluorescence emission ( ) and DOPE-NBD absorption spectrum (––), R 0 (Tyr-NBD) =22Å.
On the basis of these results, the lipid composition chosen for the inhibitor-peptide binding studies by FRET measurements was
100% DOPC at an L/P of 100, as this lipid is able to favor the transmembrane orientation for peptide H4. The problem of lateral
aggregation in the membrane nevertheless potentially remained, as in DOPC the tyrosine quantum yield was significantly lower
than in liposomes containing anionic phospholipids. The possibility of aggregation could however also be assessed, as described
below.
A FRET experiment was performed in order to determine the size of the possible aggregates. Using the Tyr15 as a donor and
NBD labeled DOPE as acceptors (NBD-DOPE), FRET efficiencies were obtained (Fig. 5B). Fitting the available theoretical models
for energy transfer between donor and acceptors in different planes [34] to our experimental data, we can recover an averaged
exclusion radius (R e ) of the donor, defined as the minimum distance between the Tyr15 and the NBD labeled phospholipids (Fig.
5A). For a monomeric peptide this value should be around 10 Å as this is approximately the sum of the radius of a α-helix backbone
and that of a phospholipid molecule. Any significant increase from this value is likely to be reporting an extensive aggregation
phenomenon.
FRET efficiencies (E) are calculated from the degree of fluorescence emission quenching of the donor caused by the presence of
acceptors.
E ¼ 1
I DA
I D
¼ 1
Z l
0
Z l
i DA ðÞdt= t i
0 D ðÞdt t
ð1Þ
i DA ðtÞ ¼i D ðtÞ:q interplanar ðtÞ
ð2Þ
where I DA and I D are the steady-state fluorescence intensities of the donor in the presence and absence of acceptors respectively.
i DA and i D are the donor decays in the presence and absence of acceptors. ρ interplanar is the FRET contribution arising from
energy transfer to randomly distributed acceptors in two different planes from the donors (two monolayer leaflets) [34].
( Z l
p
) ( ffiffiffiffiffiffiffi 1
q interplanar ¼ exp 2n 2 pl1
2 l
1 2 1 expð tb 3 Z l þR2 e
1 a6 Þ
p
)
ffiffiffiffiffiffiffi 2
a 3 da exp 2n 2 pl2
2 l
2 2 1 expð tb 3 þR2 e
2 a6 Þ
a 3 da
0
0
ð3Þ

1782 F. Fernandes et al. / Biochimica et Biophysica Acta 1758 (2006) 1777–1786
where b i =(R 0 2 /l i ) 2 τ D −1/3 , R 0 is the Förster radius, n 2 is the acceptor density in each leaflet, and l 1 and l 2 are the distance between
the plane of the donors and the two planes of acceptors. Using Eqs. (1)–(3), theoretical expectations for FRET efficiency in a
random distribution of acceptors can be calculated and converted to I DA /I D ratios. The Förster radius is given by:
R 0 ¼ 0:2108ðJj 2 n 4 / D Þ 1=6
ð4Þ
where J is the spectral overlap integral, κ 2 is the orientation factor, n is the refractive index of the medium, and ϕ D is the donor
quantum yield. J is calculated as:
Z
J ¼ f ðkÞeðkÞk 4 dk
ð5Þ
where f(λ) is the normalized emission spectra of the donor and ε(λ) is the absorption spectra of the acceptor. The numeric factor
in Eq. (4) assumes nm units for the wavelength λ and Å units for R 0 .
Assuming the already determined value for Tyr15 position in the bilayer (6 Å from the center) and the distance from the center of
the bilayer of the NBD moiety in NBD labeled phospholipids (19 Å) [35], l 1 and l 2 were determined to be 13 and 25 Å, respectively.
Using Eqs. (4) and (5) the Förster radius of the Tyr-NBD donor–acceptor pair was determined to be R 0 =22 Å. In the fitting
procedure, these values are kept fixed and the only parameter being fitted is R e. The results from the experiment along with the curve
fitted from the theoretical model, are presented in Fig. 5B. Both sets of data (L/P=50 and 100) were fitted with a value of R e =9 Å,
matching the expectation for a monomeric peptide. In case there were aggregates at the protein concentrations used these must be
small enough so that the distances between Tyr15 and the surrounding phospholipids are not affected. From Eqs. (1)–(3) and using
the same parameter values described above as well as R e =20 Å (large aggregates), a FRET simulation was performed for the same
system and the obtained donor quenching curve is compared to our experimental values in Fig. 5B.
4.2. Peptide inhibitor binding studies
Both bafilomycin A 1 and SB 242784 absorb in the UV region and can act as acceptors for tyrosine in energy transfer experiments
(spectra are shown in Fig. 6 and 7). The Förster radius for the Tyr-bafilomycin A 1 donor–acceptor pair is 20 Å, whereas for the Tyr-
SB 242784 pair it is 24 Å (Eq. (4)). A 3 mM concentration of lipid was used to ensure an almost complete incorporation of inhibitors
in the bilayer. SB 242784 partition coefficient to DOPC bilayers is 1.20×10 4 [36], and at the lipid concentration used 97% of the
inhibitor molecules are incorporated in the bilayer. The fraction of inhibitors not incorporated in the vesicles was taken into account
on the acceptor concentrations plot in Fig. 7, which depicts the Tyr15 quenching via energy transfer to SB 242784.
In contrast to SB 242784, bafilomycin A 1 is not a fluorescent molecule and partition coefficients could not be determined using
photophysical techniques. Overestimation of its partition coefficient could lead to an underestimation of bafilomycin-peptide binding
constants. However, it was recently showed by EPR of spin-labeled lipids that the macrolide molecule concanamycin A, another
powerful V-ATPase inhibitor with very similar structure to bafilomycin A 1 , readily incorporates in lipid membranes [37]. Thus, it is
to be expected from this result and from the high hydrophobic character of the molecule that the incorporation of bafilomycin A 1 at
the lipid concentrations used is close to 100%.
On the inhibitor-peptide binding assays, the L/P was kept at 100 in order to ensure minimum levels of aggregation. Experimental
FRET efficiencies obtained with both peptide H4/SB 242784 and peptide H4/bafilomycin A 1 donor/acceptor pairs (Eq. (1)), are
compared to theoretical expectations for a random distribution of acceptors (the scenario in which there is an absence of binding)
obtained from Eqs. (1) and (2) (see Figs. 6 and 7). The position of SB 242784 in DOPC bilayers was already determined from
selective quenching methodology (parallax method) to be 12.8 Å from the center of the bilayer [36], but the position of the
bafilomycin A 1 chromophore was impossible to determine by the same methodology since this molecule is not fluorescent. In Fig.
6A, the simulations corresponding to different positions of the bafilomycin A 1 chromophore inside the bilayer are shown, together
with the experimental data. It is clear that even when assuming that the Tyr15 is located on the same bilayer plane as the bafilomycin
A 1 chromophore (6 Å from the center of the bilayer), in a situation of maximal energy transfer efficiency, the experimental energy
transfer efficiencies cannot be solely explained by the unbound population of inhibitor molecules (i.e. random distribution of
acceptors) and a fraction of peptide H4 must be binding bafilomycin A 1 . It is difficult to precisely quantify the binding constant (K b )
for this process due to the uncertainty relative to the bafilomycin A 1 chromophore position in the bilayer, but a lower and a higher
limit can be determined assuming the closest and furthest position possible for bafilomycin A 1 and Tyr15 (corresponding
respectively to a position in the center and in the surface of the bilayer for bafilomycin A 1 ), and also that peptide H4/bafilomycin A 1
complexes are completely non-fluorescent (Eqs. (6) and (7)). This last assumption is valid, since for a contact interaction either by
Förster type or other transfer (exchange) mechanism, the efficiency of transfer would be 100%. The equations valid for the FRET
efficiency in a scenario of peptide H4-bafilomycin A 1 1:1 complex formation are given below:
E EXP ¼ E random þ H4 Baf ð1 E random Þ ð6Þ
½H4Š T

F. Fernandes et al. / Biochimica et Biophysica Acta 1758 (2006) 1777–1786
1783
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
½H4 Baf Š
¼ 1 K bð½Baf Š T
þ½H4Š T
Þþ ð1 þ K b ð½Baf Š T
þ½H4Š T
ÞÞ 2 4ðKb 2½Baf Š T ½H4Š T Þ
ð7Þ
½H4Š T
ð 2K b Þ½H4Š T
where [H4-Baf] is the effective concentration (molecules per lipid volume as the binding is confined to the lipid phase) of the
peptide–bafilomycin A 1 complex, and, [H4] T and [Baf] T are respectively, the effective analytical concentrations of the peptide and of
bafilomycin A 1 . E EXP is the energy transfer efficiencies obtained experimentally and E random is the theoretical energy transfer
efficiency assuming a random distribution of acceptors (Eqs. (1)–(3)). In case binding is detected, the use of effective bafilomycin A 1
concentrations will result in lower binding constants as compared to the ones of bulk concentrations, but the binding constants
obtained in this way have a larger physical meaning as they will not change with lipid concentration. Anyway they can be easily
interconverted.
Assuming a position for bafilomycin A 1 in the center of the bilayer, the fitting procedure (Fig. 6A) recovered K b =10.5±1.32 M − 1
(mol per volume of lipid). When the bafilomycin A 1 position was fixed to 20 Å from the center of the bilayer (i.e. at the surface), the
value recovered was K b =36.2±2.8 M − 1 . Therefore, the binding constant upper and lower bounds for the peptide H4–bafilomycin
A 1 complex are 10.5±1.32 13.72±3.05 M − 1 .
Fig. 6. (A) Fluorescence emission quenching of Tyr15 of peptide H4 by FRET to bafilomycin A 1 (●). Theoretical expectations (Eqs. (1–3)) for FRET in the absence of
inhibitorbindingtopeptideH4assumingpositionsforbafilomycinA 1 of12.5Å(⋯),8.5Å(–––),and4.5Å(—) fromthecenterofthebilayer.( )Fittingofequations6 and
7 to FRET data assuming a bafilomycin A 1 position of 4.5 Å from the center of the bilayer. The interval 10.5±1.34

1784 F. Fernandes et al. / Biochimica et Biophysica Acta 1758 (2006) 1777–1786
Fig. 7. Fluorescence emission quenching of Tyr15 of peptide H4 by FRET to SB 242784 (●). Curve corresponds to the theoretical expectation for FRET in the absence
of inhibitor binding to peptide H4 (see text). The range of SB 242784 to total phospholipid ratios in these experiments was 0.18% to 1.8%. Inset: overlap between
tyrosine fluorescence emission ( ) and SB242784 absorption spectrum (–––), R 0 (Tyr-SB 242784)=24 Å.
When the FRET binding assay was applied to the peptide H4/SB 242784 system, the results were substantially different (Fig.
7). With this inhibitor, the obtained FRET efficiencies matched the theoretical expectations for a random distribution of acceptors,
suggesting absence of binding between SB 242784 and the peptide. No uncertainty exists regarding this result as the position of
the inhibitor inside the bilayer is accurately know from parallax fluorescence quenching techniques (12.8 Å from the center of the
bilayer — [36]).
5. Discussion
5.1. Use of peptide H4 as a model peptide for the 4th
transmembrane segment of subunit c
Aggregation of transmembrane peptides containing a hydrophobic
sequence and flanked by basic residues has already been
reported [38–40], thus detection of aggregates of peptide H4 at
low L/P ratios (

F. Fernandes et al. / Biochimica et Biophysica Acta 1758 (2006) 1777–1786
1785
[26]. In the same way, c-subunit residues shown to be important
for bafilomycin A 1 inhibitory activity through mutational
studies [9] could not be directly involved in the binding of the
inhibitor, but still play a role in the inhibition mechanism. Here,
we proposed to further define the structural requirements for
bafilomycin A 1 and SB 242784 binding to the V-ATPase, by
focusing on the protein domain expected to be the primary
contributor to the inhibitor binding site, the putative 4th
transmembrane domain of the c-subunit. The extent of binding
efficiencies achieved using only peptides corresponding to this
section of the protein is an indicator of the relevance of the
remaining protein domains to the binding of inhibitor, the first
step in the inhibitory mechanism.
No evidence was found supporting SB 242784 association
with the peptide H4. On the other hand, bafilomycin A 1 was
shown to possess only moderate affinity for the transmembrane
peptides. Due to uncertainties in bafilomycin A 1 position in
bilayers it was impossible to rigorously quantify a binding
constant for the process, but a reliable range for this parameter
could be presented. The differences in bafilomycin A 1 binding
efficiencies for the peptide H4, H4c′ and H4 A ← E are not
significant as they fall within, or close to the error of the fits. For
H4c′ this result is somewhat surprising. Mutations on conserved
residues of the c-subunit known to confer resistance to
bafilomycin A 1 and concanamycin were shown to be ineffective
in the c′ and c′ isoforms [9]. It was proposed that if a
bafilomycin A 1 binding site was present on these isoforms, then
it must possess lower affinity than in the c isoform [9]. If this is
the case, then the differences in binding efficiencies between the
two isoforms must be explained by discrepancies in the
interaction of the inhibitor with parts of the protein other than
the highly conserved putative 4th transmembrane helix.
Although the contribution from the 4th transmembrane
segment (TM4) to the binding site affinity for bafilomycin A 1 is
not negligible, it is absolutely unable to explain the extremely
low IC 50 of bafilomycin A 1 (0.1 nM). The binding efficiencies
observed for concanamycin (IC 50 close to bafilomycin A 1 ) and
the intact isolated c-subunit [27] were more than 1000 times
larger than the ones recovered in the present study. Therefore, it
is clear that either: (i) the other transmembrane segments from
the c-subunit, namely TM1 and TM2, are essential in the
formation of an efficient inhibitor binding site, or that (ii)
interactions of TM4 with other protein segments are required for
this protein section to assume the appropriate folding which
enables the establishment of the necessary interactions for an
efficient binding between the residues of the 4th transmembrane
segment of the c-subunit and the inhibitors. Absence of SB
242784 binding to the peptide H4 could be explained by the
much lower IC 50 of this inhibitor (26 nM in chicken osteoclasts
[22]) that reflects a much lower binding site affinity for this
inhibitor, and the absence of remaining c-subunit transmembrane
segments potentiates this effect.
In this work, in addition to the detailed comparison of the
different inhibitors and peptides, it was concluded that in
contrast to some reported cases [11–16], there is a very
significant difference in binding when comparing the binding to
selected protein segments or to the whole protein. Although
studies using peptides are very relevant, clearly in the case of
the interaction between the bafilomycin A 1 or SB 242784 with
the c-subunit from V-ATPase, the whole protein architecture
and the environment that it provides, are key factors for an
efficient binding.
Acknowledgments
This work was supported by contract no. QLG-CT-2000-
01801 of the European Commission (MIVase consortium).
M. H. and M. P. are members of the COST D22 Action of
the European Union. F. F. acknowledges grant SFRH/BD/
14282/2003 from FCT (Portugal). This work was partially
funded by FCT (Portugal) under the program POCTI.
References
[1] M.E. Finbow, M.A. Harrison, The vacuolar H+-ATPase: a universal
proton-pump of eukaryotes, Biochem. J. 324 (1997) 697–712.
[2] T. Nishi, M. Forgac, The vacuolar (H+)-ATPases—Nature's most versatile
proton pumps, Mol. Cell. Biol. 3 (2002) 94–103.
[3] B. Powell, L.A. Graham, T.H. Stevens, Molecular characterization of the
yeast vacuolar H+-ATPase proton pore, J. Biol. Chem. 275 (2000)
23654–23660.
[4] M.A. Harrison, M.E. Finbow, J.B. Findlay, Postulate for the molecular
mechanism of the vacuolar H(+)-ATPase (hypothesis), Mol. Membr. Biol.
14 (1997) 1–3.
[5] B.J. Bowman, E.J. Bowman, Mutations in subunit c of the vacuolar
ATPase confer resistance to bafilomycin and identify a conserved
antibiotic binding site, J. Biol. Chem. 277 (2002) 3965–3972.
[6] G. Werner, H. Hagenmaier, H. Drautz, A. Baumgartner, H. Zähner,
Metabolic products of microorganisms. Bafilomycins, isolation,
chemical structure and biological activity, J. Antibiot. 37 (1984)
110–117.
[7] M. Harrison, B. Powell, M.E. Finbow, J.B.C. Findlay, Identification of
lipid-accessible sites on the Nephrops 16-kDa proteolipid incorporated into
a hybrid vacuolar H+-ATPase: site-directed labeling with n-(1-pyrenyl)
cyclohexylcarbodiimide and fluorescence quenching analysis, Biochemistry
39 (2000) 7531–7537.
[8] T. Páli, G. Whyteside, N. Dixon, T. Kee, S. Ball, M.A. Harrison, J.B.C.
Findlay, M.E. Finbow, D. Marsh, Interaction of inhibitors of the vacuolar
H+-ATPase with the transmembrane Vo-sector, Biochemistry 43 (2004)
12297–12305.
[9] E.J. Bowman, L.A. Graham, T.H. Stevens, B.J. Bowman, The bafilomycin/concanamycin
binding site in subunit c of the V-ATPases from Neurospora
crassa and Saccharomyces cerevisiae, J. Biol. Chem. 279 (2004)
33131–33138.
[10] M. Harrison, J. Murray, B. Powell, Y. Kim, M.E. Finbow, J.B.C. Findlay,
Helical interactions and membrane disposition of the 16-kDa proteolipid
subunit of the vacuolar H1-ATPase analyzed by cysteine replacement
mutagenesis, J. Biol. Chem. 274 (1999) 25461–25470.
[11] T.L. Lentz, V. Chaturverdi, B.M. Conti-Fine, Amino acids within residues
181–200 of the nicotinic acetylcholine receptor alpha 1 subunit in nicotine
binding, Biochem. Pharmacol. 55 (1998) 341–347.
[12] P.T. Wilson, T.L. Lentz, E. Hawrot, Determination of the primary amino
acid sequence specifying the alpha-bungarotoxin binding site on the alpha
subunit of the acetylcholine receptor from Torpedo californica, Proc. Natl.
Acad. Sci. U. S. A. 82 (1985) 8790–8794.
[13] D. Neumann, D. Barchan, M. Fridkint, S. Fuchs, Analysis of ligand
binding to the synthetic dodecapeptide of the acetylcholine receptor a
subunit, Proc. Natl. Acad. Sci. U. S. A. 83 (1986) 9250–9253.
[14] A.H. Taylor, G. Heavner, M. Nedelman, D. Sherris, E. Brunt, D. Knight, J.
Ghrayeb, Lipopolysaccharide (LPS) neutralizing peptides reveal a lipid A
binding site of LPS binding protein, J. Biol. Chem. 270 (1995)
17934–17938.

1786 F. Fernandes et al. / Biochimica et Biophysica Acta 1758 (2006) 1777–1786
[15] C. Ried, C. Wahl, T. Miethke, G. Wellnhofer, C. Landgraf, J.
Schneider-Mergener, A. Hoess, High affinity endotoxin-binding and
neutralizing peptides based on the crystal structure of recombinant
Limulus anti-lipopolysaccharide factor, J. Biol. Chem. 271 (1996)
28120–28127.
[16] A.C. Scatigno, S.S. Garrido, R. Marchetto, A 4.2 kDa synthetic peptide as
a potential probe to evaluate the antibacterial activity of coumarin drugs,
J. Peptide Sci. 10 (2004) 566–577.
[17] S.R. George, G.Y.K. Ng, S.P. Lee, T. Fan, G. Varghese, C. Wang, C.M.
Deber, P. Seeman, B.F. O'Dowd, Blockade of G protein-coupled
receptors and the dopamine transporter by a transmembrane domain
peptide: novel strategy for functional inhibition of membrane proteins in
vivo, J. Pharmacol. Exp. Ther. 307 (2003) 481–489.
[18] J.L. Popot, J. Trewhella, D.M. Engelman, Reformation of crystalline
purple membrane from purified bacteriorhodopsin fragments, EMBO J.
(1986) 3039–3044.
[19] T. Marti, Refolding of bacteriorhodopsin from expressed polypeptide
fragments, J. Biol. Chem. 273 (1998) 9312–9322.
[20] S. Ozawa, R. Hayashi, A. Masuda, T. Iio, S. Takahashi, Reconstitution of
bacteriorhodopsin from a mixture of a proteinase V8 fragment and two
synthetic peptides, Biochim. Biophys. Acta 1323 (1997) 145–153.
[21] R.W. Hesselink, R.B.M. Koehorst, P.V. Nazarov, M.A. Hemminga,
Membrane-bound peptides mimicking transmembrane Vph1p helix 7 of
yeast V-ATPase: a spectroscopic and polarity mismatch study, Biochim.
Biophys. Acta 1716 (2005) 137–145.
[22] G. Nadler, M. Morvan, I. Delimoge, P. Belfiore, A. Zocchetti, I. James, D.
Zembryki, E. Lee-Rycakzewski, C. Parini, E. Consolandi, S. Gagliardi, C.
Farina, (2Z,4E)-5-(5,6-dichloro-2-indolyl)-2-methoxy-n-(1,2,2,6,6-pentamethylpiperidin-4-yl)-2,4-pentadienamide,
a novel, potent and selective
inhibitor of the osteoclast V-ATPase, Bioorg. Med. Chem. Lett. 8 (1998)
3621–3626.
[23] S. Gagliardi, G. Nadler, E. Consolandi, C. Parini, M. Morvan, M.N. Legave,
P. Belfiore, A. Zocchetti, G.D. Clarke, I. James, P. Nambi, M. Gowen, C.
Farina, 5-(5,6-Dichloro-2-indolyl)-2-methoxy-2,4-pentadienamides: novel
and selective inhibitors of the vacuolar H+-ATPase of osteoclasts with bone
antiresorptive activity, J. Med. Chem. 41 (1998) 1568–1573.
[24] L. Visentin, L.A. Dodds, M. Valente, P. Misiano, J.N. Bradbeer, S. Oneta,
X. Liang, M. Gowen, C. Farina, A selective inhibitor of the osteoclastic
V-H+-ATPase prevents bone loss in both thyroparathyroidectomized and
ovariectomized rats, J. Clin. Invest. 103 (2000) 309–318.
[25] M.F. Manolson, H. Yu, W. Chen, Y. Yao, K. Li, R.L. Lees, J.N.M.
Heersche, The a3 isoform of the 100-kDa V-ATPase subunit is highly but
differentially expressed in large (10 nuclei) and small (5 Nuclei)
osteoclasts, J. Biol. Chem. 278 (2003) 49271–49278.
[26] Y. Wang, T. Inoue, M. Forgac, Subunit a of the yeast V-ATPase participates
in binding of bafilomycin, J. Biol. Chem. 280 (2005) 40481–40488.
[27] G. Whyteside, P.J. Meek, S.K. Ball, N. Dixon, M.E. Finbow, T.P. Kee,
J.B.C. Findlay, M.A. Harrison, Concanamycin and indolyl pentadieneamide
inhibitors of the vacuolar H+-ATPase bind with high affinity to the
purified proteolipid subunit of the membrane domain, Biochemistry 44
(2005) 15024–15031.
[28] D.F. Eaton, Reference materials for fluorescence measurement, Pure Appl.
Chem. 60 (1988) 1107–1114.
[29] R. Cerpa, F.E. Cohen, I.D. Kuntz, Conformational switching in designed
peptides: the helix/sheet transition, Fold. Des. 1 (1996) 91–101.
[30] R.W. Cogwill, Tyrosil fluorescence in proteins and model peptides, in: R.F.
Chen, H. Edelhoch (Eds.), Biochemical Fluorescence: Concepts 2, Marcel
Dekker, New York, 1976, pp. 441–486.
[31] J.W. Longworth, Excited states of proteins and nucleic acid, in: R.F.
Steiner, I. Weinryb (Eds.), Plenum Press, New York, 1971, pp. 319–484.
[32] H. Nelson, N. Nelson, The progenitor of ATP synthases was closely related
to the current vacuolar H+-ATPase, FEBS Lett. 247 (1989) 147–153.
[33] A. Chattopadhyay, E. London, Parallax method for direct measurement
of membrane penetration depth utilizing fluorescence quenching by
spin-labeled phospholipids, Biochemistry 26 (1987) 39–45.
[34] L. Davenport, R.E. Dale, R.H. Bisby, R.B. Cundall, Transverse location of
the fluorescent probe 1,6-diphenyl-1,3,5-hexatriene in model lipid bilayer
membrane systems by resonance energy transfer, Biochemistry 24 (1985)
4097–4108.
[35] F.S. Abrams, E. London, Extension of the parallax analysis of membrane
penetration depth to the polar region of model membranes: use of
fluorescence quenching by a spin-label attached to the phospholipid polar
headgroup, Biochemistry 32 (1993) 10826–10831.
[36] F. Fernandes, L. Loura, R.B.M. Koehorst, N. Dixon, T.P. Kee, M.A.
Hemminga, M. Prieto, Interaction of the indole class of V-ATPase
inhibitors with lipid bilayers, Biochemistry 45 (2006) 5271–5279.
[37] T. Páli, N. Dixon, T.P. Kee, D. Marsh, Incorporation of the V-ATPase
inhibitors concanamycin and indole pentadiene in lipid membranes, Spinlabel
EPR studies, Biochim. Biophys. Acta 1663 (2004) 14–18.
[38] J. Ren, S. Lew, J. Wang, E. London, Control of the transmembrane
orientation and interhelical interactions within membranes by hydrophobic
helix length, Biochemistry 38 (1999) 5905–5912.
[39] F.X. Zhou, H.J. Merianos, A.T. Brunger, D.M. Engelman, Polar residues
drive association of polyleucine transmembrane helices, Proc. Natl. Acad.
Sci. U. S. A. 98 (2000) 2250–2255.
[40] S. Mall, R. Broadbridge, R.P. Sharma, J.M. East, A.G. Lee, Selfassociation
of model transmembrane helix is modulated by lipid structure,
Biochemistry 40 (2000) 12379–12386.

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
IV
BINDING OF A QUINOLONE
ANTIBIOTIC TO BACTERIAL
PORIN OmpF
1. Introduction
Fluoroquinolones (quinolones) have been shown to present broad-spectrum
antimicrobial activity and are currently one of the most successful classes of drugs.
They are reported as relatively well tolerated drugs with reasonably few undesirable
effects (Albini and Monti, 2003; Stahlmann and Lode, 1999), and as a consequence
their therapeutic application is increasing. The antibiotic activity of quinolones is the
outcome of inhibition of a series of important enzymes for chromosome function and
topology (homologous type II topoisomerases, DNA gyrase, and DNA topoisomerase
IV) (Pestova et al., 2000).
First generation quinolones (nalidixic acid) were used in the treatment of
infections caused by Gram-negative bacteria. More recently, new derivatives were
developed with enhanced antibiotic activity against Gram-positive bacteria, namely
Streptococus pneumoniae, responsible for infections in the respiratory system and other
pathologies (Quiniliani et al., 1999). Some of the new generation quinolones are also
apparently effective against multidrug resistant Mycobacterium tuberculosis and were
shown to be useful in the treatment of AIDS patients infected with this pathogen
(Houston and Farming, 1994; Hooper and Wolfson, 1995). These bacteria are resistant
to a wide range of aggressive agents (acid/alcohol) due to the presence of a complex
permeability barrier (Nikaido et al., 1993).
The possible strategies for quinolones entry through the outer and inner
membranes of bacteria are via porin channels (Dechené et al., 1990; Chevalier et al.,
2000), or hydrophobic diffusion through the lipid bilayer. The porin pathway is
107

apparently the most significant as the efficiency of the quinolones is not directly related
to the water/lipid partition coefficient of the molecules (Vazquez et al., 2001). Indeed,
by either not expressing or expressing structurally changed outer membrane porins,
entrance of quinolones in bacteria is severely impaired (Mascaretti, 2003; Chevalier et
al., 2000).
OmpF is one of the porins involved in the permeation process of quinolones
across the outer membrane of bacteria (Chapman and Georgopapadakou, 1988). It was
the first membrane protein to yield crystals of size and order allowing to be analyzed by
X-ray crystallography (Garavito and Rosenbusch, 1980). This porin forms a homotrimer
in the outer membrane and each monomer presents β-barrel structure in which one of
the loops folds back to the barrel, forming a constriction zone at half the height of the
channel (Figure 1 in section 2 of this Chapter). This loop is expected to contribute
significantly to the control of the permeability of OmpF.
It is not yet know whether OmpF operates as a channel or as an enabler of
quinolones diffusion across the lipid-OmpF interface. Therefore, the knowledge of the
exact type of interaction established by OmpF and quinolones is of great biological
relevance, and could be potentially useful in the development of more effective drugs in
the future. In this chapter, the interaction of OmpF and a second generation quinolone,
ciprofloxacin (CP), was studied. CP is a 6-fluoroquinolone which structure is shown in
Figure IV-1. Recent studies suggested a 1:1 stoichiometry for OmpF-CP interaction
(Neves et al., 2005).
Figure IV-1: Chemical structure of CP.
108

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
2. CIPROFLOXACIN INTERACTIONS WITH
BACTERIAL PROTEIN OMPF: MODELLING
OF FRET FROM A MULTI-TRYPTOPHAN
PROTEIN TRIMER
ABSTRACT
The outer membrane protein F (OmpF) is known to play an important role in the uptake of
fluoroquinolone antibiotics by bacteria. In this study, the degree of binding of the fluoroquinolone
antibiotic ciprofloxacin to OmpF in a lipid membrane environment is quantified using a methodology
based on Förster resonance energy transfer (FRET). Analysis of the fluorescence quenching of OmpF is
complex as each OmpF monomer presents two tryptophans at different positions, thus sensing two
different distributions of acceptors in the bilayer plane. Specific FRET formalisms were derived
accounting for the different energy transfer contributions to quenching of each type of tryptophan of
OmpF, allowing the recovery of upper and lower boundaries for the ciprofloxacin-OmpF binding constant
(K B ). log (K B ) was found to lie in the range 3.15-3.62 or 3.58-4.00 depending on the location for the
ciprofloxacin binding site assumed in the FRET modelling, closer to the centre or to the periphery of the
OmpF trimer, respectively. This methodology is suitable for the analysis of FRET data obtained with
similar protein systems and can be readily adapted to different geometries.
109

110

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
CIPROFLOXACIN INTERACTIONS WITH BACTERIAL
PROTEIN OMPF: MODELLING OF FRET FROM A
MULTI-TRYPTOPHAN PROTEIN TRIMER
Fábio Fernandes a , Patrícia Neves b , Paula Gameiro b , Luís M. S. Loura c , Manuel Prieto a, *
a - Centro de Química-Física Molecular, Instituto Superior Técnico, Lisbon, Portugal.
b - Requimte, Departamento de Química, Faculdade de Ciências, Porto, Portugal.
c - Centro de Química and Departamento de Química, Universidade de Évora, Évora,
Portugal.
ABSTRACT
The outer membrane protein F (OmpF) is known to play an important role in the uptake
of fluoroquinolone antibiotics by bacteria. In this study, the degree of binding of the
fluoroquinolone antibiotic ciprofloxacin to OmpF in a lipid membrane environment is
quantified using a methodology based on Förster resonance energy transfer (FRET).
Analysis of the fluorescence quenching of OmpF is complex as each OmpF monomer
presents two tryptophans at different positions, thus sensing two different distributions
of acceptors in the bilayer plane. Specific FRET formalisms were derived accounting
for the different energy transfer contributions to quenching of each type of tryptophan
of OmpF, allowing the recovery of upper and lower boundaries for the ciprofloxacin-
OmpF binding constant (K B ). log (K B ) was found to lie in the range 3.15-3.62 or 3.58-
4.00 depending on the location for the ciprofloxacin binding site assumed in the FRET
modelling, closer to the centre or to the periphery of the OmpF trimer, respectively.
This methodology is suitable for the analysis of FRET data obtained with similar
protein systems and can be readily adapted to different geometries.
Abbreviations
CP-Ciprofloxacin; FRET- Förster resonance energy transfer; oPOE- n-
octylpolyoxyethylene; OmpF- Outer Membrane protein F; DMPC-Dimyristoyl-L-αphosphatidylcholine;
Trp- tryptophan; CMC- critical micellar concentration.
111

Introduction
Quinolones are broad-spectrum antibacterial agents which mechanism of action is the
inhibition of DNA gyrase and DNA topoisomerase IV enzymes that control DNA
topology and are vital for bacterial replication [1-3]. Access to the target site is a major
determinant of antibacterial activity, with the outer membrane being the major
permeability barrier in Gram negative bacteria [1]. In fact, one of the mechanisms of
resistance developed by the bacterial cell is the process of making more difficult the
access of quinolones to their target of action, by either not expressing or expressing
structurally changed outer membrane porins [3,4]. One of those porins, which
microbiology studies related to the permeation of some quinolones through the outer
membrane, is OmpF [1, 3-6]. Indeed, porin-deficient mutants of E. coli are resistant to
fluoroquinolones, although the role of OmpF, either as a channel or as an enabler of
quinolone diffusion at the OmpF/ lipid interface, has not yet been elucidated. The
relative importance and the different areas of contact of each quinolone with OmpF, is a
subject of great importance in the context of developing new molecules with less
resistance problems.
OmpF is a trimer within the membrane and it contains just two tryptophan residues per
monomer (Fig.1), Trp 214 at the lipid protein interface and Trp 61 at the trimer interface
[7]. The protein shows a maximum of emission at relatively low wavelengths, which
suggests that both tryptophans are in hydrophobic environments. This is confirmed by
experiments involving OmpF mutants [7], which lack one or both Trp’s.
Ciprofloxacin (CP) (Fig. 2) is a 6-fluoroquinolone antibiotic currently under clinical use
for which many resistances have been reported in a large number of microbial species.
The aim of the present study was to investigate the role of OmpF as a major pathway for
CP entry through the outer membrane into the bacterial cell by analysing the alteration
of OmpF fluorescence in presence of increasing concentration of CP.
In this way, the extent of Förster resonance energy transfer (FRET) between OmpF and
CP was measured, which associated with a rational modelling of the distribution of
donor and acceptors in the bilayer allowed us to quantify the extent of binding between
the protein and the antibiotic. The FRET modelling presented here is also suitable for
application to systems with different geometries, and the new FRET formalisms derived
can be readily adapted to the analysis of FRET data obtained with other large membrane
proteins presenting donor fluorophores in the protein-lipid interface.
112

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
Our study assumes significance in the overall context of the increasing problem of
bacterial resistance to the antibiotics currently in use, and the consequent need of
understanding the processes involved, in order to create new molecules with increased
antibacterial activity and less resistance problems.
Materials and methods
Ciprofloxacin (CP) was a gift from Bayer (Leverkusen, Germany). N-(2-hydroxyethyl)
piperazine-N’-ethanesulfonic acid (Hepes) was from Sigma (Sigma, St. Louis, MO).
Octylpolyoxyethylene (oPOE) was from Bachem (Bubendorf, Switzerland) and all
other chemicals were from Merck (Darmstadt, Germany). All solutions were prepared
with 10 mM Hepes buffer (0.1 M NaCl; pH 7.4). OmpF was purified from E. coli, strain
BL21 (DE3) Omp8, following published procedures (8). OmpF concentration was
estimated using the bicinchoninic acid protein assay against bovine serum albumin as
standard.
All the absorption measurements were carried out with a UNICAM UV-300
spectrophotometer equipped with a constant-temperature cell holder. Spectra were
recorded at 37ºC in 1 cm quartz cuvettes in the range 230 to 350 nm. Fluorescence
measurements were performed in a Varian spectrofluorimeter, model Cary Eclipse,
equipped with a constant-temperature cell holder (Peltier single cell holder). All the
spectra were recorded at 37ºC, under constant stirring, with slit widths of excitation and
emission of 10 nm.
Solutions
All the antibiotic solutions and proteoliposomes suspensions were prepared in 10 mM
Hepes buffer (pH 7.4, 0.1 M NaCl).
Reconstitution of OmpF in DMPC liposomes
As the final purpose for the proteoliposomes was the study of the quinolones interaction
with OmpF in a structural perspective, correct orientation of protein insertion was
important and best achieved by direct incorporation of OmpF into preformed liposomes
113

[9-12]. Having this into consideration, OmpF proteoliposomes were prepared by direct
incorporation into preformed DMPC liposomes, by a well established methodology [10,
13-15]. Briefly, an adequate volume (~ 2.6 ml) of DMPC liposome suspension (~ 2.0
mM) in Hepes buffer, prepared according usual procedures [16], is added to OmpF
(0.45 mg) in a Hepes buffer solution with 0.4% of oPOE. The lipid/protein mole ratio is
always near 1000 and the total volume of the mixture assures a final concentration of
oPOE lower than the value of its CMC (0.23%). After an efficient homogenization by
gentle stirring, the mixture is incubated 15 min at room temperature followed by 1 h on
ice. The detergent is then adsorbed onto SM2 Bio-Beads ® from Bio-Rad (Hercules, CA)
at a concentration of 0.2 g of Bio-Beads/ml, by gently shaking of the suspension during
a period of 3 h. After this time, a second portion of the same amount of Bio-Beads is
added and the suspension is again shaken for another 3 h. In the end of this period,
proteoliposomes are gently removed by decanting the Bio-Beads. The orientation of
OmpF by this reconstitution procedure is considered similar to that observed in the
bacterial membranes, based in experimental and molecular dynamics studies [17,18].
As the study to be performed is based on the fluorescence of OmpF, the presence of
protein not inserted in the membrane would lead to errors in the final results. The
suspension of proteoliposomes was therefore submitted to an ultracentrifugation (80000
g, 4 ºC, 2 h) in order to remove any traces of protein not incorporated. After this
procedure the supernatant was rejected and the pellet suspended in Hepes buffer.
Protein was quantified in the liposomes and in the supernatant. The percentage of
incorporation by this methodology was always higher than 76%. In the final step of the
procedure the proteoliposomes are sequentially extruded through 200 and 100 nm
polycarbonate membranes from Nucleopore (Kent, WA).
Quenching of OmpF fluorescence by CP
The fluorescence quenching studies were achieved by successive additions of a constant
volume (10 µl) of CP solution (~ 296 µM) to the cuvette (final concentration range: 0-
38µM) containing a constant amount of OmpF (~ 0.45 µM) incorporated in the
liposomes.
Fluorescence spectra were measured with an excitation wavelength of 290 nm. Inner
filter effects and dilution of the solution were taken into account.
114

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
Theoretical Modelling
CP quenches OmpF fluorescence through a Förster resonance energy transfer (FRET)
mechanism. FRET efficiencies (E) are calculated from the extent of fluorescence
emission quenching of the donor induced by the presence of acceptors,
IDA
E = 1− = 1−
I
D
∞
∫
DA
0
∞
0
i
∫
i
D
() t dt
() t dt
where I DA and I D are the steady-state fluorescence intensities of the donor in the
presence and absence of acceptors respectively. i DA (t) and i D (t) are the donor
fluorescence decays in the presence and absence of acceptors respectively. Although
there is a contribution of static quenching to FRET in OmpF-CP complexes, Eq.(1) still
holds since this is taken into account as described below (Eqs. 8-12).
As seen in Fig.3, CP absorbance overlaps with the OmpF’s fluorescence emission
spectrum (from both tryptophans 214 and 61), and the overlap integral (J) is calculated
as:
(1)
∫
( )
4
J = f λ ⋅ε(λ) ⋅λ ⋅dλ
(2)
where f(λ) is the normalized emission spectrum of the donor and ε(λ) is the absorption
spectrum of the acceptor.
The Förster radius (R 0 ) is given by:
R
2 −4
1/ 6
0
= 0.2108
⋅ ( J ⋅ κ ⋅ n ⋅ φ
D
)
(3)
where the orientational factor is assumed in the dynamic isotropic regime (κ 2 = 2/3), the
refractive index of the medium is n = 1.44, and φ D is the donor quantum yield. The
numeric factor in Eq.(3) assumes nm units for the wavelength λ and Å units for R 0 .
The Förster radius (R 0 ) of the Tryptophan(OmpF)-Ciproflaxacin donor-acceptor pair
was determined to be 23.3 Å. This is an average value since each of the tryptophans has
distinct fluorescence properties and therefore is expected to present different (but, in
115

any case, very similar, probably within 1-2 Å of each other) Förster radii for energy
transfer to CP.
Through an analysis of the extent of fluorescence quenching of a donor by the acceptor
in a FRET experiment it is possible to calculate a binding constant (K B ) between these
two molecules. However, care must be taken when working in solutions where
acceptors are highly concentrated, as it is often the case in experiments performed on
liposomes. In these cases, when both donors and acceptors partition to the lipid bilayer
and interact there, the concentration of acceptors around the donors increases
dramatically relatively to a situation where they are both free in solution. FRET can be
efficient at distances up to 100 Å depending on the Förster radius of the donor-acceptor
pair [19] and therefore, donors in a medium highly concentrated in acceptors, besides
being susceptible to fluorescence quenching due to formation of specific donor-acceptor
complexes, are also quenched by non-bound, nearby acceptors. Only the first
contribution is relevant for the determination of the binding affinity of the donoracceptor
complex, and as the two contributions are not additive (Eq. 4), formalisms that
accurately account for the contribution of non-bound acceptors for the experimentally
measured energy transfer efficiencies must be derived, considering the geometrical
peculiarities of the studied system. The donor fluorescence decay in the presence of
acceptors (i DA (t)) is described as:
i () t = γ ⋅i () t ⋅ρ () t ⋅ρ () t
DA D bound nonbound
( γ )
+ 1- ⋅i ( t) ⋅ρ
() t
D
nonbound
(4)
where γ is the fraction of OmpF bound to CP, ρ bound is the FRET contribution from
energy transfer to acceptors bound to Ompf, and ρ nonbound is the FRET contribution
arising from energy transfer to non-bound acceptors, randomly distributed in different
planes (i) from the donors.
Contribution to FRET of non-bound acceptors.
ρ nonbound for a cylindrically symmetric donor geometry is given by [20]:
116

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
⎡
wi
⎛
⎞⎤
2 2
⎢
wi
+ R
⎜
e 6
2 1−exp( −t⋅b
(5)
i
⋅α ) ⎟⎥
ρnonbound = ∏ ⎢exp⎜−2⋅n2 ⋅π⋅wi
⋅
dα⎟⎥
3
i ⎢
∫
α
0
⎥
⎢ ⎜
⎟
⎣ ⎝
⎠⎦
⎥
where b i =(R 2 0 /w i ) 2 τ −1/3 D , n 2 is the acceptor density in each bilayer leaflet (number of
acceptors per unit area), w i is the distance between the plane of the donors and the i-th
plane of acceptors and R e is the donor exclusion radius (which defines the area around
the donor from where the acceptors are excluded). In protein-ligand FRET studies the
exclusion radius is particularly important if the size of the protein is comparable to the
Förster radius of the donor-acceptor pair.
In order to calculate ρ nonbound , a model must be used to describe the positions of the
donors and acceptors in the bilayer. OmpF has two clear belts of aromatic residues at
the lipid/water interface separated by 25 Å [21]. Both Trp’s (61 and 214) are located in
the same aromatic belt and are assumed to be in the same plane, at 12.5 Å from the
center of the bilayer. When incorporated in liposomes, CP is located in the headgroup
region of the bilayer [22] and according to Nagle and Tristan-Nagle [23] for DMPC the
headgroups average position is 18 Å away from the centre of the bilayer. In the FRET
simulations this is considered to be the position of the plane of acceptors and w 1 and w 2
are assumed to be 5.5 Å and 30.5 Å (Fig. 4A). In our formalism we consider that there
is no partition of non-bound CP in the bilayer area occupied by the OmpF trimer apart
from the specifically bound population of antibiotic molecules.
Trp 61 is located at the trimer interface (Fig. 4B) and the exclusion area for non-bound
CP will be significant. R e (Eq. 5) of Trp 61 is assumed to be 30 Å which is the
approximate diameter of an OmpF monomer in the bilayer plane [24] (Fig. 4B).
On the other hand, Trp 214 is located in the periphery of the OmpF trimer and the
distribution of non-bound CP around it will be cylindrically asymmetric. This difference
in the distribution of acceptors sensed by the two donors (Trp 214 and Trp 61 ) can only be
accounted through the use in the FRET simulations of two different ρ nonbound
contributions.
For Trp 61 the FRET contribution is simply given by Eq.5 and setting i = 2, w 1 = 5.5 Å,
w 2 = 30.5 Å and R e = 30 Å. In the case of Trp 214 a large section of the bilayer is also
occupied by the protein trimer itself and inaccessible to acceptors. However, this
117

placement of the protein around the tryptophan is no longer symmetrical as for Trp 61 ,
and energy transfer must be accounted differently:
2 2
⎡ ⎧ ⎡ wi / wi + Rp
6
⎪ 2 ⎪
⎧ ⎡⎛ t ⎞⎛ R ⎞ ⎤
0 6 ⎪
⎫
ρ
nonbound '() t =
⎢
∏ exp −2π n2w ⎢
i
1−exp ⎢ − α ⎥ f( α)
dα
−
⎢ ⎨ ⎨ ⎬
i
⎢ ∫
⎜ ⎟⎜ ⎟
τ
2 '2 ⎢
wi / w D
w
i ⎥
⎢ ⎪
i + R ⎪ ⎝ ⎠⎝ ⎠ ⎪
d
⎣ ⎩ ⎣ ⎩ ⎣
⎦ ⎭
⎡
⎤ ⎤⎫⎤
⎛ ⎞ ⎪⎥
⎜ ⎟ ⎥
⎣ ⎦ ⎦⎭⎦⎥
1
6
t ⎛ R ⎞
0 6 −3
exp α f( α)
α dα⎥
∫ ⎢ − ⎜ ⎟ ⎥
⎬
τ
2 2 ⎢
wi / w D
wi
⎥ ⎥
i + R ⎝ ⎠⎝ ⎠ ⎪
p
(6)
ρ nonbound’ is therefore the FRET contribution to the decay of a donor located in the
surface of a protein of radius R P due to the presence of acceptors randomly distributed in
i different planes (for details on the derivation of this component and on function f(α)
see Appendix).
Using eqs. (1) and (4)–(6), theoretical expectations for FRET efficiencies for a random
distribution of acceptors can be estimated.
In the simulations and data analysis of FRET efficiencies it was considered that Trp 61
and Trp 214 contributed equally to the fluorescence intensity of the protein. Also, energy
migration among the Trp 61 does not affect the geometry of the problem, and energy
migration between Trp 61 and Trp 214 is a very inefficient process due to the distance (30
Å) between the two residues.
Contribution to FRET of specifically bound acceptors.
To quantify the contribution of specifically bound CP to FRET, it is first necessary to
relate the binding constant of this antibiotic and OmpF (K B ) to the factor γ (fraction of
OmpF bound to CP) in Eq. 4:
( 1 [ OmpF] KB
[ CP]
KB)
⎡ + ⋅ + ⋅ +
⎤
γ = ⎢
⎥ 2⋅K
2
B
⋅ OmpF
⎢
2
( 1+ [ OmpF] ⋅ KB + [ CP]
⋅KB) −4⋅KB
⋅[ OmpF] ⋅[ CP
⎥
⎣
]
⎦
[ ]
Total
(7)
where [OmpF] Total and [OmpF] are the total and unbound concentrations of OmpF, and
[CP] is the concentration of unbound CP. It is assumed that only 1:1 complexes can be
118

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
formed between OmpF and CP. The 1:1 complex model is supported by data from a
previous study [25].
In the event of binding of CP to Ompf, the fluorescence of the several Trp’s present in
one trimer will be quenched differently, as the distances between them and the antibiotic
will be also different. Therefore, also for ρ bound , two different contributions for Trp 214
and Trp 61 must be considered.
ρ = ρ + ρ
bound bound (Trp 214) bound (Trp61)
(8)
Assuming that at least one of the tryptophans of OmpF is part of a hypothetical CP
binding site and subsequently completely quenched, and that no more than one
antibiotic molecule can be specifically bound to a OmpF trimer at a given time, two
models were devised: i) the CP binding site is close to the center of the trimer and
binding of antibiotic leads to complete quenching of the three Trp 61 (distance of Trp 61 to
CP

⎛
II t ⎛ R ⎞
0
ρ () t = exp
−
bound(Trp 214)
⎜
⋅⎜ ⎟
τ
D
d
⎝ ⎝ 2 ⎠
6
⎞
⎟
⎠
(12)
Results and Discussion
Successive additions of small volumes of a CP solution to OmpF proteoliposomes
resulted as expected in a decrease of fluorescence from OmpF (Fig.6). Using Eqs 4-6 it
is possible to compare the theoretical expectation for the efficiencies of energy transfer
from the tryptophans of OmpF to unbound CP (randomly distributed) with the data
obtained experimentally. Clearly the extent of quenching of the tryptophans cannot be
exclusively explained on the basis of energy transfer to unbound CP, and a FRET
contribution to specifically associated antibiotic must be considered.
The binding constants that allow for better fits to the data points are different depending
on the model for binding that is considered (I or II). In the case of model I, that assumes
binding of CP close to the centre of the trimer, quenching is much more effective as the
binding site is in the vicinity of three tryptophans (Trp 61 ), while for model II only one
tryptophan is located near the bound antibiotic. This results in the recovery of larger K B
values for model II than for model I. Upper and lower limits for K B (that allow
respectively for better fits to the data points corresponding to the lower and higher CP
concentration ranges) were determined for each model, and are shown in Table I.
In a previous study, changes in CP absorption spectra were used to estimate a OmpF-CP
binding constant and values of log(K B ) = 3.85 ± 0.34 or 4.17 ± 0.03 were obtained
depending on the method chosen to analyze the absorption spectra changes after
addition of OmpF in a micellar environment [25]. These values fall within or close to
the range of K B obtained by us from the energy transfer data assuming binding of CP to
the periphery of OmpF (3.58 < K B < 4.00). This agreement with an alternative method
can be seen as an indication that the antibiotic binding site is likely to be located away
from the centre of the OmpF trimer where FRET would be more efficient and imply a
smaller K B .
120

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
Conclusions
In the present study, the CP fluoroquinolone is shown to associate with OmpF using
FRET methodologies. The high symmetry of the OmpF trimer allowed the analysis of
FRET data with formalisms that accounted for the different distributions of acceptors
around the two types of tryptophans present in the protein. This methodology is suitable
for application in the analysis of FRET data obtained with similar protein systems and
can be adapted to different geometries. The new FRET formalisms presented here for a
donor in the surface of a cylinder of large radius (with radius comparable to the Förster
radius) are expected to be of great assistance in the analysis of FRET data obtained with
membrane proteins presenting donor fluorophores in the protein-lipid interface.
The equilibrium constant retrieved for the association event between OmpF and CP
assuming binding of the antibiotic in the periphery of the porin is in agreement with a
value determined previously by an independent method (in a micellar environment),
indicating that the binding site of the antibiotic is likely to be located away from the
center of OmpF. Also, the FRET methodology can be applied in cases where there are
no spectral shifts, and its sensitivity is higher as compared to absorption methodologies.
121

APPENDIX – Derivation of the FRET rate between a donor at the outer radius of a
cylindrical transmembrane protein to a plane of acceptors distributed around the latter
Fig. A1 shows the geometry relevant to this problem. Our derivation of the
FRET rate law begins with the general expression for the average decay for a donor
located in the centre of a finite disk with radius R d , N , taking into account all
statistical arrangements of the N acceptor molecules located inside the disk (26):
2 2
N
w + R
6
d ⎡ t ⎛ R
0
⎞ ⎤
< ρ(t)
>
N
= exp( − t / τ) ⋅∏
∫ exp⎢−
⎜ ⎥W(R
i
)dR
i
i=
1
R
⎟
(A1)
w ⎢ τ
⎣ ⎝ i ⎠ ⎥⎦
where W(R i )dR i is the probability of finding acceptor molecule A i in the ring of inner
radius R i and outer radius R i +dR i . The acceptor distribution function is normalized, in
the sense that
∫
w
2 2
w + R d
W (R
i
)dR
i
= 1
(A2)
We now proceed as previously done by Davenport et al. (20) and Capeta et al.
(27) for similar (albeit simpler) geometries. Because all acceptors have the same
distribution function, W(R i )dR i = W(R j )dR j = simply W(R)dR, all integrals in Eq. (A1)
are identical, and denoting them by J(t), we can write
⎛ t ⎞
< ρ( t ) >
[ ] N
N
= exp
⎜−
⎟ ⋅ J ( t)
(A3)
⎝ τ
D ⎠
As pointed out by Davenport et al. [20], the probability of finding an acceptor at
a distance between R and R + dR to the donor in question, given by W(R)dR, is equal to
that of finding an acceptor in the ring between R’ and R’ + dR’ in the acceptors plane,
given by W(R’)dR’. The acceptor distribution function, in terms of R’ = Rsinθ (see Fig.
A1A), is given by
122

[ 1+
(2 / π)arcsin(
R'
/ 2R
)]
BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
⎧
p
R'
⎪
0 < R'
< 2R
'2 2
p
⎪
Rd
− Rp
W ( R'
) = ⎨
(A4)
⎪ 2R'
⎪
'2 2
2Rp
< R'
< R'
d
⎪⎩
Rd
− Rp
Note that, whereas for R’ > 2R p (where R p is the cylindrical protein radius)
W(R’) is as expected for a uniform distribution of acceptors in a planar disk (only
corrected in the denominator for the excluded area resulting from the existence of a
protein molecule inside R d ), for R’ < 2R p the more complex expression denotes that only
a fraction of the region between R’ and R’ + dR is available to acceptors (the shaded
area in Fig. A1B). Insertion of this distribution function in the definition of J(t) (the
integral in Eq. A1), together with the substitution α = cosθ = w/R, leads to
⎡
2 1
6
2w
⎛ t ⎞⎛
R0
⎞ 6
−3
J ( t)
= ∫ exp⎢
⎜−
⎟⎜
⎟ α ⎥ f ( α)
α dα
(A5)
'2 2
Rd
− R
p + ⎢⎣
⎝ τ
2 ' 2 ⎠⎝
⎠
w / w R D
w
d
⎥⎦
⎤
where
⎧
⎪
w
w
⎪
1
< α <
2 '2
2 2
w + Rd
w + 4Rp
⎪
f ( α)
= ⎨
(A6)
⎪
w
⎪ ⎛
2
1 1
⎞
< α <
⎪
⎜ w 1− α
1
+ arcsin ⎟
2 2
+
π ⎜ ⎟
w 4Rp
2
⎩ ⎝
2αRp
⎠
We now rearrange Eq. A5 into
123

2
2w
J ( t)
=
'2
R − R
+
w /
1
∫
d
2 2
w + Rp
2
p
⎛
⎜
⎜
⎝
w /
w /
⎡⎛
t
exp⎢
⎜−
⎢⎣
⎝ τ
D
2 2
w + Rp
−3
∫
α
2 ' 2
w + Rd
⎞⎛
R0
⎟⎜
⎠⎝
w
dα −
6
⎞
⎟
⎠
w /
w /
2 2
w + Rp
∫
2 ' 2
w + Rd
⎤
6
α ⎥ f ( α)
α
⎥⎦
⎪⎧
⎡⎛
t
⎨1
− exp⎢
⎜−
⎪⎩ ⎢⎣
⎝ τ
−3
⎞
⎟
dα⎟
⎠
D
⎞⎛
R0
⎟⎜
⎠⎝
w
6
⎞
⎟
⎠
⎤ ⎪⎫
6
α ⎥ f ( α)
⎬dα +
⎥⎦
⎪⎭
(A7)
and denote the integrals on the right hand side of Eq. A7 by J 1 , J 2 (t) and J 3 (t), according
to the order in which they appear in this equation. J 1 is easily calculated, leading to
2
2w
J ( t)
= 1−
[ J
2
( t)
− J
3(
t)
]
(A8)
'2 2
R − R
d
p
Being the average concentration of acceptors in the bilayer given by,
n
2
= N
'2 2
π(
R d
− R p
)
(A9)
it follows that
2
2πn2w
J ( t)
= 1−
[ J
2
( t)
− J
3(
t)
]
(A10)
N
Inserting this expression for J(t) in Eq. A1, and taking the limit (N → ∞, R’ d →
∞), one obtains the macroscopic decay law. The result is,
2
{ − 2πn
w [ J ( t)
− J ( )]}
ρ ( t)
= exp
2 2 3
t
(A11)
which is equivalent to each of the i items in the product of the right hand side of Eq. 6.
Acknowledgements:
Financial support for this work was provided by “Fundação para a Ciência e
Tecnologia” (FCT, Lisboa) through projects POCI/SAU-FCF/56003/2004 and
124

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
POCI/QUIM/57114/2004. P. N. thanks FCT for a fellowship (SFRH/ BD/ 6369/ 2001).
F. F. acknowledges grant SFRH/BD/14282/2003 from FCT (Portugal).
P.N. thanks Luminita Damian for the help with the extraction and purification of OmpF.
References
[1] J. S. Chapman, N. H. Georgopapadakou, Routes of quinolone permeation in Escherichia coli,
Antimicrob. Agents and Chemother. 32 (1988) 438-442.
[2] E. Pestova, J. J. Millichap, G. A. Noskin, L. R. Peterson, Intracellular targets of moxifloxacin: a
comparison with other fluoroquinolones, J. Antimicrob. Chemother. 45 (2000) 583-590.
[3] O. A Mascaretti, Bacteria versus Antibacterial Agents: An integrated approuch. ASM Press,
Washington D.C. 2003
[4] J. Chevalier, M. Malléa, J.-M. Pagès, Comparative aspects of the diffusion of norfloxacin,
cefepime and spermine through the F porin channel of Enterobacter cloacae, Biochem. J. 348 (2000)
223-227.
[5] L. J. V. Piddock, Y.–F. Jin, V. Ricci, A. E. Asuquo, Quinolone accumulation by Pseudomonas
aeruginosa, Staphylococcus aureus and Escherichia coli, J. Antimicrob. Chemother. 43 (1999) 61-70.
[6] K. Hirai, H. Aoyama, T. Irikura, S. Iyobe, S. Mitsuhashi, Differences in susceptibility to
quinolones of outer membrane mutants of Salmonella typhimurium and Escherichia coli, Antimicrob.
Agents and Chemother. 29 (1986) 535-538.
[7] A. H. O’Keeffe, J. M. East, A. G. Lee, Selectivity in lipid binding to the bacterial outer
membrane protein OmpF, Biophys. J. 79 (2000) 2066-2074.
[8] R. M. Gravito, J.P. Rosenbusch, Isolation and crystallization of bacterial porin, Methods
Enzymol.125 (1986) 309-328.
[9] J.-L. Rigaud, B. Pitard, D. Levy, Reconstitution of membrane proteins into liposomes:
application to energy transducing membrane proteins, Biochim. Biophys. Acta. 123 (1995) 223-246.
[10] J.-L. Rigaud, M. T. Paternostre, A. Bluzat, Mechanisms of membrane protein insertion into
liposomes during reconstitution procedures involving the use of detergents. 2. Incorporation of the lightdriven
proton pump bacteriorhodopsin, Biochemistry. 27 (1988) 2677-2688.
[11] P. Richard, J.-L. Rigaud, P. Graber, Reconstitution of CF0F1 into liposomes using a new
reconstitution procedure, Eur. J. Biochem. 193 (1990) 921-930.
[12] D. Levy, A. Gulik, A. Bluzat, J-L. Rigaud, Reconstitution of the sarcoplasmic reticulum
Ca 2+ ATPase: mechanisms of membrane protein insertion into liposomes during reconstitution procedures
involving the use of detergents, Biochim. Biophys. Acta. 1107 (1992) 283-298.
[13] M. T. Paternostre, M. Roux, J.-L. Rigaud, Mechanisms of membrane protein insertion into
liposomes during reconstituction procedures involving the use of detergents. Solubilization of large
unilamellar liposomes (prepared by reverse-phase evaporation) by Triton X-100, octyl glucoside and
sodium cholate, Biochemistry 27 (1988) 2668-2677.
125

[14] L. Plançon, M. Chami, L. Letellier, Reconstitution of FhuA, an Escherichia coli outer membrane
protein, into liposomes, J. Biol. Chem. 272 (1997) 16868-16872.
[15] J. Knol, K. Sjollema, B. Poolman, Detergent- mediated reconstitution of membrane proteins.
Biochemistry 37 (1998) 16410-16415.
[16] A. D. Bangham M. M. Standish, J. C. Watkins, Diffusion of univalent ions across the lamellae of
swollen phospholipids, J. Mol. Biol. 13 (1965) 238-252.
[17] W. Im, B. Roux, Ions and counterions in a biological channel: A molecular dynamics simulation
of OmpF porin from Escherichia coli in an explicit membrane with 1M KCl aqueous salt solution.
Journal of Molecular biology, 319 (2002) 1177-1197.
[18] E. M. Nestorovich, C. Danelon, M. Winterhalter, S. M. Bezrukov, Designed to penetrate: timeresolved
interaction of single antibiotic molecules with bacterial pores. Proc. Nat. Acad. Sci. USA, 99
(2002) 9789-9794.
[19] J. R. Lakowicz, Principles of fluorescence spectroscopy. Chapter 13, 443-475. Springer Science,
3 rd edition, New York, 2006.
[20] L. Davenport, R. E. Dale, R. E. Bisby, R. B. Cundall, Transverse location of the fluorescent
probe 1,6-diphenyl-1,3,5-hexatriene in model lipid bilayer membrane systems by resonance energy
transfer, Biochemistry 24 (1985) 4097-4108.
[21] D. W. Martin, Active unit of solubilized sarcoplasmic reticulum calcium adenosinetriphophatase:
an active enzyme centrifugation analysis, Biochemistry 22 (1983) 2276 – 2282.
[22] J. L. Vazquez, M. T. Montero, S. Merino, O. Domenech, M. Berlanga, M. Vinas, J. Hernandez-
Borrel, Location and Nature of the Surface Membrane Binding Site of Ciprofloxacin: A Fluorescence
Study, Langmuir. 17 (2001) 1009-1014.
[23] J. F. Nagle, S. Tristam-Nagle, Structure of lipid bilayers, Biochim. Biophys. Acta 1469 (2000)
159-195.
[24] S. W. Cowan, R. M. Garavito, J. N. Jansonius, J. A. Jenkins, R. Karlsson, N. Konig, E. F. Pai,
R. A. Pauptit, P. J. Rizkallah, J. P. Rosenbusch, G. Rummel, T. Schimer, The structure of OmpF porin in
a tetragonal crystal form, Structure 10 (1995) 1041-50.
[25] P. Neves, E. Berkane, P. Gameiro, M. Winterhalter, B. de Castro, Interaction between
quinolones antibiotics and bacterial outer membrane porin OmpF, Biophys.Chem. 113 (2005)123-128.
[26] P. K. Wolber, B. S. Hudson, An analytical solution to the Förster energy transfer problem in two
dimensions, Biophys. J. 28 (1979) 197–210.
[27] R. C. Capeta, J. A. Poveda, L. M. S. Loura, Non-uniform membrane probe distribution in
resonance energy transfer. Application to protein-lipid selectivity, J. Fluoresc. 16 (2005) 161-172.
126

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
Figure Legends:
Figure 1: The structure of the OmpF trimer. The positions of the two Trp residues
(Trp 214 and Trp 61 ) in each monomer are shown. Views of OmpF organization: top (A)
(8) (Reproduced by permission of Biophysical Journal) and perpendicular to membrane
axis (B). Trimer interface in (B) is assigned by T. The image was draw with PYMOL
(DeLano Scientific, San Carlos, CA, http://pymol.sourceforge.net) using PDB
coordinates 1OMF1 (22).
Figure 2: Chemical structure of ciprofloxacin.
Figure 3: Normalized fluorescence emission spectra of OmpF (---) and absorption
spectra of ciprofloxacin (▬).
Figure 4: A: Positions of donors (Trp 61 and Trp 214 ) and acceptors (ciprofloxacin) in the
DMPC bilayer. In the simulations, the OmpF monomer is approximated to a cylinder of
radius 30 Å (see text). B: Representation of the OmpF trimer as assumed in the FRET
simulations. Trp 61 is located in the trimer interface whereas Trp 214 is located in the
periphery of the OmpF trimer.
Figure 5: Representations of the OmpF trimer with possible ciprofloxacin binding sites
specified (diagonal filling) for each of chosen binding models. A: Model I) binding site
for ciprofloxacin is located near the trimer interface. Quenching of Trp 61 is complete
after antibiotic binding, while the three Trp 214 are at a distance of 30 Å from
ciprofloxacin. B: Model II) binding site for ciprofloxacin is located in the trimer
periphery near one of the Trp 214 which is completely quenched after binding. The other
Trp 214 are at a distance of 52 Å from the binding site, whereas the three Trp 61 are at a
distance of 30 Å.
Figure 6: Extent of energy transfer (Eq.1) for the OmpF tryptophan-Ciprofloxacin,
donor acceptor FRET pair. Simulation for absence of binding (random distribution of
127

acceptors) (Eqs. 4-6) (⋅-⋅-⋅). A: A model assuming binding of ciprofloxacin near the
trimer interface (Model I) was fitted to the experimental data (Eqs. 4-12). The interval
of 3.15 < log(K B ) < 3.62 was recovered for the binding constant (grey area). B: A model
assuming binding of ciprofloxacin to the OmpF periphery (Model II) was fitted to the
experimental data. The interval of 3.58 < log(K B ) < 4.00 was recovered for the binding
constant (grey area). The simulations for the upper and lower bounds of K B are
represented in both figures by a solid line (⎯) and by a dotted line (⋅⋅⋅⋅⋅), respectively.
Figure A1 – Schematic side (A) and top (B) views of a lipid bilayer containing a
cylindrical membrane protein labelled with a donor fluorophore (D) at its outer radius,
capable of FRET to acceptors (A) located in the lipid bilayer, on a plane parallel to the
bilayer plane. See text for details.
128

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
129

FIGURE 1A
FIGURE 1B
130

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
FIGURE 2
FIGURE 3
131

FIGURE 4A
FIGURE 4B
132

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
FIGURE 5A
FIGURE 5B
133

FIGURE 6A
FIGURE 6B
134

BINDING OF A QUINOLONE ANTIBIOTIC TO BACTERIAL
PORIN OmpF
FIGURE A1
135

136

INTERACTION OF HELIX-0 OF THE N-BAR DOMAIN
WITH LIPID BILAYERS
V
INTERACTION OF HELIX-0 OF
THE N-BAR DOMAIN WITH
LIPID MEMBRANES
1. Introduction
Endocytosis is a crucial phenomenon in several cellular processes, such as
receptor recycling and degradation, delivery of membrane-bound and soluble cargo to
intracellular organelles, and nutrient uptake. Clathrin-mediated endocytosis is likely to
be the best characterized of all endocytic pathways, and is responsible for recycling of
proteins from the plasma membrane. It is also important in the recycling of synaptic
vesicles. A complex assemble of clathrin and several other proteins drives invagination
of the plasma membrane and ultimately fission (Dawson et al., 2006). Therefore,
clathrin mediated endocytosis can be separated in two steps. In the first step, the
membrane must bend inwards to create a vesicular structure. Following this, the vesicle
is actively pinched off from the membrane. The last step is known to be carried out by a
protein called dynamin that polymerizes upon membrane binding and arranges into
coiled coats around the neck of the nascent vesicles, twisting (through an unknown
mechanism) in response to GTP hydrolysis, leading to fission of the vesicle (McMahon
and Gallop, 2005).
In recent years, studies have been unravelling the complex network of proteins
involved in clathrin mediated endocytosis. In the end of the last decade, De Camilli and
co-workers (Takei et al, 1998; Takei et al, 1999) observed that a protein called
amphiphysin, known to interact with dynamin during endocytosis, bounds to spherical
liposomes and induces formation of tubules both in the presence and absence of
137

dynamin. Soon after, the same group observed a similar behaviour with another protein,
endophilin, that was also know to bind dynamin (Farsad et al., 2001).
Amphiphysin and endophilin both share a homologous N-terminal amino acid
stretch which was shown to be crucial for liposome tubulation activity (Farsad et al.,
2001). This domain is highly conserved in both proteins and is found in the N-terminal
of other proteins. Due to its presence in the bin, amphiphysin and Rvs161/167 proteins
(all amphiphysin family members), it was called BAR domain (Ge and Prendergast,
2000). BAR domains are responsible for the dimerization of amphiphysin in lipid
membranes and are able to induce tubulation of liposomes isolated from the rest of the
protein (Farsad et al., 2001). The structure of the BAR domain of amphiphysin was
solved by McMahon and co-workers (Peter et al., 2004). It is a crescent-shaped dimer,
and each monomer has three kinked α-helices (Figure V-1).
Figure V-1: Crystal structure of the BAR domain of amphiphysin from Drosophila (taken from
Peter et al., 2004).
McMahon and co-workers also observed increased amphiphysin binding
efficiency to liposomes of higher curvature, and suggested that this was due to a better
fitting of the concave surface of the BAR domain dimer to the liposome surface. When
a short N-terminal stretch predicted to be an amphipatic helix (Helix-0 or H0) was
removed, liposome binding efficiencies decreased. Tubulation, however, was still
detected, albeit much less efficient. This N-terminal stretch is not present in all BAR
domains, and BAR domains containing it are called N-BAR domains (Peter et al., 2004;
Gallop and McMahon, 2005). The authors also suggested that the function of this N-
terminal segment was to insert into the bilayer as an amphipatic helix. This insertion
would lead to an asymmetry between the inner and outer leaflet causing an increase of
positive curvature, whereas the concave surface of the BAR domain acts as a scaffold to
the membrane curvature (Gallop et al., 2006; Zimmerberg and Kozlov, 2006). In fact,
the rigidity of the crescent dimer shape is crucial for liposome tubulation (Masuda et al.,
138

INTERACTION OF HELIX-0 OF THE N-BAR DOMAIN
WITH LIPID BILAYERS
2006). The N-terminal amphipatic segments were also proposed to act as mediators of
oligomerization between BAR domain dimers, increasing the efficiency of the
tubulation process (Richnau et al., 2004; Gallop and McMahon, 2005)
A model for the function of amphiphysin and other BAR proteins involved in
clathrin mediated endocytosis is represented in Figure V-2. Amphiphysin can act in
clathrin mediated endocytosis by binding to the nascent vesicle on the early stages of
invagination, either through the BAR domain itself or through interaction with other
proteins. Following this, the BAR domain can contribute to the generation of the bud
neck with the required dimensions, and as the pit matures, more amphiphysin is
recruited to the vesicle neck due to its curvature sensing properties. Dynamin is
assembled together with amphiphysin and GTP hydrolysis leads to vesicle fission
(Yoshida et al., 2004).
Figure V-2: Possible mechanism of action of amphiphysin in clathrin mediated endocytosis.
Coat components (clathrin) were omitted (taken from Yoshida et al., 2004).
139

140

INTERACTION OF HELIX-0 OF THE N-BAR DOMAIN
WITH LIPID BILAYERS
2. ROLE OF HELIX-0 OF THE N-BAR DOMAIN
IN MEMBRANE CURVATURE GENERATION
141

142

Abstract
A group of proteins with cell membrane remodeling properties are also able to change
dramatically the morphology of liposomes in vitro, frequently inducing tubulation. For a
number of these proteins, the mechanism by which this effect is exerted has been proposed to
be the embedding of amphipathic helices into the lipid bilayer. For proteins presenting BAR
domains, removal of a N-terminal amphipathic alpha-helix (H0-NBAR) results in much lower
membrane tubulation efficiency, pointing to a fundamental role of this protein segment. Here,
we studied the interaction of a peptide corresponding to H0-NBAR with model lipid
membranes. H0-NBAR bound avidly to anionic liposomes but partitionate weakly to
zwitterionic bilayers, suggesting an essentially electrostatic interaction with the lipid bilayer.
Interestingly, it is shown that after membrane incorporation, the peptide oligomerizes as an
antiparallel dimer, suggesting a potential role of H0-NBAR in the mediation of BAR domain
oligomerization. Through monitoring the effect of H0-NBAR on liposome shape by electron
microscopy it is clear that membrane morphology is not radically changed. We conclude that
H0-NBAR alone is not able to induce vesicle curvature, and its function must be related to the
promotion of the scaffold effect provided by the concave surface of the BAR domain.
Keywords: BAR proteins, membrane tubulation; amphipatic peptide; FRET; fluorescence
2

Introduction
Control of membrane remodeling is essential in clathrin-mediated endocytosis (CME)
as different types and levels of curvature are required at each stage of the budding of clathrincoated
vesicles [1,2]. Several of the proteins thought to play relevant roles in CME (dynamin,
amphiphysin, endophilin and epsin) were recently shown to induce tubulation in protein-free
spherical liposomes, indicating a potential role as mediators in the membrane remodeling
observed during CME (3, 4, 5 6).
The BAR (Bin, amphiphysin, Rvs) domain, found in amphiphysin, endophilins, and a
wide variety of other proteins with or without known function in CME (7), is able to bind lipid
membranes, generate tubulation (both in vivo and in vitro) and to sense bilayer curvature (5, 8).
This versatile domain has a banana shape and dimerizes in membranes, giving rise to a
positively charged concave surface that binds to lipid bilayers. This concave surface is likely to
be the reason why the domain presents higher affinities for high curvature liposomes in vitro.
Several BAR domains also present an N-terminal sequence that forms an amphipathic
helix upon membrane binding (9). This sequence is here referred to as helix 0 (H0-NBAR).
BAR domains presenting this sequence are called N-BAR and are able to bind to liposomes
and induce tubulation with much higher efficiency, even tough sensitivity for curvature is lost
(8). After a point mutation in H0-NBAR of a conserved hydrophobic residue (F) to an acidic
residue (E), lipid binding and tubulation were abolished for endophilin (5), and reduced for the
corresponding mutation in amphiphysin1 (8). Conservative mutations of the same residue (F to
W) had no effect (5). These results point to an important role of H0-NBAR in membrane
remodeling by N-BAR domains, and this role is likely to be dependent on membrane
embedding of H0-NBAR. The exact function of H0-NBAR in membrane tubulation is however
still elusive.
The H0 fragment from BRAP (breast-cancer-associated protein)/Bin2, one of the first
BAR domain containing proteins to be identified (10, 11), presents great homology to other N-
terminal amphipathic fragments of BAR domain-containing proteins (Fig.1). The N-BAR
domain of BRAP was already shown to tubulate liposomes in an identical fashion to other N-
BAR domains. The mutations performed on the N-BAR domain from BRAP had also
analogous effects in lipid tubulation as the corresponding mutants of the N-BAR domain of
amphiphysin, corroborating that the same liposome tubulation mechanism was shared by the
two proteins. Here we investigated the interaction of a peptide comprising the H0-NBAR
fragment of BRAP with model lipid membranes. We performed a thorough study of the effects
of partition of the N-BAR N-terminal domain to lipid membranes on both structure and
dynamics of H0-NBAR itself and the interacting lipid membranes. We show that the N-
terminal fragment of the N-BAR domain assumes in effect an alpha-helical structure upon
membrane binding and that membrane binding is dependent on the presence of anionic
phospholipids but virtually insensitive to both anionic lipid structure and liposome curvature.
Through FRET (Förster resonance energy transfer) it is demonstrated that H0-NBAR dimerizes
after incorporation in lipid membranes, providing a possible mechanism for generation of highorder
oligomers of N-BAR domains. Monitoring the fluorescence of different membrane
probes, membrane insertion of H0-NBAR is show to increase the packing of lipids both in the
hydrophobic and headgroup regions of the bilayer. Finally, we also find that H0-NBAR is
effective in inducing liposome fusion but has no liposome tubulation activity. Our results rule
out the insertion of H0-NBAR in the exposed outer membrane leaflet as the mechanism of
tubulation induced by N-BAR domains and point to a likely interplay between the membrane
binding of H0-NBAR and the scaffold provided by the concave surface of the BAR domain.
3

Methods
Materials
Peptides H0-NBAR, H0-NBAR-EDANS(5-((2-aminoethyl)amino)naphthalene-1-
sulfonic acid), H0-NBAR-FITC(fluorescein isothiocyanate), and H0-ENTH(Epsin N-terminal
homology domain) were synthesized by Genemed Synthesis (San Francisco, CA). Labeling
was achieved by conjugation on N-terminal end of the peptide. The purity was always higher
than 95%.
1-Palmitoyl-2-oleoyl-sn-phosphocholine (POPC), 1-Palmitoyl-2-oleoyl-sn-glycero-3-
[Phospho-rac-(1-glycerol)] (POPG), 1-Palmitoyl-2-oleoyl-sn-glycero-3-(phospho-l-serine)
(POPS), 1,2-Dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-
yl) (NBD-DOPE), 1-Oleoyl-2-[12-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl]-snglycero-phosphocholine
(PC-NBD), 1-Oleoyl-2-[12-[(7-nitro-2-1,3-benzoxadiazol-4-
yl)amino]dodecanoyl]-sn-glycero-3-[phospho-rac-(1-glycerol)] (PG-NBD), 1-Oleoyl-2-[12-
[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl]-sn-glycero-phosphoserine (PS-NBD),
1-Oleoyl-2-[12-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl]-sn-glycero-phosphate
(PA-NBD), 1,2-Dioleoyl-sn-glycero-3-phosphoethanolamine-N-(Lissamine Rhodamine B
Sulfonyl) (Rho-DOPE) were from Avanti Polar-Lipids (Alabaster, AL).
1,6-Diphenyl-1,3,5-hexatriene (DPH) and 5,6-carboxyfluorescein (5,6-CF) were from
Molecular Probes (Eugene, OR). Brain extract from bovine brain (Folch fraction I) was from
Sigma-Aldrich (St. Louis, MO).
Fine chemicals were obtained from Merck (Darmstadt, Germany). All materials were used
without further purification.
Liposomes preparation
The desired amount of phospholipids were mixed in chloroform and dried under a N 2(g)
flow. The sample was then kept in vacuum overnight. Liposomes were prepared with buffer
Hepes 20 mM, NaCl 150mM pH 7.4. The hydration step was performed with gentle addition
of buffer followed by freeze-thaw cycles. Large unilamellar vesicles (LUV) were produced by
extrusion through polycarbonate filters (12) in a Avestin extruder. Liposomes with 5,6-CF
encapsulated were prepared by hydration with buffer Hepes 20 mM, NaCl 150mM pH 8.4 with
5mM 5,6-CF. After extrusion the suspension was passed through a 10ml Econo-Pac 10DG
column Bio-Gel P-6DG gel with 6 kDa molecular weight exclusion) and eluted with buffer
Hepes 20 mM, NaCl 150mM pH 8.4.
CD spectroscopy
CD spectroscopy was performed on a Jasco J-720 spectropolarimeter with a 450 W Xe
lamp. Lipid suspensions were extruded using polycarbonate filters of 0.1 μm. Peptide
concentration was 40 μM.
Fluorescence Spectroscopy
4

Steady-state fluorescence measurements were carried out with an SLM-Aminco 8100
Series 2 spectrofluorimeter described in detail elsewhere (13).
Fluorescence anisotropies were determined as described in (14). Time-resolved
fluorescence measurements of H0-BAR-EDANS were carried out with a time-correlated
single-photon timing system, which is also described elsewhere (13). For steady-state
fluorescence anisotropies and time resolved fluorescence measurements with H0-NBAR-
EDANS, the excitation and emission wavelengths were 340 and 460 nm, respectively. For
time-resolved fluorescence measurements with H0-NBAR-FITC, the excitation and emission
wavelengths were 470 and 525 nm, respectively. Analysis of fluorescence intensity and
anisotropy decays were carried out as previously described (15,16). All measurements were
performed at room temperature.
Transmission electron microscopy (TEM)
TEM was carried out using a Jeol (Tokyo, Japan) Electron Microscope 100SX,
operated at 60kV. Samples were placed over copper grids covered with carbon membrane.
Negative staining was performed with uranyl acetate 1% in water. Total lipid concentration of
the mixtures was varied (ranging from 0.05 mM to 0.5 mM).
Results
The N-terminal segment of the N-BAR domain is 100% alpha-helical in anionic
liposomes.
CD measurements were performed on the unlabeled H0-NBAR peptide while in the
absence of liposomes, in the presence of zwitterionic liposomes (POPC) and anionic liposomes
(POPG) (Fig.2). CD measurements of the labeled H0-NBAR peptides (both H0-NBAR-
EDANS and H0-NBAR-FITC) produced the same results, confirming that labeling did not
change the structure of H0-NBAR (results not shown). Deconvolution of the obtained spectrum
with the CDNN CD Spectra Deconvolution v. 2.1 software revealed a predominantly
unstructured peptide in the absence of lipids and in the presence of 1mM of POPC. After
addition of 1mM of POPG the peptide presented more than 80 % alpha-helical structure,
compatible with the expected amphipathic nature of the N-terminal segment of the N-BAR
domain while interacting with lipid membranes.
Interaction of H0-NBAR with lipid bilayers is only dependent on lipid charge.
The fluorescence emission spectrum from H0-NBAR-EDANS is practically unaffected
upon interaction with lipid bilayers, as the wavelength of maximum emission intensity remains
the same and only a small broadening in the lower wavelength range of the spectrum is visible
(results not shown). EDANS emission spectrum is extremely sensitive to the environment (17)
and this result is an indication that the EDANS fluorophore is located in the hydrophilic face of
the amphipathic helix and that it remains fully exposed to the aqueous environment. There was
however a subtle difference in quantum yields (20% higher in the presence of anionic
phospholipids) and a significant change in fluorescence anisotropy, from = 0.022 in buffer
to > 0.06 in the presence of anionic phospholipids. This change in fluorescence anisotropy
5

is the result of the immobilization of the peptide in the lipid bilayer and can be used to quantify
the partition of H0-NBAR to bilayers from equation 1 (18),
where W is the fluorescence anisotropy in water, L is the fluorescence anisotropy in the
bilayer, φ W is the fluorescence quantum yield in water, φ L is the fluorescence quantum yield in
the membrane, γ L is the molar volume of the lipid, [L] is the lipid concentration and K p is the
lipid/water partition coefficient.
Figure 3A shows the dependence of on the lipid concentration for different
phospholipids. It is clear that partition to POPC liposomes is extremely low when compared to
the partition observed for both anionic phospholipids POPS and POPG.
The results of fitting Equation 1 to the data in Figure 3 are presented in Table I. Partition of
H0-NBAR to POPS and POPG liposomes is identical suggesting that the nature of the anionic
phospholipid is irrelevant for H0-NBAR interaction with liposomes. Further evidence for the
absence of a specific interaction with a class of lipids, can be obtained from a FRET assay,
where lipids (e.g. PS, PA, and PG) are derivatized with an acceptor (NBD) for Förster
resonance energy transfer (FRET) from EDANS, and dispersed (2% mol/mol concentration) in
a POPG matrix.
FRET efficiencies (E) are obtained from the extent of fluorescence emission quenching
of the donor induced by the presence of acceptors,
[ ]
[ L]
φw⋅ < r > + K ⋅γ
⋅ L ⋅φ
⋅< r >
< r >=
1+ K ⋅γ
⋅
w P L L L
P
L
(1)
IDA
E = 1− = 1−
I
D
∞
∫
DA
0
∞
0
i
∫
i
D
() t
() t
(2)
where I DA and I D are the the steady-state fluorescence intensities of the donor in the presence
and absence of acceptors respectively, and i DA (t) and i D (t) are the donor fluorescence decays in
the presence and absence of acceptors respectively.
In case there was a specific interaction between H0-NBAR-EDANS and the NBDlabeled
phospholipid, the peptide would be incorporated in the vicinity of that lipid and FRET
would increase. Since there is also liposome fusion induced by H0-NBAR-EDANS (see later),
the available two dimensional FRET formalisms may not be applicable for retrieving
quantitative information about H0-NBAR-EDANS selectivity for a specific phospholipid (19),
and a more qualitative approach was used. The results are shown in the inset of Fig. 3A. It is
clear that no significant difference exists between E obtained with any of the anionic
phospholipids. The small differences in FRET efficiencies are a result of the lack of selectivity
of H0-NBAR-EDANS for specific phospholipids.
Partition of H0-NBAR-EDANS to lipid membranes is also insensitive to the
dimensions of the liposomes and hence insensitive to the degree of curvature of the bilayer in
this range of liposome size (Fig. 3B). Even though the values of K p recovered for larger
liposomes are slightly higher (as seen in Table I), the differences are within the errors of the
fits and thus non significant.
Residual partition to POPC liposomes was detected. After long incubation times
(overnight), an increase of fusion of POPC liposomes loaded with H0-NBAR was evident
relative to blank POPC vesicles. This was clear from a FRET experiment carried out with a
mixture of liposomes loaded with 2% NBD-DOPE (donor) and liposomes loaded with 2%
6

Rho-DOPE (acceptor). The increase of FRET efficiencies after the mixture reports fusion of
the liposomes. H0-NBAR stimulated fusion for both POPG (considerably) and POPC
liposomes (slightly) (Fig. 4) and was more effective in doing so than a control peptide (H0-
ENTH), corresponding to the helix 0 of the epsin N-terminal homology domain that is also
related to membrane curvature induction (albeit in the case of epsin the presence of
phosphatidylinositol-(4,5)-bisphosphate PIP(4,5) 2 is required for membrane remodeling) [20].
It was checked that under these conditions H0-ENTH is completely bound to liposomes
(results not shown).
H0-NBAR is an antiparallel dimer in the membrane
The function of H0-NBAR in liposome tubulation mediated by the BAR domain was
recently proposed to be related to BAR domain oligomerization (7).
With the intent to verify if H0-NBAR oligomerizes in the membrane, FRET
measurements were performed using H0-NBAR-EDANS as a donor and H0-NBAR-FITC as
an acceptor (emission and absorption spectra in Supplementary Data).
For this system, the high Förster radius (R 0 ) of the EDANS-FITC donor-acceptor pair
(R 0 (EDANS-FITC) = 40 Å) entails a small contribution of energy transfer between nonoligomerized
H0-NBAR and this was taken into account in our analysis (see FRET simulation
for the monomeric hypothesis in Fig.5A). The donor fluorescence decay in the presence of
acceptors is described as,
n−1
nk -
i () t =
⎡
f ( k)ρ () t
⎤
()ρ
DA
⎢∑
⋅ ⋅i t ⋅
Dq nonbound
() t +
bound D
⎣
⎥
k=
1
⎦
n−1
⎛
1 f ( k) ⎞
⎜ −∑
i ( t) ρ t
Dq ⎟⋅ ⋅
D nonbound
⎝ k=
1 ⎠
()
[3]
where ρ bound is the FRET contribution from energy transfer to each acceptor in the oligomer
containing the donor, ρ nonbound is the FRET contribution arising from energy transfer to
acceptors that are not integrated in the same oligomer as the donor, and f Dq (k) is the fraction of
donors bound to k acceptors,
Here n is the number of units in the oligomer, k counts the number of donors, P D is the
mole ratio of donors and P A is the mole ratio of acceptors.
Through measurement of the dependence of FRET efficiencies on the acceptor/donor
ratio it is possible to conclude about the presence of oligomerization and to gather information
on the type of oligomerization found (21). The contribution for FRET from each acceptor in
the same oligomer as the donor is given by,
ρ
n
( k ) D
( ) k n k
fDq k = k⋅ ⋅P
P
⎛
6
1 ⎛ R ⎞
0
bound
= exp
−
⎜
⋅⎜ ⎟
τ0 rD-A
⎝
−
A
⎝
⎞
⎠ ⎟
⎠
[4]
[5]
7

where τ 0 is the lifetime of the donor in the absence of acceptors, and r D-A is the distance
between the donor and the acceptor inside the oligomer.
Equation 6 gives ρ nonbound for a two dimensional system (15).
nonbound 2
[
ρ = exp −n
⋅π
⋅Γ(2 / 3) ⋅
R
2
0
⎛ t ⎞
⋅⎜ ⎟
⎝τ
0 ⎠
1
3
]
[6]
where n 2 is the acceptor density in each bilayer leaflet, and Γ is the gamma function.
From Eqs. 2-6, simulations for FRET efficiencies are obtained (Fig. 5A). In the FRET
analysis, the different oligomerization models were fitted to two series of data simultaneously.
In the first series the lipid to protein ratio (L/P) equals 1000 while in the second the protein was
diluted two fold (L/P = 2000). With this methodology, uncertainty resulting from erroneous
quantification of the two FRET contributions (to acceptors within the same oligomers and to
unbound acceptors) is eliminated. The oligomerization model that provided the best fit to the
two data series was one that considered dimerization of the peptide and a distance of 43 Å
between donor and acceptor inside the dimer. This distance is in good agreement with an
antiparallel dimer as the size of a rigid and fully alpha-helical H0-NBAR is expected to be
around 49.5 Å (1.5 Å per residue). The sensitivity of the methodology chosen is clear from the
comparison with the simulations for FRET efficiencies arising from a parallel dimer (r D-A < 20
Å) and other oligomerization numbers (see Fig. 5A).
The oligomerization of H0-NBAR is dependent on membrane binding, as the peptide
exists as a monomer while in buffer. This was concluded from anisotropy decays of H0-
NBAR-FITC (Supplementary Data). The anisotropy decays were well described by two
rotational correlation times, φ 1 and φ 2 .
rt ()
⎡ ⎛−t
⎞
r ⋅ ⎢ β ⋅ exp⎜ ⎟+
⎣
0 1
= ⎝ φ1
⎠
⎛−t
⎞⎤
β2
⋅exp⎜
⎟⎥
⎝φ2
⎠⎦
[7]
Here r 0 is the fundamental anisotropy, β 1 and β 2 are the component amplitudes. The
methodology for analysis of time-resolved anisotropy decays is described in detail elsewhere
(16).
φ 1 was typically smaller than 250 ps and is expected to be related to independent and
fast movement of the FITC fluorophore. The recovered φ 2 values were between 1,45 and 1,69
ns (Fig. 5B) and are expected to correspond to the motion of the entire peptide. The rotational
correlation time can be estimated from the following equation (strictly valid for spherical
molecules),
η ⋅V
φ=
R ⋅T
[8]
8

where η is the viscosity of the medium, V is the molar volume of the molecule, R is the ideal
gas constant and T is the temperature. From Equation 8 and taking into consideration the
hydration volume for a protein (22) the expected φ 2 of monomeric H0-NBAR is 1.54 ns (Fig.
5B). Hence, H0-NBAR dynamics in solution is as expected for a monomer and is independent
of concentration up to 20 μM of peptide.
H0-NBAR does not translocate efficiently across the bilayer
When bound to the full N-BAR domain, H0-NBAR is expected to interact with only
one monolayer. The structural changes induced in the exposed monolayer and the resulting
monolayer asymmetry could be responsible for membrane bending mediated by N-BAR, in a
similar mechanism as the one observed for the ENTH domain of epsin after PIP(4,5) 2 binding
(23). In order to study if the H0-NBAR is able to induce by itself the sort of membrane bending
observed with full N-BAR domains it is necessary to determine if the peptide, after interaction
with anionic liposomes, remains bound to the outer monolayer, or if it exhibits fast
translocation across the bilayer. If H0-NBAR translocated efficiently across the bilayer, effects
on both monolayers would limit monolayer asymmetry and consequently H0-NBAR would not
behave in a comparable manner as the helix 0 in the N-terminal segment of N-BAR.
To evaluate this, the following methodology was applied. The iodide ion exhibits low
permeability across lipid bilayers (Fig. 6- Inset), and by measuring the degree of fluorescence
quenching induced by I - on H0-NBAR-EDANS it is possible to estimate the fraction of H0-
NBAR-EDANS that translocated across the bilayer. Two sets of samples were measured. In the
first, 1mM POPG liposomes were incubated with 2μM of H0-NBAR-EDANS for only 3
minutes, as this amount of time guarantees almost complete binding of the peptide to the
membrane (results not shown). In the second set of samples, the same concentrations of POPG
liposomes and H0-NBAR-EDANS were incubated for 40 min. The fluorescence intensities
were measured immediately before and after addition of KI and the Stern-Volmer plots for
fluorescence quenching were obtained (Fig. 6). It is clear that the difference between the
degree of exposure of H0-NBAR to KI is minimal between the two data sets. Assuming that
H0-NBAR-EDANS in the inside of vesicles was 100% inaccessible to KI, after 40 minutes of
incubation, there was only a maximum of 4 % translocated H0-NBAR.
H0-NBAR increases packing of anionic phospholipids at a higher degree than a control
peptide
The effect of unlabeled H0-NBAR in the packing and dynamics of phospholipids was
measured through the fluorescence anisotropy of two lipid membrane probes, NBD-DOPE and
DPH. The first is a good probe for the headgroup region of the bilayer while the second is a
well known probe of acyl chain packing. The effects of adding increasing amounts of H0-
NBAR to POPG liposomes loaded with 1 % (mol/mol) of fluorescent probe are shown in
Figure 7. The results obtained with H0-ENTH are also shown for comparison. ENTH is
thought to be able to tubulate liposomes in the presence of PIP(4,5) 2 through insertion of H0-
ENTH in the acyl-chain. In the absence of this lipid, H0-ENTH is unable to penetrate the
membrane (6). It should be stressed that the induced probe anisotropy increase is not related to
a decrease of its lifetime (Perrin equation, [14]), since the probe fluorescence intensity is
invariant (there is no fluorescence quenching), upon peptide addition (results not shown).
H0-NBAR clearly rigidifies both the acyl-chains and the headgroup region of the
bilayer. H0-ENTH has only minor effects on the fluorescence anisotropy of both probes even
9

at very high concentrations, confirming that H0-NBAR is particularly rigidifying for
phospholipid packing.
H0-NBAR is not able to tubulate lipid membranes in the absence of the scaffold domain
of N-BAR.
TEM was used to monitor the effects of H0-NBAR on liposome morphology (Fig. 8).
H0-NBAR is clearly not able to induce tubulation of liposomes composed of pure synthetic
phospholipids (POPG) (Fig. 8) or of lipid brain extracts (results not shown). The only
noticeable effect of H0-NBAR was the fusion of the liposomes, without a significant change in
their morphology or size. Different P/L ratios produced similar results.
Discussion
Proteins are believed to be able to induce membrane bending by means of three
mechanisms, namely the scaffold, local spontaneous curvature and the bilayer-couple
mechanism (24). In the scaffold mechanism, proteins present and force their intrinsic curvature
to the lipid membrane. This intrinsic curvature can be the result of tertiary structure or of the
surface from a protein network. In the local spontaneous curvature hypothesis, a shallow
insertion in the lipid membrane of an amphipathic moiety from a protein induces local
perturbation of the packing of lipid headgroups resulting in higher local curvature. Finally in
the bilayer-couple mechanism, the insertion of an amphipathic helix in the lipid bilayer could
result in an increase of the area of the monolayer where the protein is inserted that is
compensated by an increase of bilayer curvature.
The presence of H0-NBAR greatly enhances the liposome tubulation activity of BAR
domains. Several theories have been recently presented to explain the relevance of H0-NBAR
in this phenomenon. It is feasible that the sole function of H0-NBAR is the increase in
residence time of the BAR domain in the membrane by tight binding to the hydrophobic
interior of the bilayer, but it is also possible that this segment is important in the
oligomerization of BAR domains in the membrane, since cross-linking experiments suggest
that BAR domains exist in membranes as high-order oligomers [5, 25].
Therefore, N-BAR domains could induce remodeling of membranes by the scaffold, the
local spontaneous curvature and the bilayer-couple mechanisms or by a combination of them.
The concave shape of the BAR dimer can act as a scaffold, forcing the membrane curvature to
adapt to its own curvature. Insertion of the amphipathic helix could also act through either the
bilayer-couple or the local spontaneous curvature mechanism.
Here we showed that insertion of H0-NBAR in the lipid bilayer is unable to induce
significant changes in lipid membrane morphology. Recently, some authors reported that
highly amphipathic peptides induced liposomes or supported lipid bilayers to adopt tubular
structures when present at very high concentrations (L/P

ound conformation of the BAR domain through strong hydrophobic interactions with the
membrane. Although partition of H0-NBAR to anionic liposomes is very efficient, it is
unlikely to be higher than the partition of the BAR domain itself, as the concave surface of the
BAR domain dimer is already strongly positively charged. Partition of H0-NBAR to anionic
bilayers disturbs both the headgroup as the acyl-chain regions of the bilayer. H0-NBAR is
more disturbing to the bilayer than H0-ENTH, and this can be the result of H0-NBAR
dimerization in the membrane environment, as dimerization of amphipathic peptides was
previously shown to be particularly disturbing to bilayer structure (28). The dimerization of
H0-NBAR explains the detection of high-order oligomers of N-BAR domains (5,25), and if
effective in the full BAR domain, can be the mechanism by which H0-NBAR provides a more
favorable framework for tubulation mediated by N-BAR (7).
A recent molecular dynamics study (29) showed binding of N-BAR domains to a lipid
membrane resulting in generation of membrane curvature through the scaffold mechanism.
Results suggested that the main role of the N-terminal amphipathic helix of N-BAR was to
favor the orientation of the N-BAR domain that allowed direct interaction between the
membrane and the protein concave face. In effect, from our result, it is clear that the membrane
bending activity of the N-BAR domain must be achieved through the scaffold mechanism in
which the BAR domain presents its concave surface to the membrane, and forces the
membrane to adopt the same curvature, while H0-NBAR only plays a promoting role in this
process, either by: i) enhancing the membrane affinity of the full N-BAR domain; ii) mediating
N-BAR high order oligomerization and stimulating the increase of local concentrations of the
protein; iii)increasing lipid packing and facilitating curvature generation; iv) forcing the protein
to present its concave face to the membrane or by a combination of these mechanisms.
Acknowledgments
F. F. acknowledges grant SFRH/BD/14282/2003 from FCT (Portugal). A.F. acknowledges
grant SFRH/BPD/26150/2005 from FCT (Portugal) This work was funded by FCT
(Portugal) under the program POCI.
11

References
1. Farsad, K., and P. De Camilli. 2003. Mechanisms of membrane deformation. Curr.
Opin. Cell Biol. 15:372-381.
2. McMahon, H. T., and J. L. Gallop. 2005. Membrane curvature and mechanisms of
dynamic cell membrane remodeling. Nature 438:590-596.
3. Takei, K., V. Haucke, V. Slepnev, K. Farsad, M. Salazar, H. Chen, and P. De Camilli.
1998. Generation of coated intermediates of clathrin-mediated endocytosis on protein-free
liposomes. Cell 94:131-141.
4. Takei, K., V. I. Slepnev, V. Haucke, P. De Camilli. 1999. Functional partnership
between amphiphysin and dynamin in clathrin-mediated endocytosis. Nat. Cell Biol. 1:33-
39.
5. Farsad, K., N. Ringstad, K. Takei, S. R. Floyd, K. Rose, and P. De Camilli. 2001.
Generation of high curvature membranes mediated by direct endophilin bilayer
interactions. J. Cell Biol. 155:193-200.
6. Ford, M. G. J., I. G. Mills, B. J. Peter, Y. Vallis, G. J. K. Praefcke, P. R. Evans, and H.
T. McMahon, 2002. Curvature of clathrin-coated pits driven by epsin. Nature 419:361-
366.
7. Gallop, J. L., and H. T. McMahon. 2005. BAR domains and membrane curvature:
bringing your curves to the BAR. Biochem. Soc. Symp. 72:223-31.
8. Peter, B. J., H. M. Kent, I.G. Mills, Y. Vallis, P. J. G. Butler, P. R. Evans, and H. T.
McMahon. 2004. BAR domains as sensors of membrane curvature: the amphiphysin BAR
structure. Science 303:495-499.
9. Gallop, J. L., C. C. Jao, H. M. Kent, P. J. G. Butler, P. R. Evans, R. Langen, and H. T.
McMahon. 2006. Mechanism of endophilin N-BAR domain-mediated membrane
curvature. EMBO J. 25:2898-2910.
10. Ge, K., and G. C. Prendergast. 2000. Bin2, a functionally nonredundant member of the
BAR adaptor gene family. Genomics 67: 210-220.
11. Greiner, J., M. Ringhoffer, M. Taniguchi, T. Hauser, A. Schmitt, H. Döhner, and M.
Schmitt. 2003. Characterization of several leukemia-associated antigens inducing humoral
immune responses in acute and chronic myeloid leukemia. Int. J. Cancer 106:224-231.
12. Mayer, L.D., M. J. Hope, and P. R. Cullis. 1986. Vesicles of variable sizes produced
by a rapid extrusion procedure. Biochim. Biophys. Acta 858:161-168.
13. Loura, L. M. S., A. Fedorov, and M. Prieto. 2001. Fluid-fluid membrane
microheterogeneity: a fluorescence resonance energy transfer study. Biophys. J. 80: 776-
788.
14. Lakowicz J.R. 1999. Principles of fluorescence spectroscopy. Kluwer
Academic/Plenum Publishers, NY.
15. Loura, L.M.S., A. Fedorov, and M. Prieto. 1996. Resonance energy transfer in a model
system of membranes: application to gel and liquid crystalline phases. Biophys. J.
71:1823-1836.
16. de Almeida, R. F. M. , L. S. M. Loura, M. Prieto, A. Watts, A. Fedorov, and F. J.
Barrantes. 2006. Structure and dynamics of the γ-M4 transmembrane domain of the
acetylcholine receptor in lipid bilayers: insights into receptor assembly and function Mol.
Membr. Biol. 23: 305-315.
17. Hudson, E. N., and G. Weber. 1973. Synthesis and characterization of two fluorescent
sulfhydryl reagents. Biochemistry 12:4154-4161.
18. Santos, N. C., M. Prieto, and M. A. R. B. Castanho. 2003. Quantifying molecular
partition into model systems of biomembranes: an emphasis on optical spectroscopic
methods. Biochimica et Biophysica Acta 1612: 123-135.
12

19. Fernandes, F., L. M. S. Loura, R. Koehorst, R. B. Spruijt, M. A. Hemminga, A.
Fedorov, and M. Prieto. 2004. Quantification of protein-lipid selectivity using FRET:
application to the M13 major coat protein. Biophys. J. 87:344-352.
20. Itoh, T., K. Koshiba, T. Kigawa, A. Kikuchi, S. Yokoyama, and T. Takenawa. 2001.
Role of the ENTH domain in phosphatidylinositol-4,5-bisphosphate binding and
endocytosis. Science 291:1047-1051.
21. Adair, B. D., and D. M. Engelman. 1994. Glycophorin A helical transmembrane
domains dimerize in phospholipid bilayers: a resonance energy transfer study.
Biochemistry 33:5539-5544.
22. Durchschlag, H., and P. Zipper. 2001. Comparative investigations of biopolymer
hydration by physicochemical and modeling techniques. Biophys. Chem. 93:141-157.
23. Stahelin, R. V., F. Long, B. J. Peter, D. Murray, P. De Camilli, H. T. McMahon, and
W. Cho. 2003. Contrasting membrane interaction mechanisms of AP180 N-terminal
homology (ANTH) and epsin N-terminal homology (ENTH) domains. J. Biol. Chem.
278:28993-28999.
24. Zimmerberg, J., and M. M. Kozlov. 2006. How proteins produce cellular membrane
curvature. Nat. Rev. Mol. Cell Biol. 7: 9-19.
25. Richnau, N., A. Fransson, K. Farsad, and P. Aspenstrom. 2004. RICH-1 has a
BIN/Amphiphysin/Rvsp domain responsible for binding to membrane lipids and
tubulation of liposomes. Biochem. Biophys. Res. Commun. 320: 1034-1042.
26. Furuya, T., T. Kiyota, S. Lee, T. Inoue, G. Sugihara, A. Logvinova, P. Goldsmith, and
H. M. Ellerby. 2003. Nanotubules formed by highly hydrophobic amphiphilic alphahelical
peptides and natural phospholipids. Biophys. J. 84:1950-1959.
27. Domanov, Y. A., and P. K. J. Kinnunen. 2006. Antimicrobial peptides temporins B and
L induce formation of tubular lipid protrusions from supported phospholipid bilayers.
Biophys. J. 91:4427-4439.
28. Hristova, K., C. E. Dempsey, and S. H. White. 2001. Structure, location, and lipid
perturbations of melittin at the membrane interface. Biophys. J. 80:801-811.
29. Blood, P. D., and G. A. Voth. 2006. Direct observation of Bin/amphiphysin/Rvs
(BAR) domain-induced membrane curvature by means of molecular dynamics simulations
Pro. Natl. Acad. Sci. U S A. 103:15068-15072.
13

TABLE I
Membrane/Water partition coefficients (K p ) for H0-NBAR-EDANS determined from
fluorescence anisotropies (Figure 3)
Vesicles
K p
POPC * (5.6 ± 2.7) ×10 1
POPS (4.2 ± 1.0) ×10 4
POPG (diameter 30 nm) (3.2 ± 1.1) ×10 4
POPG (diameter 100 nm) (4.4 ± 0.9) ×10 4
POPG (diameter 400 nm) (5.5 ± 1.5) ×10 4
* This K p value is a lower bound assuming that the anisotropy of the peptide population in
interaction with POPC is identical as the one obtained with anionic phospholipids ( ~ 0.06).
14

Figure Legends
Fig.1: N-terminal amphipatic helix of the BAR domain of BRAP. A – Alignments of N-
terminal sequences (H0) of BRAP/Bin2 and several amphiphysins (Amph) show great degree of
homology (h-Human, d-Drosophila, r-Rat). B – Helical wheel representation of H0 from BRAP.
Numbers indicate position of the aminoacid. Hydrophobic residues are show in grey.
Fig. 2: The N-terminal sequence of the BAR domain is alpha-helical in anionic liposomes. CD
spectrum of H0 in buffer (---), in the presence of POPC (●) and POPG liposomes (―). Lipid
concentration was 1mM.
Fig. 3: Partition of H0 to bilayers is only sensitive to charge. A - Increase of fluorescence
emission anisotropy of H0-EDANS with lipid concentration. H0-EDANS in the presence of POPC
(▲), POPG (○) and POPS (■) liposomes of 100 nm diameter. INSET: Efficiencies of energy
transfer from H0-NBAR-EDANS to NBD-labeled phospholipids (PX-NBD) in POPG liposomes.
Bars show the difference in FRET efficiency (E) relative to the value obtained for PG-NBD (E PG-
NBD = 0.38). Concentration of PX-NBD in POPG bilayers was 2% (mol/mol). B - Increase of
fluorescence emission anisotropy of H0-EDANS with POPG concentration for different liposome
sizes. H0-EDANS in the presence of POPG liposomes with 30 (∆), 100 (○) and 400 (■) nm radius.
Fluorescence emission anisotropies were measured with excitation wavelength of 340 nm and
emission wavelength of 460 nm. In both panels, the lines are the fits of Equation 1 to the data, as
described in the text.
Fig. 4: H0-NBAR promotes liposome fusion in both POPC and POPG. Decrease in
fluorescence emission intensities of NBD-DOPE 15 min. after mixing of liposomes with 2% NBD-
DOPE and liposomes with 2% Rho-DOPE in the presence and in the absence of H0-NBAR.
Results obtained with H0-ENTH are shown for comparison. Fluorescence intensity in the absence
of acceptor was measured and no significant donor fluorescence bleaching was detected.
Fluorescence intensities were obtained with excitation and emission wavelengths set to 460 and
510 nm respectively.
Fig. 5: H0 forms an antiparallel dimer in POPG bilayers. A- FRET efficiencies determined
from the integration of H0-EDANS fluorescence decays in the presence of increasing fraction of
acceptors (H0-FITC) (■) at a L/P = 1000 and 2000. Simulations for FRET between monomers (···),
parallel dimers (---) or trimers (-⋅-) do not describe the data accurately. The simulation which
provided a more accurate description of the data was obtained for an antiparallel dimer (⎯) in
which EDANS and FITC are separated by 43 Å (the estimated size of H0-NBAR is 49.5 Å).
Insensitivity to L/P ratios further supports the derived formalisms. B – Dependence of the longer
rotational correlation time (φ 2 ) of the fluorescence anisotropy decay of H0-NBAR-FITC on the
concentration of peptide in buffer. The dotted line is the expected φ 2 for a H0-NBAR monomer.
The dynamics of H0-NBAR-FITC is virtually unaffected by the increase in concentration and the
recovered longer correlation time is in agreement with the rotation of a H0 monomer (see text). All
samples contained the same concentration of H0-NBAR-FITC, and different concentrations of H0-
NBAR were obtained through addition of unlabelled peptide.
Fig. 6: Translocation of H0-BAR in POPG liposomes is very slow. Stern-Volmer plots for
fluorescence emission quenching of H0-NBAR-EDANS by iodide. H0-EDANS in buffer (○), in
the presence of POPG liposomes with a 3 min. (●) and 40 min. (▲) incubation time prior to
addition of iodide quencher. Permeation of iodide across the bilayer is slow (see Inset) and after 40
min. the amount of H0-NBAR-EDANS that translocated through the bilayer is estimated to be a
maximum of 4%. INSET: Permeation of I - across POPG liposomes as measured by fluorescence
quenching of encapsulated 5,6-CF.
15

Fig. 7: H0-NBAR increases phospholipid packing. A – Effect of unlabeled H0-NBAR on the
fluorescence emission anisotropy of NBD-DOPE (•). NBD-DOPE fluorescence is sensitive to
changes in the headgroup region of the bilayer. B - Effect of unlabelled H0-NBAR on the
fluorescence emission anisotropy of DPH (•). DPH fluorescence is sensitive to changes in the acylchain
region of the bilayer. For comparison, the effect on both probes of an amphipathic peptide
which is not expected to insert in the bilayer (H0-ENTH) is also presented (○). Lines are mere
guides to the eye.
Fig. 8: H0-NBAR promotes liposome fusion but does not alter liposome morphology. A –
TEM micrograph of POPG liposomes. B – TEM micrograph of POPG liposomes after incubation
with H0-NBAR (L/P = 10).
16

FIG. 1A
FIG.1B
17

FIG. 2
18

FIG. 3A
FIG. 3B
19

FIG.4
20

FIG.5A
21

Fig. 5B
22

FIG 6.
23

FIG.7A
FIG. 7B
24

FIG. 8
A
B
25

CLUSTERING OF PI(4,5)P 2 IN FLUID PC BILAYERS
VI
CLUSTERING OF PI(4,5)P2 IN
FLUID PC BILAYERS
1. Introduction
Phosphatidylinositol-(4,5)-bisphosphate (PI(4,5)P ) (Figure VI-1) is the most
2
abundant polyphosphatidylinositide in mammalian cells. It is however only a minor
component of the plasma membrane. In human erythrocytes PI(4,5)P comprises about
2
1% of the total lipid content of the plasma membrane (McLaughlin et al., 2002),
nevertheless it is responsible for the regulation of many essential cellular functions. Its
role as a precursor of two second messengers, diacylglycerol and inositol 1,4,5-
trisphosphate is well established (van Rheenen et al., 2005), and it is now recognized to
act as signal for channel gating and for establishing sites for vesicular trafficking,
membrane remodelling and movement, and actin cytoskeletal assembly (Martin, 2001).
PI(4,5)P is also expected to act merely as a protein anchor to the plasma membrane in
2
some cases (McLaughlin, 2002).
Figure VI-1: Structure of 1,2-dioleoyl-sn-glycero-3-phosphatidylinositol-(4,5)-bisphosphate.
169

The signalling functions of PI(4,5)P
2
are performed via interactions with signalling
proteins. Several proteins with known actin regulatory properties have been shown to
interact with PI(4,5)P . N-WASP, a protein responsible for the stimulation of actin
2
nucleation is activated by binding to PI(4,5)P
2
through a short polybasic domain
(Papayannopoulos et al., 2005). Overall, it seems that free (non protein-bound)
PI(4,5)P in the plasma membrane acts as a signal for anchoring of actin to the
2
membrane (McLaughlin et al., 2002). Several proteins related to the exocytosis and
clathrin mediated endocytosis mechanisms have already been shown to specifically bind
PI(4,5)P , most notably epsin, a protein with membrane remodelling properties that was
2
shown to bind lipid membranes through interaction with PI(4,5)P (Ford et al., 2002).
2
The question of how can a single lipid species regulate so many different functions
in the cell with an apparent spatial resolution is still open. It has been hypothesized that
the distribution of PI(4,5)P in the plasma membrane is non-uniform and that pools of
2
spatially confined PI(4,5)P must exist (Martin, 2001). Several authors detected
2
evidence of cholesterol dependent localization of PI(4,5)P
2
within domains in the
plasma membrane (Pike and Miller, 1998; Liu et al., 1998; Laux et al., 2000; Botelho et
al., 2000; Kwik et al., 2003; Epand et al., 2004; Epand et al., 2005; Gokhale et al.,
2005). These observations led to the conclusion that PI(4,5)P were preferentially bound
2
to raft domains. However, some authors still disagree with the hypothesis of PI(4,5)P
2
enrichment in rafts (van Rheenen et al., 2005).
Spontaneous partition of PI(4,5)P to lipid rafts is highly unlikely, as this lipid
2
almost always presents polyunsaturated acyl-chains which are not expected to favour
incorporation in the more rigid environment found in rafts (McLaughlin et al., 2002).
One possible explanation for the detection of PI(4,5)P in these structures would be its
2
local synthesis. However, this cannot account for the levels detected, as diffusion is
expected to occur at faster rates than synthesis (Ghambir et al., 2004). The most likely
mechanism for PI(4,5)P concentration in lateral domains is that some proteins with
2
favoured partition to rafts can act as buffers, binding and passively concentrating these
lipids. A family of proteins (GMC proteins) have been identified that are likely to play
this role. GAP43, myristoylated alanine-rich C kinase substrate (MARCKS), CAP23,
170

CLUSTERING OF PI(4,5)P 2 IN FLUID PC BILAYERS
are substrates of protein kinase C that bind strongly to PI(4,5)P (Laux et al., 2000;
2
Wang et al., 2001). Although these proteins present saturated acyl-chains, this alone is
not expected to result in accumulation in rafts. It is possible that the proteins are crosslinked
via actin and this would increase significantly the preference for the cholesterol
enriched phase (McLaughlin et al., 2002). In model lipid membranes, the basic domain
of MARCKS responsible for binding to PI(4,5)P was able to sequester this
2
phospholipid into clusters through electrostatic interactions (Denisov et al., 1998; Rauch
et al., 2002; Gambhir et al., 2004).
Recently, Redfern and Gericke (2004,2005) suggested an alternative method for
PI(4,5)P clustering. According to these authors, phosphatidylinositol lipids
2
spontaneously clustered at and above physiological pH, and both in the gel and fluid
phases, when incorporated in zwitterionic bilayers. The mechanism proposed to achieve
non-uniform distribution of PI(4,5)P in the bilayer, was the establishment of hydrogen
2
bonds between phosphatidylinositol headgroups, and no external agent (cholesterol or
proteins) was required. According to the authors, the same type of clustering behaviour
was observed for phosphatidylinositol monophosphates and polyphosphates. This latter
proposal is very controversial, as the presence of large charges in the headgroups of
phosphatidylinositol phosphates must result in strong repulsion between the molecules
(Pap et al., 1995), and consequently a clustering phenomena is expected to be unlikely.
171

172

CLUSTERING OF PI(4,5)P 2 IN FLUID PC BILAYERS
2. Absence of clustering of
phosphatidylinositol-(4,5)-bisphosphate in
fluid phosphatidylcholine
173

174

Absence of clustering of phosphatidylinositol-(4,5)-
bisphosphate in fluid phosphatidylcholine
Fábio Fernandes, 1, * Luís M. S. Loura,* ,† Alexander Fedorov,* and Manuel Prieto*
Centro de Química-Física Molecular,* Instituto Superior Técnico, Lisbon, Portugal; and Centro de
Química e Departamento de Química, † Universidade de Évora, Évora, Portugal
Abstract Phosphatidylinositol-(4,5)-bisphosphate [PI(4,5)
P 2 ] plays a key role in the modulation of actin polymerization
and vesicle trafficking. These processes seem to depend
on the enrichment of PI(4,5)P 2 in plasma membrane
domains. Here, we show that PI(4,5)P 2 does not form domains
when in a fluid phosphatidylcholine matrix in the pH
range of 4.8–8.4. This finding is at variance with the spontaneous
segregation of PI(4,5)P 2 to domains as a mechanism
for the compartmentalization of PI(4,5)P 2 in the
plasma membrane. Water/bilayer partition of PI(4,5)P 2
is also shown to be dependent on the protonation state of
the lipid.—Fernandes, F., L. M. S. Loura, A. Fedorov, and M.
Prieto. Absence of clustering of phosphatidylinositol-(4,5)-
bisphosphate in fluid phosphatidylcholine. J. Lipid Res.
2006. 47: 1521–1525.
Supplementary key words PIP2 . lipid domains . fluorescence .
fluorescence resonance energy transfer
Manuscript received 13 March 2006 and in revised form 18 April 2006.
Published, JLR Papers in Press, April 21, 2006.
DOI 10.1194/jlr.M600121-JLR200
Phosphatidylinositol-(4,5)-bisphosphate [PI(4,5)P 2 ] is
found mainly in the plasma membrane, where it is a critical
regulator of several cellular functions. It plays a fundamental
role, particularly in actin polymerization and vesicle
trafficking (1, 2). These processes seem to depend on
large transient and spatially localized increases of PI(4,5)P 2
concentration, as this phospholipid constitutes only z1%
of the lipids in the plasma membrane (3).
Significant evidence indicates that the membrane
patches where localized enrichment in PI(4,5)P 2 is observed
are cholesterol-rich rafts (4–6), but partition of
these phospholipids to liquid-ordered domains is difficult
to explain, as the sn-2 acyl chain of PI(4,5)P 2 is mainly
arachidonic acid, a polyunsaturated acyl chain that is not
expected to favor partition into rafts. Local enrichment in
PI(4,5)P 2 was shown to overlap with enrichment of phosphatidylinositol-4-monophosphate
kinases in the same
membrane patches, possibly leading to localized synthesis
of the inositol (7). However, as argued previously, this effect
alone cannot explain the degree of PI(4,5)P 2 compartmentalization
observed, as diffusion away from the site of synthesis
is faster than the synthesis itself (8). However, several
proteins [myristoylated alanine-rich C-kinase substrate
(MARCKS), growth associated protein 43 (GAP43), cytoskeleton-associated
protein 23 (CAP23), and neuronal
axonal membrane protein (NAP22)] have been shown to
be responsible for the lateral sequestration of PI(4,5)P 2 to
specific domains in the membrane in a process that for some
cases was dependent on cholesterol (9–11). This phenomenon,
together with localized synthesis, provides a
plausible mechanism for PI(4,5)P 2 compartmentalization,
as large concentrations of these proteins could prevent free
diffusion of the phospholipid.
In a recent study, it was proposed that PI(4,5)P 2 compartmentalization
can be achieved simply through hydrogen
bonding between PI(4,5)P 2 head groups at or slightly
above physiological pH (corresponding to partial or complete
deprotonation of the phosphomonoester group)
(12), without the contribution of any external agent (cholesterol
or proteins). Through fluorescence resonance
energy transfer (FRET) experiments in a phosphatidylcholine
(PC) matrix in the fluid state, the authors claimed
to have detected PI(4,5)P 2 domain formation, and similar
behaviors were observed for PI(3,4)P 2 and PI(3,4,5)P 3 as
well as for phosphatidylinositol monophosphates in
another study (13). Because of the large biological relevance
of the problem and the controversy of these conclusions,
we propose in this work to investigate this issue
using a detailed experimental approach. Our results show
clearly that PI(4,5)P 2 does not form domains when in a PC
matrix in the pH range of 4.8–8.4 and that partition of
PI(4,5)P 2 to PC bilayers is largely dependent on the pH.
MATERIALS AND METHODS
POPC, 1-palmitoyl-2-oleoylphosphatidylglycerol (POPG), 1,2-
dipalmitoylphosphatidycholine (DPPC), and 7-nitro-2-1,3-ben-
Abbreviations: DPH, diphenylhexatriene; DPPC, 1,2-dipalmitoylphosphatidycholine;
FRET, fluorescence resonance energy transfer;
NBD, 7-nitro-2-1,3-benzoxadiazol; PC, phosphatidylcholine; PI(4,5)P 2 ,
phosphatidylinositol-(4,5)-bisphosphate; POPG, 1-palmitoyl-2-oleoylphosphatidylglycerol.
1 To whom correspondence should be addressed.
e-mail: fernandesf@ist.utl.pt
Downloaded from www.jlr.org by on September 3, 2007
Copyright D 2006 by the American Society for Biochemistry and Molecular Biology, Inc.
This article is available online at http://www.jlr.org Journal of Lipid Research Volume 47, 2006 1521

zoxadiazol (NBD)-PC were obtained from Avanti Polar Lipids
(Birmingham, AL). NBD-PI(4,5)P 2 was from Echelon Biosciences
(Salt Lake City, UT). Diphenylhexatriene (DPH) was obtained
from Molecular Probes (Leiden, The Netherlands). Other
fine chemicals were from Merck (Darmstadt, Germany).
Liposome reconstitution procedure
The desired amounts of phospholipids were mixed in chloroform-methanol
(1:2) and dried under an N 2(g) flow. The sample
was then kept in a vacuum overnight. Liposomes were prepared
with 20 mM HEPES, 100 mM NaCl buffer at pH 8.4 and 7.1 and
with 20 mM sodium citrate, 100 mM NaCl buffer at pH 4.8. The
hydration step was performed with the addition of buffer followed
by freeze-thaw cycles. Anisotropy measurements were performed
with large unilamellar vesicles produced by extrusion
through polycarbonate filters with a pore size of 100 nm (14).
to quantify the effective fraction of NBD-PI(4,5)P 2 incorporated
in liposomes. NBD quantum yield decreases with
hydration, and clustering of NBD-PI(4,5)P 2 leads to NBD selfquenching
(20). Consequently, it was not surprising that
NBD-PI(4,5)P 2 in buffer (micellar state) exhibited quantum
yields z100 times smaller than when incorporated in PC
liposomes (lamellar state) (Fig. 1). This difference allowed
us to use fluorescence intensity as a tool to estimate the extent
of water/bilayer partition for NBD-PI(4,5)P 2 . From
equation 3 (21), the extent of partition can be quantified
into partition coefficients (K p )
I 5 I w 1 K P 3 g L 3 [L] 3 I L
1 1 K P 3 g L 3 [L]
(Eq: 3)
Fluorescence measurements
Steady-state fluorescence measurements were carried out with
an SLM-Aminco 8100 Series 2 spectrofluorimeter described in
detail elsewhere (15). Steady-state anisotropies were determined
according to Lakowicz (16). NBD-PC and NBD-PI(4,5)P 2 anisotropies
were recorded with excitation and emission wavelengths
of 460 and 540 nm, respectively, with spectral bandwidths
of 4 nm. All measurements were performed at room temperature.
Fluorescence decay measurements of DPH (donor) were carried
out with a time-correlated single-photon timing system, which is
described elsewhere (15). All measurements were performed at
room temperature. Excitation and emission wavelengths were 340
and 430 nm, respectively.
In the FRET experiments, the donor decays in the presence of
acceptors [i DA (t)] can be described as
i DA (t) 5 i D (t) 3r interplanar (t) (Eq: 1)
where i DA (t) and i D (t) are the fluorescence decays of the donor in
the presence and absence of acceptors, respectively. r interplanar is
the FRET contribution arising from energy transfer to randomly
distributed acceptors in two different planes from the donors
(two monolayer leaflets) (17)
8
l
9
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
l 2 1 R ><
2 e
r interplanar 5 exp 24 n 2 3p3l 2 1 2 exp(2t 3 b
#
3 3a 6 )
>=
a 3 da
>:
0
>;
Downloaded from www.jlr.org by on September 3, 2007
(Eq: 2)
where b 5 (R 0 2 /l) 2 H D 21/3 ,R 0 is the Förster radius, n 2 is the
acceptor density in each leaflet, and l is the distance between the
plane of the donor and the two planes of acceptors. The Förster
radius was calculated as described elsewhere (18).
RESULTS
NBD-PI(4,5)P 2 partition to lipid bilayers
When using a labeled phospholipid as the acceptor in a
FRET study, the energy transfer efficiency will depend
strongly on its concentration inside these vesicles. As
PI(4,5)P 2 is known to generally form micelles in the aqueous
medium (19), it is essential in the type of measurements
mentioned above to determine whetherallNBD-PI(4,5)P 2
incorporates in the liposomes and, if that is not the case,
Fig. 1. Fluorescence intensity (I F ) of phosphatidylinositol-(4,5)-
bisphosphate [PI(4,5)P 2 ] as a function of POPC (A) and 1-palmitoyl-
2-oleoylphosphatidylglycerol (POPG) (B) concentration. Closed
circles, pH 8.4; open circles, pH 7.1; closed squares, pH 4.8. The
curves are fits of equation 3 to the experimental data, allowing the
determination of water/bilayer partition coefficients (K p ). 7-Nitro-
2-1,3-benzoxadiazol (NBD)-PI(4,5)P 2 concentration 5 1 mM. a.u.,
arbitrary units.
1522 Journal of Lipid Research Volume 47, 2006

where I is the fluorescence intensity, I W and I l are the fluorescence
intensities in buffer and in liposomes, respectively,
g l is the lipid molar volume, and [l] is the lipid concentration.
The curves obtained when the NBD-PI(4,5)P 2 fluorescence
intensities were plotted versus the lipid concentrations
are shown in Fig. 1. Three different pH values
(8.4, 7.1, and 4.8) corresponding to three different
protonation states of PI(4,5)P 2 (for the micellar state
Pk a2 5 6.7 for the 49 position and 7.6 for the 59 position)
were studied. The results for the different pH values were
clearly different when using POPC bilayers: a larger extent
of partition was observed when pH 8.4 was used (corresponding
to a total charge of 25), whereas partition for
pH 7.1 (24) was almost identical to that for pH 4.8 (23).
For POPG, the partition was much less dependent on pH.
The K p values obtained at pH 8.4, 7.1, and 4.8 for water/
POPC partition were (5.38 6 0.60) 3 10 4 , (1.84 6 0.14) 3
10 4 , and (2.18 6 0.18) 3 10 4 , respectively, whereas for
micelle/POPG partition, these were (1.84 6 0.33) 3 10 4 ,
(2.29 6 0.38) 3 10 4 , and (1.47 6 0.18) 3 10 3 , respectively.
Clustering of NBD-PI(4,5)P 2 in POPC
Clustering of NBD-PI(4,5)P 2 inside the POPC matrix
would result in a decrease of fluorescence intensity attributable
to self-quenching of NBD and in a decrease in
fluorescence anisotropy attributable to energy migration
(energy homotransfer) inside the clusters (20). The two
parameters were studied for increasing NBD-PI(4,5)P 2
concentrations inside a POPC matrix, as shown in Fig. 2.
NBD-PI(4,5)P 2 concentrations were corrected for the partition
coefficients determined above. As a control, the
fluorescence intensity and anisotropies of NBD-PC in
POPC and of NBD-PI(4,5)P 2 in DPPC were compared with
the data for NBD-PI(4,5)P 2 in POPC. NBD-PC in POPC
is homogeneously distributed, whereas in DPPC, clustering
is expected for NBD-PI(4,5)P 2 as a result of gel-fluid
phase separation.
It is clear that NBD-PI(4,5)P 2 and NBD-PC have identical
clustering behavior in POPC. The degree of selfquenching
and energy migration of both lipids is almost
identical and much smaller than for NBD-PI(4,5)P 2 in
DPPC at room temperature, at which clustering is observed
as a result of packing restraints in the DPPC gel
matrix (22).
We also performed a FRET experiment, using for this
purpose the fluorescence decay of the donor. The reason
for a time-dependent approach instead of a steady-state
approach is that the kinetics of the donor decay gives
information on both the distribution (homogeneity/heterogeneity)
and concentration of acceptors (15), whereas
the steady-state approach is unable to distinguish between
the two. Again, we chose POPC as the matrix lipid, the
donor was DPH, which is known to have no preference
between lipid phases (23), and the acceptor was NBD-
PI(4,5)P 2 . In this system, in the case of NBD-PI(4,5)P 2
clustering, two populations of DPH would be detectable,
one residing in a NBD-PI(4,5)P 2 -rich site and the other
in a NBD-PI(4,5)P 2 -poor area of the bilayer.
We globally fitted the FRET model for a single discrete
concentration of acceptors (homogeneous distribution of
acceptors) (equations 2, 3) to the data in Fig. 3, and for all
pH values used (only results for pH 8.4 are shown), good
quality fits were obtained (global Chi-square , 1.5). The
acceptor concentrations were the only free parameter
during these fits, and the recovered values matched closely
(610%) the bilayer concentrations expected from the
partition coefficients determined above, again confirming
the absence of NBD-PI(4,5)P 2 clustering.
DISCUSSION
Compartmentalization of PI(4,5)P 2 has been reported
several times, but its origin had always been attributed to
Downloaded from www.jlr.org by on September 3, 2007
Fig. 2. Fluorescence intensity (I F ) of NBD-PI(4,5)P 2 in
POPC (closed circles), POPG (open circles), and 1,2-dipalmitoylphosphatidycholine
(DPPC; closed diamonds)
bilayers and of phosphatidylcholine NBD-PC in POPC
(closed squares). Inset: Fluorescence anisotropy (,r.) of
NBD-PI(4,5)P 2 in POPC (closed circles) and DPPC (open
diamonds) bilayers and of NBD-PC in POPC (closed
squares). Error bars represent SD. All measurements were
performed at pH 8.4. Total lipid concentration 5 0.2 mM.
a.u., absorbance units.
Absence of PI(4,5)P 2 clustering in fluid PC 1523

Fig. 3. Fluorescence decays of diphenylhexatriene in the absence and presence of increasing concentrations
of NBD-PI(4,5)P 2 in POPC bilayers at pH 8.4 (thin lines). Global analysis of the data according to
the model for homogeneous distribution of NBD-PI(4,5)P 2 (equation 3) resulted in the fitted curves (thick
lines). The concentration of PI(4,5)P 2 was kept constant (5%) by the addition of unlabeled PI(4,5)P 2 . Total
lipid concentration 5 0.2 mM. t, time.
the interaction with the cytoskeleton and/or with specific
proteins exhibiting preferences for particular phases
inside the membrane. The recent proposal for a spontaneous
demixing of completely deprotonated PI(4,5)P 2
when in a PC matrix even at very low concentrations as a
result of intermolecular hydrogen bonding (12) suggested
an additional mechanism for localized enrichments of this
phospholipid. In this work, we tested this hypothesis
through a carefully designed approach that considered
and tested the existence of possible artifacts that could go
undetected in a phenomenological approach (mere observation
of increases or decreases in energy transfer as a
result of pH variations).
Redfern and Gericke (12) showed by differential scanning
calorimetry and Fourier transform infrared spectroscopy
that demixing occurred for mixtures of PI(4,5)P 2
and PC lipids in the gel phase. Nevertheless, in vivo,
PI(4,5)P 2 contains polyunsaturated acyl chains, and a gel
state is highly unlikely for this lipid. With this in mind,
these authors used FRET as a tool to detect the immiscibility
of PI(4,5)P 2 and PC in the fluid state at low
PI(4,5)P 2 concentrations. They used POPC as the lipid
matrix and a NBD-labeled PC as the FRET donor to a
short-chain (C6) Dipyrrometheneboron difluoride labeled
PI(4,5)P 2 . In this system, they observed variations
of energy transfer efficiency as the pH was scanned. The
authors interpreted the decrease of energy transfer efficiencies
at high pH [.Pk a2 of PI(4,5)P 2 ] as a demixing
between PI(4,5)P 2 and POPC. However, the use of a very
short-chain PI(4,5)P 2 as the acceptor in this FRET experiment
is likely to have resulted in a significant fraction
of acceptors not being incorporated in the membrane.
The lipid/water partition coefficients of this short-chainlabeled
PI(4,5)P 2 is expected to be significantly smaller
than those determined here for NBD-PI(4,5)P 2 , because
K p varies dramatically with the number of carbons in the
acyl chains of a phospholipid (24). In this case, energy
transfer efficiencies would decrease not because of the
formation of PI(4,5)P 2 clusters but as a result of changes
in the lipid/water partition coefficient of the probe, as the
PI(4,5)P 2 group becomes fully deprotonated. Redfern
and Gericke (12) acknowledge that the extent of membrane
incorporation changed considerably for different
labeled phosphatidylinositides.
Analysis of the kinetics of donor decay in a FRET experiment
was the method chosen by us to probe the
eventual clustering of PI(4,5)P 2 in a zwitterionic bilayer,
because it is capable of retrieving information on the
homogeneity of the fluorescently labeled PI(4,5)P 2 independently
of any quantitative knowledge of the probe’s
bilayer concentration. The good agreement between the
model considering a homogeneous distribution of NBD-
PI(4,5)P 2 and the data is solid proof of the absence of
PI(4,5)P 2 clustering at low PI(4,5)P 2 concentrations
(,5%) even in a completely deprotonated state. The result
of this experiment alone cannot exclude the formation
of very small PI(4,5)P 2 clusters (zR 0 for this Förster
pair 5 40 Å), but, together with the NBD energy migration
and self-quenching studies, it is clear that PI(4,5)P 2
has no lateral organization in a fluid POPC matrix. It is
Downloaded from www.jlr.org by on September 3, 2007
1524 Journal of Lipid Research Volume 47, 2006

certainly feasible that PI(4,5)P 2 /PC demixing exists in the
gel phase, as packing restraints are severely increased (22).
The difference in the lipid/water partition coefficients of
the NBD-PI(4,5)P 2 fully and partially deprotonated states
is interesting, and the same trend is expected for nonlabeled
PI(4,5)P 2 . This difference might be the result of
destabilization of the fully deprotonated PI(4,5)P 2 micellar
structure. However, it is very likely that the labeling
with NBD decreases the extent of partition to some extent,
and further studies will be necessary to determine whether
this phenomenon has physiological relevance.
F.F. acknowledges Grant SFRH/BD/14282/2003 from Fundação
para a Ciência e a Tecnologia (FCT) (Portugal). A.F. acknowledges
Grant SFRH/BPD/26150/2005 from FCT (Portugal).
This work was funded by FCT (Portugal) under the program
Programa Operacional Ciência e Inovação (POCI).
REFERENCES
1. Payrastre, B., K. Missy, S. Giuriato, S. Bodin, M. Plantavid, and M-P.
Gratacap. 2001. Phosphoinositides: key players in cell signalling, in
time and space. Cell. Signal. 13: 377–387.
2. Czech, M. P. 2003. Dynamics of phosphoinositides in membrane
retrieval and insertion. Annu. Rev. Physiol. 65: 791–815.
3. Ferrell, J., Jr., and W. Huestis. 1984. Phosphoinositide metabolism
and the morphology of human erythrocytes. J. Cell Biol. 98:
1992–1998.
4. Pike, L. J., and L. Casey. 1996. Localization and turnover of
phosphatidylinositol 4,5-bisphosphate in caveolin-enriched membrane
domains. J. Biol. Chem. 271: 26453–26456.
5. Pike, L. J., and J. M. Miller. 1998. Cholesterol depletion delocalizes
phosphatidylinositol bisphosphate and inhibits hormone-stimulated
phosphatidylinositol turnover. J. Biol. Chem. 273: 22298–22304.
6. Rozelle, A. L., L. M. Machesky, M. Yamamoto, M. H. E. Driessens,
R. H. Insall, M. G. Roth, K. Luby-Phelps, G. Marriott, A. Hall, and
H. L. Yin. 2000. Phosphatidylinositol 4,5-bisphosphate induces
actin-based movement of raft-enriched vesicles through WASP-
Arp2/3. Curr. Biol. 10: 311–320.
7. Botelho, R. J., M. Teruel, R. Dierckman, R. Anderson, A. Wells, J. D.
York, T. Meyer, and S. Grinstein. 2000. Localized biphasic changes
in phosphatidylinositol-4,5-bisphosphate at sites of phagocytosis.
J. Cell Biol. 151: 1353–1368.
8. McLaughlin, S., J. Wang, A. Gambhir, and D. Murray. 2002. PIP2
and proteins: interactions, organization, and information flow.
Annu. Rev. Biophys. Biomol. Struct. 31: 151–175.
9. Laux, T., K. Fukami, M. Thelen, T. Golub, D. Frey, and P. Caroni.
2000. GAP43, MARCKS, and CAP23 modulate PI(4,5)P2 at
plasmalemmal rafts, and regulate cell cortex actin dynamics
through a common mechanism. J. Cell Biol. 149: 1455–1472.
10. Rauch, M. E., C. G. Ferguson, G. D. Prestwich, and D. S. Cafiso.
2002. Myristoylated alanine-rich C kinase substrate (MARCKS)
sequesters spin-labeled phosphatidylinositol 4,5-bisphosphate in
lipid bilayers. J. Biol. Chem. 277: 14068–14076.
11. Epand, R. M., P. Vuong, C. M. Yip, S. Maekawa, and R. F. Epand.
2004. Cholesterol-dependent partitioning of PtdIns(4,5)P2 into
membrane domains by the N-terminal fragment of NAP-22 (neuronal
axonal myristoylated membrane protein of 22 kDa). Biochem.
J. 379: 527–532.
12. Redfern, D. A., and A. Gericke. 2005. pH-dependent domain
formation in phosphatidylinositol polyphosphate/phosphatidylcholine
mixed vesicles. J. Lipid Res. 46: 504–515.
13. Redfern, D. A., and A. Gericke. 2004. Domain formation in
phosphatidylinositol monophosphate/phosphatidylcholine mixed
vesicles. Biophys. J. 86: 2980–2992.
14. Mayer, L. D., M. J. Hope, and P. R. Cullis. 1986. Vesicles of variable
sizes produced by a rapid extrusion procedure. Biochim. Biophys.
Acta. 858: 161–168.
15. Loura, L. M. S., A. Fedorov, and M. Prieto. 2001. Fluid-fluid
membrane microheterogeneity: A fluorescence resonance energy
transfer study. Biophys. J. 80: 776–788.
16. Lakowicz, J. R. 1999. Principles of Fluorescence Spectroscopy.
Kluwer Academic/Plenum Publishers, New York.
17. Davenport, L., R. E. Dale, R. H. Bisby, and R. B. Cundall. 1985.
Transverse location of the fluorescent probe 1,6-diphenyl-1,3,5-
hexatriene in model lipid bilayer membrane systems by resonance
energy transfer. Biochemistry. 24: 4097–4108.
18. Berberan-Santos, M. N., and M. Prieto. 1987. Energy transfer in
spherical geometry. J. Chem. Soc. Faraday Trans. 83: 1391–1409.
19. Sugiura, Y. 1981. Structure of molecular aggregates of 1-(3-snphosphatidyl)-L-myo-inositol
3,4-bis(phosphate) in water. Biochim.
Biophys. Acta. 641: 149–159.
20. Prieto, M. J., M. Castanho, A. Coutinho, A. Ortiz, F. J. Aranda, and
J. C. Gomez-Fernandez. 1994. Fluorescence study of a derivatized
diacylglycerol incorporated in model membranes. Chem. Phys.
Lipids. 69: 75–85.
21. Santos, N. C., M. Prieto, and M. A. R. B. Castanho. 2003.
Quantifying molecular partition into model systems of biomembranes:
an emphasis on optical spectroscopic methods. Biochim.
Biophys. Acta. 1612: 123–135.
22. Loura, L. M. S., A. Fedorov, and M. Prieto. 2000. Membrane probe
distribution heterogeneity: a resonance energy transfer study. J.
Phys. Chem. B. 104: 6920–6931.
23. Lentz, B. R., Y. Barenholz, and T. E. Thompson. 1976. Fluorescence
depolarization studies of phase transitions and fluidity in phospholipid
bilayers. II. Two-component phosphatidylcholine liposomes.
Biochemistry. 15: 4529–4537.
24. Nichols, J. W. 1985. Thermodynamics and kinetics of phospholipid
monomer-vesicle interaction. Biochemistry. 24: 6390–6398.
Downloaded from www.jlr.org by on September 3, 2007
Absence of PI(4,5)P 2 clustering in fluid PC 1525

CONCLUSIONS
VII
CONCLUSIONS
Chapter II ‐ Protein‐Protein and Protein‐Lipid Interactions
of M13 Major Coat.
In this chapter an extensive study of the protein-protein and protein-lipid
interactions of M13 mcp was conducted. The fluorescence self-quenching data obtained
from the study with a BODIPY labelled mcp allowed us to conclude that mcp is in fact
monomeric in DOPC bilayers at the conditions used for protein purification and protein
reconstitution. The use of two different sites for labelling, resulted in identical results
excluding any possible influence of labelling on the oligomerization efficiency of mcp.
Nevertheless, as the studies with lipid bilayers presenting a different hydrophobic
thickness than mcp shows, mcp can aggregate is certain conditions. In this way, it is
possible that at much higher protein concentrations than the one used here (L/P < 50),
some aggregation takes place, although in vivo the presence of only one orientation of
mcp should contribute to keep it monomeric. This could explain the increase in the
levels of anionic lipids (cardiolipin and phosphatidylglycerol) in the lipid membrane of
E. coli during the process of infection by the bacteriophage M13. Therefore, although
anionic lipids are not required to keep mcp monomeric in normal conditions, at the very
high mcp concentrations found in the sites of phage assembly in the lipid membrane of
the host, they might contribute to stabilize the monomeric state of the protein, which is
required for correct assembly.
Fluorescence self-quenching of BODIPY labelled mcp is more efficient when mcp
is incorporated in bilayers with a thinner (positive hydrophobic mismatch) or thicker
(negative hydrophobic mismatch) hydrophobic length, than the hydrophobic domain of
mcp. This is likely to result from aggregation of mcp, and in this case aggregation in
thicker bilayers is significantly more effective than in thinner bilayers. This discrepancy
can be the result of the availability of alternative methods for protein adaptation in
positive hydrophobic mismatch conditions. In this case, the protein is able to increase
181

the tilt angle, and maintain a larger section of the hydrophobic domain inside the bilayer
(Figure I.16). When the bilayer thickness is larger than the protein hydrophobic domain,
the TM protein might find that compensation by change of tilt angle is not possible or it
is insufficient, and the only route available to decrease exposure of hydrophobic
aminoacid side-chains is aggregation. Recently, mcp was shown to change its tilt angle
when incorporated in bilayers with different hydrophobic thickness (Koehorst et al.,
2004).
Results obtained for fluorescence self-quenching of BODIPY labelled mcp
incorporated in binary lipid mixtures composed by lipids with different acyl-chain
lengths are dramatically different from the results obtained with pure lipids. The
apparent diffusion coefficients retrieved from the analysis of this data are clearly not
justified on the basis of normal diffusion of monomeric species (as it was for mcp in
DOPC), or on the basis of protein oligomerization/aggregation (as it as for mcp in
DMoPC and in DEuPC). In this case, only segregation of mcp into domains enriched in
the hydrophobically matching lipid can provide rationalization of the data. This
hypothesis is confirmed by FRET studies that are also consistent with mcp segregation
in the presence of binary mixtures of lipids with different acyl-chains. Importantly, the
lipid components used in the binary mixtures are not dramatically different, the only
difference being the presence of 4 additional carbons in the acyl-chain of one of the
lipids. This difference is not expected to result in large scale phase separation of the
lipid components such as it was detected in the FRET studies (domain size larger than
or comparable to the Förster radius of AEDANS-BODIPY, 48.8 Å). In this way,
formation of large domains enriched in the hydrophobically matching lipid must be a
consequence of incorporation of mcp. Nevertheless, experiments making use of the
properties of 1,6-diphenylhexatriene, indicate that even in the absence of protein, short
scale clustering of the lipids in the binary mixtures is expected to occur.
The same type of segregation of mcp was not detected when the acyl-chains’ lengths
of the lipid components in the binary mixtures were identical and the only difference
resides in the headgroups, confirming the crucial relevance of hydrophobic matching for
protein organization.
Concerning the quantification of the lipid selectivity of mcp, FRET proved to be a
valuable tool, and an important alternative to ESR in these type of studies. FRET was
able to detect lipid enrichment around monomeric mcp in a binary lipid mixture, while
ESR was unable to resolve this problem (Sanders et al., 1992). Relative association
182

CONCLUSIONS
constants recovered by FRET were almost identical to the ones recovered for an
aggregated form of mcp, confirming the reliability and sensitivity of the technique. mcp
showed affinity for lipids with acyl chain lengths that matched the hydrophobic length
of its TM domain, and for anionic lipids, particularly PA and PS.
Only the lipids in the immediate vicinity of the protein are sufficiently immobilized
to be detected by ESR, and as a consequence the binding affinity determined by ESR
only refers to this shell of lipids. On the other hand, FRET is not limited to this, and for
high Förster radii such as the one for the donor-acceptor pair used in this study, the
efficiency of FRET from the protein to an acceptor in the second shell of lipids is
similar to the efficiency of FRET to an acceptor in the first shell around the protein. In
this way, the compatibility of the results from FRET and ESR points clearly to the
validity of the annular model for protein-lipid interaction (single TM protein only
induces changes in lipid distribution on the first shell of lipids around it).
This FRET methodology for the quantification of protein-lipid selectivity also offers
the possibility of probing different shells of lipids around the protein by choosing
donor-acceptor pairs with different Förster radii. To increase the sensitivity of the
method to selectivity in the first shell of lipids around the protein, a small Förster radius
would be advisable, that allowed to concentrate the majority of the contributions to
donor quenching on the acceptors located there.
183

Chapter III ‐ Binding of Inhibitors to a Putative Binding
Domain of V‐ATPase.
The studies conducted allowed us to know in detail the properties of the indole type
of inhibitors in lipid membranes. The molecules studied, which correspond to the initial
(low efficiency) and final stage (high efficiency) in the development of these inhibitors
as potential drugs in osteoporosis treatment, present almost identical properties in the
liposomes environment. The preference of the lipid phase over the aqueous phase, the
position in the bilayer and the orientation of the molecules were all almost identical for
both molecules studied.
According to the results of Gagliardi et al. (1998a), the change in inhibition
efficiency of the V-ATPase inhibitors within the indole class presented some extent of
correlation to the hydrophobicity of the molecules, i.e. the more hydrophobic indole
inhibitor molecules had also a tendency to be more efficient inhibitors of V-ATPase. It
was possible that the interaction of the molecules with the lipid milieu was significantly
different within this class of inhibitors, and this could correspond to dramatically
different efficiencies of inhibition, explaining the 10 3 difference in IC 50 values of INH-1
and SB242784. In this way, the process of inhibitor binding to the lipid bilayer would
be crucial to the efficiency of the inhibition event. However, the results obtained point
to a non-determinant role of inhibitor interaction with the lipid bilayer, and the specific
molecular recognition processes with the enzyme’s inhibition site are very likely to be
responsible for dictating the large differences of inhibition efficiencies within the V-
ATPase indole inhibitors class.
The binding studies of bafilomycin and SB242784 to the peptide corresponding to
the 4 th TM helix of subunit c of V-ATPase revealed no affinity at all for SB242784 and
very low affinities for bafilomycin. These results can be seen as an indicator that the
binding pocket for the inhibitors is not solely composed of the lipid exposed face of the
4 th TM helix of subunit c, but also involves contributions of amino acids from other
domains of the protein. This proposal is in agreement with the hypothesis of the
inhibitors operating as a stone in a gear, blocking movements of the TM domains of
subunit c and interrupting proton transport.
184

CONCLUSIONS
Chapter IV ‐ Binding of a Quinolone Antibiotic to Bacterial
Porin OmpF.
In this case, the FRET analysis allowed us to conclude on a likely location for the
binding site of CP in OmpF. Two limiting positions for the binding site were
considered: i) in the trimer interface, leading to a maximum efficiency of donor
quenching; ii) in the periphery of the trimer, causing minimum quenching of donor
fluorescence due to FRET. For each of these two limiting conditions, intervals for
binding constants were retrieved. By comparison of the binding constants retrieved with
the results of an independent method that made use of changes in the absorption spectra
of CP upon binding to OmpF (Neves et al., 2005), it is concluded that the binding site
for CP is likely to be located away from the trimer interface and close to the periphery
of OmpF.
The FRET methodology developed for the analysis of the interactions between
OmpF and CP is suitable to be used in the analysis of FRET data for similar systems,
containing two populations of donors (or acceptors) separated by a distance larger or
comparable to the Förster distance of the donor-acceptor pair used, so that the analysis
is sensitive enough to be able to distinguish the contributions from these two
populations.
185

Chapter V – Interaction of Helix‐0 of the N‐BAR Domain
With Lipid Membranes
The membrane modelling properties of N-BAR domains are generally attributed to
two structural features of the BAR domain containing proteins: the concave surface of
the crescent shaped dimer, that is expected to act as a scaffold for curvature and an N-
terminal amphipatic helix or H0, that is expected to be deeply buried within the lipid
bilayer, driving monolayer asymmetry and finally curvature. This latter strategy for
curvature generation should be effective in the absence of the rest of the protein, and a
peptide comprising this amphipatic helix would be expected to produce the same
results. The results reproduced here clearly demonstrate that this is not the case. H0 is
unable to drive significant changes in liposome morphology. Therefore we can conclude
that N-BAR domains do not use their N-terminal amphipatic helices to produce bilayer
curvature in the manner often proposed (Gallop and McMahon, 2005; Gallop et al.,
2006; Masuda et al., 2006). Nevertheless, the results point to different roles of Helix-0
on membrane remodelling by N-BAR domains. The high water/lipid partition
coefficients determined for H0 are expected to have a dramatic effect in the partition
properties of the entire protein, and this alone could explain the great decreases in
liposome tubulation efficiency of BAR domains without H0. A role as a mediator of
interactions between dimers of BAR domains was also proposed for H0 (Gallop and
McMahon, 2005), and here it is shown that H0 does indeed oligomerize into dimers in
the lipid environment. Oligomerization of H0 might provide BAR domain-containing
proteins with a more efficient mechanism to colocalize, and membrane curvature
induction by scaffolding by proteins is only expected to be efficient at high local
concentrations of scaffold proteins.
186

CONCLUSIONS
Chapter VI – Clustering of PI(4,5)P2 in Fluid PC Bilayers.
Domain enrichment of PI(4,5)P 2 is expected to occur in vivo. Nevertheless the
process must require strong binding to agents (proteins) that present a strong affinity for
incorporation in cholesterol enriched domains. Clustering of charged phospholipids in
the gel state is certainly possible as energetic requirements resulting from packing
restrictions in this state can surpass the energetic penalty resulting from electrostatic
repulsion between the clustered lipids. However, the same type of phenomena is
expected to be highly unlikely in fluid bilayers, where packing restrictions are not so
severe.
The results from this study allowed us to conclude that PI(4,5)P 2 is homogeneously
distributed in fluid PC bilayers. FRET studies ruled out the existence of any large scale
domain (dimensions larger than 40 Å) enriched in PI(4,5)P 2 , while energy migration
and fluorescence self-quenching studies of a fluorescently labelled PI(4,5)P 2 clearly
indicate absence of strong short-range clustering of the lipid.
Another important result from this study are the differences in water/bilayer
partition efficiencies obtained for different protonation states of PI(4,5)P 2 . One possible
explanation for this observation is that the micellar structure of PI(4,5)P 2 is destabilized
at the completely deprotonated state (at low lamellar lipid concentrations, a micellar
population of PI(4,5)P 2 is expected to be in equilibrium with a bilayer inserted
population). These differences in partition behaviour can result in erroneous
rationalization of fluorescent data, specially taking into account that the most popular
fluorescently labelled PI(4,5)P 2 in the market presents very short acyl-chains, and will
present even smaller partition efficiencies that the ones determined here for a long chain
fluorescently labelled PI(4,5)P 2 .
187

188

FINAL CONSIDERATIONS AND PROSPECTS
VIII
FINAL CONSIDERATIONS AND
PROSPECTS
Biophysical studies are likely the only possibility of direct structural and dynamic
characterization of the type of interactions addressed in this work. The recurrent
difficulties in membrane protein expression and purification were responsible for the
popularity of the application of highly sensitive fluorescence techniques in studies of
membrane proteins. The presence of aromatic amino acids in proteins allows the
application of fluorescence techniques without requirement for labelling with a extrinsic
fluorophore. Still, when necessary, the developments in synthesis of fluorophores and
labelling techniques, provide the researcher with great flexibility in the process of
choosing the best conditions for fluorescence methodologies. Overall, the studies
presented here are a demonstration of the potential of biophysical studies, particularly
fluorescence spectroscopy studies, in the characterization of interactions between
components of biomembranes.
In this work, the use of membrane model systems, allowed for the simplification of
the number of variables in the experiments. This is essential in studies that focus on the
characterization of very specific interactions. Apart from the technical difficulties
inherent to in vivo studies (huge heterogeneity of samples, background fluorescence,
light scattering, low fluorescence intensity, etc.), the presence of a large number of
uncontrolled (and sometimes unknown) variables is highly likely to hinder the
possibility of data rationalization. The information recovered from studies with less
complex model systems can then be used for a more clear rationalization of data
gathered from in vivo studies.
The new FRET methodologies developed here, notably the methodology for
quantification of protein-lipid selectivity, are expected to be of great use in the future as
they are readily applicable to different systems. The FRET methodologies for
quantification of drug-protein binding will require only some straightforward
adaptations to the geometric constraints of the systems to be studied.
189

Some of the biological questions debated here are still open. Recently, a model for
the binding site of V-ATPase inhibitors in the c-subunit was proposed (Bowman et al.,
2006), confirming that residues in at least three different transmembrane domains were
required for effective bafilomycin inhibition. However the picture was not completely
clear, as residues in opposite sides of the same helix were found to be relevant for
bafilomycin binding pointing to indirect effects of some residues in the interaction with
the inhibitor. Although, clinical trials with SB242784 were not started due to possible
adverse effects (Nikura et al., 2005), a derivative of this molecule was shown recently to
potentiate the effect of an cytotoxic/anti-tumor drug likely due to inhibition of V-
ATPases in tumor cells (Petrangoline et al., 2006).
The study of the mechanisms behind clustering of PI(4,5)P 2 into cholesterol
dependent domains in biomembranes is a very promising field. It is clear that PI(4,5)P 2
should not spontaneously cluster or partitionate to liquid ordered domains, and this
behaviour must be connected to specific protein-lipid interactions. In this way, the study
of the lipid membrane partition properties of proteins known to bind PI(4,5)P 2 might
give us some insight into these mechanisms. The relevance of acylation characteristics
(number and length of acyl-chains in the protein), for membrane protein organization in
biomembranes, is a problem of great biological relevance.
The use of fluorescence imaging techniques in the study of protein-lipid interactions
can add additional layers of information to the results of non-space resolution
biophysical studies. The biophysical studies presented here relied mostly on
macroscopic observables, and although this type of data is, as demonstrated, highly
enriched in information, some of it is lost by averaging of fluorophore populations,
while imaging techniques allow for resolution of different populations of fluorophores.
Therefore, the application of the methodologies derived and described here to
fluorescence imaging data could help identify features of interactions between
membrane components which go undetected in the macroscopic data. In the limit,
single-molecule techniques could also be used. Imaging techniques are obviously very
useful in the observation of large scale changes (µm scale) in the lateral distribution of
membrane components and are a useful complement to FRET methodologies, which are
able to probe deviations to homogeneity in the nanometer scale. Additionally, FRET
studies under the microscope are also carried out, allowing the direct comparison
between macroscopic and microscopic data.
190

BIBLIOGRAPHY
IX
BIBLIOGRAPHY
Alberts, B., Johnson, A., Lewis, J., Raff, M., Roberts, K., and Walter, P. (2002) Molecular Biology of the
Cell. 4 th Ed. Garland Science, Oxford, UK.
Albini, A., and Monti, S. (2003). Photophysics and Photochemistry of Fluoroquinolones. Chem. Soc.
Rev. 32: 238-250.
Anantharamaiah, G. M., Jones, M. K., and Segrest, J. P. (1993) An Atlas of the Amphipatic Helical
Domains of Human Exchangeable Plasma Apolipoproteins. In The Amphipatic Helix. Ed. Epand, R. M.
CRC Press, Boca-Raton, U.S.A.
Arnold, K., Losche, A., and Gawrisch, K. (1981). 31p-NMR Investigations of Phase Separation in
Phosphatidylcholine/Phosphatidylethanolamine Mixtures. Biochem. Biophys. Acta. 645:143–148.
Auger, I. E. (1993) Computational Techniques to Predict Amphipatic Helical Segments. in The
Amphipatic Helix. Ed. Epand, R. M. CRC Press, Boca-Raton, U.S.A.
Bagatolli, L. A., and Gratton, E. (2000). Two Photon Fluorescence Microscopy of Coexisting Llipid
Domains in Giant Unilamellar Vesicles of Binary Phospholipid Mixtures. Biophys. J. 78: 290-305.
Ben-Shaul, A. (1995). Molecular Theory of Chain Packing, Elasticity and Lipid-Protein Interactions in
Lipid Bilayers. In “Structure and Dynamics of Membranes. . From Cells to Vesicles.” (Eds Lipowsky, R.,
and Sackmann, E.) North Holland, Amsterdam, The Netherlands.
Berg, J. M, Tymoczko, J. L., and Stryer, L. (2002). Biochemistry. 5 th Ed. W. H. Freeman, New York,
USA.
Bloom, M. E., and Mouritsen, O. G. (1984). Mattress Model of Lipid-Protein Interactions in Membranes.
Biophys. J. 46: 141-153.
Botelho, R. J., Teruel, M., Dierckman, R., Anderson, R., Wells, A., York, J. D., Meyer, T., and Gristein,
S. (2000). Localized Biphasic Changes in Phosphatidylinositol-4,5-Bisphosphate at Sites of Phagocytosis.
J. Cell. Biol. 151: 1353-1367.
Bowman, B. J., and Bowman, E. J. (2002). Mutations in Subunit c of the Vacuolar ATPase Confer
Resistance to Bafilomycin and Identify a Conserved Antibiotic Binding Site. J. Biol. Chem. 277: 3965-
3972.
191

Bowman, E. J., Graham, L. A., Stevens, T. H., and Bowman, B. J. (2004). The
Bafilomycin/Concanamycin Binding Site in Subunit c of the V-ATPase from Neurospora crassa and
Sacharomyces cerevisiae. J. Biol. Chem. 279: 33131-33138.
Bowman, B. J., McCall, M. E., Baertsch, R., and Bowman, E. J. (2006). A Model for the Proteolipid Ring
and Bafilomycin/Concanamycin-binding Site in the Vacuolar ATPase of Neurospora crassa. J. Biol.
Chem. 42: 31885-31893.
Brasseur, R. (1991). Differentiation of lipid-associating helices by use of three-dimensional molecular
hydrophobicity potential calculations. J. Biol. Chem. 266: 16120-16127.
Braun, P., and von Heijne, G. (1999). The Aromatic Residues Trp and Phe Have Different Effects on the
Positioning of a Transmembrane Helix in the Microsomal Membrane. Biochemistry 38: 9778-9782.
Caputo, G. A., and London, E. (2003). Cumulative Effects of Amino-Acid Substitution and Hydrophobic
Mismatch Upon the Transmembrane Stability and Conformation of Hydrophobic α-Helices.
Biochemistry 42: 3275-3285.
Caputo, G. A., and London, E. (2004). Position and Ionization State of Asp in the Core of Membrane-
Inserted α-Helices Control Both the Equilibrium between Transmembrane and Nontransmembrane Helix
Topography and Transmembrane Helix Positioning. Biochemistry 42: 3275-3285.
Chapman, J. S., and Georgopapadakou, N. H. (1988). Routes of Quinolone Permeation in Escherichia
coli. Antimicrob. Agents and Chemoter. 32: 438-442.
Chevalier, J., Malléa, M., and Pagés, J. M. (2000). Comparative Aspects of the Diffusion of Norfloxacin,
Cefepime and Spermine through the F Porin Channel of Enterobacter cloacae. J. Biochem. 348: 223-227.
Dawson, J. C., Legg, J. A., and Machesky, L. M. (2006). Bar Domain Proteins: A Role in Tubulation,
Scission and Actin Assembly in Clathrin-Mediated Endocytosis. Trends Cell Biol. 16: 493-498.
de Almeida, R. F. M., Loura, L. M. S., Fedorov, A., and Prieto, M. (2002). Nonequilibrium Phenomena in
the Phase Separation of a Two-Component Lipid Bilayer. Biophys. J. 82: 823-834.
de Planque, M. R. R., Killian, J. A. (2003). Protein-Lipid Interactions Studied With Designed
Transmembrane Peptides: Role of Hydrophobic Matching and Interfacial Anchoring. Mol. Memb. Biol.
20: 271-284.
Dechené, M., Leying, H., and Cullman, W. (1990). Role of the Outer Membrane for Quinolone
Resistance in Enterobacteria. Chemoterapy 36: 13-23.
Denisov, G., Wanaski, S., Luan, P., Glaser, M., and McLaughlin, S. (1998). Binding of Basic Peptides to
Membranes Produces Lateral Domains Enriched in the Acidic Lipids Phosphatidylserine and
192

BIBLIOGRAPHY
Phosphatidylinositol 4,5-bisphosphate: An Electrostatic Model and Experimental Results. Biophys. J. 74:
731-744.
Devaux, P. F., and Morris, R. (2004). Transmembrane Asymmetry and Lateral Domains in Biological
Membranes. Traffic 5: 241-246.
Dong, M., Bagetto, L. G., Falson, P., Le Maire, M., and Penin, F. (1997). CompleteRemoval and
Exchange of Sodium Dodecyl Sulfate Bound to Soluble and Membrane Proteins and Restoration of Their
Activities, Using Ceramic Hydroxypatite Chromatography. Anal. Biochem. 247: 333-341.
Dröse, S. and Altendorf, K. (1997). Bafilomycins and Concanamycins as Inhibitors of V-ATPases and P-
ATPases. J. Exp. Biol. 200, 1-8.
Dumas, F., Sperotto, M. M., Lebrun, M.-C., Tocanne, J.-F., and Mouritsen, O. G. (1997). Molecular
Sorting of Lipids by Bacteriorhodopsin in Dilauroylphosphatidylcholine/Distearoylphophatidylcholine
Lipid Bilayers. Biophys. J. 73: 1940-1953.
Dumas, F., Lebrun, C., and Tocanne, J-F. (1999). Is the Protein/Lipid Hydrophobic Matching Principle
Relevant to Membrane Organization and Function. FEBS Lett. 458: 271-277.
Dunnil, P. (1968). The Use of Helical Net-Diagrams to Represent Protein Structures. Biophys. J. 8: 865-
875
East, J. M., and Lee, A. G. (1982). Lipid Selectivity of the Calcium and Magnesium Ion Dependent
Adenosinetriphosphatase, Studied With Fluorescence Quenching by a Brominated Phospholipid.
Biochemistry 21: 4144-4151.
Eisenberg, D., Weiss, R. M., Terwillinger, T. C. (1982). The Helical Hydrophobic Moment: A Measure of
the Amphiphilicity of Helix. Nature 299: 371-374.
Epand, R. M. (1993) Introduction. in The Amphipatic Helix. Ed. Epand, R. M. CRC Press, Boca-Raton,
U.S.A.
Epand, R. M., Vuong, P., Yip, C. M., Maekawa, S., and Epand, R. F. (2004). Cholesterol-Dependent
Partitioning of PtdIns(4,5)P 2 into Membrane Domains by the N-terminal Fragment of NAP-22 (Neuronal
Axonal Myristoylated Membrane Protein of 22 kDa). Biochem. J. 379: 527-532.
Epand, R. M. (2005) The Role of Proteins in the Formation of Domains in Membranes. In In “Protein-
Lipid Interactions: New Approaches and Emerging Concepts.” (Eds. Mateo, C. R., Gómez, J., Villalaín,
J., Ros, J. M. G.) Springer-Verlag. Berlin, Germany.
Epand, R. F., Sayer, B. G., and Epand, R. M. (2005). Induction of Raft-Like Domains by a Myristoilated
NAP-22 Peptide and its Tyr Mutant. FEBS J. 272: 1792-1803.
193

Farsad, K., Ringstad, N., Takei, K., Floyd, S. R., Rose, K., and De Camilli, P. (2001). Generation of High
Curvature Membranes Mediated by Direct Endophilin Bilayer Interactions. J. Cell Biol. 155: 193-200.
Fattal, D. R., and Ben-Shaul, A. (1993). A Molecular Model for Lipid-Protein Interaction in Membranes:
The Role of Hydrophobic Mismatch. Biophys.J. 65: 1795-1809.
Finbow, M. E., and Harrison, M. A. (1997). The Vacuolar H + -ATPase: A Universal Proton Pump of
Eukaryotes. Biochem. J. 324: 697-712.
Ford, M.G.J., Mills, I.G., Peter, B.J., Vallis, Y., Praefcke, G.J.K., Evans, P.R. and McMahon, H.T. (2002)
Curvature of Clathrin-Coated Pits Driven by Epsin. Nature 419: 361-366.
Gagliardi, S., Nadler, G., Consolandi, E., Parini, C., Morvan, M., Legave, M. N., Belfiore, P., Zocchetti,
A., Clarke, G. D., James, I., Nambi, P., Gowen, M., and Farina, C. (1998a). 5-(5,6-Dichloro-2-indolyl)-2-
methoxy-2,4-pentadienamides: Novel and Selective Inhibitors of the Vacuolar H + -ATPase of Osteoclasts
with Bone Antiresorptive Activity. J. Med. Chem. 41: 1568-1573.
Gagliardi, S., Gatti, P. A., Belfiore, P., Zochetti, A., Clarke, G. D., and Farina, C. (1998b). Synthesis and
Structure-Activity Relationships of Bafilomycin A 1 Derivatives as Inhibitors of Vacuolar H + -ATPase. J.
Med. Chem. 41: 1883-1893.
Gallop, J. L., and McMahon, H. T. (2005). BAR Domains and Membrane Curvature: Bringing your
Curves to the BAR. Biochem. Soc. Symp. 72: 223-31.
Gallop, J. L., Jao, C. C., Kent, H. M., Butler, P. J. G., Evans, P. R., Langen, R., and McMahon, H. T.
(2006). Mechanism of Endophilin N-BAR Domain-Mediated Membrane Curvature. EMBO J. 25: 2898-
2910.
Garavito, R. M., and Rosenbusch, J. P. (1980). Three Dimensional Crystals of an Integral Membrane
Protein: An Initial X-Ray Analysis. J. Cell. Biol. 86: 327-329.
Ge, K., and Prendergast, G. C. (2000). Bin2, a Functionally Nonredundant Member of the BAR Adaptor
Gene Family. Genomics 67: 210-220.
Gennis, RB (1989). Biomembranes: Molecular Structure and Function. Springer-Verlag, New York,
USA.
Ghambir, A., Hangyás-Milhályné, G., Zaitseva, I., Cafiso, D. S., Wang, J., Murray, D., Pentyala, S. N.,
Smith, S. O., and McLaughlin, S. (2004). Electrostatic Sequestration of PIP 2 on Phospholipid Membranes
by Basic/Aromatic Regions of Proteins. Biophys. J. 86: 2188-2207.
Gil, T., Ipsen, J. H., Mouritsen, O. G. , Sabra, M. C., Sperotto, M. M., and Zuckermann, M. J. (1998).
Theoretical Analysis of Protein Organization in Lipid Membranes. Biochim. Biophys. Acta 1376: 245-
266.
194

BIBLIOGRAPHY
Glaubitz, C., Grobner, G., Watts, A. (2000). Structural and Orientational Information of the Membrane
Embedded M13 Coat Protein by 13C-MAS NMR Spectroscopy. Biochim. Biophys. Acta 1463: 151-161.
Gokhale, N. A., Abraham, A., Digman, M. A., Gratton, E., and Cho, W. (2005). Phosphoinositide
Specificity of and Mechanism of Lipid Domain Formation by Annexin A2-p11 Heterotetramer. J. Biol.
Chem. 280: 42831-42840.
Grabe, M., Wang, H., and Oster, G. (2000). The Mechanochemistry of V-ATPase Proton Pumps. Biophys.
J. 78: 2798-2813.
Harrison, M. A., Finbow M. E., and Findlay, J. B. (1997) Postulate for the Molecular Mechanism of the
Vacuolar H(+)-ATPase (Hypothesis). Mol. Membr. Biol. 14: 1-3.
Harrison, M. A., Murray, J., Powell, B., Kim, Y.-I., Finbow, M. E., and Findlay, J. B. C. (1999). Helical
Interactions and Membrane Disposition of the 16-kDa Proteolipid Subunit of the Vacuolar H + -ATPase
Analyzed by Cysteine Replacement Mutagenesis. J. Biol. Chem. 274: 25461-25470.
Harrison, M., Powell, B., Finbow, M. E., and Findlay, J. B. C. (2000). Identification of Lipid-Acessible
Sites on the Nephrops 16-kDa Proteolipid Incorporated into a Hybrid Vacuolar H + -ATPase: Site
Directed Labeling with N-(1-Pyrenyl)cyclohexylcarbodiimide and Fluorescence Quenching Analysis.
Biochemistry 39: 7531-7537.
Hardt, S. L. (1979). Rates of Diffusion Controlled Reactions in One, Two, Three Dimensions. Biophys.
Chem. 10: 239-243.
Harroun, T. A., Heller, W. T., Weiss, T. M., Yang, L., and Huang, H. W. (1999). Experimental Evidence
for Hydrophobic Matching and Membrane-Mediated Interactions in Lipid Bilayers Containing
Gramicidin. Biophys. J. 76: 937-945.
Harzer, U. and Bechinger, B. (2000). Alignment of Lysine-Anchored Membrane Peptides Under
Conditions of Hydrophobic Mismatch: a CD, 15N and 31P Solid-State NMR Spectroscopy Investigation.
Biochemistry 39: 13106-13114.
Hemminga, M. A., Sanders, J. C., and Spruijt, R. B. (1992). Spectroscopy of Lipid-Protein Interactions:
Structural Aspects of Two Different Forms of the Coat Protein of Bacteriophage M13 Incorporated in
Model Membranes. Prog. Lipid. Res. 31: 301-333.
Henry, G. D., and Sykes, B. D. (1992). Assignment of Amide 1 H and 15 N NMR Resonances in Detergent-
Solubilized M13 Coat Protein: A Model for the Coat Protein Dimer. Biochemistry 31: 5284-5297.
Hinderliter, A., Almeida, P. F. F., Creutz, C. E., and Biltonen, R. L. (2001). Domain Formation in a Fluid
Mixed Lipid Bilayer Modulated Through Binding of the C2 Protein Motif. Biochemistry 40: 4181-4191.
195

Hinderliter, A., Biltonen, R. L., and Almeida, P. F. F. (2004). Lipid Modulation of Protein-Induced
Membrane Domains as a Mechanism for Controlling Signal Transduction. Biochemistry 43: 7102-7110.
Hooper, D. C., and Wolfson, J. S. (Eds.), (1995). Quinolone Antimicrobial Agents, 2 nd Ed. American
Society from Microbiology, Washington D.C, U.S.A.
Houston, S., and Farming, A. (1994). Current and Potential Treatment of Tuberculosis. Drugs 48: 689-
708.
Ipsen, J. H., Karlstrom, G., Mouritsen, O. G., Wennerstrom, H., and Zuckermann, M. J. (1987). Phase
Equilibria in the Phosphatidylcholine-Cholesterol System. Biochem. Biophys. Acta 905:162–72.
Jensen, M. O, and Mouritsen, O. G. (2004). Lipids do Influence Protein Function – The Hydrophobic
Matching Hypothesis Revisited. Biochim. Biophys. Acta 1666: 205-226.
Jorgensen, K., and Mouritsen, O. G. (1995). Phase Separation Dynamics and Lateral Organization of
Two-Component Lipid Membranes. Biophys. J. 95: 942-954.
Kawasaki-Nishi, S. Nishi, T., and Forgac, M. (2003). Proton Translocation Driven by ATP Hydrolisis in
V-ATPases. FEBS Letters 545: 76-85.
Killian, J. A., Trouard, T. P., Greathouse, D. V., Chupin, V., Lindblom, G. (1994). A General Method for
the Preparation of Mixed Micelles of Hydrophobic Peptides and Sodium Dodecyl Sulphate. FEBS Lett.
348: 161-165.
Killian, J. A. (2003). Synthetic Peptides as Models for Intrinsic Membrane Proteins. FEBS Lett. 555:
134-138.
Kitamura, A., Kiyota, T., Tomohiro, M., Umeda, A., Lee, S., Inoue, T., and Sugihara, G. (1999).
Morphological Behaviour of Acidic and Neutral Liposomes Induced by Basic Amphiphilic Alpha-Helical
Peptides with Systematically Varied Hydrophobic-Hydrophilic Balance. Biophys. J. 76: 1457-1468.
Kuhn, A., Wickner, W., and Kreil, G. (1986). The Cytoplasmic Carboxy Terminus of M13 Procoat is
Required for the Membrane insertion of its Central Domain. Nature 322: 335-339.
Klausner, R. D., and Wolf, D. E. (1980). Selectivity of Fluorescent Lipid Analogues for Lipid Domains.
Biochemistry 19: 6199-6203.
Knol, J., Sjollema, K., and Poolman, B. (1998). Detergent-Mediated Reconstitution of Membrane
Proteins. Biochemistry 37: 16410-16415.
Korlach, J., Schwille, P., Webb, W.W., Feigenson, G.W. (1999). Characterization of Lipid Bilayer
Phases by Confocal Microscopy and Fluorescence Correlation Spectroscopy. Proc. Natl. Acad. Sci.
U.S.A. 96: 8461-8466.
196

BIBLIOGRAPHY
Koehorst, R. B. M., Spruijt, R. B., Vergeldt, F. J. and Hemminga, M. A. (2004). Lipid Bilayer Topology
of the Transmembrane α-Helix of M13 Major Coat Protein and Bilayer Polarity Profile by Site-Directed
Fluorescence Spectroscopy. Biophys. J. 87: 1445-1455.
Kwik, J., Boyle, S., Fooksman, Margolis, L., Sheetz, M. P., and Edidin, M. (2003). Membrane
Cholesterol, Lateral Mobility, and the Phosphatidylinositol 4,5-Bisphosphate-Dependent Organization of
Cell Actin. PNAS USA. 100: 13964-13969.
Lark, M. W., James, I. E. (2002). Novel Bone Antiresorptive Approaches. Curr. Opin. Pharmac. 2: 330-
337.
Laux, T., Fukami, K., Thelen, M., Golub, T., Frey, D., and Caroni, P. (2000). GAP43, MARCKS, and
CAP23 Modulate PI(4,5)P 2 at Plasmalemmal Rafts, and Regulate Cell Cortex Actin Dynamics Through a
Common Mechanism. J. Cell Biol. 149: 1455-1471.
Lee, A.G. (2003). Lipid-Protein Interactions in Biological Membranes: a Structural Perspective.
Biophys. Biochim. Acta 1612: 1-40.
Lehtonen, J. Y. A., Holopainen, J. M., and Kinnunen, P. K. (1996). Evidence for the Formation of
Microdomains in Liquid Crystalline Large Unilamellar Vesicles Caused by Hydrophobic Mismatch of the
Constituent Phospholipids. Biophys. J. 70: 1753-1760.
Lehtonen, J. Y. A., and Kinnunen, P. K. J. (1997). Evidence for Phospholipid Microdomain Formation in
Liquid Crystalline Liposomes Reconstituted With Eschericia coli Lactose Permease. Biophys. J. 72:
1247-1257.
Liu, F., Lewis, R. N. A. H., Hodges, R. S., and McElhaney, R. N. (2002). Effect of Variations in the
Structure of a Polyleucine-Based α-Helical Transmembrane Peptide on its Interactions with
Phosphatidylcholine Bilayers. Biochemistry 87: 2470-82
Lentz, B. R., Barenholz, Y. and Thompson, T. E. (1976). Fluorescence Depolarization Studies of Phase
Transitions and Fluidity in Phospholipid Bilayers. 2. Two-component Phosphatidylcholine Liposomes.
Biochemistry 15:4529–4537.
Lentz, T. L., Chaturverdi, V., and Conti-Fine, B. M. (1998). Amino Acids Within Residues 181-200 of the
Nicotinic Acetylcholine Receptor Alpha1 Subunit Involved in Nicotine Binding. Biochem. Pharmacol. 55:
341-347.
Lewis, B. A., and Engelman, D. M. (1983). Bacteriorhodopsin Remains Dispersed in Fluid Phospholipid
Bilayers Over a Wide Range of Bilayer Thickness. J. Mol. Biol. 166:203–210.
Liu, Y., Casey, L., and Pike, L. J. (1998). Compartmentalization of Phosphatidylinositol 4,5-
Bisphosphate in Low-Density Membrane Domains in the Absence of Caveolin. Biochem. Biophys. Res.
Commun. 245: 684-690.
197

Lodish, H., Berk, A., Zipursky, S. L., Matsudaira, P., Baltimore, D., and Darnell, J. E. (2000). Molecular
Cell Biology. 4 th Ed. W. H. Freeman, New York, USA
Mall, S., Broadbridge, R., Sharma, R. P., Lee, A. G., and East J. M. (2000). Effects of Aromatic Residues
at the Ends of Transmembrane Alpha-Helices on Helix Interactions With Bilayers. 39: 2071-2078.
Mall, S., East, J. M., and Lee, A. G. (2002). Transmembrane α-Helices. in Current Topics in Membranes
52: 339-371
Marsh, D. (1990). Lipid-Protein Interactions in Membranes. FEBS Lett. 268: 271-375.
Marsh, D. (1996). Peptide Models for Membrane Channels. Biochem, J. 315: 345-361.
Marsh, D., Horváth, L. I., Swamy, M. J., Mantripragada, S., and Kleinschmidt, J. H. (2002). Interactions
of Membrane-Spanning Proteins With Peripheral and Lipid-Anchored Membrane Proteins: Perspectives
From Protein-Lipid Interactions. Mol. Memb. Biol. 19: 247-255.
Martin, T. F. J. (2001). PI(4,5)P 2 Regulation of Surface Membrane Traffic. Curr. Opin. Cell Biol. 13: 493-
499.
Mascaretti, O. (2003). Bacteria Versus Antibacterial Agents: An Integrated Approach. ASM Press,
Washington D. C.
Masuda, M., Takeda, S., Sone, M., Ohki, T., Mori, H., Kamioka, Y., and Mochizuki, N. (2006).
Endophilin BAR Domain Drives Membrane Curvature By Two Newly Identified Structure-Based
Mechanisms. EMBO J. 25: 2889-2897.
Mateo, C. R., de Almeida, R. F. M., Loura, L. M. S., and Prieto, M. (2006). From Lipid Phases to
Membrane Protein Organization: Fluorescence Methodologies in the Study of Lipid-Protein Interactions.
in “Protein-Lipid Interactions: New Approaches and Emerging Concepts.” (Eds. Mateo, C. R., Gómez,
J., Villalaín, J., Ros, J. M. G.) Springer-Verlag. Berlin, Germany.
McIntosh, T. J., Vidal, A., and Simon, S. A. (2002). The Energetics of Peptide-Lipid Interactions:
Modulation by Interfacial Dipoles and Cholesterol. in Current Topics in Membranes 52: 205-253.
McLaughlin, S., Wang, J., Gambhir, A., and Murray, D. (2002). PIP 2 and Proteins: Interactions,
Organization, and Information Flow. Annu. Rev. Biophys. Struct. 31: 151-175.
McMahon, H. T., and Gallop, J. L. (2005). Membrane Curvature and Mechanisms of Dynamic Cell
Membrane Remodeling. Nature 438: 590-596.
Meijer, A. B., Spruijt, R. B., Wolfs, C. J. A. M., and Hemminga, M. A. (2001). Membrane-Anchoring
Interactions of M13 Major Coat Protein. Biochemistry 40: 8815-8820.
198

BIBLIOGRAPHY
Mishra, V. K., Palgunachari, M. N., Segrest, J. P., and Anantharamaiah, G. M. (1994). Interactions of
Synthetic Peptide Analogs of the Class A Amphipathic Helix With Lipids. Evidence for the Snorkel
Hypothesis. J. Biol. Chem. 269: 7185-7191.
Monné, M., Nilsson, I., Johansson, M., Elmhed, N., and von Heijne, G. (1998). Positively and Negatively
Charged Residues Have Different Effects on the Position in the Membrane of a Model Transmembrane
Helix. J. Mol. Biol. 284: 1177-1183.
Morein, S., Killian, J. A., and Sperotto M. (2002). Characterization of the Thermotropic Behaviour and
Lateral Organization of Lipid-Peptide Mixtures by a Combined Experimental and Theoretical Approach:
Effects of Hydrophobic Mismatch and Role of Flanking Residues. Biophys. J. 82: 1405-1417.
Mouritsen, O. G., and Bloom, M. (1984). Mattress Model of Lipid-Protein Interactions in Membranes.
Biophys. J. 46: 141-153.
Mouritsen, O. G., and Bloom, M. (1993). Models of Lipid-Protein Interactions in Membranes. Annu.
Rev. Biophys. Bioeng. 22: 145-171.
Mouritsen, O.G. (2005). Life – As a Matter of Fat. The Emerging Science of Lipidomics. Springer,
Heildelberg, Germany.
Nadler, G., Morvan, M., Delimoge, I., Belfiore, P., Zocchetti, A., James, I., Zembryki, D., Lee-
Rycakzewski, E., Parini, C., Consolandi, E., Gagliardi, S., and Farina, C. (1998). (2Z,4E)-5-(5,6-dichloro-
2-indolyl)-2-methoxy-n-(1,2,2,6,6-pentamethylpiperidin-4-yl)-2,4-pentadienamide, a Novel, Potent and
Selective Inhibitor of the Osteoclast V-ATPase. Bioorg. Medic. Chem. Lett. 8: 3621-3626.
Nagle, J. F., and Tritram-Nagle, S. (2000). Structure of Lipid Bilayers. Biochim. Biophys. Acta 1469:
159-195.
Neumann, D., Barchan D., Fridkint, M., and Fuchs, S. (1986). Analysis of Ligand Binding to the Synthetic
Dodecapeptide of the Acetylcholine Receptor a Subunit. Proc. Natl. Acad. Sci. USA 83: 9250-9253.
Neves, P., Berkane, E., Gameiro, P., Winterhalter, M., and de Castro, B. (2005). Interaction Between
Quinolones Antibiotics and Bacterial Outer Membrane Porin OmpF. Biophys. Chem. 113: 123-128.
Nielsen, C., Goulian, M., and Andersen, O. S. (1998). Energetics of Inclusion-Induced Bilayer
Deformations. Biophys. J. 74: 1966-1983.
Nieva, J. L., Goni, F. M., and Alonso, A. (1993). Phospholipase C-Promoted Membrane Fusion.
Retroinhibition by the End-Product Diacylglycerol. Biochemistry 32: 1054-1058.
Nikaido, H., Kim, S. H., and Rosenberg, E. Y. (1993). Physical Organization of Lipids in the Cell Wall of
Mycobacterium Chelonae. Mol. Microbiol. 8: 1025-1030.
199

Nikura, K., Takeshita, N., and Takano, M. (2005). A Vacuolar ATPase inhibitor, FR167356 Prevents
Bone Resorption in Ovariectomized Rats With High Potency and Specificity: Potential for Clinical
Application. J. Bone Miner. Res. 20: 1579-1588.
Nishi, T., and Forgac, M. (2002). The Vacuolar (H + ) - ATPases – Nature’s Most Versatile Proton Pumps.
Mol. Cell. Biol. 3: 94-103.
O´Keeffe, A. H., East, J. M., and Lee, A. G. (2000). Selectivity in Lipid Binding to the Bacterial Outer
Membrane Protein OmpF. Biophys. J. 79: 2066-2074.
Páli, T., Whyteside, G., Dixon, N., Kee, T. P., Ball, S., Harrison, M. A., Findlay, J. B. C., Finbow, M. E.,
and Marsh, D. (2004). Interaction of Inhibitors of the Vacuolar H + -ATPase With the Transmembrane V o -
sector. Biochemistry. 43: 12297-12305.
Pap, E. H. W., van Hoek, A., Wirtz, K. W. A., and Visser, A. J: W. G. (1995). Quantitative Analysis of
Lipid-Lipid and Lipid-Protein Interactions in Membranes by Use of Pyrene-Labeled Phosphoinositides.
Biochemistry 34: 9118-9125.
Papayannopoulos, V., Co, C., Prehoda, K. E., Snapper, S., Taunton, J., and Lim, W. A. (2005). A
Polybasic Motif Allows N-WASP to Act as a Sensor of PIP 2 Density. Moll. Cell 17: 181-191.
Peelen, S. J. C. J., Sanders, J. C., Hemminga, M. A. and Marsh, D. (1992). Stoichiometry, Selectivity and
Exchange Dynamics of Lipid-Protein Interaction with Bacteriophage M13 Coat Protein Studied by Spin
Label Electron Spin Resonance. Effects of Protein Secondary Structure. Biochemistry 31: 2670-2677.
Pérez-Gil, J., Cruz, A., and de la Serra, J. B. (2005). Lateral Membrane Structure and Lipid-Protein
Interactions. in “Protein-Lipid Interactions: New Approaches and Emerging Concepts.” (Eds. Mateo, C.
R., Gómez, J., Villalaín, J., Ros, J. M. G.) Springer-Verlag. Berlin, Germany.
Pestova, E., Millichap, J. J., Noskin, G. A., and Peterson, L. R. (2000). Intracellular Targets of
Moxifloxacin: A Comparison With Other Fluoroquinolones. J. Antimicrob. Chemother. 45: 583-590.
Peter, B. J., Kent, H. M., Mills, I.G., Vallis, Y., Butler, P. J. G., Evans, P. R., and McMahon, H. T.
(2004). BAR Domains as Sensors of Membrane Curvature: the Amphiphysin BAR Structure. Science 303:
495-499.
Petrangoline, G., Supino, R., Pratesi, G., Bo, L. D., Tortoreto, M., Croce, A. C., Misiano, P., Belfiore, P.,
Farina, C., and Zunino, F. (2006). Effect of a Novel Vacuolar-H + -ATPase Inhibitor on Cell and Tumor
Response to Camptothecins. Chemot. Antib. Gene Ther. 318: 939-946.
Pike, L. J., and Miller, J. M. (1998). Cholesterol Depletion Delocalizes Phosphatidylinositol
Bisphosphate and Inhibits Hormone-Stimulated Phosphatidylinositol Turnover. J. Biol. Chem. 273:
22298-22304.
200

BIBLIOGRAPHY
Pluschke, G., Hirota, Y, and Overath, P. (1978). Function of Phospholipids in Escherichia Coli.
Characterization of a Mutant Deficient in Cardiolipin Synthesis. J. Biol. Chem. 253: 5048-5055.
Popot, J.-L., and Engelman, D. M. (1990). Membrane Protein Folding and Oligomerization: The Ttwo-
Stage Model. Biochemistry 29: 4031-4037.
Popot, J. L. (1993). Integral Membrane Protein Structure: Transmembrane Alpha-Helices as
Autonomous Folding Domains.Curr. Opin. Struct. Biol. 3: 532-540.
Powl, A. M., East, J. M., and Lee, A. G. (2003). Lipid-Protein Interactions Studied by Introduction of a
Tryptophan Residue: The Mechanosensitive Channel MscL. Biochemistry 42: 14306-14317.
Quintiliani, R., Owens, R. C., and Grant, E. M. (1999). Clinical Role of Fluoroquinolones in Patients
With Respiratory Tract Infections. Infect. Dis. Clin. Pract. 8: 28-41.
Rauch, M. E., Ferguson, C. G., Prestwich, G. D., and Cafiso, D. S. (2002). Myristoylated Alanine-Rich C
Kinase Substrate (MARCKS) Sequesters Spin-labeled Phosphatidylinositol 4,5-Bisphosphate in Lipid
Bilayers. J. Biol. Chem. 277: 14068-14076.
Redfern, D. A., and Gericke, A. (2004). Domain Formation in Phosphatidylinositol
Monophosphate/Phosphatidylcholine Mixed Vesicles. Biophys. J. 86: 2980-2992.
Redfern, D.A., and Gericke, A. (2005). pH-Dependent Domain Formation in Phosphatidylinositol
Polyphosphate/Phosphatidylcholine Mixed Vesicles. J. Lipid Res. 46: 504-515.
Ren, J., Lew, S., Wang, J., and London, E. (1997). Transmembrane Orientation of Hydrophobic α-
Helices is Regulated Both by the Relationship of Helix Length to Bilayer Thickness and by the
Cholesterol Concentration. Biochemistry 36: 10213-10220.
Ren, J., Lew, S., Wang, J., and London, E. (1999). Control of the Transmembrane Orientation and
Interhelical Interactions Within Membranes by Hydrophobic Length. Biochemistry 38: 5905-5912
Richnau, N., Fransson, A., Farsad, K., and Aspenstrom, P. (2004). RICH-1 has a BIN/Amphiphysin/Rvsp
Domain Responsible for Binding to Membrane Lipids and Tubulation of Liposomes. Biochem. Biophys.
Res. Commun. 320: 1034-1042.
Ried, C., Wahl, C., Miethke, T., Wellnhofer, G., Landgraf, C., Schneider-Mergener, J., and Hoess, A.
(1996). High Affinity Endotoxin-Binding and Neutralizing Peptides Based on the Crystal Structure of
Recombinant Limulus Anti-Lipopolysaccharide Factor. J. Biol. Chem. 271: 28120-28127.
Rigaud, J.-L., Paternoste, M.-T., and Pitard, B. (1995). Reconstitution of Membrane Proteins Into
Liposomes: Application to Energy Transducing Membrane Proteins. Biochim. Biophys. Acta. 1231: 223-
246.
201

Rodan, G. A. and Martin, T. J. (2000). Therapeutic Approaches to Bone Diseases. 289: 1508-1414.
Rodriguez, N., Pincet, F., and Cribier, S. (2005) Giant Vesicles Formed by Gentle Hydration and
Electroformation: A Comparison by Fluorescence Microscopy. Coll. Surf. B: Biointerfaces 42: 125-130.
Russel, M. (1991). Filamentous Phage Assembly. Mol. Microbiol. 5: 1607-1613.
Sackmann, E. (1995). Biological Membranes Architecture and Function in Structure and Dynamics of
Membranes. From Cells to Vesicles. Eds Lipowsky, R., and Sackmann, E. North Holland, Amsterdam,
The Netherlands.
Sanders, J. C., Ottaviani, M. F., van Hoek, A., Visser, A. J. W. G., and Hemminga, M. A. (1992). A Small
Protein in Model Membranes: A Time Resolved Fluorescence and ESR Study on the Interaction of M13
Coat Protein With Lipid Bilayers. Eur. Biophys. J. 21: 305-311.
Santos, N. C., Prieto, M., and Castanho, M. A. R. B. (2003). Quantifying Molecular Partition into Model
Systems of Biomembranes: an Emphasis on Optical Spectroscopic Methods. Biophys. Biochim. Acta
1612: 123-135.
Scatigno, A. C., Garrido, S. S., and Marchetto, R. (2004). A 4.2 kDa Synthetic Peptide as a Potential
Probe to Evaluate the Antibacterial Activity of Coumarin Drugs. J. Peptide Sci. 10: 566-577.
Schiffer, M., and Edmundson, A. B. (1967). Use of Helical Wheels to Represent the Structures of
Proteins and to Identify Segments with Helical Potential. Biophys. J. 7: 121-135.
Seddon, J. M., and Templer, R. H. (1995). Polymorphism of Lipid-Water Systems in Structure and
Dynamics of Membranes. From Cells to Vesicles. Eds Lipowsky, R., and Sackmann, E. North Holland,
Amsterdam, The Netherlands.
Seddon, A. M., Curnow, P., and Booth, P. J. (2004). Membrane Proteins, Lipids and Detergents: Not Just
a Soap Opera. Biochim. Biophys. Acta 1666: 105-117.
Seelig, J. (2004). Thermodynamics of Lipid-Peptide Interactions. Biochim. Biophys. Acta 1666: 40-50.
Shimshick, E. J., and McConnell, H. M. (1973). Lateral Phase Separation in Phospholipid Membranes.
Biochemistry. 12:2351–2360.
Silvius, J. R. (1990). Calcium-induced Lipid Phase Separations and Interactions of
Phosphatidylcholine/Anionic Phospholipid Vesicles. Fluorescence Studies Using Carbazole-Labeled and
Brominated Phospholipids. Biochemistry. 24: 2930-2938.
Simons, K., and Ikonen, E. (1997). Functional Rafts in Cell Membranes. Nature 387:569-572.
Singer, S. J., and Nicolson, G. L. (1972). The Fluid Mosaic Model of the Structure of Cell Membranes.
Science 175: 720-731.
202

BIBLIOGRAPHY
Sperotto, M. M., and Mouritsen, O. G.(1991). Monte Carlo Simulation Studies of Lipid Order Parameter
Profiles Near Integral Membrane Proteins. Biochemistry 59: 261-270.
Spruijt, R., and Hemminga, M. A. (1991). The in Situ Aggregational and Conformational State of the
Major Coat Protein of Bacteriophage M13 in Phospholipid Bilayers Mimicking the Inner Membrane of
Host Escherichia Coli. Biochemistry 30: 11147-11154.
Spruijt, R., Wolfs, C., and Hemminga, M. (1999). M13 Phage. in “Encyclopedia of Molecular Biology.”
(Ed. Creighton, T. E.), Wiley-Interscience, New York, U.S.A.
Spuijt, R. B., Wolfs, C. J. A. M., Verveer, J. W. G., and Hemminga, M. (1996). Putative Transmembrane
Domain of the Major Coat Protein of Bacteriophage M13. Biochemistry 35: 10383-10391.
Stahlmannn, R., and Lode, H. (1999). Toxicity of Quinolones. Drugs 58: 37-42.
Stopar, D., Jansen, K. A. J., Páli, T., Marsh, D., and Hemminga, M. A. (1997a). Membrane Location of
Spin-Labeled M13 Major Coat Proteins Determined by Paramagnetic Relaxation Agents. Biochemistry
36: 8261-8268.
Stopar, D., Spruijt, R., Wolfs, C. J. A. M., and Hemminga, M. A. (1997b). In Situ Aggregational State of
M13 Bacteriophage Major Coat Protein in Sodium Cholate and Lipid Bilayers. Biochemistry 36: 12268-
12275.
Stopar, D., Spruijt, R., Wolfs, C. J. A. M., and Hemminga, M. A. (1998). Mimicking Initial Interactions
of Bacteriophage M13 Coat Protein Disassembly in Model Membrane Systems. Biochemistry 37: 10181-
10187.
Stopar, D., Spruijt, R., Wolfs, C. J. A. M., and Hemminga, M. A. (2003). Protein-Lipid Interactions of
Bacteriophage M13 Major Coat Protein. Biochim. Biophys. Acta 78451: 1-11.
Takei, K., Haucke, V., Slepnev, V., Farsad, K., Salazar, M., Chen, H. and De Camilli, P. (1998).
Generation of Coated Intermediates of Clathrin-Mediated Endocytosis on Protein-Free Liposomes. Cell
94: 131-141.
Takei, K., Slepnev, V. I., Haucke, V., and De Camilli, P. (1999). Functional Partnership Between
Amphiphysin and Dynamin in Clathrin-Mediated Endocytosis. Nat. Cell Biol. 1:33-39.
Tanford, C. (1980). The Hydrophobic Effect. Wiley, New York, USA.
Taylor, J. W. (1990). Peptide Models of dynorphin A(1-17) Incorporating Minimally Homologous
Substitutes for the Potential Amphiphilic β-Strand in Residues 7-15. Biochemistry 29: 5364-5373
Taylor, A. H., Heavner, G., Nedelman, M., Sherris, D., Brunt, E., Knight, D., and Ghrayeb, J. (1995).
Lipopolysaccharide (LPS) Neutralizing Peptides Reveal a Lipid A Binding Site of LPS Binding Protein.
203

J. Biol. Chem. 270: 17934-17938.
Teitelbaum, S. L. (2000). Bone Resorption by Osteoclasts. Science 289: 1504-1508.
van Rheenen, J., Achame, M. E., Janssen, H., Calafat, J., and Jalink, K. (2005). PIP 2 Signaling in Lipid
Domains: A Critical Re-evaluation. EMBO J. 24: 1664-1673.
Väänänen, K. (2005). Mechanism of osteoclast mediated bone resorption—rationale for the design of new
therapeutics. Adv. Drug Deliv. Rev. 57: 959-971
Vázquez, J. L., Merino, S., Domenech, Ò, Berlanga, M., Viñas, M., Montero, M. T., and Hernández-
Borrell, J. (2001). Determination of the Partition Coefficients of a Homologous Series of Ciprofloxacin:
Influence of the N-4 Piperazinyl Alkylation on the Antimicrobial Activity. Int. J. Pharm. 220: 53-62.
Visentin, L., Dodds, L. A., Valente, M., Misiano, P., Bradbeer, J. N., Oneta, S., Liang, X., Gowen, M.,
and Farina, C. (2000). A Selective Inhibitor of the Osteoclastic V-H + -ATPase Prevents Bone Loss in Both
Thyroparathyroidectomized and Ovariectomized Rats. J. Clin. Invest. 103: 309-318.
Wallin, E., Tsukihara, T., Yoshikawa, S., von Heijne, G., and Elofsson, A. (1997). Architecture of Helix
Bundle Membrane Proteins. An Analysis of Cytochrome c Oxidase from Bovine Mitochondria. Protein
Sci. 6: 808-815.
Wang, J., Arbuzova, A., Hangyás-Milhályné, G., and McLaughlin, S. (2001). The Effector Domain of
Myristoylated Alanine-rich C Kinase Substrate Binds Strongly to Phosphatidylinositol 4,5-Bisphosphate.
J. Biol. Chem. 276: 5012-5019.
Webb, M. S., Hui, S. W., and Steponkus, P. L. (1993). Dehydration Induced Lamellar-to-Hexagonal-II
Phase Transitions in DOPE/DOPC Mixtures. Biochim. Biophys. Acta. 1145:93-104.
Weiss, T. M., van der Wel, P. C. A., Killian, J. A., Koeppe II, R. E., and Huang, H. W. (2003).
Hydrophobic Mismatch between Helices and Lipid Bilayers. Biophys. J. 84: 379-385.
White, S. H., and Wimley, W. C. (1999). Membrane Protein Folding and Stability: Physical Principles.
Annu. Rev. Biophys. Biomol. Struct. 28: 319-365.
Whyteside, G., Meek, P. J., Ball, S. K., Dixon, N., Finbow, M. E., Kee, T. P., Findlay, J. B. C., and
Harrison, M .A. (2005). Concanamycin and Indolyl Pentadieneamide Inhibitors of the Vacuolar H+-
ATPase Bind with High Affinity to the Purified Proteolipid Subunit of the Membrane Domain.
Biochemistry. 44: 15024-15031.
Wiener, M. C., and White, S. H. (1992). Structure of a Fluid Dioleoylphosphatidylcholine Bilayer
Determined by Joint Refinement of X-Ray and Neutron Diffraction Data. III. Complete Structure.
Biophys. J. 61:434-447.
204

BIBLIOGRAPHY
Williamson, I. M., Alvis, S. J., East, J. M., and Lee, A. G. (2002). Interaction of Phospholipids with the
Potassium Channel KscA. Biophys. J. 83: 2026-2038.
Wilson, P. T., Lentz, T.L., and Hawrot, E. (1985). Determination of the Primary Amino Acid Sequence
Specifying the Alpha-Bungarotoxin Binding Site on the Alpha-Subunit of the Acetylcholine Receptor
From Torpedo Californica. Proc. Natl. Acad. Sci. USA 82: 8790-8794.
Wimley, W. C., and White, S. H. (1996). Experimentally Determined Hydrophobicity Scale for Proteins
at Membrane Interfaces. Nat. Struct. Biol. 3: 842-848.
Wimley, W.C., Creamer, T.P., and White, S.H. (1996) Solvation Energies of Amino Acid Residues and
Backbone in a Family of Host Guest Pentapeptides. Biochemistry 35: 5109-5124.
Yau, W.-M., Wimley, W. C., Gawrisch, K., and White, S. H. (1998). The Preference of Tryptophan for
Membrane Interfaces. Biochemistry 37: 14713-14718.
Yeagle, P. L. (1993). The Membranes of Cells. 2 nd Ed. Academic Press, San Diego, USA.
Yeagle, P. L., Bennet, M., Lemaître, V., and Watts, A. (2007). Transmembrane Helices of Membrane
Proteins May Flex to Satisfy Hydrophobic Mismatch. Biochim. Biophys. Acta 1768: 530-537.
Yokoyama, K., Nakano, M., Imamura, H., Yoshida, M., and Tamakoshi, M. (2003). Rotation of the
Proteolipid Ring in the V-ATPase. J. Biol. Chem. 278: 24255-24258.
Yoshida, Y., Kinuta, M., Abe, T., Liang, S., Araki, K., Cremona, O., Di Paolo, G., Moriyama, Y.,
Yasuda, T., De Camilli, P., and Takei, K. (2004). The Stimulatory Action of Amphiphysin on Dynamin
Function is Dependent on Lipid Bilayer Curvature. EMBO J. 23: 3483-3491.
Zimmerberg, J. (2000). Are the Curves in All the Right Places Traffic 1: 366-368.
Zimmerberg, J., and Kozlov, M. M. (2006). How Proteins Produce Cellular Membrane Curvature. Nat.
Rev. Mol. Cell Biol. 7: 9-19.
205