14.11.2012 Views

Antibody Engineering Conference, San Diego - MorphoSys

Antibody Engineering Conference, San Diego - MorphoSys

Antibody Engineering Conference, San Diego - MorphoSys

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

A New <strong>Antibody</strong> Library Concept (AB <strong>Engineering</strong>, <strong>San</strong> <strong>Diego</strong>, Dec. 2011)<br />

Stefanie Urlinger, Director R&D, <strong>MorphoSys</strong> AG<br />

Page 1 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Why a New <strong>Antibody</strong> Library?<br />

From: „Building Better <strong>Antibody</strong> Therapeutics“ by Patrick McGee, Drug Discovery<br />

and Development, 2007 (quotation Stephen Demarest, Biogen Idec):<br />

„The construction of nearly every human or humanized antibody drug candidate<br />

necessitates some unnatural protein engineering.“ […]<br />

„The only companies with success in this area are those that included stability and<br />

solubility as a key component of their library designs.“ […]<br />

„And while there have been many recent successes in this area, there have been<br />

numerous and costly failures over the past 15 years because stability was not always<br />

considered a key issue.“<br />

Building quality directly into the antibody library:<br />

Page 2 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


AGENDA<br />

1. Features of Ylanthia<br />

2. Realization of the Concept<br />

3. Quality Control<br />

4. Library Performance<br />

Page 3 2011 <strong>Antibody</strong> <strong>Engineering</strong> © © <strong>MorphoSys</strong> AG AG


Analyzing the Natural Human <strong>Antibody</strong><br />

Repertoire<br />

Ranking of most prominent<br />

variable region genes and<br />

VH/VL pairings:<br />

� Extraction of V gene usage and<br />

VH/VL pairing frequencies from<br />

own B-cell sequencing efforts and<br />

several publications analyzing<br />

both, autoimmune and healthy<br />

people*.<br />

� Analysis of VH/Vk as well as<br />

VH/Vλ pairings.<br />

Ranking according to natural<br />

occurrence.<br />

* Wardemann H. et al. Science (2003); Yurasov S. et al. JEM (2005); Tsuiji M. et al. JEM (2006); Yurasov S. et al. JEM (2006);<br />

Tiller T. et al. Immunity (2007); Mietzner B. PNAS (2008); Kofer J. et al: unpublished/ personal communication;<br />

Brezinschek H.P. et al. JCI (1997); Demaisson C. et al. Immunogenetics (1995); Foster S.J. at al. JCI (1997)<br />

VH<br />

Page 4 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG<br />

V�


Selection of Top 20 HC and 20 LC Germline<br />

Genes<br />

Human antibody repertoire:<br />

� 40 functional Vkappa<br />

� 30 Vlambda<br />

� 50 VH segments germline encoded.<br />

In silico pre-selection of 20 HC & LC:<br />

� Natural prevalence in human rearranged<br />

antibodies<br />

� Diversity in variable region gene families<br />

� Diversity in canonical CDR structures<br />

� Number of potential post-translational<br />

modification sites (PTMs)<br />

� Isoelectric point (pI)<br />

In vitro testing of 400 HC/LC pairs to<br />

select for pairs with most favorable<br />

biophysical properties<br />

* strong T-cell epitope only with rare DRB1*0411 allele<br />

Page 5 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Ylanthia: Distinct Heavy/Light Chain Pairing<br />

� Single framework libraries:<br />

HC/LC pair selected for biophysical properties;<br />

limited structural repertoire<br />

� Multiple framework synthetic and PCR-based<br />

libraries: High structural diversity; but random<br />

HC/LC pairing might yield to, e.g., antibody<br />

instability<br />

� Ylanthia: 36 distinct HC/LC pairs<br />

pre-selected for favorable biophysical<br />

properties<br />

� High structural diversity through main canonical<br />

CDR conformations of HC & LC<br />

� Only stable, well expressing HC/LC combinations<br />

Page 6 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Reduction of Potential Post-Translational<br />

Modification Sites (PTMs)<br />

� Fully germline human antibodies contain a relevant number of potential PTMs,<br />

especially in CDR-H1 and CDR-H2<br />

� Ylanthia features fully germline and “no PTM” CDR-H1 and -H2 sequences<br />

� Completely removed from CDR-H1 & -H2:<br />

� Deamidation sites: NS, NG, NH<br />

� Isomerization sites: DS, DG, DD<br />

� Cleavage sites: DP, DQ, NS<br />

� Oxidation sites: M<br />

� Glycosylation sites: NxT, NxS<br />

� CDR-H3: Asn (N) is omitted and Asp (D), Gln (Q) and Met (M) are decreased<br />

Page 7 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


AGENDA<br />

1. Features of Ylanthia<br />

2. Realization of the Concept<br />

3. Quality Control<br />

4. Library Performance<br />

Page 8 2011 <strong>Antibody</strong> <strong>Engineering</strong> © © <strong>MorphoSys</strong> AG AG


Screening 400 HC/LC Pairs for Biophysical<br />

Properties<br />

Fab (20 HC x 20 LC = 400 combinations)<br />

Pool cloning<br />

Fab display vector Fab expression vector<br />

(1) Phage display (2) Expression yield<br />

(3) Thermal stability<br />

(4) Serum stability<br />

Ranking; Exclusion of unfavorable HC/LC pairs or frameworks<br />

Pre-selection of ~100 HC/LC combinations<br />

� Fab and IgG1 expression & purification<br />

� Biophysical characterization<br />

� Expression yields (Fab & IgG1)<br />

� Aggregation propensity<br />

� Apparent Tm determination<br />

� Physical stress testing<br />

� Serum stability<br />

IgG1 (20 HC x 20 LC = 400 combinations)<br />

Single antibody cloning<br />

Human IgG1 expression vector<br />

(5) Expression yield<br />

(6) Serum stability<br />

Final master gene selection<br />

30-40 distinct HC/LC pairs.<br />

Page 9 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Screening for Fab Phage Display Levels<br />

Step 1: Gene synthesis of 20 HCs and 20 LCs (including gene and codon optimization)<br />

Step 2: Pool sub-cloning into CysDisplay and Fab expression vectors<br />

Step 3: Set-up of predictive screening assays for Fab display levels on phage, Fab expression<br />

levels and Fab thermostability from crude bacterial expression samples<br />

Step 4: ELISA-based screening of HC/LC pairs; sequence analysis for identification of ≥ 75% of<br />

HC/LC pairs (> 50% at least double determination)<br />

Example: Determination of Fab phage display levels in ELISA format<br />

cultivation of E. coli<br />

carrying phagemid;<br />

helper phage infection<br />

and antibody phage<br />

production in 96-well<br />

format<br />

phage capture<br />

via anti-g8p<br />

phage capture<br />

via anti-Fab<br />

anti-g8p<br />

detection<br />

calculation of relative<br />

Fab display rates<br />

(g8p vs. Fab-specific<br />

signal) using<br />

reference phage<br />

prep<br />

Page 10 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Screening for Fab Phage Display Levels<br />

VH1<br />

VH3<br />

VH4,5,6<br />

kappa<br />

lambda<br />

Framework combinations are selected for high display levels in CysDisplay format.<br />

HCs showing low display are excluded (VH1-2, VH3-73, VH4-31, -39).<br />

Page 11 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Screening for Thermostability<br />

VH1<br />

VH3<br />

VH4,5,6<br />

kappa<br />

lambda<br />

Framework combinations are selected for apparent Fab thermostability.<br />

Instable HCs and LCs are excluded (e.g., VH1-2, VK1-17, VK2-30).<br />

Page 12 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


…<br />

Screening for IgG1 Expression Levels<br />

Step 5: Set-up of sandwich ELISA for IgG1 quantification from mammalian cell culture supernatants<br />

Step 6: Pool subcloning of HC & LC into mammalian hIgG1 expression vectors (“2-vector system”)<br />

Step 7: Cross-transfection of 20 HC with 20 LC into HEK.EBNA cells and IgG1 expression<br />

Step 8: IgG1 quantification in 384-well format<br />

Example: Determination of IgG1 levels from cell culture supernatants<br />

LC<br />

cross-transfection<br />

of 20 heavy chain<br />

with 20 light chain<br />

plasmids into<br />

HEK.EBNA<br />

HC<br />

HEK.EBNA<br />

cultivation and<br />

antibody<br />

production of<br />

400 pairs<br />

mouse antihuman<br />

IgG1<br />

capture from<br />

cell culture<br />

supernatants<br />

detection via<br />

anti-human IgG1<br />

specific<br />

biotinylated<br />

antibody (not<br />

cross-reactive<br />

with mouse Ig)<br />

calculation of<br />

relative IgG1<br />

expression levels<br />

via signal<br />

obtained with<br />

reference<br />

antibody<br />

Page 13 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Screening for IgG1 Expression Levels<br />

VH1<br />

VH3<br />

VH4,5,6<br />

kappa<br />

lambda<br />

Framework combinations are selected for human IgG1 expression levels<br />

and low expressing HCs are excluded (e.g., VH1_2, VH4_31).<br />

Page 14 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Combination of all Screening Parameters<br />

low Fab display<br />

low Fab expression level<br />

low Fab thermostability<br />

moderate hIgG1 expression<br />

high Fab display<br />

moderate Fab expression level<br />

high Fab thermostability<br />

high hIgG1 expression<br />

Selection of ~100 HC/LC pairs with favorable properties for research scale expression,<br />

purification and biophysical testing.<br />

Page 15 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


In Depth Biophysical Characterization of<br />

Purified HC/LC Pairs<br />

� 100 HC/LC pairs pre-selected by ELISA-based screenings were<br />

expressed and purified in mg amounts, both in Fab and IgG1 formats:<br />

� Quantification of expression yields<br />

� Determination of monomeric content by analytical SEC<br />

� Apparent Tm by Thermofluor<br />

� Serum stability testing<br />

� Stress testing (turbidity/particle formation upon physical & pH stress)<br />

Ylanthia frameworks: 36 HC/LC pairs with optimal biophysical properties,<br />

plus featuring a diverse set of canonical CDR conformations for<br />

broad epitope coverage.<br />

Page 16 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Ylanthia CDRs: Slonomics Inside<br />

Page 17 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


CDR-H3: JH Gene Contribution<br />

� Predominantly<br />

used HC joining<br />

region in<br />

rearranged human<br />

antibodies: JH4<br />

YFDY…<br />

� Long CDR-H3s:<br />

JH6 usage<br />

YYYYYGMDV…<br />

increases with<br />

length<br />

Page 18 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Ylanthia CDR-H3 Design<br />

� Lengths 6 to 17 with JH4 based design<br />

� JH4: dominant joining region in short to medium length human CDR-H3s<br />

� Lengths 12 to 17 have additional JH6 based design<br />

� JH6 gains importance with increasing CDR-H3 length<br />

� In CDR-H3 Asn (N) is completely omitted; Asp (D) and Met (M) are decreased to abolish or<br />

reduce occurrence of critical PTM sites<br />

Page 19 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


CDR-H3 Length Distribution<br />

� Twelve different CDR-<br />

H3 lengths ranging<br />

from 6 to 17 amino<br />

acids<br />

� Covering 80% of the<br />

natural human CDR-H3<br />

diversity (according to<br />

Zemlin et al., 2003)<br />

� Lengths 6 to 17:<br />

JH4 specific design<br />

� Lengths 12 to 17:<br />

additional JH6 specific<br />

design<br />

Page 20 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


AGENDA<br />

1. Features of Ylanthia<br />

2. Realization of the Concept<br />

3. Quality Control<br />

4. Library Performance<br />

Page 21 2011 <strong>Antibody</strong> <strong>Engineering</strong> © © <strong>MorphoSys</strong> AG AG


Library Size and Correctness<br />

� Sub-library sizes vary between 5E+8 and 4E+9<br />

clones.<br />

� Overall correctness: ~85%.<br />

Total library size:<br />

1.3E+11 independent clones with more than<br />

hundred billion fully correct human antibodies!<br />

Page 22 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Library QC by Next Generation Sequencing:<br />

Redundancy<br />

� 454 sequencing of VH<br />

of the unselected<br />

library (performed using<br />

pan-selective primers<br />

recognizing all HCs;<br />

kappa and lambda<br />

libraries „tagged“)<br />

� Proprietary software for<br />

data evaluation:<br />

More than 151.000<br />

Ylanthia VH sequences<br />

analyzed<br />

Number of replicates:<br />

High sequence diversity: From more than 151.000 sequences 94% were CDR-<br />

H3 unique (< 6% sequences found in duplicate).<br />

Page 23 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Library QC by Next Generation Sequencing:<br />

CDR-H3 Length Distribution<br />

� Analysis of more than 151.000 unselected Ylanthia VH sequences:<br />

Unselected CDR-H3 length distribution matches design.<br />

Page 24 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Library Features – Correctness<br />

� QC of final library:<br />

� Sequence analysis of VL<br />

and corresponding VH of at<br />

least 48 clones per sub-<br />

library; in total almost 4000<br />

antibodies sequenced and<br />

evaluated.<br />

Overall library correctness:<br />

~85%<br />

failed<br />

false<br />

TRIM TRIM Slonomics<br />

kappa lambda LC HC<br />

total sequences 2367 1482 3849 3849<br />

mixed or bad sequence 4.3% 4.4% 4.4% 4.8%<br />

seq break<br />

before KpnI or XhoI<br />

0% 0% 0% 3.6%<br />

analyzed sequences 2265 1416 3681 3523<br />

frameshifts 6.2% 8.8% 7.2% 3.0%<br />

mixed in AND after<br />

CDR3<br />

0.6% 0.8% 0.7% 2.2%<br />

stop mutation<br />

(w/o frameshift)<br />

0.3% 0.1% 0.2% 0%<br />

multiple inserts 0% 0% 0% 4.0%<br />

functional<br />

clones<br />

93% 90% 92% 91%<br />

undesired CDR3 length 0.6% 0.2% 0.5% 0.2%<br />

undesired aa 0.7% 0.5% 0.6% 0.4%<br />

correctness 92% 90% 91% 90%<br />

Page 25 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


AGENDA<br />

1. Features of Ylanthia<br />

2. Realization of the Concept<br />

3. Quality Control<br />

4. Library Performance<br />

Page 26 2011 <strong>Antibody</strong> <strong>Engineering</strong> © © <strong>MorphoSys</strong> AG AG


<strong>Antibody</strong> Selections – Snapshot*<br />

*beta-testing in progress<br />

� So far, antibodies from<br />

27/36 tested VH/VL<br />

pairs identified<br />

� Different targets favor<br />

different VH/VL pairs<br />

� A minimum of 94<br />

antibodies identified<br />

per target from a<br />

limited number of<br />

selections and<br />

analyzed sequences of<br />

the germline version<br />

only (only 36 VH/VL<br />

pairs of 72 tested)<br />

Page 27 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Biophysical Properties of Selected Antibodies<br />

Isoelectric Point (pI)<br />

The calculated pI values of the selected antibodies are mostly above 8.5,<br />

even for lambda antibodies.<br />

� The pI values were<br />

calculated using VNTI<br />

software.<br />

� Rational CDR3 design<br />

results in an increase of pI<br />

values in most selected<br />

antibodies (as compared<br />

to the initial VH/VL pair;<br />

or ) .<br />

Page 28 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Biophysical Properties of Selected Antibodies<br />

Thermostability<br />

� IgG1 Melting Temperature<br />

(Tm) was determined by<br />

Thermofluor.<br />

� The median thermal<br />

stability for each family<br />

equals the parental<br />

antibody ( or ) that<br />

was initially used for<br />

selecting the VH/VL pairs.<br />

The Melting Temperature of the selected antibodies clusters around the apparent<br />

Tm of the initial HC/LC pairs. 90% of tested antibodies show Tm >68°C.<br />

Page 29 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Biophysical Properties of Selected Antibodies<br />

IgG1 Production Yields<br />

More than 95% of the antibodies express more than 10 mg IgG1 per liter in a three<br />

day transient mammalian expression culture as determined after purification.<br />

Average production yields: ~25 mg/L.<br />

� Human IgG1 antibodies<br />

were transiently expressed<br />

for 3 days in HKB11 cells<br />

in shake flasks and<br />

purified via Protein A.<br />

Page 30 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


Biophysical Properties of Selected Antibodies<br />

Aggregation Propensity<br />

90% of the IgGs show


Ylanthia ® – Moving Further Towards<br />

Quality by Design<br />

� New antibody library platform*<br />

� 36 fixed HC/LC combinations:<br />

� High structural diversity<br />

� HC/LC pairs pre-selected for optimal<br />

biophysical properties<br />

� More than hundred billion fully human and<br />

entirely correct antibodies<br />

� Full flexibility for engineering;<br />

rapid Ig conversion<br />

� Slonomics inside<br />

� Applies Slonomics for tailored antibody<br />

optimization (“arYla ® ”)<br />

*beta-testing in progress<br />

Ailanthus spec: Tree of the Gods,<br />

Tree of Heaven<br />

Page 32 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG


www.morphosys.com<br />

HuCAL Page 33 2011 <strong>Antibody</strong> <strong>Engineering</strong><br />

© <strong>MorphoSys</strong> AG<br />

® , HuCAL GOLD ® , HuCAL PLATINUM ® , CysDisplay ® , RapMAT ® , AutoCAL ® , arYla ® and Ylanthia ® are registered trademarks of <strong>MorphoSys</strong> AG.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!