Real time PCR - European Pharmaceutical Review
Real time PCR - European Pharmaceutical Review
Real time PCR - European Pharmaceutical Review
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
TECHNOLOGY /<br />
KNOWLEDGE /<br />
INNOVATION ISSUE ONE 2009<br />
Part One:<br />
Lab<br />
Automation<br />
Series<br />
Using PAT for<br />
Stem Cell<br />
manufacture<br />
<strong>Real</strong> <strong>time</strong><br />
<strong>PCR</strong><br />
HCS<br />
Technology<br />
SEE IT...<br />
CLICK IT!<br />
@<br />
<strong>PCR</strong><br />
HCS
TECHNOLOGY /<br />
KNOWLEDGE /<br />
INNOVATION ISSUE ONE 2009<br />
Part One:<br />
Lab<br />
Automation<br />
Series<br />
Using PAT for<br />
Stem Cell<br />
manufacture<br />
<strong>Real</strong> <strong>time</strong><br />
<strong>PCR</strong><br />
HCS<br />
Technology<br />
SEE IT...<br />
CLICK IT!<br />
@<br />
<strong>PCR</strong><br />
HCS
EDITORIAL BOARD<br />
Dr Anthony Davies<br />
HCA Research Facility Manager<br />
Trinity College Dublin<br />
Dr Sheraz Gul<br />
Vice President and Head of Biology<br />
<strong>European</strong> ScreeningPort GmbH<br />
Dr. Anne Katrin Werenskiold<br />
Project Manager<br />
Interaction Proteome<br />
Introduction<br />
Ken Leiper<br />
Benson Associates<br />
Dr. Gordon Alton<br />
Head of Kinase Screening<br />
Pfizer Global Research and Development<br />
Matthew Moran<br />
Director of the Irish <strong>Pharmaceutical</strong> and<br />
Chemical Manufacturers Federation<br />
Dr. Michael J Dunn<br />
SFI Research Professor of Biomedical Proteomics,<br />
University College Dublin<br />
Don Clark<br />
Pfizer Global Research & Development<br />
RUSSELL PUBLISHING LTD<br />
Founder Ian Russell<br />
Managing Director Vivien Cotterill-Lee<br />
Commissioning Editor Carrie Lancaster<br />
Publications Assistant Pippa McCartney<br />
Editorial Manager Craig Waters<br />
Assistant Editorial Manager Sarah Wills<br />
Sales Director Gary King<br />
Sales Executive Jay Vencatasen<br />
Publishing Manager Samantha Hay<br />
Production Team Brian Cloke<br />
Jat Garcha<br />
Front Cover Artwork Drew Hillier<br />
<strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> Digital is published to address new<br />
technologies, developments and advancements within the<br />
pharmaceutical manufacturing industries throughout Europe. The<br />
readership of <strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> Digital comprises key senior<br />
management involved in decision making and procurement functions.<br />
<strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> Digital is available by subscription at £90<br />
for a year (six issues). Subscription enquiries to Pippa McCartney:<br />
pmccartney@russellpublishing.com or telephone +44 (0) 1959 563311<br />
<strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> Digital:<br />
Published by Russell Publishing Ltd<br />
Court Lodge, Hogtrough Hill, Brasted,<br />
Kent, TN16 1NU, UK<br />
Tel: +44 (0) 1959 563311<br />
Fax: +44 (0) 1959 563123<br />
Email: pharma@russellpublishing.com<br />
www.europeanpharmaceuticalreview.com<br />
ISSN 1759-1279<br />
Copyright rests with the publishers.<br />
All rights reserved<br />
©2009 Russell Publishing Limited<br />
Independent audit<br />
watchdog service for<br />
printed publications<br />
SUBSCRIBE<br />
Registered Office as above.<br />
Russell Publishing Ltd, is registered as a Limited<br />
Company in England, Number 2709148<br />
VAT Number GB 577 8978 47<br />
Carrie Lancaster<br />
Commissioning Editor<br />
<strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong><br />
email: clancaster@russellpublishing.com<br />
Let’s get digital!<br />
Welcome to Issue 1/2009 of <strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> Digital.<br />
As we continue to take our place in the ‘digital age’ I hope that<br />
our first issue of <strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> Digital,<br />
published way back in November, was able to excite and inspire our<br />
readers as always. I also hope that it provided an insightful look at<br />
how we see our publication in the digitalised future. We are very<br />
happy so far with how <strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> Digital is<br />
looking and hope to make a few more improvements in the future….<br />
So keep reading!<br />
In this issue we are welcoming Dr Sheraz Gul from the <strong>European</strong><br />
ScreeningPort, not only as an author, but also as a new member of<br />
our Editorial Board. Sheraz will be writing a ‘Lab Automation Series’<br />
for us throughout 2009 which will feature in every issue of<br />
<strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> Digital. In his first instalment he<br />
looks at automation solutions for High Throughput Screening.<br />
To read more see page 25.<br />
We are also featuring an article which looks at Thermal Analysis<br />
and its use as a Process Analytical Tool, written by Dr Simon<br />
Gaisford. Simon is currently a Senior Lecturer at the University of<br />
London and he is also a co-founder of Synectix <strong>Pharmaceutical</strong><br />
Solutions Ltd. To view Simon’s article please visit page 28.<br />
Of course if you would like to contribute to a future issue of<br />
<strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> Digital, please do not hesitate to<br />
contact me via email at clancaster@russellpublishing.com.<br />
<strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> can guarantee its circulation<br />
is 11,994 (December 2007). The publication is ABC audited.<br />
This is an independent verification that our circulation is genuine.<br />
No part of this publication may be reproduced, stored in a retrieval system or transmitted in any<br />
form or by any means electronic, mechanical, photo copying, recording or other wise without the<br />
prior written permission of the copyright holder. Published February 2009. While the publishers<br />
believe that all information contained in this publication was correct at the <strong>time</strong> of going to press,<br />
they can accept no liability for any inaccuracies that may appear or loss suffered directly or<br />
indirectly by any reader as a result of any advertisement, editorial, photographs or other material<br />
published in <strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong> Digital.<br />
Carrie Lancaster<br />
Commissioning Editor<br />
www.europeanpharmaceuticalreview.com<br />
03
Amaxa ®<br />
Nucleofector ®<br />
Transfection Technology<br />
Now Available from Lonza<br />
Guaranteed to perform with many of our primary cells<br />
Visit www.lonza.com/amaxa for details<br />
Get More With Your Cells<br />
Clonetics ® and Poietics ® Primary Cells and Services<br />
Lonza has the most extensive offering of cryopreserved primary<br />
human and animal cells including epithelial, endothelial, fibroblast,<br />
keratinocyte, bone, neural, muscle, hepatocyte, stem and immune<br />
cells.<br />
Take advantage of our value-added services that are included when<br />
you use Lonza’s Clonetics® and Poietics® Primary Cells.<br />
Choose from our large selection of 5 or more different lots for<br />
each cell type based on your donor requirements<br />
Reserve and test specific lots of cells to ensure performance<br />
in your application prior to purchase<br />
Reserve large quantities of cells and media for large projects<br />
with a dedicated Lonza Reserve Manager<br />
Attend our Lonza Discovery Webinars to gain more product<br />
and application knowledge with primary cells and assays<br />
Transfection of primary cells with up to 90% efficiency using<br />
Amaxa® Nucleofector® Technology<br />
For more information, go to www.lonza.com/certifiedcells<br />
Unless otherwise noted, all trademarks herein are marks of the Lonza Group or its affiliates.<br />
© Copyright 2008, Lonza Walkersville, Inc. www.lonza.com
ISSUE ONE 2009<br />
Contents<br />
Index to Advertisers<br />
03 INTRODUCTION<br />
Lets get digital!<br />
Carrie Lancaster,<br />
Commissioning Editor,<br />
<strong>European</strong> <strong>Pharmaceutical</strong><br />
<strong>Review</strong> Digital<br />
25<br />
LAB AUTOMATION<br />
Article 1:<br />
Lab Automation<br />
series<br />
Dr Sheraz Gul, Vice President<br />
and Head of Biology,<br />
<strong>European</strong> ScreeningPort<br />
Almac Group..................................... 02<br />
Lonza..................................................... 04<br />
Thermo Fisher Scientific<br />
(Abgene Products).................. 07<br />
USB Europe ........................................ 11<br />
Fluidigm................................................ 15<br />
Thermo Fisher Scientific<br />
(Cellomics Products).............. 17<br />
MDS Analytical Technologies... 23<br />
IDBS....................................................... 27<br />
Mettler Toledo .................................. 31<br />
Buchi...................................................... 34<br />
To Advertise in the next issue,<br />
contact Gary King<br />
Tel: +44 (0) 1959 563311<br />
email: gking@russellpublishing.com<br />
06 q<strong>PCR</strong><br />
<strong>Real</strong> <strong>time</strong><br />
<strong>PCR</strong><br />
Dr. Frédérique Ponchel,<br />
Senior Research Fellow,<br />
Leeds Institute of Molecular<br />
Medicine, University of Leeds<br />
28<br />
THERMAL ANALYSIS<br />
Calorimetry as a<br />
Process Analytical<br />
Tool for micronising<br />
pharmaceuticals<br />
Simon Gaisford, Lecturer<br />
in Pharmaceutics,<br />
University of London<br />
32 PAT<br />
PAT and QbD<br />
aspects on stem cell<br />
manufacture<br />
Professor Carl-Fredrik<br />
Mandenius, Head of the<br />
Division of Biotechnology and<br />
Mats Björkman, Head of the<br />
Division of Assembly<br />
Technology/Production<br />
Engineering, Linköping<br />
University, Sweden<br />
16<br />
HIGH CONTENT SCREENING<br />
Open source data management in<br />
High Content Screening technology<br />
Karol Kozak, Head of Data Handling Facility, Dr Gabor Csucs, Head<br />
of the High-thorughput/high-content screening facility, LMC-RISC,<br />
ETH Zurich, Switzerland and Andrzej Firkowski, Professor of Chemical<br />
Technology at the Technical University of Radom, Director of the<br />
Faculty of Materials Science and Technology and a member of the<br />
Advisory Board, Noster-IT, Dresden, Germany<br />
38 PHARMAINFOCUS<br />
A REGULAR ROUND-UP OF<br />
INDUSTRY PRODUCTS,<br />
INNOVATION, NEWS AND<br />
DEVELOPMENTS<br />
05<br />
www.europeanpharmaceuticalreview.com
q<strong>PCR</strong> ISSUE 2009<br />
<strong>Real</strong> <strong>time</strong> <strong>PCR</strong><br />
Dr. Frédérique Ponchel, Senior Research Fellow, Leeds Institute of Molecular Medicine, University of Leeds<br />
Polymerase Chain Reaction (<strong>PCR</strong>) has revolutionised molecular biology, and is now a wellstandardised<br />
procedure and its applications are wide. <strong>Real</strong> <strong>time</strong> <strong>PCR</strong> is a more recent development<br />
of the initial <strong>PCR</strong> reaction, which has provided a solution to the limitation of standard <strong>PCR</strong> with<br />
regards to quantification. By using fluorescence, which allows the quantification of <strong>PCR</strong> products at<br />
each cycle of the reaction, real <strong>time</strong> <strong>PCR</strong> avoids the “plateau” effect of standard <strong>PCR</strong>. Two main<br />
approaches have been developed, the TaqMAn assay and the SYBR-Green I dye. Alongside real<strong>time</strong><br />
<strong>PCR</strong> as such, an “end point assay” has also evolved, most often referred to as allelic<br />
discrimination, allowing the detection of a specific sequence (i.e. genetic polymorphism). The<br />
technology is now widely used for DNA and RNA quantification as well as genetic polymorphism<br />
detection. It can be applied to the detection and quantification of plain mRNA, for splice variant,<br />
alternative promoter usage, allele specific expression etc. For DNA, gene amplification, gene<br />
deletion or rearrangement, allelic copy number can be measured. It is also widely used for the<br />
detection of pathogens or contaminants in food or other types of products.<br />
Figure 1: <strong>PCR</strong> can amplify a small amount of template DNA (or RNA) into large quantities<br />
in a few hours. This is performed by mixing the DNA with primers on either side of the<br />
DNA (forward F and reverse R), Taq polymerase, free nucleotides (dNTPs), and buffer.<br />
The temperature is then alternated between hot and cold to denature and reanneal the<br />
DNA, with the polymerase adding new complementary strands each <strong>time</strong>.<br />
Adapted from http://www.slideworld.org/viewslides.aspx/<strong>Real</strong>%20Time%20<strong>PCR</strong>%20principle%20%20chemistry%20%20application%20-%2066144<br />
The advent of Polymerase Chain<br />
Reaction (<strong>PCR</strong>) almost 30 years<br />
ago has revolutionised<br />
molecular biology. The principles of<br />
<strong>PCR</strong> are simple but rely on a<br />
particular type of DNA poplymerase,<br />
the Taq polymerase (from Thermus<br />
aquaticus). This form of the enzyme<br />
is thermophile and its polymerase<br />
activity is able to withstand<br />
extremely high temperatures allowing<br />
cycles of high and low temperature.<br />
The <strong>PCR</strong> reaction amplifies a low copy<br />
number nucleic acid template (DNA<br />
or RNA) into detectable amount,<br />
allowing its analysis (see Figure 1).<br />
<strong>PCR</strong> is now a well-standardised<br />
procedure and its application are wide.<br />
<strong>Real</strong> <strong>time</strong> <strong>PCR</strong> is a more recent<br />
development of the initial <strong>PCR</strong><br />
reaction and its popularity has never<br />
ceased growing since the late 90s as<br />
illustrated by the increasing numbers<br />
of publication using this technique<br />
(see Table 1 on page 11). It resolved<br />
the main limitation of the standard<br />
06<br />
www.europeanpharmaceuticalreview.com
© 2009 Thermo Fisher Scientific Inc. All rights reserved.<br />
All trademarks are the property of Thermo Fisher Scientific Inc. and its subsidiaries.<br />
Tired of waiting for Q<strong>PCR</strong> results<br />
Adopting a Fast Q<strong>PCR</strong> protocol can save a lot of <strong>time</strong>. Thermo<br />
Scientific ABsolute Fast Q<strong>PCR</strong> Mixes have been specially developed to<br />
deliver optimal Q<strong>PCR</strong> results in a fraction of the <strong>time</strong>.<br />
Fast enzyme activation and thermal cycling for<br />
rapid collection of high quality Q<strong>PCR</strong> data<br />
Unique blue mix for error free reaction<br />
setup and confidence in assay results<br />
Highly optimized mix for industry<br />
leading assay reproducibility, sensitivity and specificity<br />
Range of probe-based fast Q<strong>PCR</strong> master<br />
mixes individually optimized for use on all major Q<strong>PCR</strong> platforms<br />
<br />
Thermo Scientific ABsolute Fast Q<strong>PCR</strong><br />
mixes for rapid, reliable probe-based<br />
detection of DNA and cDNA targets.<br />
To request a free sample of ABsolute Fast Q<strong>PCR</strong> Mix visit:
q<strong>PCR</strong> ISSUE 2009<br />
Figure 2: In a real <strong>time</strong> <strong>PCR</strong> assay, a forward and a reverse primer are used to generate a<br />
<strong>PCR</strong> reaction. Two fluorescent dyes, a reporter die in 5’ (R) and a quencher die in 3’ (Q)<br />
are attached to the “Taqman” probe. When both dyes are attached to the probe, the<br />
fluorescent emission of the reporter dye is quenched (top panel). During each extension<br />
cycle, the Taq DNA polymerase cleaves the probe in small fragments in order to proceed<br />
to primer extension. This frees the reporter dye from the template DNA and the<br />
proximity of the rest of the probe (middle panel). Once separated from the quencher, the<br />
reporter emits its characteristic fluorescence throughout the DNA polymerisation<br />
(bottom panel). This fluorescence is measured at each cycle of the <strong>PCR</strong> (Adapted from: ABI web site)<br />
(see later allelic discrimination).<br />
Two main approaches have been<br />
developed to detect fluorescence<br />
directly proportionally to the amount<br />
of the <strong>PCR</strong> product. The most popular,<br />
the TaqMAn assay or 5’-nuclease<br />
assay, is based on the specific<br />
hybridisation and degradation of a<br />
dually-labeled probe to the <strong>PCR</strong><br />
product itself (see Figure 2). It relies<br />
on a 5’-exonuclease activity of the<br />
Taq polymerase, which cleaves the<br />
probe freeing a fluorescent dye that<br />
can be measured at each cycle of the<br />
assay. Several fluorescent dyes are<br />
available, Vic, Tet, Fam etc. Another<br />
approach uses the SYBR-Green I dye,<br />
which emits fluorescence only when<br />
bound to double stranded DNA<br />
(i.e. <strong>PCR</strong> products) (see Figure 3).<br />
Although both assays are<br />
potentially rapid and sensitive, their<br />
principles of detection and<br />
optimisation are different, as is the<br />
resulting price per assay.<br />
Independently of the chemistry<br />
used, read-outs are given as the<br />
number of <strong>PCR</strong> cycles necessary to<br />
achieve a given level of fluorescence<br />
or “cycle threshold” (Ct).<br />
Fluorescence is recorded at each<br />
<strong>PCR</strong> technique. A classic <strong>PCR</strong> reaction<br />
is limited by a “plateau” effect<br />
terminating prematurely the reaction<br />
and usually due to the exhaustion of<br />
reagents in the mix (primers, dNTP).<br />
When the target is abundant this<br />
plateau limits the quantification<br />
capability of the assay. In contrast<br />
when the template has a very low<br />
initial copy number, its abundance at<br />
the end of 40 cycles may still not be<br />
sufficient to be detectable using<br />
classic electrophoresis gels. Classic<br />
<strong>PCR</strong> is therefore efficient in allowing<br />
positive/negative detection but its<br />
range is limited when quantification<br />
is involved.<br />
The use of fluorescence allows the<br />
analysis of the amount of <strong>PCR</strong><br />
products at each cycle of the reaction<br />
(i.e. real <strong>time</strong> <strong>PCR</strong>) as opposed to at<br />
the reaction end however, this can<br />
also be done for particular application<br />
Figure 3: SYBR-green molecules binding double stranded DNA (courtesy of ABI) and <strong>PCR</strong><br />
reaction <strong>PCR</strong> product in the presence of SYBR-Green. At each cycle of the <strong>PCR</strong>, the SYBRgreen<br />
die binds only to the double stranded DNA only and emits fluorescence directly in<br />
relation with the amount of DNA. (Adapted from: <strong>Real</strong>-Time <strong>PCR</strong>. Chapter 8: <strong>Real</strong>-<strong>time</strong><br />
<strong>PCR</strong> using SYBR® Green I.<br />
By Dr Frederique Ponchel. Talyor &Francis. 2006 Oxford Press)<br />
08<br />
www.europeanpharmaceuticalreview.com
SUBSCRIBE<br />
ISSUE<br />
2009 q<strong>PCR</strong><br />
Figure 4: Total fluorescence and transformation. At each cycles of the <strong>PCR</strong>, total<br />
florescence is measured across a large range of wavelength (top left corner, individual<br />
traces for fluorescence measures at four different cycles) at the end of the annealing /<br />
elongation phase (59oC) from cycle 5 till cycle 40 (bottom left corner). Individual traces<br />
indicate the three main components in this example. The pick to the right represents the<br />
buffer dye, ROX, which remain stable with <strong>time</strong>. The pick to the left increase with <strong>time</strong><br />
and represents the amount of fluorescence release specifically at each cycle when<br />
quencher dye (TAMRA) is cleaved away from the reporter dye (FAM in this example).<br />
Total fluorescence traces are then transformed into individual fluorescence components<br />
(top right corner) where specific fluorescence are followed with <strong>time</strong> trough the<br />
progression of the <strong>PCR</strong>. Here, ROX fluorescence is stable (yellow line) indicating a<br />
reaction with no mis-happening. The TAMRA fluorescence (green line) represents the<br />
quencher and is also stable across the reaction. FAM (red line) increases with <strong>time</strong>,<br />
representing the <strong>PCR</strong> product. Finally, the multi-components analysis is transformed into<br />
an amplification plot (bottom right corner) as described in more details in Figure 5.<br />
phase of the <strong>PCR</strong> (linear part of the<br />
Log-fluorescence curve). As a Ct is<br />
proportional to the logarithm of the<br />
initial amount of target in a sample,<br />
the relative concentration of one<br />
target with respect to another is<br />
reflected in the difference in cycle<br />
number (ΔCt = Ct sample – Ct reference )<br />
necessary to achieve the same level<br />
of fluorescence. A glossary of real<br />
<strong>time</strong> <strong>PCR</strong> terminology is available on<br />
the web. The main benefit of real <strong>time</strong><br />
<strong>PCR</strong> which made its popularity<br />
(in addition to the fact that the assay<br />
last about two hours) is that its<br />
sensitivity usually allows detection<br />
over a 6 to 7 log scale (see Figure 6<br />
on page 10). Furthermore, two<br />
reactions can be multiplexed in the<br />
same tube allowing a house-keeping<br />
gene to be measured under the exact<br />
same conditions as the test gene.<br />
In addition to real-<strong>time</strong> <strong>PCR</strong> as<br />
such, which allows quantification of<br />
<strong>PCR</strong> products at each cycle of the<br />
reaction, an “end point assay” ((see<br />
Figure 7 on page 10) most often<br />
referred to as allelic discrimination<br />
cycle across the whole wavelength<br />
range (see Figure 4). The algorithm<br />
within the real <strong>time</strong> <strong>PCR</strong> software<br />
then calculates the specific<br />
contribution to the total fluorescence<br />
of the reporting die, the quencher and<br />
the passive reference (see Figure 4).<br />
This is then transformed in an<br />
amplification plot where fluorescence<br />
is represented at each cycle of the<br />
<strong>PCR</strong> (see Figure 5). During the initial<br />
cycles, the fluorescence signal<br />
emitted by the die (either by the<br />
reporter of a TaqMan probe or by<br />
SYBR-green I) is usually too weak to<br />
register above background. During<br />
the exponential phase of the <strong>PCR</strong>, the<br />
fluorescence doubled at each cycle.<br />
A precise fluorescence doubling at<br />
each cycle is an important indicator<br />
of a well-optimised assay. After<br />
30-35 cycles, the intensity of<br />
fluorescent signal usually began to<br />
plateau, indicating that the <strong>PCR</strong> had<br />
reached a saturation status. Cts are<br />
best recorded during the exponential<br />
Figure 5: <strong>PCR</strong> amplification plot under a log-phase scale. During the initial cycles<br />
(between 0 and ~15 cycles), the fluorescence signal emitted by the die is usually too weak<br />
to register above background. However, this is a very important phase of the assay as it<br />
determines how the algorithm calculating Cts at each cycle should consider as specific<br />
fluorescence above background to allocate values to the reporting die (FAM, VIC, SYBRgreen…)<br />
as opposed to the quencher (TAMRA in TaqMan assay) and the reference<br />
fluorescence (ROX). During the exponential phase of the <strong>PCR</strong>, the fluorescence precisely<br />
doubled at each cycle. After 30-35 cycles, the intensity of fluorescent signal began to<br />
plateau, indicating that the <strong>PCR</strong> had reached a saturation status. Cts are best recorded<br />
during the exponential phase of the <strong>PCR</strong> (linear part of the Log-fluorescence curve).<br />
(Adapted from: <strong>Real</strong>-Time <strong>PCR</strong>. Chapter 8 : <strong>Real</strong>-<strong>time</strong> <strong>PCR</strong> using SYBR® Green I.<br />
By Dr Frederique Ponchel. Talyor &Francis. 2006 Oxford Press)<br />
GO TO CONTENTS PAGE 09
q<strong>PCR</strong> ISSUE 2009<br />
assay) and a melting curve analysis<br />
(see Figure 8 on page 12) have also<br />
been developed, which are more<br />
qualitative. Both assay allow the<br />
detection of a specific sequence<br />
(i.e. genetic polymorphism) rather<br />
Figure 6: These two plots illustrate the high sensitivity of real <strong>time</strong> <strong>PCR</strong>. On the left, plots<br />
show fluorescence traces obtained over 6 log dilution of the same sample and on the<br />
right, the linearity over the detection rage. (Adapted from: <strong>Real</strong>-Time <strong>PCR</strong>. Chapter 8:<br />
<strong>Real</strong>-<strong>time</strong> <strong>PCR</strong> using SYBR® Green I.<br />
By Dr Frederique Ponchel. Talyor &Francis. 2006 Oxford Press)<br />
than their quantification, however,<br />
this has also been done to calculate<br />
the expression of specific alleles.<br />
This particular real-<strong>time</strong> <strong>PCR</strong><br />
application has contributed to the<br />
development of high throughput<br />
allelic discrimination assays.<br />
Since the original release of the<br />
first real <strong>time</strong> <strong>PCR</strong> machines by ABI in<br />
the late 90s, a number of platforms<br />
have been developed by several<br />
companies (Biorad, Roche, Hybaid,<br />
Eppendorf …). Similarly, there are now<br />
many commercially available TaqMan<br />
“off the shelf” assays designed by<br />
companies, several real <strong>time</strong> grade<br />
Taq polymerase, several TaqMann<br />
chemistry kits or SYBR-green kits<br />
(ABI, Quiagen, Sigma, Roche, Biorad,<br />
BioScience corp., MJ Research and<br />
probably many others). Today, there<br />
are no reason to chose one over the<br />
other however, once an assay has<br />
been optimised with one chemistry,<br />
changing kit requires re-optimisation.<br />
Multiple software for designing<br />
primers and probes have been<br />
developed due to the increasing<br />
popularity of the technology.<br />
Accordingly, the risk that real-<strong>time</strong><br />
<strong>PCR</strong> technology may be used<br />
inappropriately or at least without<br />
proper optimisation and<br />
understanding of its limitations, has<br />
raised considerably.<br />
Applications using the real<br />
<strong>time</strong> <strong>PCR</strong> technology<br />
Figure 7: In this assay, the same forward and reverse primers, are used to generate a <strong>PCR</strong><br />
reaction. Two Taqman” probes are use in the same assay however each of them is<br />
specific for one allele of the gene, with either a perfect hybridisation on one allele but<br />
allowing the presence of a mismatch on the second allele. When the polymerase<br />
approaches, the probes are lifted from the template. If the strength of the hybridisation is<br />
sufficient to maintain the probe on the template the 5’ exonuclease activity cleaves the<br />
probe and releases fluorescence. In presence of the mismatch, the probe is getting away<br />
before the exonuclease activity can cleave it therefore preventing the release of the<br />
fluorescent dye. (Adapted from: ABI web site).The ratio of fluorescence for each allele<br />
(bottom graph,) is then calculated at the end of the assay and either biased to one or the<br />
other allele (homozygotes: allele 1 red, allele 2 blue, negative control black) or mixed<br />
(heterozygote, green).<br />
Quantification of genomic DNA<br />
Medical diagnostic or prognostic is<br />
one of the areas where real <strong>time</strong><br />
<strong>PCR</strong> has really made a difference.<br />
There are several situations where<br />
quantifying DNA is important.<br />
The short response <strong>time</strong> to obtain<br />
an outcome can often be of great<br />
importance for patients, particularly<br />
when considering that the alternative<br />
methods may some<strong>time</strong>s require<br />
live cells or long delays. Gene<br />
rearrangements or translocations<br />
have been studied using real-<strong>time</strong><br />
<strong>PCR</strong>. For example, the major and<br />
minor break-points of the<br />
10<br />
www.europeanpharmaceuticalreview.com
SUBSCRIBE<br />
ISSUE<br />
2009 q<strong>PCR</strong><br />
chromosome 14 to 18 translocation<br />
t(14:18)(q32;q21) have been analysed<br />
in non-Hodgkins lymphomas using a<br />
TaqMan assay 1 . Gene amplification<br />
used to be detected by lengthy<br />
techniques such as Southern blotting<br />
or FISH. The use of the SYBR-Green I<br />
assay for the detection of gene<br />
amplification has had an important<br />
impact for patients as results are now<br />
produced rapidly (in less than one<br />
day). The diagnosis and treatment of<br />
a number of severe conditions has<br />
therefore benefited from the<br />
development of such assays based<br />
(i.e. for example, the c-Myc<br />
amplification as a diagnostic marker<br />
in neuroblastome (see Figure 9 on<br />
page 12) the multi-drug resistance<br />
phenotype in tumours resulting from<br />
the amplification of the MDR gene 2 ,<br />
cyclin D1 amplification, a prognostic<br />
marker in breast carcinoma 3 and<br />
MDM2 gene amplification, essential<br />
in the pathogenesis of a variety of<br />
tumours 4 . The detection of deletion<br />
(particularly when such genetic<br />
Table 1: Literature search using “real-<strong>time</strong> <strong>PCR</strong>” as keyword only<br />
Year Papers [ISS] Paper and abs tracts [PubMed]<br />
1995 none 5<br />
1996 6 35<br />
1997 25 56<br />
1998 38 75<br />
1999 110 101<br />
2000 250 216<br />
2001 451 653<br />
2002 756 1,153<br />
2003 1,257 2,101<br />
2004 1,504 3,373<br />
2005 4,632 5,264<br />
2006 5,628 6,811<br />
2007 6,411 12,505<br />
2008 6,364 15,426<br />
abnormalities could only be detected<br />
unequivocally by FISH) is also<br />
possible using real <strong>time</strong> <strong>PCR</strong>. We<br />
were able to detect the deletion of<br />
one allelic copy of a particular exon of<br />
the OPA1 gene (see Figure 10 on page<br />
13). Similar SYBR green assay could<br />
be applied to the detection of tumour<br />
suppressor gene deletion in cancer<br />
biopsies or when inherited<br />
susceptibility to cancer is suspected<br />
(i.e. for genes such as p53, Rb, WT1,<br />
APC, VHL or BRCA1), therefore<br />
benefiting genetic counselling with<br />
a quick and reliable assay to detect<br />
carriers of such deletions. Finally, we<br />
have recently adapted the technique<br />
to the detection of gene duplication in<br />
a case of mental retardation (data<br />
recently submitted for publication).<br />
®<br />
HotStart-IT.<br />
Less Heat. No DNA Damage.<br />
We couldn’t be more specific.<br />
USB’s novel hot start <strong>PCR</strong> method<br />
requires no extensive heat denaturation<br />
step. The result is no damage to<br />
precious samples with increased<br />
specificity and yield.<br />
HotStart-IT ® Taq DNA Polymerase<br />
• Based on primer sequestration–a novel method<br />
developed by USB scientists<br />
• Unique protein binds primers to prevent mispriming<br />
• Primers are released during heat denaturation<br />
Benefits<br />
High Sensitivity<br />
• Detects
q<strong>PCR</strong> ISSUE 2009<br />
Figure 8: Two sequence specific primers (top left corner), one for each allele of the polymorphic sequence to analyse, are used to<br />
generate <strong>PCR</strong> product, one being 10 (non-specific) base pair longer than the other. At then end of a <strong>PCR</strong> reaction, the two <strong>PCR</strong> product<br />
are therefore of different length as illustrated on a classic electrophoresis (top right corner). A melting curve assay relies on the slower<br />
release of SYBR-green from the longer <strong>PCR</strong> product. At the end of the reaction, <strong>PCR</strong> products are slowly heated. When double DNA<br />
start melting, SYBR-green is released and the fluorescence decrease accordingly (bottom left panel, fluorescence intensity). Speed of<br />
release is then calculated through a mathematical conversion (dF/dT), providing a melting curve where the pick is indicative of the<br />
melting temperature of the <strong>PCR</strong> product. Each sequence has a different melting point however; allele discrimination is brought by the<br />
additional 10 bp which provide an identifiable shift of melting point (bottom right corner).<br />
(Adapted from: <strong>Real</strong>-Time <strong>PCR</strong>. Chapter 8: <strong>Real</strong>-<strong>time</strong> <strong>PCR</strong><br />
using SYBR® Green I by Dr Frederique Ponchel. Talyor &Francis. 2006 Oxford Press).<br />
In medical and biological research,<br />
the technique has also provided a lot<br />
of advances. Similar applications<br />
were used for similar and additional<br />
purposes such as assessing gene<br />
copy number in transgenic animals<br />
for examples 5 . One particular area<br />
that has considerably benefited from<br />
accurate quantification of DNA is<br />
chromatin immuno-precipitation (ChIP).<br />
ChIP is a powerful tool for the study<br />
of protein/DNA interactions. In this<br />
technique, specific protein/DNA<br />
complexes are immunoprecipitated<br />
using an antibody directed against the<br />
protein of interest. The proteins are<br />
then removed and the DNA is purified<br />
and analysed to determine which<br />
DNA fragments, but most importantly<br />
how much, was bound to the protein<br />
(see Figure 11 on page 14).<br />
Another area that has<br />
considerably benefited from real <strong>time</strong><br />
<strong>PCR</strong> technology is the detection or<br />
identification of pathogens or<br />
contaminats. Toxins, virus (such as<br />
Hepatitis B presence or viral load 6 )<br />
or bacteria (Chlamydia trachomatis<br />
presence and/or identification),<br />
genetically modified organisms<br />
can be detected/quantified either<br />
with real <strong>time</strong> <strong>PCR</strong> or end point<br />
Figure 9: Detection of gene amplification in DNA extracted from neuroblastoma tumor<br />
samples. We applied the SYBR-Green I assay to tumour samples for the detection of the<br />
MYC-N gene amplification. These samples had been tested previously by standard<br />
procedures for the diagnosis of neuroblastoma using fluorescence in-situ hybridisation<br />
(FISH)9. <strong>Real</strong> <strong>time</strong> <strong>PCR</strong> (black bars) compared well with the established amplification<br />
factor recorded for these tumours by FISH (open bars) and confirmed the presence of<br />
MYC-N amplification in six tumours. Adapted from BMC Biotechnology, 2003 5<br />
12<br />
www.europeanpharmaceuticalreview.com
SUBSCRIBE<br />
ISSUE<br />
2009 q<strong>PCR</strong><br />
assays. Tests were developed to<br />
detect mycotoxins, T-toxin,<br />
phytoestrogens such as Coumestrol<br />
and several other contaminants,<br />
particularly in food.<br />
Quantification of RNA<br />
This obviously is the area that has<br />
most immediately beneficiated from<br />
real <strong>time</strong> <strong>PCR</strong>. It is well accepted that<br />
measures of gene expression is now<br />
quantitative rather than semiquantitative.<br />
Two methods exist, one<br />
absolute comparing gene expression<br />
to known standards and a second,<br />
relative to an endogenous reference<br />
(usually a house keeping gene).<br />
Outside its plain use for gene<br />
expression studies (in cells in<br />
response to stimuli for example), real<br />
<strong>time</strong> <strong>PCR</strong> has been instrumental<br />
notably in transcriptome studies for<br />
confirming micro-array work, when<br />
limited amount of material is available<br />
(small tissue biopsy studies), or for<br />
large studies necessitation high<br />
thorough-put and automatism with<br />
<strong>PCR</strong> robots. This has notably<br />
warranted the development and<br />
success of the micro-fluidic card<br />
systems. It was recently argued 7 that<br />
advances in the tumour biology have<br />
now provided molecular targets for<br />
diagnosing and treating cancer.<br />
Clustering algorithms have revealed a<br />
molecular diversity that forms a new<br />
taxonomy with diagnostic, prognostic<br />
and possibly therapeutic significance.<br />
The challenge in pathology today is<br />
to develop and implement such<br />
molecular classifications in routine<br />
clinical practice. This requires<br />
objective, robust, and cost-effective<br />
techniques and real-<strong>time</strong> <strong>PCR</strong> offers<br />
such attractive features.<br />
Discrimination assay<br />
Genotyping using end point assays or<br />
melting curves analysis assays<br />
provides rapid results (raw biological<br />
samples to SNP genotyping results in<br />
less than one hour), highly accurate in<br />
virtually any sample (low amount of<br />
material or poor quality samples) and<br />
an increasing number of publications<br />
Figure 10: Pedigree of an Autosomal Dominant blindness family with a deletion of the<br />
entire OPA1 gene. Carriers of the deletion are indicated with filled symbols and normal<br />
individuals with open symbols 10 . <strong>Real</strong>-<strong>time</strong> was used in a copy number assay for the OPA1<br />
gene using the SYBR-Green I dye. Quantification of the OPA1 gene deletion in 13<br />
members of the family: Non-carriers showed a ratio close to 1 for the GAPDH versus<br />
OPA1 quantification. In contrast, all carriers of the deletion showed a clear reduction in<br />
this ratio. Adapted from BMC Biotechnology, 2003 5<br />
(see for example any issues of Clinical<br />
Chemistry, Molecular Diagnostic,<br />
Nucleic Acid Research…) testifies<br />
how instrumental real <strong>time</strong> <strong>PCR</strong> has<br />
been in this field. The success of the<br />
technology has also promoted the<br />
development of new molecular tools<br />
such as minor grove binding,<br />
molecular beacon or scorpion probes.<br />
Genotyping assay for SNP are now<br />
available in array format, (TaqMan®<br />
SNP Genotyping Assays by ABI,<br />
Illumina Chip by Affimetrix and<br />
several others) making it<br />
straightforward to perform genome<br />
wide SNP genotyping studies.<br />
Importantly, melting curve analysis<br />
assays, scanning a particular<br />
sequence where polymorphisms are<br />
suspected, were shown to accurately<br />
reveal mutations/SNPs 8 .<br />
References<br />
1. Luthra, R. and L.J. Medeiros, 5 '-> 3 '<br />
exonuclease-based real-<strong>time</strong> <strong>PCR</strong><br />
methods for detecting the t(14;18)<br />
and t(11;14) in non-Hodgkin's<br />
lymphomas. J. Clin. Ligand Assay,<br />
2000; 23: 6-14.<br />
2. Ramachandra, C. and S. Melnick,<br />
Multidrug resistance in human<br />
tumors: molecular diagnosis and<br />
clinical significance. Mol. Diag,<br />
1999; 4: 81-94.<br />
3. Vos, C.B.J., N.T. Ter Haar, J.L.<br />
Peterse, G.J. Cornelisse and M.J.<br />
Van de Vijver, Cyclin D-1 gene<br />
amplification and overexpression<br />
are present in ductal carcinoma in<br />
situ of the breast. J. Pathol., 1999;<br />
187: 279-284.<br />
4. Gunther, T., R. Schneider-Stock,<br />
C. Hackel, H.U. Kasper, M. Pross,<br />
GO TO CONTENTS PAGE 13
q<strong>PCR</strong> ISSUE 2009<br />
Figure 11: Chromatin immuno-precipitation (ChIP). In this technique, intact cells are fixed using formaldehyde, which cross-links and<br />
preserves protein/DNA interactions. The DNA is then sheared into small, uniform fragments using sonication and specific<br />
protein/DNA complexes are immunoprecipitated using an antibody directed against the DNA-binding protein of interest.<br />
Following immunoprecipitation, cross-linking is reversed, the proteins are removed by treatment with Proteinase K and the DNA is<br />
purified. The DNA is then analysed to determine which DNA fragments were bound by the protein of interest.<br />
(Adapted from Chromatin Immunoprecipitation Kits, Active Motif North America)<br />
A. Hackelsberger, H. Lippert, and A.<br />
Roessner, Mdm2 gene amplification<br />
in gastric cancer correlation with<br />
xpression of mdm2 protein and<br />
p53 alterations. Mod. Pathol., 2000;<br />
13: 621-626.<br />
5. Ponchel, F., C. Toomes, K. Bransfield,<br />
F. Leong, S. Field, S. Douglas, S. Bell,<br />
V. Combaret, A. Puisieux, A.<br />
Mighell, P. Robinson, C. Inglehearn, J.<br />
Isaacs, and A. Markham, <strong>Real</strong>-<strong>time</strong><br />
<strong>PCR</strong> based on SYBR-green<br />
fluorescence: An alternative to the<br />
TaqMan assay for a relative<br />
quantification of gene<br />
rearrangements, gene amplifications<br />
and micro gene deletions. BMC<br />
Biotechnology, 2003; 3: 18.<br />
6. Yeh, S., Tsai, J. Kao, C. Liu, T. Kuo, M.<br />
Lin, W. Huang, S. Lu, J. Jih, and D.<br />
Chen, Quantification and genotyping<br />
of hepatitis B virus in a single<br />
reaction by real-<strong>time</strong> <strong>PCR</strong> and<br />
melting curve analysis. J Hepatol<br />
2004; 41: 659-666.<br />
7. Bernard, P.S. and C.T. Wittwer, <strong>Real</strong>-<br />
Time <strong>PCR</strong> Technology for Cancer<br />
Diagnostics. Clin Chem 2002. 48:<br />
1178-1185.<br />
8. Stromqvist Meuzelaar, L.,<br />
K. Hopkins, E. Liebana and A.J.<br />
Brookes, DNA Diagnostics by<br />
Surface-Bound Melt-Curve<br />
Reactions. Mol Diag 2007. 9. 30-41.<br />
9. De Preter, K., F. Speleman, V.<br />
Combaret, J. Lunec, G. Laureys,<br />
B. Eusson, N. Francotte,<br />
A. Pearson, A. De Paepe, N. van<br />
Roy, and J. vandersompele,<br />
Quantification of MYC-N, DDX1 and<br />
NAG gene copy number in<br />
neuroblastoma using real <strong>time</strong><br />
quatitative <strong>PCR</strong> assay. Mod. Pathol.<br />
2002; 15: 150-166.<br />
10. Marchbank, N., J. Craig, P. Leek,<br />
M. Toohey, A. Churchill, A.<br />
Markham, D. mackey, C. Toomes,<br />
and C. Inglehearn, Deletion inthe<br />
OPA1 gene in a dominant optic<br />
atrophy family: evidence that<br />
haploinsufficiency is the cause of<br />
disease. J Med Genet 2002; 39: 0-3.<br />
Dr. Frédérique Ponchel<br />
Dr. Frédérique Ponchel currently works as<br />
a RCUK Senior Academic Research Fellow<br />
at the Leeds Institute of Molecular<br />
Medicine at Leeds University. Prior to this<br />
Frédérique has worked at The University of<br />
York as a postdoctoral fellow. Originally<br />
from France, Frédérique studied<br />
Biochemistry and Genetics in Paris.<br />
@<br />
email the author<br />
14<br />
www.europeanpharmaceuticalreview.com
Single Cell Gene Expression from Fluidigm<br />
96 GENES FROM A SINGLE CELL<br />
96 SINGLE CELLS SIMULTANEOUSLY<br />
9,216 Q<strong>PCR</strong> CURVES PER RUN<br />
Would you like to analyze 96 genes in parallel on a single cell, with<br />
the sensitivity of real-<strong>time</strong> <strong>PCR</strong> but without the pipetting complexity<br />
and high reagent costs<br />
Fluidigm’s 96.96 Dynamic Array — an integrated circuit for biology<br />
— might be the answer. This system can easily determine gene<br />
expression levels in samples ranging from circulating tumor cells<br />
to stem cells. To learn more about the advantages of single cell<br />
gene expression on Fluidigm’s 96.96 Dynamic Arrays, please visit<br />
www.Fluidigm.com/single_cell.<br />
BIOMARK 96.96 DYNAMIC ARRAYS FOR GENOTYPING & GENE EXPRESSION<br />
ACCELERATE<br />
<br />
<br />
® Primers & Probes<br />
ENABLE<br />
<br />
<br />
<br />
<br />
Sales: Europe/EMEA: +31 20 578 8853 | biomarkeurope@fluidigm.com<br />
North America: 1 650.266.6170 | biomark@fluidigm.com<br />
Japan/Korea: +81 3 3555 2351 | biomarkasia@fluidigm.com<br />
Asia: +65 9431 3790 | biomarkasia@fluidigm.com<br />
www.fluidigm.com<br />
© Fluidigm Corporation. All rights reserved. Fluidigm, the Fluidigm logo, and<br />
BioMark are trademarks or registered trademarks of Fluidigm Corporation in the U.S.<br />
and other countries. All other trademarks are the property of their respective owners.<br />
FOR RESEARCH USE ONLY.
HIGH CONTENT SCREENING ISSUE 2009<br />
Open source data<br />
management in High<br />
Content Screening<br />
technology<br />
Karol Kozak, Head of Data Handling Facility, Dr Gabor Csucs, Head of the High-thorughput/high-content screening facility, LMC-RISC, ETH<br />
Zurich, Switzerland and Andrzej Firkowski, Professor of Chemical Technology at the Technical University of Radom, Director of the Faculty of<br />
Materials Science and Technology and a member of the Advisory Board, Noster-IT, Dresden, Germany<br />
Screening for biological or chemical compounds with a favourable biological effect is at the core of<br />
the modern drug discovery process. High Content Screening (HCS) is increasingly used for the<br />
automated evaluation of spatio-temporally resolved multiple biochemical and morphological<br />
parameters in cellular systems. HCS data include comprehensive information about the bioactive<br />
molecules, the targeted genes, and images as well as their extracted data matrices after acquisition.<br />
This paper will describe an open-source data management solution for integrating, sharing, analysis<br />
and processing HCS data using a distributed service-oriented architecture.<br />
Fluorescent microscopy has<br />
enabled multifaceted insights<br />
into the detail and complexity of<br />
cellular structures and their functions<br />
for well over two decades. As an<br />
essential prerequisite for a systematic<br />
phenotypical analysis of gene<br />
functions in cells at a genome-wide<br />
scale, the throughput of microscopy<br />
had to be improved through<br />
automation. HCS is defined as<br />
multiplexed functional screening<br />
based on imaging multiple targets in<br />
the physiologic context of intact cells<br />
by extraction of multicolour<br />
fluorescence information.<br />
Simultaneous staining in three or four<br />
colours allows the extraction of<br />
various parameters from each cell<br />
quantitatively as well as qualitatively<br />
such as intensity, size, distance or<br />
distribution (spatial resolution).<br />
The parameters might be referenced<br />
to each other, for example the use of<br />
nuclei staining to normalise other<br />
signals against cell number, or<br />
particular parameters might verify or<br />
exclude each other.<br />
An essential factor in the success<br />
of high content screening projects is<br />
the existence of algorithms and<br />
software that can reliably and<br />
automatically extract information<br />
from the masses of captured images.<br />
Typically, nuclei are identified and<br />
masked first. Then, areas around the<br />
nuclei are determined or the cell<br />
boundaries are searched to mask the<br />
cell shape. For popular HCS assays at<br />
the sub-cellular level such as cell<br />
cycle analysis (mitotic index),<br />
cytotoxicity, apoptosis, micronuclei<br />
detection, receptor internalisation,<br />
protein translocation (membrane to<br />
cytosol, cytosol to nucleus, and vice<br />
versa), co-localisation, cytoskeletal<br />
arrangements, or morphological<br />
analysis at the cellular level such as<br />
neurite outgrowth, cell spreading,<br />
cell motility, colony formation, or tube<br />
formation, ready-to-use scripts are<br />
available and need only some fineadjustment<br />
for the particular cell<br />
line and/or conditions of the assay.<br />
Presently available analysis<br />
methodologies for large-scale RNAi<br />
(siRNA) data sets typically rely on<br />
ranking data and are based on single<br />
image descriptor (feature) or<br />
significance value 1,10,12 . However,<br />
identifying patterns of image<br />
descriptors and grouping genes into<br />
classes based on multiparamteric<br />
16<br />
www.europeanpharmaceuticalreview.com
© 2008 Thermo Fisher Scientific Inc. All rights reserved. All trademarks are the property of Thermo Fisher Scientific<br />
Inc. and its subsidiaries.<br />
What does Productivity mean to you<br />
For many, choosing better leads, eliminating problem<br />
compounds early, understanding disease targets better or<br />
predicting human toxicity is the route to discovery productivity.<br />
Did you know, the Thermo Scientific Cellomics high content<br />
screening (HCS) platform is addressing these issues in<br />
laboratories around the world, every day Did you also know<br />
that based on an analysis of the literature, users of our platform<br />
generate 5 <strong>time</strong>s the number of peer reviewed publications<br />
compared to all the other HCS platforms put together.<br />
Learn more about how we can redefine your HCS productivity<br />
and take advantage of our special offer to receive either one<br />
HCS Reagent Kit or one Redistribution CryoRedi 2-pack, at half<br />
price, visit www.thermo.com/hcs<br />
Our complete HCS solution.<br />
Flexible, modular instrumentation<br />
Powerful, easy to use image analysis software<br />
Open standards data management<br />
HCS Reagent Kits and Redistribution cell lines<br />
Moving science forward
HIGH CONTENT SCREENING ISSUE 2009<br />
Figure 1: High Content Screening data flow<br />
analysis might provide much greater<br />
insight into their biological function<br />
and relevance (see Figure 1).<br />
Many free and commercial<br />
software packages are now available<br />
to analyse HCS data sets using<br />
classifiers, although it is still difficult<br />
to find a single off-the-shelf<br />
software package that answers.<br />
Statistical open source software<br />
packages such as BioConductor<br />
(www.bioconductor.org) provide large<br />
collections of methods suitable for<br />
HCS data analysis. However, their<br />
command-line usage can be too<br />
demanding for users without<br />
adequate computer knowledge. As an<br />
alternative, software packages where<br />
users can upload their data and<br />
receive their processed results are<br />
becoming increasingly common:<br />
Weka 16 , CellAnalyzer 2 , CellHTS 3 ,<br />
TreeView 13 have all been published<br />
within the last year. Unfortunately,<br />
these services often allow only limited<br />
freedom in the choice and<br />
arrangement of processing steps.<br />
Other, more flexible tools, such as<br />
Eclipse 6 , KNIME 11 , JOpera 5 , operate<br />
either stand-alone or require<br />
considerable computer knowledge<br />
and extra software to run through the<br />
web. In order to make use of the vast<br />
variety of data analysis methods<br />
around, it is essential that such an<br />
environment is easy and intuitive to<br />
Figure 2: General concept of a pipeline node. The component properties are described by<br />
the input metadata, output metadata and user defined parameters or transformation<br />
rules. The input and output ports can have one or more incoming or outgoing metadata<br />
or images.<br />
18<br />
www.europeanpharmaceuticalreview.com
SUBSCRIBE<br />
ISSUE<br />
2009 HIGH CONTENT SCREENING<br />
Figure 3: A schematic experimental flow of HCS data in a HC/DC workflow and corresponding nodes.<br />
use, allows for quick and interactive<br />
changes to the analysis process and<br />
enables the user to visually explore<br />
the results. To meet these challenges<br />
data pipelining environments have<br />
gathered incredible momentum over<br />
the past years. These environments<br />
allow the user to visually assemble<br />
and adapt the analysis flow from<br />
standardised building blocks, which<br />
are then connected through pipes<br />
carrying data or models. An<br />
additional advantage of these<br />
systems is the intuitive, graphical way<br />
to document what has been done.<br />
Open source data<br />
pipeline system<br />
Pipeline (workflow) systems could<br />
become crucial for enabling biologists<br />
doing large scale experiments to<br />
deal with this data explosion. The<br />
workflow is termed abstract in that it<br />
is not yet fully functional but the<br />
actual components are in place and in<br />
the requisite order. In general,<br />
workflow systems concentrate on<br />
creation of abstract process workflows<br />
to which data can be applied when<br />
the design process is complete.<br />
In contrast, workflow systems in the<br />
life sciences domain are often based<br />
on a dataflow 8 model, due to the<br />
data-centric and data-driven nature<br />
of many scientific analyses.<br />
A comprehensive understanding of<br />
biological phenomena can be achieved<br />
only through the integration of all<br />
available biological information and<br />
different data analysis tools and<br />
applications. Workflow environment<br />
allows screening researchers to<br />
perform the integration themselves<br />
without involving any programming.<br />
Workflow system allows the<br />
construction of complex in silico<br />
experiments in the form of workflows<br />
and data pipelines. Data pipelining is<br />
a relatively simple concept. Any<br />
computational component or node<br />
has data inputs and data outputs.<br />
Data pipelining views these nodes as<br />
being connected together by ‘pipes’<br />
through which data flows (see<br />
Figure 2 on page 18).<br />
In a workflow controlled data<br />
pipeline, as the data flows, it is<br />
transformed and raw data is analysed<br />
to become information and the<br />
collected information gives rise to<br />
knowledge. The concept of workflow<br />
is not new and it has been used by<br />
many organisations, over the years,<br />
to improve productivity and increase<br />
efficiency. A workflow system is<br />
highly flexible and can accommodate<br />
any changes or updates whenever<br />
new or modified data and<br />
corresponding analytical tools<br />
become available. Currently, there<br />
are few workflow systems available<br />
in life sciences. Definitely InforSense<br />
KDE 4 and Pipeline Pilot 7 are state of<br />
the art workflow systems helping<br />
faster and efficient research in life<br />
sciences domain. However due<br />
to the high costs involved, these<br />
are still not accessible to many<br />
research institutions – a major<br />
contributor to scientific research.<br />
In this paper we describe open-source<br />
workflow package called HC/DC<br />
(High Content Data Chain) and some<br />
of the design aspects of the<br />
underlying architecture and briefly<br />
sketch how nodes for HCS data can<br />
be incorporated.<br />
GO TO CONTENTS PAGE 19
HIGH CONTENT SCREENING ISSUE 2009<br />
HC/DC<br />
The architecture of HC/DC was<br />
designed based mostly on eclipse<br />
plugin framework and Eclipse-KNIME<br />
data workflow system. HC/DC is a<br />
functional node set, working together<br />
with package KNIME and ImageJ<br />
presented. A plug-in for opening and<br />
processing proprietary HCS files<br />
(library, numeric results and images)<br />
was developed within the KNIME<br />
environment. All those open source<br />
components (Eclipse environment,<br />
KNIME, R-Project, Weka and ImageJ)<br />
were chosen for its platformindependence,<br />
openness, simplicity,<br />
■<br />
single image. Software supports<br />
all image types which are<br />
supported by ImageJ and ImageJ<br />
plug-ins<br />
Computation: Dataflow pipelines<br />
dictate that each processor be<br />
executed as soon as its data<br />
inputs are available, and<br />
processors that have no data<br />
dependencies amongst each other<br />
can be executed concurrently.<br />
They are used for integrating data<br />
from different sources, data<br />
capture, preparation and analysis<br />
pipelines, and populating scientific<br />
models or data warehouses.<br />
■<br />
■<br />
automatically reacting to changed<br />
environmental circumstances<br />
Modularity: Processing units and<br />
containers should not depend on<br />
each other in order to enable easy<br />
distribution of computation and<br />
allow for independent<br />
development of different image<br />
processing algorithms<br />
Easy expandability: In HC/DC, like<br />
in Knime, it is easy to add a new<br />
microscope, data analysis, image<br />
processing software nodes or<br />
views and distribute them through<br />
a simple plug-in mechanism<br />
without the need for complicated<br />
install/reinstall procedures. In<br />
order to achieve this, a data<br />
processing consists of a pipeline<br />
of nodes, connected by edges that<br />
transport data.<br />
Figure 3 on page 19 shows in schematic<br />
way an example of HCS data analysis<br />
flow and corresponding nodes.<br />
Figure 4: Workflow Projects manager and export/import possibilities from submenu.<br />
and portability. They are also the<br />
fastest pure Java image, dataprocessing<br />
programs currently<br />
available. The programs have built-in<br />
command recorder, editor, and Java<br />
compiler; therefore, it is easily<br />
extensible through custom plug-ins.<br />
The pipeline model of HC/DC<br />
describes the exact behaviour of the<br />
workflow when it is executed. The<br />
nodes of HC/DC are designed with<br />
main principles:<br />
■ Resource type: The source of data<br />
can be collection high level<br />
images familiar to the user or<br />
■<br />
■<br />
Control flows directly dictate the<br />
flow of process execution, using<br />
loops, decision points etc<br />
Interactivity: Nodes execution<br />
could be wholly automatic or<br />
interactively steered by the user.<br />
Data flows are combined by simple<br />
drag&drop from a variety of<br />
processing units. Customised<br />
applications can be modeled<br />
through individual data subpipelines<br />
Adaptivity: The nodes and<br />
workflow design or instantiation<br />
can be dynamically adapted “in<br />
flight” by the user or by<br />
Workflow Projects<br />
Workflow Projects manager shows all<br />
currently defined workflow projects in<br />
HC/DC. The context menu of the<br />
project manager allows you to create<br />
new projects, import existing<br />
workflows into the workbench and<br />
export own workflows (see Figure 4).<br />
Each project is saved in the<br />
workspace. The workspace is located<br />
in your HC/DC installation directory<br />
by default (see below how to change<br />
this). On the file system a workflow<br />
project consist of several files and<br />
folders to store the current status of a<br />
flow; this structure is not visible in the<br />
Workflow Projects Manager as it<br />
represents internal information not<br />
intended to be changed manually.<br />
HCS nodes<br />
Workflows in HC/DC (based on<br />
KNIME) are essentially graphs<br />
connecting nodes, or more formally, a<br />
direct acyclic graph (DAG). The<br />
Workflow Manager allows to insert<br />
new nodes and to add directed edges<br />
(connections) between two nodes. It<br />
also keeps track of the status of<br />
20<br />
www.europeanpharmaceuticalreview.com
SUBSCRIBE<br />
ISSUE<br />
2009 HIGH CONTENT SCREENING<br />
nodes (configured, executed, ...) and<br />
returns, on demand, a pool of<br />
executable nodes. This way the<br />
surrounding framework can freely<br />
distribute the workload among a<br />
couple of parallel threads or – in the<br />
future – even a distributed cluster of<br />
servers. Each Node can have an<br />
arbitrary number of views associated<br />
with it.<br />
Plate Viewer<br />
Plate Viewer (PV) guarantees the<br />
identification of library and well<br />
position of a specific compound on a<br />
plate. The history of location of each<br />
compound in the screen, run and<br />
replicate along with reformatting<br />
information are recorded and<br />
reconstructed by PV. Within the<br />
GUI the user may select the library,<br />
plate and if desired, compounds<br />
data derived from specific 96,<br />
384 or 1536-well plate. Once a<br />
plate is selected, a window is opened<br />
in a plate viewer that provides<br />
functions easy navigation within the<br />
plate that helps extracting of<br />
comprehensive information from<br />
wells about particular compounds<br />
(see Figure 5).<br />
HiLite controls on the plate<br />
Through receiving events from a<br />
HiLiteHandler (and sending events to<br />
it) it is possible to mark selected<br />
points in such a view to enable visual<br />
brushing. Views can range from<br />
simple table views to more complex<br />
views on the underlying data (e.g.<br />
scatterplots, parallel coordinates) or<br />
the generated model (e.g. decision<br />
trees, rules).<br />
Join library, image results, image<br />
processing results<br />
This node joins two or three data<br />
sources in one matrix. The join on the<br />
two sources is carried out so that the<br />
first source table (from the first, top<br />
input port) provides the left part of<br />
the output table and the second table<br />
(bottom input port) provides the<br />
columns for the right part. Thus, the<br />
output table has as many rows as<br />
Figure 5: Plate viewer plug-in. Visualisation of image processing parameters in heatmap<br />
with access to library metadata.<br />
both of the input tables (given<br />
that both tables contain exactly the<br />
same row identifier (IDs)) and as<br />
many columns as the sum of both<br />
column counts. If a row ID only occurs<br />
in one of the two tables, the remaining<br />
part (the column that should have<br />
been provided by the other table) is<br />
filled with missing records.<br />
Cluster support<br />
Due to the modular architecture it is<br />
easy to designate specific nodes to be<br />
run on separate machines. But to<br />
accommodate the increasing<br />
availability of multi-core machines,<br />
the support for shared memory<br />
parallelism also becomes increasingly<br />
important. KNIME offers a unified<br />
framework to parallelise data parallel<br />
operations. Sieb et al. (2007) 14<br />
describe further extensions, which<br />
enable the distribution of complex<br />
tasks such as cross validation on a<br />
cluster or a GRID.<br />
Image processing nodes<br />
Image processing can process sets of<br />
images parallel. Of particular interest<br />
to HCS assay development is the<br />
ability to load sequences of images to<br />
create what ImageJ15 calls a “Stack”.<br />
We describe standard nodes of<br />
HC/DC for image processing:<br />
■ Data I/O: generic image file reader<br />
and image writer which overwrite<br />
■<br />
■<br />
■<br />
■<br />
images or create new output<br />
collection of images<br />
Image Selections: The ImageK<br />
provides tools for selecting regular<br />
and irregular areas (called Regions<br />
Of Interest or ROIs) on an image.<br />
Several types of selections such as<br />
rectangles, circles, poly-line, and a<br />
“magic wand” are available.<br />
Selections can be measured,<br />
filtered, filled or drawn<br />
Colour and level adjustment: Basic<br />
brightness/contrast and<br />
minimum/maximum level<br />
adjustments are available as<br />
nodes. This includes the allowance<br />
for colour adjustments by<br />
separately manipulating R, G, and<br />
B (as well as C, M, and Y)<br />
Other image adjustments: A range<br />
of standard image adjustments<br />
such as rotation, resizing (with or<br />
without interpolation), cropping,<br />
duplicating, zooming, and<br />
renaming are fully supported<br />
Image Filtering: Several standard<br />
image filters are included in ImageK.<br />
Some of these include “Gaussian<br />
Blur”, “Median”, “Mean”, and<br />
“Unsharp Mask”, among others.<br />
There is also the built in capability<br />
of specifying a user-defined<br />
convolution mask. Dozens of<br />
user-contributed plugins provide<br />
access to many other filters<br />
including Wavelet filters. Fourier<br />
GO TO CONTENTS PAGE 21
HIGH CONTENT SCREENING ISSUE 2009<br />
■<br />
■<br />
■<br />
■<br />
■<br />
transforms and frequency filtering<br />
are also part of the standard<br />
ImageJ package. A wide variety of<br />
mathematical functions may be<br />
easily performed on image pixels.<br />
For example, one may add,<br />
subtract, multiply, or divide by a<br />
constant, perform logarithms,<br />
square root, reciprocal, or find the<br />
absolute value of a selection of<br />
pixels in an image<br />
Image Calculator: The “Image<br />
Calculator” allows images to be<br />
combined using one of several<br />
mathematical functions, including<br />
add, subtract, multiply, and divide<br />
as well as AND, OR, XOR, min,<br />
max, and average. This is useful<br />
for performing calibrations such as<br />
dark subtraction and flat fielding<br />
on astronomical images. The<br />
image calculator accepts stacks as<br />
input, thus a stack of images can<br />
be batch reduced<br />
Measuring Images: A wide<br />
variety of measurements including<br />
Centre of Mass, Min/Max Gray<br />
Value, Mean Gray Value,<br />
Standard Deviation, and Area can<br />
be performed on an image or a<br />
stack of images. Measurements<br />
appear as a table in a popup<br />
window executed from node<br />
submenu. The measurements<br />
can be copied and pasted<br />
directly into a spreadsheet where<br />
they can be sorted or analysed<br />
by KNIME nodes in the same<br />
environment<br />
Image Display Nodes: Image scale,<br />
for example arcminutes per pixel,<br />
can easily be set using<br />
configuration popup window<br />
Results Visualisation: scatter<br />
plot, histogram, parallel<br />
coordinates, multidimensional<br />
scaling, rule plotters<br />
Misc: scripting nodes.<br />
Microscope image readers<br />
Microscope readers offers automated<br />
loading of image data from many HCS<br />
microscopes or cellular analysis<br />
system: Opera, MD Micro, BD<br />
Pathway, CellWorx, ArrayScan,<br />
ScanR. Those nodes can be also very<br />
easily customised.<br />
Library reader<br />
This node can be used to read RNAi<br />
data from an ASCII file or URL<br />
location. It can be configured to read<br />
in various formats. When you open<br />
the node's configuration dialog and<br />
provide a filename, it will try to guess<br />
the reader's settings by analysing the<br />
beginning of the file. Check the<br />
results of these settings in the<br />
preview table. If the data shown is not<br />
correct or an error is reported, you<br />
can adjust the settings manually (see<br />
below). When the node is executed it<br />
reads in the entire file and caches it in<br />
a temporary file for faster access by<br />
the connected successor nodes. It<br />
also stores all possible values it came<br />
across for each column.<br />
Image viewer<br />
Based on a previously developed<br />
microscope image converter called<br />
Screening Image Browser (SIB) 9<br />
which provided us with a convenient<br />
way to view digital microscope slides<br />
produced via screen direct from<br />
image storage. We were able to<br />
read different image formats: TIFF 8<br />
and 16 bit (BD Pathway, MD<br />
ImageXpress), FLEX Evotec<br />
Technologies, LSM Zeiss, LEI Leica,<br />
TIFF ArrayScan, CellWorx. Using this<br />
driver we can read images during a<br />
screening process in microscope<br />
format direct from microscope<br />
storage and provide them into<br />
pipeline of HC/DC. Image driver in<br />
same <strong>time</strong> extract metadata from<br />
microscope scan and display those<br />
information on image.<br />
Nodes for small molecule screening<br />
In provided pipeline, system based on<br />
Knime, exists a collection of plugins<br />
for chemical structure handling.<br />
Those nodes can read different file<br />
formats with molecules like Mol2,<br />
SDF file and creates a column with<br />
each molecule in a new row. The<br />
molecules' names are used as row<br />
IDs as long as they are unique<br />
(otherwise an artificial name is<br />
created based on the row index).<br />
View molecules can be displayed as<br />
3D depictions of molecular<br />
structures. Other collection of nodes<br />
can wrap all data cells of a column<br />
into Smiles cells. This node does not<br />
validate the structures but simply<br />
marks the strings as being of Smiles<br />
type. Node “Generates 2D<br />
coordinates” is able to display<br />
chemical structure in a 2D viewer.<br />
Implementation<br />
HC/DC was implemented in Java,<br />
which is one of the most widely used<br />
languages for platform independent<br />
programming. The functional modules<br />
are realised as subroutines which are<br />
either fully coded in Java or include<br />
calls to R, Weka, ImageJ scripts or<br />
external programs. The method of<br />
coding depends on the statistical<br />
complexity and the need for speed.<br />
For example, the VSN normalisation<br />
method is available within the pipeline<br />
as an R package, while a permutation<br />
program that calculates P-values is<br />
written in Python to achieve a speed<br />
increase of several magnitudes<br />
compared with the R or Java<br />
equivalents. To allow the parallel<br />
execution of multiple analysis tasks,<br />
HC/DC can run on a cluster and<br />
automatically swaps large jobs to idle<br />
nodes. No special requirements are<br />
necessary to use the service and results<br />
can be viewed through a range of web<br />
browsers on any major operating<br />
system. HC/DC like KNIME is built<br />
upon Eclipse, employing its wealth of<br />
functionality in a variety of ways.<br />
A key concept behind Eclipse is its<br />
use of plugins which can be added<br />
onto an existing installation to provide<br />
additional functionality. The existing<br />
installation does not need to know<br />
about these extension plugins<br />
beforehand, it just has to offer a<br />
so-called extension point where other<br />
plugins can register themselves and<br />
offer additional functionality. At<br />
Eclipse's base/core, there is just a small<br />
run<strong>time</strong> engine that executes plugins<br />
and determines their dependencies.<br />
22<br />
www.europeanpharmaceuticalreview.com
More<br />
Imaging<br />
Speed<br />
Introducing the IsoCyte laser scanning cytometer, the latest<br />
addition to MDS Analytical Technologies’ complete solution for<br />
cellular imaging. The IsoCyte system offers high-throughput, large<br />
format, laser scanning for a wide variety of imaging applications.<br />
See more and see it faster with the IsoCyte system.<br />
More Speed<br />
Image-based assays from acquisition to analysis in 5 minutes<br />
More Data<br />
Large depth of field (±200 μm) eliminating setup <strong>time</strong> or<br />
focusing<br />
Whole well/whole plate imaging for detecting rare events and<br />
improving statistics<br />
More Applications<br />
Cell and bead-based homgoeneous assays, arrays, FRET by<br />
anisotropy, 3-D colony assays, and whole organism imaging<br />
Visit www.moleculardevices.com/isocyte for more information.<br />
now part of MDS Analytical Technologies<br />
tel. +44-118-944-8000 | www.moleculardevices.com
HIGH CONTENT SCREENING ISSUE 2009<br />
Conclusion<br />
We described data pipeline<br />
framework to speed up the HCS data<br />
analysis and management. Largescale<br />
HCS data analysis needs flexible<br />
workflow based integration of<br />
different components and subprocesses<br />
from diverse formats<br />
(library, image readers, microscope<br />
nodes, image processing results, data<br />
mining) which can provide in silico<br />
experimental design through visual<br />
programming and execution on grids.<br />
Pipeline system in HCS is a very new<br />
concept and still evolving and the<br />
final goal is a distributed and a<br />
ubiquitous environment which can<br />
integrate all automated microscopes,<br />
available image processing packages,<br />
bioinformatics databases and data<br />
from other large scale experiments<br />
(proteomics, microarray, flow<br />
cytometry, new sequencing, etc).<br />
References<br />
1. Boutros, M., Kiger A.A., Armknecht S.,<br />
Kerr K., Hild M., Koch B., Haas S. A.,<br />
Paro R. & Perrimon N. Genome-Wide<br />
RNAi Analysis of Growth and Viability<br />
in Drosophila Cells Science 303,<br />
832–835 (2004).<br />
2. CellAnalyzer Project.<br />
http://www.cellprofiler.org<br />
3. CellHTS the open-source Bioconductor/R<br />
package and cellHTS49.<br />
http://www.dkfz.de/signaling/cellHTS.<br />
4. ChemSense, TextSense and BioSense<br />
(InforSense),<br />
http://www.inforsense.com/<br />
5. C. Pautasso and G. Alonso. The JOpera<br />
Visual Composition Language,<br />
Journal of Visual Languages and<br />
Computing, in press, Nov. 2004.<br />
C. Pautasso and G. Alonso.<br />
6. Eclipse Foundation, Eclipse 3.1<br />
Documentation, http://www.eclipse.org.<br />
7. Hassan, M., Brown, R.D., Varma, S.,<br />
Brien, O., Rogers, D., 2006.<br />
Cheminformatics analysis and learning<br />
in a data pipelining environment. Mol.<br />
Divers. 10 (3), 283–299.<br />
8. Hudson, T.C., Stapleton, A.E., Brown, J.L.,<br />
2004. Codifying bioinformatics processes<br />
without programming. Drug Discov.<br />
Today: BIOSILICO 2 (4), 140–148.<br />
9. Karol Kozak, Marta Kozak, Eberhard<br />
Krausz: SIB: database and tool for the<br />
integration and browsing of large scale<br />
image high-throughput screening data.<br />
IEEE. Lectures Proceedings. BIDM '06.<br />
10. Kittler, R., Putz, G., Pelletier, L., Poser, I.,<br />
Heninger, A.K., Drechsel, D., Fischer, S.,<br />
Konstantinova, I, Habermann, B., Grabner,<br />
H., Yaspo, M.L., Himmelbauer, H., Korn,<br />
B., Neugebauer, K., Pisabarro, M.T. &<br />
Buchholz, F. An endoribonucleaseprepared<br />
siRNA screen in human cells<br />
identifies genes essential for cell division.<br />
Nature 432, 1036–1040, (2004);<br />
doi:10.1038/nature03159<br />
11. KNIME (Universit¨at Konstanz),<br />
http://knime.org<br />
12. Moffat, J., Grueneberg, D., Yang, X., Kim,<br />
S., Kloepfer, A., Hinkle, G., Piqani, B.,<br />
Eisenhaure, T., Luo B. & Grenier, J. A.<br />
Lentiviral RNAi Library for Human and<br />
Mouse Genes Applied to an Arrayed Viral<br />
High-Content Screen. Cell, Volume 124,<br />
Issue 6, Pages 1283-1298 (2006).<br />
13. Saldanha AJ, Java Treeview—extensible<br />
visualization of microarray data.<br />
Bioinformatics 2004 20(17):3246-3248;<br />
doi:10.1093/bioinformatics/bth349<br />
14. SIEB C., MEINL T., and BERTHOLD, M. R.<br />
(2007): Parallel and distributed data<br />
pipelining with KNIME. Mediterranean<br />
Journal of Computers and Networks,<br />
Special Issue on Data Mining<br />
Applications on Supercomputing and Grid<br />
Environments.<br />
15. Wayne Rasband, ImageJ,<br />
http://rsb.info.nih.gov/ij/<br />
16. Witten IH, Frank E. Data Mining:<br />
Practical Machine Learning Tools and<br />
Techniques, (2005) 2nd edn. San<br />
Francisco: Morgan Kaufmann.<br />
@<br />
email the author<br />
Karol Kozak<br />
Karol Kozak has been influential in the<br />
development of data handling and data<br />
mining tools for High Throughput, High<br />
Content Screening (HCS) over the few<br />
years at Max Planck Institute of Molecular<br />
Cell Biology and Genetic in Dresden<br />
(Germany) and ETH Zurich (Switzerland).<br />
Currently, as Head of Data Handling<br />
Facility, he plays a leading role in defining<br />
the strategy for organising management<br />
large scale biological data. During PHD<br />
<strong>time</strong> Karol Kozak became an expert in<br />
statistical pattern recognition applied to<br />
large scale screening data.<br />
Prof Andrzej Firkowski<br />
Prof Andrzej Firkowski received a M.S.<br />
from Technical University of Breslau<br />
(Poland) in 1976. He received his Ph.D. in<br />
Chemical Engineering from the Technical<br />
University of Zwickau (Germany) in 1998.<br />
Currently Andrzej Firkowski is a Professor<br />
of Chemical Technology at the Technical<br />
University of Radom and Director of the<br />
Faculty of Materials Science and<br />
Technology. Prof. Firkowski is the author of<br />
many articles and he holds 14 patents<br />
deriving from his research. Prof. Firkowski<br />
has business experience from a 20-year<br />
career as president & chief executive<br />
officer (CEO).<br />
Dr. Gabor Csucs<br />
Dr. Gabor Csucs studied<br />
Physics/Biophysics in Budapest Hungary.<br />
He completed his PhD at the University of<br />
Zurich with optical biosensors as a subject<br />
in 1998. After some years as a postdoc in<br />
the BioMicroMetrics Group at the ETH<br />
Zurich he established the Light<br />
Microscopy Centre at the ETH Zurich in<br />
2003. From 2006 is heading also the<br />
High-thorughput/high-content screening<br />
facility of the ETH Zurich.<br />
24<br />
www.europeanpharmaceuticalreview.com
SUBSCRIBE<br />
ISSUE<br />
2009 LAB AUTOMATION<br />
Article 1:<br />
Lab Automation series<br />
Dr Sheraz Gul, Vice President and Head of Biology, <strong>European</strong> ScreeningPort<br />
Welcome to the first Drug Discovery Developments section of <strong>European</strong> <strong>Pharmaceutical</strong> <strong>Review</strong><br />
Digital. The purpose of the updates is to bring to the attention of the reader advances that have been<br />
made in the life science and pre-clinical Drug Discovery research efforts. Case studies will be<br />
discussed in relation to topics that offer the potential to increase the efficiency and likelihood of<br />
success in Drug Discovery. In this edition, advances in the design and development of automation<br />
solutions for High Throughput Screening (HTS) will be discussed.<br />
The use of small molecule<br />
libraries in HTS campaigns to<br />
identify compounds that are<br />
capable of modulating the activity of<br />
the target under investigation has for<br />
the previous decade been widely<br />
exploited by the pharmaceutical<br />
industry. More recently, screening has<br />
also been adopted by others such as<br />
academic research organisations.<br />
A key requirement for executing HTS<br />
campaigns is access to a compound<br />
library. There are significant<br />
differences in the sizes of the<br />
compound libraries that the<br />
pharmaceutical industry and<br />
academic research organisations<br />
possess or have access to and this<br />
trend is likely to continue for reasons<br />
such as the significant costs<br />
associated with the acquisition of the<br />
libraries, the requirement to purchase<br />
and maintain an appropriate and<br />
reliable hardware and IT<br />
infrastructure to manage the libraries,<br />
the need to perform appropriate<br />
quality control experiments on the<br />
compounds to monitor their integrity<br />
and replacing deteriorated or<br />
depleted stocks. These extra<br />
functionalities require significant<br />
financial investment which usually<br />
can only be fully funded by the<br />
pharmaceutical industry. As a result<br />
of the variation in the sizes of<br />
compound libraries, researchers<br />
require varying degrees of automated<br />
solutions. In those cases where a few<br />
‘‘There are significant<br />
differences in the sizes of the<br />
compound libraries that the<br />
pharmaceutical industry and<br />
academic research<br />
organisations possess or have<br />
access to and this trend is<br />
likely to continue for reasons<br />
such as the significant costs<br />
associated with the<br />
acquisition of the libraries’’<br />
relatively simple manual tasks are<br />
carried out, off the shelf solutions are<br />
usually sufficient. At the other<br />
extreme, there may be a need to cater<br />
for throughputs in excess of 100,000<br />
tests per day. In such cases, suppliers<br />
are often required to provide custom<br />
made solutions and they can be (i) in<br />
the form of an integrated package<br />
from a variety of suppliers or (ii) an<br />
entire solution provided by a sole<br />
supplier. The solutions for those<br />
requiring complex high throughput<br />
capabilities inevitably end up being<br />
financially costly not just in terms of<br />
their initial purchase price, but also in<br />
terms of preventative maintenance<br />
and service contracts.<br />
High Content Screening<br />
In the pharmaceutical industry, High<br />
Content Screening (HCS) is becoming<br />
firmly established and increases the<br />
repertoire of technologies that are<br />
envisaged to aid the reduction in the<br />
high attrition rates in Drug Discovery.<br />
All too often, the identification of<br />
compounds that exhibit the ability to<br />
modulate the activity of a<br />
therapeutically relevant target in<br />
isolation fail to translate their<br />
behavior when evaluated in a cellular<br />
context. Compounds identified from<br />
screening activities against libraries<br />
carried out in a HCS setting may be<br />
better starting points for drug<br />
discovery efforts.<br />
Whereas up until recently HCS<br />
was used after the execution of an<br />
GO TO CONTENTS PAGE 25
LAB AUTOMATION ISSUE 2009<br />
Typical images obtained using the Operetta. HeLa cells stained with HOECHST 33342 (blue), Alexa 488 labeled anti-tubulin antibody<br />
and Alexa 647 labeled anti phosphohistone H3 antibody. Control cells (left) and cells treated with 30 μM cytochalasin (right).<br />
‘‘The newly launched<br />
Operetta by Perkin Elmer<br />
provides a complete solution<br />
for new and experienced users<br />
of HCS systems for use in the<br />
life sciences area for studying<br />
and characterising molecular<br />
and morphological events in<br />
cells and tissues in 96 and<br />
384 micro-titre plates’’<br />
HTS on a small selection of<br />
compounds, it is now possible to use<br />
HCS in a much higher capacity and<br />
speed such that in some cases it can<br />
replace altogether the more routinely<br />
used and relatively simple in vitro<br />
biochemical assays. This has been<br />
possible as a result of the fusion of<br />
the outcomes of the advances in<br />
microscopy, image acquisition and<br />
analysis software, computer<br />
processing power, integration into<br />
automated platforms, molecular<br />
biological techniques to construct<br />
tagged target proteins with suitable<br />
labels such as Green Fluorescent<br />
Protein and the development of<br />
fluorescent dyes. HCS screening<br />
offers the possibility of evaluating the<br />
effect of substances (usually small<br />
molecule compounds) on both<br />
phenotypic end-points as well as on<br />
individual cellular events. The newly<br />
launched Operetta by Perkin Elmer<br />
provides a complete solution for new<br />
and experienced users of HCS<br />
systems for use in the life sciences<br />
area for studying and characterising<br />
molecular and morphological events<br />
in cells and tissues in 96 and 384<br />
micro-titre plates. This is also a<br />
valuable tool making it feasible for<br />
more researchers to become involved<br />
in carry out life science research,<br />
target and biomarker identification<br />
and cytotoxicity profiling.<br />
Automated preparation of<br />
assay plates containing<br />
compounds for screening<br />
purposes<br />
Automated compound handling has<br />
usually been treated separately from<br />
the liquid handling of bulk reagents<br />
for assays. Compounds are usually<br />
stored in 100% v/v DMSO as the<br />
solvent for their dissolution in microtitre<br />
plates. Usually, assays will<br />
contain DMSO at concentrations in<br />
the region of 1% v/v, therefore a<br />
100-fold dilution of compound will<br />
need to be carried out. Thus, for an<br />
assay with a volume of 10μl, the<br />
volume of the compound solution<br />
required will be 0.1μl which is usually<br />
added to assay plates first. This is<br />
then followed by the addition of bulk<br />
biological reagents e.g. in the case of<br />
a simple in vitro assay, one addition of<br />
substrate (5μl) and a second addition<br />
of enzyme (5μl) to initiate the<br />
reaction and a third addition (10μl) to<br />
stop the reaction prior to reading the<br />
micro-titre plate. The types of liquid<br />
handling equipment employed to<br />
dispense compounds and bulk<br />
reagents are usually kept separate<br />
as the volumes required are<br />
substantially different.<br />
Numerous technologies have been<br />
developed that are capable of reliable<br />
dispensing of sub-microlitre volumes<br />
of compounds in DMSO whilst<br />
maintaining sufficient throughput for<br />
the production of plates containing<br />
compounds for screening activities.<br />
‘‘The solutions for those<br />
requiring complex high<br />
throughput capabilities<br />
inevitably end up being<br />
financially costly not just in<br />
terms of their initial purchase<br />
price, but also in terms of<br />
preventative maintenance<br />
and service contracts’’<br />
One such example is that based on<br />
capillary action using specially<br />
designed tips for fixed volume<br />
dispensing (Genomic Solutions®<br />
Hummingbird). In this case, as the<br />
same tips are re-used they need to be<br />
26<br />
www.europeanpharmaceuticalreview.com
ISSUE<br />
2009 LAB AUTOMATION<br />
washed sufficiently after each dispense to prevent<br />
cross contamination. This washing is usually done<br />
with DMSO (the same solvent that is used to dissolve<br />
the compounds) to prevent their precipitation within<br />
the tips and blocking them. A more recent<br />
development for dispensing low volumes of solutions<br />
of DMSO is based on a contactless method using<br />
acoustic droplet ejection and is exploited in the<br />
Labcyte Echo.<br />
As a result of the contactless dispense method there<br />
is no need for the use of disposable tips, the need for<br />
‘‘Numerous technologies have been<br />
developed that are capable of reliable<br />
dispensing of sub-microlitre volumes of<br />
compounds in DMSO whilst maintaining<br />
sufficient throughput for the production<br />
of plates containing compounds for<br />
screening activities’’<br />
washing is overcome, no waste DMSO is generated<br />
and compound adsorption is obviated. This type of<br />
dispensing for the preparation of compound plates is<br />
becoming a standard within the <strong>Pharmaceutical</strong><br />
industry. Also, in an effort to streamline the<br />
preparation of assay plates containing compounds<br />
and the dispensing of bulk biological reagents, Labcyte<br />
have developed the POD 810 system. This is<br />
capable of preparing assay plates containing<br />
compounds and has an integrated liquid handling<br />
system for the dispensation of bulk reagents.<br />
@<br />
email the author<br />
Dr Sheraz Gul<br />
Dr Sheraz Gul is Vice President and Head of Biology at<br />
<strong>European</strong> ScreeningPort, Hamburg, Germany. He is responsible<br />
for the management and development of Medium and High<br />
Throughput Screening activities for academic partners across<br />
Europe. He has 12 years research and development experience<br />
in both academia (University of London) and industry<br />
(GlaxoSmithKline <strong>Pharmaceutical</strong>s). This has ranged from the<br />
detailed study of catalysis by biological catalysts (enzymes and<br />
catalytic antibodies) to the design and development of assays<br />
for High Throughput Screening. He is the co-author of<br />
numerous papers, chapters and the Enzyme Assays: Essential<br />
Data handbook. He can be contacted on<br />
sheraz.gul@screeningport.com<br />
GO TO CONTENTS PAGE
THERMAL ANALYSIS ISSUE 2009<br />
Calorimetry as a<br />
Process Analytical Tool<br />
for micronising<br />
pharmaceuticals<br />
Simon Gaisford, Lecturer in Pharmaceutics, University of London<br />
Drug delivery via the pulmonary route is becoming increasingly popular, but brings a unique set of<br />
formulation challenges. In particular the small particle sizes (between 2-5 μm) required to facilitate<br />
deposition in the lung frequently mean that size-reduction steps (such as milling) are required<br />
during processing of constituents (usually the active principal) prior to formulation.<br />
Typically either ball mills or air-jet<br />
mills are employed to achieve<br />
the desired reduction in particle<br />
size. In the former case ceramic or<br />
metal balls are placed in a container<br />
with the sample and whole apparatus<br />
is rotated; the size, weight and number<br />
of balls can be varied as can the<br />
number of revolutions per minute<br />
and the total milling <strong>time</strong>. In the latter<br />
case compressed air is used to<br />
agitate particles, causing size reduction<br />
Figure 1: Particle size distribution as a function of number of micronisation cycles for<br />
α-lactose monohydrate (air-jet milled with compressed nitrogen, 50 psi)<br />
by particle attrition; the pressure of the<br />
air can be varied and centrifugal force<br />
determines when the particles are<br />
ejected from the mill. On a commercial<br />
scale, batches may be milled a number<br />
of <strong>time</strong>s to ensure the desired particle<br />
size distribution has been achieved.<br />
An example (for an air-jet milled<br />
sample) is given in Figure 1.<br />
Powders with such a small particle<br />
size distribution can be difficult to<br />
aerosolise; many dry powder inhaler<br />
(DPI) formulations thus include a<br />
large, coarse substrate (typically<br />
lactose) on which the micronised<br />
drug is located, bound by a force of<br />
adhesion. When the patient inspires,<br />
the turbulent air-flow aerosolises the<br />
powder blend and then causes<br />
deaggregation of the drug and its<br />
carrier. The large carrier particles<br />
impact the back of the throat and<br />
the micronised drug enters the lung.<br />
Such an approach is convenient,<br />
because the inhaler device is breath<br />
actuated and hence the aerosolisation<br />
step is coordinated with inspiration<br />
28<br />
www.europeanpharmaceuticalreview.com
SUBSCRIBE<br />
ISSUE<br />
2009 THERMAL ANALYSIS<br />
(a common problem with pressurised<br />
metered dose inhalers (pMDI)).<br />
As has been discussed in the<br />
literature 1 , the processes involved in<br />
particle size reduction during milling<br />
are complex. The mechanical forces<br />
imparted to a crystalline material<br />
during milling result initially in a<br />
reduction in particle size, as the<br />
material fractures along natural fault<br />
lines and crystal defects. Eventually<br />
there comes a point at which the bulk<br />
material can no longer fracture and<br />
maximum particle size reduction has<br />
been achieved. Post this point,<br />
mechanical forces are still being<br />
applied and must be absorbed and<br />
dissipated by the sample; these<br />
processes result initially in production<br />
of surface dislocations and eventually<br />
generation of amorphous regions; so<br />
called process-induced disorder.<br />
As has been well documented<br />
elsewhere 2,3 , the amount of<br />
amorphous material formed in this<br />
way is usually small (of the order of a<br />
few percent w/w); however, the nature<br />
of the force applied (impaction)<br />
means that this amorphous material<br />
must necessarily be located on the<br />
surface of the milled material and,<br />
hence, although the milled material is<br />
predominantly crystalline it behaves<br />
as if it were entirely amorphous.<br />
Newell at al 4 published some inverse<br />
phase gas chromatography (IGC) data<br />
(see Table 1) that convey this point. It<br />
is inevitable that this amorphous<br />
material will crystallise, probably over<br />
a relatively short <strong>time</strong>frame as it is<br />
located on a crystalline substrate that<br />
can act as a seed. Crystallisation of<br />
the surface may have disastrous<br />
consequences for the function of the<br />
product 5 , primarily because the force<br />
of adhesion between drug and carrier<br />
will change, altering the balance of<br />
forces that is so important in<br />
modulating product performance from<br />
batch-to-batch and within one batch<br />
over <strong>time</strong>. As such, it is imperative to<br />
understand the mechanisms and<br />
kinetics of crystallisation, such that its<br />
effects can be ameliorated. Often this<br />
means conditioning (under humidity<br />
or some other suitable plasticising<br />
vapour) the material post micronising<br />
to force crystallisation.<br />
If it is true that particle size<br />
reduction occurs before, and not<br />
concomitantly with, generation of<br />
process induced disorder then it is<br />
apparent that it would be possible to<br />
micronise a sample without<br />
generating any amorphous material.<br />
This would remove the need for<br />
downstream conditioning and ensure<br />
consistent product performance. It is<br />
not possible to optimise milling by<br />
following particle size reduction<br />
alone, since there is no particle sizing<br />
technique that can simultaneously<br />
detect formation of amorphous<br />
content. Returning to the data shown<br />
in Figure 1 (page 28), it is tempting to<br />
set up a process involving four to five<br />
micronising steps, since this seems<br />
to ensure a consistent particle size<br />
distribution; such an approach<br />
does nothing to report the generation<br />
of process induced disorder and<br />
can lead to the batch-to-batch<br />
variability in product performance<br />
noted above.<br />
Here, we show that following the<br />
micronising process with an<br />
additional technique, isothermal<br />
microcalorimetry, affords extra data<br />
that allows optimisation of the<br />
micronising <strong>time</strong>. In addition, the data<br />
allow the hypothesis that particle size<br />
reduction occurs before the<br />
generation of process induced<br />
disorder to be tested.<br />
Figure 2: Calibration plot of amorphous content versus heat of crystallisation for<br />
salbutamol sulphate, determined with IGPC<br />
The model drug chosen was<br />
salbutamol sulphate (SS), since this is<br />
a drug commonly formulated for<br />
inhalation. Crystalline SS was<br />
prepared by precipitation from<br />
acetone while amorphous SS was<br />
prepared by spray-drying. Crystalline<br />
SS was ball-milled for increasing<br />
periods of <strong>time</strong>; Laser-diffraction<br />
particle sizing (Mastersizer, Malvern<br />
Instruments Ltd) was used to<br />
measure particle size distributions<br />
and isothermal gas perfusion<br />
calorimetry (IGPC, based in a TAM<br />
Table 1: Dispersive surface energy data (recorded with inverse-phase gas<br />
chromatography) for various lactose samples [4].<br />
Sample Dispersive surface energy (mJ/m 2 )<br />
Amorphous lactose (spray-dried) 37.1 (2.3)<br />
Crystalline lactose monohydrate 31.2 (1.1)<br />
Mixture (ca. 1% w/w amorphous) 41.6 (1.4)<br />
Milled (ca. 1% w/w amorphous) 31.5 (0.4)<br />
GO TO CONTENTS PAGE<br />
29
THERMAL ANALYSIS ISSUE 2009<br />
system, TA Instruments LLC) was<br />
employed to quantify the extent of<br />
any disorder.<br />
The use of IGPC for amorphous<br />
content has become popular and the<br />
technique offers many advantages 6 .<br />
Briefly, in a gas-perfusion experiment<br />
the sample is held in an ampoule<br />
while a carrier gas flows over the<br />
sample. The relative vapour pressure<br />
(RVP) of the gas is controlled by<br />
proportional mixing of dry (0% RVP)<br />
and wet (100% RVP) gas streams.<br />
The system can thus be programmed<br />
to follow a specific RVP program.<br />
IGPC is convenient for quantifying<br />
amorphous material; the sample is<br />
held dry and allowed to reach thermal<br />
equilibrium. The RVP is then<br />
increased (usually to ca. 90%) which<br />
results in solvent uptake by the<br />
sample. This plasticises the sample,<br />
lowering the glass transition<br />
temperature and causing<br />
recrystallisation. The RVP is then<br />
returned to zero, which dries the<br />
sample. Subtraction of the areas<br />
under the wetting and drying peaks<br />
thus indicates the presence of any<br />
irreversible events.<br />
A calibration curve of amorphous<br />
content versus heat of crystallisation<br />
was prepared by mixing proportional<br />
mass ratios of the crystalline and<br />
amorphous SS samples. Full<br />
experimental details are available<br />
upon request.<br />
The calibration curve obtained of<br />
SS amorphous content versus<br />
measured heat was linear (see<br />
Figure 2 on page 29). It was thus<br />
possible to use the data to quantify<br />
extents of disorder in milled SS<br />
samples, although it should be noted<br />
that the use of this type of calibration<br />
curve has some limitations 6,7 .<br />
Principally, the material used to<br />
prepare the calibration curve (a<br />
mixture of wholly amorphous and<br />
wholly crystalline particles) is<br />
physically distinct from the study<br />
material, which in this case will have<br />
disordered regions (that is, a<br />
combination of crystal dislocations<br />
and amorphous areas) forming a<br />
Figure 3: Particle size and amorphous content data as a function of milling <strong>time</strong> for ball<br />
milled salbutamol sulphate<br />
corona around an otherwise<br />
crystalline substrate. The magnitude<br />
of this effect will vary according to the<br />
sample studied. In the case of<br />
lactose6 significant issues arise<br />
because of the existence of 2 anomers<br />
and 3 crystal forms and the effect of<br />
muta-rotation must be considered. For<br />
SS the case is much simpler and<br />
hence confidence in the correlation<br />
between the data sets is higher, since<br />
there are no isomeric forms. Since the<br />
calibration curve was prepared with<br />
totally amorphous material (because<br />
it was spray-dried), the values<br />
reported below for milled SS samples<br />
are expressed in amorphous content<br />
(% w/w) but it is noted that this term<br />
really means extent of disorder and<br />
accounts for crystal dislocations as<br />
well as true amorphous regions.<br />
SS samples were milled for various<br />
<strong>time</strong> periods up to one hour. Samples<br />
were subsequently removed from the<br />
mill and analysed for particle size and<br />
amorphous content immediately (to<br />
reduce the risk of relaxation or<br />
recrystallisation increasing the<br />
particle size or reducing amorphous<br />
content value). The average particle<br />
size and extent of disorder data are<br />
plotted in Figure 3. Over short milling<br />
<strong>time</strong>s there is a sharp reduction in<br />
particle size, with no increase in<br />
extent of disorder detected with the<br />
calorimeter. After ca. five minute<br />
milling <strong>time</strong>, the particle size<br />
distribution remains constant and<br />
there is a concomitant increase in<br />
extent of disorder.<br />
The data clearly shows that the<br />
hypothesis that particle size reduction<br />
occurs before generation of<br />
amorphous content is correct. The<br />
immediate benefit of this is that it<br />
appears feasible to control the<br />
micronising process to produce a<br />
product with a defined particle size<br />
distribution and a crystalline surface.<br />
Although we did not attempt to make<br />
similar measurements on a larger<br />
scale, there would seem to be no<br />
reason why this outcome would not<br />
apply to larger ball milling apparatus;<br />
the break-point <strong>time</strong> at which<br />
maximum particle size reduction is<br />
achieved would simply increase.<br />
Although not explicitly explored in<br />
this study, IGPC analysis may have<br />
application to air-jet milling also. In<br />
this case, samples are ejected from<br />
the mill under centrifugal force, but<br />
changing the pressure of the input<br />
gas would modulate ejection <strong>time</strong> and<br />
attrition rate.<br />
One further conclusion can be<br />
drawn from the data in Figure 3.<br />
At longer milling <strong>time</strong>s, the extent of<br />
disorder starts to reach a plateau<br />
while the particle size starts to<br />
increase. It is likely that at these<br />
longer milling <strong>time</strong>s the conditions in<br />
the mill are such that the ambient RH<br />
or, more likely, temperature become<br />
30<br />
www.europeanpharmaceuticalreview.com
SUBSCRIBE<br />
ISSUE<br />
2009 THERMAL ANALYSIS<br />
sufficiently high to cause some<br />
crystallisation of the amorphous<br />
regions. This in turn would tend to<br />
cause particle agglomeration. Since<br />
extended milling will result in<br />
continual formation of disordered<br />
regions, it seems likely that the<br />
system will reach a dynamic<br />
equilibrium, with both amorphous<br />
content and particle size remaining<br />
constant. This means that, for SS at<br />
least, it would not be possible to<br />
prepare a totally amorphous sample<br />
simply by extended ball-milling.<br />
Of course, these limiting values<br />
may not be ideal from the perspective<br />
of DPI product performance. Further<br />
study could correlate DPI product<br />
performance with extent of disorder<br />
and hence the calorimeter could be<br />
used to optimise the milling process<br />
of the drug prior to formulation.<br />
To summarise, the data shows<br />
that milling of a crystalline<br />
pharmaceutical results initially in a<br />
reduction in particle size with no<br />
measurable surface disorder.<br />
Continued milling does not reduce<br />
particle size further; rather, the<br />
surface of the material becomes<br />
disordered. This disorder can act to<br />
change the surface energy and, as a<br />
direct consequence, may alter DPI<br />
product performance. Controlling this<br />
process starts with being able to<br />
monitor the formation of these<br />
disordered regions. The study has<br />
shown that isothermal calorimetry<br />
has the potential to accomplish this.<br />
Acknowledgements<br />
I would like to thank Mansa Dennison<br />
and Matthew Jones from the School<br />
of Pharmacy, University of London<br />
and Mark Saunders from Synectix<br />
<strong>Pharmaceutical</strong> Solutions Ltd.<br />
References<br />
1. Brodka-Pfeiffer, K., Langguth, P., Graß, P. and<br />
Hausler, H. (2003). Influence of mechanical<br />
activation on the physical stability of salbutamol<br />
sulphate. Eur. J. Pharm. Biopharm., Vol 56,<br />
pp 393-400.<br />
2. Feeley, J.C, York, P., Sumby, B.S., Dicks, H., 1998.<br />
Determination of surface properties and flow<br />
characteristics of salbutamol sulphate, before and<br />
after micronisation. Int. J. Pharm. 172, 89-96.<br />
3. Hogan, S.E., Buckton, G., 2000. The quantification<br />
of small degrees of disorder in lactose using<br />
solution calorimetry. Int. J. Pharm. 207, 57-64.<br />
4. Newell, H.E., Buckton, G., Butler, D.A., Thielmann,<br />
F., Williams, D.R., 2001. Pharm. Res. 18, 662-666.<br />
5. Hancock, B., Zografi, G., 1997. Characteristics and<br />
significance of the amorphous state in<br />
pharmaceutical systems. Int, J. Pharm. 86, 1-12.<br />
6. Ramos, R., Gaisford, S. and Buckton, G. (2005).<br />
Calorimetric determination of amorphous content<br />
in lactose; A note on the preparation of calibration<br />
curves. Int. J. Pharm., Vol 300, pp 13-21.<br />
7. Gaisford, S. and Ramos, R. (2007). Calorimetry for<br />
amorphous content quantification. Eur. Pharm. Rev.,<br />
Issue 3, pp 46-52.<br />
Simon Gaisford<br />
Dr Gaisford is a Senior Lecturer in<br />
Pharmaceutics and runs a laboratory<br />
dedicated to the application of thermal<br />
analysis in the development of medicines.<br />
He is a founding partner and Director of<br />
Synectix <strong>Pharmaceutical</strong> Solutions Ltd, and<br />
Chair of the Thermal Methods Group<br />
(Royal Society of Chemistry).<br />
@<br />
email the author<br />
Thermal Analysis<br />
Excellence<br />
METTLER TOLEDO<br />
sets the standards<br />
in thermal analysis<br />
as with its worldclass<br />
balances.<br />
Mettler-Toledo AG, CH-8603 Schwerzenbach<br />
Tel. +41-44-806 77 11<br />
www.mt.com/ta
PAT ISSUE 2009<br />
PAT and QbD aspects on<br />
stem cell manufacture<br />
Professor Carl-Fredrik Mandenius, Head of the Division of Biotechnology and Mats Björkman, Head of the Division of Assembly<br />
Technology/Production Engineering, Linköping University, Sweden<br />
Process analytical technology (PAT) and quality-by-design (QbD) considerations should be introduced<br />
in emerging manufacturing processes for clinical-grade stem cell products. The currently applied<br />
procedures for stem cell differentiation and expansion are developed for research and discovery<br />
purposes. Protocols must be adapted to current GMP standards and quality aspects would benefit<br />
considerably from being implemented in these procedures according to QbD principles and by PAT,<br />
especially if this is done at an early stage of process development and manufacturing. By applying<br />
modern conceptual design science methodology the quality of the manufacturing design can be<br />
efficiently enhanced.<br />
We have previously in this<br />
journal described how<br />
modern design science<br />
concepts can be adapted to address<br />
critical issues according to the PAT<br />
and QbD principles for<br />
pharmaceutical production 1,2 . Here we<br />
apply these ideas to a new area of<br />
biotechnology-related production –<br />
the manufacture of stem cells.<br />
The production of stem cells is<br />
today in its infancy. It is therefore<br />
<strong>time</strong>ly to apply established modern<br />
methods for design and optimisation<br />
to the manufacturing of stem cells<br />
and stem cell-derived products and,<br />
in particular, to do that with PAT and<br />
QbD aspects in mind 3-5 . This is a<br />
challenging possibility for stem cell<br />
R&D companies which intend to<br />
Figure 1: Overview of the operational steps in the processing of stem cells. Flow chart<br />
diagram represents both the manufacturing line to undifferentiated stem cells and the<br />
line to a differentiated cell type derived from the same hESC (for explanation, see text).<br />
transform their established cells and<br />
cell lines into bio-products which can<br />
be reliably and efficiently<br />
manufactured according to current<br />
Good Manufacturing Practice (GMP)<br />
and in compliance with regulatory<br />
directives and guidelines from the<br />
<strong>European</strong> Commission, FDA, EMEA,<br />
and other medicinal authorities 6-8 .<br />
This is especially valid for clinicalgrade<br />
stem cell products for cell<br />
therapy applications where the<br />
derived cells are to be transplanted<br />
into a patient 9 . Moreover, application<br />
of stem cells in the drug discovery and<br />
development process must also meet<br />
very high standards to be useful for<br />
toxicity testing and other preclinical<br />
studies where the results shall be<br />
included in a drug application file 10 .<br />
Stem cell manufacturing is<br />
ethically sensitive when the cells<br />
come from a human source,<br />
especially when it is of embryonic<br />
origin, as human embryonic stem<br />
cells (hESC). The ethical<br />
complications of using human<br />
embryos as the starting material of<br />
32<br />
www.europeanpharmaceuticalreview.com
SUBSCRIBE<br />
ISSUE<br />
2009 PAT<br />
a bioprocess further emphasises the<br />
need of a thorough analysis of testing<br />
and optimisation possibilities with<br />
modern PAT methods.<br />
The variety of procedural<br />
alternatives for propagation,<br />
expansion, differentiation,<br />
preservation and characterisation of<br />
hESC are indeed multi-parametrical<br />
and need to be adapted, refined and<br />
optimised to satisfy the demands<br />
of the regulators, including the<br />
expectations inherent in guidelines<br />
of QbD and PAT.<br />
Stem cell technology<br />
The first step in the preparation of<br />
hESC and hESC derived products is to<br />
enzymatically digest the inner cell<br />
mass of embryonic blastocytes 11-12 .<br />
The isolated cell mass is placed on a<br />
layer of growth-inhibited mouse<br />
fibroblast feeder cells in tissue culture<br />
dishes. After one or two weeks the<br />
outgrown cells from the inner cell<br />
mass are isolated, by dissection or<br />
enzymatic treatment, and transferred<br />
to new culture dishes for further<br />
growth in an un-dissociated cellular<br />
state. The feeder cells provide<br />
differentiation inhibitors which hinder<br />
the stem cells to differentiate.<br />
The stability of the cell culture is<br />
monitored through characterising<br />
specific biomarker molecules,<br />
typically cell surface proteins<br />
(e.g. SSEA-3/4, Oct3/4 and Nanog)<br />
which indicate that the cell mass<br />
remains in the undifferentiated state 13 .<br />
Additional characterisation and<br />
control methods are appropriate to<br />
use, such as gene array analyses, flow<br />
cytometry other methods 14,15 .<br />
If the production goal is to<br />
produce undifferentiated stem cells,<br />
the characterised cells are isolated<br />
and further expanded under<br />
controlled conditions (see Figure 1 on<br />
page 34). Up to 100 passages have<br />
been reported provided efficient<br />
inhibition factors are supplied 10 .<br />
If the goal of the production is a<br />
particular cell type, a well defined<br />
differentiation protocol ensues in a<br />
stepwise procedure of propagation<br />
Figure 2: The Hubka-Eder model where the general model is supported by a biological<br />
systems entity to clearly depict how a biological system interacts and effects the other<br />
systems in the transformation occurring during a production process.<br />
and isolation of the cells. The added<br />
differentiation factors and the culture<br />
conditions applied stimulate the cells<br />
to develop into progenitor cells and<br />
matured cells from which the desired<br />
cell type is isolated. This procedure<br />
takes an additional two to four weeks.<br />
Crucial for a successful result is<br />
continuing characterisation of<br />
biomarker molecules, now others that<br />
are characteristic for the desired cell<br />
type. Supporting the biomarker data<br />
are other functionality tests for the<br />
cell type. The cells are going through<br />
several isolations by manual<br />
microscopic selection and dissection.<br />
Furthermore, the cells are<br />
cryopreserved intermittently and<br />
preservation and shipping procedures<br />
require careful handling of the cells.<br />
The flow chart in Figure 1 (see<br />
page 32) summarises the steps in<br />
the derivation of the stem cells<br />
with accompanying analytical<br />
characterisation. This chart of<br />
procedural steps is the blue print<br />
which should be transformed and<br />
scaled-up to regular manufacturing.<br />
Figure 3: The HE model adapted to a simplified manufacturing where embryonic stem<br />
cells are the starting material. This example refers to a manufacturing process producing<br />
a particular differentiated cell type.<br />
GO TO CONTENTS PAGE 33
PAT ISSUE 2009<br />
Design theory<br />
Of the conceptual models that are used<br />
in design theory 16-18 much attention has<br />
been given to the Hubka-Eder (HE)<br />
model 19-20 due to its generality and<br />
applicability to any technical system for<br />
manufacturing. Most applications are<br />
found in mechanical and electrical<br />
engineering although the theoretical<br />
model is as applicable for chemical and<br />
biological products.<br />
The HE-model is built around the<br />
transformation process (TrP) and its<br />
applied technologies (Tg) to transform<br />
the inputs, being the sums of<br />
operands (ΣOd1), i.e. the raw<br />
materials, properties (ΣPr1) and<br />
secondary components (ΣSecIn), into<br />
outputs (ΣOd2, ΣPr2, and ΣSecOut)<br />
(see Figure 2). The model then<br />
describes and analyses all effects that<br />
are exerted on the TrP. That includes<br />
the technical systems involved (ΣTS),<br />
all information systems (ΣIS) and all<br />
management and goal-setting systems<br />
(ΣM&GS). Thus, processing machines<br />
such as blending vessels, freezers, and<br />
robots are example of technical units<br />
and systems ΣTS) which transform<br />
chemical or biological reactants and<br />
additives by causing reaction,<br />
interaction or separation but without<br />
being transformed themselves. Data<br />
collection systems such as data<br />
processing software and batch<br />
reporting procedures are included<br />
within the ΣIS while protocols,<br />
procedures and quality guidelines<br />
such as GMP and operational SOPs<br />
for the TrP act to manage to reach set<br />
manufacturing goals and quality<br />
criteria in the ΣM&GS.<br />
The purpose with the design model<br />
is to carry out a systematical<br />
investigation of effects in all steps<br />
involved. The interaction chart in<br />
Figure 2 (see page 33) illustrates the<br />
top level of approaching the system; a<br />
more detailed chart on a second or<br />
third level would elucidate in a much<br />
more thorough and informative way<br />
the multitude of interacting effects.<br />
This analysis should then lead to the<br />
identification of alternative, potentially<br />
better design solutions. By this<br />
systematic approach too early fixations<br />
with un-reflected solutions is avoided.<br />
The HE-model has recently been<br />
adapted by us for biotechnology<br />
products and systems by adding a<br />
separate biological systems entity,<br />
ΣBS, to the design concept with the<br />
objective of deepening the functional<br />
analysis of the cross-interactions<br />
between the participating<br />
subsystems 21 . By that, complex<br />
biotechnology processes such as cell<br />
culturing procedures that exploit the<br />
transformation capacity of the<br />
biological systems per se, are easier<br />
to describe and analyse thoroughly.<br />
As we have brought up in a separate<br />
paper 1 , the concepts of PAT and<br />
QbD benefit from being analysed<br />
within the HE-modeling, since PAT<br />
includes the technical systems which<br />
should according to the PATguidelines<br />
3 be used to direct the<br />
design of the manufacturing process<br />
and the QbD principles should<br />
interconnect and affect the product<br />
and manufacturing quality through its<br />
early integration in the system.<br />
Figure 3 on page 33 illustrates how<br />
the embryonic stem cell<br />
transformation procedures in principle<br />
could be related in a HE model. Note
SUBSCRIBE<br />
ISSUE<br />
2009 PAT<br />
especially the fundamentally different<br />
way of describing the steps of the<br />
procedures compared to a typical<br />
laboratory manual. Unlike a SOP the<br />
functions of the steps are separated<br />
and thoroughly investigated in terms<br />
of effects on the procedural steps.<br />
Note also the influence of the<br />
active environment. The hESC process<br />
is subject to longterm exposure of<br />
contaminating disturbances from the<br />
environment – not possible to foresee,<br />
but to counteract. The biological<br />
variation, which is so important to<br />
measure or detect is included in the<br />
active environment.<br />
The depiction here is also<br />
integrating PAT and QbD in the<br />
conceptual design, in accordance<br />
with the regulators’ mindset, i.e. to<br />
bring in the quality aspects directly<br />
into the design process and with the<br />
help of available powerful PAT<br />
methods and tools.<br />
It is our interpretation of PAT<br />
that the HE modeling methodology<br />
can be considered as one of the<br />
useful PAT tools.<br />
Manufacturing aspects on<br />
stem cell production<br />
Scaled-up manufacturing of hESC and<br />
hESC-derived products requires<br />
access to exclusive biological<br />
systems, perfectly controlled and<br />
contamination-free manufacturing<br />
facilities, traceability of raw materials,<br />
xeno-free media, specialised<br />
equipment for handling cell culture, a<br />
variety of analytical instrument and<br />
quality control devices and, most<br />
importantly, very skilled and well<br />
trained staff and operators. The<br />
equipment should preferably be as<br />
automated as possible and the<br />
manufacturing should be performed<br />
in a clean room surrounding to reduce<br />
the risk for contamination.<br />
Most of the manufacturing<br />
equipment is commercially available<br />
such as freezing equipment, robots,<br />
incubators, microscopes, analysers,<br />
autopipets, vessels and other culturing<br />
devices. Consumables, single-use<br />
plastics, reagents and media are<br />
Figure 4: The HE depiction of the hESC manufacturing process in Figure 3 is detailed with<br />
additional effecting units to illustrate how the HE analysis is carried on for further<br />
investigating the interactions.<br />
supplied by numerous companies.<br />
The components are mostly used<br />
based on standard operating<br />
procedures and validated GMP<br />
protocols for preparation and testing.<br />
The procedures and protocols<br />
developed during the research and<br />
development phases should be<br />
framed into a large scale and adapted<br />
as much as possible to automation.<br />
This requires extensive further<br />
validation and reconsideration of<br />
effects caused by the scaled-up<br />
manufacturing systems.<br />
For example, procedures that can<br />
be automated in cell culturing robotic<br />
systems are expansion and<br />
maintenance of multiple cell lines,<br />
various sub-culturing steps,<br />
expanding cell numbers through the<br />
seeding of a number of flasks,<br />
harvesting and plating of cells for<br />
assays and screening, incubation in<br />
different flasks and plate formats, cell<br />
counting and viability measurement,<br />
processing of multilayer flasks, and<br />
automated plating. Successful<br />
automation of such procedures<br />
is accomplished for example in the<br />
Automation Partnership systems 22,23 .<br />
Other important functions the<br />
scaled-up manufacturing system<br />
should include are:<br />
■<br />
■<br />
■<br />
■<br />
■<br />
■<br />
Quality control of cell banks<br />
Quality control of input materials<br />
(culture media, factors,<br />
consumables)<br />
Quality control methods for<br />
manufactured cells (monitoring<br />
biomarkers; Functional testing of<br />
final cells, genomic analyses,<br />
histological analyses, microscopy,<br />
flow cytometry methods)<br />
Cryopreservation procedures<br />
(deep freezers for preserving<br />
cell materials at various stages of<br />
the process)<br />
Isolation procedures (dissection in<br />
microscopy, enzyme treatments)<br />
Preservation/storage of the<br />
product and shipping procedures<br />
Transforming the stem<br />
technology into an efficient<br />
manufacturing system by<br />
using conceptual design<br />
The complexity of these technical<br />
procedures for stem cell production<br />
makes it obvious that the<br />
transformation process into an<br />
efficient bio-mechatronic<br />
manufacturing system is difficult,<br />
<strong>time</strong>-consuming and, in the end, very<br />
cost dependent.<br />
To exploit modern manufacturing<br />
system development methods may<br />
GO TO CONTENTS PAGE 35
PAT ISSUE 2009<br />
appear as an inevitable stage for<br />
reaching a competitive manufacture.<br />
However, the design theory is<br />
originally shaped and predominantly<br />
used for “cars and freezers”. The<br />
Hubka-Eder model has, to our<br />
knowledge, never before been<br />
adapted for and applied to sensitive<br />
biological systems in a mechatronic<br />
environment.<br />
It is most likely that a<br />
manufacturing facility for hESC<br />
products needs to be flexible in terms<br />
of different final products, i.e. hESCderived<br />
cell types or undifferentiated<br />
hESC materials of a particular cell line.<br />
The customers will no doubt have<br />
varying demands depending on the<br />
purpose of the cells (for clinical-grade<br />
use, for drug testing etc). The quantity<br />
of the cells required will vary<br />
considerably over <strong>time</strong>. The<br />
distribution/shipping conditions will be<br />
more or less demanding. A hESC<br />
manufacturer may, of course, specialise<br />
on certain customers and cell types.<br />
However, from a manufacturing<br />
engineering perspective, an interesting<br />
question is how to make a robust<br />
flexible manufacturing system that can<br />
cope with the varying demands in order<br />
to be as competitive as possible on the<br />
market. The demands for robustness<br />
imply that robots should be used<br />
instead of human operators whenever<br />
possible. The variability and risk for<br />
mistakes due to monotonous and<br />
repetitive work are much smaller for<br />
robotic systems than for human<br />
operators.<br />
In Figure 4 (see page 35) the<br />
stem cell production stages are<br />
transformed into the Hubka-Eder<br />
design model on a more detailed<br />
level. The shown example concerns a<br />
manufacture of a differentiated cell<br />
type: from the isolation of embryonic<br />
cells of blastocytes, their<br />
propagations on feeder cells,<br />
Figure 5: Examples of the relationship between the functional and anatomic structures for<br />
the stem cell expansion. The left part shows the functions, i.e. to acquire cells from<br />
supernumerous embryos, isolating the cells from the inner cell mass with the desired<br />
cellular properties, outgrowing these isolates successfully so that the undifferentiated<br />
state is maintained, following by repeated passages to expand the undifferentiated cell<br />
culture and where all these steps are monitored appropriately with characterisation<br />
methods. In the right part of the figure the technical means to accomplish these functions<br />
are added where the physical apparatuses and other equipment/disposables are added.<br />
As illustrated several anatomic units are required to accomplish a particular function. In<br />
reality, many charts of this sort are needed in the HE analysis.<br />
expansion of undifferentiated cells<br />
and further propagation and<br />
differentiation. It should be<br />
understood that the scheme is an<br />
illustration that should be further<br />
detailed to allow a useful analysis of<br />
effects. HE-models should be used for<br />
analysing different levels of detail of<br />
the manufacturing system. You start<br />
with an overall model which is then<br />
further developed into one or several<br />
models that are more detailed. Of<br />
particular interest for the PAT and<br />
QbD issues is how analytical devices<br />
and instruments are utilised and<br />
distributed in order to evolve the<br />
most beneficial effects on the design<br />
solution. The key PAT components are<br />
obviously the tools to characterise the<br />
cells at various stages. The effects of<br />
these analytical procedures, their<br />
precision and just-in-<strong>time</strong> capacity<br />
could show their most favourable use<br />
for the functional purpose.<br />
The QbD methods, especially the<br />
definition of design space and control<br />
space variables in the procedural<br />
steps of robots, freezing equipment<br />
and <strong>time</strong> lapses are decisive<br />
parameters for establishing the<br />
optimal manufacturing system. The<br />
number of alternative solutions will,<br />
especially when considering these<br />
aspects, grow significantly which<br />
underscores the need for a systematic<br />
approach when designing the system.<br />
In Figure 5 the HE methodology<br />
for analysing the relationship between<br />
the functional purpose of the<br />
production steps and the anatomy of<br />
operation units in the design is<br />
illustrated. The figure exemplifies a<br />
limited part of a chart of functional<br />
and anatomic steps in stem cell<br />
manufacture – the propagation of<br />
undifferentiated stem cells. The chart<br />
should, in a complete analysis, be<br />
supported by ranking tables that<br />
systematically assess the alternative<br />
design solutions, and their benefits<br />
and capacities. The advantage with<br />
this, if it is done at an early stage of<br />
development, is that there is a good<br />
chance of adapting the design in the<br />
pre-production scaled-down stage.<br />
To include effects of the defined<br />
design space and being supported by<br />
PAT tools is definitely an advantage.<br />
Conclusion<br />
It is not mentioned in the regulators’<br />
guidelines for PAT and QbD that a<br />
bio-mechatronic design tool is one<br />
of the necessary tools for successful<br />
design of pharmaceutical processes.<br />
However, many of the inherent<br />
criteria of such tools are manifested<br />
in the HE design approach. It is an<br />
example of a scientific method and a<br />
tool for knowledge management and<br />
knowledge analysis, albeit not by<br />
using traditional knowledge<br />
management tools.<br />
This short paper has highlighted<br />
some of the specific characteristics<br />
of hESC manufacture which need<br />
36<br />
www.europeanpharmaceuticalreview.com
SUBSCRIBE<br />
ISSUE<br />
2009 PAT<br />
extensive attention for a further<br />
exploitation of the hESC technology<br />
into successful products. Here, a<br />
systematic PAT/QbD inspired<br />
approach can be especially rewarding<br />
from the regulators’ perspective as<br />
well as the manufacturers’. To employ<br />
the methodology of manufacturing<br />
systems design may then be a good<br />
way to avoid unnecessary costs and<br />
delays in the development.<br />
Several features can be envisaged<br />
as typical in hESC manufacturing –<br />
high demands on robust flexibility,<br />
long-<strong>time</strong> contracts, and very small<br />
volumes of bioproducts with an<br />
unusually high price. This is in<br />
contrast to most other traditional<br />
areas where the HE model is applied.<br />
However, need for flexibility, the<br />
ongoing development and increasing<br />
competition, the benefits of<br />
automation, and the need of control of<br />
the environment and quality are all<br />
features of importance in most existing<br />
industrial manufacturing systems.<br />
Acknowledgement<br />
The authors would like to thank<br />
VINNOVA (The Swedish<br />
Governmental Agency for<br />
Innovation Systems) for valuable<br />
financial support.<br />
References<br />
1. Mandenius CF, Derelöv M, Detterfelt J, Björkman<br />
M (2007) Process analytical technology and<br />
design science, Eur Pharm Rev 3, 74-80.<br />
2. Mandenius CF (2006) Process analytical<br />
technology in biotechnology, Eur Pharm<br />
Rev 11, 69-76.<br />
3. Federal Drug and Food Administration (USA)<br />
Guidance for industry, process analytical<br />
technology (PAT) a framework for innovative<br />
pharmaceutical development, manufacture and<br />
quality assurance, FDA, 2004<br />
4. International Conference on harmonization of<br />
technical requirements for registration of<br />
pharmaceuticals for human use (ICH). Document<br />
ICH Q9 Quality Risk Management, 2006<br />
5. International Conference on harmonization of<br />
technical requirements for registration of<br />
pharmaceuticals for human use (ICH). Document<br />
ICH Q10 Note for guidance on <strong>Pharmaceutical</strong><br />
Quality Systems, 2008<br />
6. von Tigerstrom BJ (2008) The challenge of<br />
regulatory stem cell-based products, Trends<br />
Biotechnol 26, 653-658.<br />
7. <strong>European</strong> Commission, Directive of the <strong>European</strong><br />
Parliament and of the Council of 31 March<br />
2004 on setting standards of quality and<br />
safety for donation, procurement, testing,<br />
processing, preservation, storage and<br />
distribution of human tissues and cells, E.C.<br />
Directive 2004/23 (2004)<br />
8. Federal Drug and Food Administration (USA)<br />
Content and <strong>Review</strong> of Chemistry, Manufacturing,<br />
and Control (CMC) Information for Human<br />
Somatic Cell Therapy Investigational New Drug<br />
Applications (INDs), FDA, 2008<br />
9. Unger C, Skottman H, Blomberg P, Dilber MS,<br />
Hovatta O (2008) Good manufacturing practice<br />
and clinical-grade human embryonic stem cell lines,<br />
Human Mol Genetics 17, R48-R53.<br />
10. Sartipy P, Björquist P, Strehl R, Hyllner J (2007)<br />
The application of human embryonic stem cell<br />
technologies to drug discovery, Drug Discovery<br />
Today 12, 688-699.<br />
11. Ameen C, Strehl R, Björquist P, Lindahl A,<br />
Hyllner J, Sartipy P (2008) Human embryonic<br />
stem cells: current technologies and emerging<br />
industrial applications, Crit Rev Oncol Hematol<br />
65:54–80.<br />
12. Ström S, Inzunza J, Grinnemo KH, Holmberg K,<br />
Matalainen E, Strömberg AM, Blennow E,<br />
Hovatta O (2007) Mechanical isolation of inner<br />
cell mass is effective in derivation of new human<br />
embryonic stem cell lines, Human Reproduction 22,<br />
3051-3058.<br />
13. Adewumi O, et al (2007) Characterization of<br />
human embryonic stem cell lines by the<br />
international stem cell initiative, Nature Biotechnol<br />
25:803-816.<br />
14. Narkilahti S, Rajala K, Pihlajamäki H, Suuronen R,<br />
Hovatta O, Skottman H (2007) Monitoring and<br />
analysis of dynamic growth of human embryonic<br />
stem cells, comparison of automated<br />
instrumentation and conventional culturing<br />
methods, Biomed Eng Online 6:11, 1-8<br />
15. Chang KH, Zandstra PW. Quantitative screening of<br />
embryonic stem cell differentiation: endoderm<br />
formation as a model. Biotechnol Bioeng.<br />
2004;88:287–298.<br />
16. Pahl G, Beitz W (1996) Engineering design, a<br />
systematic approach, Springer Verlag, Berlin.<br />
17. Rozenburg N F M, Eekels J (1996) Product design,<br />
fundamentals and methods, John Wiley & Sons,<br />
Chicester<br />
18. Ulrich KT, Eppinger SD (2004) Product design and<br />
development, 3rd edition, McGawn-Hill, New York.<br />
19. Hubka V, Eder WE (1988) Theory of technical<br />
systems, a total concept theory for engineering<br />
design, Springer Verlag, Berlin.<br />
20. Hubka V, Eder WE (1996) Design science, Springer<br />
Verlag, Berlin.<br />
21. Derelöv M, Detterfelt J, Björkman M,<br />
Mandenius CF (2008) Engineering design<br />
methodology for bio-mechatronic products,<br />
Biotechnol Prog 24, 232-244.<br />
22. The Automation Partnership, Cambridge (UK),<br />
http://www.automationpartnership.com/<br />
23. Drake RAL, Oakeshott RBS (2005) Smart cell<br />
culture, <strong>European</strong> Patent Application EP 1598415<br />
Prof Carl-Fredrik Mandenius<br />
Prof Carl-Fredrik Mandenius is head of<br />
Division of Biotechnology at Linköping<br />
University. His main research interests<br />
include biochemical and bio-production<br />
engineering, bioprocess monitoring and<br />
control, stem cell technology, and<br />
biosensor technology. He has previously<br />
been a director for process R&D at<br />
Pharmacia AB and is presently coordinator<br />
of two EU-networks on hESC-derived<br />
models for drug testing.<br />
Prof Mats Björkman<br />
Prof Mats Björkman is head of the Division<br />
of Assembly Technology / Production<br />
Engineering at Linköping University,<br />
Sweden. His main research interests<br />
include design and operation of flexible<br />
manufacturing systems and equipment.<br />
Another research area is the interface and<br />
integration between manufacturing and<br />
product design, including design science.<br />
The research has developed from<br />
traditional mechanical industry to also<br />
include areas as electronic manufacturing,<br />
and manufacturing of biotech equipment<br />
and pharmaceutical products.<br />
@<br />
email the author<br />
GO TO CONTENTS PAGE 37
PharmaINFOcus<br />
A REGULAR ROUND-UP OF INDUSTRY PRODUCTS, INNOVATION, NEWS AND DEVELOPMENTS<br />
@<br />
email Carrie Lancaster clancaster@russellpublishing.com<br />
Fluidigm systems revolutionise quantative <strong>PCR</strong><br />
Fluidigm systems alleviate the burdens of<br />
intensive pipetting and high reagent costs<br />
through the use of integrated fluidic circuits<br />
(IFC) — integrated circuits for biology.<br />
Thus, Fluidigm systems are igniting exciting<br />
advances for traditional and emerging<br />
applications in areas such as:<br />
■ Large-scale relative gene expression<br />
quantification<br />
■ Single cell gene expression<br />
■ Copy number variation<br />
■ Library quantification for next<br />
generation sequencing<br />
■ Gene Expression Quantification at<br />
Incredible Throughput.<br />
An IFC called the BioMark 96.96<br />
Dynamic Array has the capacity to<br />
generate 9,216 data points simultaneously<br />
(96 samples x 96 primer-probe pairs),<br />
while requiring 100-<br />
fold less pipetting.<br />
This capability is ideal<br />
for validation of<br />
candidate genes<br />
obtained through<br />
microarray screening<br />
and for analysis of<br />
single cell samples.<br />
These capabilities are<br />
enabling a revolution<br />
in what types of gene<br />
expression studies<br />
are possible.<br />
sensitivity; however, its implementation has<br />
historically been too <strong>time</strong>-consuming for<br />
routine laboratory use. Now, a Fluidigm IFC<br />
called the 12.765 Digital Array has brought<br />
the promise of digital <strong>PCR</strong> to labs<br />
worldwide. The process requires only<br />
minutes of hands-on <strong>time</strong>, and results are<br />
available in just a few hours. Since its<br />
introduction, the digital array has been<br />
proven in multiple applications requiring<br />
extreme <strong>PCR</strong> sensitivity, including the study<br />
of variations in copy number, where it<br />
allows scientists to distinguish between<br />
four and five copies, and sample<br />
preparation for next generation sequencing,<br />
where it allows scientists to eliminate <strong>time</strong><br />
consuming and expensive titration steps.<br />
Click here for more information about life<br />
science with Fluidigm IFCs.<br />
About Fluidigm<br />
Fluidigm develops, manufactures and<br />
markets proprietary Integrated Fluidic<br />
Circuit (IFC) systems that significantly<br />
improve productivity in life science<br />
research. Fluidigm’s IFCs enable the<br />
simultaneous performance of thousands of<br />
sophisticated biochemical measurements<br />
in extremely minute volumes. These<br />
“integrated circuits for biology” are made<br />
possible by miniaturising and integrating<br />
liquid handling components on a single<br />
microfabricated device. Fluidigm’s IFC<br />
systems, consisting of instrumentation,<br />
software and single-use IFCs, increase<br />
throughput, decrease costs and enhance<br />
sensitivity compared to conventional<br />
laboratory systems. Fluidigm products have<br />
not been cleared or approved by the Food<br />
and Drug Administration for use as a<br />
diagnostic and are only available for<br />
research use.<br />
Digital <strong>PCR</strong> made fast<br />
and easy<br />
Digital <strong>PCR</strong> represents<br />
a quantum leap in<br />
q<strong>PCR</strong> resolution and<br />
Figure 1: Heat Map. Data obtained with 24 samples and 96 gene<br />
expression assays on the 96.96 dynamic array<br />
Figure 2: Examples of real-<strong>time</strong> curves<br />
from the 96.96 Dynamic Array<br />
Thermo Fisher Scientific expands the utility of<br />
high content for toxicity testing<br />
Thermo Fisher Scientific has expanded their range of Thermo<br />
Scientific Cellomics BioApplication image processing tools for<br />
toxicology studies. The new Comet v3.0 BioApplication<br />
automates the detection and analysis of the single cell gel<br />
electrophoresis assay often used by toxicologists studying the<br />
amount of DNA damage caused by compound or radiation<br />
exposure. This assay, along with the well established<br />
Micronucleus BioApplication, permits scientists to more fully<br />
profile compounds for both cytotoxicity and genotoxicity<br />
endpoints in a more automated, less subjective manner.<br />
Additionally, Thermo scientists have enabled their high content<br />
platform to study whole organisms with their soon to be released<br />
ZebraTox v3.0 BioApplication which addresses acute toxicity<br />
seen in developing zebrafish embryos.<br />
As a new tool in organism research, the BioApplication<br />
allows for multiple morphological measurements to<br />
determine shape changes in zebrafish embryos along with<br />
capability to look at fluorescent measurements of targets such<br />
as in the Zygogen Z-tag SM fluorescent embryos.<br />
www.thermofisher.com<br />
38<br />
www.europeanpharmaceuticalreview.com
PharmaINFOcus<br />
A REGULAR ROUND-UP OF INDUSTRY PRODUCTS, INNOVATION, NEWS AND DEVELOPMENTS<br />
MDS Pharma Services appoints<br />
Dr. Jessica Liu to Lead Clinical<br />
Operations in Asia<br />
MDS Pharma Services, has appointed<br />
Jessica Liu, M.D., to the position of<br />
Clinical Operations Director and Head of<br />
the Asia-Pacific region for its late-stage<br />
clinical trial management division. The<br />
appointment supports the MDS business<br />
unit’s well-established and growing<br />
presence there and strengthens its ability<br />
to serve clients in an increasingly<br />
important emerging geography.<br />
“Dr. Liu brings many years of<br />
extensive and relevant global<br />
experience in medicine, clinical<br />
research and the pharmaceutical<br />
industry to this important position,”<br />
said MDS Pharma Services President<br />
David Spaight. “Her appointment<br />
reflects our commitment to provide the<br />
highest levels of quality and service to<br />
our clients in a region that has emerged<br />
as a key arena for global clinical<br />
research. Her clinical research<br />
experience encompasses a number<br />
of therapeutic areas in which MDS<br />
Pharma Services specialises,<br />
including oncology, diabetes and<br />
infectious diseases. ”<br />
Prior to joining MDS Pharma<br />
Services, Dr. Liu was with Bristol-Myers<br />
Squibb (BMS) for ten years, most<br />
recently in Brussels where she<br />
managed international BMS projects.<br />
@<br />
email Carrie Lancaster clancaster@russellpublishing.com<br />
IDBS Data Management Solutions deliver proven<br />
efficiency and award winning quality<br />
After 20 years as a leading provider of<br />
data management solutions to research<br />
and development industries, IDBS<br />
continues to strengthen its profitability<br />
and grow its product development.<br />
Among significant business advances<br />
made in 2008, the company experienced<br />
rapid sales growth of its electronic<br />
workbook suite (E-WorkBook) and<br />
increased adoption of its high throughput<br />
screening solution ActivityBase XE.<br />
Neil Kipling, founder and CEO of<br />
IDBS, commented. "In particular, the<br />
market gave a very positive reception to<br />
our chemistry tool set, which is<br />
competing strongly and replacing<br />
She began her career as a resident<br />
physician in the Endocrinology<br />
Department of the Peking Union<br />
Medical College Hospital in Beijing<br />
followed by 15 years in clinical<br />
research, including leadership roles in<br />
clinical operations, project<br />
management and medical affairs.<br />
Dr. Liu holds a Medical Doctorate<br />
degree from China’s prestigious<br />
University of Beijing and a Diploma in<br />
<strong>Pharmaceutical</strong> Medicine from the<br />
University of Basel in Switzerland.<br />
www.mdsps.com<br />
Jessica Liu<br />
existing legacy solutions that have not<br />
kept pace with the changing<br />
environment within R&D."<br />
He added: "With our clients already<br />
reporting 20-30% efficiency savings<br />
directly accountable to the<br />
implementation of our solutions, we<br />
will continue to work hard to maximise<br />
their benefit."<br />
Unlike its competitors, IDBS<br />
continues to invest significantly in<br />
product development. "Our<br />
competitors believe that they can retain<br />
their market share and maintain<br />
customer satisfaction by freezing<br />
development or asking their customer<br />
DxS appoints<br />
Jeff Devlin as new Chief<br />
Operating Officer<br />
DxS Ltd, a personalised medicine company<br />
and leaders in the provision of companion<br />
diagnostics, has recently announced the<br />
appointment of Jeff Devlin as Chief<br />
Operating Officer.<br />
Prior to DxS, Jeff held positions as<br />
Executive VP and Executive Committee<br />
member of Shire <strong>Pharmaceutical</strong>s,<br />
where he played a key role in strategy<br />
and integration. Before this he was a<br />
Partner in Ernst and Young’s life science<br />
division. Jeff has a BSc in Physics from<br />
the University of Edinburgh and an<br />
MSc in Marketing from the University<br />
of Strathclyde.<br />
Dr. Stephen Little, CEO of DxS, said:<br />
“Jeff’s appointment significantly<br />
strengthens the DxS management team<br />
and with his strong background in<br />
operations and strategy, his expertise<br />
will be critical as the company continues<br />
its planned expansion.”<br />
The appointment is one of a number<br />
planned for 2009, continuing DxS’<br />
expansion which has seen the<br />
company’s workforce more than double<br />
in size in the last twelve months with<br />
new manufacturing space taken at their<br />
Manchester site and a global distribution<br />
deal agreed with Roche Molecular<br />
Diagnostics for their TheraScreen®<br />
diagnostic kits.<br />
www.dxsdiagnostics.com<br />
base to fund even the most basic<br />
product improvements," said Glyn<br />
Williams, VP of Product Management.<br />
"We believe that we serve our client<br />
base better by listening and investing<br />
up-front in the solutions we know they<br />
require, then offering real products with<br />
demonstrable value."<br />
A recipient of two Queen’s<br />
Awards for Enterprise, IDBS has also<br />
recently been awarded the Select<br />
Science.Net Scientist’s Choice Award<br />
for its E-WorkBook Suite and a<br />
prestigious Frost & Sullivan award for<br />
the quality and innovation of its<br />
professional services. www.idbs.com<br />
GO TO CONTENTS PAGE 39