11.01.2015 Views

Real time PCR - European Pharmaceutical Review

Real time PCR - European Pharmaceutical Review

Real time PCR - European Pharmaceutical Review

SHOW MORE
SHOW LESS

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

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

Saved successfully!

Ooh no, something went wrong!