Incorporation Of Developability Into Cell Line Selection Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

The pharmaceutical industry is coming under increasing pressure to deliver effective drug products, yet reduce sales prices. Small molecule product pipelines are not delivering effective lead products quick enough, thus, large pharmaceutical companies are investing huge amounts of money in the biotechnology sector. Of these emerging biopharmaceutical products, the majority of most successful products have been monoclonal antibodies, for example Herceptin and Avastin, and thus therapeutic antibody technologies are attracting significant attention and therefore investment. Of most interest has been the apparent ability to construct common manufacturing platforms, from which a range of antibody products could be produced. In this way, development times and costs can be reduced. In addition, domain antibodies have recently been discovered and investigated as a possible future way of augmenting antibody mode of action and pharmacokinetics, allowing the engineering of different antibody activities, as well as possibly being used independently as therapeutic agents.

These antibody products are most often produced as heterologous proteins in, one of a variety of, cell lines at large scales in a bioreactor. However, at these large manufacturing volumes, there is limited opportunity to perform cell culture optimisation and process development due to the high initial capital required and significant running costs at such scales. Therefore, if such a manufacturing process can be 'scaled down', this will provide an opportunity to perform such process development work at much higher throughputs whilst still reducing costs. The major hurdle in such work lies in creating a small scale model that accurately recreates the environment experienced at large scale in the scaled down device such that it is possible to produce relevant data that can subsequently be used to scale up to large scale production. Whilst there are a number of such systems currently on the market, no such device has been shown to accurately predict the large scale process and thus be used in a widespread fashion to screen or optimise mammalian cell culture processes.

A further challenge is the utility of accurate analytical equipment that can be used in conjunction with the scaled down system. Furthermore, it is desirable to determine biological markers that are indicative of productivity to allow for increased accuracy for cell line selection or optimisation. In addition, it would be advantageous to integrate any such scaled down model with robotic handling units, thus enabling far more parallel fermentation processes running in parallel in large scale screening and development experiments.

The aim of this area of research is, in essence, to accelerate product development timelines, and decrease associated costs, and therefore bring potentially life saving drugs to market quicker at a reduced selling price, thus increasing product availability.

1.1 Pharmaceutical Industry

Whilst the pharmaceutical industry market is continuing to increase with time, the rate of growth within the market is declining over that same period, as shown in Figure 1.1. This is as a result of the present failure of the 'big pharma' blockbuster drug business model; product development pipelines are failing to create products to meet the needs of the therapeutic market, most of the 'easier' indications have already been targeted and older product patents are expiring thus generic drugs are subsequently competing for that same market.

Figure 1.1. Graph showing global pharmaceutical industry sales data compared to that of growth in the sector over the past eight years (Raw data current as of March 2009, taken from IMS Health Market Prognosis, as shown in the Appendix Table 1.A)

A recent article from IMS Health indicated three challenges to existing pharmaceutical companies and helps explain the current surge in investment in biotechnology companies; in particular a transition in growth rate, market segment and R&D focus (IMS Health, 21st Century Pharma). Firslty, the global growth rate has decreased; major markets have moderated whilst "pharmerging" markets, e.g. China, Brazil, Mexico, Turkey, have rapidly increased (IMS Health, 21st Century Pharma), thus shifting the indications that pharmaceutical companies need to target. The focus of pharmaceutical markets has shifted dramatically; traditionally being focussed on large numbers of patients within the primary care sector, with treatments for conditions such as infectious diseases, high cholesterol or blood pressure, now shifting towards small niche groups of patients, requiring innovative, molecularly targeted products for more complicated diseases such as cancer or rheumatoid arthritis (IMS Health, 21st Century Pharma). Finally, pharmaceutical companies have relied heavily on the blockbuster drug model, at the middle of the last decade 44.3 percent of growth was attributed to blockbusters, thus leading to a shift in R&D strategy; however, now as blockbuster R&D targets have decreased pharmaceutical companies must expand their product portfolios and redirect R&D budgets (IMS Health, 21st Century Pharma). Thus the decline in growth in the market and recent successes with biologics, in particular antibody based therapeutics, has resulted in a huge investment of traditional 'big pharma' companies in biopharmaceuticals.

1.2 Biopharmaceutical Industry

Pharmaceutical companies are becoming increasingly reliant on biopharmaceuticals, or biologics, product sales as a means of generating revenue, as shown in Figure 1.2. This is largely due to recent successes in the field, hence driving up sales and interest in these products.

Figure 1. 2. Illustration of the increasing influence that biopharmaceutical sales have on total global pharmaceuticals sales (IMS Health; Biogenerics: A Difficult Birth)

Indeed global prescription sales of biotech drugs increased 12.5 percent in 2007 to more than $75 billion, and the global biotech market grew at nearly double the rate of the global pharmaceutical market, which increased only 6.4 percent that same year (IMS Health, Press Release). In 2007, 22 biopharmaceuticals generated sales exceeding $1 billion in 2007, compared with just six products in 2002, and in 2007 biopharmaceuticals represented 25% of the total pharmaceutical drug development pipeline (IMS Health, Press Release). One attractive feature of biotechnology products is the perceived lower associated risk; a recent analysis showed that the success rate of biopharmaceutical medicines had an overall higher success rate over chemically-derived medicines at 30% success rate when compared to 21.5% success (Simoens, 2009); however, it should also be noted that biopharmaceutical products had a lower Phase III clinical trial success rate (Simoens, 2009). In market forecasts for 2010 it has been predicted that 60% of revenue growth of 'big pharma' will be from biopharmaceutical products; with a predicted combined annual growth rate for biopharmaceuticals at 13% in comparison to just 0.9% for small molecule products, and with mAb therapies driving growth and reinvestment in biopharmaceutical products (Scott, 2009).

The demand for mAbs triggered parallel efforts to increase production capacity through construction of large bulk manufacturing plants as well as improvements in cell culture processes to raise product titres (Kelley, 2009), with factors such as fermentation titre and overall yield deemed critical determinants of economic success (Farid, 2007). There is now increased pressure for the cost-effective manufacture of antibodies given the fact that they are administered at high doses and show an increasing market potential (Farid, 2007). Thus it is imperative to establish and implement an efficient small scale cell culture model in a high throughput format that can be used to screen for high producing, high potency, cell lines and rapidly advance them through development

1.3 Cell Culture Processes

The advent of genetic manipulation led to the ability to manufacture heterologous protein products in a variety of cell lines (Rai and Padh, 2001). The choice of cell line largely depends on the product purpose. Different cell lines have advantages over other cell lines as well as different engineering challenges, in terms of the design of the production scale vessel and subsequent purification of the product. Many different cell types have been investigated in the production of heterologous proteins as therapeutic agent (Lee and Lee, 2005, Rai and Padh, 2001). It is paramount that the cell line selected is not only capable of producing the protein properly, i.e. correctly folded with the necessary post-translational modifications required for functionality, but also at sufficient levels of productivity so that the manufacturing costs are not too high. Table 1.3 summarises some of the commonly used cell types along with some brief comparisons between them with regard to heterologous protein production as therapeutic agents.

Table 1.3. Summary of different cell lines. Table adapted from Lee and Lee, 2005 with information from Rai and Padh, 2001.

Cell Type




High growth rate

High cell densities achievable

Inexpensive media

Misfolds large or complex protein products

No protein glycosylation

No post-translational modifications

Must ensure removal of bacterial endotoxins from final product


High growth rate

Correct folding of proteins

Correct protein glycosylation

High culture cost

Currently no manufacturing facilities or regulatory experience for this cell type


High growth rate

High cell densities achievable

Inexpensive media

Certain post translational modifications achievable

Often hypermannosylates the protein product rendering it ineffective


Already used to make pharmaceuticals

Slow growth rate

Expensive to culture


Correct folding of proteins

Correct protein glycosylation

Correct post-translational modification

Expensive media

Slow growth rate

'Shear sensitive' cells due to lack of cell wall

For the production of heterologous proteins as biopharmaceuticals, especially when they are large, mutlidomain glycoproteins, i.e. as in the case for a full length monoclonal antibody (mAb), it is often necessary to use mammalian cells, as they often difficult to fold and require significant amounts of post-translational modifications, which is not always possible in other cell types. One significant development with regard to recombinant glycoprotein production is the work by Glycofi (owned by Merck), who are genetically engineering yeast cells so they might perform the necessary glycosylation post-translational modifications as occurs with human cells (Hamilton and Gerngross, 2007). This group have pioneered the glycoengineering of the yeast Pichia pastoris, which has led to the production of fully humanized sialylated glycoproteins (Hamilton and Gerngross, 2007). As described in table 1.3, as yeast cells have very high growth rates and inexpensive media requirements, by integrating the ability to perform human-like glycosylation protein modifications, this is potentially the one future expression system that could rival mammalian cell culture for the production of recombinant glycoprotein products, in particular mAb therapeutics.

Ideally, recombinant glycoprotein products will have high structural fidelity with the 'natural' product, i.e. with regards to protein folding and post-translational modifications, given that significant alteration of structural fidelity will result in 'altered self' and the potential to be immunogenic, consequently causing anti-therapeutic antibody responses which can impact therapeutic efficacy, i.e. reduced circulation time in blood, and might precipitate adverse reactions (Jefferis, 2009). Thus, whilst screening cells as potential manufacturing scale candidates, it is not only essential to analyse for high productivity in terms of amount of protein expressed, but also in terms of the quality of the protein expressed, a very high titre is offset by the fact that the protein product itself is not therapeutically active or in fact immunogenic. All currently licensed mAb products are manufactured in one of three mammalian cell lines: Chinese hamster ovary (CHO) cells, mouse NSO myeloma cells or mouse Sp2/0 myeloma cells (Jefferis, 2009). There is further evidence that both NSO and Sp2/0 affix additional oligosaccharide residues in some recombinant mAb products, but the CHO cell line does not, and therefore may be preferred over these alternative mammalian cell hosts (Jefferis, 2009)

The industrial scale culture of cells expressing a product of interest will most often occur in a bioreactor, which will attempt to optimise the conditions for the growth of the cells, and expression of the desired product. There are many reactor types that may be employed, but the two most often used are the stirred tank reactor (STR) and airlift reactors (Doran, 1999). An STR is able to adequately mix, even at large volumes, providing homogenous culture conditions for the cells, i.e. uniform temperature, pH, dissolved oxygen, and nutrients. When using 'fragile' mammalian cells, there have been fears that high hydrodynamic shear, especially in the impeller discharge zone, may damage the cells. This led to the development of airlift reactors. Airlift reactors have to be very tall in order to operate effectively, and thus are awkward to install in manufacturing facilities, and while they do not generate particularly efficient mixing they do create high kLa values. However, with the advent of shear protectants, like Pluronic F-68, airlift reactors have largely been made redundant in favour of STR's. In addition some research has suggested that cell death is more dependent on bubble damage than shear as a result of the impeller, as shown in Figure 1.3; energy dissipation released by bubble bursting is in the range of that required for cell damage. The chosen bioreactor may be operated in a variety of fermentation strategies, e.g. batch, fed-batch and perfusion, whereby the aim is to extend the culture length and thus increase the amount of time the cells are at their highest density and producing large amounts of the product of interest (Bailey and Ollis, 1986).

Figure 1.3. Effects of various energy dissipation values on cells (Heath and Kiss, 2007)

The successful manufacture of a heterologous protein in a production scale bioreactor will require significant optimisation before full productivity can be reached. It would be incredibly costly and time consuming to perform all process development and optimisation at large scale, therefore small scale mimics of large scale bioreactors have been developed to allow assessment of the process and optimisation, and subsequent scale up to production conditions.

1.4 mAb's

Within the biopharmaceutical sector there has been a range of highly successful products, however the overriding driver in the sector has been as a result of the success of mAb's (Pavlou and Belsey, 2005). More than 20 recombinant antibody molecules have currently been licensed for therapeutic indications, largely for cancers and chronic diseases (Jefferis, 2009) The following table has been compiled to highlight the number of mAb's that exist within the top 10 biopharmaceuticals in the world market. In addition it is estimated that 30% of new drugs to be licecned in the next decade will be based on antibody products (Jefferis, 2009)

Table 1.4. Table of the top 10 Biologics products as of 2007 (Table created with data from IMS Health, MIDAS with additional product information from La Merie - Top 20 Biologics 2006). rHU - recombinant humanised, EPO - erythropoietin, PEG - Polyethylene glycol, G-CSF - Granulocyte colony-stimulating factor and rec - recombinant.

Top 10 Products

Product Type



Sales (US$ Billions)

% Market Share

% Growth (Constant US$)

Global Biotech Market








TNF Fusion Protein

Amgen/ Wyeth/ Takeda

Autoimmune Diseases





Synthetic EPO


Renal and cancer anaemia





Recombinant Chimeric mAb

Centocor (J&J)/ Schering-Plough

Autoimmune Diseases





Recombinant Chimeric mAb

Genentech/ Biogen/ Roche/

Leukemias and Lymphomas







Neutropenia (Lack of white blood cells)





rHu mAb

Roche/ Genentech

Metastatic Breast Cancer






Johnson and Johnson

Renal and cancer anaemia







Renal and cancer anaemia





rHu mAb

Roche/ Genentech

Metastatic Colorectal Cancer





rec fully Human mAb


Autoimmune Diseases




Total Top 10







As shown in Table 1.4, half of the top 10 products are mAb products, and together these 10 products account for half the total global biotechnology market, thus highlighting the significant impact this group of products has on the biopharmaceutical market. Note that Enbrel is not technically a mAb, but is a fusion protein combining a TNF receptor and an antibody constant region. Antibodies are complex protein molecules, see Figure 1.4.a), which have the ability to bind to foreign organisms and thus mark them for destruction by other agents of the immune system.

In order to combat the vast array of pathogens that an individual may encounter, lymphocytes of the adaptive immune system have evolved to recognise a great range antigens, small protein fragments (Janeway et al, 2005) that might either be on the surface of or within a pathogenic organism. The antigen-recognition molecules of B cells are the immunoglobulins (Ig), which are themselves produced in a vast range of antigen specificities, with each B cell producing an Ig of a single specificity (Janeway et al, 2005). Ig molecules can be membrane bound on the surface of B cells, and are thus known as B-Cell Receptors (BCR's) or secreted from terminally differentiated B cells, or plasma cells, as antibody molecules (Janeway et al, 2005). Antibody molecules have two distinct functions; the first is to bind specifically to antigens of the pathogen, whether it be a surface protein from a virus or a toxic product from a pathogen, and the second function is to recruit other cells or molecules of the immune system to destroy the pathogen once the antibody is bound to it (Janeway et al, 2005). As such the antibody molecule is composed of two different regions, see Figure 1.4.c), one that binds the antigen and the other that brings about a certain response. The variable region, or V-region, is the antibody binding domain, and the great degree of variation that occurs in this part of the protein means that a huge range of antigens may be recognised, and this interaction can be very specific (Janeway et al, 2005). The region that brings about the effector functions is much less varied and is therefore known as the constant region, or C region.

All antibodies are composed from paired heavy and light polypeptide chains, and within these immunoglobulins there are five classes based on their different constant regions: IgM, IgD, IgG, IgA and IgE, with IgG being the most abundant form (Janeway et al, 2005), and therefore most relevant as a therapeutic agent. IgG molecules are approximately 150 KDa and are composed of two different polypeptide chains: the Heavy or H chain is approximately 50 KDa and the Light or L chain is approximately 25 KDa (Janeway et al, 2005). Each IgG molecule is constructed from two light chains and two heavy chains, with disulphide bonds linking the two heavy chains, and each heavy chain to a light chain (Janeway et al, 2005). The IgG molecule is a Y-shape with each variable, antigen binding domain at the ends of the arms of the Y-shape. In this way, the antibody molecule is also capable of cross-linking antigens and binding to them more stably (Janeway et al, 2008).

Jefferis p356:

Explains different function for rMabs

Brekke and Sandlie, 2003 - Therapeutic antibodies for human diseases at the dawn of the twenty-first century

Figure from this article

Figure 1.4. Illustrations of typical features that an antibody possess as represented in the form of an IgG antibody firstly by a) ribbon diagram highlighting how the heavy and light chains interact, b) block diagram showing the structural features of the antibody where C = Constant Region, V = Variable Region, L = Light Chain and H = Heavy Chain and c) simplified block diagram highlighting the key features of an antibody. Images from Janeway et al, 2005

1.4.1 dAb's

Whilst mAb's have proven very successful as therapeutic agents, there has been successive innovation in the field which has led to the development of variations of these Ig molecules, including fragment antibodies, fAb's, or single-chain variable fragments, scFv's (ref). Figure 1.4.1 shows some of the multiple different isoforms from natural Ig molecules that may be generated as therapeutic agents (Enever et al, 2009) which are currently on the market or in development, along with some novel variations often combining multiple components from different Ig molecules and/or sysnthetic components.

Conventional recombinant antibody fragments contain one antigen-binding VH-VL pairing. At 57 kDa, a Fab fragment comprises a VH-CH1 polypeptide disulphide-bonded to a VL-CL polypeptide. At 27 kDa, a scFv fragment contains only the VH domain fused to the VL domain via a polypeptide linker (Holt et al, 2003)

Figure 1.4.1. Schematic representation of different Ig molecules that have been identified as potential therapeutic agents (Enever et al, 2009)

- find another picture

Domain antibodies, or dAb's are of 11-15 kDa and can either exist as an isolated antibody VH domain, as shown in Figure 1.4.2 or as an isolated antibody VL domain (Holt et al, 2003). Each dAb will therefore retain three of the six naturally occurring complementarity determining regions, CDR's, from the original VH-VL pairing, these side chains are highlighted in red in Figure 1.4.2.

Figure 1.4.2. Schemiatic representation of a domain antibody, or dAb (Holt et al, 2003)

Domain antibodies are a particular new and exciting development in antibody technologies as they not only could be used independently as a therapeutic agent but also may be employed as a way of augmenting mAb mode of action and pharmacokinetics. In the former context, a human dAb product is, for example, being investigated for HIV treatment, and it is thought that due to the small molecular weight of the product that the agent may be capable of penetrating into virally infected tissues ( There are further considerations to take with the use of dAb's as independent therapeutic agents, due to their high binding capacity yet small size, these products have been thought to have problems with 'stickiness' (Ref). In terms of using dAb's as mAb modulators, Domantis has recently been granted a patent for AlbudAb which will aim to extend the serum half life of compounds, in particular, antibodies, and thus increase product efficacy (Domantis Patent - ""a=vHYPERLINK ""&HYPERLINK "" ""&HYPERLINK ""hl=enHYPERLINK ""&HYPERLINK ""gl=ukHYPERLINK ""&HYPERLINK ""sig=AHIEtbQLZjqUPMEjONfHR-2qacsHlD_OFA).

1.5 Cell Culture Process Development

Focus on Mammalian Cell - need for improved fermentation titres, etc.

Higher Titre's -- Less USP - less DSP

Fewer Contaminants (e.g. HCP)

Historic view at how cell culture processes have been improved (figure from Principles of fermentation processes notes - fermenter productivity and operating economics - Gary Lye, p8) - note the incredibly long process development timelines - by the time the process is optimised the patent may be coming to the end - need much faster, higher throughput development pathways

Birch and Racher, 2006 'Antibody Production' review:

Hacker et al, 2009

Matasci et al, 2009

Examples of cell lines typically used for mAb/dAb production, e.g. CHO, NS0, hybridomas

- often try to use one cell line for a variety of products as the characteristics become well understood: genetics (transfection/selection), growth profiles, protein production, product purification

Heterologous protein production

- integrate gene into chromosome / transient gene expression

- methods of gene integration - electroporation, plasmids, viral vectors, etc.

- methods used to select for transfectants - antibiotic resistance, ability to produce a metabolite (glutamine synthetase/dhfr)

- grow cell up to high cell densities and obtain high productivity

- any one area can be improved to help improve titres

Key is that in the past processes were developed progressively (again refer to figure from notes) - need for high throughput systems to investigate and optimise all process choices much quicker and therefore acquire an optimised manufacturing process much earlier in the product lifetime.

Cell culture research usually focussed on increasing yield and decreasing costs

Increasing yield traditionally by increasing life span of culture - usually by replenishing media as cell viability decreases

Wooley and Al-Rubeai, 2009 - The isolation and identification of a Secreted Biomarker Associated with Cell Stress in Serum-Free CHO Cell Culture

1.6 Commercial Drivers to Accelerate Product/Process Development

Herceptin, a mAb therapeutic agent used in the treatment of metastatic breast cancers, costs in the region of £60,000 per year per patient (Waltz, 2005). Whilst there is increasing pressure from government and health bodies to drop sales prices, manufacturers argue they must recoup their losses from expensive R&D procedures, on average it costs $1.2 billion dollars and takes approximately 10 years to develop a new biopharmaceutical (Tufts CSDD, Press Release). Furthermore, referring to Figure 1.6.1, the 10 pharmaceutical companies spend, on average, approximately a quarter of all their revenue on R&D. Therefore, to meet such pressure yet remain strong economically, companies must find a way to accelerate their R&D programs and increase product pipeline throughput. To highlight this factor, even looking at the top 10 biologics (2003-2008 LaMerie), Figure 1.6.2, it takes, on average, around two years before development costs are recouped, assuming an average development cost of a biologic product of $1.2 billion. Should selling prices need to fall it would take even longer, and might even make some of these products non-viable. All these factors are driving for increased throughput and pace in the R&D process, without losing quality of data that might risk product safety and therefore patient well-being.

Historically, R&D has proceeded via incremental process improvements, however it will become crucial to attempt to hit much higher productivity levels much earlier on in the R&D process. Such a development may be achieved by the scaling down of the manufacturing vessel that is used to grow the cell line producing the heterologous protein product. In doing so, cells can be grown, transfected, screened and selected in an environment that represents that in which they will be cultured at the manufacturing scale. This will allow for much more accurate selection of high producing clonal lines and further accuracy in screening cells in optimisation studies for media development or growth condition studies.

Figure 1.6.1. Graph to highlight the significant investment required by pharmaceutical companies for ongoing R&D projects. Information from 2008, published October 2009 from Scrip's Pharmaceutical Company League Tables - 2009, raw data in Appendix Table 1.B

Figure 1.6.2. Graph to highlight the significant investment required to develop a new biologic product, and the resulting success that it can achieve shown by the top selling biologics product 2003-2008 Figure taken from Purvis, 2009, which is based on data from La Merie, Top 20 Biologics 2008.

1.6.1 Technologies for Accelerating Upstream Process Development

Creation of small scale models to allow for faster development: media, cell line, DSP, etc., both prior to product launch and parallel to manufacturing

Better for company - economics/faster to clinical trials, etc.

Better for patient - faster drugs to market, cheaper drugs (companies spent less in dev.), increase availability

Scale down

HT - faster development time

less material - cheaper

Really provides excellent tool for cell line selection/process development - however, there are a unique set of problems that are faced at this scale (e.g. evaporation, etc.)

Necessary to recreate conditions of the large scale equipment when scaling - e.g. centrifugation studies at UCL - feed zone of large scale centrifuges - high shear zone- not present in scaled down (bench top) centrifuges - therefore use 'shear device' to artificially recreate the large scale shearing forces exerted in the centrifuge feed zone before performing the scaled down centrifuge operation itself.

Focus on fermentation and discovery/development of cell line for product

Need to evaluate needs and limitations at the large, production scale, for example is shear going to be an issue, and this will form a critical aspect of the scale down principle adopted, i.e. scale down with equivalent tip speed

Typical scale criteria for fermentation systems:

Handbook of industrial mixing: science and practice By Edward L. Paul, Victor A. Atiemo-Obeng, Suzanne M. Kresta, 2003

Encyclopedia of Bioprocess Technology - Fermentation, Biocatalysis, and Bioseparation, Edited by Flickinger 2003

Equivalent P/V

Maintaining geometric similarity

Equal impeller tip speeds

Constant mixing times

Maintain kLa

Xing et al, 2009 Scale up analysis for a CHO cell culture process

Geometric similarity, constant kLa, and constant specific impeller pump rate (Qs)

Geometric similarity, constant kLa, and constant maximum shear (impeller tip speed, Vtip)

Constant kLa, constant impeller tip speed, and constant Qs

Give examples from literature of some cases where a fermentation process has been scaled up/down, the parameter(s) adopted and the results.

Hewitt and Nienow, 2007: The scale-up of microbial batch and fed-batch fermentation processes

Kensy, F., Zang, E., Faulhammer, C., Tan, R.-K., Büchs, J. Validation of a high-throughput fermentation system based on online monitoring of biomass and fluorescence in continuously shaken microtiter plates (2009)

Gill, N.K., Appleton, M., Baganz, F., Lye, G.J. Design and characterisation of a miniature stirred bioreactor system for parallel microbial fermentations (2008)

1.6.2 Overview of High Throughput Systems (HTS) for Cell Culture

Pilot Plant Scale

Mini Bioreactors

Shake Flasks

Microwell Plates

Microfluidic Bioreactors

Trade off between degree of info that can be obtained from each run and no. of fermentations that can be run in parallel - Figure from Doig et al, 2006

Examples of different commercial HTS's for fermentation: SimCell, m24, Sixfors, TubeSpin bioreactors, etc. comparison between them online monitoring capabilities, sparging, volume, etc. like figure from Betts and Baganz, 2006 - Miniature bioreactors: current practices and future opportunities

Figure 1.6.3. Trade off between experimental throughput and the information that can be obtained from each fermentation (Doig et al, 2006)

Decreasing scale:

ADV - increase no. of parallel runs, therefore decrease time required decrease amount of process materials required therefore decrease cost

DIS - decrease ease of online measurement (traditional probes cannot be used) and harder to take samples for offline analysis (not as much volume), harder to recreate the engineering environment of the largest scale bioreactor with decreasing scale

1.7 HTS for mammalian cell culture

à Mention limited number of papers on the topic, the lack of details regarding how scale down was actually performed, i.e. on what principle is scale down achieved and include the table summarising the different models

Throughout, there is a lack of definition to the scale down parameters used. Often see figures where, for example, matching growth profiles have been generated, however, there is no mention of how this has been achieved and whether the culture conditions are even the same for different factors investigated. More work is required in this area to create a truly representative scaled down system that can easily be used in high throughput screening experiments, yet compatible with analytical tools and have the option for automated robotic handling integration.

Cell Line

Reactor Type

Scale Up Analysis



CHO (expressing recombinant protein)

m24 24 Deep Well MTP

pH, DO + Temp sensors

Thermal heat conductor

0.2 mm sparge membrane

m24 (5mL WV) and 3L (2L WV) Bioreactor

Scale up criteria not specified

Matched parameters:




Offline pH







Problems with foaming

Uneven gas distribution

Chen et al, 2008 (Genentech)

SP 2/0 Mouse Myeloma

3L (Rushton, no baffles) w/ 75L, 300L, 2500L (2 x pitched blade)

Not geometrically equivalent

Step-wise Prediction, Constant:

impeller tip speed,

shear rate at tip,

overall circulation time

mixing time

Temp, pH + DO are volume independent \keep same

Use predictions to make accurate estimate of conditions where variables will be similar (if not maintained exactly w/ scale) - i.e. 1 and 4 impracticable at largest scale

End up scaling with constant rpm

Yang et al, 2007 (Immunomedics)

CHO (Antibody-fusion protein)

5000L (3 x marine) scaled from 5 and 20L bioreactors

- Geometrically similar

Address particular issues at very large scale:

Bulk Mixing

Oxygen Transfer

CO2Removal Rate

Developed equations for CHO cell line:





Observed pH and DO gradients - mixing time more dependent (3-fold) on culture volume than

effort to reduce overall feeding volume

kLa more dependent on air flow rate than P/V

Current configuration only capable of supporting 7 x 106cells ml-1

Increased headspace airflow rate 200% - no effect

- bioreactor configuration may need to be changed to allow higher sparge flow rates

Xing et al, 2009 (Bristol-Myers Squibb)


a) Volumetric mass transfer coefficient (kLa) determination from 50mL (20mL WV) to 2000L (1000L WV)

b) CFD Analysis of free-surface from 50mL (20mL WV) to 30L (13.4L)

a) Correlated kLa values against shaking speed for a range of vessel volumes

b) Matched experimentally and by CFD simulation the free-surface shapes experienced at different volumes and shaking speeds.

Increase in kLa at higher shaking speeds was mainly due to an increased kLvalue - highlights dominant effect of free-surface turbulence on gas transfer in orbitally shaken bioreactors

- Feasibility of orbital shaken tech for mammalian cell culture up to 1000L (WV)

Zhang et al, 2009



Evaluating the use of energy dissipation rate (e)values to quantify shear in a range of different bioprocessing environments: pipe, annulus, parallel plates, rectangular channel, contracting flow (pipette) and bioreactor

2L vessel (1.6L WV) used for Particle Tracking Velocimetry (PTV) to determine the distribution of ewithin the impeller discharge zone

For a given impeller type, emaxin the impeller region can be approximated using a non-dimensional constant and to a specific non-dimensional location relative to the impeller:

If specific geometric ratios are maintained - E doesn't alter significantly over a large range of impeller speeds (for either RT or PBT)

Energy dissipation rate is always high in the impeller discharge stream

- of mechanical energy added to the vessel, 43.5% is dissipated in the impeller region for RT and 70.5% for PBT

This work fails to take into account the effect of gas-liquid interfaces on the energy dissipation rates

Mollet et al, 2004


Put contents of table into text

Further examples of mammalian cell scale up/down work (i.e. Legmann et al, 2009 - A predictive high-throughput scale-down model of monoclonal antibody production in CHO cells, Prokop et al, 2009 - NanoLiterBioReactor: Long-Term Mammalian Cell Culture at Nanofabricated Scale, Zhang et al, 2010 - Use of orbital shaken disposable bioreactors for Mammalian cell cultures from the milliliter-scale to the 1,000-liter scale.)

Girard et al, 2001 and Strobel et al, 2001 from Barret et al, 2009 - Microwell Engineering Characterisation for Mammalian Cell Culture Process Development - uses matched average energy dissipation rates (P/V)

Crucial to create a small scale model that accurately predicts the large scale equipment - evaluate the use of different scale up criteria that might be employed

Nick, UCL - matched mixing time

Barrett - engineering characterisation of microwell plates

Sara - Matched hydrodynamic shear for scale up

Highlight common problems (e.g. evaporation issue at small scale leading to problems with osmolality, increasing significance of certain forces (e.g. surface tension) with decreasing volume)

Compare to different set of challenges faced at very large scale Nienow

Nienow, 2006 - Very large scale challenges - mixing time (non-homogeneity and concentration gradients) and CO2 removal

thus often useful to have an intermediary scale up, i.e. HTS to pilot scale to production scale, would be very difficult to directly scale from miniature bioreactor to production scale, i.e. 1000L wv or greater

1.7.1 HTS Analytical Tools

An inherent problem working with small scale cell culture devices is that there is less volume for traditionally designed probes to fit into and there is a further limitation on the amount of sample that can be removed for offline analysis. Therefore, to fully utilises a small scale device effectively, alternative analytical methods must be employed hat allow for online monitoring without affecting the culture environment.

e.g. ways of measuring cell growth, pH, DO, CO2 etc. - online/offline

As with a traditional bioreactor the will then be the option to respond to any online measurements and influence the culture environment. However, this will be limited by the device chosen, for example, if you want to change the agitation speed for a single well of a microwell plate, this cannot be done independently of the rest of the plate.

Alternative is not to monitor online - may prefer to set up a standard cell culture fermentation in the desired HTS device and then screen all clonal lines at a given protocol, analyse the results and then send the lead producers onto a subsequent round of screening in devices with online monitoring and control.

Examples of companies that provide analytical equipment and the sorts of devices they are compatible with, e.g. Presens

Analytical probes must not be too expensive, i.e. single use microwell plates with integrated analytical probes, either probes can be taken out and re-used or they must be very cheap

Identification of protein biomarkers for bioprocess applications, e.g. Wooley and Al-Rubeai, 2009: The Isolation and Identification of a Secreted Biomarker Associated With Cell Stress in Serum-Free CHO Cell Culture

Chaperones, UPR (XBP1) eIF2a

Product Quality

Product Glycosylation - Jefferis p360 (high productivity might compromise post-translational modification machinery and result in low product quality - need to screen at initial cell line selection

Protein aggregation

Markers (Biomarkers - genotype/phenotype)



Cell Size

Membrane Integrity

Cell cycle markers


Brooks, 2009: Strategies for Analysis of the Glycosylation of Proteins: Current Status and Future Perspectives

Henriques, et al, 2009 - Monitoring Mammalian Cell Cultivations for Monoclonal Antibody Production Using Near-Infrared Spectroscopy

1.7.2 HTS and Automation

Automated systems can be invaluable both in cell culture steps and for analytical stages. As well as increasing speed, they will also be more accurate and consistent in comparison to a human performing the same manipulation. With regard to implementing a HTS in a process optimisation scenario, without automated capacity costs will increase rapidly on increased staff requirements, and also human error will most likely increase with greater number of operations, particularly when dealing with small, intricate devices. In comparison, with automation, personnel numbers can be reduced, thus reducing development costs, but also frees up personnel for more complex tasks such as data analysis and interpretation. The key to making full use of HT cell culture systems will be if the culture is automated and can be integrated with HT automated analytical tools to screen all runs and collate data, thus while increasing speed and throughput, precision and reliability do not deteriorate.

There are a variety of systems commercially available that could be employed to assist in HT automation of cell culture operation and analysis. Liquid handling robotic platforms, e.g. Tecan (Tecan website), might be employed for cell culture, feed additions during culture and sample removal for analysis in a Microwell plate format system. Similarly, The Automation Partnership (TAP) offer a variety of automated cell culture robotic units, varying from the SelecT for T-175 flasks, the Cellmate for roller bottle culture to the new advanced microbioreactor (ambr) that cultures cells in a 24 well microplate format (10-15mL) (TAP website). MicroReactor Technologies offer a similar product to the ambr (TAP) called the m24 microbioractor, again operating in a 24 well microplate format (1-7mL working volume) (MicroReactor Technologies website).

1.8 Critical Evaluation of Literature

One of the main issues with the current literature regarding the scale down of mammalian cell culture has been the lack of scaling down with regard to scientific and engineering reasoning. The key failure is to critically evaluate the scale down environment, then devise a suitable scale down parameter that may then be used to establish high throughput screening experiments in an environment that accurately replicates the production scale equipment. On the other hand, there have been many studies that show multiple graphs of seemingly matched growth kinetics without any given explanation as to how this has been achieved, and on what basis the authors believe this method to be successful in achieving such scaled down culture.

In this area of research, I believe that the successful culture of mammalian cells in a small scale format can only be gainfully applied if the culture format has firstly been thoroughly characterised both experimentally and theoretically. Given the understanding of the device in terms of mixing, oxygen transfer, etc., suitable scaling criteria may then be implemented in order to reproduce culturing cells in the small scale format as they grow in the large scale device. The overall aim will be to integrate a small scale high throughput culture format that accurately predicts the production scale culture environment, using, preferably online, analytical tools that can detect the most significant cell culture factors in an automated fashion for ongoing culture control and analysis.

1.9 Aim and Objectives

1.9.1 Aim

Establish small scale culture of GSK CHO DG44 (dhfr -) cells expressing a mAb product using the m24 (MicroReactor Technologies, owned by Pall Life Sciences) microbioreactor, to enable the rapid selection of robust and scaleable cell lines following thorough engineering characterisation of the chosen small scale culture format.

1.9.2 Objectives

Experimental characterisation of the m24 cell culture format in terms of mixing, oxygen transfer, energy dissipation or hydrodynamic shear.

Theoretical characterisation of the m24 cell culture format using Computational Fluid Dynamics (CFD).

Develop cell culture model to accurately mimic production scale process operation and culture performance thus achieving scale down from 2L (wv) bioreactor, itself representative of GSK's 1200L wv production scale bioreactor, to the m24 format and achieve comparable growth kinetics in terms of matched viable cell density, cell viability, antibody productivity and metabolite levels.

Identify phenotypic and genotypic 'biomarkers' that can be used for rapid cell line selection, and in particular utilise SELDI-TOF mass spectrometry.

Use of flow cytometry to assay cultured cells in terms of cell cycle distribution and apoptosis and identification of appropriate high throughput assays to determine product quality.











Total World Market (Current US$ in Billions)









Growth Over Previous Year ($Constant US$ Growth (%))









Table 1.A. All information current March, 2009 (IMS Health Market Prognosis)


Total World Market Sales (US$ Millions)

R&D Expenditure (US$ Millions)

R&D Expenditure as a Percentage of Total Sales (%)

























Johnson &Johnson




Merck &Co








Eli Lilly




Table 1.B. Scrip's Pharmaceutical Company League Tables - 2009