The Evolution of Protein Expression and Cell Culture

Published on: 
BioPharm International, BioPharm International-10-01-2007, Volume 20, Issue 10

It is commonly believed that technologies in the next 10–15 years will enable sequencing an individualized human genome for less than $1,000. With innovations like these, the twenty-first century will certainly belong to biotechnology. From an industrial standpoint, the discovery of therapeutic molecules and the development of cell lines and processes to produce these molecules will be of paramount importance. This article describes various approaches that have been prevalent in the industry or are likely to be used in the future for generating cell lines with desirable traits and developing high titer cell culture processes.


The biopharmaceutical industry has seen an extraordinary evolution over the last 30 years since the first successes of recombinant DNA technology. Today, it represents the highest growth rate sector for the pharmaceutical industry. Developments in protein expression systems and cell culture technologies, along with major advances in analytical characterization capabilities, are at the core of the biopharmaceutical industry's rapid growth. This article presents an overview of how protein expression and cell culture have evolved over the last 20 years and highlights the emerging areas for development in each of these fields.

It is commonly believed that technologies in the next 10–15 years will enable sequencing an individualized human genome for less than $1,000. With innovations like these, the twenty-first century will certainly belong to biotechnology. From an industrial standpoint, the discovery of therapeutic molecules and the development of cell lines and processes to produce these molecules will be of paramount importance. This article describes various approaches that have been prevalent in the industry or are likely to be used in the future for generating cell lines with desirable traits and developing high titer cell culture processes.


The factors that dictate the choice of a cell line include the expression level, economic considerations, and desired product quality for a given biological function. Because of the ease and the speed with which E. coli processes could be scaled up, this bacterium was a dominant host until the mid 1990s. Fermentation systems for E. coli with capacities in the range of 40,000 L were operational in the 1980s. As the need for producing properly folded protein molecules and antibodies with appropriate glycosylation patterns grew, the industry shifted toward the use of eukaryotes, particularly mammalian cells. The processes developed in the 1980s using mammalian cells often resulted in lower product yields. However, product yields of 1–2 g/L are currently commonplace, and industrial claims of product titers as high as 10 g/L have been recently made. The two cell lines that are primarily used for recombinant protein production in the biotechnology industry are Chinese hamster ovary (CHO) and mouse myeloma (NS0) cells. At the same time, extensive use of and research in other expression systems, including bacteria,1 yeasts,2 insect cells,3 plants,4 human cells,5 and transgenic animals,6 are underway. Some notable applications that have emerged in the recent past include the glycoengineered yeast strains for controlled glycosylation,7 the use of duckweeds for producing protein therapeutics,8 and the cell-free production of recombinant proteins.9 Because the majority of investigational and clinical biotechnology products are currently being produced in mammalian cells, this article focuses on the use of mammalian cells to produce protein and antibody therapeutics.


Expression vectors commonly use a strong viral or cellular promoter and the gene of interest is generally isolated as a cDNA without introns. Because splicing is known to affect the cytoplasmic transport and translation of mRNA,10 most expression vectors will also include at least one intron sequence, usually located between the promoter and the cDNA coding sequence. The gene expression may also be improved by using the more abundant tRNA codons.11

It is known that the transcription of the inserted recombinant gene in the mammalian cell genome often can be influenced by the site of integration of the gene of interest in the mammalian cell genome. This may in turn affect the specific productivity and clonal stability of the resulting cell line. The incorporation of the recombinant gene in the host genome is a random process and several approaches have been demonstrated to either locally modify the chromatin structure or target a specific site of integration. Several protective cis-regulatory elements can be used to locally modify the chromatin structure used, e.g., insulators,12 locus control regions,13 scaffold or matrix attachment regions,14 ubiquitous chromatin opening elements,15 conserved antirepressor elements,16 and butyrate.17

The integration of the recombinant gene in the host genome is random, which makes it an inefficient process (approximately 1 in 10,000 events results in a successful recombination) and the chromosomal site of insertion and the copy number cannot be predicted.18 To overcome the problems associated with randomness of recombinant gene insertion in the host genome, the use of episomal vectors has been demonstrated as a viable strategy for recombinant protein production.19 However, the most notable development has been the demonstration of site-specific homologous recombination, which can be achieved by using recombinase enzymes (e.g., bacteriophage P1 Cre or yeast Flp enzymes). These enzymes exchange DNA between the transfected plasmid and the host genome at the sites thought to be relevant for high expression.20


A number of reports are available that demonstrate the use of genetic engineering to improve the growth and product formation potential of the host cell. The integration of several anti-apoptotic genes including E1B-19K and Aven (in CHOcells) have been reported to improve the performance of mammalian cell culture.21 The other examples include the over expression of cell cycle regulatory proteins cyclin E and E2F-1 independently to activate the proliferation of CHO cells in the absence of serum and external growth factors.22 The co-expression of cell proliferation genes and anti-apoptotic genes has also been reported to improve cell culture performance.23 Targeted approaches to affect the cell adhesion,24 antibody secretion,25 and post-translational modifications of antibodies26 have also been demonstrated.


The production scheme of generating a clonal population to produce the protein of interest is well established. The host cell is transfected with the plasmids containing the recombinant gene with the necessary regulatory elements and the gene conferring the selection pressure. The most popular selector genes are dihydrofolate reductase (DHFR) and glutamine synthetase (GS). For these systems, the expression of recombinant proteins can be significantly increased by exposing cells to gradual increases in concentrations of methotrexate (MTX) or methionine sulfoximine (MSX). The surviving cells will frequently contain significantly higher copies of the integrated plasmids in host chromosomes and will produce more protein compared to unamplified cells as a result. However, the specific productivity will vary among clones and the identification of high producers will require the screening of hundreds or thousands of cell lines.

Table 1 depicts the evolution of the productivity of mammalian cell lines over the last 20 years as a result of cell engineering, media optimization, and bioreactor process development efforts.

Table 1. Productivity increases over the last 20 years in mammalian cell culture



The survivor cells after selection are often subjected to serial dilution in secondary containers like 96-well plates, and the resulting cells that show as single colonies subsequently are expanded in larger containers like 12-or 24-well plates. The selection of the final clone often takes a long time because the clonal stability is as important as the protein/antibody titer achieved with a given clone. The other traits of interest may be the cell sensitivity to osmolality and pCO2, as the cells may have to counteract high osmotic pressure and elevated dissolved carbon dioxide levels at large scale. To speed up the process of finding the right clone, a number of biotechnology companies have invested in highly automated, commercially available systems. These systems combine image processing and highly mechanized robotics to identify and pick the mammalian colonies of interest. The other technique that has gained popularity for single-cell cloning is the cell sorting based on florescence activated cell sorting (FACS). After the single-cell clones have been isolated, the identification of high producers still remains a time consuming process. The feasibility of using FACS in conjunction with the use of a nonfluorescent reporter protein molecule that can be detected by a fluorescent antibody also has been demonstrated to screen the high producers from the sorted single-cell clones in 96-well plates.27 Although the subject of single-cell cloning has generated a lot of academic interest, the fundamental mechanisms that dictate clonal productivity and stability are still unknown.

With the advent of the transcriptomics and proteomics era, researchers are attempting to understand the fundamental mechanisms underlying the differences between the cell lines producing low and high amounts of the protein of interest. Seth et al. used the DNA microarrays and two-dimensional gel electrophoresis (2DE) for eleven GS-NS0 cell lines to show that major functional class changes between low and high producers are in protein synthesis and cell death/growth.28 A recent report showed that there is a positive correlation between the specific productivity and the ratio of heavy- to light-chain mRNA expression of an antibody produced by a GS-NS0 cell line. The report also showed that the use of Northern and Southern analyses can highlight the presence of any abnormal mRNA species (in terms of molecular size) and can also indicate if the functional integrity of the site of insertion of the transgene has been compromised.29 The set of information generated from the above mentioned techniques can serve as useful markers to identify the overproducers among various clones early in the clone selection phase and there is no doubt that such techniques will be used to a greater extent in the coming years. In addition to understanding the mechanisms of cell productivity, it is equally important to understand how the stability of gene expression is conferred to a clone. This area of protein expression is also not well understood and there is a clear need to determine the molecular markers that can help identify stable cell lines early in the project cycle. There is also a need to enhance the understanding, at the chromatin level, of the specific regions of the genome that confer genetic stability to a transgene as well as the transgene integration approaches that can specifically target these regions.


After a clone is selected, the next task is to develop the growth medium. Simple, defined media for bacterial systems have been developed since the initial days of biotechnology manufacturing. Such media contain a carbon source (e.g., glucose, glycerol), a nitrogen source, a phosphorous source, mineral salts, and buffering components. Complex media, using yeast extract, casein hydrolysate, etc., are also common. Compared to bacterial media, the cell culture medium is rather complex. In the past, fetal bovine serum (FBS) was an essential component for the propagation of mammalian cells. Many industrial mammalian cell growth media developed in the 1990s did not use serum. They did, however, include multiple serum fractions containing hormones, growth factors, lymphokines, cytokines, transport proteins, attachment proteins, serum albumin, and lipid supplements. Such a medium that is not supplemented with serum but includes the discrete protein and bulk protein fractions is commonly known as serum free medium. Due to bovine spongiform encephalopathy (BSE) concerns, there is a strong motivation to develop processes without any serum derivative. Such processes, often called animal-protein-free processes are currently the most prevalent and often form the bases of current platforms being developed by many companies. The media for these processes may contain complex ingredients like hydrolysates. The latest industrial development has been to create formulations that do not depend upon the inclusion of any complex raw materials. Such completely defined media have been successfully developed.30 It remains to be seen, however, if the industry will universally adopt the use of completely defined media, because the complex media ingredients are often believed to be the key determinants of productivity.

The cell culture medium typically contains glucose, amino acids, vitamins, bulk ions, lipids, phospholipid precursors, nucleotides, buffering components, protective agents like Pluronic surfactants, antioxidants, and reducing agents. For serum-free formulations, complex ingredients may include various serum supplements and protein molecules like insulin and transferrin. For animal-protein-free and completely defined media, water insoluble components like cholesterol are delivered to cells by carrier molecules like cyclodextrin.31 Scientists now have been able to successfully grow NS0 cells, which were originally thought to be cholesterol auxotrophs, without cholesterol.32 It also has been shown that the cholesterol dependence of cholesterol dependent NS0 cells is due to epigenetic silencing because of the methylation of the CpG-rich region upstream of the transcription start site of the gene Hsd17b7 in NS0 cells.33

The tedious task of medium development is typically accomplished in shake flasks and small-scale reactors. Two- to three- liter bioreactors are most commonly used for this purpose. The key to speedy medium development is the use of statistical designs that help screen and optimize the concentrations of critical components. For this reason, it is highly desirable to use a high throughput system. The implementation of such systems for mammalian cells has been successfully demonstrated.34 These developments are occurring in conjunction with promising developments in the area of high throughput systems for analytical measurements, e.g., systems for measuring antibody yield in 96-well format.

Although limited, some reports in the literature describe how microarray analysis can help assist with the medium design. Allison et al. designed a serum-free formulation for growing the normal human fibroblast, WI-38, which normally requires the presence of FBS in the growth medium. The microarray analysis for the cultures grown in serum-containing medium indicated the expression of 17 receptors. The inclusion of ligands of four of these 17 receptors resulted in a formulation that could support the growth of the cells without serum.35 The future belongs to the combined approaches of genomics, proteomics, and metabolomics for the design of media for mammalian cells. Such an approach, where the combined analysis of transcriptomic, proteomic, and protein interaction data was used, has been explored to better understand the galactose metabolism of S. cerevisae.36


The demand of therapeutic proteins and antibodies will keep increasing the scale at which these molecules have to be manufactured. A number of industrial biologics are currently manufactured by cell culture at the 10,000–20,000 L scale. The scale-up of cell culture processes still remains poorly understood, however, and it often leads to altered yields and altered product quality attributes. Although a number of empirical approaches are used for scale-up, the exact scale-up criterion is still unknown. The scale-up of processes in the past often was constrained because the mammalian cells were thought to be highly sensitive to shear. The notion of shear sensitivity of mammalian cells, particularly the ones used in industry, is slowly fading away and this certainly gives scientists more flexibility in scale-up. With no gas–liquid interfaces present, the local energy dissipation rate that animal cells would encounter in typical bioprocessing situations typically is orders of magnitude lower than the rate that has been experimentally demonstrated to catastrophically damage mammalian cells.37 Nonetheless, the non-lethal effects to cells of high energy dissipation rates still need to be fully established.

The likely reasons for imprecise scale-up and scale-down may lie in either raw material differences or dissimilar bioreactor hydrodynamics at different scales. Variation caused by raw materials might result from differences in manufacturing processes used to produce the raw materials for large scale, lot-to-lot variability, trace impurities present in the raw materials, or media preparation issues at large scale. Understanding variation caused by raw materials is challenging because of limited analytical capabilities, resources, and knowledge of process sensitivity to different analytes. Equally challenging is to understand how different fluid microenvironments affect cellular behavior. Computational fluid dynamics is an exciting technology that can be effectively used to understand bioreactor hydrodynamics.38 It should be borne in mind, however, that such technology must be used along with experimental data generated at different scales or in different bioreactor configurations at a given scale. The combination of computational and experimental data holds the key to precisely define a scale-up criterion.

Over the past decade, the use of disposable technologies has become an integral part of biotechnological processes. Currently, disposable bioreactors with volumes up to 1,000 L of cell culture are commercially available.39 Disposable bioreactors with small volumes (up to 20 L) are particularly popular because they provide an alternative to using multiple shake flasks or spinner flasks for cell expansion. The perceived benefits of using disposable (single-use) technologies are listed in Table 2. The industry also has seen tremendous improvements in available technologies for cell growth and viability measurements. Automatic instruments for cell counting and viability measurements are now commonplace, and the next phase will likely involve the implementation of online probes for measuring viable cell density for commercial processes. The inclusion of viable cell density probes is important because multiple decisions, including the sequential transfer of cultures in a bioreactor scale-up train and the feeds addition in a production bioreactor, often are dependent on the viable cell density of the culture.

Table 2. Rationale for using disposable systems


As the number of commercial products in the biotechnology industry continues to grow, so does the emphasis on maintaining consistent process and product quality. The control of raw material consistency is critical and is often challenging in the case of complex raw materials. For complex materials like fetal bovine serum, serum fractions, or hydrolysates, it may be hard to pinpoint the reason an excursion is observed. This is because of a lack of sufficient understanding of the specific components present in these complex raw materials and how they affect various cellular activities.

The operation of commercial biopharmaceutical processes generates large amounts of online and offline data that are often underused. The management of these data is challenging; equally challenging is strategizing how best to use the data. The use of multivariate analysis as a methodology has gained popularity recently.40 The technique is capable of incorporating raw material as well as process and product data and is useful in identifying clusters and the state of control of a process. Most importantly, the analysis can be used to identify the potential levers that can be used to troubleshoot or improve a process. When combined with analytical and experimental data, the technique can also be used as a raw material control strategy.

Scale-down is an equally important area, because using a good laboratory-scale model is an easy way to troubleshoot and improve an existing commercial-scale process. An ideal scale-down model will result in cell culture performance similar to that observed at large scale, not only in terms of growth, viability, and titer, but also in consistent metabolism and product quality attributes. Scale-down is an uncharted territory, however, and is generally not a straightforward task. As is the case for understanding scale-up, combined analyses using computational, genomic, and proteomic approaches hold the key to thoroughly understanding the issues related to scale-down.


There has been enormous growth in the areas of protein expression, and cell culture over the last 30 years. Although bacterial systems are rather established, cell culture still remains poorly understood. Table 3 summarizes the historical practices, current and future trends, and expectations for the cell culture processes used in the biotechnology industry. With the introduction of newer methods and technologies targeted toward understanding the fundamental mechanisms of cell physiology and metabolism, scientists are gradually inching toward a time when they will be able to maneuver cell culture to their needs rather than letting the cells dictate their next path forward.

Table 3. Historical practices and future trends in protein expression and cell culture

Antonio R. Moreira, PhD, is vice provost for academic affairs and a professor of chemical and biochemical engineering at the University of Maryland, Baltimore County, Baltimore, MD, 410.455.6576,


1. Graumann K, Premstaller A. Manufacturing of recombinant therapeutic proteins in microbial systems. Biotechnol J. 2006;1(2):164–186.

2. Gasser B, Mattanovich D. Antibody production with yeasts and filamentous fungi: on the road to large scale? Biotechnol Lett. 2007;29(2):201–212.

3. Safdar A, Cox MM. Baculovirus-expressed influenza vaccine. A novel technology for safe and expeditious vaccine production for human use. Expert Opin Investig Drugs. 2007;16(7):927–934.

4. Boehm R. Bioproduction of therapeutic proteins in the 21st century and the role of plants and plant cells as production platforms. Ann N Y Acad Sci. 2007;1102:121–134.

5. Bout A, Hateboer G, Verhulst KC, Uytdehaag AG, Schouten GJ, inventors; Holland C, assignee. Recombinant protein production in a human cell. US patent 10,234,007. 2006 Nov.

6. Salamone D, Baranao L, Santos C, Bussmann L, Artuso J, Werning C, et al. High level expression of bioactive recombinant human growth hormone in the milk of a cloned transgenic cow. J Biotechnol. 2006;124(2):469–472.

7. Hamilton SR, Davidson RC, Sethuraman N, Nett JH, Jiang Y, Rios S, et al. Humanization of yeast to produce complex terminally sialylated glycoproteins. Science. 2006;313(5792):1441–1443.

8. Stomp AM. The duckweeds: a valuable plant for biomanufacturing. Biotechnol Ann Rev. 2005;11:69–99.

9. Calhoun KA, Swartz JR. Energizing cell-free protein synthesis with glucose metabolism. Biotechnol Bioeng. 2005;90(5):606–613.

10. Le Hir H, Nott A, Moore MJ. How introns influence and enhance eukaryotic gene expression. Trends Biochem Sci. 2003;28(4):215–220.

11. Barrett JW, Sun Y, Nazarian SH, Belsito TA, Brunetti CR, McFadden G. Optimization of codon usage of poxvirus genes allows for improved transient expression in mammalian cells. Virus Genes. 2006 Aug;33(1):15–26.

12. Pikaart MJ, Recillas-Targa F, Felsenfeld G. Loss of transcriptional activity of a transgene is accompanied by DNA methylation and histone deacetylation and is prevented by insulators. Genes Dev. 1998 Sep 15;12(18):2852–62.

13. Li Q, Harju S, Peterson KR. Locus control regions: Locus control regions: coming of age at a decade plus. Trends Gene. 1999 Oct; 15(10):403–8.

14. Chattopadhyay S, Pavithra L. MARs and MARBPs: key modulators of gene regulation and disease manifestation. Subcell Biochem. 2007;41:213–30.

15. Benton T, Chen T, McEntee M, Fox B, King D, Crombie R, et al. The use of UCOE vectors in combination with a preadapted serum free, suspension cell line allows for rapid production of large quantities of protein. Cytotechnol. 2002 Jan;38(1–3):43–46.

16. Kwaks THJ, Barnett P, Hemrika W, Siersma T, Sewalt RGAB, Satijn DPE, et al., Identification of anti-repressor elements that confer high and stable protein production in mammalian cells. Nat Biotechnol. 2003;21(7):553–558.

17. Cuisset L, Tichonicky L, Delpech M. A protein phosphatase is involved in the inhibition of histone deacetylation by sodium butyrate. Biochem Biophys Res Commun. 1998 May 29;246(3):760–4.

18. Gorman C, Bullock C. Site-specific gene targeting for gene expression in eukaryotes. Curr Opin Biotechnol. 2000 Oct;11(5):455–60.

19. Davies A, Greene A, Lullau E, Abbott WM. Optimisation and evaluation of a high-throughput mammalian protein expression system. Protein Expr Purif. 2005 Jul;42(1):111–21.

20. Bode J, Schlake T, Iber M, Schübeler D, Seibler J, Snezhkov E, Nikolaev L. The transgeneticist's toolbox: novel methods for the targeted modification of eukaryotic genomes. J Biol Chem. 2000 Sep–Oct;381(9–10):801–13.

21. Figueroa B, Jr., Ailor E, Osborne D, Hardwick JM, Reff M, Betenbaugh MJ: Enhanced cell culture performance using inducible anti-apoptotic genes E1B-19K and Aven in the production of a monoclonal antibody with Chinese hamster ovary cells. Biotechnol Bioeng. 2007;97(4):877–892.

22. Lee KH, Sburlati A, Renner WA, Bailey JE. Deregulated expression of cloned transcription factor E2F-1 in Chinese hamster ovary cells shifts protein patterns and activates growth in protein-free medium. Biotechnol Bioeng. 1996;50(3):273–279.

23. Ifandi V, Al-Rubeai M. Regulation of cell proliferation and apoptosis in CHO-K1 cells by the coexpression of c-Myc and Bcl-2. Biotechnol Prog. 2005 May–Jun;21(3):671–7.

24. Jaluria P, Betenbaugh M, Konstantopoulos K, Frank B, Shiloach J. Application of microarrays to identify and characterize genes involved in attachment dependence in HeLa cells. Metab Eng. 2007 May;9(3):241–51.

25. Borth N, Mattanovich D, Kunert R, Katinger H. Effect of increased expression of protein disulfide isomerase and heavy chain binding protein on antibody secretion in a recombinant CHO cell line. Biotechnol Prog. 2005 Jan–Feb;21(1):106–11.

26. Umaña P, Jean-Mairet J, Moudry R, Amstutz H, Bailey JE. Engineered glycoforms of an antineuroblastoma IgG1 with optimized antibody-dependent cellular cytotoxic activity. Nat Biotechnol. 1999 Feb;17(2):176-80.

27. DeMaria CT, Cairns V, Schwarz C, Zhang J, Guerin M, Zuena E, et al. Accelerated clone selection for recombinant CHO cells using a FACS-based high-throughput screen. Biotechnol Prog. 2007 Mar–Apr;23(2):465–72.

28. Seth G, Philp RJ, Lau A, Jiun KY, Yap M, Hu WS. Molecular portrait of high productivity in recombinant NS0 cells. Biotechnol Bioeng. 2007 Jul 1;97(4):933–51.

29. Barnes LM, Bentley CM, Moy N, Dickson AJ. Molecular analysis of successful cell line selection in transfected GS-NS0 myeloma cells. Biotechnol Bioeng. 2007 Feb 1;96(2):337–48.

30. Zhang, Jinyou, Robinson, David. Development of animal-free, protein-free and chemically-defined media for NS0 cell culture. Cytotechnol. 2005 Jun;48(1–3):59–74.

31. Walowitz JL, Fike RM, Jayme DW. Efficient lipid delivery to hybridoma culture by use of cyclodextrin in a novel granulated dry-form medium technology. Biotechnol Prog. 2003 Jan–Feb;19(1):64–8.

32. Burky JE, Wesson MC, Young A, Farnsworth S, Dionne B, Zhu Y, et al. Protein-free fed-batch culture of non-GS NS0 cell lines for production of recombinant antibodies. Biotechnol Bioeng. 2007 Feb 1;96(2):281–93.

33. Seth G, Ozturk M, Hu WS. Reverting cholesterol auxotrophy of NS0 cells by altering epigenetic gene silencing. Biotechnol Bioeng. 2006 Mar 5;93(4):820–7.

34. Ge X, Hanson M, Shen H, Kostov Y, Brorson KA, Frey DD, Moreira A., Rao G. Validation of an optical sensor-based high-throughput bioreactor system for mammalian cell culture. J Biotechnol. 2006 Apr 10;122(3):293–306.

35. Allison DW, Aboytes KA, Fong DK, Leugers SL, Johnson TK, Loke HN, et al. Development and optimization of cell culture media: Genomic and proteomic approaches. BioProcess Int. 2005 Jan;3(1):38–45.

36. Ideker T, Thorsson V, Ranish JA, Christmas R, Buhler J, Eng JK, et al., Integrated genomic and proteomic analyses of a systematically perturbed metabolic network. Science. 2001 May 4;292(5518):929–34.

37. Mollet M, Ma N, Zhao Y, Brodkey R, Taticek R, Chalmers JJ. Bioprocess equipment: characterization of energy dissipation rate and its potential to damage cells. Biotechnol Prog. 2004 Sep–Oct;20(5):1437–48.

38. Dhanasekharan KM, Sanyal J, Jain, A, Haidari A. A generalized approach to model oxygen transfer in bioreactors using population balances and computational fluid dynamics. Chem Eng Sci. 2005;60(1):213–218.

39. Morrow, KJ. Disposable bioreactors gaining favor. Gen Eng Biotechnol News. 2006 Jun 15;26(12):42–45.

40. Kirdar, AO, Conner JS, Baclaski J, Rathore AS. Application of multivariate analysis toward biotech processes: Case study of a cell-culture unit operation. Biotechnol Prog. 2007;23(1),61–67.