Reviews pubs.acs.org/jpr
Application of Multi-Omics Techniques for Bioprocess Design and Optimization in Chinese Hamster Ovary Cells Amy Farrell,† Niaobh McLoughlin,† John J. Milne,† Ian W. Marison,‡ and Jonathan Bones*,† †
Characterisation and Comparability Laboratory, NIBRT − The National Institute for Bioprocessing Research and Training, Foster Avenue, Mount Merrion, Blackrock, Co. Dublin, Ireland ‡ Laboratory of Integrated Bioprocessing, Dublin City University, Glasnevin, Dublin 9, Ireland ABSTRACT: Significant improvements in the productivity and quality of therapeutic proteins produced in Chinese hamster ovary (CHO) cells have been reported since their establishment as host cells for biopharmaceutical production. Initial advances in the field focused on engineering strategies to manipulate genes associated with proliferation, apoptosis, and various metabolic pathways. Process engineering efforts to optimize culture media, batch-feeding strategies and culture conditions, including temperature and osmolarity, were also reported. More recently, focus has shifted toward enhancing process consistency and product quality using systems biology quality by design-based approaches during process development. Integration of different data generated using omics technologies, such as genomics, transcriptomics, proteomics and metabolomics, has facilitated a greater understanding of CHO cell biology. These techniques have enabled the provision of global information on dynamic changes in cellular components associated with different phenotypes. Using systems biology to understand these important host cells at the cellular level will undoubtedly result in further progression in the development and expression of biopharmaceutical products in CHO cells. KEYWORDS: Chinese Hamster ovary cells, bioprocessing, cell engineering, industrial proteomics, metabolomics, CHO genomics, process design spanning almost 30 years.1,7 Second, CHO cells have desirable process performance characteristics such as high expression rates, rapid growth and the ability to grow in both adherent and serum-free suspension cell cultures.7,8 The successful adaption of CHO cells to grow in serum-free media in suspension culture reinforced their status as the number one mammalian cell host for biopharmaceutical production. In addition to alleviating regulatory concerns regarding high-risk adventitious agents, the omission of serum has facilitated the transition to chemically defined media. Fully chemically defined serum-free growth media has simplified downstream processing due to a reduction in contaminants incorporated into the product, thereby minimizing batch-to-batch variation.9,10 Third, low specific productivity associated with recombinant protein production in mammalian cell lines has been circumnavigated using well-characterized and commonly used cell transfection, gene amplification and clone selection methods,8 which are discussed in Section 2 below. Fourth, high batch yields have been achieved in CHO cells through the optimization of batchfeeding strategies, resulting in prolonged viability, higher viable cell density and increased specific productivity and overall yield.11,12 Finally, the most significant attribute of CHO cells relating to biopharmaceutical production is their ability to
1. INTRODUCTION The biopharmaceutical sector has become a significant and ever-increasing segment of the global pharmaceutical industry. The overall market for protein-based therapeutics reached U.S. $99 billion in 2009 and further grew to U.S. $163 billion in 2012.1,2 More than 200 biopharmaceutical products have achieved market approval for the treatment of ailments including cancer, chemotherapy-induced neutropenia, diabetes mellitus, hepatitis, multiple sclerosis, rheumatoid arthritis, endstage renal disease and other orphan conditions, with greater than 350 additional products currently in clinical trials.1,3 Mammalian cell lines, including those derived from Chinese hamster ovary (CHO) cells, baby hamster kidney (BHK) cells, mouse myeloma (NS0), human embryonic kidney (HEK-293) and human retinal cells, have been established as production platforms for recombinant therapeutic proteins.4,5 Despite the number of mammalian cell lines available, CHO cells have emerged as the most popular choice for recombinant protein production, with approximately 45% of new products approved in the four year period from 2006 to 2010 produced in CHO cells.1 Currently, 7 out of 10 of the world’s top-selling biopharmaceutical drugs are expressed in CHO cells.6 The widespread use of CHO cells in recombinant protein production can be attributed to a number of important factors. First, CHO cells have an established record of product safety, quality and efficacy and a long history of regulatory approval © XXXX American Chemical Society
Received: March 5, 2014
A
dx.doi.org/10.1021/pr500219b | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Reviews
improved since the late 1980s from expression levels of 50 mg/L to routine concentrations of 2−5 g/L currently.7 Improvements in productivity have been largely achieved through CHO cell engineering, clonal selection, optimization of processing conditions, including batch-feeding strategies and developments in chemically defined and serum-free media.12 Aligned with efforts regarding recombinant protein productivity, the enhancement of consistent therapeutic protein quality is of great interest. The quality of most protein-based biopharmaceuticals is greatly affected by variability in PTMs, which can have a vast impact on their therapeutic properties.18 Glycosylation, the most common PTM in CHO-produced recombinant protein products, involves the incorporation of oligosaccharides into the therapeutic protein.12 The addition of N- and O-linked glycans to recombinant proteins imparts functional diversity to the protein that can impact cellular processes in development, immune reactions and cell signaling pathways.19 Consequently, numerous CHO cell engineering strategies have focused on producing genetically modified CHO mutants with enhanced glycosylation capabilities.20−22
produce human-like post translational modifications (PTMs), in particular, glycosylation. This attribute results in an increase in efficacy and durability of the therapeutic protein and a minimization in associated safety concerns.13 CHO cells are envisaged to remain as the dominant production system for recombinant therapeutics for years to come. The large database of product quality and safety data and the extensive work completed in optimizing these production systems suggests that the investment required to establish an alternative production system, comparable with CHO cells, using transgenic animals or microbial, plant-based, or other human cell lines would not be cost-effective.7 Despite reported progress in improving therapeutic protein productivity, poor characterization of the cellular machinery of CHO cells has been achieved.14 Characterization of the fundamental biological mechanisms of CHO cells in industrial cell culture may provide a foundation for further knowledgedriven improvement of manufacturing processes.15 Hence, current research focus has shifted toward the enhancement of process consistency and associated recombinant product quality.7 This new direction has led to the implementation of a systems biology approach to gain a global understanding of CHO cell physiology and consequently a heightened knowledge of exogenous recombinant protein production. Systems biology has been greatly facilitated by the development of multi-omic tools including genomic, transcriptomic, proteomic and metabolomic techniques and through advances in analytical technologies, in particular, mass spectrometry. In the past, advances in CHO systems biology have been hampered by the lack of genomic data. Consequently, the publication of the CHO-K1 genome sequence in 2011 represented a major milestone in CHO cell systems biology.13 The more recent publication of the Chinese hamster draft genome and the genomes of six CHO cell lines from the CHO-K1, DG44 and CHO-S lineages will considerably further advance our capabilities to characterize and better understand these important production hosts and will ultimately facilitate improvements in productivity and biopharmaceutical product quality.16 This review describes several engineering strategies that have been employed for the improvement of recombinant therapeutic protein productivity and quality. Emerging multiomic approaches used to characterize CHO cell production systems at the systems biology level are outlined. The evolving body of information gathered using multi-omic techniques will be instrumental in expediting improvements in the production of therapeutic proteins in CHO cells through facilitation of knowledge-based decisions for future process manipulations and cell engineering strategies.
2.1. CHO Cell Engineering
Improvements to CHO cell lines through cell engineering have primarily involved the downregulation or overexpression of individual genes to optimize cellular processes within the cell line. The principal targets for enhancing culture longevity, improving the specific growth rate and increasing the specific productivity include the induction of cell proliferation at the initial stages of culture and the reduction of cell death at the end of cell culture. Such approaches, which have largely been achieved through anti-apoptosis engineering and through proliferation-, metabolic- and unfolded protein response (UPR)-based engineering,8 are briefly discussed in this section. For a broader review of CHO cell engineering, the reader is directed toward comprehensive reviews of the field by Kim and co-workers8 and Lim et al.12 Targeting apoptotic genes to generate apoptosis-resistant CHO cell lines has been a successful strategy for the generation of efficient cell lines for therapeutic protein production. Several genes that demonstrate anti-apoptotic activity have been identified and are shown in Figure 1, including Bcl-2 and BclxL. The protein products of these genes delay the onset of apoptosis by binding to pro-apoptotic proteins (Bim, Bad, Bax and Bak) in the mitochondrial-mediated apoptosis signaling pathway, thereby disturbing the activation of caspases.23,24 High-producing CHO cell lines have been generated following transfection with Bcl-xL to increase resistance to apoptosis.24 Conversely, an apoptosis-resistant CHO cell line was generated following silencing of the pro-apoptotic genes Bax and Bak in an interferon-γ (IFN-γ)-producing CHO cell line.25 Wong et al. generated four genetically modified apoptosis-resistant CHO cell lines using small interfering RNA (siRNA) knockdown of the pro-apoptotic genes Alg-2 and Requiem and by overexpression of the anti-apoptotic protein Faim accompanied by overexpression of a dominant-negative form of the antiapoptotic protein Fadd, as presented in Figure 1.26 Because of their reported protective effects against a range of chemical and physical apoptotic stimuli, the ability of heat shock proteins (HSPs) to extend culture viability and to improve recombinant protein production has been investigated. Lee and co-workers engineered CHO cells producing IFN-γ to overexpress HSP27 and HSP70.27 These engineered cell lines displayed a more gradual loss of viability and a 2.5-fold
2. CHO ENGINEERING STRATEGIES CHO cell lines were established as popular hosts for recombinant protein production following the adaption of cell lines to achieve high-level recombinant protein expression. The incorporation of mutations in the dihydrofolate reductase (dhf r) gene led to the production of two of the most widely used CHO cell lines in the biopharmaceutical industry, CHODXB11 (also known as CHO-DUKX) and CHO-DG44.17 The isolation of high-producing CHO cell lines signaled a continued effort to optimize therapeutic protein production to further increase the yield and quality of these products. One of the early shortcomings of recombinant protein production in CHO cells, low volumetric productivity, has been significantly B
dx.doi.org/10.1021/pr500219b | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Reviews
Figure 1. Signaling of apoptosis during CHO cell culture via the death receptor-mediated, mitochondria-mediated and endoplasmic reticulummediated apoptosis signaling pathways.26
have gained momentum in recent years. Baik and co-workers used metabolic engineering to produce bioengineered heparin in CHO cells.31 Heparin is synthesized through the same biosynthetic pathway as heparan sulfate, a naturally produced molecule in CHO cells. Therefore, it may be possible to engineer CHO cells to produce pharmacological heparin, although the production of bioengineered heparin in CHO cells has yet to be realized.32 Emerging systems biology tools, as discussed below, have facilitated a greater understanding of metabolic pathways at the molecular and cellular level.33 Increased knowledge of metabolic pathways in CHO cells has paved the way for manipulation to enhance the growth of CHO cells in culture. A metabolic engineering strategy was employed by Le et al. to alter the expression of mGLUT5 in CHO cells, thereby controlling sugar consumption, decreasing lactate production and ultimately inhibiting apoptosis and prolonging culture duration.34 Despite limited use in increasing product yield in mammalian cells, metabolic engineering was reported to improve the quality of recombinant glycoproteins by manipulating glycosyltransferase and glycosidase expression.35 Glycoproteins are produced by cellular processes involving several families of proteins, including glycosidases, glycosyltransferases, nucleotide sugar synthesis enzymes and transporters. These mechanisms are highly variable and give rise to glycoprotein heterogeneity. Consequently, the various protein
improvement in IFN-γ titer when compared to that of parental cell lines. The engineered cells also achieved an extension of culture times from 36 to 72 h in fed-batch bioreactor cultures and a corresponding delay in caspase-2, -3, -8 and -9 activities. Caspase activation is known to affect both the quantity and quality of recombinant proteins and consequently the caspase cascade systems that mediate apoptotic signaling have been considered for cell engineering approaches. Apoptosis is known to occur via the death receptor- or mitochondrial-mediated pathways and is instigated via the initiator caspases, caspase-8 and caspase-9, respectively.28 Inhibition of caspase-8 and caspase-9 has been shown to enhance the viability of CHO cells.29 Manipulation of genes and proteins associated with proliferation has also proven to be an effective approach to increase viable cell concentrations. C-myc, which regulates cell proliferation, has been shown to successfully increase proliferation rates and result in a high integral viable cell number when transfected into CHO cell lines.30 Other cellular engineering targets that have been investigated in recombinant CHO (rCHO) cells include proteins in the secretory pathway and the unfolded protein response. Metabolic engineering of a recombinant CHO cell line involves upregulation and/or downregulation of specific proteins in a metabolic pathway, resulting in a novel product.19 Such approaches to production of non-protein bioproducts C
dx.doi.org/10.1021/pr500219b | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Reviews
2.2. CHO Process Engineering Strategies
families involved in glycosylation have been targeted in cellular engineering efforts to control heterogeneity or to produce glycoproteins incorporating desirable characteristics.12 For instance, N-acetyllactosaminide 3-α-galactosyltransferase-1 exemplifies a possible engineering target because it is responsible for the synthesis of the terminal galactose-α-1,3galactose (α-Gal), an oligosaccharide epitope capable of eliciting adverse clinical events in patients. Although α-Gal is present in CHO cells, the expression of this gene varies greatly in different CHO cell lines as well as in subclonal populations from the same cell line.36 Thus, determination of α-Gal activity during clonal selection or monitoring and control of α-Gal levels during therapeutic protein production may be beneficial to patients. Biological effector mechanisms such as antibody-dependent cellular cytotoxicity (ADCC), a cell-mediated immune defense mechanism, are known to be influenced by the N-linked oligosaccharide attached to asparagine 297 of the Fc region of immunoglobulin G (IgG) and are reportedly enhanced by the removal of core fucose from N-glycans attached to asparagine 297 in the CH2 domain.22,37 Accordingly, glycosylation engineering efforts have focused on the design of monoclonal antibodies (mAbs) with improved Fc function.12,38 Zinc-finger nuclease technology has been used to alter fucosylation and sialylation of recombinant therapeutic products.22,39 A CHO glycosylation mutant, CHO-gmt5, produced by zinc-finger nuclease technology was employed to generate recombinant glycoproteins incorporating N-glycans deficient in fucose and sialic acid.22 The resulting improvement in ADCC rendered CHO-gmt5 an advantageous cell line for recombinant protein production.21,22,39 CHO glycosylation mutant (CHO-gmt) cell lines are reported to overcome the inherent heterogeneity associated with glycoproteins due to genetic defects in their glycosylation machinery, leading to recombinant glycoproteins with consistent and unique glycan structures.21 Tummala et al. used siRNA as an alternative to metabolic engineering of host cells to produce a mAb with decreased levels of core fucosylation.40 When added to bioreactors, siRNA duplexes targeting gmds and f ut8 succeeded in producing a modified mAb with improved biological function, without the requirement for cell-line development or adverse effects on cell growth, therapeutic protein productivity, or product quality.40 Another popular approach to cellular engineering that has gained great interest in recent years is the use of micro-RNA (miRNA). miRNA is non-coding single-stranded RNA of between 18 to 25 nucleotides in length that regulates gene expression by binding to the 3′ UTR regions of target mRNA, by either inhibiting or degrading the translation of target mRNA.8,41 miRNA has the potential to affect cell growth, metabolism, apoptosis, secretion and specific productivity of recombinant therapeutic proteins without contributing to translational burden in the host cell.19 miRNA has emerged as an attractive alternative to single-gene targeting in cellular engineering of CHO cells because it has the ability to bind to several different mRNAs, thereby potentially modulating multiple complex regulatory pathways simultaneously.8 Furthermore, overexpression of miRNA contribute less metabolic burden to the cell compared to that of a protein that may occupy endogenous translation machinery, possibly impacting specific productivity.42
In addition to cellular and metabolic engineering approaches to improve recombinant viable cell concentration and specific productivity in CHO cells, attempts to increase titers through the optimization of bioprocesses have also been performed. CHO cells were traditionally grown in adherent cell culture containing fetal bovine serum (FBS), which has been widely used as a growth supplement due to its high levels of stimulatory factors.43 In recent years, a trend has emerged to end the use of plasma-derived supplements in the production of therapeutic proteins. Removal of animal-derived components arose from regulatory concerns regarding the potential for transmission of infectious agents, such as bovine spongiform encephalopathy (BSE) and Creutzfeldt−Jakob disease (CJD) and from the inherent heterogeneity and unreliable supply associated with serum products44,45 The ease with which CHO cell lines adapt to serum-free suspension cell culture has allowed them to dominate the domain of recombinant therapeutic protein mass production.8 Serum-free media has been greatly enhanced through the addition of various supplements including sodium butyrate,46,47 zinc sulfate,48 adenosine49 and hydrolysates.50 Media supplementation with sodium butyrate (NaBu) has been used as an effective method to enhance specific productivity of recombinant proteins such as mAbs, erythropoietin (EPO) and tissue plasminogen activator (t-PA) in CHO cells.46 However, butyrate treatment is known to significantly inhibit cell growth and possibly lead to apoptosis.47 For this reason, NaBu is frequently used in combination with anti-apoptotic agents or used in the culture of cells that overexpress apoptosis suppressor genes.51 Zinc sulfate is commonly utilized to increase the expression of human growth hormone (hGH) in serum-free CHO cell culture. Simultaneous addition of NaBu and zinc sulfate to the cell culture of CHO cells expressing hGH was shown to result in an increase in production of the recombinant protein when compared to that of untreated cultures.52 Supplementation of CHO cells producing IFN-γ with adenosine was demonstrated to enhance productivity of recombinant IFN-γ through two likely mechanisms. Adenosine causes growth arrest in CHO cell culture and leads to an increase in intracellular ATP in mammalian cells, thereby enhancing protein expression capacity.49 Kim and colleagues studied the effect of Ca2+ and Mg2+ concentration in serum-free culture on the activation of recombinant factor IX (aFIX) produced in CHO cells.53 Ca2+ at 0.5 mM was found to inhibit the conversion of aFIX to the undesirable activated factor IX (FIXa) at the end stage of cell culture when compared to that of cell culture containing 1.0 mM Ca2+. The various effects of these media supplements in different cell culture denote the importance of using well-defined media for the production of recombinant therapeutics. For this purpose, many commercially available chemically defined media have been developed for the large-scale culture of CHO cells.54 Serum-free media supplemented with hydrolysates, which contain peptides, vitamins, free amino acids and trace elements, is commonly used in industry for large-scale culture of CHO cells.50 The use of hydrolysates in serum-free culture media is required to achieve viable cell densities and specific productivity comparable to that of CHO cell media containing FBS. However, batch-to-batch variation also exists for hydrolysates. Therefore, despite the regulatory-friendly nature of serum-free media, the use of serum-free media containing hydrolysates D
dx.doi.org/10.1021/pr500219b | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Reviews
and N-glycosylation of therapeutic proteins produced in CHO cells was investigated by Burleigh et al., who paired metabolic flux analysis with measurement of intracellular sugar nucleotides.66 CHO cells producing human chorionic gonadotrophin (HCG) were grown in a continuous culture system containing alternating glutamine concentrations of 8 and 0 mM, generating four steady metabolic states. The use of metabolic flux analysis at the different steady states demonstrated the metabolic changes that occurred in response to varying concentrations of glutamine and highlighted the possibility of manipulating cells into key metabolic states, potentially improving the quality of therapeutic proteins. To encourage the biopharmaceutical industry to optimize the efficiency of their manufacturing operations, the U.S. Food and Drug Administration (FDA) published a regulatory framework for process analytical technology (PAT) to promote real-time monitoring and control of process parameters that are critical to product quality.67 Likewise, the International Conference on Harmonisation (ICH) recommended the use of PAT for the timely measurements of quality and performance attributes critical to end-product quality.68 Consequently, many techniques have been evaluated for their suitability in monitoring CHO cell culture, including fluorescence spectroscopy,57 nearinfrared spectroscopy,69 flow cytometry70 and Raman spectroscopy.71−73 When combined with chemometric approaches, such as principal component analysis (PCA) and partial leastsquares regression methods (PLS), the PAT methods described have formed robust approaches for routine quality evaluation and prediction of cultivation variables.69,72 In addition to improvements in recombinant therapeutic yields through chemical means, efforts have also been made to increase viability and productivity of CHO cells through physical means, including the use of low-temperature cell culture,74−76 hyperosmotic pressure77 and microcarriers.78−80 Use of microcarriers in large-scale production of adherent cell cultures has also been evaluated as an alternative to serum-free culture. Tharmalingam et al. demonstrated a 2.6- to 2.8-fold increase in β-IFN titer following production in CHO cells entrapped in Cytopore microcarriers when compared to that of regular suspension culture.78 Optimized conditions employed were found to have no apparent effect on protein glycosylation and minimal aggregation of the hydrophobic therapeutic protein, β-IFN. In a follow-up study, optimized culture conditions were applied for the culture of CHO cells producing tissue plasminogen activator (t-PA).79 Microcarriers were found to result in a 2.5-fold increase in volumetric production of both β-IFN and t-PA when compared to that from cells grown in an equivalent suspension culture. Enhancement of volumetric product titer is achieved as a result of the protective environment established in the macroporous microcarriers, shielding cells from high shear forces associated with stirred bioreactor cultures. Breguet and co-workers investigated four types of alginate/poly-L-lysine microcapsules to determine the impact on microcapsule properties, size and mechanical stability on cell density and therapeutic protein productivity in CHO cell cultures.80 Liquid core microcapsules were found to be unsuitable for long-term cultures, whereas, in contrast, semiliquid microcapsules achieved greater mechanical resistance in addition to realizing a 2-fold increase in recombinant human secretory component of immunoglobulin A. Recently, Costa and co-workers examined the impact of microcarriers on glycosylation of mAb produced by adherent CHO-K1 cells.81 In addition to a reduction in glycoform
does not address the desire for a well-defined, robust and reproducible manufacturing process.55 Formulation of appropriate cell culture media necessitates the combination of a variety of components to achieve the desired physiological environment for effective culturing of mammalian cells. The composition of cell culture media is reported to be the most significant factor in determining the efficiency of industrial CHO cell processes; consequently, careful selection of media components is of paramount importance to ensure optimum end-product yield and quality.56 Similarly, the generation of rapid, robust, consistent and nondestructive analytical methods for the analysis of media quality and cell culture variance are important to ensure the efficacy of culture media and hence the quality and yield of therapeutic protein products.57 Fed-batch cultures have been proven to successfully increase viability and peak cell density and to enhance recombinant protein yield.58 Reducing lactate production is a major aim of batch-feeding strategies, as lactate metabolism produces lactic acid, a major byproduct of cell culture known to inhibit cell growth and reduce cell viability. A reduction in lactate production may be achieved by various approaches, including the manipulation of the glucose concentration of the media,59 the use of alternative carbon sources,60 the downregulation of lactate dehydrogenase and pyruvate dehydrogenase kinases,61 or the enhancement of glucose introduction into the tricarboxylic acid (TCA) cycle.62 Ma et al. combined the use of a fed-batch process with chemically defined media to improve both productivity and lactate metabolism in NS0 and CHO cells.62 In addition to a depletion of lactate for both cell lines, an 82% increase in CHO product titer and a 115% increase in NS0 product titer were observed. Dynamic online batch-feeding strategies have been developed that reduce glucose or glutamine concentration, thereby inhibiting lactate formation and extending viability in CHO cell cultures.11,63 Wong and colleagues reported an enhancement in IFN-γ productivity in CHO cells following a low glutamine fed-batch strategy. The 10-fold improvement in IFN-γ yield was attributed to a reduction in ammonia and lactate produced during cell culture.11 The impact of a dynamic online fed-batch strategy on the glycosylation quality of IFN-γ produced in CHO cells was evaluated. Very low concentrations of glutamine (
