A perspective-driven and technical evaluation of machine learning in bioreactor scale-up: A case-study for potential model developments.

Bioreactor scale-up and scale-down have always been a topical issue for the biopharmaceutical industry and despite considerable effort, the identification of a fail-safe strategy for bioprocess development across scales remains a challenge. With the ubiquitous growth of digital transformation technologies, new scaling methods based on computer models may enable more effective scaling. This study aimed to evaluate the potential application of machine learning (ML) algorithms for bioreactor scale-up, with a specific focus on the prediction of scaling parameters. Factors critical to the development of such models were identified and data for bioreactor scale-up studies involving CHO cell-generated mAb products collated from the literature and public sources for the development of unsupervised and supervised ML models. Comparison of bioreactor performance across scales identified similarities between the different processes and primary differences between small- and large-scale bioreactors. A series of three case studies were developed to assess the relationship between cell growth and scale-sensitive bioreactor features. An embedding layer improved the capability of artificial neural network models to predict cell growth at a large-scale, as this approach captured similarities between the processes. Further models constructed to predict scaling parameters demonstrated how ML models may be applied to assist the scaling process. The development of data sets that include more characterization data with greater variability under different gassing and agitation regimes will also assist the future development of ML tools for bioreactor scaling.

Recent advances in the production of recombinant factor IX: bioprocessing and cell engineering.

Appropriate treatment of Hemophilia B is vital for patients' quality of life. Historically, the treatment used was the administration of coagulation Factor IX derived from human plasma. Advancements in recombinant technologies allowed Factor IX to be produced recombinantly. Successful recombinant production has triggered a gradual shift from the plasma derived origins of Factor IX, as it provides extended half-life and expanded production capacity. However, the complex post-translational modifications of Factor IX have made recombinant production at scale difficult. Considerable research has therefore been invested into understanding and optimizing the recombinant production of Factor IX. Here, we review the evolution of recombinant Factor IX production, focusing on recent developments in bioprocessing and cell engineering to control its post-translational modifications in its expression from Chinese Hamster Ovary (CHO) cells.

Perfusion culture of Chinese Hamster Ovary cells for bioprocessing applications.

Much of the biopharmaceutical industry's success over the past 30 years has relied on products derived from Chinese Hamster Ovary (CHO) cell lines. During this time, improvements in mammalian cell cultures have come from cell line development and process optimization suited for large-scale fed-batch processes. Originally developed for high cell densities and sensitive products, perfusion processes have a long history. Driven by high volumetric titers and a small footprint, perfusion-based bioprocess research has regained an interest from academia and industry. The recent pandemic has further highlighted the need for such intensified biomanufacturing options. In this review, we outline the technical history of research in this field as it applies to biologics production in CHO cells. We demonstrate a number of emerging trends in the literature and corroborate these with underlying drivers in the commercial space. From these trends, we speculate that the future of perfusion bioprocesses is bright and that the fields of media optimization, continuous processing, and cell line engineering hold the greatest potential. Aligning in its continuous setup with the demands for Industry 4.0, perfusion biomanufacturing is likely to be a hot topic in the years to come.

Coagulation factor IX analysis in bioreactor cell culture supernatant predicts quality of the purified product.

Coagulation factor IX (FIX) is a complex post-translationally modified human serum glycoprotein and high-value biopharmaceutical. The quality of recombinant FIX (rFIX), especially complete γ-carboxylation, is critical for rFIX clinical efficacy. Bioreactor operating conditions can impact rFIX production and post-translational modifications (PTMs). With the goal of optimizing rFIX production, we developed a suite of Data Independent Acquisition Mass Spectrometry (DIA-MS) proteomics methods and used these to investigate rFIX yield, γ-carboxylation, other PTMs, and host cell proteins during bioreactor culture and after purification. We detail the dynamics of site-specific PTM occupancy and structure on rFIX during production, which correlated with the efficiency of purification and the quality of the purified product. We identified new PTMs in rFIX near the GLA domain which could impact rFIX GLA-dependent purification and function. Our workflows are applicable to other biologics and expression systems, and should aid in the optimization and quality control of upstream and downstream bioprocesses.

S-Trap Eliminates Cell Culture Media Polymeric Surfactants for Effective Proteomic Analysis of Mammalian Cell Bioreactor Supernatants.

Proteomic analysis of bioreactor supernatants can inform on cellular metabolic status, viability, and productivity, as well as product quality, which can in turn help optimize bioreactor operation. Incubating mammalian cells in bioreactors requires the addition of polymeric surfactants such as Pluronic F68, which reduce the sheer stress caused by agitation. However, these surfactants are incompatible with mass spectrometry proteomics and must be eliminated during sample preparation. Here, we compared four different sample preparation methods to eliminate polymeric surfactants from filtered bioreactor supernatant samples: organic solvent precipitation; filter-assisted sample preparation (FASP); S-Trap; and single-pot, solid-phase, sample preparation (SP3). We found that SP3 and S-Trap substantially reduced or eliminated the polymer(s), but S-Trap provided the most robust cleanup and highest quality data. Additionally, we observed that SP3 sample preparation of our samples and in other published data sets was associated with partial alkylation of cysteines, which could impact the confidence and robustness of protein identification and quantification. Finally, we observed that several commercial mammalian cell culture media and media supplements also contained polymers with similar mass spectrometry profiles, and we suggest that proteomic analyses in these media will also benefit from the use of S-Trap sample preparation.

Heat shock protein 27 overexpression in CHO cells modulates apoptosis pathways and delays activation of caspases to improve recombinant monoclonal antibody titre in fed-batch bioreactors.

CHO cells are major production hosts for recombinant biologics including the rapidly expanding recombinant monoclonal antibodies (mAbs). Heat shock protein 27 (HSP27) expression was observed to be down-regulated towards the late-exponential and stationary phase of CHO fed-batch bioreactor cultures, whereas HSP27 was found to be highly expressed in human pathological cells and reported to have anti-apoptotic functions. These phenotypes suggest that overexpression of HSP27 is a potential cell line engineering strategy for improving robustness of CHO cells. In this work, HSP27 was stably overexpressed in CHO cells producing recombinant mAb and the effects of HSP27 on cell growth, volumetric production titer and product quality were assessed. Concomitantly, HSP27 anti-apoptosis functions in CHO cells were investigated. Stably transfected clones cultured in fed-batch bioreactors displayed 2.2-fold higher peak viable cell density, delayed loss of culture viability by two days and 2.3-fold increase in mAb titer without affecting the N-glycosylation profile, as compared to clones stably transfected with the vector backbone. Co-immunoprecipitation studies revealed HSP27 interactions with Akt, pro-caspase 3 and Daxx and caspase activity profiling showed delayed increase in caspase 2, 3, 8 and 9 activities. These results suggest that HSP27 modulates apoptosis signaling pathways and delays caspase activities to improve performance of CHO fed-batch bioreactor cultures.

Combined in silico modeling and metabolomics analysis to characterize fed-batch CHO cell culture.

The increasing demand for recombinant therapeutic proteins highlights the need to constantly improve the efficiency and yield of these biopharmaceutical products from mammalian cells, which is fully achievable only through proper understanding of cellular functioning. Towards this end, the current study exploited a combined metabolomics and in silico modeling approach to gain a deeper insight into the cellular mechanisms of Chinese hamster ovary (CHO) fed-batch cultures. Initially, extracellular and intracellular metabolite profiling analysis shortlisted key metabolites associated with cell growth limitation within the energy, glutathione, and glycerophospholipid pathways that have distinct changes at the exponential-stationary transition phase of the cultures. In addition, biomass compositional analysis newly revealed different amino acid content in the CHO cells from other mammalian cells, indicating the significance of accurate protein composition data in metabolite balancing across required nutrient assimilation, metabolic utilization, and cell growth. Subsequent in silico modeling of CHO cells characterized internal metabolic behaviors attaining physiological changes during growth and non-growth phases, thereby allowing us to explore relevant pathways to growth limitation and identify major growth-limiting factors including the oxidative stress and depletion of lipid metabolites. Such key information on growth-related mechanisms derived from the current approach can potentially guide the development of new strategies to enhance CHO culture performance.

Two-dimensional difference fluorescence gel electrophoresis to verify the scale-up of a non-affinity-based downstream process for isolation of a therapeutic recombinant antibody.

For therapeutic antibody production Protein A chromatography is often replaced by non-affinity-based purification sequences, which are considered as more economical. 2-D DIGE was applied for evaluation of scale-up of non-affinity based process of a humanized monoclonal antibody, anti-Rh(D) IgG(1), in comparison with other conventional analytical methods, like SDS-PAGE, Western blot, or SEC. Due to a high sensitivity of this technique (125 pg protein/spot) and high dynamic range of five orders of magnitude, low molecular weight impurities were detected in purified samples. Cation exchange chromatography was efficient capture step for IgG(1) purification in laboratory and pilot scale. The differences between samples after first purification step in laboratory and pilot scale were compensated with second purification step where almost the same protein pattern was observed. 2-D DIGE is a helpful tool for monitoring of purification effects and for scale-up verification of downstream processes.

Engineering mammalian cells in bioprocessing - current achievements and future perspectives.

Over the past 20 years, we have seen significant improvements in product titres from 50 mg/l to 5-10 g/l, a more than 100-fold increase. The main methods that have been employed to achieve this increase in product titre have been through the manipulation of culture media and process control strategies, such as the optimization of fed-batch processes. An alternative means to increase productivity has been through the engineering of host cells by altering cellular processes. Recombinant DNA technology has been used to over-express or suppress specific genes to endow particular phenotypes. Cellular processes that have been altered in host cells include metabolism, cell cycle, protein secretion and apoptosis. Cell engineering has also been employed to improve post-translational modifications such as glycosylation. In this article, an overview of the main cell engineering strategies previously employed and the impact of these strategies are presented. Many of these strategies focus on engineering cell lines with more efficient carbon metabolism towards reducing waste metabolites, achieving a biphasic production system by engineering cell cycle control, increasing protein secretion by targeting specific endoplasmic reticulum stress chaperones, delaying cell death by targeting anti-apoptosis genes, and engineering glycosylation by enhancing recombinant protein sialylation and antibody glycosylation. Future perspectives for host cell engineering, and possible areas of research, are also discussed in this review.

Identification of cellular genes critical to recombinant protein production using a Gaussia luciferase-based siRNA screening system.

Development of high-throughput functional genomic screening, including siRNA screening, provides a novel approach for quick identification of critical factors involved in biological processes. Here, we apply this strategy to search for cellular genes involved in recombinant protein production. Since most of biopharmaceutical proteins are secreted proteins, we develop a cell-based reporter assay using a secreted luciferase, Gaussia luciferase (Gluc), as the reporter. Human embryonic kidney 293 (HEK293) cells transiently transfected with the Gluc reporter plasmid are used to screen our siRNA panel. Three cellular genes, CCAAT/enhancer binding protein gamma (CEBPG), potassium channel tetramerisation domain containing 2 (KCTD2), transmembrane protein 183A (TMEM183A), were isolated from the screening. Production of erythropoietin (EPO) was significantly inhibited when CEBPG, KCTD2, and TMEM183A were knocked down. Furthermore, overexpression of CEBPG is shown to significantly improve production of recombinant EPO, interferon gamma, and monoclonal antibody in HEK293 and Chinese hamster ovary cells. Collectively, this novel Gluc-based siRNA screening system is proven to be a useful tool for investigation of secreted protein production in mammalian cells.

Regulation of XBP-1 signaling during transient and stable recombinant protein production in CHO cells.

X-box binding protein 1 (XBP-1) is a key regulator of cellular unfolded protein response (UPR). The spliced isoform of XBP-1, XBP-1S, is a transcription activator, which is expressed only when UPR is induced. However, the impact of recombinant protein production on the regulation of XBP-1 signaling in CHO cells is not well understood. In this report, we cloned the Chinese hamster XBP-1 homolog to aid the investigation of the interplay between protein productivity, culture conditions, and endogenous XBP-1 signaling in CHO cells. Interestingly, expression of XBP-1S is detected in the non-producing and unstressed CHO-K1 cells. Transient expression of recombinant erythropoietin reveals a positive correlation between XBP-1 mRNA abundance and protein production level. However, such a correlation is not observed in batch cultivation of stable producing cell lines. The increased XBP-1 splicing is detected in late-phase cultures, suggesting that induction of XBP-1S may be a result of nutrient limitations or other environmental stresses rather than that of increased intracellular accumulation of recombinant proteins. Our data suggest that XBP-1 is a key determinant for the secretory capacity of CHO cells. Understanding its dynamic regulation hence provides a rational basis for cellular engineering strategies to improve recombinant protein secretion.

Overexpression of heat shock proteins (HSPs) in CHO cells for extended culture viability and improved recombinant protein production.

It has been widely reported that CHO cells undergo apoptosis in culture, despite supplementation of nutrients through fed-batch strategies. Improvement of cell viability in culture can effectively improve recombinant protein yield through extension of the culture's production lifespan, especially at high cell densities. Heat shock proteins (HSPs) have been reported to demonstrate anti-apoptotic effects against a wide range of physical and chemical stimuli through their ability to bind and act as antagonists to critical apoptotic molecules. CHO-IFN-gamma cells, expressing recombinant human interferon-gamma (IFN-gamma), were engineered to overexpress two HSPs (HSP27 and HSP70) either individually or in combination. In fed-batch bioreactor cultures, the engineered cell lines exhibited a more gradual viability loss and extension of culture times of 36-72h, with corresponding delays in escalation of caspases 2, 3, 8 and 9 activities, compared to the control cultures utilizing cells transfected with the vector backbone. The extension in culture times translated to a 2.5-fold improvement in IFN-gamma production over controls in fed-batch cultures. These results suggest that overexpression of HSPs represents a promising generic strategy for the development of robust CHO cell lines resistant to apoptotic insults and possessing improved culture characteristics to enhance recombinant glycoprotein yields.

Two-dimensional fluorescence difference gel electrophoresis for comparison of affinity and non-affinity based downstream processing of recombinant monoclonal antibody.

Although Staphylococcus Protein A (SpA) affinity chromatography is the state of the art capture step for antibody purification, non-affinity methods are more economical. We used two-dimensional fluorescence difference gel electrophoresis (2-D DIGE) to evaluate the purification of a recombinant IgG(1) antibody from cultured cells, with two different processes: (1) SpA capture followed by cation-exchange chromatography (CEX); and (2) CEX capture, followed by anion exchanger, then hydrophobic interaction chromatography. Efficiencies were similar in sodium dodecylsulphate polyacrylamide gel electrophoresis and size-exclusion chromatography; however, 2-D DIGE revealed higher efficiency with SpA than with CEX capture. Thus, 2-D DIGE is a valuable tool for downstream process development.

Identification and expression analysis of miRNAs during batch culture of HEK-293 cells.

MicroRNAs (miRNAs) are small non-coding RNAs of about 20-24 nucleotides in length. They regulate gene expression negatively and have been implicated in a wide variety of biological processes. To identify potential miRNAs that may influence the growth and proliferation of mammalian cells cultured in bioreactors, we applied miRNA microarray expression profiling technology to batch cultures of HEK293 cells in protein free media. In our study, we identified miRNAs that were differentially expressed during the exponential and stationary phases, 13 of these showed distinct up regulation trends while 1 exhibited down regulation. These miRNAs have been implicated in cellular differentiation, growth arrest and apoptosis. Specifically, miR-16 and let-7b are potentially useful in the enhancement of bioreactor cell cultures.

2-D DIGE to expedite downstream process development for human monoclonal antibody purification.

Two-dimensional fluorescence difference gel electrophoresis (2-D DIGE) is an established method for assessing protein expression strategies, understanding pathogenesis mechanisms, characterizing biomarkers, and controlling therapeutic processes. We applied 2-D DIGE to facilitate the development of a purification process for a recombinant IgG1 antibody against Rhesus D antigen expressed by Chinese hamster ovary cells. The variability of two expression clones as well as the influence of cell viability on the host-cell protein pattern was assessed quantitatively. Up to 800 different spots were identified. 2-D DIGE showed that differences in cell viability had more influence on the protein expression pattern than did the expression clone itself. After purification of the IgG from different culture supernatants, the protein patterns on 2-D DIGE were identical, indicating the validity of purification scheme.

Adaptation of hybridomas to protein-free media results in a simplified two-step immunoglobulin M purification process.

An IgM antibody was purified from hybridoma supernatant containing serum using a three-step purification process comprising of tangential flow filtration, anion-exchange chromatography and size-exclusion chromatography. Recovery and purity were significantly improved upon adaptation of the hybridoma to serum-free media. The process could even be simplified by omitting the initial tangential flow filtration step. Even with a two-step purification process a purity of >98% and a recovery of >60% was obtained.

Modeling amino acid metabolism in mammalian cells-toward the development of a model library.

Amino acids are necessary to mammalian cell cultures both for protein synthesis and as an energy source. In this study, we present an unstructured mathematical model describing (i) cell growth and death kinetics and (ii) metabolism of glucose and 19 amino acids for HEK-293 and CHO IFN-gamma cell cultures. The proposed mathematical framework is in good agreement with experimental data for both cell lines. It accommodates the inclusion of expressions for other cellular activities, such as the production of recombinant viral vectors or proteins, and can be used as the basis for the development of a model library for mammalian cell cultures.

Transcriptional profiling of batch and fed-batch protein-free 293-HEK cultures.

Dynamic nutrient feeding to control glutamine at low levels in protein-free fed-batch cultures of 293-human embryonic kidney (HEK) cells achieved cell concentrations of 6 x 10(6) cells/ml. This represented a 4-fold improvement in cell concentration compared to batch cultures. Reduction in glutamine and glucose consumption, as well as lactate and ammonia production, were also observed in these fed-batch cultures. High virus production titers of 3 x 10(11) pfu/ml were achieved in fed-batch cultures which were 10,000-fold higher than batch cultures. An investigation of the transcriptional regulation of the metabolic changes associated with the batch and the low-glutamine fed-batch cultures using DNA microarray was conducted. This analysis provides better understanding of the transcriptional regulatory mechanism resulting in the observed physiological changes. Transcriptional profiling of cells from the mid-exponential, late exponential and stationary phases of both the batch and fed-batch were undertaken using an 18,000 element human chip. Transcriptional profiles were ontologically classified to provide a global view of the genetic changes. Furthermore, a pathway-oriented analysis focusing on cellular metabolism was conducted to reveal the dynamic regulation of genes related to amino acid metabolism, tRNA synthetases, TCA cycle, electron transport chain and glycolysis.

Transcriptome and proteome profiling to understanding the biology of high productivity CHO cells.

A combined transcriptome and proteome analysis was carried out to identify key genes and proteins differentially expressed in Chinese hamster ovary (CHO) cells producing high and low levels of dhfr-GFP fusion protein. Comparison of transcript levels was performed using a proprietary 15K CHO cDNA microarray chip, whereas proteomic analysis was performed using iTRAQ quantitative protein profiling technique. Microarray analysis revealed 77 differentially expressed genes, with 53 genes upregulated and 24 genes downregulated. Proteomic analysis gave 75 and 80 proteins for the midexponential and stationary phase, respectively. Although there was a general lack of correlation between mRNA levels and quantitated protein abundance, results from both datasets concurred on groups of proteins/genes based on functional categorization. A number of genes (20%) and proteins (45 and 23%) were involved in processes related to protein biosynthesis. We also identified three genes/proteins involved in chromatin modification. Enzymes responsible for opening up chromatin, Hmgn3 and Hmgb1, were upregulated whereas enzymes that condense chromatin, histone H1.2, were downregulated. Genes and proteins that promote cell growth (Igfbp4, Ptma, S100a6, and Lgals3) were downregulated, whereas those that deter cell growth (Ccng2, Gsg2, and S100a11) were upregulated. Other main groups of genes and proteins include carbohydrate metabolism, signal transduction, and transport. Our findings show that an integrated genomic and proteomics approach can be effectively utilized to monitor transcriptional and posttranscriptional events of mammalian cells in culture.

Transcriptional profiling of apoptotic pathways in batch and fed-batch CHO cell cultures.

Chinese Hamster ovary (CHO) cells are regarded as one of the "work-horses" for complex biotherapeutics production. In these processes, loss in culture viability occurs primarily via apoptosis, a genetically controlled form of cellular suicide. Using our "in-house" developed CHO cDNA array and a mouse oligonucleotide array for time profile expression analysis of batch and fed-batch CHO cell cultures, the genetic circuitry that regulates and executes apoptosis induction were examined. During periods of high viability, most pro-apoptotic genes were down-regulated but upon loss in viability, several early pro-apoptotic signaling genes were up-regulated. At later stages of viability loss, we detected late pro-apoptotic effector genes such as caspases and DNases being up-regulated. This sequential regulation of apoptotic genes showed that DNA microarrays could be used as a tool to study apoptosis. We found that in batch and fed-batch cultures, apoptosis signaling occurred primarily via death receptor- and mitochondria-mediated signaling pathways rather than endoplasmic reticulum-mediated signaling. These insights provide a greater understanding of the regulatory circuitry of apoptosis during cell culture and allow for subsequent targeting of relevant apoptosis signaling genes to prolong cell culture.