Engineering the Future: Cell-Line Development Discussions at BPI West

View PDF

Big data, next-generation sequencing (NGS), automation, and powerful analytical technologies are helping biopharmaceutical companies achieve the parallel goals of cell-line development: precise integration of transgenes into production cells with documented monoclonality, shortening timelines and increasing productivity. Presentations in the most recent Biotech Week Boston (BWB) and the BPI Conference US West (BPI West) programs have demonstrated that there is no one way to accomplish those goals. Combining several of the approaches highlighted herein just might take biopharmaceutical production to new levels that not long ago would have been deemed impossible.

In BPI’s April 2023 featured report, we reported on discussions in Boston, MA, this past fall (1). The BPI Conference program within BWB included both a preconference workshop on CLD and topical talks within the upstream production track. Presenters often spoke about how advanced gene-editing techniques can increase process productivity without compromising product quality. For example, biopharmaceutical companies increasingly are applying transposase-based gene editing to increase gene-integration targeting efficiency and accelerate CLD workflows. That is helping to reduce the number of subcloning/pool-screening steps that programs require.

Some speakers highlighted the importance of high gene-copy numbers for optimized expression. Many talks highlighted the industry’s needs for technologies that enable continuous and/or real-time control of cell-culture parameters, especially for truly on-line process analytical technology (PAT). Many current systems need to improve interfaces between bioreactors and automated samplers/analyzers. BioPhorum representatives announced publication of a risk-assessment guide for raw materials used in CLD and other upstream processes. And some presenters called for greater attention to compositional chemistry, highlighting how such factors influence cell growth, consumption of media nutrients, and other key parameters.

Boosting Productivity, Viability, Stability, and Predictability
Held on 27 February through 2 March 2023, the BPI West program in San Diego, CA, included a full track devoted to CLD. Many themes followed from the BWB discussions a few months before, but the overall emphasis shifted toward applying advanced genetic-engineering techniques to improve cell stability, productivity, predictability, and viability. Technological potentialities touted over recent years seem to be coming to fruition in big ways in the post-COVID era.

In his keynote address, “Why Do We Need ’Omics, and What Do We Get Out of It?” Nathan Lewis (University of California, San Diego) set the tone for three days of presentations. Having been on the forefront of efforts to sequence the CHO genome, his laboratory continues to develop systems-biology and genome-editing techniques to map and engineer cell pathways that control mammalian cell growth, metabolism, protein synthesis, and glycosylation toward developing improved protein-expression hosts.

“Over the past decade,” he explained in his abstract, “cell-line development and engineering for CHO cells [have] fully entered the genome era wherein tools for studying and optimizing CHO have become easily accessible. However, practitioners have found the utility of the various ’omics technologies to be hit and miss.” To light the way ahead, Lewis focused on how such data can be used to facilitate decision-making in both CLD and bioprocessing.

Using CHO cells to express the entire human secretome, Lewis’s team found that some of those >2,200 genes expressed well and others did not — and results did not necessarily correlate with transgene integration. Size and other features of the proteins made some difference. But the key deciding factor was associated with the secretory pathway of CHO cells, themselves. HEK293 cells performed better for many of the proteins studied. Comparing the secretory “machinery” of both cell lines revealed some host-cell proteins that could be overexpressed to improve results, and that’s what the laboratory is working on now.

“We need to use these models to make ’omics actionable.” Lewis predicted that such studies will be part of future CLD processes, in which data analysis from ’omics experiments provides target genes for implementation in engineered cell lines.

As leader of the CLD programs in Boehringer Ingelheim’s (BI’s) biologics pipeline, Simon Fischer presented “The Power of Transposases — How an Ancient Cellular Mechanism Revolutionized Modern Biopharmaceutical Development.” He pointed out that until just recently in the industry, transgenes were integrated only randomly into host-cell genomes, which always produced highly variable integration patterns. Those results made time-consuming and labor-intensive clone screenings necessary to identifying high-performing and genetically stable cell lines. He outlined how the application of transposase-mediated transgene integration marks a paradigm shift that will accelerate biologics development timelines substantially for the future.

Although “speed to clinic” was increasingly important in biopharmaceutical development over the years leading up to 2020, the COVID-19 pandemic challenged companies to move even faster. Fischer said that transposase technology has proven to be a game-changer for doing so in CLD. “Plug-and-play” technologies are compatible with CHO cell lines and process platforms, providing a means for improving speed, performance, and cell-line stability. BI is taking new projects to 80-L scale and early supply of drug substance in under six weeks — and to 2,000-L scale for good manufacturing practice (GMP) supply in under three months. That provides materials for phase 1 clinical studies (“first in human”) in about six months and GMP supply within nine months.

In “Next-Generation CHO Cell Line Development: Higher Titers, Shorter Timelines, and Optimized Product Quality,” Christoph Zehe (Sartorius) described other potentially game-changing technologies. As worldwide demand for protein therapeutics grows steadily, he said, molecular complexity and product-quality requirements are increasing along with the pressure to reduce both cost and time of development. To address those challenges, his company developed version 5.0 of its 4Cell CHO CLD technology to deliver stable producer clones expressing ≤10 g/L within nine weeks (from DNA to a research cell bank, RCB). Zehe highlighted CellCelector technology for early high-throughput clone generation with a model-based preselection assay for identification of clones with increased productivities at nanowell stage; an automated scale-down model in 96-well plates for broad early screening of clones in fed-batch conditions to select high producers; application of modern gene-editing technologies during CLD to enable optimized product-quality profiles (e.g., afucosylated antibodies for improved bioactivity); and intensified fed-batch process designs that further boost final titers or shorten process duration.

In “Evaluation of Genetic Rearrangements in Site-Specific Integration Cell Lines,” Barbara Tevelev (Pfizer) used two case studies to show that “off-target” genetic rearrangement in site-specifically integrated (SSI) cells either had no effect on product quality, cell-line productivity, or stability — or actually could improve them. When traditionally monitored attributes remain consistent and well controlled, she said, then variation in genotype should be considered acceptable. That’s because it would be expected to have no impact on product quality. Although SSI clones are easier to characterize than their randomly integrated counterparts, Tevelev explained, that does not necessarily mean that they must be characterized to a greater extent.

In “Leap-in to the Future of Cell-Line Development,” Claes Gustafsson (ATUM Bio) included case studies highlighting his company’s integrated platform covering generation of antigens; maturation, developability assessment, engineering, and humanization of antibodies; and stable cell-line generation. He described how each step in the process uses technological breakthroughs in synthetic biology, machine learning, laboratory information management system (LIMS) data integration, robotics, and transposase engineering to help ensure maximum efficiency. The “Leap-in” transposase technology also was highlighted by Simon Fischer (above) and by Valentina Ciccarone of MacroGenics in her presentation, “Cell Line Development for Bispecific Molecules: Random Versus Targeted Integration.”

In “Inducible Systems for Recombinant Protein Expression in CHO Cells,” Anna Klug (Just-Evotec Biologics) pointed out that under pressure to reduce the cost of recombinant protein therapeutics, companies are placing significant focus on increasing cellular productivity. But higher levels of expression, she said, increase metabolic burden while inhibiting growth and cell-line stability, especially if an expressed molecule is toxic to the cells that make it. To address that problem, her company developed an inducible expression system to reduce the metabolic burden and limit the ill effects of toxic therapeutic proteins expressed in CHO cells, ultimately resulting in overall higher expression.

Klug’s group constructed a tetracycline-inducible system that limits protein expression in CHO cells to the production phase, thus obtaining high-expressing pools for both Fc-fusion and MAb proteins. Applying the same technology to an already-developed cell line provided isolated pools with improved protein expression and productivity of a protein that was known to be toxic to the expressing cells. Klug demonstrated the functionality of a tetracycline repression system and other inducible expression systems to improve and expand protein expression capabilities, even for multiple proteins simultaneously.

In “Patching Automation Gaps in Cell-Line Development Workflows,” Anna Venne (Biogen) said that accelerating process development while reducing costs requires high-throughput automation to streamline processes and digitize data operations. As a computer programmer at Biogen, she worked on a suite of tools including Python scripts and an autofeeding program for Ambr 15 systems (Sartorius) to improve robotics integration to that end. “This new repertoire of automation technology enables our scientists to make data-driven decisions by alleviating benchtop time and reducing clerical errors.” Such errors had caused run failures in the past, leading to setbacks and suboptimal clone selections. Autofeeding has eliminated over eight hours of manual operation per run, and the company plans to enable the same method of data transfer and automated entry for pH adjustments in the future.

In “Microfluidic-Chip–Based Single-Cell Cloning To Accelerate Biologic Production Timelines,” Jonathan Diep (Amgen) described his company’s application of the Berkeley Lights Beacon instrument in an early single-cell cloning process to accelerate the CLD timeline by 30% while still identifying clones with acceptable manufacturability. He was followed by Aurora Fabry-Wood (Berkeley Lights), who offered more detail on the microfluidic chips and an automated process for sorting and cloning CHO cells, performing miniaturized assays, and recovering top performers. That method reduces the number of clones that must be expanded and characterized while increasing the likelihood of finding a cell line that makes high titers of manufacturable product. Two case studies demonstrated how miniaturized quality assays enabled selection of top-performing clones within five days of cloning. And Falguni Patel (AbbVie) followed with more case studies about troubleshooting and lessons learned in “Accelerate FIH Timeline for Biologics Pipeline Development with Berkeley Lights Beacon Technology.”

In “Production of Afucosylated Antibodies in CHO Cells By Coexpression of an Anti-FUT8 Intrabody,” Laurence Delafosse (National Research Council of Canada) highlighted the dependency of IgG effector functions on N-glycans in the Fc portion of antibody structures. He described a method for developing next-generation therapeutic antibodies with enhanced effector functions by engineering cells to express an anti-alpha-(1,6)-fucosyltransferase “intrabody” alongside the protein of interest. That intrabody was engineered to reside in the endoplasmic reticulum and Golgi apparatus of the cells, where it efficiently reduces FUT8 activity and therefore limits core fucosylation in the Fc N-glycans of coexpressed antibodies. “This quick and efficient technology allows for production of high levels (>2 g/L) of afucosylated IgGs with strongly enhanced effector function,” said Delafosse, “for which the level of fucosylation can be selected.” Unlike other approaches, he added, this platform doesn’t require special fucose-free media/feeds.

In “Increased Efficiency of Clonal Selection and Shortened CLD Workflows with Solentim ICON,” Kayla Bean (AGC Biologics) described her company’s implementation of the Solentim ICON system from Advanced Instruments into its CLD workflow — which in a familiar refrain, “allowed us to significantly decrease our timelines while increasing our high-throughput screening methods to pinpoint the top-producing clones.” The technology enabled AGC to advance its MAb platform rapidly to higher levels of productivity than were previously possible.

In “Cell-Line Characterization for Cell-Line Selection Using Next-Generation Sequencing and RNA Sequencing,” Christina Tsai (IGM Biosciences) explained that CLD processes have been using proteomic approaches and traditional Sanger sequencing to characterize cell lines for over a decade. As NGS technology has advanced in recent years, characterization of cell lines for bioproduction has become achievable. IGM Biosciences uses such technology to determine the mutations and integration sites of transgenes, then evaluate the genotypes of cell lines to understand correlations between cell growth and productivity. In addition, Tsai said, her group explores epigenetic sequencing to determine the expression and stability of different cell lines.

In “Applications of Next-Generation Sequencing To Benefit the Development and Characterization of Manufacturing Chinese Hamster Ovary (CHO) Cell Lines,” Jie Zhu (AstraZeneca) pointed to the importance of generating clonally derived production cell lines with high fidelity of therapeutic protein sequences to ensure a robust process and consistent product quality going forward. Her team combined NGS with proteogenomic assays to characterize the genetic features of production cell lines. She reported that such studies “enable assessment of the probability of clonal derivation for legacy cell lines, identification of chromosomal integration sites and novel genetic regulatory elements, and detection of low-level single-nucleotide variants (SNVs) and misspliced mRNA species during cell-line development.” Results also provide insights for engineering optimal expression platforms for novel and difficult-to-express proteins. AstraZeneca plans to use the technology in support of regulatory filings —replacing Southern blotting and cDNA analysis on MCBs and cells at the limit of in vitro age (LIVCA) — and to evaluate long-range sequencing technology for transgene structure and integration site analysis. Zhu said that the same technology can be applied to adventitious-agent detection and can be combined with ’omics studies to identify cell-line engineering targets.

In “Evaluating Use of Stable Pool or Pool of Clones for Tox Production To Enable Speed to Clinic,” Jessica Pan (Merck) compared the cell-culture performance and product-quality attributes of three different proteins from stable pools, pools of clones, and top clones for three different molecules. She showed that using stable pools or pools of clones for early supply of toxicology material shortened Merck’s timeline to first-in-human studies by about four months. Now Pan’s team is working to demonstrate pool stability over time, to perform large-scale confirmation runs and multiple batches to demonstrate consistency in process performance and product quality, and to demonstrate comparability among clonally derived cell banks and those generated from stable pools.

In “Population Dynamics, Phenotypic Heterogeneity, and Age: Shifting Expression Patterns in Stable and Unstable Clonally Derived CHO Populations,” Theodore Peters (Seagen) highlighted the problem with clonality that many experts have argued about in recent years as regulators have emphasized it. “CHO cells lines have significant phenotypic variability,” he said, “even when derived from a single cell progenitor.” Such variability correlates with the propensity of some cell lines to exhibit production instability over time. Peters and his colleagues characterized RNA expression from both stable and unstable cell lines using single-cell RNA sequencing. Results showed that even clonally derived cell lines are complex metapopulations of cells whose make-up changes significantly over several generations.

Achieving the Unachievable
The last day of the CLD track in San Diego ended with some impressive forward-looking talks from two young technology companies founded by industry veterans. They shared data that hinted at where cell-line engineering and productivity may be going in the near future.

Based in Hong Kong, Great Bay Bio combines biotechnology and information technology toward making significant improvements in bioprocessing productivity. The company combines robotics with artificial intelligence (AI) and machine learning tools — trained on millions of internally generated data points — with methods of site-specific integration and cell culture media development. The result is precision adjustment of cell-based production.

In “AI + Site-Specific Integration Technology = Predictable Cell Pool for Drug Development,” Michael Chen (Great Bay Bio) described how his company uses AI-enabled platforms to predict real-world results — and he showed how those results have confirmed the predictions. The software scans hundreds of thousands of individual cells within minutes to identify top performers, which are collected automatically one at a time for clonal culturing. That shortens CLD time while predicting product quality, making early evaluation of feasibility and risk reduction possible for licensees and service clients.

Larry Forman (CHO Plus) was a process developer at Genentech long before it was part of the Roche Group. His talk recalled the message of Nathan Lewis’s keynote two days before, highlighting what may be a key reason for CHO cells’ naturally limited productivity. In “Novel Cell Engineering Platform for Creating High-Productivity Cells for Therapeutic Protein Production and for Other Purposes,” he demonstrated the utility and versatility of a platform that can provide a 5- to 10-fold increase in stable protein productivity (to 117 pg/cell/day) over 60 generations. It’s general knowledge in CLD, he pointed out, that just adding more copies of a transgene does not necessarily bring higher productivity. “At some point, the cells just can’t produce anymore. In fact, they might suffer from all those additional copies.”

The bottleneck is in translation of those genes. It turns out that CHO cells are naturally deficient in endoplasmic reticuli. Through repeated cell fusions, CHO Plus randomly amplifies whole chromosomes, then screens for cells presenting the greatest numbers of endoplasmic reticuli — which gives them ability to make more proteins. Forman showed that the process has yielded expression titers >20 g/L in large-scale, fed-batch CHO cell cultures — and 9× increases in adenoassociated virus productivity of HEK293 cells, “with the potential for up to 100× improvement.” He also showed growth improvements by >50% for two engineered CHO lines, both of which achieved higher cell concentration and longer sustained viability in culture.

A future of double-digit baseline titers — or even more? Cell-culture process and media optimization methods have brought the industry this far. Precision engineering could bring unprecedented advancements that heretofore were considered to be unachievable.

1 Gazaille B, Scott C. Introduction: Cell-Line Development Discussion at Biotech Week Boston. BioProcess Int. 21(4) 2013: S1–S4, S16;

Further Reading
Aziz RK, Breitbart M, Edwards RA. Transposases Are the Most Abundant, Most Ubiquitous Genes in Nature. Nucl. Acids Res. 38(13) 2010: 4207–4217;

Grabundzija I, et al. Comparative Analysis of Transposable Element Vector Systems in Human Cells. Molec. Ther. 18(6) 2010: 1200–1209;

Gustafsson C, et al. Engineering Genes for Predictable Protein Expression. Prot. Expr. Purif. 83(1) 2012: 37–46;

Lewis N, et al. Genomic Landscapes of Chinese Hamster Ovary Cell Lines As Revealed By the Cricetulus griseus Draft Genome. Nat. Biotechnol. 31, 2013: 759–765;

Lin D, et al. CHOmics: A Web-Based Tool for Multi-Omics Data Analysis and Interactive Visualization in CHO Cell Lines. PLOS Comput. Biol. 22 December 2020;

Rajendra Y, Peery RB, Barnard GC. Generation of Stable Chinese Hamster Ovary Pools Yielding Antibody Titers of Up to 7.6 g/L Using the PiggyBac Transposon System. Biotech. Prog. 32(5) 2016: 1301–1307;

Rupp O, et al. A Reference Genome of the Chinese Hamster Based on a Hybrid Assembly Strategy. Biotechnol Bioeng. 115(8) 2018: 2087–2100;

Ivics Z, Izsvák Z. A Whole Lotta Jumpin’ Goin’ On: New Transposon Tools for Vertebrate Functional Genomics. Trends Genetics 21(1) 2005: 8–11;

Cheryl Scott is cofounder and senior technical editor of BioProcess International (part of Informa Connect Life Sciences); 1-212-600-3429; You can access video recordings of many BPI conference presentations — along with more video content from the Informa portfolio — online at