How to choose and build a high-performing cell line
A practical guide to host selection, the CHO lineage, gene-integration systems, and modern clonal cell line development
Authors: Jack Crawford (Demeetra) & Magnus Gustafsson (NorthX Biologics) | June 2026
Summary
This article provides an overview of CHO cell line development for recombinant protein production, with a focus on monoclonal antibodies and other complex biologics. It covers expression system selection, the development and relevance of CHO cells, and key gene integration strategies used in modern cell line engineering.
It also outlines how contemporary clonal cell line development is performed, including selection, screening and stability assessment, and describes the benefits of integrating cell line development with downstream process development and GMP manufacturing. This integrated approach can reduce timelines, minimise technology transfer risk and improve continuity across the development program.
Introduction
Cell line development (CLD) is the process of creating a stable cell line for recombinant protein production. The DNA encoding a therapeutic protein is introduced into a host cell, and through rounds of selection, screening and characterization a single clone is identified that produces the molecule efficiently, stably, and with the right quality attributes. Because the stable cell line becomes the factory for the entire commercial life of the drug, decisions made here lock in titer, product quality, genetic stability and cost of goods for a decade or more.
1. Choosing an expression system for recombinant protein production
Four host systems account for most recombinant protein manufacturing today. The right choice depends on the molecule, its required post-translational modifications, the scale, and the regulatory path.
E. coli expression system: high yield for non-glycosylated proteins
Bacterial expression launched the biotech industry. Recombinant human insulin became the first rDNA therapeutic approved by the FDA in 1982, followed by recombinant human growth hormone in 1985, both produced in E. coli.
E.coli is fast, inexpensive and yields very high production levels, but it does not naturally glycosylate proteins and often deposits product in inclusion bodies. It remains ideal for small, non-glycosylated molecules (insulin, peptides, and antibody fragments), but not for full-length antibodies.
Yeast expression systems: scalable production with non-human glycosylation
Yeasts such as Saccharomyces cerevisiae and Pichia pastoris grow to high cell densities at low cost and efficiently secrete proteins. Yeast has been used for recombinant insulin and vaccine antigens.
Its native glycosylation, however, is hyper-mannosylated and not human-like without extensive engineering, which limits its use for therapeutic antibodies.
HEK293 cells: human-like expression for complex proteins
HEK293 is a human embryonic kidney line created in Alex van der Eb’s laboratory and characterized by Frank Graham (J. Gen. Virol. 1977); the “293” refers to Graham’s 293rd experiment. Because it is human, it produces the most authentically human glycosylation, which makes it the preferred host for many gene therapies, viral vectors, and complex proteins. For conventional antibodies, it generally yields less than CHO and has historically drawn more regulatory scrutiny as a manufacturing host.
CHO cells: the leading platform for therapeutic protein production
Chinese Hamster Ovary cells produce roughly 70% of all approved recombinant therapeutic proteins. The first CHO-derived therapeutic was Activase (alteplase, recombinant tissue plasminogen activator), approved by the FDA in November 1987, the molecule that established mammalian cell culture as a viable manufacturing platform.
CHO remains the dominant platform due to several factors: it performs human-compatible glycosylation; it grows to very high density in chemically defined, serum-free suspension culture and scales to bioreactors of 20,000 L; it is refractory to most human viruses, a major safety advantage; it has mature, well-understood selection systems (DHFR/methotrexate amplification and glutamine synthetase); and regulators have reviewed hundreds of CHO filings, so the path is well paved. Modern CHO processes routinely reach 5–10 g/L.
When to use CHO cells in biopharmaceutical development
For full-length monoclonal antibodies, multi-specifics, Fc-fusions and most recombinant glycoproteins destined for human clinical trials, CHO is the default: it offers the best combination of titer, scalability, human-like product quality and regulatory familiarity. E. coli is preferred for simple non-glycosylated proteins, and HEK293 for gene therapy vectors and certain complex human proteins. For a typical antibody program, CHO is almost always the answer.
2. CHO cell line history and advances in GS knockout technology
The CHO story begins in 1957, when Theodore Puck isolated cells from the ovary of a Chinese hamster and established the original CHO line. Over the following decades, derivatives were adapted for industrial use: CHO-K1, the DHFR-deficient CHO-DG44 and CHO-DXB11 lines that enabled methotrexate-driven gene amplification, suspension-adapted CHO-S, and glutamine-synthetase (GS) knockout host lines.
Selection chemistry drove much of this evolution. The GS system uses glutamine synthetase as a metabolic selection marker: cells survive in glutamine-free medium only if the transfected GS gene is active, linking survival to expression of the gene of interest. Because CHO cells carry their own endogenous GS, early GS systems needed the inhibitor methionine sulfoximine (MSX) to suppress background. Knocking out the endogenous GS gene removes that need, giving cleaner, more stringent selection and faster isolation of high producers.
The state of the art today is a double GS knockout. Recent work (Limia et al., Biotechnology Progress, 2026) showed that CHO-K1 carries not only the primary GS gene on chromosome 5 but also a GS pseudogene on chromosome 1 that can contribute residual activity. Using its high-fidelity Cas-CLOVER editing system, Demeetra sequentially knocked out both loci to create the first commercially available double GS knockout CHO host (CleanCut GS CHO), with no off-target mutations detected across 40 predicted sites. The result is exceptionally stringent selection — only genuinely high-expressing cells survive — which shortens screening and improves the odds of finding stable, high-titer clones. Recently, Demeetra has achieved 18g/L in a pool expressing trastuzumab.
3. Gene integration strategies in CHO cell line development
Photo provided by Demeetra
Transposon-based gene integration is preferred in CHO cell line development because transposons preferentially integrate into accessible, open chromatin rather than tightly packed genomic regions. This increases the likelihood of transcriptionally active integration sites, supporting higher titers and the generation of stable, high-producing cell lines.
Random integration
The classic approach: a plasmid carrying the gene of interest plus a selection marker is transfected, integrates at unpredictable genomic locations, and selection pressure kills cells that did not integrate. Its strength lies in simplicity and the chance of finding rare, exceptionally high producers. However, that simplicity is offset by a substantial downstream burden: the very low efficiency turns screening into a numbers game which can consume months. Frequent genetic instability means the program may need to fall back to earlier clones or, in a worst-case scenario, repeat major portions of the clone isolation and screening process, adding significant time, cost, and uncertainty to development timelines.
Transposons
Transposon systems use an enzyme (transposase) to cut a defined cassette flanked by target sequences called inverted terminal repeats (ITRs) and paste it into the genome. Compared with plasmid random integration, transposons integrate intact, multi-copy cassettes more efficiently into transcriptionally active sites, giving higher-expressing, more uniform pools faster. They are the pragmatic middle ground between random integration and fully targeted approaches and are widely used for early material and for stable manufacturing lines (Section 4).
Targeted integration (landing pads)
Targeted integration places the gene at a pre-validated genomic “hot spot”, a landing pad, using recombinases (Flp/FRT, Cre/lox) or CRISPR/Cas9. This sharply reduces clone-to-clone variability and shortens screening, because expression is governed by a known, favourable locus. The trade-off is extremely low integration efficiency, which extends timelines and limits access to stable pools for early-stage material needs. Although clone-to-clone variation is reduced, single-cell cloning remains required, and inserting only one or two gene copies often results in limited titer and productivity.
Most next-generation platforms have moved away from pure random integration toward transposon-based or targeted strategies; trading unpredictability for more consistent integration site quality and expression. The major transposon systems in active use each have different mechanisms achieving that, with meaningful differences in efficiency, cargo capacity, and especially IP accessibility.
4. Transposon technologies used in cell line development
Several transposon/transposase pairs are used in biopharmaceutical CLD. They differ in cargo capacity, integration behaviour and — importantly for commercial programs — licensing terms.
Sleeping Beauty (SB): a reconstructed fish transposon and the first active synthetic transposon in vertebrates; well characterised, with a relatively modest cargo capacity. SB has a known tendency to integrate into transcriptionally silent regions of the genome, resulting in lower expression and reduced titer, a limitation that has led to its declining use in CLD. The system also carries licensing requirements.
PiggyBac (and hyperactive Super PiggyBac): derived from a moth transposon; carries large cargo, integrates in multiple copies, and leaves a precise, seamless excision footprint. It is one of the most efficient transposon systems; however, commercial rights and access for CLD applications carry complex, restrictive, and costly licensing terms that limit its availability.
Leap-In Transposase: engineered from a frog transposon; has established IND filings and regulatory precedent, and has demonstrated competitive titers in some programs. However, the system requires significantly higher copy numbers per cell to achieve titers comparable to piggyBac and Harbor-IN, which raises questions about integration site quality. Licensing terms are also expensive, adding cost and complexity for CLD applications.
Harbor-IN: a codon optimized mRNA format license-free transposase that integrates large payloads at high efficiency and supports high-titer stable expression. Harbor-IN has comparable efficiency and integration profile to piggyBac, but available for unrestricted commercial use, with no royalties, milestones, or annual fees.
5. Modern cell line development workflow for monoclonal antibodies
A contemporary antibody CLD campaign follows a defined sequence, with clonality assurance and stability built in from the start:
- Vector design and host selection: codon-optimized heavy- and light-chain cassettes are built into the chosen integration vector(s) and paired with an appropriate host (e.g. a GS-knockout CHO line).
- Transfection and selection: DNA and transposase components are delivered by electroporation or lipid reagents; metabolic or antibiotic selection eliminates non-integrants. With a double GS knockout host, glutamine-free selection is rapid and highly stringent.
- Single-cell cloning with documented clonality: pools are deposited as single cells using FACS, microfluidic dispensers or limiting dilution, with high-resolution day-0 imaging providing the documented assurance of monoclonality that regulators require.
- Clone screening and ranking: hundreds to thousands of clones are screened for titer, growth and product quality; increasingly, omics and machine-learning tools help rank clones from limited early data.
- Stability and productivity assessment: top clones are passaged 60+ generations to confirm that titer and product quality are maintained over the production timeline.
- Research / master cell bank and characterisation: the lead clones are banked and characterised (identity, glycan profile, aggregate and charge variants), and the data package supports the IND.
With optimized platforms, what once took 9–12 months from DNA to research cell bank can now be done in 3–4 months.
6. From cell line development to GMP manufacturing: an integrated CDMO approach
At NorthX Biologics, we integrate Demeetra‘s CLD technology directly into our CDMO workflow, so a monoclonal antibody program moves from gene to GMP drug substance under continuous ownership. The CleanCut GS CHO double-knockout host paired with the Harbor-IN transposase has delivered fed-batch titers up to 18 g/L and cell-specific productivity above 100 pg/cell/day, with stability maintained over extended passaging.
Running CLD and downstream development together removes the technology-transfer handoffs that typically sit between a cell-line vendor, a process-development group and a manufacturer (each of which would add weeks of re-validation and risk of lost knowledge). Keeping the work in one organization ensures continuity of process understanding, allows for analytical methods and quality systems to carry across phases, and can compress early-phase timelines by several months. This is relevant because it’s often the difference between hitting or missing an IND date.
The commercial structure matters as much as the biology. Since both CleanCut GS CHO and Harbor-IN are license-free, a program built on this platform is far simpler to manage, and carries no royalties, milestones or annual fees on the cell line. Manufacturing costs and commercial terms are contained within the service agreement, not attached to the therapeutic asset, preserving the full commercial value of the program as it advances. For a high-titer, clean-selection CHO system, that combination of strong technical performance and unencumbered commercial terms is a meaningful advantage when planning a full antibody program.
Integrating cell line development with GMP manufacturing enables faster, more robust development and reduces overall program risk.
References
1. Puck, T.T., Cieciura, S.J., & Robinson, A. (1958). Genetics of somatic mammalian cells. III. Long-term cultivation of euploid cells from human and animal subjects. Journal of Experimental Medicine, 108(6), 945–956. https://doi.org/10.1084/jem.108.6.945
2. Graham, F.L., Smiley, J., Russell, W.C., & Nairn, R. (1977). Characteristics of a human cell line transformed by DNA from human adenovirus type 5. Journal of General Virology, 36(1), 59–74. https://doi.org/10.1099/0022-1317-36-1-59
3. U.S. FDA / FDA history sources. Humulin approval, 1982; Genentech. Protropin approval, October 18, 1985.
4. Smithsonian / FDA regulatory sources. Activase/alteplase approval, November 13, 1987.
5. Gomez Limia, C. et al. (2026). Sequential, chromosome-specific glutamine synthetase double knockout with Cas-CLOVER establishes enhanced CHO platforms for cell line development. Biotechnology Progress. https://doi.org/10.1002/btpr.70113
6. Demeetra AgBio. (2026). Harbor-IN: License-Free Transposase for High-Titer Biologics Production. Company product information.