HEK293 vs CHO Cells for Protein Expression

Why Cell Line Choice Matters in Mammalian Protein Expression

Selecting a mammalian expression system is not a default decision. The cell line you choose influences post-translational modifications (PTMs), secretion efficiency, protein folding fidelity, downstream purification complexity, and—critically for therapeutic candidates—the immunogenic potential of the final product. Both HEK293 and CHO cells are capable of producing complex, correctly folded, disulfide-bonded glycoproteins that prokaryotic or insect systems cannot reliably deliver. The question is which system best matches your protein's biology and your project's timeline.

For proteins that require human-type sialylation or whose activity is sensitive to specific N-glycan structures, the glycosylation differences between HEK293 and CHO are not cosmetic—they can alter receptor binding affinity, serum half-life, and Fc effector function. For proteins where glycosylation is largely irrelevant (e.g., non-glycosylated cytokines expressed for structural studies), throughput and cost per milligram often dominate the decision. Understanding the biology of each line first makes that cost-benefit calculation straightforward.

Biology of HEK293 Cells

Origin and Genomic Context

HEK293 (Human Embryonic Kidney 293) cells were established in 1973 by Frank Graham through adenovirus 5 (Ad5) transformation of primary human embryonic kidney cells. The "293" designation refers to the number of the experiment in Graham's notebook. The line retains several neuronal gene expression signatures, which is somewhat unexpected given its kidney origin—a detail that occasionally affects the background expression of signaling proteins in certain assay contexts. Importantly, HEK293 cells carry approximately 64 chromosomes per cell, reflecting the chromosomal instability introduced during transformation.

Multiple HEK293 sub-lines exist. HEK293T contains the SV40 Large T antigen, enabling episomal replication of vectors carrying the SV40 origin of replication and dramatically increasing transient expression titers. HEK293E carries the Epstein-Barr virus nuclear antigen 1 (EBNA1), serving a similar role for oriP-based vectors. Expi293F is a suspension-adapted variant optimized for high-density culture. Each sub-line has distinct utility, but they all share the core human glycosylation machinery described below.

Growth Characteristics

• Doubling time: ~24 hours under standard adherent conditions; suspension-adapted variants (e.g., Expi293F) maintain similar kinetics at densities up to 3–5 × 10⁶ cells/mL.
• Culture format: Adherent growth on tissue-culture plastic is standard; suspension adaptation is readily achieved and commercially available as pre-adapted lines.
• Transfection efficiency: Among the highest of any mammalian cell line—routinely >90% with standard polyethylenimine (PEI) or lipid-based reagents.
• Serum dependency: Standard 293 cells grow in DMEM + 10% FBS; suspension variants use chemically defined, serum-free media, removing lot-to-lot variability concerns.
• Temperature optimum: 37°C, 5% CO₂; lowering to 32–34°C post-transfection reduces apoptosis and is frequently used to extend the production window during transient runs.

Biology of CHO Cells

Origin and Genomic Context

Chinese Hamster Ovary (CHO) cells were derived in 1957 by Theodore Puck from a biopsy of the ovary of an adult Chinese hamster (Cricetulus griseus). The original line was proline-auxotrophic (CHO-Pro⁻), a property later exploited for dihydrofolate reductase (DHFR)-based gene amplification. The CHO genome is highly plastic—CHO cells have a modal chromosome number of approximately 20, but significant copy-number variation and chromosomal rearrangements are common across sub-lines and after prolonged culture. This genomic instability has practical consequences: stable CHO clones must be screened rigorously to identify high-producing, genetically stable clones.

Key CHO sub-lines used in protein production include CHO-K1 (parental), CHO-DG44 (DHFR-deficient, used with DHFR selection/amplification), CHO-DXB11 (hemizygous DHFR deletion), and CHO-S (suspension-adapted). GS-CHO systems using glutamine synthetase (GS) as a selectable marker are widely used in industry, particularly following licensing of the GS system originally developed at Lonza.

Growth Characteristics

• Doubling time: ~18–24 hours; some suspension-adapted lines achieve shorter doubling times under fed-batch conditions.
• Culture format: Originally adherent; industrially relevant sub-lines (CHO-S, CHO-DG44) are fully suspension-adapted and can be scaled from shake flasks to 20,000-liter bioreactors.
• Transfection efficiency: Lower than HEK293 for standard transient transfection—typically 30–70% depending on reagent and sub-line. This makes CHO less attractive for rapid transient screening.
• Scalability: Extensively validated in fed-batch and perfusion bioreactor modes; specific productivity (qP) values of 20–60 pg/cell/day are achievable in optimized stable clones.
• Regulatory history: CHO cells have been used to manufacture licensed biologics since Activase (tissue plasminogen activator) was approved in 1987. Over 70% of approved recombinant therapeutic proteins are produced in CHO cells, providing an established regulatory framework.

Glycosylation: The Critical Biochemical Difference

N-linked glycosylation is the PTM most likely to differ consequentially between HEK293 and CHO systems. Both cell lines perform core N-glycan addition in the endoplasmic reticulum and subsequent processing in the Golgi, but the terminal glycan structures they produce diverge in ways that affect protein activity, immunogenicity, and pharmacokinetics.

HEK293 Glycosylation Profile

HEK293 cells express the full complement of human glycosyltransferases, including ST6Gal-I, which adds α-2,6-linked sialic acid (N-acetylneuraminic acid, Neu5Ac) to N-glycans. Human serum glycoproteins predominantly carry α-2,6-linked sialic acids. HEK293-produced proteins therefore closely replicate the sialylation pattern of endogenous human proteins. Additionally, HEK293 cells do not express a functional CMP-N-acetylneuraminic acid hydroxylase (CMAH) gene, meaning they do not incorporate N-glycolylneuraminic acid (Neu5Gc)—a non-human sialic acid that can trigger anti-Neu5Gc immune responses in humans.

Key HEK293 glycan features:

• Predominantly complex-type bi- and tri-antennary N-glycans with terminal α-2,6 sialylation
• Absence of Neu5Gc (human-compatible)
• Low levels of high-mannose glycoforms compared to some expression systems
• O-glycosylation patterns that closely mirror human cell O-glycosylation

CHO Glycosylation Profile

CHO cells lack ST6Gal-I activity and instead express ST3Gal enzymes, producing predominantly α-2,3-linked sialic acids rather than the α-2,6-linked form found on human glycoproteins. CHO cells also express a low level of functional CMAH, meaning a fraction of sialic acids can be present as Neu5Gc—though this is generally at low levels and has not prevented regulatory approval of CHO-derived therapeutics. CHO cells do not synthesize the immunogenic Galα1-3Gal (alpha-gal) epitope and do not add bisecting GlcNAc at the level seen in some other expression systems.

Key CHO glycan features:

• Predominantly α-2,3-linked sialylation (differs from human serum glycoproteins)
• Trace levels of Neu5Gc possible
• Absence of alpha-gal epitope
• Glycan heterogeneity can be reduced through media optimization and cell line engineering
• Glycoengineered CHO lines (e.g., expressing ST6Gal-I) have been developed to produce more human-like α-2,6 sialylation

When Glycosylation Differences Matter

For most research-grade proteins—where the goal is to study binding, measure activity, or generate structural data—the α-2,3 vs. α-2,6 sialylation difference between CHO and HEK293 is unlikely to affect your conclusions. However, the difference becomes significant in the following contexts:

• Fc-fusion proteins and antibodies: IgG1 Fc contains a conserved N-glycan at Asn297 whose composition modulates FcγR and complement binding. Both cell lines produce acceptable Fc glycoforms for most research purposes, but specific effector function studies may require careful characterization.
• Proteins with lectin-binding activity or partners: Sialic acid linkage type can alter galectin and siglec binding, affecting results in cell-based assays.
• Half-life studies: The asialoglycoprotein receptor (ASGPR) recognizes terminal galactose exposed upon desialylation; proteins with higher sialic acid content (as seen in HEK293 products) may show extended serum half-life in functional assays.
• Therapeutic development candidates: Regulatory agencies expect glycan characterization; choosing HEK293 for early-phase material can simplify comparability arguments if the eventual commercial process also uses a human-compatible glycosylation system.

Transient vs. Stable Expression: Speed, Yield, and Cost

Transient Expression

In transient transfection, plasmid DNA enters cells and is expressed from non-integrated episomal copies. Expression peaks within 48–96 hours and typically declines over 7–14 days as the episomal DNA is diluted through cell division or degraded. Transient expression is fast—from sequence-confirmed plasmid to purified protein in 2–3 weeks is routine—and requires no clone selection. It is the standard approach for small-scale functional screening, structural biology sample preparation, and producing proteins where milligram quantities suffice.

HEK293 for transient expression: HEK293T and Expi293F are the dominant platforms for high-titer transient production. Expi293F cultures transfected with ExpiFectamine can routinely yield 250–1000 mg/L of secreted IgG-format proteins, with some reports exceeding 1 g/L under optimized conditions. For non-antibody proteins, yields of 10–200 mg/L are typical depending on the construct. The high transfection efficiency of HEK293 cells means that protein is detectable within 24 hours and peak harvest can be taken at 5–7 days post-transfection.

CHO for transient expression: Transient CHO expression (e.g., using ExpiCHO-S) has improved substantially with the availability of optimized reagents. ExpiCHO-S transfections can achieve titers comparable to Expi293F for antibodies, and some reports show superior performance for certain glycoprotein classes. However, the workflow is slightly more demanding in terms of media optimization, and transient CHO is used less routinely than transient HEK293 in most research laboratories.

Stable Cell Line Development

Stable expression involves integration of the transgene into the host genome, followed by selection and clonal expansion of high-producing cells. This process typically takes 8–16 weeks from transfection to a characterized, research-grade stable pool or clonal cell line. The advantages are consistency, scalability, and—for manufacturing—the ability to build a cell bank with defined characteristics.

CHO for stable expression: CHO is the industry standard for stable biopharmaceutical production. DHFR- and GS-mediated gene amplification systems can drive specific productivity to levels that make large-scale fed-batch processes economically viable. Stable CHO clones have been scaled from 2-liter bench bioreactors to 20,000-liter manufacturing vessels with maintained productivity profiles. The extensive regulatory and process development database for CHO stable cell lines is a significant practical advantage when developing a clinical-stage biologic.

HEK293 for stable expression: Stable HEK293 lines can be generated and are used in research settings, but achieving the specific productivity levels and genetic stability required for manufacturing is less well-established than in CHO. For research applications requiring stable, continuous supply of a protein without the full regulatory framework of a biopharmaceutical program, stable HEK293 pools are a viable option.

Head-to-Head Comparison Table

Applications Where Each System Has a Clear Advantage

HEK293 Protein Production: Where It Excels

• Structural biology: Cryo-EM and X-ray crystallography projects often require milligram quantities of homogeneous, correctly folded glycoprotein on a short timeline. Transient HEK293 expression meets this need without the lead time of stable cell line development.
• Protein interaction and binding studies: When assaying binding to human cell-surface receptors or serum components where glycan-lectin interactions may be relevant, the human-type glycosylation of HEK293 reduces the risk of artifact.
• Assay development antigen production: Producing a glycoprotein antigen for ELISA, SPR, or cell-based assay development where human PTMs are needed for antibody cross-reactivity testing.
• Early-phase antibody discovery campaigns: Rapid antigen production for immunization or screening, where turnaround speed outweighs any benefit of stable expression.
• Proteins sensitive to over-expression stress: HEK293 cells generally handle transient over-expression well, with relatively low basal unfolded protein response compared to some other mammalian lines.

CHO Protein Production: Where It Excels

• Biopharmaceutical manufacturing: When a project is progressing toward IND-enabling studies or clinical manufacturing, establishing a stable CHO cell line with a defined production process is the expected path. The regulatory database for CHO is unmatched among mammalian expression systems.
• Large-scale continuous production: Stable CHO fed-batch and perfusion processes can deliver product titers of 3–10 g/L in optimized manufacturing formats, making them economically viable for commercial-scale production.
• Long-term supply agreements: When a consistent, characterized supply of a recombinant protein is needed over years—for reagent supply, reference standard production, or commercial diagnostics—stable CHO cell banks provide the required consistency.
• Proteins requiring gene amplification: For constructs that are difficult to express at high levels, DHFR or GS-mediated amplification in CHO can rescue productivity in ways that have fewer established analogues in HEK293 systems.

Practical Considerations for Choosing Between Systems

Construct Design

Vector backbones differ between systems. For HEK293T transient expression, vectors with the SV40 origin (recognized by the Large T antigen) provide episomal replication and higher copy number. For ExpiCHO-S transient runs, standard CMV-promoter vectors without SV40 ori are used. Signal peptides also matter: the mouse Ig kappa leader and human IL-2 signal peptide generally perform well in both systems, but it is worth testing two signal sequences in parallel if secretion efficiency is suboptimal.

Protein Complexity and PTMs

If your protein of interest contains multiple disulfide bonds, GPI anchors, or complex O-glycosylation (e.g., mucin-type), both HEK293 and CHO are capable platforms, but HEK293 may more faithfully reproduce human-specific O-glycan core structures given its human cellular machinery. For proteins requiring specific chaperone-assisted folding that is known to vary by cell type, a small pilot transfection in both systems followed by analytical comparison (SEC, native PAGE, functional assay) is more informative than any a priori rule.

Downstream Purification

Both cell lines can be grown in chemically defined, serum-free media, simplifying downstream processing. CHO cells tend to release lower levels of host cell DNA per unit volume in optimized processes, which can simplify purification. HEK293 cells grown in serum-free suspension media present a similar profile. For Fc-containing proteins (antibodies, Fc-fusions), Protein A affinity chromatography captures product from both expression systems with equivalent efficiency. For non-Fc proteins, the purification tag (His, Strep, FLAG) and the chromatography strategy are independent of the expression host. A practical overview of downstream purity assessment can be found in our guide to activity-based protein characterization assays.

Quality Control and Characterization

Regardless of which mammalian expression system you use, secreted recombinant proteins should be characterized for: (1) molecular weight and aggregation state by SEC-HPLC or analytical ultracentrifugation; (2) glycan profile by LC-MS or HILIC-HPLC if glycosylation is functionally relevant; (3) endotoxin level by LAL assay if the protein will be used in cell-based assays; and (4) biological activity by a relevant functional assay. Skipping glycan characterization when comparing results obtained from HEK293 vs. CHO-produced material is a common source of irreproducible data in the literature. For reference reagents and characterized recombinant proteins from our in-house production pipeline, see the Triple Point Biologics recombinant protein catalog.

HEK293 and CHO Expression at Triple Point Biologics

Triple Point Biologics offers custom recombinant protein production in both HEK293 and CHO mammalian expression systems, selected based on the requirements of the specific project. For research-grade material needed on a 2–4 week timeline, transient HEK293 (or transient ExpiCHO) expression is the standard starting point. For projects requiring stable, long-term supply, a scaled manufacturing process, or a CHO-based cell line that can be transferred into a GMP-compatible workflow, CHO stable expression is the appropriate path.

Our team works from gene synthesis or customer-supplied sequence through vector construction, expression screening, protein production, and purification to a customer-defined specification. Typical deliverables include SEC purity data, SDS-PAGE under reducing and non-reducing conditions, endotoxin quantification, and a functional activity result where a validated assay exists. Expression system selection is part of the initial project scoping conversation, informed by the protein class, required quantity, timeline, and downstream application. Explore our custom protein production services for a full description of process options.

For antibody-based detection reagents—including anti-protease and anti-inhibitor antibodies validated for Western blot—Triple Point Biologics has maintained production and validation programs since 1994. Details on available validated antibody targets are listed in the antibody catalog.

Frequently asked questions