CHO vs HEK293 for Early BsAb Expression

Bispecific antibodies (BsAbs) add a second heavy and light chain, so the expression decision is not just "which cell line" — it is how much material, how fast, and how much mispairing the purification can tolerate. The two standard answers are transient HEK293 (screening-grade, days) and stable CHO (development-grade, months). Use them as a sequence, not a competition.

Decision 1 — Transient HEK293 for screening (speed)

Choose transient HEK293 (Expi293) when you need screening-grade BsAb in days, not months, across a panel of variants. DNA in → purified protein out is typically 7–14 days (typical range: ~5–7 days transfection-to-harvest plus ~2–5 days purification). Typical yields are 10–300 mg/L — enough for binding, potency, and early stability screens, not for PK/tox work.

• Choose Expi293 over plain HEK293 for higher titers and reproducibility at small scale: it is a suspension-adapted, high-transfection-efficiency line built for this niche.
• Cost is low upfront — no cell-line development, no months of lead time — though per-litre reagent cost exceeds CHO at scale.
• Caveat: HEK293 glycosylation resembles but is not identical to CHO; keep the system consistent across a screen or Fc-effector comparisons become confounded.

Decision 2 — Stable CHO for development (quality/GMP path)

Choose stable CHO when the molecule has survived screening and you need development-grade material: consistent quality, manufacturability, and a GMP path. Cell-line development (transfection → single-cell cloning → clone selection → banks → production) typically runs 3–6+ months, and stable clones typically reach 1–5+ g/L (typical ranges; transient runs usually stay below ~1 g/L). At scale, CHO amortizes cost and cuts lot-to-lot variation — the property that makes early PK/tox and comparability studies interpretable.

• Choose CHO-S / ExpiCHO when you want to probe manufacturability early or when the final molecule demands CHO-compatible glycan control (e.g., fucosylation strategies).
• Glycosylation is the quality lever: CHO glycoengineering is mature and regulatory precedent overwhelmingly favors CHO product. If the destination is GMP, the stable CHO decision is *when*, not *whether*.

Decision 3 — Chain-pairing formats and their expression consequences

BsAbs must pair two different heavy chains and — unless designed otherwise — two different light chains. The format decides how much mispairing burden the expression and purification system must absorb.

• Knob-into-hole (KiH): CH3 mutations force heavy-chain heterodimerization, but light-chain mispairing remains unless combined with a shared light chain or CrossMab-type swaps.
• Common light chain: one shared LC eliminates light-chain mispairing by design. Choose it when both arms can tolerate a single LC (e.g., arms from the same germline family); combine with KiH for the heavy chains.
• Fab-arm exchange (FAE): two half-antibodies with matched CH3 mutations are expressed separately and assembled *in vitro*, giving near-homogeneous product. Choose FAE when the arms need different light chains and the half-molecules are stable.

Practical consequence: formats with two different light chains (no FAE, no CrossMab) waste yield on mispaired byproducts and need heavier purification. Expect SEC-purified monomer ≥90% for well-behaved formats; if you cannot reach it transiently, fix the format *before* committing 6 months of CHO development — aggregation and mispairing amplify with scale, they do not improve.

Decision 4 — Practical go/no-go rules (yield, aggregation, timeline budget)

Cost rule of thumb: transient HEK293 is cheaper in total for the first gram(s); stable CHO becomes cost-efficient once you need grams and reproducible lots.

Escalate when: the timeline budget is exceeded, yields stay <10 mg/L or monomer stays <80% across ≥2 construct iterations, or material is needed for in vivo/tox studies — that is the point to bring in a CDMO or dedicated stable-cell-line provider.

What a service deliverable should include

A BsAb expression service should return more than a vial. Minimum deliverable:

• Purified BsAb — SEC-purified monomer, with yield (mg/L and total mg), concentration, volume, and endotoxin (≤1 EU/mg if intended for in vivo use).
• Basic QC — reducing/non-reducing SDS-PAGE or CE confirming chain composition, SEC chromatogram (aggregation), intact mass spec confirming correct pairing (key for BsAbs), and a functional binding readout (ELISA or SPR) against both targets.
• Full provenance — expression system (Expi293 vs CHO-S), production scale, construct and format details, and a COA. No expression data, no COA, no deal.

References

• [Ridgway et al., 1996 — Knobs-into-holes engineering of antibody CH3 domains (PubMed)](https://pubmed.ncbi.nlm.nih.gov/8969519/)
• [Labrijn et al., 2014 — Controlled Fab-arm exchange for stable bispecific IgG1 (Nat Biotechnol)](https://www.nature.com/articles/nbt.2821)
• [Thermo Fisher — Expi293 Expression System (vendor technical notes)](https://www.thermofisher.com/us/en/home/life-science/protein-biology/protein-expression/transient-protein-expression/expi293-expression-system.html)
• [Thermo Fisher — ExpiCHO / CHO-S Expression Systems (vendor technical notes)](https://www.thermofisher.com/us/en/home/life-science/protein-biology/protein-expression/transient-protein-expression/expicho-expression-system.html)
• [ProteoGenix — Stable cell line generation timeline FAQ](https://us.proteogenix.science/monoclonal-antibody-production/stable-cell-line-generation/faqs-stable-cell-lines/timeline-production/)
• [Nature Outlook — Ramping up therapeutic antibody development](https://www.nature.com/articles/d42473-021-00208-1)
• [WuXi Biologics — Mammalian transient and stable expression capabilities](https://collateral.wuxibiologics.com/view/319540890/1/)