A CLDN18.2 cell line is almost never a gastric cell line. Antibody discovery and CAR-T programmes targeting claudin 18.2 overwhelmingly engineer the antigen into an antigen-null background — HEK293, HeLa, CHO-K1 or K562 — because the property that makes CLDN18.2 an attractive drug target is exactly what makes it awkward at the bench. This application note explains why the blank-cell strategy wins, how to choose the background and the expression modality, and why surface flow cytometry — not western blot — is the readout that decides whether your model is fit for purpose.
Why CLDN18.2 has no soluble form you can coat on a plate
Claudin 18.2 (CLDN18.2) is one of the most actively pursued solid-tumour antigens in gastric and gastro-oesophageal junction (GEJ) adenocarcinoma. It is a splice variant of claudin-18 whose expression in healthy adult tissue is essentially restricted to differentiated gastric mucosa, and even there it is buried within tight junctions and largely inaccessible to circulating antibody (Sahin et al., Clin Cancer Res 2008, PMID 19047087 · Türeci et al., Gene 2011, PMID 21571049). When gastric epithelium transforms, polarity is lost and CLDN18.2 becomes exposed on the cell surface. That differential accessibility is the therapeutic window behind zolbetuximab (SPOTLIGHT, Lancet 2023, PMID 37068504 · GLOW, Nat Med 2023, PMID 37524953) and behind CLDN18.2-directed CAR-T (Qi et al., Nat Med 2022, PMID 35534566).
The bench problem follows directly from the structure. CLDN18.2 is a tetraspan membrane protein: four transmembrane helices with two short extracellular loops. Those loops adopt their functional conformation only when the protein is folded within a lipid bilayer. There is no soluble ectodomain to express, purify and coat onto a plate. Antibodies raised or screened against denatured protein or synthetic loop peptides frequently fail to recognise the native cell-surface form — and the reverse is equally true.
The antigen has to be presented on a cell. Every downstream decision in this note — background line, delivery route, readout — follows from that single constraint.
Why a CLDN18.2 cell line is built in an antigen-null background
HEK293, HeLa, CHO-K1 and K562 share one decisive property for this work: they carry no meaningful endogenous claudin-18. They are chosen because they are blank. Starting from zero and adding a single defined variable delivers five things a naturally expressing gastric line cannot.
Choosing the background line: HEK293, HeLa, CHO-K1 or K562
The right blank cell is dictated by the assay format, not by preference. All four are available in the BioHippo cell lines catalog.
| Line | Origin & format | Why it is chosen | Typical assay |
|---|---|---|---|
| HEK293 / HEK293T | Human embryonic kidney · adherent | Exceptionally high transient expression; fastest route from nucleic acid to surface antigen | Rapid antigen generation, flow binding, hit triage |
| HeLa | Human cervical adenocarcinoma · adherent | Easy to transfect, flat morphology, extremely well characterised | Confocal localisation, adherent binding, internalisation |
| CHO-K1 | Chinese hamster ovary · adherent or suspension | Non-human background reduces the chance an anti-human candidate scores through an unrelated human surface protein; established platform for stable lines | Antibody screening, specificity panels, stable antigen lines |
| K562 | Human CML · suspension · MHC class I low | Standard artificial target cell; suspension format suits co-culture and low MHC I limits allo-recognition confounds | CAR-T and NK cytotoxicity, cytokine release, CD107a |
| Jurkat reporter | Human T-cell leukaemia · suspension | Reporter (e.g. NFAT-driven) readout of CAR engagement without primary T cells | CAR activation screening, construct ranking |
HeLa is a legitimate and widely used choice for binding and imaging. Note that as a cervical adenocarcinoma line it supplies the protein without gastric context — see the limitations section below.
mRNA, plasmid DNA or a stable CLDN18.2 cell line?
Having chosen a blank background, the next decision is how to deliver the antigen. These routes are not interchangeable — each answers a different question, and most sustained programmes use more than one.
| Attribute | Synthetic mRNA | Plasmid DNA | Lentiviral / stable line |
|---|---|---|---|
| Delivery requirement | Cytoplasm only | Must reach the nucleus | Integrates into genome |
| Time to surface antigen | Hours | 24–72 h | Weeks (selection, cloning) |
| Fraction of cells positive | High and relatively uniform | Variable; often a mixed population | Uniform after clonal selection |
| Suspension / hard-to-transfect cells | Generally yes | Often poorly | Yes (transduction) |
| Density titratable by dose | Yes — cleanly | Poorly | Fixed per clone |
| Duration | Transient (days) | Days, unless selected | Permanent |
| Expression drift over passage | Not applicable | Not applicable | A real risk — requires monitoring |
| Best suited to | Screening campaigns, killing assays, density-threshold studies, rapid variant panels | Construct prototyping, cheap adherent work, stable-line intermediate | Long-running programmes needing one consistent reagent cell for months |
Why mRNA suits the counter-screen panel specifically
Matched variant panels on demand
Because mRNA bypasses the nucleus and expresses within hours, a three-arm panel — CLDN18.2, CLDN18.1 and mock — can be produced from one parental culture in a single session, with identical handling for each arm. Dose-titrating the same transcript then converts that panel into a density series, so specificity and sensitivity are read from one experimental system rather than two.
Use when you are triaging antibody candidates, generating defined target cells for cytotoxicity assays, or measuring the antigen-density threshold of a CAR construct.
Transient expression is a time window, not a steady state
Surface density rises and then declines. Binding, cytotoxicity and quantification assays must be anchored to a time point established by an initial time-course in your own hands — and that time point should be reported in methods.
Flow cytometry vs western blot for surface CLDN18.2
These two methods are not competing versions of the same measurement. They answer different questions, and only one of them is the question that matters here.
| Flow cytometry (non-permeabilised) | Western blot | |
|---|---|---|
| Question answered | Is correctly folded antigen on the outer surface of intact cells, on how many cells, and at what density? | Is protein of roughly this molecular weight present in this lysate, in total? |
| Sample state | Live, intact, membrane preserved | Lysed, denatured, reduced |
| Resolution | Single cell | Bulk population average |
| Surface vs intracellular | Yes, by definition | No |
| Reports % of cells positive | Yes | No |
| Conformational epitopes | Yes | Generally no |
| Absolute antigen density | Yes, with calibration beads | Relative intensity only |
| Molecular weight / truncation | No | Yes |
1. Western blot is blind to topology — the decisive failure
Lysis destroys the membrane. Protein retained in the endoplasmic reticulum, misfolded or aggregated produces the same band as protein correctly trafficked to the plasma membrane. A construct can therefore give a strong, convincing blot while presenting zero antigen accessible to an antibody or a CAR-T cell. Since surface accessibility is the entire point of the model, a method that cannot report it cannot validate it.
2. It averages away transfection efficiency
A blot cannot distinguish 100% of cells expressing at 1× from 20% of cells expressing at 5×; the band looks the same. Those populations behave entirely differently in a cytotoxicity assay — the second contains 80% antigen-negative cells that cap the apparent killing plateau and are easily misread as CAR failure.
3. It destroys the epitope you care about
SDS and reducing agents denature the protein. Therapeutically relevant anti-CLDN18.2 antibodies recognise conformational epitopes on extracellular loop 1 that exist only in the folded, membrane-embedded state, and many give no western signal at all. A negative blot with a conformation-dependent antibody is therefore uninterpretable — it is not evidence that the antigen is absent.
4. It cannot separate CLDN18.1 from CLDN18.2
The two splice variants differ at the N-terminus but are closely similar in size and are not resolved from one another by migration on a standard gel. The most consequential specificity question in the field — is my binder selective for 18.2 over 18.1? — is simply not addressable by molecular weight. Flow cytometry on separately transfected populations answers it directly, in one experiment, with an isogenic comparator.
Handling caveat for claudins on SDS-PAGE
Polytopic membrane proteins including claudins are prone to heat-induced aggregation; boiling can drive them into high-molecular-weight aggregates that fail to enter the gel, producing false negatives or smears. Common practice is to heat at reduced temperature rather than boil.
Membrane proteins also migrate anomalously on SDS-PAGE because of differential detergent binding, so apparent molecular weight should not be over-interpreted (Rath et al., PNAS 2009, PMID 19181854).
What flow cytometry adds that nothing else does
- Absolute antigen density. Calibrated bead standards (MESF or antibody-binding-capacity beads) convert fluorescence intensity into an estimate of molecules per cell (Wang & Hoffman, Curr Protoc Cytom 2017, PMID 28055116). Because CAR activity is a threshold function of density, this number is the experiment rather than a refinement of it.
- Trafficking efficiency, without a blot. Stain one aliquot non-permeabilised (surface only) and a matched aliquot permeabilised (total). The ratio reports what fraction of expressed protein actually reached the surface — the quantity a western blot is often incorrectly used to infer.
- Live-cell gating. Viability dye exclusion removes dead and dying cells, which bind antibody non-specifically and are a major source of false positives. Lysates carry that artefact silently.
- Multiplexing and sorting. Antigen, viability and a transfection marker can be read simultaneously, and positive cells can be sorted for downstream use.
Where western blot still earns its place
The honest position is that these methods are complementary. A programme that abandons blotting entirely will miss product integrity (flow tells you an epitope is present, not whether the protein is full length — truncation and degradation are visible only by size), total cellular protein as a denominator when quantifying trafficking, and antibody characterisation (establishing linear-epitope reactivity is a prerequisite for qualifying a reagent for western or IHC use). Neither method displaces immunohistochemistry for tissue: flow requires a single-cell suspension, while IHC preserves architecture and is the established approach for scoring CLDN18.2 in gastric tumour tissue, including in the clinical setting (PMID 37068504). The correct framing is flow for cells, IHC for tissue, western for identity and size.
A defensible validation sequence for a CLDN18.2 cell line
Controls to run every time: isotype-matched control antibody · mock-treated cells · viability dye · secondary-only (where applicable) · a positive-control antibody of known reactivity.
What an engineered CLDN18.2 cell line cannot tell you
Overstating what a transfected line represents is the most common interpretive error in this area. Three limitations should be stated explicitly in any methods section.
The protein is present, but out of context
HeLa is cervical; HEK293 is embryonic kidney; CHO is hamster ovary. None runs a gastric differentiation programme, forms a properly polarised gastric epithelium, or supplies the native junctional partners and membrane environment CLDN18.2 normally sits within. The molecule is there and correctly folded; the biology around it is not. That is acceptable when the question concerns the molecule — does my antibody bind, how tightly, how selectively. It is not acceptable when the question concerns junctional behaviour or physiology, which requires a polarisable model or a gastric line.
Overexpression can flatter a binder
On a transfected, non-polarised cell, antigen is distributed across the whole surface and can reach densities well above those found in tissue. In a real polarised epithelium, CLDN18.2 is sequestered within tight junctions and substantially less accessible. Binding data generated on engineered cells is a screening and ranking tool — not a prediction of tumour accessibility in vivo.
Naturally expressing lines have their own constraints
Gastric adenocarcinoma lines retaining endogenous CLDN18.2 are valuable for confirmation but poor primary screening tools: expression is comparatively low, can drift across passages, offers no isogenic negative control, cannot be titrated, and is often accompanied by other claudin family members that complicate specificity calls.
Verify before citing
Reported CLDN18.2 status of specific gastric carcinoma lines varies between publications and between laboratory stocks. Confirm expression in your own hands by flow or IHC before treating any line as a positive control, and cite the source of the line and its authentication status.
Frequently asked questions
Why not simply buy recombinant CLDN18.2 protein and coat a plate?
Because there is no soluble form that retains the relevant conformation. The antibody-accessible surface of CLDN18.2 consists of two short extracellular loops whose structure depends on four transmembrane helices anchoring them in a bilayer. Purified or peptide forms present epitopes that frequently do not correspond to the native antigen, in either direction. Cell-surface display is the only routine way to present the real thing.
Should I use mRNA or a plasmid for a CLDN18.2 cell line?
It depends on how long you need the cells. Choose mRNA for speed, uniformity, suspension or hard-to-transfect cells, and any experiment requiring a titratable antigen density — screening campaigns, cytotoxicity assays, CAR threshold studies. Choose plasmid DNA, or more commonly lentiviral transduction, when you need a permanent line still available in six months. Many programmes maintain both: a stable line as the workhorse and mRNA for rapid variant panels and density series.
My western blot shows strong expression but my antibody does not bind by flow. Which is wrong?
Most likely neither. The usual explanation is that the protein is expressed but not trafficked to the plasma membrane — present in the lysate and therefore on the blot, but absent from the outer surface where the antibody must reach it. Test this directly by staining matched permeabilised and non-permeabilised aliquots: intracellular-positive with surface-negative confirms a trafficking problem rather than an expression problem. A second possibility is that your antibody recognises a linear epitope on the blot where a conformational one is required on the cell.
How many cells need to be antigen-positive for a valid cytotoxicity assay?
There is no universal threshold, but the percentage must be measured and reported, because antigen-negative cells set a ceiling on achievable killing that is easily mistaken for poor construct performance. Where transfection is incomplete, either sort the positive fraction or normalise the readout to the measured positive percentage. This is a further argument for a flow-based readout: the number is produced automatically.
Why does the CLDN18.1 counter-screen matter so much?
CLDN18.1 is expressed in normal lung epithelium, so a binder that engages both splice variants carries a pulmonary on-target, off-tumour risk. Because the variants differ over only a short region, selectivity cannot be assumed from sequence similarity or inferred from a western blot — it has to be demonstrated on cells expressing each variant in the same background.
Is HeLa an appropriate choice, given it is not gastric?
For binding, specificity and imaging work, yes — the question being asked is about the molecule, and HeLa supplies a clean, well-characterised, easily transfected background. For questions about junctional localisation or gastric physiology it is not appropriate, and a polarisable or gastric model is required. State the limitation in methods rather than implying tissue relevance the model does not have.
Which background line should I pick for a CAR-T killing assay?
K562 is the conventional artificial target: it grows in suspension, which suits co-culture, and its low MHC class I expression limits allo-recognition confounds when using primary T cells from an unmatched donor. Deliver the antigen by mRNA if you want a density series in the same experiment, and read the percentage of antigen-positive target cells on the same day you run the killing assay.
Reagents for this workflow
| Workflow role | Reagent |
|---|---|
| Background (antigen-null) cells |
HEK293 · HeLa · CHO-K1 · K562 BHC11100722 · BHC10901792 · BHC11100061 · BHC11100712 |
| Antigen source |
Claudin 18.2 mRNA (BHN20152499, GenCefe Biotech) Cap1 (m7GpppNm) with 100% N1-methylpseudouridine substitution; supplied lyophilised and non-encapsulated. Confirm current grade, pack size and intended use on the product page. |
Browse the full mRNA reagents collection and the cell lines catalog, or send a quote request with your background line, target variant and readout if you need a combination that is not listed. All products are supplied for Research Use Only; they are not for use in diagnostic or therapeutic procedures.
References
- Sahin U, Koslowski M, Dhaene K, et al. Claudin-18 splice variant 2 is a pan-cancer target suitable for therapeutic antibody development. Clin Cancer Res 2008;14(23):7624–34. PMID 19047087
- Türeci O, Koslowski M, Helftenbein G, et al. Claudin-18 gene structure, regulation, and expression is evolutionary conserved in mammals. Gene 2011;481(2):83–92. PMID 21571049
- Shitara K, Lordick F, Bang YJ, et al. Zolbetuximab plus mFOLFOX6 in patients with CLDN18.2-positive, HER2-negative, untreated, locally advanced unresectable or metastatic gastric or gastro-oesophageal junction adenocarcinoma (SPOTLIGHT): a multicentre, randomised, double-blind, phase 3 trial. Lancet 2023;401(10389):1655–68. PMID 37068504
- Shah MA, Shitara K, Ajani JA, et al. Zolbetuximab plus CAPOX in CLDN18.2-positive gastric or gastroesophageal junction adenocarcinoma: the randomized, phase 3 GLOW trial. Nat Med 2023;29(8):2133–41. PMID 37524953
- Qi C, Gong J, Li J, et al. Claudin18.2-specific CAR T cells in gastrointestinal cancers: phase 1 trial interim results. Nat Med 2022;28(6):1189–98. PMID 35534566
- Majzner RG, Rietberg SP, Sotillo E, et al. Tuning the antigen density requirement for CAR T-cell activity. Cancer Discov 2020;10(5):702–23. PMID 32193224
- Rath A, Glibowicka M, Nadeau VG, Chen G, Deber CM. Detergent binding explains anomalous SDS-PAGE migration of membrane proteins. Proc Natl Acad Sci USA 2009;106(6):1760–5. PMID 19181854
- Wang L, Hoffman RA. Standardization, calibration, and control in flow cytometry. Curr Protoc Cytom 2017;79:1.3.1–1.3.27. PMID 28055116
Research Use Only. Not for use in diagnostic or therapeutic procedures. Products described here are supplied by third-party manufacturers and made available through BioHippo; manufacturer attribution, current specifications, grade and intended-use statements are given on each product page and should be confirmed before purchase. This note is informational and does not constitute experimental protocol validation for any specific application — assay conditions, time points, controls and cell line authentication must be established and verified by the end user. Clinical trial and regulatory references are current as of publication and should be re-confirmed.
