Buying oral, gingival and mucosal tissue: the epithelia outside the guidelines

The short answer: mucosal tissue models are real, shipped products from the same two suppliers that dominate skin — and almost none of them sit inside a regulatory guideline. Reconstructed epidermis has OECD TG 439 and TG 431. Cornea-like epithelium has TG 492. Oral, gingival, oesophageal, bladder and vaginal epithelium have supplier characterisation data and a literature.

That single fact should change how you buy in this category. When there is no externally-set acceptance criterion, the specification burden moves onto you, and the questions you ask in the RFQ are the only quality control the purchase will ever have.

The second thing to know: the keratinised/non-keratinised distinction is the whole ballgame. Oral mucosa is not thin skin. Buccal and vaginal epithelium are non-keratinised, gingival and hard-palate epithelium are keratinised, and permeability differs by orders of magnitude between them. A model that is keratinised when your target site is not will under-predict absorption dramatically, and it will do so silently.

What is actually available

The table makes the structural point on its own. The same supplier, the same production platform, the same cold chain — and the regulatory column flips depending on which epithelium you order. Bladder and oesophageal models in particular are genuinely niche products with real users and no guideline behind them.

Keratinised or not: the specification that decides your answer

Oral mucosa is regionally differentiated in a way that matters more than most buyers expect, and the same is true across the mucosal category.

Non-keratinised — buccal mucosa, floor of mouth, soft palate, oesophagus, vaginal epithelium. No stratum corneum. Comparatively permeable, which is exactly why sublingual and buccal drug delivery exists at all.

Keratinised — gingiva, hard palate, dorsal tongue. A cornified surface layer and a substantially tighter barrier.

Three consequences:

For drug delivery, the target site is the specification. A buccal film, a sublingual tablet and a gingival gel are three different permeability problems. If you are developing buccal delivery and you test on a gingival model, you will under-predict flux — and the result will look clean and reproducible, which is what makes the error expensive.

For irritation and safety, the exposure site drives the choice. A mouthwash contacts both. A toothpaste concentrates at the gingival margin. A denture adhesive sits on keratinised palate. Suppliers catalogue gingival and oral models separately because these are separate questions.

Ask which configuration you are being sold. The parallel is exact with cornea: OECD TG 492 requires an RhCE construct to be non-keratinised with at least three layers of viable epithelial cells, because a keratinised surface would not model the cornea. Nobody imposes that requirement on an oral model, so you must impose it yourself. Put the word in the RFQ — keratinised or non-keratinised — and get the histology to confirm it.

Working without a guideline

This is the category’s defining problem. For reconstructed epidermis you can cite TG 439. For corneal epithelium you can cite TG 492’s published ET50 and IC50 batch release ranges. For oral, gingival, oesophageal, bladder and vaginal models there is no equivalent public number to hold a supplier to.

What replaces it, in descending order of strength:

1. The supplier’s own release criterion, in units. Ask what is measured on every batch before release, in what units, with what acceptance range, on what instrument. A supplier that answers “TEER above X Ω·cm²” or “ET50 between A and B minutes on 0.3% Triton X-100” is giving you a real specification. A supplier that answers “each batch is QC’d” is giving you nothing.
2. Histology per lot, or per product. At minimum: number of viable cell layers, presence or absence of a cornified layer, and evidence of correct differentiation markers. For the guideline models this is fixed by the guideline; here it is a question.
3. A published reference set. Peer-reviewed papers using the specific model for an endpoint close to yours are the closest available substitute for validation, and both suppliers maintain reference libraries. Ask for the ones matching your assay rather than the total count.
4. Your own positive control. Run a known irritant or a known permeant on arrival and build your own historical range. Over three or four lots this becomes the acceptance criterion nobody published for you. It is also the only one guaranteed to be measured on your instrument, in your hands.

The reporting consequence. If the study supports a regulatory submission, the absence of a guideline is not fatal but it does shift the burden: you will be defending the model’s suitability rather than citing a guideline number. Say so in the study plan at the start, not in response to a question at the end.

Infection models: the category’s strongest use case

The most distinctive thing mucosal models do is host bacterial and viral infection, and this is where they are worth the sourcing effort.

MatTek positions EpiVaginal for HIV-1 and other sexually transmitted disease research and for microbicide testing, and catalogues oral candidiasis as a named application for its oral mucosal range. Both are genuine live-pathogen challenge models on human 3D tissue.

Practical consequences that catch people out:

The antibiotic and antifungal question is upstream. Tissue models are cultured with antibiotics and antifungals as standard — for its EpiAirway line MatTek publishes gentamicin at 5 µg/mL and amphotericin B at 0.25 µg/mL, with agent-free variants available and agents removed at least three days before shipment. You cannot wash a tissue into being antifungal-free at the bench, and residual amphotericin B in a Candida challenge is an experiment that cannot work. Order the variant upstream, and treat that three-day removal window as a hard lead-time item.

Biosafety approval is the long pole, not the tissue. HIV-1 work needs containment, institutional approval and trained staff. The tissue is the easy part of that procurement and it is routinely the only part that gets scheduled.

Barrier integrity is the readout, so measure it before you infect. As in the gut models, a pre- and post-challenge barrier measurement is what separates pathogen-driven barrier disruption from a monolayer that was never intact.

Colonisation is not infection. Ask what the model demonstrates — adhesion, invasion, hyphal penetration, cytokine response, epithelial damage — and over what timescale. These are different endpoints with different validation behind them.

Formats, kits and the shipping pattern

Mucosal models come from the same production lines and the same cold chain as their suppliers’ skin products, and the operational pattern is the one described across this hub.

The reference case is MatTek’s published EpiDerm specification: a standard kit of 24 tissues, 9 mm and 22 mm individual inserts plus 96-well HTS plates on a 0.4 µm pore substrate, shipped on medium-supplemented agarose gels, dispatched on a fixed weekly cadence with a shelf life of days including transit. EPISKIN publishes an analogous format range for its epithelial models — small, medium and large formats plus HTS 24-well and 96-well plates.

What to verify for a mucosal order specifically, because these are lower-volume products:

• Whether your format exists for this model. A 96-well HTS version of the flagship skin model does not guarantee a 96-well version of the bladder model.
• Production frequency. Niche models may be produced less often than weekly. That directly changes your order-by date — see production calendars and shelf life.
• Minimum order quantity. If the kit is 24 tissues and your design needs 6, you are buying 24.
• Whether partial kits or mixed-model shipments are possible, which matters when your study spans oral and gingival models in one run.

Culture duration: ask, do not assume

Mucosal epithelia turn over rapidly in vivo, and the reconstructed models are correspondingly short-lived compared with the airway products described in airway and respiratory models, where suppliers publish months of stable culture.

Do not carry an assumption across from one tissue to another. The specific question is: for how many days is this model validated in culture, on what feeding schedule, with what medium, before the histology degrades? For an acute irritation assay measured in hours this hardly matters. For a five-day candidiasis challenge or a repeated-dose delivery study it is the constraint that decides whether the design is possible at all.

What a mucosal model purchase does not include

• Agent-free medium variants, unless ordered upstream with the three-day removal window respected.
• The pathogen, its characterised stock, and the containment to work with it.
• Barrier instrumentation — TEER electrodes and a chamber matched to the insert diameter.
• Franz cells or permeation hardware for delivery work, and the analytical chemistry behind it.
• Histology, unless the supplier includes it per lot — for a model with no guideline behind it, budget for your own.
• A validated acceptance criterion. In this category you are usually building one.

Specification checklist for a mucosal tissue quote

• Which epithelium, named anatomically: buccal, gingival, hard palate, oesophageal, bladder, vaginal, ectocervical.
• Keratinised or non-keratinised, stated explicitly and confirmed by histology.
• Number of viable cell layers, and the differentiation markers demonstrated.
• Batch release criterion, in units, with the acceptance range and the measurement method.
• Validated culture duration, in days, with the feeding schedule and medium.
• Medium variant: antibiotic-free, antifungal-free, phenol-red-free — ordered upstream.
• Format and insert diameter, confirmed to exist for this specific model, not just for the supplier’s flagship.
• Kit size and minimum order quantity, worked against your replicate count.
• Production frequency and order-by date for this model specifically.
• Reference papers using this model for an endpoint close to yours.
• For infection work: the pathogen, the endpoint, the challenge duration and the containment level.
• For delivery work: the receiver fluid, the sampling schedule and the bioanalytical method.

Where to go next

• Engineered tissue map — where mucosal models sit among the availability clusters.
• Corneal and ocular models — the adjacent non-keratinised epithelium, and the contrast case of what a guideline-backed model looks like.
• Airway and respiratory models — published TEER release criteria and much longer culture windows.
• Production calendars and shelf life — why niche models are harder to schedule than flagship ones.
• Buying reconstructed human epidermis — the fully validated end of the same supply chain.

How we can help

The recurring failure in this category is a keratinisation mismatch that nobody notices, because the tissue performs beautifully and answers the wrong question. The second is discovering at the bench that the standard model ships with amphotericin B in it, three days after the antifungal-free variant would have had to be ordered.

We pin the anatomical site and the keratinisation state before anything is quoted, get each supplier’s actual batch release criterion in units rather than a QC assurance, confirm the format and production frequency exist for the specific niche model rather than the flagship, and where no guideline exists we help you build the internal acceptance range that will have to serve instead.

Tell us the anatomical site, the endpoint, whether live pathogens are involved and how long the tissue has to survive.