Cell Line Choice Decides the Answer

Two laboratories test the same peptide for the same activity and reach opposite conclusions. Both ran the experiment competently. The difference is that one used a cell line expressing the relevant receptor and the other used a line that does not express it at all — and neither paper says so, because neither thought it needed stating.

The model is part of the result

A cell line is not a neutral container in which a compound is observed. It is a specific population with a particular expression profile, a particular metabolic state and a particular history, and the result obtained is a property of the compound and that system together.

This is the reason a result cannot be quoted without its model. “Compound X increases proliferation” is not a finding; “compound X increases proliferation in this line under these conditions” is. The first sentence is the one that circulates.

Receptor expression, which decides most peptide experiments

Peptides acting through a receptor can only act where that receptor is present. Whether a given line expresses it is an empirical question with a frequently surprising answer, and expression levels also drift with passage number and culture conditions.

Two failure modes follow, in opposite directions. A negative result in a line lacking the receptor says nothing about the compound — it is a statement about the model. And an overexpressing line, engineered to carry far more receptor than any natural tissue, can produce a robust response at concentrations where a natural expression level would produce nothing measurable, which shifts apparent potency in the sense discussed in Ki, IC50 and EC50.

Receptor subtype matters as much as presence. Where a family of related receptors exists, selectivity between them is the whole question, as it is for the melanocortin receptors in MC1R and MC4R selectivity — and a line expressing the wrong subtype answers a different question than the one asked.

Species is not a detail

Many peptide sequences differ between species, and the receptors differ more. A human sequence tested against a rodent receptor is a cross-species experiment whether or not it is described as one.

Sometimes the sequences are identical and this does not arise. Often they are not, and the difference is precisely what makes the comparison uninterpretable — the situation described for kisspeptin in kisspeptin-10 and KISS1R, where the human and rodent peptides are not the same molecule.

Immortalised, primary and engineered

• Immortalised lines divide indefinitely because something in their growth control is broken. That is what makes them usable and also what makes them abnormal, and the abnormality is often in exactly the pathways under study.
• Primary cells are closer to the tissue they came from and behave less consistently, vary between donors, and change as they are passaged.
• Engineered lines answer a narrow question cleanly — does the compound act at this receptor — while telling you nothing about whether the receptor is present anywhere it would matter.

None of these is the correct choice in general. Each is correct for a different question, and the common error is reading a result from one as though it came from another.

Conditions that change the answer without changing the line

Serum concentration in the medium alters how much compound is free rather than bound, which shifts effective concentration in the way described in why a concentration is not a dose. Confluence changes signalling behaviour. Passage number changes expression. Atmospheric oxygen, which is what a standard incubator provides, is far above what most tissues experience.

Peptidase activity in serum is a specific concern for this class: a peptide can be degraded during the incubation, so the exposure differs between a serum-containing and a serum-free condition even at the same nominal concentration.

Identity and contamination, which are still real problems

A substantial proportion of published work has used misidentified or cross-contaminated lines — cells that are not what the paper says they are. Authentication is available and inexpensive, and is still not universal.

Mycoplasma contamination is the other long-running issue. It is invisible, it alters cellular metabolism and signalling, and it is common enough that a result from an untested culture carries a real possibility of being a result about the contaminant. Neither problem is exotic; both are ordinary enough that a methods section silent on them has left a question open — the same reading discipline as for the controls in vehicle controls.

What to look for, and what to record

Reading a paper, the useful questions are which line, from where, at what passage, whether the relevant receptor is expressed and at what level, whether authentication and contamination testing were done, and what the serum conditions were. A study that answers those can be compared with another that answers them. A study that answers none cannot be compared with anything.

Running the experiment, the same list is what makes the work reproducible by somebody else, and it sits alongside the material-side record described in inventory labels and records. Between the two, the compound and the system are both documented — which is the minimum for a result that means the same thing in a year’s time as it does today, and the starting point for the wider problem set out in why two laboratories get different results.