What a Cell Viability Assay Actually Measures

A compound is added to cells, a viability assay is run, and the signal falls. The conclusion drawn is that cells died. What the assay measured was the rate at which something in the well reduced a dye, and the distance between those two statements is where most of the trouble in this literature lives.

What the common assays actually detect

The tetrazolium assays and their resazurin-based relatives all work the same way: a dye is added, something in the cell reduces it, and the reduced form is measured by colour or fluorescence. The reduction is carried out largely by NAD(P)H-dependent oxidoreductases, so the signal is a measure of reductive metabolic capacity in the well.

That correlates with cell number only under an assumption that is rarely stated and frequently false: that metabolic activity per cell is constant. Where it holds, the assay is a reasonable proxy for how many living cells are present. Where the compound under test changes metabolism, it stops being a proxy for anything.

The circularity that matters for this catalogue

This is not an abstract concern for peptide work. A compound that acts on mitochondrial function will change the rate of dye reduction directly, without any cell dying or dividing.

Run a tetrazolium assay on a mitochondrially-targeted compound and the result is partly a readout of the mechanism being studied, not an independent check on whether the cells survived it. The same applies to anything affecting NAD availability, since NAD(P)H is what performs the reduction — the coenzyme role set out in what a coenzyme is and is not and the targeting described in SS-31 and cardiolipin.

A signal that goes up in that situation may mean more cells, healthier cells, or the same cells reducing dye faster. The assay cannot distinguish them, and no amount of replication will make it.

Viability, proliferation and cytotoxicity are three questions

• Viability — what proportion of the cells present are alive. Properly answered by membrane integrity or a live/dead discrimination, counted per cell.
• Proliferation — whether the population grew. Requires a count, or a marker of division.
• Cytotoxicity — whether the compound killed cells. Properly answered by measuring something released from dying cells, such as a cytoplasmic enzyme appearing in the medium.

A metabolic dye assay is used as a proxy for all three and is a direct measure of none. A reduced signal is consistent with fewer cells, with the same number of quieter cells, or with cells that stopped dividing — and those are different findings.

Artefacts that produce a signal with no biology behind them

• Direct chemical reduction. A reducing compound in the well can reduce the dye without any cellular involvement at all. Ascorbate and thiol-containing material are the classic cases, and a cell-free control containing compound and dye is the check — the same role played by the solvent condition in vehicle controls.
• Optical interference. A coloured compound in the well adds absorbance. This is a live concern for a copper complex, which is visibly coloured, as noted in GHK and GHK-Cu.
• Precipitation. Material that comes out of solution scatters light and shifts the reading, which is one of several reasons the solution behaviour in peptide aggregation is worth establishing first.
• Edge effects. Evaporation from perimeter wells concentrates everything in them over a long incubation.

The endpoint problem

These assays are read at one time point, which makes them a snapshot of a process. A compound that delays division produces the same reduced signal at 24 hours as one that killed a fraction of the cells, and the two diverge completely by 72 hours.

A single time point therefore cannot distinguish cytostatic from cytotoxic. Where the distinction matters, the answer is more time points or a different assay, not a more precise reading of the same one.

What makes a result interpretable

Two assays with different mechanisms, agreeing. A metabolic readout paired with a membrane-integrity readout answers a different question each and constrains the interpretation in the way described in confirming with two independent methods — the same orthogonality principle that governs analytical work.

Alongside that: a cell-free control to exclude direct reduction, a compound-only optical control, more than one time point, and an actual cell count somewhere in the study to anchor what the signal corresponds to.

The concentration in the well is its own question

Viability assays are usually run over 24 to 72 hours, which is long enough for the exposure to stop resembling what was added. A peptide adsorbs to the plate, as described in adsorptive loss to surfaces, and is degraded by peptidases in serum-containing medium.

The practical consequence is that an apparent lack of effect at a stated concentration may be a lack of exposure rather than a lack of activity. Where a compound shows nothing across a whole range, the possibility worth excluding before concluding anything is that little of it was still present and free by the time the readout was taken.

Reading a reported result

The useful questions are which assay was used, whether the compound could plausibly interfere with its chemistry, whether a cell-free control was run, at what time point the reading was taken, and whether any orthogonal measure was included.

A paper reporting “viability” with a tetrazolium assay, one time point and no interference control has measured dye reduction and called it survival. That may be correct, and nothing in the experiment establishes it.