What a Western Blot Can and Cannot Establish

A western blot is a photograph of a membrane, and it is read as a measurement. The gap between those is wide enough that the technique has been at the centre of most published disputes about image handling, and wide enough that the number placed beneath a band deserves more scepticism than it usually gets.

What the method does

Proteins are separated by size in a gel, transferred to a membrane, and probed with an antibody against the protein of interest. A labelled secondary antibody generates signal, and the resulting band’s position indicates apparent size while its intensity indicates, approximately, how much was there.

Each of those steps is an opportunity for the relationship between band intensity and protein quantity to stop being proportional, and several of them are not under the operator’s control.

Why it is semi-quantitative at best

Transfer efficiency varies with protein size and across the membrane. Antibody binding is not linear with antigen concentration across the whole range. Detection chemistry saturates: chemiluminescent signal on film has a limited dynamic range, and a strong band that has saturated the detector cannot be distinguished from a stronger one.

The practical consequence is that a twofold difference in band intensity does not reliably indicate a twofold difference in protein. Comparisons are only meaningful within the linear range of the detection, which requires establishing where that range is — a step almost never reported.

The specific problem with peptides

The technique was built for proteins, and short peptides behave badly in it.

Standard gel systems resolve poorly below roughly ten kilodaltons, and a peptide of one or two kilodaltons will run near or off the dye front. Transfer is worse: small peptides pass through a standard membrane rather than binding to it, so a proportion of the material ends up in the buffer. Smaller pore membranes and modified protocols exist, and their use is worth checking for in any blot claiming to show a short peptide.

Much of the time, a band attributed to a peptide in this size range warrants asking whether the system used could have retained it at all — a question separate from whether the antibody is specific.

Antibody specificity, which is the load-bearing assumption

A band at the expected position is treated as the protein of interest. What it establishes is that something of approximately that apparent size was recognised by the antibody.

The controls that make the attribution defensible are a sample known to lack the target — a knockout, a knockdown, or a cell type that does not express it — and a peptide competition, in which pre-incubating the antibody with its immunising peptide abolishes the band. The cross-reactivity considerations are the same ones that govern ELISA, with the added complication that a blot at least shows apparent size, which an immunoassay does not.

Multiple bands are common and are frequently cropped out. Their presence is information: it indicates the antibody recognises more than one species, which bears on whether the chosen band is the right one.

Loading controls and normalisation

A housekeeping protein is probed alongside the target to correct for differences in how much was loaded. The assumption is that the housekeeping protein does not change with treatment, which is frequently untested and sometimes false.

Two further constraints are routinely ignored. The loading control must itself be within its linear range, and abundant housekeeping proteins saturate easily — a saturated control cannot correct anything. And total protein staining of the membrane is a better normaliser than any single protein, because it does not depend on an assumption about one gene’s behaviour.

Image handling, where the line sits

Adjusting brightness and contrast across an entire image is acceptable and should be disclosed. Adjusting part of an image is not. Splicing lanes from different parts of a gel, or from different gels, requires a visible dividing line and a statement in the legend.

The reason these conventions exist is that the manipulations are easy, invisible in the final figure, and change what the blot appears to show. A figure with no visible lane boundaries, uniform background and perfectly clean bands is worth reading with the same care as an unusually tidy chromatogram, per reading a chromatogram.

Detection chemistry changes what the figure looks like

Chemiluminescent detection produces signal from an enzymatic reaction that is brightest immediately and decays, so exposure timing affects which bands appear and how strong they look. A long exposure that makes a faint band visible may have saturated the strong ones in the same lane.

Fluorescent detection has a considerably wider linear range and does not decay the same way, which makes it the better choice where quantitation is the point. The practical marker in a paper is whether more than one exposure is shown or mentioned; a single chemiluminescent exposure presented as quantitative has not demonstrated that the bands compared were within range.

What it can and cannot establish

It can show that an immunoreactive species of approximately a given size is present, that it is present in one condition and not another, and — with the linear range established and proper normalisation — that it changed by roughly some amount.

It cannot establish absolute quantity, cannot establish identity on its own, and cannot reliably detect a small peptide in a standard system. Where identity is the question, mass spectrometry answers it directly, in the way set out in reading a sequence from fragment ions.