What a Western Blot Can and Cannot Establish

Picture a figure in which a peptide of a couple of kilodaltons appears as a crisp band on an ordinary blot. Before anyone asks whether the antibody was specific, a blunter question is available: could that membrane have held the peptide at all? Frequently the answer is no, and the band belongs to something else. That gap is the technique in miniature. A membrane gets photographed, and the photograph gets read as though it were a number.

Small peptides and the equipment that loses them

Everything about the method was designed around proteins. Drop below roughly ten kilodaltons and standard gel systems stop resolving cleanly; at one or two kilodaltons the material travels with the dye front or past it. Transfer is the harsher failure. Rather than binding, short peptides go straight through an ordinary membrane and finish up in the buffer, so a fraction of what was loaded never reaches the surface being probed.

Membranes with smaller pores exist, and so do adapted protocols. Whether either was used is worth confirming in any blot claiming to show a short peptide, because the competing explanation for a tidy band is that the band belongs to something the system could actually retain.

What a band entitles you to say

Run the mechanism forward. Proteins separate by size in a gel, move onto a membrane, and meet an antibody raised against the target. A labeled secondary antibody generates the signal. Position reports apparent size; intensity reports, loosely, abundance.

Notice what that chain delivers. A band sitting where one was expected tells you that some species of roughly that apparent size was bound by your antibody. Identity is an inference layered on top, and it rests on controls rather than on position.

Two controls carry the weight. First, material known to lack the target: a knockout, a knockdown, or a cell type that never expresses it. Second, competition. Pre-incubate the antibody with its immunizing peptide and the band should disappear. Cross-reactivity here behaves as it does in ELISA, with one advantage. A blot at least reports apparent size; an immunoassay reports none.

Extra bands turn up constantly and get cropped away just as often. Cropping discards evidence. More than one band means the antibody sees more than one species, which is exactly the fact bearing on whether the chosen band is the right one.

Why the intensities do not divide

Every step between sample and film is a chance for proportionality to break, and the operator controls only some of them. Transfer efficiency shifts with protein size and varies from place to place across the membrane. Antibody binding tracks antigen concentration linearly over part of the range and not over all of it. Detection saturates outright: chemiluminescent signal captured on film has a narrow dynamic range, and once a band has maxed out the detector, nothing separates it from a band that ought to have been stronger still.

So doubled intensity is not doubled protein. Any comparison holds only inside the linear window of the detection method, and locating that window is a step almost never reported.

Exposure is an editorial decision

Chemiluminescence comes from an enzymatic reaction that peaks early and then fades. When the operator chooses to capture it therefore decides which bands surface and how emphatic they look. Pull a long exposure to rescue something faint and the strong bands beside it may already be off-scale.

Fluorescent detection sidesteps both problems: a considerably wider linear range, and no comparable decay. Where numbers are the point, it is the better instrument. In a published figure the tell is whether a second exposure appears, or is even mentioned. A single chemiluminescent exposure offered as quantitative has not shown that the compared bands were in range.

The assumption buried in the loading control

Probing a housekeeping protein alongside the target is meant to absorb differences in how much material went into each lane. It rests on that housekeeping protein holding steady under treatment, which is rarely tested and sometimes untrue.

Two further requirements go missing even more often:

• The control has to sit within its own linear range. Housekeeping proteins are abundant and saturate readily, and a saturated control corrects nothing whatsoever.
• Staining total protein on the membrane normalizes better than any individual protein can, since it carries no assumption about how one gene behaves.

Where the line falls in image editing

Brightness and contrast may be adjusted, provided the adjustment covers the entire image and the legend discloses it. Adjusting a region is not permitted. Lanes assembled from different areas of one gel, or from separate gels, need a dividing line the reader can see, plus a statement in the legend.

These rules exist because such edits take seconds, leave no trace in the finished figure, and change what the blot appears to demonstrate. Uniform background, immaculate bands, no lane boundaries anywhere: treat that figure with the suspicion you would bring to an implausibly neat trace, as set out under reading a chromatogram.

The ledger

Where identity is the open question, mass spectrometry answers it directly, along the lines described in reading a sequence from fragment ions.