The Western Blot and Everything That Can Go Wrong

The western blot has an unfortunate reputation as a technique that either works or does not, with little in between and no obvious way to find out which stage let you down. That reputation is undeserved. A blot is a chain of six operations, each of which either succeeds or fails independently, and each failure mode leaves a distinctive mark on the final image. A completely blank film means something different from a film that is uniformly grey, and both mean something different from a film with a beautiful band at the wrong molecular weight.

Learning to read those signatures backwards is the difference between troubleshooting and guessing. The alternative, repeating the whole protocol with one variable changed and hoping, is how weeks disappear.

This piece walks the chain forwards, explaining what each step is physically doing and what it looks like when it does not do it, then collects the common failure patterns into a diagnostic list.

Separating Proteins by Size in a Gel

The gel is a mesh of polyacrylamide with a pore size determined by how much acrylamide was polymerised into it. Proteins driven through that mesh by an electric field are retarded according to how awkwardly they fit through the pores, so smaller proteins travel further in a given time. That is the whole principle, and everything else in the first step exists to make it true.

Proteins in their native state do not behave this way, because their mobility depends on their own folded shape and their own net charge, which vary unpredictably. Sodium dodecyl sulfate solves both problems at once. It is a detergent that unfolds proteins into extended rods and coats them with negative charge roughly in proportion to their length, so that charge-to-mass ratio becomes approximately constant across every protein in the sample. A reducing agent is usually added to break disulfide bonds, since a protein held in a compact shape by internal cross-links will run as though it were smaller than it is.

Sample preparation is where the first failures happen. Proteases released when cells are lysed will chew up the target unless inhibitors are present and the sample is kept cold, and the signature of this is a smear below the expected band or a ladder of fragments. Incomplete denaturation shows up as bands at multiples of the expected weight, because dimers and higher oligomers survived. Overloading a lane distorts the band into a smile or a bowed front and makes neighbouring lanes unreliable.

Acrylamide percentage matters more than beginners expect. A gel optimised to resolve small proteins will run large ones as an unresolved clump at the top, and a low-percentage gel lets small proteins run off the bottom entirely. Choosing the percentage to bracket the target is the cheapest single improvement available.

Transferring Proteins to a Membrane

Proteins inside a gel are inaccessible to antibodies, so they must be moved onto a membrane surface where they can be probed. An electric field applied perpendicular to the gel drives them sideways out of the mesh and onto a sheet of nitrocellulose or polyvinylidene fluoride pressed against it.

The two membrane chemistries differ in ways worth knowing. Nitrocellulose binds protein readily, is easy to handle wet and gives low background, but it is brittle when dry. PVDF binds more protein per unit area, tolerates stripping and reprobing better, and is physically tougher, but it is hydrophobic and must be wetted with methanol before it will accept anything at all. A PVDF membrane that was not properly pre-wetted produces a blank blot, and this is one of the most common single-cause failures in the whole procedure.

Transfer is also where size works against you in the opposite direction from the gel. Small proteins leave the gel quickly and can be driven straight through the membrane and out the other side, while very large proteins may never leave the gel at all. Adding methanol to the transfer buffer improves binding but shrinks the gel and hinders the exit of large proteins; reducing methanol and adding a small amount of detergent helps large proteins move at the cost of some binding.

Air bubbles trapped between gel and membrane block transfer locally, producing round blank spots that are unmistakable once you have seen one. The remedy is rolling the sandwich flat rather than pressing it.

The single most useful habit here is to verify transfer directly with a reversible total-protein stain on the membrane before proceeding. It takes minutes and it converts the most common ambiguous failure, a blank film, into a known quantity.

Blocking and Why It Prevents Noise

A membrane is designed to bind protein indiscriminately, which is exactly the problem once the sample proteins are on it. Every unoccupied site will happily bind the antibodies you add next, and an antibody stuck to bare membrane generates signal identical to an antibody stuck to its target. Blocking fills those sites with a cheap, irrelevant protein first.

Skimmed milk powder and bovine serum albumin are the traditional choices, and they are not interchangeable. Milk is a complex mixture that includes phosphoproteins, and it will compete with phospho-specific antibodies and suppress the very signal you are trying to see. Blots probed for phosphorylation are therefore blocked in albumin as a matter of routine. Milk also contains biotin, which matters if the detection system uses a biotin-streptavidin step. Conversely, some antibodies give cleaner results in milk than in albumin for reasons that are largely empirical.

Both under-blocking and over-blocking cause trouble. Insufficient blocking leaves sites open and produces a grey membrane with poor contrast. Excessive blocking, particularly with high concentrations or long incubations, can mask epitopes and cut genuine signal, giving a clean but faint result. Blocking buffer that has been reused or kept too long grows bacteria, and the products of that growth create speckled backgrounds.

Buffer composition carries through the rest of the procedure. Blocking agent is usually retained in the antibody dilutions to keep the membrane occupied while the specific binding happens.

Primary and Secondary Antibody Layers

Detection is a two-layer system for reasons of both specificity and amplification. The primary antibody recognises the target protein. The secondary antibody recognises the primary, is raised against the species the primary came from, and carries the enzyme that eventually produces the signal.

That arrangement means several secondary molecules can decorate each primary, multiplying the signal, and it means one labelled secondary reagent serves any number of primaries from the same host species. It also means the species relationships must be tracked carefully. A secondary raised against mouse will bind mouse immunoglobulin wherever it finds it, including endogenous immunoglobulin in a sample from mouse tissue, which is a classic source of unexpected bands.

Antibody validation is the uncomfortable part. A primary antibody is a reagent whose specificity is a claim, and that claim is frequently weaker than the datasheet implies. The honest tests are a knockout or knockdown sample where the band should disappear, a positive control lysate known to express the target, and a lane with primary omitted to confirm that the secondary is not producing the band on its own. Blots published without at least one of these are relying on trust.

Concentration and incubation conditions trade signal against noise. Too much primary produces non-specific bands; too little produces nothing. An overnight incubation in the cold at low concentration usually gives a better signal-to-noise ratio than a short incubation at high concentration, because specific binding has time to reach equilibrium while weak non-specific interactions are less favoured.

Detection Chemistry and Exposure Time

Most blots are read by chemiluminescence. The secondary antibody carries horseradish peroxidase, which oxidises a luminol substrate in the presence of peroxide, and the oxidised product emits light as it relaxes. That light is captured either on film or by a cooled digital camera.

The essential property of this system is that it is enzymatic and therefore consumes substrate. Emission rises quickly, peaks, and then declines as local substrate is exhausted, which is why a strong band can fade between a first and second exposure while a weak one appears to strengthen. It also means the relationship between the amount of protein and the amount of light is only approximately linear, and it flattens badly at the high end when the enzyme is saturated.

Overexposure is worth naming separately because it looks like success. A band so intense that it hollows out in the centre, appearing white with dark edges, has exceeded the dynamic range completely, and any quantification from it is fiction. Digital imagers that report saturated pixels make this visible; film does not, which is one of the main arguments for imaging over film.

Loading Controls and Normalisation

A blot compares lanes, and lanes only compare fairly if they received equivalent amounts of protein. Total protein is measured before loading, but pipetting errors, incomplete lysis and differential transfer all break the assumption afterwards, so a control is probed to check it.

The traditional approach uses an abundant housekeeping protein, typically a cytoskeletal or glycolytic enzyme, on the assumption that its expression does not change with the experimental treatment. That assumption is the weak point. Those proteins are not inert, and several are known to shift with cell density, stress, differentiation state and metabolic manipulation. Using one as a denominator when the treatment changes it produces a normalised result that moves in the wrong direction.

Abundance creates a second problem. Housekeeping proteins are often present at concentrations far above the target, so at an exposure where the target is visible the control is deep into saturation. A saturated control cannot report a loading difference, because it looks identical whether the lane received the correct amount or half again as much.

Total protein normalisation avoids both issues by staining everything on the membrane and using the summed signal from the lane as the denominator. It is more robust, it uses the full lane rather than a single band, and it does not depend on any assumption about a particular gene. It has become the expectation in careful work, and it is worth adopting even when a housekeeping band is also shown.

Reading Common Failure Patterns

The productive question after a failed blot is not what to change but what the film is telling you. Work backwards through the chain and most results narrow quickly to one or two stages.

A blank film with a good total-protein stain on the membrane means the proteins arrived and detection failed, which points at antibody, substrate or enzyme. A blank film with a blank stain means they never arrived, which points at transfer or at the gel. A grey wash of background with the target faintly visible means blocking and washing, not the primary. A band at an unexpected weight is the most interesting case, because it may be a genuine post-translationally modified form, a cleavage product, an isoform, or an antibody binding something else entirely, and only a knockout or knockdown control resolves that question.

Two habits prevent most repeat failures. The first is to change one variable at a time, which sounds obvious and is routinely violated when a frustrating blot is repeated with new antibody, new buffer and a different exposure simultaneously. The second is to keep the intermediate evidence: the total-protein stain, the exposure series rather than the single best film, and a note of which reagent lots were used.

There is also a limit worth accepting. A western blot is a semi-quantitative technique. It is excellent at answering whether a protein is present, roughly what size it runs at, and whether it changed substantially between conditions. It is poor at resolving small differences, and a figure claiming a modest fold-change from densitometry on a single blot is claiming more precision than the chemistry supports. Knowing where that boundary sits is as much a part of the technique as the pipetting.