Loading controls on a western blot

A loading control is the argument that a difference between lanes is not just a difference in how much sample you put down. On a western blot that argument is often a second band, and often it should have been a picture of the whole lane instead. This page is a selection guide. How the protein gets from gel to membrane, and which transfer controls come first, is western blot from gel to membrane. How to decide that the primary antibody is even the right reagent is choosing and checking a primary antibody. DNA gels do not use these protein controls. Their ladder logic is in agarose gel electrophoresis for DNA.

The photograph is the membrane step itself: gloved hands lifting a sheet with forceps over a buffer tray, a rocking shaker behind. No face is in that frame. The sheet those hands are holding is where a loading argument has to be visible. Antibodies and stains are enquiry classes in the molecular biology catalogue. The comparison you need to defend belongs on the quote request.

What you are actually comparing

Write the biological sentence before you pick the control. "Did this treatment change the target, per cell, in the same culture?" is not the same sentence as "Did these two tissues contain the same mass of this protein?" or "Did the immunoprecipitation recover the partner?" A housekeeping protein can serve one of those sentences and sabotage another.

Equal volume is not equal protein. A lysate from treated cells that grew less densely is a smaller load if you pipetted the same microlitres. Quantify the lysate with a method you trust, load similar mass, and then still show that the membrane received it. Quantification in the tube does not survive a bubble during transfer.

A housekeeping band is a biological object

Actin, tubulin and GAPDH are the names people reach for because antibodies exist and the bands are strong. Strength is the hazard. A control that saturates the film in every lane cannot reveal a twofold load error. Choose a control whose signal sits in the same linear window as the target, or accept that you are looking at a ceiling.

Stability is the other hazard. A protein can be a fine loading control in resting fibroblasts and a terrible one in a differentiation time course, a starvation, a drug that remodels the cytoskeleton, or a panel of tissues where its abundance is the biology. UniProt will show you the protein's identity and a sequence mass. It will not promise that your treatment leaves it alone. If the paper you are following used actin, that is a precedent, not a measurement in your lysates. When the treatment is known to touch that pathway, pick something else or stop using a single band.

Housekeeping proteins also have to be separable from the target if you will strip, cut, or overlay. A control that co-migrates with the target cannot be read on the same blot after one detection. Fluorescent secondaries from different hosts are one way to keep them apart. A strip is another, with the failures that stripping carries. Cutting the membrane between size windows works when the masses really differ and you marked the sheet before it was sliced.

Total protein is the other argument

A total-protein stain of the transfer membrane reports the lane rather than one polypeptide's fortunes. Reversible stains such as Ponceau, stain-free methods tied to a particular gel chemistry, and fluorescent membrane stains are the classes. They answer "did similar protein arrive and stay?" They do not answer "is the target specific?" A blank stain means you are not ready to normalise anything. The transfer failed, the load was below what that stain sees, or the sheet never activated.

Use the stain inside its linear range. A lane flooded with signal is the same ceiling problem as a saturated actin band. Background between lanes has to be handled the same way on every blot you will compare. The Human Proteome Organization's public discussion of protein measurement is a reminder that normalisation is part of the claim's quality, not a decorative strip under the figure.

A selection procedure

List what the treatment is known to change. Cross off any "control" protein on that list. If the comparison is across tissues or cell types, assume housekeeping levels differ until you have shown they do not in these samples.

Decide whether the claim is equal loading or equal biology. Equal loading points at total protein. A single-band control is reasonable only when you can defend its stability and its dynamic range for this experiment.

Match abundance. A control you detect in one second while the target needs ten minutes is not in the same window. Dilute the control antibody, shorten its exposure, or choose a less abundant control. Record the exposure that was still grey, not clipped white.

Match the membrane. The control has to be on the sheet that holds the target. A loading control run on a different gel, last week, does not normalise this transfer. The prestained molecular weight ladder shows orientation and the fact of transfer. It is not a loading control. Loading dye in the sample buffer is not one either.

Then bind the antibody rules. A loading-control antibody is still a primary antibody. Host, dilution, and a secondary that matches that host all apply. A secondary-only strip that lights up at the control's size means the control band is not the control. The antibody guide is the checklist. Do not skip it because the band is familiar.

Failure modes

A control band that appears only in some lanes can be biology, a transfer bubble, or a primary that did not cover the sheet. The rocking tray in the photograph only helps if buffer actually moves over the whole membrane. A dry edge stains dark and makes the end lane look overloaded. Look at the total-protein picture before you invent a dose response.

A control that goes up exactly when the target goes up may mean you loaded more, or that both proteins responded. You cannot see which from the target alone. That is the point of having an independent load argument. If total protein is flat and the housekeeping band moves, report the load from the stain and treat the housekeeping change as data, not as the denominator.

Normalising a saturated target to a tidy control produces tidy nonsense. Expose a series. Keep the frame in which both signals are still inside range. Most blots are better at presence and approximate size than at a fold change, which is the same limit the western guide states.

Safety and research limits

The hazards are the blot's hazards: methanol if PVDF was activated, acrylamide if you cast, and the biosafety level of the lysate. Forceps and a wet tray, as in the photograph, are the ordinary handling. This selection guide does not approve a diagnostic claim. A loading control that behaves does not make an unvalidated antibody a clinical test. Institutional biosafety practice is the authority for the cells. The WHO Laboratory biosafety manual, 4th edition is background for that decision, not a permit.

Incubations that ran hotter than the protocol

In a warm laboratory the "room temperature" incubation is warmer than the room the protocol was written in. Background on the control and the target can rise differently, so a ratio from a hot afternoon is a shaky thing to compare with a blot made in a cooler week. Note the incubation temperature. A cooled room or a shorter time is a documented change. Do not pool those blots into one figure and call the housekeeping ratio a biological constant. If a power cut stops the shaker, the sheet sits in antibody unevenly. That membrane is a poor loading argument. Repeat the probe or the blot.

What to put in an enquiry

Name the sample (lysate, tissue, immunoprecipitate), the comparison (treatment, genotype, fraction), the target mass, and whether you need a housekeeping antibody or a total-protein stain compatible with your membrane. Say if the treatment is already known to change actin, tubulin or GAPDH. Ask for host species so the secondary matches, and ask whether a quotation is possible. The nucleic acid analysis pathway is the neighbouring path when the same project also orders nucleic-acid reagents. A control described only as "loading" has not told anyone which claim it is supposed to carry.