Cross-reactivity and related proteins

Protein families share domains. An antibody raised against one member can bind the cousins, and the assay will report the cousin as if it were the protein on the label. Species is a second axis, not a footnote to the first. A reagent directed at a human protein may or may not bind the mouse orthologue, and a datasheet sentence is a claim you retest in the matrix you actually run. This overview is that check. Formats and curves, which cannot substitute for it, are in ELISA formats, controls and readout. The broader habit of doubting a primary reagent is choosing and checking a primary antibody.

Who needs the check before the first real sample

Run it when the target is a member of a family, which is most signalling proteins, most receptors and a great many enzymes. Run it when the samples are a different species from the immunogen. Run it when the conclusion is "this member changed and the other member did not". That conclusion is exactly the one a cross-reactive antibody invents or hides.

The decision you are allowed to take, once the related-protein lane is quiet, is that this format does not confuse the target with that named relative at the dose you tested. You are not allowed to extend the sentence to every untested paralogue, every species, or every fixation. Specificity is a list of negatives you actually ran.

Two axes, drawn separately

Paralogues arise by duplication. They often share a domain layout and differ in the loops and the termini. An immunogen taken from a shared domain is a recipe for family-wide binding. An immunogen taken from a unique terminus can be selective and can also fail the moment a splice isoform clips that terminus. You do not get to guess which you bought. You compare sequences.

Orthologues are the same gene in another species, insofar as "same" survives evolution. Percent identity across the whole protein is a weak guide to a short epitope. Twenty identical residues in the immunogen matter more than ninety percent identity everywhere else. A mouse protein can be an excellent orthologue on a domain the antibody does not see, and a single substitution in the epitope can abolish binding. The reverse, unexpected binding, happens when the epitope was the conserved piece.

Write both axes in the notebook as separate rows. "Reacts with mouse" does not answer "reacts with human paralogue B". "Does not see paralogue B" does not answer "works on the rat orthologue".

A sequence check before anyone coats a plate

Open the target entry and the nearest relatives on UniProt. Confirm you have the accession the antibody claims, including the isoform. Repeat the lookup on NCBI Protein if the supplier quoted an older identifier. Align the immunogen, or the stated epitope, against each relative. Mark identical stretches long enough to be an epitope, a handful of residues and up. A perfect match is a prediction of cross-reactivity, not a decoration.

If the supplier withholds the immunogen, you cannot do this alignment. You can still do the empirical lane. Withholding the peptide does not reduce the biological risk. It only reduces your ability to predict which lane to run. Choose the closest paralogues by domain architecture instead, and say in the methods that the epitope was undisclosed.

Splice isoforms deserve a line of their own. An antibody against an exon the short isoform lacks will look beautifully specific and will also look like a biological loss of protein whenever the short isoform is the one the cell expressed. That is a real molecular fact and a wrong biological sentence if you claimed the whole gene went away.

The lane that has to be quiet

Empirical test, same format as the assay you will trust. For a blot, load the target lysate, a lysate or recombinant of the paralogue, and the orthologue if species is in scope. For an ELISA, coat or spike purified target and purified relative at matched molar amounts, including an amount at the top of the range you care about. A relative tested only at a trace dose does not clear a relative that is abundant in the sample.

A cell that expresses the paralogue and not the target is often more honest than a pure protein, because the matrix matches. A pure protein is more honest about dose. Do both when the claim is important. Keep a secondary-only lane so a band at the immunoglobulin mass is not called a paralogue.

How other groups document such a panel is scattered through public protocols on protocols.io. Copy the habit of naming accessions. Do not copy their conclusion about your antibody.

Interpret the quiet lane tightly. Quiet means no signal above the secondary-only background at that load. It does not mean the antibody would stay quiet at a tenfold higher load, or in a different buffer, or after a different retrieval. A visible band means you measure a sum. You can still use the antibody if an orthogonal separation, such as a distinct mass on the blot, lets you score only the target. An ELISA usually has no such separation. A cross-reactive ELISA antibody is the wrong reagent for a member-specific claim, however handsome the standard curve.

Sticky proteins that are not relatives

A band at an absurd mass, a smear, or a signal that vanishes when the secondary is omitted is not a paralogue. It is stickiness or immunoglobulin. Do not spend a week aligning sequences for a band the secondary-only lane already explains. Do not do the opposite either: a clean secondary-only lane and a second specific-looking band at the paralogue's mass is exactly the family problem. Confirm the mass against the sequence record before you name the band.

Charge-based sticking often changes when salt or detergent changes. Epitope sharing does not. If a harsher wash removes the extra band and keeps the target, you learned something about stickiness. If both bands fade together, you have not separated them. Say so.

A standard curve of the pure target will not mention any of this. Samples that contain the paralogue will read high, recovery of a spike can still look acceptable, and the biology will be wrong. The related-protein well is the control the curve forgot.

Lysates, recombinant proteins and institutional rules

A recombinant relative expressed in bacteria or in a mammalian cell is a research material with the hazards of the expression system: antibiotics in the bacterial work, and the biosafety level of the cells. A lysate from human material keeps that assignment after you decide it is "just a control lane". Follow the institutional rule. The WHO Laboratory Biosafety Manual is a public reference for how such rules are framed. A cross-reactivity panel is not a diagnostic identification of a protein in a patient sample.

A recombinant control that travelled through heat

The related protein is often the reagent you ordered once and then trusted for a year. If the cold chain broke on the dock, the protein may have aggregated or lost the fold the antibody sees. A false "quiet lane" from a dead recombinant is how a cross-reactive antibody gets a pass. When a shipment arrives warm, or after a power cut thawed the control box, re-establish that the recombinant is still detectable by a method that does not depend on the antibody under test, such as a tag antibody or a stained gel of the expected mass. Then repeat the cross-reactivity lane. A quiet lane from a missing protein is not specificity.

Humidity and repeated thaw do the same damage more slowly. Aliquot the relative. A control that has been thawed six times is a different dose from the one you calculated.

What the enquiry should name

Name the target accession, the species you will sample, and the paralogues you already know you must exclude. Paste nothing sentimental. Paste the accessions. Say whether you need the immunogen sequence disclosed so you can align it, and whether you need a recombinant of the relative as well as the antibody. The reagents and chemicals catalogue is a place to look at reagent classes. The molecular biology pathway is the methods context. A cross-reactivity check can be discussed through the quote request. Ask the reply to state which species and which family members were actually tested. A sentence that says "specific" without that list is a description of hope.