Choosing and checking a primary antibody

Choosing and checking a primary antibody is an experiment, not a catalogue filter. The reagent binds an epitope. Your sample contains thousands of other surfaces. A datasheet photograph is evidence about the supplier's sample, under the supplier's conditions. Your job is to show that the band, the stain, or the ELISA signal in your hands depends on the target you named. Plate formats and curves are the companion pillar ELISA formats, controls and readout. This guide is for research use. It is not a diagnostic validation.

What to read before you buy

Start with the target as a sequence, not as a gene symbol that several species share. UniProt and Ensembl let you check whether the immunogen peptide exists in the species and the isoform you study. An antibody raised against a human peptide can fail on a mouse protein that differs at that stretch, or it can cross-react with a related protein that kept the stretch. Write the accession you checked.

Host species has to work with the detection system. A mouse primary antibody on mouse tissue, detected with an anti-mouse secondary, will paint endogenous immunoglobulin unless you have a method that avoids that. The host should also differ from the capture or secondary reagents in a sandwich assay so the enzymes bind the right layer.

Clonality is a trade. A monoclonal antibody is one epitope, one reproducible binding site, and a risk that the epitope is modified, cleaved, or shared. A polyclonal antibody is a mixture that can survive partial epitope loss and can change when a new animal is immunised. Recombinant antibodies are a class that can be defined by sequence, which makes lot changes easier to discuss. None of these labels replaces a gel or a slide.

Application claims are narrow. An antibody that works on a denatured blot may fail on a fixed tissue section, because fixation hides or destroys epitopes. The reverse happens too. The photograph of a stained slide is a reminder that tissue staining has its own controls: morphology, a negative tissue, and a no-primary slide. Do not transfer a dilution from a blot to a section and call it the same assay.

How to check it with samples you understand

A positive control expresses the target for a reason you can explain: a lysate from a cell line known to express it, a recombinant protein, or a tissue where independent evidence says the protein is present. A negative control lacks it for a reason you can explain. A knockout or knockdown is the strongest genetic control. Editing resources such as the Addgene CRISPR guide are a public orientation to how research laboratories build that kind of cell, not a protocol you can skip validating. If you do not have a knockout, a cell or tissue that genuinely does not express the transcript is a weaker but still useful negative, provided you did not choose it only because the antibody was blank there.

The no-primary control uses blocking and the secondary detection exactly as in the real samples, without the primary antibody. Signal there belongs to the secondary, to endogenous enzyme, or to sticky tissue. Blocking buffers, discussed for plates in the ELISA article, matter on blots and on slides for the same reason. Change the blocker when the no-primary slide is already stained. Do not only shorten the development.

An orthogonal check uses a different physical principle. A band that appears with two antibodies against different epitopes, or a band that matches a mass-spectrometric identification of that slice, is stronger than a second aliquot of the same antibody. Peptide pre-absorption can reduce a specific signal and can also reduce a non-specific signal if the non-specific binder likes that peptide. Treat it as one clue.

Dilution series tell you where the specific signal remains and the extra bands fade, or where everything fades together. The second pattern is not specificity. Record the lot number. A new lot of a polyclonal antibody is a new reagent.

Where a standard curve belongs

Once the signal depends on the target, quantification can be discussed. On a blot, densitometry needs a linear range and a loading control that is stable in your treatment. On an ELISA, a standard curve, built as the format article describes, converts signal to concentration inside that curve. Spike recovery and dilution parallelism still apply. A curve does not repair a no-primary well that is already bright, and blocking does not repair an antibody that binds the wrong protein in both the standard and the sample.

Failure modes

A single band at the "right" size in every sample, including the negative, is a cross-reactive protein or a loading artefact. Extra bands that vanish in the knockout may be true fragments. Extra bands that remain are not. Tissue edges that stain only at the margin are often drying artefacts. Nuclear stain with an antibody against a secreted protein is a reason to stop and look at the secondary and the antigen retrieval, not a reason to rewrite the cell biology.

Isotype controls match the antibody class and do not match the specificity. They help with some flow-cytometry backgrounds and do not prove a band on a blot. Trusting a dilution posted for a different tissue is a common way to over-stain. Find the dilution in your system.

Safety, storage and the enquiry

Preservatives such as sodium azide are toxic and also inhibit horseradish peroxidase if they are carried into a detection step that uses that enzyme. Fluorophores fade. Fixed tissue and human samples follow your biosafety rules. This page does not interpret a stain as a diagnosis.

Antibodies are cold-chain reagents. Record the arrival temperature. Repeated freeze-thaw cycles of a small vial kill some clones and leave others. Aliquot on arrival if the datasheet says to freeze, and keep the lot identity on the aliquot. In a humid cold room, labels fall off. A power cut that warms a refrigerator is a reason to re-run the positive and negative controls before a large experiment. Do not assume a vial that stayed "cool to the touch" is unchanged.

For a sourcing note, name the target accession, the species of the sample, the application, the host species you can detect, any fixation, and the positive and negative samples you will use to check the lot. Start from the reagents and chemicals catalogue and the molecular biology pathway. Use the quote request to ask whether a quotation is possible. A family name on a product line is not a promise of a particular clone, and this site does not claim to hold that vial or to validate it for you. Ask for the immunogen and the application data with the offer.

Method notes on protocols.io are a reasonable place to see how other groups document antibody controls. Keep your own images with the lot numbers attached.