Sandwich direct and competitive ELISA

Sandwich, direct and competitive ELISA are three different ways of arranging the same idea: an antibody binds an epitope, an enzyme or a fluorophore reports that binding, and a plate reader turns the report into a number. The arrangement decides whether the number rises or falls as the analyte rises, and which blank is meaningful. The pillar page ELISA formats, controls and readout sets the controls and the curve. This guide is the format choice itself, for a research sample. It is not a diagnostic method and it does not authorise a clinical claim.

Who is choosing, and what the choice is allowed to support

Use this page when you are about to coat a plate and you still have to say which format matches the analyte. A protein with two accessible epitopes in a messy matrix is a sandwich problem. A purified antigen you already hold is often a direct or indirect problem. A hapten, a drug-like small molecule, or any analyte that cannot host two antibodies at once is a competitive problem. If you only have one antibody, you cannot invent a sandwich.

The number supports a comparison among the wells you built, and a concentration only inside a standard curve prepared in a matrix you can defend.

What each arrangement is physically doing

In a direct assay the antigen is adsorbed to the plastic and one labelled antibody binds it. The label is already on the primary reagent, so there is no second animal's immunoglobulin in the well. Signal rises as more antigen is coated, or as more labelled antibody binds a fixed coat. The format is only as specific as that one antibody, and the coat has to be clean enough that the plastic is not the thing being measured.

An indirect assay keeps the antigen coat and the unlabelled primary antibody, then adds a labelled secondary antibody raised against the primary host. Several enzyme molecules can hang from one primary, so the colour or the luminescence is often stronger. The secondary is also a new source of nonspecific binding. A well that received secondary but no primary is the control that shows whether that extra layer is quiet. People sometimes use the indirect arrangement to measure antibody already present in a sample: the antigen is coated, the sample supplies the primary, and a secondary against the sample species reports it. That is still an indirect assay. It is a different question from measuring the antigen.

A sandwich coats a capture antibody, lets analyte from the mixture bind, washes away what did not bind, and detects the captured analyte with a second antibody on a different epitope. Specificity comes from the requirement that both reagents recognise the same molecule. Sequence context for those epitopes, including isoforms a peptide immunogen might have missed, belongs on a record such as UniProt. If the two antibodies compete for one face of the protein, the sandwich never forms and the plate stays quiet.

A competitive assay limits the antibody or limits a labelled version of the analyte. Sample analyte and the labelled competitor fight for the same binding sites. As the sample analyte rises, less labelled competitor stays bound, and the signal falls. Some competitive plates coat the antigen and let sample analyte compete with that coat for a limited antibody. The direction is the same: more analyte, less signal. Read that direction before you call a pale well a low result.

Reagent classes the format forces you to name

You need a plate whose plastic will adsorb the coat you intend. High-binding polystyrene is the usual class for protein coats. A plate sold for cell culture is a different surface and often a poor coat. You need a coating buffer in the pH window the protocol states, commonly a slightly alkaline carbonate or a neutral phosphate class, and you follow the time and temperature written for that coat. Overnight in the cold and a shorter incubation at 37 Celsius are both used. They are not interchangeable across kits.

Capture and detection antibodies are separate reagents in a sandwich, even when both came from the same animal species, and they must not be the same clone unless you have a reason the epitope is repeated. Detection may be directly labelled with horseradish peroxidase, alkaline phosphatase, a fluorophore, or biotin. A biotin detection layer then needs an avidin or streptavidin enzyme. That amplification fails in matrices rich in biotin. The antibody guide choosing and checking a primary antibody is where specificity of each reagent is argued before you spend a plate proving a format.

Blocking protein, wash buffer, substrate and stop solution are format-independent classes, but the blank you use them in is not. Standards must be the analyte, not a related protein you hope is close enough, diluted in a matched matrix when you can obtain one. Keep the plate map in the same file as the read. Method notes on protocols.io show how other groups structure that map.

A workflow that branches when the blank misbehaves

Coat the capture antibody or the antigen, block, and only then decide the plate is worth samples. The branch point is the control that should be quiet or, in a competitive assay, the zero that should be loud.

If a sandwich zero-analyte well is already as dark as the low standards, stop. The detection antibody is binding the capture antibody or the plastic. Changing the sample dilution will not repair that. Change the blocker, the wash, or the antibody pair, and rebuild the curve. If a sandwich positive control is quiet while a direct coat of the same detection antibody on purified antigen is loud, the capture step or the epitope pairing has failed. Do not add more substrate to a sandwich that never formed.

If an indirect secondary-only well is dark, the secondary is the assay. Dilute it, change the block, or pick a secondary that does not see immunoglobulin in the sample. If a competitive zero standard is pale, the labelled competitor or the limited antibody has failed, and a pale sample well cannot be interpreted as high analyte. If the competitive top standard is as dark as the zero, competition did not happen. The antibody was in excess, or the competitor was not actually labelled.

Only when those controls sit where the format says they should sit do you back-calculate samples, and only between standards.

When the number points the wrong way

The format failure that wastes a whole study is ranking samples in the competitive direction as if they were sandwiches. Export the raw signal next to the back-calculated value so a reviewer can see the direction. The second failure is a sandwich built on one epitope. Both antibodies bind, just never at once, and every sample looks like the zero. A purified analyte spiked into the matrix should climb. If the spike does not climb, the pair is not a pair.

A third failure is matrix. A standard curve in buffer can be steep and pretty while serum, lysate or medium suppresses the same calibrator. Spike recovery and a dilution that lands on the same curve are the checks. If they fail, report a rank or change the matrix treatment. Do not publish the buffer curve as the sample concentration.

Edge wells and a substrate that has already coloured belong to every format. They are not format-specific, but a competitive curve compresses when the enzyme runs too far, because the zero and the high standard both slam into the top of the reader. Stop the reaction while the zero is still inside the instrument's linear window.

Preservatives, acids and the research boundary

Antibody stocks often contain sodium azide. Stop solutions for colorimetric substrates are often dilute acid. Substrates can irritate skin and eyes. Treat the waste as the safety officer's class, not as ordinary buffer. Human and animal specimens that might contain pathogens are an institutional biosafety decision. The WHO Laboratory Biosafety Manual is a public reference for how laboratories frame that decision. It is not a permit for this plate, and a research format is not a diagnostic approval.

Heat, humidity and a reader that stopped

In a laboratory without steady cooling, write the bench temperature beside a coating or substrate step, because a hot hour shortens colour development. Competitive assays feel that first: the zero must stay off the flat top of the reader. Humidity and a loose lid concentrate edge wells, so an edge-only layout can fake a treatment effect. If power returns after a stopped colorimetric plate has waited, read it and record the delay. A missed luminescent read is gone. Do not restart the substrate and treat it as the same kinetic.

What to put in the enquiry

Name the analyte, whether it is a protein or a small molecule, the species, the sample matrix, and how many antibodies you already have with non-overlapping epitopes. Say whether you expect signal to rise or fall, and the concentration window the experiment must resolve. Mention biotin, phosphate modifications, or a sample species whose immunoglobulin will meet the secondary. The reagents and chemicals catalogue is a place to look at buffer, plate and related reagent classes. The molecular biology pathway is the wider methods context. A format can be discussed through the quote request. Ask for the antibody pair and the curve direction in the reply. Do not read a catalogue line as a plate already shown to work in your matrix.