How does an ELISA work and which protein to coat?

In this article, I shortly explain the basic principle of an ELISA style binding assay and discuss the first considerations in creating an ELISA style binding assay.

How does an ELISA work?
Normally assays are executed in 96 well plates (very common: NUNC Maxisorp). One protein is immobilised to the plastic by hydrophobic interactions and the remaining bindings sites are blocked by a blocking agent. Subsequently the second protein is exposed to the well either in a concentration series for determination of affinity constants or as a single concentration for screening purposes. After washing away unbound proteins, the bound protein is detected specifically by an antibody (in our days often tag specific). This first antibody is then detected by a secondary antibody coupled to an HRP or a fluorescent dye, which in turn either catalyse a colour reaction or is directly measured by a fluorescent reader.
Initial thoughts…
To develop an ELISA assay, you first decide on which protein will be coated (the ligand) and which will stay in solution. Ideally, you try both options – but this is often not possible:
1) The coated protein (=ligand)
The ligand (the protein you are going to coat) should be rather pure. If you use an unsual ligands (like DNA, peptides or membrane proteins) you need to check, whether your compound really immobilises on the ELISA plate (oligos and small peptides do normally not!). If not use you might try a different plastic surface or perform a sandwich ELISA [link]. In my experience apart from the above mentioned exceptions coating is not problematic – however, if it is (anticipated), the best way around the problem is to use the protein as the analyte instead
2) The soluble protein (=analyte)
The analyte does not need to be as pure as your ligand, as you will detect it specifically with the antibody. It is not unusual to use blood serum or other complex mixtures as the analyte. However, as you normally have no way to know the proper concentration of your protein in complex mixtures, only simple yes/no answer can be derived from this kind of assay (1). To detect your bound protein, you need a specific antibody. If you use a protein tag as an epitope, you have to be sure that your ligand does not have the same epitope. Cutting protein tags off, might sometimes alleviate this problem; however, in my experience there is often enough uncut protein left to give strong “background” problems. If you have no good antibody, biotinylation [link] of your protein might help, although this procedure might turn your protein incapable of binding. Your ligand should not tend to aggregate or precipitate, as this will give false positive signals. You might consider using such a protein (like collagens, which I use a lot) as coat instead.
Additionally, protein amounts might also influence your decision. While you use only 500ng of your protein per well for coating typically 2-4 µg are necessary for high analyte concentration.

read more