The immunochemistry of sandwich ELISAs--I. The binding characteristics of immunoglobulins to monoclonal and polyclonal capture antibodies adsorbed on plastic and their detection by symmetrical and asymmetrical antibody-enzyme conjugates.

Radiolabelled bovine IgG1, IgG2, SIgA and IgM and heavy-chain specific polyclonal and monoclonal antibodies to these isotypes were employed as models to investigate immunochemical aspects of sandwich enzyme immunoassays (ELISAs). The titration plots obtained by measuring enzyme activity paralleled those obtained when the binding of radiolabelled immunoglobulins to solid-phase capture antibodies was quantitated. As predicted from the Mass Law, the percentage of labelled immunoglobulin which was bound remained constant over the range in which the sandwich ELISA titration was linear on a log-log plot. Also as predicted from the Mass Law, increasing the solid-phase concn of polyclonal antibodies by affinity purification increased the linear region of the log-log ELISA plot and the corresponding region over which a constant percentage of immunoglobulin binding was observed. When used as capture antibodies adsorbed on plastic at equal concns, the best monoclonal antibodies were 1/8- less than 1/16 as effective as their polyclonal counterparts in binding iodinated bovine immunoglobulins; these differences can be directly interpreted to result from an 8 and greater than 16-fold higher functional, relative affinity of the polyclonal reagents. Steric hindrance was shown to occur when symmetrical sandwich ELISAs, i.e. capture and detection antibody are both heavy-chain specific, are used to measure monomeric but not IgM immunoglobulins. The use of an asymmetrical configuration, i.e. anti-Fab antibody-enzyme conjugates, avoids this problem. Symmetrical conjugates based on the avidin-biotin system, horseradish peroxidase or alkaline phosphatase, were less effective than their asymmetrical (anti-Fab) counterparts. Evidence that the lower activity of symmetrical conjugates was due to steric hindrance was illustrated using horseradish peroxidase-antibody conjugates of different sizes. Sandwich assays using affinity-purified, polyclonal solid-phase antibodies and an asymmetrical conjugate were judged to be immunochemically and economically optimal. Using an asymmetrical configuration, the non-linear nature of sandwich ELISA titration plots is the predictable result of changing antibody to antigen ratios in an antibody-limiting system, and not the result of steric hindrance of the detection system.