A solid-phase ELISA was set up to measure the direct binding capacity (BC) of different, commercially available, purified human IgA preparations to plates coated with human fibronectin (FN). It was found that secretory, polymeric, and, to a much lesser extent, monomeric IgA exhibited elevated FN-BC as compared to their BC to plates coated with bovine serum albumin. This binding was specific since not observed with human IgG or IgM antibodies. In addition, we noted that this interaction was dose dependent, Ca2+ dependent, saturable, and not covalent, was inhibited by soluble FN, but not by a prior incubation of FN-coated plates with anti-human fibronectin antibodies, and appeared to involve on the dimeric FN other structures than its heparin-binding, collagen-binding, or C1q-binding domains. Similar experiments conducted with normal plasma indicated that plasma IgA, but not plasma IgG or IgM, was also capable of significant binding to FN-coated plates. In contrast, serum IgA did not significantly bind to those plates under otherwise identical experimental conditions. Thus, the coagulation process induces a strong decrease in the FN-BC of circulating IgA, which implies the necessity of using plasma rather than serum to study such interactions. The apparent molecular weight of plasma IgA interacting with FN-coated plates ranged between 450 and 900 kd, and its major binding characteristics were quite similar to those observed with purified polymeric IgA. The FN-BC of plasma IgA was then measured by the same ELISA in 30 patients with primary IgA nephropathy (IgAN) and in 23 healthy controls. The mean FN-BC of plasma IgA was significantly higher in patients than in normal controls. This enhancement was due mainly to the augmentation in the concentration of circulating "macromolecular" IgA and was significantly correlated with the plasma levels of IgA-FN complexes. However, the pathogenetic role of these findings was probably not determinant since similar observations were made in alcoholic liver cirrhosis without urinary abnormalities and since the FN-BC of plasma IgA or the plasma levels of IgA-FN complexes were not correlated with the various biological parameters of evolutivity of primary IgAN. In conclusion, these studies suggest that the ability of polymeric IgA to directly bind to FN is involved in the formation of circulating IgA-FN complexes and that this normal binding process, although enhanced in IgAN, is probably not responsible for kidney injury, at least in the patients studied.