Abstract. We have tested for antibodies against human and pig eye muscle membrane antigens in the serum of patients with Graves' ophthalmopathy using an enzyme-linked immunosorbent assay (ELISA). Several different membrane preparations were used as source of putative antigen including a 100 000 × g pellet, a pellet depleted of the 100 000 × g (microsome) fraction, and solubilized membranes. With eye muscle membrane pellets there were no significant differences for either serum or immunoglobulins between patients with ophthalmopathy, those with autoimmune thyroid disorders without eye disease, and normal subjects for either human or pig membranes, although tests were positive determined from the upper limit of normal in a few patients with or without eye disease. This was the case regardless of the enzyme-antibody conjugate used, the membrane protein concentration or serum or immunoglobulin dilution. Pre-absorption of tissue fractions, serum, or immunoglobulins, with red blood cells or liver powder, eye muscle membranes or skeletal muscle membranes did not significantly reduce background binding which was often very high, or enhance the difference between patients with ophthalmopathy and normal subjects. It was found that non-specific binding to the plastic surface of the microplates and/or tissue proteins, the presence, in human tissues, of blood-derived immunoglobulins which gave strong reactions in the ELISA, and variable fixation of membrane pellets to the plates were factors which made ELISA unsatisfactory when crude membrane pellets were used as antigen. When eye muscle membranes solubilized with standard agents including SDS, Triton X-100 and deoxycholine were tested, again no differences were demonstrated between patients with Graves' ophthalmopathy and normal subjects. However, when membranes were solubilized with the zwitterionic agent 'CHAPSO' approximately 20% of patients with Graves' ophthalmopathy and smaller proportions of patients with thyroid disease without apparent eye involvement had positive tests with both human and pig eye muscle. While availability of human monoclonal antibodies should soon allow the isolation and purification of soluble eye muscle membrane autoantigens for use in sensitive and specific antibody tests, the ELISA, as presently used with crude tissue fractions, appears too variable and prone to non-specific immunoglobulin-binding for routine clinical use.
