Molecular Structure of Thyroxine in Relation to Its Binding by Human Serum Albumin *

In previous work the existence of free thyrox-ine in human serum was verified by dialysis through cellophane; the concentration of the un-bound hormone in normal human serum was found to approximate one-thousandth of the total thy-roxine (1). This direct measurement of free thyroxine was in reasonable agreement with the theoretical value computed from data on the per cent of total thyroxine bound to albumin in serum and from the binding constants for the interaction between thyroxine and human serum albumin (2, 3). This theoretical value of expected free thy-roxine concentration was in accordance with the earlier calculations of Robbins and Rall (4), based upon Lein's equilibrium dialysis studies of the interaction between thyroxine and bovine se-rum albumin (5). Recent studies of the binding of thyroxine by albumin in our own (6) and in other laboratories (7-11) have indicated the existence of a single strong binding site on the protein molecule rather than a primary class of four equal binding loci, as originally proposed (2, 3). The data, to be re-ported separately in detail (6), suggest that the albumin molecule possesses a single strong bind-ing site with an association constant approximating 500,000, a secondary class of three or four sites of intermediate affinity for thyroxine, and a ter-tiary class of many weaker sites. This conceptual revision does not necessitate appreciable alteration' of the calculations of expected concentration of free thyroxine in serum (1, 3). There are ap-proximately 10,000 albumin molecules to one of thyroxine in human serum. Under these circum-stances of low thyroxine to albumin ratio (i.e., at theoretical thyroxine: albumin ratios approach-ing zero), the original model of a primary class of