Following the discovery by Adams and Purves in 19561 of an abnormal substance in the blood of patients with hyperthyroidism, renewed interest has developed concerning the pathogenesis of Graves' disease. The occurrence of this substance, now referred to as the long-acting thyroid stimulator (LATS), has been confirmed by others,2 3 and it has been found in the serum of the majority of patients with hyperthyroidism. It is almost always detectable in those patients with exophthalmos and localized pretibial myxedema.4 Despite the suggestion that this factor is of etiologic importance in Graves' disease, little has been known of its nature or origin, and some doubt has been expressed whether it actually represents a distinct entity.5 Various attempts to concentrate or isolate the active principle have been only partially successful. The percolation method for the purification of thyrotropin has failed.6 Recently, a 5- to 10-fold concentration by means of filtration through Sephadex gel was described.' This report describes a method of purification of the lollg-acting thyroid stimnulator in which the activity was found to reside exclusively in the gamma globulin fraction of the serum proteins. Biologic activity was further identified in subunits of the gamma globulin; however, these fragments, in contrast to the intact material, produced an early peak response similar to that obtained with thyrotropin. Materials and Methods.-Sources of plastma: The starting material used in the purification procedure was lyophilized citrated plasma or lyophilized serum from two patients with hyperthyroidism, exophthalmos, and consistently high titers of LATS-activity in the unaltered serum. One patient had pretibial myxedema; neither patient had an abnormal distribution of serum proteins as determined by paper electrophoresis. Purification procedure: The lyophilized plasmna, or serumir, was dissolved in distilled water to make a 2.5% solution. Almmonium sulfate was added to a 40% concentration at 4?C. The precipitate was allowed to form overnight in the cold, collected by centrifugation, then dissolved and dialyzed against 0.1. M sodium chloride for 12 hr. The solution was then dialyzed 12 hr against 0.005 M phosphate buffer, pH 7.0. The euglobulin precipitate was removed, and the clear protein solution was passed through a 2.4-cm X 20-cmIm column packed with diethylaminoethylcellulose (DEAE-cellulose) which previously had been equilibrated with the phosphate buffer. The column load of the DEAE-cellulose (exchange capacity, 0.7 mEq/gm) was limited to 1.0 gm of protein. The unabsorbed protein fraction (gamma globulin) was dialyzed free of phosphate and was stored in the lyophilized state; the absorbed protein was removed by the passage of 0.2 M potassiun phosphate through the column, dialyzed against 0.1 M sodium chloride, and lyophilized. Neutralization studies: The active gamma globulin was dissolved in saline, mixed with an equal volume of rabbit antibody to human 7S gamma globulin (Hyland Laboratories, Lot. no RP 763F), and incubated at 37?C for 90 min. The solution was then held at 4?C overnight. Although no visible precipitate formed, the solution was centrifuged and the upper portion removed for assay. A control sample of active gamma globulin was treated in an identical fashion with the exception that saline was substituted for the rabbit antibody. Reduction of the active gamma globulin: The method for the reduction and fractionation of ganmmna globulin described by Fleischman et al.8 was used with minor modifications. A 2% solution of