A Golgi retention signal in a membrane-spanning domain of coronavirus E1 protein

G protein has been shown to form a noncovalently associated homotrimer before its exit from the ER (6) . We tested the oligomeric structure of Gml on sucrose gradients after a pulse-chase label (Fig. 3) . After a 10 min label, wildtype G protein was -50% trimer (8S) and 50% monomer (4S), consistent with the results of Doms et al. (6, 7) . After 20 min ofchase, all the G protein was found in the 8S trimer peak. Although apparently a monomer after the 10 min label, Gml pelleted after the 20 min chase. Other centrifugation conditions suggested this oligomer was between 15 and 20S (data not shown) . Several mutant G proteins that are grossly misfolded were also shown to pellet under the standard gradient conditions (7; and unpublished results), but unlike Gml, they pelleted immediately after the pulse label. The simplest interpretation ofour results is that Gml was retained specifically by the ml sequence. However, we cannot distinguish whether inability to trimerize resulted in retention of Gml in a subcompartment of the ER (near the Golgi region) or if the large oligomers (with or without other proteins) were the result of specific retention in the Golgi complex. These points will be discussed below.