The trafficking of GLUT4, a facilitative glucose transporter, is examined in transfected CHO cells. In previous work, we expressed GLUT4 in neuroendocrine cells and fibroblasts and found that it was targeted to a population of small vesicles slightly larger than synaptic vesicles (Herman, G.A, F. Bonzelius, A.M. Cieutat, and R.B. Kelly. 1994. Proc. Natl. Acad. Sci. USA. 91: 12750-12754.). In this study, we demonstrate that at 37 degrees C, GLUT4-containing small vesicles (GSVs) are detected after cell surface radiolabeling of GLUT4 whereas uptake of radioiodinated human transferrin does not show appreciable accumulation within these small vesicles. Immunofluorescence microscopy experiments show that at 37 degrees C, cell surface-labeled GLUT4 as well as transferrin is internalized into peripheral and perinuclear structures. At 15 degrees C, endocytosis of GLUT4 continues to occur at a slowed rate, but whereas fluorescently labeled GLUT4 is seen to accumulate within large peripheral endosomes, no perinuclear structures are labeled, and no radiolabeled GSVs are detectable. Shifting cells to 37 degrees C after accumulating labeled GLUT4 at 15 degrees C results in the reappearance of GLUT4 in perinuclear structures and GSV reformation. Cytosol acidification or treatment with hypertonic media containing sucrose prevents the exit of GLUT4 from peripheral endosomes as well as GSV formation, suggesting that coat proteins may be involved in the endocytic trafficking of GLUT4. In contrast, at 15 degrees C, transferrin continues to traffic to perinuclear structures and overall labels structures similar in distribution to those observed at 37 degrees C. Furthermore, treatment with hypertonic media has no apparent effect on transferrin trafficking from peripheral endosomes. Double-labeling experiments after the internalization of both transferrin and surface-labeled GLUT4 show that GLUT4 accumulates within peripheral compartments that exclude the transferrin receptor (TfR) at both 15 degrees and 37 degrees C. Thus, GLUT4 is sorted differently from the transferrin receptor as evidenced by the targeting of each protein to distinct early endosomal compartments and by the formation of GSVs. These results suggest that the sorting of GLUT4 from TfR may occur primarily at the level of the plasma membrane into distinct endosomes and that the organization of the endocytic system in CHO cells more closely resembles that of neuroendocrine cells than previously appreciated.