We demonstrated that Themis constitutively associates with Grb2 in thymocytes. Themis2 also associates with Grb2, therefore Grb2-association must be an important feature of Themis family proteins. In the recent peptide inhibition study, the PRS motif of Themis was shown to directly ABT-263 interact with the C-terminal SH3 motif of Grb2. The result from immunoprecipitation experiments using our own anti-Themis monoclonal antibody also supported direct binding of Grb2 to the PRS motif of Themis. As has been recently reported, we also observed defective association of the DPRS mutant Themis with Grb2 in thymocytes. More surprisingly, not only DPRS, but also all other mutants lost the ability to associate with Grb2. These results suggest that the intra-molecular interaction of these domains is crucial to form the proper threedimensional structure to associate with Grb2. Moreover, all of the five mutants in the present study lost not only Grb2 binding, but also tyrosine-phosphorylation. It should be noted that these five mutants also lost weak basal phosphorylation of the protein in unstimulated thymocytes. From our sequential immunoprecipitation experiments, a half of Themis binds to Grb2, and about one tenth of Grb2 binds to Themis. A recent study with T cellspecific Grb2-deficient mice revealed that Grb2 is critical for positive selection. Although Grb2 has been reported to associate with several molecules other than Themis, deficiency of Shc or FAK does not inhibit positive selection. Involvement of Sos in positive selection was also recently shown to be nonobligatory because Sos1/Sos2 DKO exhibited normal positive selection. Therefore, Themis and Grb2 may be cooperatively required for positive selection. Further analysis of Themis/Grb2 DKO would clarify whether Themis and Grb2 are functionally complementary during positive selection. We demonstrated that the PRS, NLS, Core1, and Core2 motifs were all required for Grb2-association, tyrosine-phosphorylation, and positive selection, whereas NLS and CABIT1 are required for nuclear translocation. The CABIT1 domain might be important for the interaction of Themis with nuclear transporter proteins. So far, the causal relationship between positive selection, Grb2- association, and tyrosine-phosphorylation is not yet understood, however, our results demonstrated that all of these are tightly correlated to one another. On the contrary, the current study demonstrated that positive selection and nuclear translocation of Themis are not directly correlated. As a matter of fact, it should be noted that deletion of the NLS motif exhibited the least effect on positive selection compared to the deletion of other motifs, indicating that nuclear translocation of Themis may be less important for positive selection. Finally, this is the first report that describes the significance of the CABIT domain in cellular events.