This study was approved by the ethics committee of this hospital

was not observed in immunoprecipitates prepared from cells from which HA-tagged GABARAPL2 was absent. As Jurkat-T cells express GIMAP proteins endogenously, we used these cells to BHI1 chemical information address the question of whether an interaction could be detected between GIMAP6 and GABARAPL2 when expressed at endogenous levels. Using novel monoclonal antibodies that we prepared to the two proteins, we found that, after cross-linking, a proportion of the material immunoreactive with our anti-human GIMAP6 antibody in Western blotting acquired a slower mobility. The apparent size shift was consistent with the earlier experiments using the internally biotinylated GIMAP6 construct as well as with the molecular weight of GABARAPL2. Probing of a second Western blot run in parallel with an antibody against GABARAPL2 revealed a cross-linked product of the same size, suggesting strongly that this corresponds to the same complex. Interestingly, the principal complex formed by GABARAPL2 appears to be that with GIMAP6, although some material is cross-linked into a higher molecular weight complex. Unfortunately, we have not been able to demonstrate directly an interaction between the two endogenous proteins by co-immunoprecipitation. This could be due to the inability of our antibodies to recognise the native complex in vivo or could reflect the transient nature of the interaction between the proteins expressed at endogenous levels, which is captured by cross-linking but which is too brief to survive a normal immunoprecipitation protocol. As GABARAPL2 has been shown to have key roles in autophagy, we investigated the possibility that induction of autophagy might allow us to detect GIMAP6GABARAPL2 interaction. However, although starvation of Jurkat-T cells induced autophagy, as assessed by the formation of 18509334 MAP1LC3B-II from MAP1LC3B-I, we were still unable to show co-immunoprecipitation of GABARAPL2 with GIMAP6. To address the issue of whether the interaction between GIMAP6 and GABARAPL2 was direct or required additional proteins, glutathione-S-transferase or a GST-GIMAP6 fusion protein were expressed in Escherichia coli and the two proteins purified on glutathione beads. The beads carrying the proteins were then incubated with bacterially expressed purified GABARAPL2. After extensive washing, the GST proteins were eluted from the beads with glutathione and the eluates assessed for the presence of associated GABARAPL2. Although GST was expressed at much higher levels than the GST-GIMAP6 fusion protein, whereas no GABARAPL2 was found to 18201139 elute with GST, it was detected in the GST-GIMAP6 eluates, indicating that the interaction between GIMAP6 and GABARAPL2 is direct and does not require other proteins. Specificity of the GIMAP6-GABARAPL2 interaction Both GIMAP6 and GABARAPL2 are members of multigene families and, in both cases, individual members of these families are predicted to perform related functions. It was therefore important to determine the selectivity of the observed interaction. Using pull-downs from co-transfected HEK293T cell lysates, we first tested the ability of GABARAPL2 to discriminate between mammalian GIMAP family members. Biotin-tagged human GABARAPL2 proved to be highly selective for GIMAP6 amongst a panel of all of the human GIMAP proteins. Next we tested the ability of biotin-tagged GIMAP6 to pull down HA-tagged human Atg8 family members, namely MAP1LC3A, MAP1LC3B, MAP1LC3C, GABARAP, GABARAPL1, and GABARAPL2. In this case, whilst GIMAP6 interacted s