To investigate the interaction of PICK1

with Arf1, we performed GST-PICK1 pull-down assays with a constitutively active mutant of Arf1 (Arf1Q71L) or a nucleotide-binding-defective mutant (Arf1T31N) expressed in COS cells. GST-PICK1 interacts specifically with the constitutively INCB024360 mouse active Arf1Q71L mutant, showing negligible binding to Arf1T31N, suggesting that the PICK1-Arf1 interaction is GTP dependent (Figure 1A). To test this further, we carried out GST-PICK1 pull-down assays with purified his6-Arf1 in the absence of other proteins and in the presence of either nonhydrolyzable GTP (GTPγS) or guanosine diphosphate (GDPβS). Arf1 binds PICK1 only in the presence of GTPγS, demonstrating a direct GTP-dependent interaction of Arf1 with PICK1 (Figure 1B). To investigate the PICK1-Arf1 interaction in native tissue, we carried out coimmunoprecipitations (co-IPs) from neuronal extracts using PICK1 antibodies in the presence or absence of GTPγS.

Arf1 interacts with PICK1 only in the presence of GTPγS, demonstrating that a GTP-dependent PICK1-Arf1 complex exists in neurons (Figure 1C). The GluA2-PICK1 interaction is unaffected by the presence of GTPγS (Figure 1C and Figure S1A available online). Since a major function of PICK1 is regulating AMPAR trafficking via an interaction with the GluA2 subunit, we assessed whether PICK1 can complex with both GluA2 and Arf1 simultaneously. Co-IP from cultured neuronal extracts using anti-GluA2 antibodies demonstrates that Arf1 is in a

GTP-dependent complex with GluA2 (Figure 1D). The Apoptosis Compound Library GluA2-Arf1 complex is disrupted following transduction of neurons with Sindbis virus expressing a peptide (pep2-EVKI) that inhibits AMPAR-PICK1 interactions (Terashima et al., 2004 and Terashima et al., 2008), demonstrating that Arf1 associates with GluA2 via PICK1 (Figures 1D and S1B). To confirm AMPAR subunit specificity of this interaction, we carried out co-IP experiments from transfected HEK293 all cells. Endogenous Arf1 forms a complex with PICK1flag and mycGluA2 but not mycGluA1 (Figure S1C). To analyze the subcellular distribution of Arf1, we carried out differential detergent fractionation of synaptosomes prepared from brain tissue. It has previously been shown that PICK1 is present in synaptosomal fractions as well as PSD fractions I and II but not the core PSD III fraction (Rocca et al., 2008 and Xia et al., 1999). Arf1 shows a strikingly similar distribution to PICK1, demonstrating that both proteins are found in the same subcellular fractions, and are both loosely associated with the postsynaptic density (Figure 1E). Arf6 has also been implicated in AMPAR trafficking during LTD (Scholz et al., 2010), so we investigated whether this related protein binds PICK1. GST pull-downs demonstrate that Arf6 does not interact with PICK1 (Figure 1F).