Phosphorylation of Stargazin by Protein Kinase A Regulates Its Interaction with PSD-95*

Stargazin is the first transmembrane protein known to associate with AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionate) glutamate receptors (AMPARs) and regulate their synaptic targeting by two distinct mechanisms, specifically via delivery of AMPARs to the surface membrane and synaptic targeting of these receptors by binding to PSD-95/SAP-90 and related PDZ proteins. However, it is not known whether and how this stargazin-mediated synaptic targeting of AMPARs is regulated. Stargazin interacts with the PDZ domains of PSD-95 through the C-terminal PDZ-binding motif. The stargazin C terminus contains a consensus sequence for phosphorylation by cAMP-dependent protein kinase A (PKA). Phosphorylation site-specific stargazin antibodies reveal that the stargazin C terminus is phosphorylated at the Thr-321 residue in heterologous cells and in vivo. Stargazin phosphorylation is enhanced by the catalytic subunit of PKA. Mutations mimicking stargazin phosphorylation (T321E and T321D) lead to elimination of yeast two-hybrid interactions, in vitrocoimmunoprecipitation, and coclustering between stargazin and PSD-95. Phosphorylated stargazin shows a selective loss of coimmunoprecipitation with PSD-95 in heterologous cells and limited enrichment in postsynaptic density fractions of rat brain. These results suggest that phosphorylation of the stargazin C terminus by PKA regulates its interaction with PSD-95 and synaptic targeting of AMPARs.


From the Department of Biological Sciences, Korea Advanced Institute of Science and Technology, Daejeon 305-701, Korea
Stargazin is the first transmembrane protein known to associate with AMPA (␣-amino-3-hydroxy-5-methylisoxazole-4-propionate) glutamate receptors (AMPARs) and regulate their synaptic targeting by two distinct mechanisms, specifically via delivery of AMPARs to the surface membrane and synaptic targeting of these receptors by binding to PSD-95/SAP-90 and related PDZ proteins. However, it is not known whether and how this stargazin-mediated synaptic targeting of AMPARs is regulated. Stargazin interacts with the PDZ domains of PSD-95 through the C-terminal PDZ-binding motif. The stargazin C terminus contains a consensus sequence for phosphorylation by cAMP-dependent protein kinase A (PKA). Phosphorylation site-specific stargazin antibodies reveal that the stargazin C terminus is phosphorylated at the Thr-321 residue in heterologous cells and in vivo. Stargazin phosphorylation is enhanced by the catalytic subunit of PKA. Mutations mimicking stargazin phosphorylation (T321E and T321D) lead to elimination of yeast two-hybrid interactions, in vitro coimmunoprecipitation, and coclustering between stargazin and PSD-95. Phosphorylated stargazin shows a selective loss of coimmunoprecipitation with PSD-95 in heterologous cells and limited enrichment in postsynaptic density fractions of rat brain. These results suggest that phosphorylation of the stargazin C terminus by PKA regulates its interaction with PSD-95 and synaptic targeting of AMPARs.
Stargazin is a transmembrane protein that interacts with AMPARs 1 and regulates their synaptic targeting (1,2). The stargazer, a spontaneous mutant mouse (3) with defects in the stargazin gene (Cacng2) (4), displays an absence seizure (also known as petit-mal or spike-wave) and, as the name implies, a head-tossing movement, probably because of a defect in the vestibular system (3). In addition, stargazer mice develop an ataxic gait (3) and severe impairment in classical eye-blink conditioning (5), probably because of a cerebellar malfunction. Both mRNA and protein levels of brain-derived neurotrophic factor are selectively reduced in cerebellar granule cells of stargazer mice (5,6). Stargazin, a protein with a calculated molecular mass of 36 kDa, contains four putative transmembrane domains and a cytosolic C terminus, and its primary structure is closely related to that of the ␥ subunits of voltagegated calcium channels (7)(8)(9)(10). Stargazin (or ␥-2) associates with neuronal calcium channel subunits in vivo (11) and inhibits calcium channel activity by increasing steady-state inactivation (4,7,11,12).
The functional association between stargazin and AMPAR was initially ascertained by the observation that postsynaptic AMPAR currents are selectively impaired in cerebellar granule cells of stargazer mice (13). A subsequent study revealed that stargazin mediates synaptic targeting of AMPARs by two distinct mechanisms (14). Stargazin initially interacts with AM-PARs and assists their translocation to the extrasynaptic surface membrane. Next, the AMPAR-stargazin complex is targeted to synaptic sites by binding to PSD-95 and related PDZ proteins. In support of this hypothesis, a stargazin mutant lacking the last four residues (stargazin⌬C) rescues extrasynaptic but not synaptic AMPAR currents in cerebellar granule cells of stargazer mice (14). However, little is known about whether the stargazin-mediated synaptic targeting of AMPARs is regulated and, if so, what these regulatory mechanisms involve.
The C terminus of stargazin contains the end sequence RRT-TPV, which belongs to the class I PDZ-binding motif, (S/T)XV (S/T, Ser or Thr; X, any aa residue; V, hydrophobic residue) (15)(16)(17). Interestingly, the RRTT sequence of the C terminus additionally corresponds to the consensus sequence for phosphorylation by PKA, (R/K)(R/X)X(S/T), suggesting that Thr at the Ϫ2 position (RRTTPV, designated T321) is phosphorylated by PKA. The crystal structure of the PDZ3 domain of PSD-95 (class I) complexed with the C terminus of CRIPT, a PSD-95binding protein that ends with the QTSV sequence (18), reveals that the Thr residue at the Ϫ2 position interacts with His-372 of PDZ3 (15). Specifically, the hydroxyl oxygen of the Thr forms a hydrogen bond with the N-3 nitrogen of His-372. Therefore, phosphorylation of T321 at the stargazin C terminus may weaken the interaction between stargazin and the PDZ domains of PSD-95. Consistently, earlier data demonstrate that phosphorylation of the Ser residue at the Ϫ2 position of Kir2.3 (an inward rectifier potassium channel) by PKA disrupts interactions with PSD-95 (19).
Here we report that PKA phosphorylates the stargazin C terminus and consequently regulates its interaction with PSD-95. In view of the hypothesis that stargazin regulates synaptic targeting of AMPARs by binding to PSD-95 and related PDZ proteins, our results suggest that phosphorylation of the C terminus of stargazin regulates AMPAR synaptic targeting.

EXPERIMENTAL PROCEDURES
Antibodies-Anti-fusion protein stargazin antibody (Stg-Cyto 1222, guinea pig polyclonal) was generated using an H6 fusion protein containing two copies of the C-terminal cytoplasmic region of stargazin (aa 203-323) as immunogen. Anti-peptide stargazin antibody (Stg-C-term 1217, rabbit polyclonal) was generated using a synthetic peptide mimicking the last 10 aa residues of stargazin (CNTANRRTTPV). The underlined cysteine residue was added for coupling to keyhole limpet hemocyanin or the SulfoLink column (Pierce). Affinity purification of antibodies was performed using the SulfoLink column coupled with the peptides. To generate phosphorylation site-specific stargazin antibodies (Stg-pT321 1218 rabbit and 1229 guinea pig polyclonal), the same synthetic stargazin C-terminal peptide with phosphorylated T321 was employed as immunogen. For affinity purification of the Stg-pT321 antibody, antisera were passed through the SulfoLink column coupled with the phophorylated peptide followed by a column coupled with unphosphorylated C-terminal peptide as described previously (20). Rabbit polyclonal EGFP (1167) and PSD-95 (SM55) antibodies were generated using H6-EGFP (aa 1-240) and H6-PSD-95 (aa 77-299) (21) as immunogens. The PSD-95 (HM319) antibody is described in the literature (21).

RESULTS
Stargazin Is Phosphorylated at T321-To determine whether the T321 residue of stargazin is the phosphorylation site, we generated phosphorylation site-specific (termed Stg-pT321) antibodies (1218 rabbit and 1229 guinea pig) using the last 10 residues of stargazin with phosphorylated T321 as immunogen. Stg-pT321 antibodies were tested against stargazin expressed in heterologous cells (Fig. 1). Immunoblotting of COS cell lysates transfected with EGFP-tagged stargazin (EGFP-stargazin) with Stg-pT321 antibodies revealed a single protein band of about 66 kDa (38 kDa stargazin ϩ 28 kDa EGFP) (Fig. 1A). In contrast, mutant stargazin (T321A) lacking the hydroxyl group for phosphorylation was not detected by Stg-pT321 antibodies (Fig. 1A). Preincubation of Stg-pT321 antibodies with excess free phosphorylated (but not unphosphorylated) peptide prevented the recognition of EGFP-stargazin (wild-type) (Fig. 1B). Moreover, pretreatment of the membrane with -phosphatase abolished recognition of the EGFPstargazin (wild-type) by Stg-pT321 antibodies (Fig. 1B). These results indicate that stargazin is phosphorylated at T321 in heterologous cells.
T321 of Stargazin Is Phosphorylated in the Brain-To examine whether T321 of stargazin is phosphorylated in vivo, we performed immunoblotting analyses on protein samples from rat brain with various stargazin antibodies (Fig. 2). Both Stg-pT321 antibodies (1218 and 1229) recognized a major band of about 38 kDa corresponding to the size of reported stargazin (4) in crude synaptosomal (P2) and small membrane (S2) fractions. Similarly sized stargazin bands were detected by additional stargazin antibodies raised against the C-terminal peptide (the last 10 residues with unphosphorylated T321, Stg-C-term) and the entire C-terminal cytoplasmic region (aa 203-323; Stg-Cyto). Our data indicate that stargazin is phosphorylated at the T321 residue in vivo. It is possible that Stg-pT321 and Stg-C-term antibodies may recognize closely related proteins such as the ␥-3 and ␥-4 subunits of voltage-dependent calcium channels (7, 8, 10), because they have similar C-terminal sequences (the last 7 residues are identical) and calculated mo- The Stargazin C Terminus Is Phosphorylated by PKA-To determine whether the stargazin C terminus is a substrate of PKA, we examined phosphorylation levels of this region of stargazin in COS cells doubly transfected with stargazin and the PKA catalytic subunit (wild-type and an inactive mutant) (Fig. 3). When COS cells singly transfected with stargazin were immunoblotted with Stg-pT321 antibodies, a relatively small but significant amount of phosphorylated stargazin was detected, suggesting that stargazin is basally phosphorylated in COS cells (Fig. 3). Stargazin phosphorylation was markedly increased upon coexpression with the PKA catalytic subunit but not with an inactive catalytic subunit (25) (Fig. 3). These results indicate that PKA phosphorylates the C terminus of stargazin at T321.
Mutations Mimicking Phosphorylation of the Stargazin C Terminus at T321 Disrupt Interactions with PSD-95 and SAP97 in the Yeast Two-hybrid Assay-To investigate whether phosphorylation of T321 of stargazin affects its binding to PSD-95, we generated two mutations mimicking the PKAphosphorylated state of the protein (RRTDPV and RRTEPV; mutations are underlined). When these stargazin mutations were tested for their interactions with the PDZ domains of PSD-95 and SAP97 in the yeast two-hybrid assay, neither mutant bound PSD-95 or SAP97 (Fig. 4). In control experiments, the wild-type stargazin C terminus specifically interacted with the PDZ domains of PSD-95 and SAP97 but not with unrelated PDZ domains of GRIP1, an AMPAR-interacting multi-PDZ protein (26) (Fig. 4). Additional mutant stargazin (RRTTPA) lost the ability to bind both PSD-95 and SAP97, indicating that the C terminus of stargazin specifically interacts with the PDZ domains of the PSD-95 family members. These findings demonstrate that the stargazin mutations mimicking T321 phosphorylation disrupt interactions with members of the PSD-95 family.
Mutations Mimicking T321 Phosphorylation Disrupt Biochemical Association and Coclustering between Stargazin and PSD-95 in Heterologous Cells-As an independent test of the effects of stargazin mutations on the stargazin-PSD-95 interaction, we performed coimmunoprecipitation experiments on COS cell lysates doubly transfected with EGFP-stargazin (wild-type and mutants) and PSD-95 (Fig. 5). PSD-95 antibodies pulled down, in addition to their cognate antigen PSD-95, wild-type stargazin (RRTTPV) but none of the mutant stargazins tested (RRTDPV, RRTEPV, and RRTTPA) (Fig. 5). In control immunoprecipitation, singly transfected stargazin (wild-type) was not brought down by PSD-95 antibodies (Fig.  5). These results indicate that stargazin mutations mimicking T321 phosphorylation disrupt the biochemical association between stargazin and PSD-95 in heterologous cells.
Coexpression of stargazin and PSD-95 in heterologous cells results in the formation of clusters where the two proteins colocalize (14). We examined the effects of stargazin mutations on the coclustering between stargazin and PSD-95 (Fig. 6). Wild-type EGFP-stargazin when coexpressed with PSD-95 in COS cells formed clusters in which both proteins colocalized (Fig. 6A), consistent with previous results (14). However, none of the stargazin mutants (RRTDPV, RRTEPV and RRTTPA) formed typical clusters on coexpression, and both proteins were diffusely distributed throughout the cells (Fig. 6, B-D). These data suggest that T321 phosphorylation of stargazin regulates the interaction between stargazin and PSD-95, in agreement with the yeast two-hybrid results.
Phosphorylated Stargazin Selectively Loses Its Interaction with PSD-95-To investigate whether phosphorylation of stargazin reduces its interaction with PSD-95, we performed coimmunoprecipitation experiments with COS cells triply transfected with stargazin (wild-type), PSD-95, and the PKA catalytic subunit (Fig. 7). Immunoprecipitation of COS cell lysates with PSD-95 antibodies did not bring down any detectable amount of phosphorylated stargazin, as visualized by Stg-pT321 antibodies (Fig. 7, top panel). In contrast, PSD-95 antibodies brought down a significant amount of total stargazin, as revealed by EGFP antibodies (Fig. 7, middle panel), suggesting that unphosphorylated stargazin retains the ability to coimmunoprecipitate with the PSD-95. These findings further confirm that phosphorylation of stargazin at the C-terminal T321 residue inhibits its interaction with PSD-95.
Phosphorylated Stargazin Is Minimally Enriched in Postsynaptic Density (PSD) Fractions-Stargazin is enriched in detergent-insoluble PSD fractions along with AMPAR subunits and PSD-95 (14). To test whether phosphorylated stargazin shows an altered enrichment in the PSD, we performed immunoblot analysis on PSD fractions of rat brain (Fig. 8). Stg-pT321 antibodies revealed that phosphorylated stargazin is minimally enriched in PSD fractions. In particular, phosphorylated stargazin was not detectable in the PSD III fraction, a core of the PSD extracted with Triton X-100 and sarcosyl detergents. In contrast, Stg-Cyto and Stg-C-term antibodies showed a significant enrichment of total stargazin in all PSD fractions. These results indicate that phosphorylated stargazin is less tightly associated with the PSD. DISCUSSION Our results demonstrate that phosphorylation of stargazin at T321 by PKA inhibits its interaction with PSD-95. The C-terminal sequence of stargazin (NRRTTPV) is identical or similar to those in other closely related ␥-subunits of voltagedependent calcium channels (NRRTTPV in ␥-3 and ␥-4 and NRKTTPV in ␥-8) that are expressed in the brain (7,8,10), suggesting similar interactions with PSD-95 and phosphorylation by PKA. Conversely, stargazin interacts with all known members of the PSD-95 family (Fig. 4) (14), which display diverse distribution patterns in the brain (16). These results suggest that the PKA-dependent phosphorylation of stargazin at T321 regulates interactions between various members of the stargazin/␥ and PSD-95 families. To our knowledge, this is the second report thus far showing that phosphorylation of the C-terminal PDZ-binding motif regulates PDZ interactions.
We propose three possible roles of stargazin phosphorylation. First, stargazin phosphorylation may regulate the synaptic targeting of AMPARs by modulating interactions between the stargazin C terminus and the PDZ domains of PSD-95 and related PDZ proteins. Consistently, stargazin⌬C rescues extrasynaptic but not synaptic AMPAR currents in cerebellar granule cells of stargazer mice (14). In cultured hippocampal neurons, stargazin⌬C is diffusely distributed and down-regulates synaptic AMPAR currents (14).
Second, phosphorylation may regulate the stability of stargazin on the synaptic surface membrane. Phosphorylated stargazin may lose its ability to interact with synaptic PDZ anchors and consequently diffuse away laterally from the synaptic surface or become internalized. Consistently, PSD-95 markedly suppresses internalization of its binding partners including Kv1.4 potassium channel (27) and ␤1-adrenergic receptor (28). Stargazin phosphorylation may not be the only factor that determines its synaptic stability, if stargazin delivered to synaptic sites still remains associated with AMPARs, which are known to interact with their own anchors such as GRIP/ABP (26,29,30). However, a mutant GluR2 subunit of AMPARs lacking GRIP binding loses its stability at the synaptic surface  , Stg-pT321), whereas a significant amount of total stargazin was pulled down with PSD-95 (middle panel, EGFP) suggesting that phosphorylated stargazin selectively loses its ability to interact with PSD-95. IB, immunoblot; Input, 3%. (31), suggesting that stargazin may dissociate from GluR2 after initial synaptic targeting.
Third, stargazin phosphorylation may regulate translocation of AMPARs to the cell surface. Stargazin phosphorylation may release the stargazin-AMPAR complex from cytosolic PDZ anchors and facilitate its delivery to the surface membrane. Interestingly, phosphorylation of the Ser-845 residue of GluR1, a subunit of AMPARs that binds to stargazin (14), correlates well with its surface association in cultured neurons (32). In addition, PKA activation reduces N-methyl-D-aspartic acid-induced GluR1 endocytosis, and PKA inhibition reduces GluR1 reinsertion following NMDA treatment (32). It is known that Ser-845 of GluR1 is phosphorylated by PKA (20,33) and that this phosphorylation increases the peak open probability of AM-PARs (34). However, little is known about its involvement in GluR1 trafficking. Stargazin phosphorylation by PKA may be a molecular mechanism that explains the effects of PKA on AM-PAR surface expression.