Group I Metabotropic Glutamate Receptors Bind to Protein Phosphatase 1C

The modulation of neurotransmitter receptors by kinases and phosphatases represents a key mechanism in controlling synaptic signal transduction. However, molecular determinants involved in the specific targeting and interactions of these enzymes are largely unknown. Here, we identified both catalytic γ-isoforms of protein phosphatase 1C (PP1γ1 and PP1γ2) as binding partners of the group I metabotropic glutamate receptors type 1a, 5a, and 5b in yeast cells and pull-down assays, using recombinant and native protein preparations. The tissue distribution of interacting proteins was compared, and protein phosphatase 1C was detected in dendrites of retinal bipolar cells expressing the respective interacting glutamate receptors. We mapped interacting domains within binding partners and identified five amino acids in the intracellular C termini of the metabotropic glutamate receptors type 1a, 5a, 5b, and 7b being both necessary and sufficient to bind protein phosphatase 1C. Furthermore, we show a dose-dependent competition of these C termini in binding the enzyme. Based on our data, we investigated the structure of the identified amino acids bound to protein phosphatase 1C by homology-based molecular modeling. In summary, these results provide a molecular description of the interaction between protein phosphatase 1C and metabotropic glutamate receptors and thereby increase our understanding of glutamatergic signal transduction.

The correct targeting and localization of proteins to synaptic specializations represents an important biological mechanism to regulate neuronal excitability. Increasing evidence underlines the importance of macromolecular signaling complexes, where functionally related proteins such as ion channels, neurotransmitters receptors, kinases, and phosphatases are arranged in close vicinity and are physically anchored to the synaptic cytoskeleton. Therefore, identification and characterization of interactions between synaptically localized proteins adds substantially to our understanding of molecular mechanisms of neurotransmission.
Receptors for glutamate are divided in ion channel-associated (ionotropic) receptors of the AMPA-, Kainate-, and NMDAtype and in G-protein coupled (metabotropic) receptors. Al-though ionotropic glutamate receptors mediate fast synaptic transmission, metabotropic glutamate receptors (mGluRs) 1 modulate neuronal excitability and development, synaptic plasticity, transmitter release, and memory function using a variety of intracellular second messenger systems (1). The eight known members of this protein family are sub-divided into three groups, based on sequence homology, associated second messenger systems, and pharmacological properties (2). mGluR1 and mGluR5 (group I) stimulate phospholipase C, are selectively activated by quisqualate, and are generally expressed perisynaptically at postsynaptic sites (3)(4)(5)(6)(7). mGluR2 and mGluR3 (group II) are negatively coupled to adenylyl cyclase and do not show a specific preference for pre-or postsynaptic neurons (8 -10). mGluR4, mGluR7, and mGluR8 (group III) are also negatively coupled to adenylyl cyclase and are primarily found at the active zone where they are suggested to function as glutamate auto-receptors (10,(11)(12)(13)(14)(15)(16). Of the group III mGluRs the expression of mGluR6 is restricted to the retina, where it is localized postsynaptically at retinal photoreceptor to bipolar cell synapses, transmitting the "light on" signal in visual signal transduction (17,18).
Here, we investigated the binding between the two PP1C isoforms PP1␥1 and PP1␥2 and the group I metabotropic glutamate receptors mGluR1 and mGluR5. We compared the expression profile between PP1␥1/2 and interacting mGluRs and showed in the retina that PP1␥1 is present in the same subcellular compartments as mGluR1 and mGluR5. Furthermore, we mapped PP1␥1-binding amino acids in the interacting glu-tamate receptors mGluR1a, mGluR5a/b, and mGluR7b and investigated their three-dimensional structure using homology-based molecular modeling.

EXPERIMENTAL PROCEDURES
All animal experiments were performed in compliance with the guidelines for the welfare of experimental animals issued by the Federal Government of Germany, the National Institutes of Health, and the Max Planck Society.
Molecular Biology-Domains of mGluR C termini were PCR-amplified from rat brain cDNA and sub-cloned in the yeast two-hybrid bait vector pBTM116 or fused to the coding sequence of glutathione Stransferase (GST) in pET41 (Novagen, Madison, WI). The complete coding regions of rat PP1␥1 and PP1␥2 were also generated by PCR, ligated in the yeast two-hybrid prey vector pVP16, or fused to the T7-epitope of pET21 (Novagen). Mutation and deletion constructs of mGluR C termini and PP1␥1/2 were generated by PCR cloning techniques. All constructs were sequenced to check for PCR errors.
Binding Assays-For GST pull-down assays pET21 and pET41 constructs were transformed in Escherichia coli BL21(DE3)pLysS, and protein expression was induced by adding 1 mM isopropyl-beta-D-thiogalactoside (Sigma-Aldrich). Fusion proteins were purified under native conditions, immobilized to glutathione-Sepharose beads, and incubated with the cytosolic fractions of E. coli expressing PP1␥1 as described (31,32). C-terminal domains of mGluR1a, -5a, and -5b were also synthesized in vitro using the T7 promoter of pET41 according to the RiboMAX RNA Production Kit and the Flexi Rabbit Reticulocyte Lysate System (Promega) in the presence of [ 35 S]methionine (Amersham Biosciences). To obtain comparable conditions in competition experiments, E. coli protein extracts of similar protein concentrations, as measured at 280 nm, were used. For these samples the total volume was adjusted to 400 l (defined as 100%) by adding protein extract of non-transformed E. coli BL21(DE3)pLysS. Bound proteins were separated on SDS-PAGE and visualized with Coomassie Brilliant Blue R-250 (Serva, Heidelberg, Germany) or detected by Western blotting using a monoclonal anti-T7 antibody and the enhanced chemiluminescence system (ECL; Amersham Biosciences). Unless otherwise stated all reagents were purchased from Novagen.
For binding assays with native proteins, rat brains were homogenized in 0.32 M Saccharose, 20 mM Tris-HCl, pH 7.4, containing DNaseI and protease inhibitors (Roche Diagnostics) and centrifuged at 700 ϫ g for 5 min as described (32). The supernatant was collected and centrifuged at 30,000 ϫ g for 30 min, and 5 mg of the resulting supernatant were incubated with glutathione-Sepharose beads, coated with GST or GST fusion proteins, for 5 h under slow agitation. Interaction partners were analyzed on Western blots using specific immunesera for PP1␥1 (1:1000; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and PKC-(1:10000; Sigma-Aldrich).
Immunocytochemistry-Adult mice were deeply anesthetized with halothane and decapitated. Retinal cryostat sections with a thickness of 12 m were prepared for immunocytochemical labeling by the indirect fluorescence method as described (14). The retinal sections were incubated with primary antibodies directed against PKC-␣ (1:100; Dunn Labortechnik, Asbach, Germany) and PP1␥1 (1:1000; kindly provided by Paul Greengard, Rockefeller University, New York; see Ref. 33) overnight at room temperature. Binding of primary antibodies was detected by goat anti-mouse or goat anti-rabbit IgG secondary antibodies coupled to Alexa TM 594 (red fluorescence) or Alexa TM 488 (green fluorescence) (1:500; Molecular Probes, Eugene, OR). In control experiments, either the primary or secondary antibody was omitted, resulting in a complete loss of specific immunoreactivity. Sections were examined with a Zeiss confocal laser scanning microscope equipped with argon and HeNe lasers (LSM5 Pascal; Zeiss, Oberkochen, Germany).
Yeast Two-hybrid Techniques-Protein interactions were tested in the yeast strain L40 (Invitrogen) as described (31). In brief, yeast strains expressing bait constructs were transformed with prey constructs and vice versa, and protein interactions were monitored by the activation of His3 and LacZ reporter genes on selection plates. Binding affinities were calculated according to the Yeast Protocols Handbook (Clontech, Palo Alto, CA) using o-nitrophenyl-␤-D-galactopyranoside (Sigma-Aldrich) as substrate.
Comparative Modeling-The crystal structure of PP1␥1 in complex with a RRVSFA peptide (34) was used as template to model PP1␥1 binding sequences (KSVSWS, KSVTWA, KSVTWY) identified in this study. Models were generated by substituting side chains of the RRVSFA ligand using the Sybyl 6.5 program package (Tripos, Inc.). Noncovalent interactions were improved by 200 steps of conjugate gradient energy minimization using the Powell algorithm (35). The stereochemistry and nonbonded contacts were evaluated using the programs PROCHECK (36) and WHAT_CHECK (37). Protein-ligand contacts were analyzed with LIGPLOT (38) using default settings.

PP1␥1
Binds to mGluR1a and mGluR5a/b-Preliminary experiments of our laboratory indicated binding of PP1␥1 to mGluR5 in yeast cells. Therefore, interactions between PP1␥1 and the group I metabotropic glutamate receptors mGluR1a, mGluR5a, and mGluR5b were now tested in GST pull-down assays using immobilized C-terminal domains of mGluR1a, mGluR5a, and mGluR5b C termini fused to GST. Loaded glutathione-Sepharose beads were incubated with E. coli protein extracts or soluble native proteins from rat brain, and bound PP1␥1 was identified by Western blots using specific antibodies. Both recombinant and native PP1␥1 interacted with mGluR1a, mGluR5a, and mGluR5b (Fig. 1A). The previously described interaction of PP1␥1 with mGluR7b, and absence of interaction with mGluR7a (31), served as positive and negative controls, respectively.
To ensure that GST fusion proteins did not unspecifically bind to proteins of the extracts, we analyzed their interaction with the -isoform of protein kinase C (PKC-). This enzyme was present in the soluble protein fraction of rat brain (32), and a specific immune serum was available in the laboratory. Interaction between PKC-and metabotropic glutamate receptors has not been reported, and consistently we did not detect its binding to mGluR C termini (Fig. 1A).
Next, we tested whether the observed interactions between PP1␥1 and mGluR1a and mGluR5a/b C-terminal domains also occurred with the native forms of the metabotropic glutamate receptors. As a source for mGluR1a and the two mGluR5 isoforms we used microsomal membrane proteins purified from rat brain that also contained high concentrations of PP1␥1 (Fig.  1B, left panel). Immunoprecipitation experiments with specific antibodies for mGluR1a and mGluR5a/b demonstrated binding of PP1␥1 to both receptor types (Fig. 1B, right panel). Antibodies against PP1␥1 and PKC-were used as positive and negative controls, respectively.
Tissue Distribution of mGluR1a, mGluR5a/b, and PP1␥1/2-Next, we compared the expression profiles of PP1␥1, its isoform PP1␥2, and of mGluR1a, mGluR5a, and mGluR5b performing semi-quantitative reverse transcriptase-PCR. To reduce variability between samples, the same primer pair was used to amplify both splice variants of mGluR5. Furthermore, DNA products representing mGluR5a and mGluR5b, as well as PP1␥1 and PP1␥2, were visualized using an oligonucleotide hybridizing to both isoforms. mGluR1a and mGluR5 isoforms were present in all regions of the central nervous system (CNS) examined, although in different amounts (Fig. 2). The isoform distribution between mGluR5a and mGluR5b varied between CNS regions. mGluR5a was predominantly expressed in olfactory bulb and cerebellum, whereas mGluR5b was higher concentrated in cortex, hippocampus, and thalamus. PP1␥1 was present in all CNS regions examined, whereas transcripts for PP1␥2 were more diversely expressed, being most abundant in cerebellum (Fig. 2). Outside the CNS, PP1␥1, and PP1␥2 seemed to be ubiquitously expressed, whereas only mGluR1a was detected in lung.
PP1␥1 Is Present at Retinal Synapses Expressing mGluR1a and mGluR5a/b-Co-expression of binding partners represents a pre-requisite for protein interactions. In the retina, the localization of mGluR1a and mGluR5a/b was analyzed in detail (39). Therefore, we tested whether PP1␥1 is present in the same sub-cellular compartments as described for mGluR1a and mGluR5a/b. Staining of vertical sections of adult mouse retinae with an antiserum against PP1␥1 showed expression of PP1␥1 in the two synaptic layers of the retina, the outer (OPL) and inner (IPL) plexiform layer (Fig. 3A). No staining for PP1␥1 was found on somata of interneurons in the inner nuclear layer (INL) and of ganglion cells in the ganglion cell layer (GCL). In the IPL, PP1␥1 immunoreactivity was homogeneously distributed in small immunofluorescent puncta, suggestive of a synaptic localization of the enzyme. In the OPL, small immunofluorescent puncta (arrows in Fig. 3A) were visible above larger aggregates of PP1␥1 staining (arrowheads). In this layer, the small synaptic terminals of rod photoreceptors are located above the large cone photoreceptor synaptic terminals. Thus, the small puncta indicated PP1␥1 expression at rod synapses, whereas the larger immunoreactive aggregates suggested expression of PP1␥1 in neurons postsynaptic to cone photoreceptors.
Group I mGluRs are expressed by the postsynaptic neurons of photoreceptors, especially in the dendrites of rod bipolar cells postsynaptic to the rod photoreceptor synaptic terminals (39). Therefore, sections were double-labeled with antibodies against PP1␥1 (green) and against PKC-␣ (red), a marker for rod bipolar cells (Fig. 3B). The rod bipolar cell somata are located in the INL, their dendrites extend into the OPL toward the synaptic terminals of the rod photoreceptors, and their axons terminate in the innermost part of the IPL close to the GCL. Comparison of the staining for PP1␥1 in the OPL with the dendritic structures of the labeled rod bipolar cells by confocal laser-scanning microscopy unambiguously demonstrated that PP1␥1 was expressed in the dendrites of rod bipolar cells (Fig. 3, C and D). This is best seen in Fig. 3E, showing higher power views of the merge of the two stainings for selected areas (boxes) in the OPL.
Mapping of Interacting Domains within mGluR C Termini and PP1␥1/2-Based on our findings, we proposed the existence of PP1␥1/2 binding motifs located in the C-terminal domains of mGluR1a, mGluR5a, and mGluR5b. To map interact-ing domains within the binding partners, we analyzed the interaction between PP1␥1 and mGluR C termini in yeast cells by the ability of transformants to grow on selective media upon activation of His3 and LacZ reporter genes. The complete C termini of mGluR1a, mGluR5a, and mGluR5b showed transactivation of the reporter genes, which had been also noticed by other groups (21,25). Therefore, we used C-terminally truncated constructs for our studies (see stars in Fig. 4A) that were able to bind recombinant and native PP1␥1 in GST pull-down assays (data not shown). In yeast cells PP1␥1 interacted with mGluR1a*, mGluR5a*, and mGluR5b*, whereas no binding was observed for mGluR1b and mGluR1c (Fig. 4A). The PP1␥1 binding characteristics of mGluR7a and mGluR7b have been reported before (31) and were used as controls for the assay. The relative intensities of the interactions were estimated using a quantitative ␤-galactosidase assay and are presented as arbitrary ␤-galactosidase units (horizontal columns in Fig. 4A).
To identify PP1␥1 binding regions within interacting mGluR C termini, their identical and splice-specific domains were tested individually. The proximal region of the mGluR1a C terminus, identical to the corresponding regions in mGluR1b and mGluR1c, showed no interaction with PP1␥1, whereas the distal, isoform-specific sequence of mGluR1a* (mGluR1a*sp) was sufficient for the binding (Fig. 4B). In contrast, the isoform-specific cassette inserted into the C terminus of mGluR5b did not interact with PP1␥1, whereas a region identical between mGluR5a and mGluR5b (mGluR5*sp) contained enough information for the binding. Again, reported interactions of mGluR7a and mGluR7b C-terminal domains were used as positive and negative controls. From these results we concluded that the PP1␥1 binding motif is located in the splice-specific distal region of mGluR1a and mGluR7b, whereas the isoformspecific sequence of mGluR5b does not seem to be involved in the binding.
We also mapped domains in PP1␥1 and PP1␥2 mediating the binding to mGluR1a and mGluR5a/b C termini, following a classification of functional domains described for PP1␣ (40). PP1␥1/2 was divided in the N terminus, a conserved core region, and the C-terminal part that was further sub-divided into a region needed for functional expression of PP1␣ (residues 270 -297; see Ref. 41), the part proximal of the splice site, and the two distal domains different between PP1␥1 and PP1␥2 (Fig. 4C).
N-and C-terminal deletion constructs of PP1␥1/2 showed identical binding characteristics for all three tested mGluR C termini (Fig. 4C). Deletion of 40 N-terminal amino acids abolished interactions, whereas removal of the isoform specific C termini of PP1␥1/2 had no effect. C-terminal deletions up- Identification of PP1␥1 Binding Motifs in Interacting mGluR C Termini-To map PP1␥1 binding motifs in mGluR C termini, we compared the amino acid sequences of the minimal interacting C-terminal domains of mGluR1a, mGluR5a/b, and mGluR7b, as determined in Fig. 4B. This alignment revealed one region of high similarity, ranging from position 888 to 895 of mGluR1a (Fig. 5A). It seemed unlikely that Ser-888 takes part in the PP1␥1 binding motifs, because its deletion in mGluR7b did not alter the binding to PP1␥1 (31). Thus, we proposed PP1␥1 binding motifs consisting of the amino acids KSVSW for mGluR1a and of the sequence KSVTW for both mGluR5 isoforms and for mGluR7b (bold characters in Fig. 5A).
We tested this hypothesis by examining a series of deletion and mutant constructs for their ability to bind PP1␥1 in yeast cells and GST pull-down assays. First, we deleted the proposed PP1␥1 binding motif, as well as left and right flanking sequences. Consistent with our hypothesis, residues N-or Cterminal to the binding motif had no effect on the interactions (Fig. 5, B-D). In contrast, deleting the five amino acids of the proposed binding motifs completely abolished the interaction with PP1␥1. The estimated binding affinities of the mutated constructs were similar to those observed for the wild type sequences (horizontal columns in Fig. 5, B-D), except for mGluR7b-sp⌬YTI, probably because of two prolines adjacent to the PP1␥1 binding motif. We verified the observed binding characteristics of the deletion mutants in GST pull-down experiments (Fig. 5E) and concluded that the amino acid sequences KSVSW and KSVTW contributed substantially to the PP1␥1 binding motifs in mGluR C termini.
To assess the contribution of each individual residue of these motifs for the protein interactions, each amino acid of the PP1␥1 binding motifs was replaced independently by alanine. Except for serine at position ϩ2 of the motifs, all mutants failed to interact with PP1␥1, indicating that amino acid side chains, rather than the protein backbone forms the PP1␥1 binding surface (Fig. 6). Although the hydroxyl group of serine at position ϩ2 seemed not to participate in the binding surface contacting PP1␥1, an amino acid at this position might be needed to space the neighboring interacting residues correctly. To test this hypothesis, we deleted this serine in all three mGluR C termini, which prevented their binding to PP1␥1 (Fig. 6). Therefore, an amino acid at this position, although not directly needed to interact with PP1␥1, might be required for the correct three-dimensional structure of the PP1␥1 binding motif.
Identified Protein Phosphatase Binding Motifs Are Necessary and Sufficient to Interact with PP1␥1-To analyze whether the identified five amino acids in the mGluR1a, mGluR5a/b, and mGluR7b C termini are indeed sufficient to bind PP1␥1, we introduced their binding motifs into the C terminus of mGluR7a that did not interact with PP1␥1 (see Figs. 1 and 4). The sequences KSVSW and KSVTW were inserted in mGluR7a at a position corresponding to the location of the mGluR7b motif (Fig. 7A). Although the wild type sequence of mGluR7a-sp did not bind PP1␥1, mGluR7a-sp carrying the motifs KSVSW and KSVTW interacted in yeast cells (Fig. 7A) and in GST pull-down assays (Fig. 7B). Thus, the identified PP1␥1 binding motifs are necessary and sufficient for the protein interactions.
Next, we used the C terminus of mGluR7b, containing the identified KSVTW motif, to compete the binding of PP1␥1 with group I mGluRs. First, PP1␥1 was bound to the C termini of mGluR1a, mGluR5a, and mGluR5b immobilized on glutathione-Sepharose beads (Fig. 8, A-C, rows 1). These beads were then incubated with increasing concentrations of the Histagged C terminus of mGluR7b containing the KSVTW motif, or with His-mGluR7a, not interacting with PP1␥1 as control (Fig. 8, A-C, rows 2). mGluR7b competed with immobilized mGluR1a*, mGluR5a*, and mGluR5b* for the binding to PP1␥1 and displaced the enzyme from the coated glutathione-Sepharose beads into the supernatant. The remaining amount of PP1␥1 bound to mGluR1a*, mGluR5a*, and mGluR5b* was analyzed by separating the coated glutathione-Sepharose beads from the supernatant and subsequent detection of PP1␥1 on Western blots (Fig. 8, A-C, rows 3). Displaced PP1␥1, bound to His-mGluR7b in the supernatant, was analyzed on Western blots after pulling down the proteins using Ni-NTA beads (Fig.  8, A-C, rows 4). We also controlled that PP1␥1 itself did not bind to GST-or Ni-NTA beads, and that the His-tagged mGluR7 C termini did not interact with the glutathione-Sepharose beads (data not shown).
Increasing concentrations of mGluR7b were able to compete with mGluR1a, mGluR5a, and mGluR5b for interaction with PP1␥1 in a dose-dependent manner, whereas mGluR7a had no effect (Fig. 8). Because the amino acid sequence KSVTW of mGluR7b is present in mGluR5a/b and homologous to the sequence KSVSW of mGluR1a (see Fig. 5A), these results strongly indicate an important contribution of these motifs in PP1␥1 binding. Comparative Modeling of the Identified PP1␥1 Binding Motifs-Co-crystallization of PP1␥1 with the peptide RRVSFA, corresponding to amino acids 64 -69 of the PP1C regulatory G-subunit (G M ), demonstrated the formation of a ␤-strand comprising residues 67-69 of G M that is incorporated into a ␤-sheet of PP1C (34). This sequence is similar to the described PP1␥1/2 binding motifs KSVSW and KSVTW located in mGluR C termini. To obtain a three-dimensional representation of the in-teraction between mGluR C termini and PP1␥1, we modeled the identified PP1␥1 binding motifs in a complex with the enzyme, based on the reported structure of the PP1␥1/G M complex (34). Because this study analyzed the structure of six amino acids of G M (RRVSFA) bound to PP1␥1, we included the analogous sixth position of the mGluR sequences (see Fig. 5) in our modeling.
Inspection of the modeled complexes among mGluR1a, mGluR5a/b, mGluR7b, and PP1␥1 revealed that side chain substitutions and subsequent energy minimizations only resulted in very minor structural changes compared with the PP1␥1/G M interaction (Fig. 9, A-E). Analysis of the stereochemistry and nonbonded contacts confirmed that even for the sterically most demanding PP1␥1 binding motif KSVTWY, carrying a tyrosine instead of an alanine at position ϩ6, no steric strain was imposed on the complex by the modeling procedure, and all specific side chain interactions reported for the RRVSFA peptide (34) were preserved (Fig. 9, E and F).
At the first position of the binding motifs all modeled peptides contain a lysine, capable of forming a hydrogen bond to the carboxyl group of the PP1␥1 residue Glu-287 (Fig. 9F), similar to the corresponding arginine in the crystallized PP1␥1/G M complex (34). The amino acid present at the second position showed neither in the reported crystal structure nor in the modeled peptides any specific side chain interactions ( Fig.  9), consistent with our experimental data in which this residue could be replaced by alanine (Fig. 6). The observation that this position tolerated a high sequence variability but no deletion (Fig. 6) could be explained by the extended backbone conformation of the peptide forcing the side chains of adjacent residues to point into opposite directions. Deletion of the second residue in the PP1␥1 binding motifs would disrupt this pattern resulting in the loss of interactions formed by lysine and valine residues in position ϩ1 and ϩ3 (Fig. 9F).
The side chains of the two hydrophobic residues of the motifs, valine and tryptophan at positions ϩ3 and ϩ5, were directed toward the PP1␥1 surface and formed numerous hydrophobic contacts (Fig. 9, E and F). The binding site of valine was mainly formed by the PP1␥1 side chains of Ile-169, Leu-243, Leu-289, and Cys-291, whereas that for tryptophan was formed by side chains of Phe-257, Met-283, Cys-291, and Phe-293. Interestingly, the latter binding pocket that was occupied by a phenylalanine in the crystal structure of the PP1␥1/G M complex was sufficiently large to accommodate also a tryptophan side chain without requiring side chain rearrangement (Fig. 9E). Incorporation of tryptophan also did not require a displacement of the backbone or the hydrogen bonds between residues of the modeled motifs and Leu-289 and Cys-291 in PP1␥1.
The threonine at position ϩ4 exhibited a highly solventexposed side chain and was part of a hydrogen-bonding network in which a water molecule bridges its side chain hydroxyl and the backbone carbonyl of residue ϩ2 to the main-chain carbonyl group of Thr-288 of PP1␥1 (Fig. 9F). The same interaction was reported for the side chain hydroxyl group of a serine at the corresponding position in the PP1␥1/G M complex (Fig. 9F) but was disrupted upon introducing an alanine (Fig.  6). The residue at position ϩ6 is also solvent-exposed but did not form specific side chain interactions, consistent with the sequence variability observed at this position and its possible deletion without affecting binding behavior (Fig. 5). DISCUSSION Metabotropic glutamate receptors are dynamically regulated by associated proteins that are grouped together in macromolecular signaling complexes. The intracellular C termini of mGluRs define specific receptor isoforms and are prominent targets for interacting proteins. Here, we identified the C ter- FIG. 8. Identified binding motifs in mGluR C termini compete for PP1␥1. PP1␥1 was bound to the C termini of mGluR1a (A), mGluR5a (B), and mGluR5b (B) immobilized on glutathione-Sepharose (rows 1). Identical amounts of beads were then incubated with increasing concentrations of His-mGluR7b, containing the KSVTW motif, or with His-mGluR7a for control (rows 2). Competition among mGluR1a*, mGluR5a*, mGluR5b*, and mGluR7b, but not mGluR7a, displaced PP1␥1 from the coated glutathione-Sepharose beads into the supernatant. Remaining amounts of PP1␥1 bound to mGluR1a*, mGluR5a*, and mGluR5b* were analyzed by pulling down the glutathione-Sepharose beads and subsequent detection of PP1␥1 on Western blots, as described in Fig. 1 (rows 3). Displaced PP1␥1 interacted with His-mGluR7b, but not with His-mGluR7a, and was detected on Western blots using Ni-NTA beads (rows 4). The different pull-downs of PP1␥1 are indicated by the names of the GST or His fusion proteins in parentheses. Protein concentrations of GST-and His-tagged fusion proteins are shown on Coomassie-stained SDS-PAGE (arrowheads). mini of the group I mGluRs, mGluR1a, and mGluR5a/b, as binding partners of the PP1C isoforms PP1␥1/2. In an earlier study we showed that mGluR7b, a member of the group III mGluRs, also interacts with PP1␥1/2 (31). Thus, in contrast to the specific binding between type I mGluRs and proteins of the Homer family (19,22), the interaction between mGluRs and PP1␥1/2 is not restricted to a specific receptor group.
We demonstrated co-expression of PP1␥1/2 and interacting mGluRs in several regions of the mammalian CNS but could find nearly no overlapping distribution in organs outside the CNS. The observed expression patterns of type I mGluRs are consistent with published data, showing that mGluR5a is predominantly expressed in the olfactory bulb, whereas in cortex and hippocampus mGluR5b is the main isoform (20). In the retina, we found PP1␥1 in dendrites of rod bipolar cells postsynaptic to rod photoreceptor terminals. Because mGluR1a and mGluR5a/b are expressed in rod bipolar cell dendrites (39), this co-localization of PP1␥1 and group I mGluRs in the same subcellular compartment suggests their interaction in vivo. PP1␥1 and associated mGluRs are also co-localized in the same subcellular compartments of neurons in other brain areas, e.g. in hippocampal interneurons of the stratum oriens and in dendrites of cerebellar purkinje cells (7,(42)(43)(44)(45).
Of the eight known mGluR types, mGluR1a, mGluR5a/b, and mGluR7b were identified as PP1␥1/2 binding partners in this and in a previous study (31). We mapped amino acids within the mGluR C termini mediating the protein interactions and uncovered an identical architecture of the PP1␥1/2 binding domains, indicating that the protein interactions occur via similar molecular mechanisms. This hypothesis is supported by the finding that identical domains of PP1␥1/2 bound to the mGluR C termini. Finally, we modeled the peptide sequences of the identified PP1␥1 binding motifs of mGluR1a (KSVSW) and of mGluR5a/b and mGluR7b (KSVTW) in a complex with the enzyme, based on the reported crystal structure of the PP1␥1/G M complex (34). The obtained three-dimensional representations favors an extended backbone conformation forcing the side chains of adjacent residues alternatively pointing to the solvent and to the surface of PP1␥1, which is consistent with our experimental data.
As reported in a previous study, an alanine scan covering the isoform-specific distal region of mGluR7b revealed no contribution of amino acids outside its PP1␥1 binding motif KSVTW (46), consistent with our observation that the sequences KS-VSW and KSVTW were both necessary and sufficient to interact with PP1␥1. Thus, the identified sequences KSVSW and KSVTW seem to be the complete PP1␥1 binding motifs in mGluR C termini. These motifs share common characteristics with an already discussed PP1C binding motif (R/K)(V/I)XF (X ϭ any amino acid) found in many proteins interacting with the enzyme (34).
Interestingly, the recently identified PP1C nuclear-binding protein p99 contains the same sequence motif as present in mGluR5a/b and mGluR7b C termini (47). The authors provided FIG. 9. Proposed structure of identified PP1␥1 binding motifs. A, solvent accessible surface area and electrostatic potential of PP1␥1 in complex with the RRVSFA peptide of the regulatory G-subunit or with the mGluR C-terminal sequences KSVSWS (B), KSVTWA (C), and KSVTWY (D). Peptides are shown in stick representation, and the protein surface is colored according to the electrostatic potential (red, most negative; blue, most positive). Coordinates in A were taken from Ref. 33. E, overlay of the RRVSFA template (red) with the sterically most demanding KSVTWY sequence (green). The solvent accessible surface area of PP1␥1 is shown in white. F, schematic representation of the contacts between PP1␥1 and the KSVTWY sequence. The peptide is shown in ball-and-stick representation with carbon, oxygen, nitrogen, and sulfur atoms colored in black, red, blue, and yellow, respectively. Bonds of the protein backbone are shown in black, and those of the side chains in blue. Green dotted lines indicate polar interactions between the peptide and PP1␥1, with their distance given in Angstroms. The corresponding residues of PP1␥1 are highlighted by gray rectangles. Hydrophobic interactions are indicated by red dashes for peptide atoms and red circles for the corresponding residues of PP1␥1. evidence for a physical interaction between PP1␥1 and a 16amino acid peptide containing the sequence KSVTW. Exchanging its tryptophan by alanine prevented binding of the peptide to PP1␥1. This is in agreement with our study, where mutation of tryptophan residues in PP1␥1/2 binding motifs disrupted the protein interactions. In contrast, changing the same tryptophan into phenylalanine did not influence the interaction with PP1␥1 (47), indicating that an aromatic system might be needed in the last position of the binding motif. Based on their data, the authors proposed a PP1C binding motif consisting of the sequence RKSVTW (47). To us it seems unlikely that the N-terminal arginine would be needed to bind PP1␥1, because the corresponding amino acid positions in mGluR1a and mGluR5a/b contain a glycine and a glutamine in mGluR7b. Furthermore, deletion of these residues did not influence binding to PP1␥1 significantly.
In summary, our results provide a molecular description of the PP1␥1/2 binding sites on interacting mGluRs and identify the new PP1C binding motifs KSVSW and KSVTW. Because mGluR-mediated signal transduction depends largely on its association with regulatory proteins such as kinases and phosphatases, this study provides a molecular basis for the understanding of metabotropic glutamatergic signaling.