J Biol Chem, Vol. 274, Issue 50, 35845-35854, December 10, 1999

Brain Actin-associated Protein Phosphatase 1 Holoenzymes Containing Spinophilin, Neurabin, and Selected Catalytic Subunit Isoforms^*

Leigh B. MacMillan^§, Martha A. Bass, Nikki Cheng, Eric F. Howard^¶, Masaaki Tamura^¶, Stefan Strack, Brian E. Wadzinski^**, and Roger J. Colbran

From the  Department of Molecular Physiology and Biophysics, ^¶ Department of Biochemistry, ^** Department of Pharmacology, and  Center for Molecular Neuroscience, Vanderbilt University School of Medicine, Nashville, Tennessee 37232-0615

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

We previously characterized PP1bp134 and PP1bp175, two neuronal proteins that bind the protein phosphatase 1 catalytic subunit (PP1). Here we purify from rat brain actin-cytoskeletal extracts PP1_A holoenzymes selectively enriched in PP1₁ over PP1 isoforms and also containing PP1bp134 and PP1bp175. PP1bp134 and PP1bp175 were identified as the synapse-localized F-actin-binding proteins spinophilin (Allen, P. B., Ouimet, C. C., and Greengard, P. (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 9956-9561; Satoh, A., Nakanishi, H., Obaishi, H., Wada, M., Takahashi, K., Satoh, K., Hirao, K., Nishioka, H., Hata, Y., Mizoguchi, A., and Takai, Y. (1998) J. Biol. Chem. 273, 3470-3475) and neurabin (Nakanishi, H., Obaishi, H., Satoh, A., Wada, M., Mandai, K., Satoh, K., Nishioka, H., Matsuura, Y., Mizoguchi, A., and Takai, Y. (1997) J. Cell Biol. 139, 951-961), respectively. Recombinant spinophilin and neurabin interacted with endogenous PP1 and also with each other when co-expressed in HEK293 cells. Spinophilin residues 427-470, or homologous neurabin residues 436-479, were sufficient to bind PP1 in gel overlay assays, and selectively bound PP1₁ from a mixture of brain protein phosphatase catalytic subunits; additional N- and C-terminal sequences were required for potent inhibition of PP1. Immunoprecipitation of spinophilin or neurabin from crude brain extracts selectively coprecipitated PP1₁ over PP1. Moreover, immunoprecipitation of PP1₁ from brain extracts efficiently coprecipitated spinophilin and neurabin, whereas PP1 immunoprecipitation did not. Thus, PP1_A holoenzymes containing spinophilin and/or neurabin target specific neuronal PP1 isoforms, facilitating efficient regulation of synaptic phosphoproteins.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Phosphorylation/dephosphorylation of synaptic proteins is regulated by neurotransmitters and is a major modulator of synaptic function. For example, Ca²⁺/calmodulin-dependent protein kinase II and protein phosphatase 1 are two major dendritic proteins that play key roles in hippocampal long-term potentiation and long-term depression, respectively (1). Both of these forms of synaptic plasticity display remarkable synapse-specificity; only synapses stimulated with an appropriate activity pattern undergo long-term potentiation or long-term depression, whereas other synapses in the same postsynaptic neuron remain unaffected. These observations imply that synaptic activity stimulated signal transduction molecules and events are restricted (or targeted) specifically to synapses. While mechanisms for synaptic targeting of neuronal protein kinases are being elucidated (2, 3), comparatively little is known about mechanisms for synaptic targeting of protein phosphatases.

The monomeric catalytic subunit of protein phosphatase 1 (PP1)¹ likely does not represent a significant fraction of the total enzyme in any tissue; diverse tissue-specific PP1-binding proteins (PP1bps) are thought to confer specific regulatory properties and subcellular localization to each cellular PP1 holoenzyme (reviewed in Ref. 4). For example, in skeletal muscle, G_M-regulatory subunits target PP1 to glycogen particles and sarcoplasmic reticulum. Phosphorylation of G_M by protein kinase A dissociates PP1 and thereby inhibits the dephosphorylation of glycogen-associated substrates. Conversely, phosphorylation of G_M by an insulin-sensitive protein kinase activates the associated PP1 and promotes dephosphorylation of substrate proteins associated with the glycogen particle (4). More than 10 proteins that bind the catalytic subunit of PP1 in non-neuronal tissues have been identified (5). A conserved sequence motif V-X-F/W in PP1bps, often preceded by basic residues, makes important interactions with PP1 (6-8). However, there are undoubtedly additional contacts that play a role in the interaction and in modulating phosphatase activity (see "Discussion").

The four isoforms of PP1 (, , ₁, ₂) are expressed in neuronal tissues and exhibit distinct cellular distributions and subcellular localizations (9-11). For example, PP1₁ is selectively enriched at synapses in cultured rat cortical neurons, whereas PP1 is enriched in the soma (11). Immunohistochemical staining of rat brain sections also indicates that PP1 and PP1₁ are enriched in synaptic layers and dendritic spines (10, 11), whereas PP1 is relatively enriched in somatic layers (11). Although neuronal PP1 inhibitor proteins, such as DARPP-32, have been characterized in detail (reviewed in Ref. 12), proteins responsible for selective targeting of PP1 isoforms have not been identified. We characterized four PP1bps with molecular masses of 75, 134, 175, and 216 kDa in whole rat forebrain extracts which were enriched in isolated postsynaptic densities (13). Interestingly, PP1bp134 and PP1bp175 exhibited isoform selectivity, binding to the , ₁, ₂ isoforms better than the  isoform in gel overlays (13). Recent screening of brain cDNA libraries using yeast two-hybrid approaches with a PP1 probe identified two novel neuronal PP1bps, spinophilin, a dendritic spine-localized protein (14), and PNUTS, a PP1bp found in the nucleus (15). However, there is only limited information available about native spinophilin and PNUTS proteins, and their roles in isoform-selective targeting of PP1 has not been investigated.

Here we report the purification and characterization of a novel neuronal PP1 holoenzyme containing PP1₁ associated with PP1bp134 and PP1bp175, and identify the PP1bps as spinophilin (14, 17)² and neurabin (16), respectively, two homologous actin-binding proteins. We show that spinophilin and neurabin selectively interact with PP1₁ over PP1, suggesting that they are at least in part responsible for the enrichment of PP1₁ at synapses.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Forebrain Extraction-- Forebrains were rapidly dissected following euthanasia, quick-frozen in liquid N₂, and then stored at 80 °C until needed. All subsequent procedures were performed at 4 °C. Forebrains were partially thawed in 10 ml per forebrain of Buffer A (10 mM Tris-HCl, pH 7.5, 1 mM EGTA, 1 mM EDTA, 1 mM DTT, 0.2 mM PMSF, 1 mM benzamidine, 40 mg/liter soybean trypsin inhibitor, 10 mg/liter leupeptin) plus 150 mM KCl and then homogenized using a Polytron followed by 15 passes in a motorized Teflon/glass homogenizer. After centrifugation at 35,000 × g for 25 min, the supernatant (S1) was discarded. The pellet (P1) was re-homogenized in 10 ml/forebrain of Buffer B (1 mM Tris-HCl, pH 7.5, 0.5 mM EGTA, 0.5 mM EGTA, 1 mM DTT, 0.2 mM PMSF, 1 mM benzamidine, 20 mg/liter soybean trypsin inhibitor, 5 mg/liter leupeptin) and then dialyzed for 2 h against Buffer B. Similar low ionic strength conditions have been shown previously to destabilize the F-actin cytoskeleton, solubilizing actin-binding proteins (23). After centrifugation at 100,000 × g for 60 min, the supernatant (S2) was saved and the pellet (P3) was re-homogenized in 10 ml/forebrain of Buffer B containing 1% Triton X-100. After mixing gently for 1 h followed by centrifugation at 100,000 × g for 60 min, the supernatant (S3) was removed and the pellet (P3) was resuspended in Buffer B.

Purification of PP1_A Holoenzymes-- All procedures were performed at 4 °C. Combined S2/S3 extracts from 10-20 rat forebrains (see above) were applied to a Fast-Flow Q-Sepharose ion exchange column (100 ml) equilibrated in Buffer C (10 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 0.1% (v/v) Triton X-100, 0.5 mM benzamidine, 0.1 mM PMSF, 20 mg/liter soybean trypsin inhibitor, and 5 mg/liter leupeptin). The column was sequentially washed with 250 ml of Buffer C and then with 250 ml (each) of Buffer C adjusted to 0.25, 0.40, and 0.5 M NaCl, collecting 20-ml fractions. PP1bps were most abundant in the 0.4 M NaCl eluate, coeluting with PP1 activity and immunoreactivity, and were precipitated from pooled Q-Sepharose fractions by slowly adding ammonium sulfate (313 g/liter). After stirring for 30 min, precipitated proteins were collected by centrifugation. The precipitate was redissolved in ×0.1 the original volume of Buffer B and 2-ml aliquots were size-fractionated on a FPLC Superdex 200 column (100 ml) equilibrated in Buffer C, collecting 2 ml fractions at 1 ml/min. PP1 activity and immunoreactivity eluted from Superdex 200 in two peaks (400-600 kDa, 100 kDa), and PP1bps were mostly associated with the larger PP1 species (13). The 400-600-kDa pools from 2 or 3 Superdex 200 columns were applied directly to a 1-ml FPLC Mono Q (0.5 ml/min) or a 8-ml Source Q (4 ml/min) ion exchange column equilibrated in Buffer C; the columns were eluted with a NaCl gradient, collecting 1- or 8-ml fractions, respectively. PP1bp134, PP1bp175, and PP1 co-eluted with a peak of PP1 activity at 0.3-0.4 M NaCl (13). Peak fractions were pooled and mixed overnight with microcystin-Sepharose (0.5 ml) (generous gift of Dr. C. Macintosh, Dundee) equilibrated in Buffer C. After washing with 0.5 M NaCl in Buffer C (10 ml), bound proteins were eluted with 3 M NaSCN, collecting 0.5-ml fractions. Fractions containing PP1bp134, PP1bp175, and PP1 were pooled, concentrated in Centricon 30 devices, and then dialyzed against 10 mM HEPES, pH 7.5, 30% (v/v) glycerol, 0.1 M NaCl, 1 mM EDTA, 1 mM DTT, 0.1% Triton X-100.

DIG-PP1 Overlay and Immunoblotting-- Samples were analyzed for PP1bps by DIG-PP1 overlay assay, and for the presence of protein phosphatase catalytic subunits by immunoblotting, as described (13). Briefly, samples were resolved by SDS-PAGE, electroblotted to nitrocellulose membranes, and then probed with either 2.5-10 nM digoxigenin-conjugated PP1₁ or with antibodies specific to the indicated proteins (see below). Blots were developed using alkaline phosphatase-conjugated secondary antibodies and colorimetric reagents.

PP1 Activity Assay-- Protein phosphatase activities were measured essentially as described (18). During purification from brain extracts, PP1 activity is defined as glycogen phosphorylase a phosphatase activity measured in the presence of 2.5 nM okadaic acid. Similar definitions were used previously to define PP1 activity in brain extracts (18, 28). Although 2.5 nM okadaic acid is generally considered to block >90% of PP2A-like activities, certain forms of PP2A may have reduced sensitivity to okadaic acid (22). Thus, poorly defined PP2A-like species may contribute to our operationally defined PP1 activity, particularly at earlier stages of the purification. The activity of recombinant PP1₁ (bacterially expressed; generous gift of Dr. E. Y. Lee) was determined in the absence of okadaic acid and in the presence of 0.2 mM MnCl₂.

Amino Acid Sequencing-- About 80 µg of purified PP1_A holoenzyme was solubilized in SDS and fractionated by polyacrylamide gel electrophoresis (mini-gels 0.75-mm thick, 3-cm wide lane; 7.5% acrylamide). The gel was electroblotted to polyvinylidene difluoride membrane (Immobilon P, Millipore) for 3 h at 1 A (4 °C) in 10 mM CAPS, pH 11, containing 10% (v/v) methanol. Two widely separated vertical strips (2 mm each) were excised from the lane and analyzed by DIG-PP1 overlay to identify PP1bps, and the remaining membrane was stained with Coomassie Blue. The stained band corresponding to PP1bp134 (estimated 5-10 µg of protein) was excised, washed thoroughly in water, reduced, and alkylated before being incubated with 0.2 µg endo-Lys-C (15 h at 37 °C). Solubilized proteolytic fragments of PP1bp134 were fractionated by reversed-phase high performance liquid chromatography (C18 column) using a gradient of acetonitrile in 0.05% trifluoroacetic acid buffer. Three discretely resolved peptide peaks (detected by absorbance at 214 nm) were subjected to Edman degradation using a Procise 492 protein sequencer (PE Biosystems).

Isolation of cDNAs Encoding Spinophilin and Neurabin-- Double stranded DNA fragments of spinophilin and neurabin were generated by reverse transcriptase-polymerase chain reaction (Promega Access) using rat brain total RNA. Oligonucleotide primer pairs (spinophilin: 5'-CATTTCAGCACCGCACCGATC and 5'-CCAGCGCCCTTTCTCCTGCTC; neurabin: 5'-GAGGGCTCCCAGCAGAGTAGG and 5'-CACTTCCGGTACTGGCACAGC; 5'-GTCCAAGGCCTGCAAGTTCGG and 5'-GCACACTCCACTCATGGACGG) were designed based on sequences available in the data bases (spinophilin, AF016252; neurabin, U72994). The polymerase chain reaction products served as templates for the synthesis of ³²P-labeled DNA probes (Stratagene Prime-It II), which were used to screen a rat brain cDNA library in ZAPII by hybridization, according to the manufacturer's instructions (Stratagene). Hybridizing cDNA clones were plaque-purified, excised, and transferred to pBluescript II SK(+), and their identity confirmed by partial sequencing. Full-length cDNAs were assembled from partial clones using restriction sites.

Expression of Spinophilin and Neurabin in HEK293 Cells-- Full-length spinophilin and neurabin cDNAs, with or without an N-terminal Myc epitope tag (MEQKLISEEDL) (19), were subcloned into a pCMV4 eukaryotic expression vector containing the cytomegalovirus promoter (generous gift of Drs. M. Wilson and L. E. Limbird, Vanderbilt University). Automated DNA sequencing (Vanderbilt Center for Molecular Neuroscience) confirmed the complete nucleotide sequences of these constructs. HEK293 cells were transfected by calcium phosphate precipitation (13.6 µg of DNA/100-mm dish) essentially as described (20). Cells were scraped into lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.5% Triton X-100, 0.1 mM PMSF, 1 mM benzamidine, 20 mg/liter soybean trypsin inhibitor, and 5 mg/liter leupeptin; 1.25 ml/100-mm dish), probe-sonicated (2 × 20-s bursts), and then centrifuged (100,000 × g, 30 min) to generate soluble cell extracts. Extracts were either used immediately, or stored at 80 °C until required.

Expression of Glutathione S-Transferase (GST) Fusion Proteins-- Regions of cDNAs encoding the indicated residues of spinophilin and neurabin were either isolated by polymerase chain reaction using primers incorporating appropriate restriction sites, or obtained by digestion at naturally occurring restriction sites, and then inserted into pGEX-2T or -4T prokaryotic expression vectors (Amersham Pharmacia Biotech). Bacteria (DH5) harboring the expression vector were induced with 0.5 mM isopropyl-1-thio--D-galactopyranoside for 3-5 h and GST fusion proteins were isolated from soluble extracts using glutathione-agarose essentially according to the directions of the manufacturer. A GST-G_M fusion protein containing residues 1-240 of the glycogen targeting subunit of PP1 (21) was generously provided by R. Terry and Dr. S. Shenolikar (Duke University). Protein concentrations were determined using the Bio-Rad protein assay with bovine serum albumin as standard; in PP1 inhibition assays, concentrations refer to estimates of the full-length fusion protein by Coomassie Blue staining following SDS-PAGE.

Antibodies-- Rabbit antibodies raised against divergent sequences from the C termini of PP1 isoforms were described previously (11, 13). Similar antibodies were raised against the identical peptides in sheep. The rabbit and sheep antibodies exhibited similar isoform-specificity in immunoblotting experiments using either whole antiserum or affinity purified antibodies: antibodies raised to PP1 and PP1₁ sequences were specific for the corresponding isoform, whereas the antibodies raised to a PP1 sequence recognized all PP1 isoforms and are designated pan-PP1 antibody (11) (data not shown). PP2A catalytic subunit antibodies were either from Transduction Laboratories (mouse monoclonal) or were raised against a PP2A_C peptide sequence (PNVTRRTPDYFL) in sheep. Sheep PP2A_C antibodies exhibited similar specificity to previously described rabbit antibodies against the same peptide (22) (data not shown). Antibodies to the Myc epitope tag (purified by ammonium sulfate precipitation from mouse ascites fluid) were a generous gift from Drs. J. Flick and R. Wisdom (Vanderbilt University).

Rabbit antisera were raised to GST fusion proteins containing divergent regions of either spinophilin (residues 286-390) or neurabin (residues 146-453). Rabbits were given an intranodal injection of 0.2 mg of fusion protein in Freund's complete adjuvant, followed after 21 days by a 0.2-mg intranodal boost in Freund's incomplete adjuvant. Animals were first boosted with 0.1 mg of fusion protein in incomplete adjuvant at subcutaneous and intramuscular sites 14 days after the initial injection, and subsequently at 20-day intervals. All the studies in this manuscript utilize whole antiserum from the first or second bleed, performed 10 days after booster injections. Rabbit injections and bleedings were performed by Upstate Biotechnology.

Preparation of Mixed Brain Protein Phosphatase Catalytic Subunits-- Native brain protein phosphatase catalytic subunits were separated from their associated subunits and proteins as follows. Fresh rat forebrain S1 extract (180 ml) (see above) was quick-frozen and stored at 80 °C until used. The extract was slowly thawed and then 313 g/liter ammonium sulfate were slowly added and the extract was stirred at 4 °C for 30 min. Precipitated proteins were collected by centrifugation for 30 min at 28,000 × g, and then resuspended in 20 ml of Buffer X (20 mM Tris-HCl, pH 7.5, 1 mM DTT, 0.1 mM EGTA, 2 mM MgCl₂, 0.5 mM benzamidine, 0.1 mM PMSF, 20 mg/liter soybean trypsin inhibitor, 5 mg/liter leupeptin) containing 10% glycerol using a Polytron homogenizer. Room temperature ethanol (100 ml; 95%) containing 1 mM PMSF was added and the suspension was then immediately centrifuged for 5 min at 28,000 × g. The pellet was extracted with 2 × 20-ml aliquots of Buffer X by homogenization using a Polytron. Ammonium sulfate (430 g/liter) was added to the pooled solubilized proteins, and precipitated proteins were collected by centrifugation (35,000 × g for 20 min). The pellet was redissolved in 10 ml of Buffer X and then dialyzed overnight against Buffer X. After dialysis, the preparation was clarified at 35,000 × g for 20 min and then stored at 80 °C in aliquots. To estimate concentrations of PP1 and PP1₁ in this preparation, aliquots were immunoblotted in parallel with known amounts of recombinant PP1 isoforms using isoform-specific antibodies. In two independent experiments, the concentrations were estimated at 1-1.5 µM and 0.5-0.75 µM for PP1 and PP1₁, respectively (total protein concentration 2.96 mg/ml).

Glutathione-agarose Sedimentation Assays-- Approximately 8 µg of the indicated GST fusion proteins were mixed for 1 h at 4 °C with 50 µl of crude protein phosphatase catalytic subunit mixture (see above) in 14 ml of binding buffer (50 mM Tris-HCl, pH 7.5, 0.2 M NaCl, 0.1% Triton X-100, 0.25 mg/ml bovine serum albumin). The final concentration of GST fusion protein was about 17 nM; PP1 and PP1₁ were present at 2-5 nM each, together with an unknown concentration of PP1. About 20 µl of a 50:50 slurry of glutathione-agarose was added and the incubation was continued overnight at 4 °C. The resin was sedimented, washed with 4 × 5-ml aliquots of binding buffer, and then transferred to a fresh microcentrifuge tube. To concentrate protein phosphatases that did not associate with GST fusion proteins, supernatants from glutathione-agarose sedimentation were incubated for 2 h at 4 °C with 20 µl of a 50:50 slurry of microcystin-agarose (Upstate Biotechnology Inc.), and the resin was then sedimented. Proteins associated with glutathione-agarose and with microcystin-agarose were solubilized by boiling in SDS-PAGE sample buffer and aliquots were analyzed by immunoblotting.

Immunoprecipitations-- Combined S2/S3 rat forebrain extracts or HEK293 cell soluble extracts (see above) were diluted to 1 mg/ml protein in IP Buffer (50 mM Tris-HCl, pH 7.5, 0.15 M NaCl, 0.5% Triton X-100) and 0.5-ml aliquots were precleared using 20 µl of a 1:1 slurry of protein A-Sepharose or GammaBind Plus Sepharose (Amersham Pharmacia Biotech). The supernatant was mixed with 5 µl of the indicated rabbit or sheep antiserum (or preimmune serum) or mouse ascites fluid for 1 h at 4 °C. After addition of 20 µl of protein A-Sepharose (rabbit or mouse antibodies) or GammaBind Plus Sepharose (sheep antibodies), respectively, incubations were continued overnight at 4 °C. Resin was collected by microcentrifugation, and then washed with at least 4 × 1 ml of IP Buffer; during the last wash resin was transferred to a new microcentrifuge tube. Immune complexes were solubilized in SDS-PAGE sample buffer. Aliquots of immune complexes and immune supernatants, as well as an aliquot of the initial extract, were analyzed by immunoblotting.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Purification of PP1_A-- In order to identify previously characterized PP1bps (13), we undertook their purification from rat forebrain extracts. This paper focuses on PP1bp134 and PP1bp175, which were partially solubilized by re-homogenization of an initial particulate fraction (P1) in a low ionic strength buffer (13) to destabilize the actin cytoskeleton (23). Re-extraction of the remaining particulate material (P2) with low ionic strength buffer containing 1% Triton X-100 solubilized additional PP1bp134 and PP1bp175 (S3) (data not shown). Significant amounts of PP1 activity and immunoreactivity also were present in S2 and S3 (13). The combined S2/S3 extract is referred to as a crude actin cytoskeletal extract and was used as starting material for purification. Our goal was to purify complex(es) (or holoenzymes) containing PP1bp134 and PP1bp175 associated with PP1. Therefore, the purification was monitored by a combination of DIG-PP1 overlay assays for PP1bps, immunoblotting for PP1, and by PP1 activity assays ("Experimental Procedures"). Some samples also were monitored by immunoblotting using antibodies to the catalytic (C) and regulatory (A; PR65) subunits of PP2A.

Table I and Fig. 1 summarize data from a single purification representative of 5 others (see "Experimental Procedures"). About 0.1 mg of total protein was reproducibly obtained from the microcystin-agarose column starting with only 10 g rat forebrain. The final preparations contained three major Coomassie Blue-stained proteins with apparent molecular masses of about 37, 134, and 175 kDa. Careful comparison of stained gels with DIG-PP1 overlays and PP1 immunoblots confirmed that the major proteins corresponded to PP1, PP1bp134, and PP1bp175, respectively. In some gels PP1bp134 could be resolved into 2 or 3 distinct bands (see below). Analysis of digital scans of stained gels using NIH Image estimated a molar ratio of PP1 to total PP1bps (PP1bp134 plus PP1bp175), of 1.0:0.95, suggesting that total PP1bps are present in about a stoichiometric ratio with PP1. The molar ratio of PP1bp134 to PP1bp175 was approximately 7:1. Additional minor proteins present in the final sample of some preparations also bound DIG-PP1 in overlay assays (see Fig. 1), suggesting that there was a variable amount of proteolysis of PP1bps during purification (see below). Since this complex was isolated from actin-cytoskeletal extracts, and since the two PP1bps were previously identified as F-actin-binding proteins (see below), we termed it PP1_A, a putative native form of brain PP1, which likely contains multiple oligomeric species (see "Discussion").

View larger version (50K): [in this window] [in a new window]   Fig. 1.   Purification of PP1A, an actin-associated PP1 holoenzyme. A PP1 holoenzyme was purified from rat forebrain (see "Experimental Procedures"). Aliquots of pools from each purification stage were analyzed by SDS-PAGE followed by either Coomassie Blue staining (top), DIG-PP1 overlay (middle), or immunoblotting using a pan-PP1 antibody (bottom). S2+S3, crude actin cytoskeletal extract (15 µg); Q-Seph., 0.4 M NaCl eluate from fast flow Q-Sepharose (15 µg); Amm. Sulf., resuspended ammonium sulfate precipitate (15 µg); Sd200, 400-600-kDa pool from Superdex 200 (15 µg); Mono Q, pooled fractions from Mono Q (5 µg); MC-Seph., sodium thiocyanate eluate from microcystin-Sepharose (5 µg, top, and 1 µg, middle and bottom). This preparation is representative of five purifications.

Although the amounts of protein obtained in different preparations and their appearance on Coomassie Blue-stained polyacrylamide gels were very reproducible, folds purification (30-500-fold from the S2/S3 extract) and specific activities (3-106 nmol/min/mg) of final preparations were quite variable based on activity assays. There are likely to be several factors contributing to this variability and the relatively low values for specific activity and fold purification: 1) S2/S3 extracts contained other forms of PP1 that were separated by chromatography on size exclusion and ion exchange columns (see Figs. 2 and 6 of Ref. 13). For example, chromatography on Superdex 200 enriches PP11 immunoreactivity approximately 10-fold (see below), but enriches the specific activity of PP1 only 2-fold (Table I). 2) Limited trypsinolysis of the final preparation resulted in 2-3-fold activation toward phosphorylase a (not shown), similar to that seen in partially purified samples (see Fig. 8 in Ref. 13) (see below). 3) The activity of the final preparation is stimulated 2-3-fold by 1 mM MnCl₂. 4) Sodium thiocyanate elution of the microcystin-agarose column may denature and irreversibly inactivate some of the PP1. 5) Certain PP2A-like enzymes may contribute to activities measured in extracts and at early purification steps, despite the presence of 2.5 nM okadaic acid in the assays (see "Experimental Procedures"). Later chromatographic steps resolve PP2A from PP1 (see also Ref. 13), such that the A and C subunits of PP2A are not typically detected in the final preparation on silver-stained SDS-PAGE gels, or by immunoblotting (not shown).

Identification of PP1bps in the PP1_A Holoenzyme-- PP1bp134 was subjected to endo-Lys-C digestion and three purified peptides were subjected to amino acid sequencing (Fig. 2). A total of 52 Edman degradation cycles were performed, yielding only 50 residues of PP1bp134 sequence since amino acid assignments were ambiguous for two sequencer cycles. When the sequences were initially obtained, BLAST analysis revealed no known mammalian protein sequences with similarities to all three peptides. However, subsequent data base searches identified two newly added sequences (Fig. 2). The peptides were 100% identical to portions of spinophilin, which was identified as a PP1bp by yeast two-hybrid analysis (14) and was independently purified as an F-actin-binding protein (17).² In addition, the peptide sequences were 84% identical to portions of neurabin, a spinophilin homolog which was purified as an F-actin-binding protein and cloned (16). The apparent molecular weights of spinophilin and neurabin by SDS-PAGE were noted to be considerably larger than sizes predicted from their amino acid sequences (16, 17), but are similar to the apparent molecular weights of PP1bp134 and PP1bp175, respectively. While interaction of PP1 with spinophilin was established previously (14), association of PP1 with neurabin has not been documented. These data show that PP1bp134 is spinophilin and suggest that PP1bp175 may be neurabin.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Amino acid sequences of PP1bp134 identify spinophilin and neurabin. Three endo-Lys-C derived peptide fragments of PP1bp134 (pep 25, pep 20, and pep 24) were subjected to amino acid sequencing (see "Experimental Procedures"), yielding a total of 50 definitive amino acid assignments from 52 cycles, including the inferred lysines at the beginning of each peptide (in parentheses). BLAST analysis revealed that these sequences were identical to spinophilin and had similarity to neurabin (42/50 identity). Domain maps of spinophilin and neurabin are shown together with amino acid identities between the indicated domains (14, 16, 17). Circles (3 in spinophilin, 2 in neurabin) indicate minimal consensus PP1-binding motifs, as identified by screening of a random peptide library (8). The black circle indicates the functionally active PP1-binding motif (identified in later figures), within a light gray box indicating a larger region of high homology.

To confirm the identity of PP1bp134 and PP1bp175, antibodies were raised to GST fusion proteins containing divergent regions of spinophilin and neurabin (see "Experimental Procedures"). Immunoblotting of forebrain extracts with anti-spinophilin sera, but not preimmune serum from the same animal yielded a major band at about 134 kDa, plus several very minor bands (Fig. 3A). Preincubation of anti-spinophilin serum with antigen blocked all of the signals, but preincubation with a GST fusion protein containing the corresponding region of neurabin had no significant effect (Fig. 3A). Furthermore, the 134-kDa immunoreactive band co-purified with the actin-associated PP1 (Fig. 3B), although smaller immunoreactive bands appeared during purification. Thus, anti-spinophilin serum specifically detects brain spinophilin, which is highly enriched in PP1_A. Immunoblotting brain extracts with anti-neurabin serum, but not a normal rabbit serum, yielded a major 175-kDa band, a doublet (or triplet) at about 134 kDa and several additional minor bands (Fig. 3A). Preincubation of anti-neurabin serum with GST-neurabin antigen blocked detection of all of these bands, but preincubation with a corresponding GST-spinophilin fusion protein protected only one band in the 134-kDa doublet/triplet (asterisk in Fig. 3A). Thus, anti-neurabin serum detects brain neurabin and exhibits very minor cross-reactivity with spinophilin. All proteins recognized by the anti-neurabin serum were enriched in the PP1_A preparation (Fig. 3B), although additional smaller immunoreactive bands were generated during purification. The smaller proteins detected by the anti-spinophilin and anti-neurabin sera are likely proteolytic fragments of spinophilin and neurabin (also see above and Fig. 1) although we cannot eliminate the possibility that they represent uncharacterized homologs. In combination, the data confirm that spinophilin and neurabin are components of the PP1_A holoenzyme.

View larger version (34K): [in this window] [in a new window]   Fig. 3.   PP1A contains spinophilin and neurabin and is selectively enriched in PP11. Panel A, characterization of antibodies to spinophilin and neurabin. Rat forebrain homogenate (300 µg) was fractionated on a preparative SDS-polyacrylamide mini-gel (7.5% acrylamide) and transferred to nitrocellulose. Vertical strips of the membrane were probed with spinophilin or neurabin antisera (-Sp286-390 or -Nb146-453, respectively) in the absence or presence of the indicated GST fusion proteins or GST alone (40 µg each) or with preimmune antiserum (PIS). An asterisk marks a band detected with -Nb146-453 that is specifically protected by GST-Sp(151-444), indicating minor cross-reactivity of -Nb146-453 with spinophilin. Panel B, pools from a PP1A purification were immunoblotted using antibodies to neurabin, spinophilin, PP11, or PP1 (top to bottom, respectively). Homog., 15 µg; S2+S3, 15 µg; Amm. Sulf., 15 µg; Sd200, 15 µg; Source Q, 4 µg; MC-Seph., 1 µg. Abbreviations are defined in the legend to Fig. 1, except for "Homog.," whole forebrain homogenate.

Spinophilin and Neurabin Directly Bind PP1 and Also to Each Other-- To establish that spinophilin and neurabin are PP1bps and to investigate potential interactions between these proteins, their cDNAs were expressed in HEK293 cells with or without a Myc epitope tag at their amino termini (see "Experimental Procedures"). Cells were transfected with single expression constructs (or an empty vector) and then cell extracts were subjected to DIG-PP1 overlay analysis. Recombinant spinophilin and neurabin both bound DIG-PP1 and exhibited electrophoretic mobilities identical to brain PP1bp134 and PP1bp175, respectively (Fig. 4A). Anti-spinophilin or anti-neurabin sera also recognized the corresponding HEK293 cell-expressed proteins (data not shown). Both spinophilin and neurabin interacted with endogenous HEK293 cell PP1 because immunoprecipitation of Myc-spinophilin or Myc-neurabin from singly transfected cell extracts resulted in co-precipitation of PP1 (Fig. 4B, left panel). Analysis of extracts from cells co-expressing Myc-spinophilin with neurabin, or Myc-neurabin with spinophilin, indicated that the PP1bps interacted with each other; in both cases Myc immunoprecipitates contained the untagged protein in addition to its Myc-tagged counterpart (Fig. 4B, right panel) and PP1 (not shown). Thus, both spinophilin and neurabin are authentic PP1bps which exhibit very similar electrophoretic mobilities to PP1bp134 and PP1bp175.

View larger version (43K): [in this window] [in a new window]   Fig. 4.   Analysis of spinophilin and neurabin expressed in HEK293 cells. HEK293 cells were transfected with pCMV4 expression vectors containing full-length cDNAs encoding spinophilin and neurabin, without or with the inclusion of sequences encoding an N-terminal Myc epitope tag; "mock" cells were transfected with empty vector. Panel A, cell lysates (15 µg) were analyzed by DIG-PP1 overlay (top) and immunoblotting using Myc (middle) or pan-PP1 (bottom) antibodies. A brain S2/S3 extract (20 µg) was analyzed in parallel. Arrows indicate previously identified PP1bps in the brain extract (13) as well as recombinant spinophilin (Sp) and neurabin (Nb). Panel B, extracts of mock transfected cells or of cells expressing Myc-spinophilin, Myc-neurabin, spinophilin, or neurabin, either alone (left panel) or in combinations indicated (right panel), were immunoprecipitated with Myc antibodies. Immune complexes were analyzed by DIG-PP11 overlay and/or by immunoblotting using Myc, pan-PP1, spinophilin, or neurabin antibodies, as indicated.

Identification of the PP1-binding Domain in Spinophilin and Neurabin-- We sought to determine domains in neurabin and spinophilin responsible for interaction with PP1. Previous studies using multiple approaches have revealed a minimal consensus PP1-binding motif of -V-X-F/W- often preceded by basic residues (6-8). Interestingly, spinophilin contains three sequences that conform to the minimal consensus (residues 187-189, 449-451, and 576-578), two of which are conserved in corresponding regions of neurabin (residues 458-460 and 585-587) (Fig. 2). To determine the relative roles of the different consensus sequences in PP1 binding, a battery of GST fusion proteins containing various residues from either spinophilin or neurabin were screened using the gel overlay assay (Fig. 5). GST-spinophilin proteins that contained the central consensus motif exhibited strong binding of DIG-PP1, whereas binding was not detected with proteins lacking this motif but containing either of the other two consensus PP1-binding motifs. These data indicate that only proteins containing the central consensus PP1-binding motif in spinophilin (residues 427-470) are functional in binding PP1. Similar analyses of GST-neurabin fusion proteins indicated that the PP1-binding domain of neurabin contains the corresponding consensus PP1-binding motif (residues 436-479), whereas the more C-terminal consensus PP1-binding motif is not functional (Fig. 5).

View larger version (44K): [in this window] [in a new window]   Fig. 5.   DIG-PP1 overlay analysis of GST-spinophilin and GST-neurabin. Top panels, GST fusion proteins (approximately 1 and 5 µg in left and right panels, respectively) containing corresponding residues of spinophilin (Sp) or neurabin (Nb) (indicated in diagrams above according to the scheme in Fig. 2) were analyzed by SDS-PAGE followed by Coomassie Blue staining. Dots on the figure indicate intact fusion proteins; additional stained bands represent proteolytic degradation products. Bottom panels, the same GST fusion proteins (approximately 0.1 and 0.5 µg in left and right panels, respectively) were analyzed for binding to PP1 in gel overlays using 2.5 nM DIG-PP11.

Effect of Spinophilin and Neurabin on PP1 Activity-- Trypsinolysis of PP1 holoenzymes often results in rapid degradation of PP1bps, but only a few C-terminal residues are removed from PP1 (for discussion of this technique see Refs. 13 and 37). Since proteolytic degradation of free PP1 has little effect on its activity toward phosphorylase a, the 2-3-fold increase in PP1_A activity following limited trypsinolysis (see above) suggests that PP1 activity is inhibited by interaction with spinophilin and/or neurabin. Therefore, the effect of various concentrations of GST fusion proteins containing different portions of spinophilin or neurabin on PP1 activity was determined.

Fusion proteins containing the functional PP1-binding domain of spinophilin (residues 427-470) or neurabin (residues 436-479) inhibited PP1 activity (Fig. 6), but not the activity of purified PP2A_C (not shown). Interestingly, the smallest GST fusion proteins containing a functional PP1-binding motif (GST-Nb(436-479) and GST-Sp(427-470)) were relatively weak inhibitors (about 20% inhibition at 400 nM). Extension of the fusion proteins toward the N- and/or C terminus significantly increased the potency of inhibition. The most potent (IC₅₀ 23 nM) and complete (>80%) inhibition was observed with GST-Nb(146-616), which includes the central variable domain through the PDZ domain. GST-Sp(394-484) and GST-Nb(146-493) inhibited PP1 with similar IC₅₀ (200-300 nM). GST fusion proteins lacking the functional PP1-binding domain did not significantly inhibit PP1 (Fig. 6). In comparison, a GST-G_M fusion protein only inhibited about 35% of PP1 activity at 2 µM. Proteolytic sensitivity of several GST-spinophilin fusion proteins (e.g. those used in the left panel of Fig. 5) precluded analysis of their inhibitory properties.

View larger version (26K): [in this window] [in a new window]   Fig. 6.   Inhibition of PP11 by GST fusion proteins containing portions of spinophilin or neurabin. Top panel, PP11 was assayed in the presence of various concentrations of GST fusion proteins containing the indicated residues from neurabin, GST-GM(1-240), or GST alone. The data are representative of at least three independent experiments. Bottom panel, table summarizing inhibition studies using GST-spinophilin, GST-neurabin, and GST-GM fusion proteins. Bars indicate domains and residues included in the fusion proteins according to the scheme used in Fig. 2; black circle, functional PP1-binding motif; white circles, additional consensus PP1-binding motif. The maximum inhibition at 2 µM fusion protein (or at 0.4 µM fusion protein where indicated **) is shown; figures in parentheses (negative numbers) indicate activities above the control at this concentration. Observed IC50 values are included where applicable. Mean ± S.E (n = 3 or 4) are reported.

Recombinant Spinophilin and Neurabin Selectively Interact with PP1 Isoforms-- We previously demonstrated that PP1₁, but not PP1, selectively co-purifies with PP1bp134 and PP1bp175 on Superdex 200 columns (13). In the present studies, the final PP1_A preparation was enriched in spinophilin and neurabin and was selectively enriched in PP1₁ over PP1 (Fig. 3B). Therefore, we tested the binding of PP1 isoforms to recombinant spinophilin and neurabin. In overlay assays using DIG-PP1₁ (2.5 nM) as the ligand, bacterially expressed PP1 , ₁, and ₂ isoforms effectively competed for binding to both spinophilin and neurabin (EC₅₀ values ranged from 2 to 25 nM in three experiments). In contrast, bacterially expressed PP1 was a relatively weak competitor (10 ± 6% inhibition at 100 nM competitor, n = 3). Similar data were previously reported using spinophilin and neurabin in brain extracts (see Fig. 5 in Ref. 13).

Glutathione-agarose sedimentation assays were used to investigate whether spinophilin and neurabin exhibit selectivity among native brain PP1 isoforms. After incubation of GST fusion proteins containing the functional PP1-binding domain from spinophilin or neurabin with a mixture of protein phosphatase catalytic subunits (i.e. containing native PP1₁ and PP1, as well as PP1, PP2A_C, and likely other phosphatase catalytic subunits), phosphatase catalytic subunits sedimented with glutathione-agarose were analyzed by immunoblotting. PP1₁, but no detectable PP1 or PP2A_C, was co-sedimented with GST-Nb(436-479) or GST-Sp(427-470) (Fig. 7). GST fusion proteins containing longer domains from neurabin (e.g. residues 146-493 or 146-616) or spinophilin (e.g. residues 394-484) exhibited similar selectivity for co-sedimentation of PP1₁ over PP1 (not shown). In contrast, neither GST alone nor GST-Nb fusion proteins lacking the PP1-binding domain (e.g. residues 146-453) were able to co-sediment any PP1 isoform or PP2A_C (Fig. 7 and not shown). Both PP1 and PP1₁, but not PP2A_C, co-sedimented with a GST fusion protein containing the PP1-binding domain from G_M (Fig. 7), which is not known to exhibit any isoform selectivity, demonstrating that PP1 in these samples was capable of binding to certain PP1bps. It appears that GST-G_M may selectively bind PP1 over PP1₁, but more quantitative studies will be necessary to establish this point since PP1 was about 2-fold more concentrated than PP1₁ in the protein phosphatase catalytic subunit preparation (see "Experimental Procedures"). Note that under the conditions used (low protein concentrations), binding of PP1 to GST fusion proteins was not quantitative. These data demonstrate that recombinant fragments of spinophilin and neurabin selectively bind native PP1₁ over native PP1 in vitro.

View larger version (54K): [in this window] [in a new window]   Fig. 7.   Selective binding of PP11 by GST-neurabin and GST-spinophilin. GST alone, GST-Nb(436-479), GST-Sp(427-470), or GST-GM(1-240) were incubated with a crude protein phosphatase catalytic subunit mixture and resulting complexes were sedimented with glutathione-agarose (pellets). Unbound protein phosphatase catalytic subunits were recovered on microcystin-agarose (supernatants) (see "Experimental Procedures"). Proteins were solubilized in SDS-PAGE sample buffer and analyzed (about 25% of total per lane) by immunoblotting for the indicated protein phosphatase catalytic subunits. A Ponceau-stained nitrocellulose membrane was scanned prior to immunoblotting and is shown below to reveal the amount of sedimented GST fusion proteins; the 66-kDa protein in all four lanes is bovine serum albumin from the binding buffer. Black arrowheads indicate full-length GST-GM, GST-Nb/GST-Sp, and GST alone (top, middle, and bottom, respectively). The open arrowhead indicates major proteolytic fragments of GST-Sp(427-470) and GST-Nb(436-479) which do not bind to DIG-PP1 in gel overlay assays (see Fig. 5).

Native Spinophilin and Neurabin Selectively Associate with PP1₁-- Interactions between native spinophilin, native neurabin, and specific PP1 isoforms in brain extracts was examined by coimmunoprecipitation assays using antisera to spinophilin, neurabin, PP1, or PP1₁ (Fig. 8). PP1₁, but not PP1 or PP2A_C, was enriched in spinophilin immunoprecipitates; neurabin (including putative proteolytic fragments) also was coprecipitated. Similarly, PP1₁ and spinophilin, but not PP1 or PP2A_C, were enriched in neurabin immunoprecipitates. Parallel immunoprecipitations using PP1₁ antibodies coprecipitated both spinophilin and neurabin, whereas immunoprecipitations with PP1 or PP2A_C antibodies failed to coprecipitate significant amounts of spinophilin or neurabin (Fig. 8). In these experiments PP1₁ and PP1 were almost quantitatively and specifically immunodepleted from extracts by the corresponding antiserum, indicating that the failure to detect spinophilin or neurabin in PP1 immune pellets did not result from selective precipitation of a PP1 pool not associated with these proteins. In combination, these data provide compelling evidence that spinophilin and neurabin form a complex with each other and with PP1₁ in neurons.

View larger version (79K): [in this window] [in a new window]   Fig. 8.   Selective coimmunoprecipitation of PP11, spinophilin, and neurabin from rat forebrain extracts. A combined S2/S3 extract was incubated with rabbit antibodies to spinophilin or neurabin or preimmune serum (PIS), or in parallel with sheep antibodies to PP1, PP11, or PP2AC, and immune complexes were isolated (see "Experimental Procedures"). Aliquots of immune complexes (P, about 25% of total), immune supernatants (S, about 1.5% of total), and the input (15 µg of protein; about 1.5% of total) were immunoblotted for the indicated proteins. Lower portions of some membranes were cut vertically in the center of the input lane (indicated by line): the left and right halves were probed with affinity purified sheep and rabbit antibodies to the indicated PP1 isoform, respectively. IgGH and IgGL indicate heavy and light chains, respectively, of IgG used for immunoprecipitation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The present results suggest a molecular mechanism for selective localization of PP1 isoforms at synapses, namely via their association with spinophilin and/or neurabin. We previously detected four brain PP1bps which were enriched in isolated postsynaptic densities, and demonstrated that two of these proteins (PP1bp134 and PP1bp175) were associated with unidentified isoforms of PP1 in brain extracts (13). The present studies significantly extended this prior work by purifying a brain holoenzyme containing three major proteins, PP1bp134, PP1bp175, and PP1 (Fig. 1, Table I). The two PP1bps were identified as spinophilin and neurabin, respectively (Figs. 2 and 3). Spinophilin was initially identified as a PP1bp in a yeast two-hybrid screen (14), and independently as an F-actin-binding protein (17). Its homolog, neurabin also was purified as an F-actin-binding protein and cloned (16). Here we show for the first time that neurabin, like spinophilin, binds to PP1 (Figs. 4-8). The actin-binding properties of neurabin and spinophilin likely account, at least in part, for their enrichment in isolated postsynaptic densities (13), which contain significant amounts of actin (39). Since the PP1 complex was purified from actin-enriched brain extracts and both spinophilin and neurabin bind F-actin, we termed this form of the enzyme PP1_A, although it may also associate with other organelles or structures by as yet poorly defined mechanisms (e.g. the PDZ domain).

Purified PP1_A is likely to consist of a mixture of different oligomeric forms, based on the following lines of evidence: 1) PP1_A elutes from Superdex 200 in a 400-600-kDa peak (13), and at no step in the purification are PP1bp175 and PP1bp134 separated from each other (not shown); 2) spinophilin and neurabin can be coimmunoprecipitated from HEK293 cell extracts expressing both proteins (Fig. 4, right panel) and from brain extracts (Fig. 8); and 3) recombinant neurabin or spinophilin were shown to self-oligomerize into 400-600-kDa complexes, likely dimers, trimers, or tetramers (16, 17). In combination, these data suggest that purified PP1_A contains spinophilin·neurabin heteromultimers, as well as spinophilin·spinophilin, and possibly neurabin·neurabin homomultimers, presumably formed by interactions between coiled-coil and/or PDZ domains. Since the overall molar ratio of PP1 to (spinophilin + neurabin) is approximately 1:1 (see "Results"), it is likely that PP1 binds to all molecules of spinophilin and neurabin in purified PP1_A, whether they are in homo- or heteromultimeric complexes.

Analysis of PP1-binding to GST fusion proteins containing diverse regions of spinophilin and neurabin suggests that the interactions involve a conserved domain containing a consensus PP1-binding motif (Figs. 5-7). This motif presumably interacts with PP1 in the same binding pocket occupied by the peptide analog of the PP1-binding domain from G_M in a co-crystal structure (7). The shortest GST fusion proteins containing this domain from spinophilin or neurabin were only weak inhibitors of PP1 activity (<20% inhibition at 400 nM), similar to the inhibition observed with a GST-G_M fusion protein (Fig. 6). This is consistent with the crystal structure, in which the hydrophobic binding groove for the PP1-binding motif is opposite the catalytic site (7). However, extension of fusion proteins toward the N- and/or C termini resulted in significantly increased potency of inhibition (Fig. 6). In contrast, fusion proteins lacking the PP1-binding motif had no significant effect on PP1 activity. A paper published during the preparation of this article made similar observations using only GST-spinophilin fusion proteins, and also showed that mutation of the PP1-binding motif abrogated inhibition (24). Together with the present results, these data indicate that the PP1-binding motif is essential, but not sufficient, for potent inhibition of PP1; presumably additional inhibitory contacts with PP1 are made involving regions of spinophilin/neurabin outside of the minimal PP1-binding fragments. In this sense, inhibition of PP1 by spinophilin and neurabin may resemble that by inhibitor-1 and DARPP-32, which potently inhibit PP1 by making multiple interactions (25-27). However, one obvious difference is that inhibitor-1 and DARPP-32 must be phosphorylated to become potent PP1 inhibitors, whereas spinophilin and neurabin need not be phosphorylated. It will be interesting to determine whether phosphorylation of spinophilin and/or neurabin affects the binding and/or activity of associated PP1, and also to investigate the effects of spinophilin and neurabin on dephosphorylation of physiological substrates (see below).

An intriguing aspect of this work is the selectivity of spinophilin and neurabin for interactions with PP1₁ over PP1 (Figs. 3, 7, and 8). PP1bp134 and PP1bp175 were previously shown to preferentially bind to , ₁, and ₂ isoforms over the  isoform in gel overlay assays (13). In addition, we recently demonstrated that PP1₁ and spinophilin are enriched in actin cytoskeletal extracts from hindbrain, whereas PP1 is enriched in microtubule extracts (11). In the present studies, the analysis of binding specificity of spinophilin and neurabin was restricted to PP1₁ and PP1 because specific antibodies were only available for these isoforms. However, gel overlay data (13) suggest that spinophilin and neurabin also bind PP1, and possibly PP1₂, in neurons. Spinophilin was first isolated as a PP1-binding protein (14) and PP1 is localized to dendritic spines and enriched in isolated postsynaptic densities similarly to PP1₁ (10, 11, 13, 18), whereas PP1 is enriched in cell bodies (11). Furthermore, both spinophilin and neurabin were localized to synapses of cultured neurons, although it was unclear whether the localization was pre- or postsynaptic, or both (16, 17). Although PP1₂ is present in brain extracts, much less information is available concerning its localization (11). While the mechanism for selective somatic targeting of PP1 is unclear, the isoform-selectivity of spinophilin and neurabin in vitro is consistent with localization of  and ₁ isoforms to synapses.

The present observations raise the question as to a biochemical explanation for the PP1 isoform selectivity of spinophilin and neurabin. As mentioned above, 6 residues surrounding the PP1-binding motif (Arg⁶⁴-Arg-Val-Ser-Phe-Ala) from G_M make multiple interactions with a hydrophobic channel on PP1 that is opposite the catalytic site (7). Residues (Ile¹⁶⁹, Leu²⁴³, Phe²⁵⁷, Leu²⁸⁹, Cys²⁹¹, and Phe²⁹³) in this hydrophobic channel of PP1 are highly conserved between the four mammalian PP1 isoforms, suggesting that the consensus PP1-binding motifs of spinophilin (Arg⁴⁴⁷-Lys-Ile-His-Phe-Ser) and neurabin (Arg⁴⁵⁶-Lys-Ile-Lys-Phe-Ser) make similar interactions with PP1. Residues surrounding the PP1-binding motif may make additional contacts with , ₁, and ₂ isoforms that are not favored with the  isoform. This may result from conformational differences in the variable C-terminal tails of PP1 isoforms. The C terminus of PP1₁ (residues 300-323), as well as 6 variable N-terminal residues, were not resolved in the crystal structure (7). Perhaps the C-terminal tail of the  isoform interferes with spinophilin/neurabin binding to PP1, whereas C-terminal tails of other isoforms do not. Alternatively, C-terminal tails of , ₁, and ₂ isoforms may be more directly involved in the interaction. However, this seems unlikely since antibodies raised against the C-terminal sequence of PP1₁ were able to coimmunoprecipitate spinophilin and neurabin (Fig. 8). The precise mechanism for isoform-selective interactions will be the subject of future studies.

PP1 targeting (and possibly regulation) and F-actin binding appear to be functions of spinophilin and neurabin. The presence of PDZ domains in both proteins, and proline-rich putative SH3-binding domains in spinophilin (14), suggests additional protein-protein interactions occur in cells. In fact, the neurabin PDZ domain has been shown to interact with p70 S6 kinase (29), whereas the spinophilin PDZ domain interacts with the rat homolog of what was initially believed to be the protein encoded by the lin10 gene in Caenorhabditis elegans (30, 31). However, the authentic lin10 gene was recently shown to encode an unrelated protein (32), making the identity and role of the spinophilin-binding protein unclear. Residues 100-371 of spinophilin also interact with the third intracellular loop of the D2 dopamine receptor (33). Since spinophilin and neurabin form heteromultimers they could assemble a great diversity of distinct regulatory complexes. The presence of PP1 in these complexes may allow for efficient regulation of associated functions by dephosphorylation.

The physiological functions and substrates of PP1_A holoenzyme(s) are not yet established. Neurabin has been implicated in neurite formation in developing neurons in culture (16), presumably due to its roles in PP1 targeting, F-actin binding and/or other protein-protein interactions. In mature neurons, there is substantial evidence linking PP1-mediated dephosphorylation of AMPA-type glutamate receptors in dendrites to certain forms of long-term depression (34, 35). A peptide corresponding to spinophilin residues 438-461 was shown to potentiate AMPA receptor currents in neurons (36). This peptide was presumed to disrupt targeted PP1 holoenzymes involving spinophilin resulting in reduced dephosphorylation of colocalized AMPA receptors, and thereby sustained phosphorylation. However, it is likely that PP1 holoenzymes involving other targeting proteins, such as neurabin, also were disrupted by this spinophilin peptide. Another PP1 substrate in dendritic spines is Thr²⁸⁶-autophosphorylated constitutively active calcium/calmodulindependent protein kinase II, which is dephosphorylated by PP1 when the kinase is associated with postsynaptic densities (18, 38); interestingly, one role of calcium/calmodulindependent protein kinase II during long-term potentiation induction is believed to be potentiation of AMPA receptors (40). Thus, synaptic targeting of PP1 by spinophilin and/or neurabin, in the form of PP1_A holoenzymes, appears to provide coordinated regulation of several physiologically relevant substrates. Further investigation of the interactions between PP1, spinophilin, and neurabin will likely provide fundamental insight into the biochemical mechanisms of long-term depression and other forms of synaptic plasticity.

	FOOTNOTES

^* This work was supported in part by National Institutes of Health Research Grants GM47973 and NS37508 (to R. J. C.) and GM51366 (to B. E. W.). The VUMC Peptide Sequencing and Amino Acid Analysis Shared Resource is supported by NCI Grant CA68485 and the E. Bronson Ingram Cancer Center. The VUMC Cell Imaging Core is supported by NCI Grant CA68485 and NIDDK Grant DK20593.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ Supported by a postdoctoral fellowship from the Tennessee affiliate of the American Heart Association.

Established Investigator of the American Heart Association. To whom correspondence should be addressed: Rm. 702, Light Hall, Dept. of Molecular Physiology and Biophysics, Vanderbilt University School of Medicine, Nashville, TN 37232-0615. Tel.: 615-936-1630; Fax: 615-322-7236; E-mail: roger.colbran@mcmail.vanderbilt.edu.

² Spinophilin was initially identified by yeast two-hybrid screening of a rat brain cDNA library, and was named based on its immunohistochemical enrichment in dendritic spines (14). The same protein was independently purified and cloned as an F-actin binding protein and named neurabin II (17) based on its homology to neurabin (16). We elected to use spinophilin to name this protein in the present paper since it was the first published name.

	ABBREVIATIONS

The abbreviations used are: PP1, catalytic subunit of protein phosphatase 1; DTT, dithiothreitol; PMSF, phenylmethylsulfonyl fluoride; PP1bp, PP1-binding protein; PP2A_C, catalytic subunit of protein phosphatase 2A; PAGE, polyacrylamide gel electrophoresis; GST, glutathione S-transferase; CAPS, 3-(cyclohexylamino)propanesulfonic acid; DIG, dioxigenin; G_M, glycogen-targeting subunit of protein phosphatase 1; AMPA, -amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES