The Neuronal Actin-binding Proteins, Neurabin I and Neurabin II, Recruit Specific Isoforms of Protein Phosphatase-1 Catalytic Subunits*

Neurabins are protein phosphatase-1 (PP1) targeting subunits that are highly concentrated in dendritic spines and post-synaptic densities. Immunoprecipitation of neurabin I and neurabin II/spinophilin from rat brain extracts sedimented PP1γ1 and PP1α but not PP1β. In vitro studies showed that recombinant peptides representing central regions of neurabins also preferentially bound PP1γ1 and PP1α from brain extracts and associated poorly with PP1β. Analysis of PP1 binding to chimeric neurabins suggested that sequences flanking a conserved PP1-binding motif altered their selectivity for PP1β and their activity as regulators of PP1 in vitro. Assays using recombinant PP1 catalytic subunits and a chimera of PP1 and protein phosphatase-2A indicated that the C-terminal sequences unique to the PP1 isoforms contributed to their recognition by neurabins. Collectively, the results from several different in vitro assays established the rank order of PP1 isoform selection by neurabins to be PP1γ1 > PP1α > PP1β. This PP1 isoform selectivity was confirmed by immunoprecipitation of neurabin I and II from brain extracts from wild type and mutant PP1γ null mice. In the absence of PP1γ1, both neurabins showed enhanced association with PP1α but not PP1β. These studies identified some of the structural determinants in PP1 and neurabins that together contribute to preferential targeting of PP1γ1 and PP1α to the mammalian synapse.

PP1␥1, are expressed in all tissues (4) with PP1␥2, an alternately spliced product of the PP1␥ gene, present predominantly in testes (5,6). Immunocytochemistry using isoform-specific antibodies suggested that expression of PP1 isoforms varied in different brain regions where they are also localized to different subcellular compartments (6,7). For example, PP1␤ was the predominant isoform associated with microtubules in the neuronal cell body, whereas PP1␥1 and PP1␣ were preferentially concentrated in dendritic spines (6,8). Furthermore, by analyzing endogenous PP1 movement during the cell cycle, Andreassen et al. (9) showed that the distribution of PP1 isoforms in cells was highly dynamic. This placed new emphasis on understanding the mechanisms that target individual PP1 isoforms to cellular organelles.
Isolation of PP1 bound to skeletal muscle glycogen (10) and myosin (11) established the paradigm that regulatory or targeting subunits bound to PP1 catalytic subunits dictate the subcellular localization, substrate recognition, and hormonal control of PP1. The search for PP1 regulators that control functions as diverse as protein synthesis, gene expression, cell division, and motility has thus far yielded more than 50 PP1binding proteins (12). Whereas myosin phosphatases from skeletal (11) and cardiac muscle (13) copurified with both PP1␣ and ␤ (14), the smooth muscle myosin phosphatase bound exclusively PP1␤ (15)(16)(17). The molecular basis by which the myosinbinding subunits and other regulators selected specific PP1 isoforms remains unknown.
PP1 plays a key role in regulating synaptic transmission in mammalian neurons (18). PP1␥1 and PP1␣, but not PP1␤, were enriched in dendritic spines (6,8), where they associated with the actin-rich structure known as the post-synaptic density (PSD) (19). Recent studies (20) also suggested a heterogeneity of spines that contained either PP1␣ alone or both PP1␣ and PP1␥1. The neuronal actin-binding proteins, neurabin I and neurabin II/spinophilin, were also localized in PSD and bind PP1 (19,21,22). Thus, neurabins were excellent candidates for recruiting specific PP1 isoforms to the synapse. Prior studies showed that anti-neurabin immune complexes from rat brain contained PP1␥1 but excluded PP1␤ (23). Far Westerns of brain extracts using recombinant PP1 catalytic subunits as probes suggested that neurabins displayed a significant preference for PP1␥1 and PP1␣ over PP1␤ (24). Our studies highlighted a domain in both neurabins that displayed PP1 isoform selectivity similar to the native proteins. Subsequent analyses of chimeric neurabins focused attention on sequences in neurabin I that regulated PP1 activity and determined PP1␤ binding. Mutation of the PP1 catalytic subunit also suggested that its C terminus participated in neurabin binding. Finally, neurabin complexes from tissues of a PP1␥ null mouse confirmed their preference for PP1␥1 and PP1␣ over PP1␤. These studies provided the first experimental evidence for structural contributions from both PP1 and its regulator, neurabin, in selective targeting of PP1␥1 and PP1␣ to the neuronal actin cytoskeleton.
Expression of Recombinant Neurabins-pRSET-B encoding hexahistidine-NrbI-(374 -516) was transformed into BL21(DE3)pLysS Escherichia coli (Stratagene). Bacteria were grown in LB media to an A 600 ϭ 0.6, and protein expression was induced by addition of 1 mM isopropyl-1-thio-␤-D-galactopyranoside to the media and further incubation for 2 h at 30°C. Following the lysis of bacterial cells using sonication, hexahistidine fusion protein was purified on Ni ϩ -nitrilotriacetic acid-agarose according to the manufacturer's instructions (Qiagen).
The pGEX-5X-2 plasmids were transformed into BL21-Codon Plus-RIL (Stratagene), and the bacteria were grown in LB media to A 600 ϭ 0.6. Protein expression was induced by the addition of 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside to the media and continued growth for 6 h at 24°C. Cells were lysed by two passages through a French press (at 1000 pounds/square inch), and GST fusion proteins were affinity-puri-fied using glutathione-Sepharose according to the manufacturer's instructions (Amersham Biosciences).
Sedimentation Assays-Rat brain was obtained from Pel-Freeze, and brain and testes from ϩ/ϩ and PP1␥ Ϫ/Ϫ mice were provided by Susan Varmuza, University of Toronto (3). The tissue was homogenized in 50 mM Tris-HCl, pH 7.5, containing 5 mM EDTA, 5 mM EGTA, 10 mM NaCl, 1% w/v deoxycholate, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 1 g/ml each of aprotinin, leupeptin, and pepstatin using a Dounce homogenizer. Following centrifugation at 100,000 ϫ g for 60 min, the supernatant was dialyzed against 100 volumes of 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 10 mM NaCl, 0.1 mM phenylmethylsulfonyl fluoride, and 0.1 mM benzamidine for 3 h with one change of buffer. The lysates were used either immediately or frozen in liquid N 2 and stored at Ϫ80°C.
For immunoprecipitation, anti-NrbI or NrbII antibody (10 l) was mixed with rat brain (2.5 mg of total protein) or mouse brain lysate (4 mg of total protein) for 1 h at 4°C. A 1:1 slurry of protein A-agarose (Bio-Rad) and protein G-Sepharose 4B (Sigma) (25-l total bed volume) was added, and the mixture was incubated for 1 h. The beads were washed 4 times with NETN-250 (250 mM NaCl, 1 mM EDTA, 10 mM Tris-HCl, pH 7.5, and 0.5% v/v Nonidet P-40), and the bound proteins were eluted using 25 l of SDS sample buffer prior to SDS-PAGE.
For sedimentation or pulldown assays, glutathione-Sepharose (25-l bed volume) was equilibrated with Tris-buffered saline. GST-neurabin proteins were added, and the volume was adjusted to 300 l with Tris-buffered saline prior to incubation for 1 h at 4°C. The beads were washed 2 times with Tris-buffered saline, and rat brain lysate (5 mg total protein) was added and the mixture incubated for a further 1 h at 4°C. The beads were washed 4 times with NETN-250, and the bound proteins were eluted using 25 l of SDS sample buffer prior to SDS-PAGE on 12% (w/v) acrylamide gels. Coomassie Blue staining verified the loading of GST-neurabins on beads, and the specific PP1 isoform sedimented was quantified by immunoblotting and densitometry using Densitometer SI and ImageQuant software (Amersham Biosciences).
Immunopurification of Rat Brain PP1 Isoforms-Affinity-purified sheep IgGs specific for PP1␤ or PP1␥ 1 (23) were covalently coupled to Affi-Gel 15 resin (Ϸ1 mg of IgG per ml) as described by the manufacturer (Bio-Rad). The anti-␤ and anti-␥ 1 resins (0.1 ml packed beads) were incubated at 4°C for 4 h with 0.25 ml of a rat brain protein phosphatase catalytic subunit preparation containing a mixture of PP1 and PP2A catalytic subunits in 2.5 ml of 50 mM Tris-HCl, pH 7.5, 0.2 M NaCl, 0.1% Triton X-100, 1 mM dithiothreitol, 2 mM MnCl 2 (Buffer A), and 0.25 mg/ml bovine serum albumin. The resin was washed 4 times with 5 ml of Buffer A, and bound PP1s were eluted sequentially with 3 M magnesium chloride (3ϫ 0.3-ml aliquots) and then 3 M sodium isothiocyanate (3ϫ 0.3-ml aliquots), both in Buffer A. The eluted samples were dialyzed separately against 20 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 0.1 mM EGTA, 2 mM MnCl 2 , 10% glycerol before storage at Ϫ80°C. By using affinity-purified rabbit antibodies to specific PP1 isoforms (6) and recombinant PP1 isoforms as standards, the purity of the immunopurified rat brain PP1 isoforms was verified, and protein concentration was estimated (0.5-2.5 g/ml).
Purification of Native and Recombinant PP1-PP1 catalytic subunit was purified from rabbit skeletal muscle according to DeGuzman and Lee (28), and purified PP2A catalytic subunit was provided by Brian Wadzinski, Vanderbilt University.
Protein Phosphatase Assays-Phosphorylase b obtained from Calzyme Laboratories, Inc., was phosphorylated using phosphorylase kinase (Invitrogen) and [␥-32 P]ATP (specific activity 3000 Ci/mmol, PerkinElmer Life Sciences) and used as substrate in a protein phosphatase activity as described in Shenolikar and Ingebritsen (32). GST-GluR1 C-terminal tail (provided by Michael Ehlers, Duke University (33)) was phosphorylated by purified PKA (bovine heart) in the presence of [␥-32 P]ATP, and the unincorporated 32 P was removed by dialysis. Assays with PP1 and PP2A were performed for 10 min at 37°C in 50 mM Tris-HCl, pH 7.5, 1 mg/ml bovine serum albumin, 1 mM EDTA, and 0.1% ␤-mercaptoethanol using 20 M 32 P-labeled phosphorylase a or 0.2 M GluR1 as substrates and restricted to a maximum of 20% release of [ 32 P]phosphate to ensure linearity.
Peptides that disrupt PP1 binding to targeting subunits modified AMPA currents in striatal neurons (36), suggesting that a PP1 complex dephosphorylated the ␣-amino-5-hydroxy-3-methyl-4-isoxazole propionate (AMPA) receptor. Mutation of FIG. 1. Neurabin-PP1 complexes in rat brain. NrbI and NrbII were immunoprecipitated from deoxycholate extracts of rat brain (2.5 mg of total protein) as described under "Materials and Methods." The preimmune serum from rabbits immunized with NrbII was used as a control. Labeled across the top of the blots are immunoprecipitates and the input extract (1 and 10% total available for sedimentation), which were subjected to SDS-PAGE on 8.75 (for neurabins) or 12% (for PP1) polyacrylamide gels, transferred to polyvinylidene difluoride membranes, and immunoblotted with monospecific antibodies against NrbI, NrbII, PP1␣, PP1␤, and PP1␥1, as indicated to the left of the blots. the mouse NrbII gene abolished NrbII expression and impaired AMPA receptor regulation (37) consistent with a deficit in GluR1 (an AMPA receptor subunit) phosphatase activity. Thus, we analyzed the dephosphorylation of the cytoplasmic C-terminal tail of the GluR1 subunit fused to GST that was phosphorylated in vitro by PKA (Fig. 3B). Both GST-NrbI-(374 -516) and GST-NrbII-(354 -494) inhibited dephosphorylation of GST-GluRI by skeletal muscle PP1, although requiring higher concentrations than those needed to inhibit its phosphorylase phosphatase activity (Fig. 3A). PP1 binding was essential as the mutant GST-NrbI-(374 -516; AAAA) failed to inhibit GluRI dephosphorylation by PP1. At similar concentrations, GST-NrbII-(354 -494) also failed to inhibit PP2A-mediated dephosphorylation of GST-GluRI. These data suggested that the interaction of neurabins with PP1 via the KIXF motif regulated its activity against neuronal substrates.
Analysis of Chimeric Neurabins-To analyze the neurabin sequences that mediated PP1 binding and regulation, we generated chimeric neurabins that exchanged regions of NrbI and NrbII, schematically shown in Fig. 4A. NrbI and NrbII sequences C-terminal to site 1 are highly conserved, whereas those N-terminal to this site diverge significantly. Two other sites, 2 and 3, were selected to narrow down the functional domains. PP1 pulldowns from rat brain lysates (Fig. 4B) and inhibition of skeletal muscle PP1 (Fig. 4, C and D) were used to characterize the chimeric neurabins.
We also analyzed the neurabin chimeras in protein phosphatase assays, which are more readily quantified than the pulldowns and detected both PP1 activity and binding. We utilized a phosphorylase phosphatase assay in the presence of physiological salt concentrations (100 -150 mM KCl or NaCl) shown previously to enhance differences in the properties of modified PP1 regulators (21,38). GST-NrbI-(374 -516) was an ϳ10-fold more potent inhibitor of phosphorylase phosphatase activity of rabbit muscle PP1 catalytic subunit than the equivalent NrbII peptide, GST-NrbII-(354 -494) in the presence of 100 mM KCl. In the same assay, GST-NrbCH1 containing NrbI sequences N-terminal to the KIXF motif displayed similar efficacy to GST-NrbI-(374 -516) as a PP1 inhibitor. In contrast, GST-Nrb-CH2, which contained NrbII N-terminal sequences, showed the decreased inhibitory potency of GST-NrbII-(354 -494). Switching further N-terminal sequences, at site 3 (Fig. 4A), yielded GST-NrbCH5 and GST-NrbCH6, which demonstrated PP1 inhibition with IC 50 values similar to GST-NrbI-(374 -516) and GST-NrbII-(354 -494) respectively. This result was essentially identical in 0 (Fig. 4D) and 100 mM KCl conditions. These data demonstrated that sequences encompassed by sites 1 and 3 not only facilitated the PP1␤ recognition by NrbI but also accounted for its enhanced activity as a PP1 regulator.
Regulation of PP1 Isoforms by Neurabin Peptides-To inves-tigate the regulation of other PP1 isoforms by the recombinant neurabins, we established an immunopurification procedure using isoform-specific antibodies to isolate the individual PP1 catalytic subunits from rat brain. As anticipated, the immunopurified PP1␥1 and PP1␤ were inhibited in a dose-dependent manner by both GST-NrbI-(436 -479) (Fig. 5A) and GST-NrbII-(427-470) (Fig. 5B) with the NrbI peptide showing increased activity as a PP1 inhibitor. In contrast to PP1␥1, which was inhibited by the NrbI and NrbII peptides with an IC 50  NrbII-(354 -494) were aligned using BoxShade. Identical amino acids are shown in black, and homologous residues are shown in gray. Three sites, x, y, and z were chosen to generate the GST-fused NrbI/NrbII chimeras described in A. The dashed region between sites k and l mediates certain PP1 regulatory properties of NrbI and is described in the text. B, sedimentation of PP1 from rat brain lysates using increasing concentrations of the chimeric neurabins. The bound proteins were eluted with SDS sample buffer and analyzed by Western immunoblotting for the major PP1 isoforms. C, inhibition of phosphorylase phosphatase activity of rabbit skeletal muscle PP1 by NrbI (diamonds), NrbII (triangles) and the chimeras, NrbCH1 (circles), and NrbCH2 (squares). D, comparison of PP1 inhibition by NrbI (diamonds), NrbII (triangles), and NrbCH5 (circles) and NrbCH6 (squares). The experiment in C was conducted in 100 mM KCl (see text), and the experiment in D included no KCl. C and D show representative profiles from at least three independent experiments carried out in duplicate and that varied by less than 5%.
The principal differences in primary structures of the human PP1 isoforms are located near their N and C termini with the most significant difference being in their C-terminal sequences (Fig. 6A, top). PP2A, which as shown above does not associate with neurabins, shows even greater differences in this region. To evaluate the contribution of PP1 C-terminal sequences in PP1/neurabin interactions, we compared WT PP1␣ with a chimeric phosphatase, CHRM2, which contained the N terminus of PP1␣ fused to the C terminus of PP2A (29,31). Whereas both enzymes showed similar specific activity as phosphorylase phosphatases, the dose-response curve for inhibition of CHRM2 by His-NrbI-(374 -516) was 100-fold right-shifted compared with that of WT PP1␣ (Fig. 6B). Similar results were obtained using GST-NrbII-(354 -494) (data not shown). Earlier work (39) showed that, compared with PP1␣, the CHRM2 phosphatase was poorly sedimented by GST-NrbII, indicating that effective neurabin binding required PP1 C-terminal sequences. This suggests that the NrbI and NrbII central domains interacted differently with the differing C-terminal sequences in PP1 isoforms, thereby contributing to their preferential binding to PP1␥1 and to a lesser extent PP1␣.
Neurabin-PP1 Complexes in PP1␥ Ϫ/Ϫ Mouse Brains-The availability of a PP1␥ null mouse (3) provided a unique opportunity to analyze further the selectivity of neurabins for PP1 isoforms. Immunoblotting established that both NrbI and NrbII are expressed in brains from WT and mutant mice (Fig.  7A). NrbII, but not NrbI, is expressed in non-neuronal tissues, albeit at lower levels than brain, and its expression was unaltered in testes from WT and mutant animals. A pan anti-PP1 antibody showed that the overall PP1 levels were much higher in brain than in testes. Total PP1 levels in both tissues were slightly elevated by inactivation of the mouse PP1␥ gene. As PP1␤ levels were identical, this reflected a specific increase in PP1␣ that may compensate for the loss of PP1␥1 and PP1␥2 in these tissues from the mutant mice.
As seen in immunoprecipitations from rat brain (Fig. 1), anti-NrbI and NrbII immunoprecipitates from the mouse tissue contained both NrbI and NrbII (Fig. 7B). Although NrbI immune complexes contained significant NrbII, the NrbII complexes contained much less NrbI, suggesting that in contrast to rat brain, a significant portion of NrbII in mouse brain may exist as homodimers. Both neurabins coimmunoprecipitated with PP1␣ and PP1␥1 from WT brain lysates, and as anticipated, little or no PP1␤ association was seen. Although neurabin hetero-and homodimers were unchanged in lysates from mutant mice, in the absence of PP1␥1 a considerable increase in PP1␣ recruitment by the two neurabins was noted. This was particularly obvious in the NrbII immunoprecipitates (Fig. 7B). Although similar results were seen in testes (data not shown), the lower levels of spinophilin in testes (Fig. 7A) made similar comparative analyses more difficult. DISCUSSION In the nervous system, protein phosphatase-1 (PP1) regulates short term events such as the phosphorylation status of receptors, ion channels, and signaling protein, as well as long term events requiring changes in protein translation, gene expression, and neuronal morphology that together modify neuronal plasticity (18, 40 -42). PP1 activity controls both long term potentiation or LTP (43,44) and long term depression or LTD (41,45) in the hippocampal synapse. Biochemical (19,25) and immunological (8) studies showed high concentrations of PP1, specifically PP1␥1 and PP1␣, in the actin-rich structure known as the PSD, where it was ideally positioned to regulate substrates such as NMDA (46,47) and AMPA subtypes of glutamate receptors (36,48), calcium (49) and potassium (50) channels, and CaMKII (25,44). As seen with its substrate, CaMKII (51), PP1 localization at post-synaptic sites was controlled by neuronal activity (45).
Disruption of the neurabin II gene also showed that PP1 targeting was essential for the regulation of AMPA receptors and LTD in the mouse hippocampus (37). Neurabin complexes from rat brain selectively bound the two PP1 isoforms, PP1␥1 (23) and PP1␣ (Fig. 1), present in the PSD. Furthermore, fluorescent recovery after photobleaching experiments suggested that both neurabins shuttled rapidly on and off the actin-cytoskeleton (53). Thus, neurabins were excellent candidates for the dynamic targeting of PP1␥1 and PP1␣ to PSDs to regulate synaptic plasticity.
Although isolation of PP1 regulators from mammalian tissues suggested that they selectively associated with PP1 isoforms, analysis of recombinant PP1 catalytic subunits has thus far failed to show significant differences in their regulation by endogenous regulators. For example, the apparent selectivity of the smooth muscle myosin-binding subunit for PP1 isoforms in vitro (34) was no longer observed when the recombinant PP1 catalytic subunits were "refolded" through their association with inhibitor-2, a putative PP1 chaperone (34,54,55). The overexpression of PP1 catalytic subunits and regulators in cultured cells also failed to establish the PP1 isoform specificity for NIPP-1, a nuclear inhibitor of PP1 that was isolated from calf thymus nuclei with PP1␣ (56). The overexpressed NIPP-1 bound equally to PP1␣, PP1␤, and PP1␥1 (57). In contrast, our in vitro studies examined PP1-neurabin complexes in tissue extracts, analyzed the regulation of immunopurified native PP1 isoforms and recombinant PP1 catalytic subunits, and suggested that Nrb sequences flanking the KIXF motif dictated selectivity for PP1 isoforms similar to that seen with neurabins in vivo. As neurabins homo-and heterodimerize (23,65) and bind polymerized actin as well as other cellular proteins (58 -62), our studies focused on the structural determinants intrinsic to neurabins that dictated PP1 recognition independent of any contributions from other cellular factors.
Having focused on a central domain that bound PP1, we generated several NrbI/NrbII chimeras that exchanged regions N-terminal to the KIXF motif. This enabled us to map a region that permitted PP1␤ binding in vitro, a unique property of NrbI. This region of NrbI N-terminal to the KIXF motif altered PP1 recognition as defined by both binding and enzyme regulation. These experiments, however, did not reveal specific determinants in neurabins that distinguished PP1␥1 from PP1␣, as this property was displayed by all recombinant neurabins analyzed. On the other hand, analysis of recombinant PP1 catalytic subunits and a PP1/PP2A chimera suggested that some of the PP1 selectivity shown by neurabins may arise from their association with the unique C-terminal sequences found in individual PP1 catalytic subunits. We speculated that the region C-terminal to the KIXF motif shared in the neurabin peptides preferentially recognized the C-terminal sequences unique to PP1␥1 and PP1␥2, which showed equivalent binding to recombinant neurabins (data not shown). The differences in the C-terminal tails of PP1␣ and particularly PP1␤ (see Fig. 6A) may decrease their affinity for neurabins.
Disruption of the PP1␥ gene eliminated PP1␥1 and PP1␥2 expression in all tissues of the PP1␥ Ϫ/Ϫ mice. The PP1␥ gene deletion did not modify the expression of neurabins but slightly elevated overall PP1 levels (most likely due to upregulation of PP1␣) in tissues from the mutant mice. In the absence of PP1␥1, a significant increase in neurabin-PP1␣ complexes was noted in tissue extracts from PP1␥ Ϫ/Ϫ mice. Interestingly, even in the PP1␥ Ϫ/Ϫ mouse tissues, which contain only the ␣ and ␤ isoforms of PP1, PP1␤ was completely excluded from cellular neurabin complexes. Yet, in our in vitro assays, NrbI showed measurable association with PP1␤. This dichotomy may reflect the fact that the current in vitro assays deliberately focused on the small, essential PP1binding domain in neurabins. Alternatively, other factors in vivo may play a role in PP1␤ exclusion from neurabin complexes. Future studies will explore how the subcellular distribution of neurabins and/or the recruitment of additional cellular proteins to the neurabin-PP1 complex may further contribute to the distinct PP1 isoform selectivity neurabin displays in mammalian tissue.
The ability of neurabins to form PSD complexes selectively containing PP1␥1 and PP1␣ may have important implications for signaling at the mammalian synapse. Recent studies (20) have noted distinct populations of dendritic spines in the primate cortex, containing either PP1␣ alone or both PP1␣ and PP1␥1. Interestingly, dopamine D1 receptors were only found in spines containing both PP1 isoforms, suggesting unique signaling properties of the PP1␣/PP1␥1-containing synapses. As both neurabins bind PP1␣ and PP1␥1, form hetero-and homodimers, and rapidly shuttle on and off the actin cytoskeleton, this raises the intriguing possibility that expression and targeting of NrbI and NrbII at the PSD may directly contribute to the signaling complexity, heterogeneity, and dynamics of spines in the mammalian brain. In this regard, physiological signals such as steroid hormones in the fly (63) and rat (64) and Rac1 signaling (53) in tissue culture cells can alter neurabin expression and localization. Our studies show that NrbI differs from NrbII in both PP1 binding and regulation. Moreover, NrbI, but not NrbII, can be phosphorylated in vivo (65) and in vitro (21) by PKA at a serine adjacent to the KIKF motif to attenuate PP1 association. This suggests a complex and dynamic regulation of the PP1-neurabin complexes in the mammalian synapse. Understanding of the molecular interactions between neurabin and PP1 isoforms combined with the availability of mice lacking NrbII (37) and PP1␥1 (3) should yield insights into the role of post-synaptic neurabin-PP1 complexes in signaling and morphology of dendritic spines.