Address correspondence and reprint requests to Dr A. J. Baucum, Department of Molecular Physiology and Biophysics, Vanderbilt Kennedy Center, and Vanderbilt Brain Institute, Vanderbilt University School of Medicine, Vanderbilt University, 23rd Ave South and Pierce, 702 Light Hall, Nashville, TN 37232, USA.
Distinct physiological stimuli are required for bidirectional synaptic plasticity in striatum and hippocampus, but differences in the underlying signaling mechanisms are poorly understood. We have begun to compare levels and interactions of key excitatory synaptic proteins in whole extracts and subcellular fractions isolated from micro-dissected striatum and hippocampus. Levels of multiple glutamate receptor subunits, calcium/calmodulin-dependent protein kinase II (CaMKII), a highly abundant serine/threonine kinase, and spinophilin, a F-actin and protein phosphatase 1 (PP1) binding protein, were significantly lower in striatal extracts, as well as in synaptic and/or extrasynaptic fractions, compared with similar hippocampal extracts/fractions. However, CaMKII interactions with spinophilin were more robust in striatum compared with hippocampus, and this enhanced association was restricted to the extrasynaptic fraction. NMDAR GluN2B subunits associate with both spinophilin and CaMKII, but spinophilin-GluN2B complexes were enriched in extrasynaptic fractions whereas CaMKII-GluN2B complexes were enriched in synaptic fractions. Notably, the association of GluN2B with both CaMKII and spinophilin was more robust in striatal extrasynaptic fractions compared with hippocampal extrasynaptic fractions. Selective differences in the assembly of synaptic and extrasynaptic signaling complexes may contribute to differential physiological regulation of excitatory transmission in striatum and hippocampus.
The striatum and hippocampus control different forms of learning (Amso et al. 2005; Berke et al. 2009). The majority (~95%) of neurons in the striatum are γ-aminobutyric acid-containing medium spiny neurons (MSNs) (Huang et al. 1992; Kreitzer and Malenka 2008), whereas glutamatergic pyramidal neurons predominate in hippocampus. Although bidirectional synaptic plasticity [i.e. long-term potentiation (LTP) and long-term depression (LTD)] is thought to play a key role in the function of both brain regions, there are substantial differences in the underlying mechanisms. For example, N-methyl-d-aspartate receptor (NMDAR)- and calcium/calmodulin-dependent protein kinase II (CaMKII)-dependent LTP has been extensively studied in hippocampal CA1 pyramidal neurons (Bear and Malenka 1994; Malenka 1994; Nicoll and Malenka 1995; Malenka and Bear 2004; Lisman et al. 2012), whereas LTP in striatal MSNs can only be reliably observed when NMDAR activity is enhanced (Calabresi et al. 1992; Jia et al. 2010). Moreover, these physiological synaptic differences between brain regions extend to pathological situations. For example, Rett Syndrome and Alzheimer disease are associated with a decrease in dendritic spine density in hippocampal neurons (Chapleau et al. 2009; Penzes et al. 2011), whereas Parkinson's disease is associated with decreased spine density in striatal MSNs (Stephens et al. 2005; Zaja-Milatovic et al. 2005). However, the molecular basis for these distinct synaptic properties is not well-understood.
Differences in the localization, expression, and/or interactions of proteins that modulate postsynaptic signaling may contribute to the unique physiological properties and pathological susceptibilities of striatal and hippocampal neurons. For example, transgenic mice lacking postsynaptic density-95 (PSD-95), the prototypical postsynaptic scaffolding protein, have decreased spine density in striatal MSNs, but increased spine density in CA1 hippocampal pyramidal neurons (Vickers et al. 2006). Total tissue levels of the alpha isoform of CaMKII are somewhat higher in hippocampus compared with striatum (Erondu and Kennedy 1985), whereas total levels of the actin- and CaMKII-binding protein, α-actinin-2, are higher in striatum compared with hippocampus (Wyszynski et al. 1998). However, to the best of our knowledge, there are no studies directly comparing interactions between signaling proteins in striatum and hippocampus.
We recently found that spinophilin targets protein phosphatase 1 (PP1) to CaMKII in adult striatum (Baucum et al. 2012). Here, we report that the association of CaMKII with the spinophilin-PP1 complex is significantly greater in adult striatum compared with hippocampus. The enhanced striatal association was detected in an extrasynaptic, but not synaptic fraction. Moreover, extrasynaptic NMDAR GluN2B subunits are more robustly associated with both spinophilin and CaMKII in striatum compared with hippocampus. These differences in protein-protein interactions in specific subcellular compartments may contribute to the distinct physiological properties and/or pathological susceptibilities of striatal and hippocampal neurons.
Adult, male (1.8–7-month old) C57Bl6/J mice (Jackson Laboratories, Bar Harbor, ME, USA) were decapitated. Neostriatum (referred to as striatum) or hippocampus was dissected and either used fresh or frozen on dry ice and stored at −80°C until processed. To minimize post-mortem differences, hippocampus and striatum were dissected from the same animals at the same time and processed in parallel. Total time from decapitation to homogenization or freezing is approximately 90 s. All animal protocols were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the NIH, and were approved by the Vanderbilt Institutional Animal Care and Use Committee.
CaMKII goat antibody was previously described (McNeill and Colbran 1995). Commercially available antibodies are listed in Table S1.
Tissue homogenization: total lysates and low-ionic strength Triton-soluble fraction
Fresh or frozen mouse striata or hippocampus were homogenized in 2 ml of a low-ionic strength buffer (all values are w/v unless otherwise noted: 2 mM Tris-HCl pH 7.4, 2 mM EDTA, 2 mM EGTA, 1 mM dithiothreitol (DTT), 0.2 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 10 μg/ml leupeptin, 10 μM pepstatin, 20 μg/ml soybean trypsin inhibitor, 1 μM microcystin, and 1% (v/v) Triton X-100) using a Teflon-glass tissue grinder (Wheaton, Millville, NJ, USA) either by hand or with a motorized plunger. Total homogenates were adjusted to the same protein concentration in each experiment (0.84–1 mg/ml) as measured using the Bradford protein assay. Because of the labile nature of Thr286 phosphorylation, we only quantified Thr286 phosphorylation from freshly prepared striata or hippocampi homogenized in buffers containing additional phosphatase inhibitors (1 mM NaVO4 and 0.5 nM cypermethrin) and immediately mixed with 4X sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer (0.25 M Tris-HCl, 8% SDS, 40% glycerol (v/v), 0.032% bromophenol blue, 100 mM DTT). The remaining homogenate was incubated at 4°C for 30–60 min and then centrifuged at 9000 g for 10 min at 4°C. An aliquot of the supernatant (Triton-soluble fraction), was mixed with 4X SDS-PAGE sample buffer and the remaining Triton-soluble fraction was immunoprecipitated or incubated with GST spinophilin fusion protein (see below). We have previously shown that several PSD proteins are efficiently solubilized using this procedure (Baucum et al. 2010).
Subcellular fractions were prepared as previously described (Gustin et al. 2010). Briefly, fresh or frozen striatum or hippocampus was homogenized in an isotonic buffer (150 mM KCl, 50 mM Tris HCl pH 7.5, 1 mM DTT, 0.2 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 1 μM pepstatin, 10 mg/l leupeptin, 1 μM microcystin) without detergent and normalized to equal protein concentrations (0.34–0.6 mg/ml). The homogenate (1 ml) was passed through a 22 gauge needle two times, incubated with rocking at 4°C for 30 min and then centrifuged at 100 000 g for 1 h. The pellet was resuspended in the isotonic buffer containing 1% (v/v) Triton X-100, triturated until homogenous, and then incubated with rocking at 4°C for 30 min. Lysates were then centrifuged at 18 403 g and the supernatant (‘Extrasynaptic’ S2 fraction) was saved for immunoprecipitation (see below) or mixed with 4X SDS-PAGE sample buffer for direct loading on SDS-PAGE gels. The pellet was resuspended in isotonic buffer containing 1% Triton X-100 and 1% deoxycholate and sonicated (‘Synaptic’ S3 fraction) and then saved for immunoprecipitation (see below) or mixed with 4X SDS-PAGE sample buffer.
Immunoprecipitations and GST pulldowns
Immunoprecipitations were performed with goat spinophilin (2–3 μg) or goat CaMKII (1.8–4 μg) antibody as previously described (Brown et al. 2008; Baucum et al. 2010). Immunoprecipitations from fractionation experiments were performed using magnetic protein A/G beads (ThermoScientific, Rockford, IL, USA). Pulldowns using GST-spinophilin fusion proteins [GSTSpN1 (amino acids 1–154), GSTSpN2 (amino acids 151–300), or GSTSpC1 (amino acids 665–817)] were performed as previously described (Baucum et al. 2012). A single spinophilin band is detected in immunoblots of total lysates, whereas immunoprecipitated spinophilin sometimes migrates as a doublet, presumably because of limited proteolysis during immunoprecipitation.
Immunoblots were developed as previously described using either enhanced chemiluminescence and X-ray film or infra-red fluorescence and the Odyssey system (LiCor Biosciences, Lincoln, NE, USA) (Baucum et al. 2012).
Densitometry was performed using Image J (NIH, Bethesda, MD, USA) on images linearly adjusted for brightness and contrast. Normalization of signals in total lysates or specific subcellular fractions was performed by dividing the immunoblotted protein band density by the density of total protein stain (Ponceau S) (Gustin et al. 2010). For co-immunoprecipitations, levels of individual co-precipitating proteins were normalized to the amount of primary immunoprecipitated protein (e.g. CaMKII immunoreactivity in spinophilin immunoprecipitates was normalized to spinophilin immunoreactivity in the spinophilin immunoprecipitates). In co-immunoprecipitations from specific subcellular fractions, immunoreactivity of the co-precipitating protein was normalized to levels of the primary immunoprecipitated protein as well as the amount of the co-precipitated protein in the input of the corresponding subcellular fraction. For pulldowns, individual co-precipitating proteins were normalized to the amount of the spinophilin GST protein on the gel as measured by Ponceau S stain as well as the amount in the low-ionic strength, Triton-soluble fraction.
To allow for comparison across multiple gels, a ratio was obtained by dividing the immunoreactivity in striatal samples by immunoreactivity values in either the corresponding hippocampal fraction or the hippocampal S2 fraction (to compare S2 and S3 values together) analyzed on the same gel. An n of one to three animals was quantified per gel from two to three sets of experiments. All data were normalized to immunoreactivity in the hippocampal S2 fraction, transformed, and plotted on an antilog scale to allow for symmetrical comparisons of both increased and decreased ratios. The number of animals per analysis is shown on each graph.
For comparisons of hippocampus and striatum within a fraction, an unpaired Student's t-test was performed if the variances between the two groups were not significantly different. An unpaired t-test with Welch's correction was applied if there was a significant difference in the variances. When data were evaluated for comparison of hippocampus and striatum from S2 to S3 fractions together, a two-way anova followed by an uncorrected Fisher's LSD post hoc test was used.
Differences in total striatal and hippocampal synaptic protein levels
To begin to understand mechanisms underlying differences in synaptic regulation between striatum and hippocampus, we compared total tissue levels of several proteins that have been functionally implicated in excitatory synaptic modulation. Whole striatal and hippocampal lysates from adult mice were loaded at equal protein concentrations, as evidenced by approximately equal staining using Ponceau S (Fig. 1a). Immunoblotting (Fig. 1b) revealed that total levels of the postsynaptic density marker, PSD-95, were not significantly different between the two brain regions (Fig. 1c). Levels of the presynaptic vGlut1 transporter were significantly lower in striatum compared with hippocampus (Fig. 1d), whereas levels of vGlut2 were significantly higher in striatum compared with hippocampus (Fig. 1d). Levels of the GluN1 and GluN2B NMDAR subunits were 30–60% lower in striatum compared with hippocampus (Fig. 1e), whereas total levels of GluA1 and GluA2, subunits of the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR), were 60–80% lower in striatal lysates (Fig. 1f).
CaMKII is a key regulator of AMPARs and NMDARs (Derkach et al. 1999; Hayashi et al. 2000; Sessoms-Sikes et al. 2005; Lu et al. 2010). Total levels of CaMKIIα and CaMKIIβ were approximately 40% lower in striatum than in hippocampus (Fig. 2a and b). Interestingly, levels of CaMKIIα phosphorylation at Thr286 (normalized to total CaMKIIα) were significantly higher in striatal, than in hippocampal, lysates (Fig. 2c), but phosphorylation of CaMKIIβ at the equivalent residue (Thr287) was not significantly different (Fig. 2d).
PP1 activity is a key determinant of CaMKII autophosphorylation, glutamate receptor phosphorylation, and synaptic plasticity (Strack et al. 1997; Blitzer et al. 1998; Hu et al. 2007; Mullasseril et al. 2007; Farinelli et al. 2012), and PP1 targeting by neurabin and spinophilin is important for normal plasticity (Yan et al. 1999; Allen et al. 2006). While there was no difference in the total levels of neurabin between the two brain regions (Fig. 2e), total PP1 catalytic subunit levels were ∼twofold higher in striatal compared with hippocampal lysates (Fig. 2f). In contrast, striatal levels of a major synaptic PP1 targeting subunit, spinophilin, were ∼25% lower (Fig. 2g).
Differential association of CaMKII with spinophilin in striatum and hippocampus
We recently found that CaMKII associated with spinophilin in a low-ionic strength, Triton-soluble fraction that largely solubilizes PSD proteins (Baucum et al. 2010, 2012). Interestingly, CaMKII association with spinophilin was more robust in the striatal, compared with hippocampal, low-ionic strength, Triton-soluble fraction (Fig. 3a). While the association of PP1 with spinophilin was not significantly different between brain regions (Fig. 3b), there was a more robust association of PP1 with CaMKII in striatum compared with hippocampus (Fig. 3c).
CaMKII directly and indirectly binds to two N-terminal regions and one C-terminal region in spinophilin (Baucum et al. 2012). Interestingly, GSTSpN1 (residues 1–154) (Fig. 3d) and GSTSpN2 (residues 151–300) (Fig. 3e), but not GSTSpC1 (residues 665–817) (Fig. 3f), bound to more striatal CaMKII than hippocampal CaMKII after normalizing for the lower striatal CaMKII levels in the low-ionic strength, Triton-soluble fraction.
Differential levels and interactions of proteins in the extrasynaptic and synaptic fractions
Glutamate receptor subunits and dendritic signaling/scaffolding proteins localize both synaptically and extrasynaptically, in some cases serving distinct functions (Petralia et al. 2010; Gladding and Raymond 2011). To better understand differences between striatum and hippocampus, we also compared protein levels in extrasynaptic (S2) and/or synaptic (S3) fractions isolated from the two brain regions (see 'Methods'). PSD-enriched proteins such as PSD-95, GluN2B, GluN2A, and spinophilin were substantially enriched in synaptic fractions from both striatum and hippocampus, but were also detected in extrasynaptic fractions (Fig. 4a, Figure S1). Levels of the canonical PSD marker, PSD-95, were modestly lower in the striatal extrasynaptic, but not striatal synaptic, fraction compared with the corresponding hippocampal fractions (Fig. 4b). GluA1 was expressed at lower levels in the extrasynaptic fraction isolated from striatum compared with hippocampus (Fig. 4c), with only weak detection in the synaptic fraction that could not be reliably quantified. GluN2A, GluN2B, and GluN1 levels were lower in both fractions isolated from striatum compared with hippocampus (Fig. 4d). Striatal levels of CaMKII were significantly lower in both the extrasynaptic and synaptic fractions compared with hippocampal fractions (Fig. 4e), whereas striatal levels of spinophilin were significantly lower only in the extrasynaptic fraction (Fig. 4f).
To investigate the subcellular localization of the spinophilin/CaMKII complex observed in striatum and hippocampus, we immunoprecipitated spinophilin from both synaptic and extrasynaptic fractions. There was a significantly more robust interaction between spinophilin and CaMKII in the extrasynaptic (S2) compared with synaptic (S3) fraction in both brain regions (Fig. 4g). Moreover, significantly more CaMKII co-precipitated with spinophilin in striatal compared with hippocampal extrasynaptic fractions, while there was no difference in the association of spinophilin and CaMKII between the synaptic fractions (Fig. 4g).
Enhanced striatal association of Myosin Va with CaMKII and spinophilin in the extrasynaptic fraction
Myosin Va interacts with CaMKII (Costa et al. 1999) as well as with N-terminal domain of spinophilin (Baucum et al. 2010). Myosin Va was detected in both subcellular fractions isolated from striatum or hippocampus, but was more enriched in the synaptic fraction (Figure S2). While there were similar levels of myosin Va in synaptic fractions isolated from striatum and hippocampus, myosin Va levels in striatal extrasynaptic fractions were significantly lower than in hippocampal extrasynaptic fractions (Fig. 5a). However, despite these lower levels, the interaction of myosin Va with both CaMKII (Fig. 5b) and spinophilin (Fig. 5c) was greater in striatal extrasynaptic, but not synaptic, fractions compared with the corresponding hippocampal fractions.
Enhanced association of GluN2B with spinophilin and CaMKII in the extrasynaptic fraction
Both myosin Va and CaMKII can associate with NMDARs (Strack and Colbran 1998; Husi et al. 2000; Bayer et al. 2006). Therefore, we compared the association of NMDARs with CaMKII in striatal and hippocampal subcellular fractions. The CaMKII-GluN2B complex was substantially enriched in synaptic fractions of both brain regions relative to the corresponding extrasynaptic fraction, consistent with an important role for GluN2B in subcellular targeting of CaMKII (Bayer et al. 2006). However, GluN2B-CaMKII association also was significantly more robust in the striatal extrasynaptic fraction compared with the hippocampal extrasynaptic fraction (Fig. 6a, Figure S3a and b). Notably, NMDAR GluN1 and GluN2A subunits were also more robustly associated with CaMKII in striatal extrasynaptic S2 fractions relative to the hippocampal S2 fraction, whereas only GluN2A was more robustly associated with CaMKII in the synaptic S3 fraction from striatum relative to hippocampus (Figure S3c–f).
Spinophilin plays a key role in PP1-dependent regulation of AMPAR and NMDAR currents as well as striatal LTD (Feng et al. 2000; Allen et al. 2006). However, to the best of our knowledge no one has evaluated the physical interactions between spinophilin and NMDARs ex vivo. We detected GluN2B in spinophilin immunoprecipitates. In contrast with the CaMKII-GluN2B complex (Fig. 6a), GluN2B association with spinophilin was enriched in the extrasynaptic fraction compared with the synaptic fraction (Fig. 6b). Moreover, GluN2B association with spinophilin also was more robust in striatal extrasynaptic fractions compared with hippocampal extrasynaptic fractions, whereas there was no significant difference between synaptic fractions from the two brain regions (Fig. 6b, Figure S3g and h).
There is a general consensus that long-term plasticity of excitatory synaptic transmission is important for different forms of learning and memory. Extensive studies in hippocampus have identified molecular mechanisms and multiple proteins that contribute to control of synaptic plasticity. The precise subcellular localization and regulation of proteins such as NMDARs, AMPARs, CaMKII, and PP1 are important for normal synaptic plasticity. While these proteins are expressed across many brain regions, it is becoming increasingly apparent that there can be substantial differences in types of plasticity that can be induced using similar stimulation paradigms (see Introduction). Here, we begin to explore biochemical mechanisms that may underlie such differences by demonstrating brain region-selective differences in the expression levels and interactions of key postsynaptic proteins in synaptic and extrasynaptic fractions. We suggest that such differences in postsynaptic signaling architecture contribute to the distinct physiological properties of hippocampal and striatal excitatory synapses.
Differential expression and distribution of excitatory synaptic proteins between brain regions
Currently available biochemical approaches do not allow for global analysis of protein levels and protein complexes in specific cell types. However, comparison of striatal and hippocampal homogenates represents an interesting model to begin to understand cell-type differences because ∼95% of striatal neurons are γ-aminobutyric acid-containing MSNs (Huang et al. 1992; Kreitzer and Malenka 2008), whereas glutamatergic pyramidal neurons predominate in hippocampus. Excitatory inputs are critical in both cell types, and were the focus of our study, although we cannot exclude possible minor contributions to our biochemical analyses from proteins localized to additional types of synapses and/or neurons. Thus, while our results do not identify cell specific differences, the results give insight into more global, systems level, brain region specific differences in protein expression and interactions.
The levels of synaptic proteins in whole extracts of brain regions may reflect the overall density of excitatory synapses, the levels of these proteins at individual synapses, or a combination of both factors. Terminals expressing vGlut2 in the striatum originate from thalamic neurons, but vGlut2 is expressed at low levels at hippocampal synapses. In contrast, vGlut1 is expressed in hippocampal pyramidal neurons and corticostriatal neurons (Bellocchio et al. 1998; Kaneko et al. 2002; Fremeau et al. 2004). Consistent with this, total levels of vGlut2 were > twofold higher in the striatum compared with the hippocampus (Fig. 1d). In contrast, total levels of vGlut1 were slightly lower in the striatum compared with the hippocampus (Fig. 1d).
The canonical postsynaptic density scaffolding protein PSD-95 localizes to both synaptic and extrasynaptic fractions (Petralia et al. 2010) and is critical for targeting AMPARs and NMDARs to modulate synaptic plasticity (Ehrlich and Malinow 2004; Kopec et al. 2007). While there was no significant difference in total lysate PSD-95 levels (Fig. 1), PSD-95 levels were significantly, if modestly, lower (∼31%) in the striatal extrasynaptic fraction compared with hippocampus, with a trend for slightly higher PSD-95 levels in the striatal synaptic fraction (Fig. 4b). A lack of a robust difference in synaptic PSD-95 levels may indicate that the overall density of excitatory synapses is similar in these two brain regions, although it is also possible that differences in the amount of PSD-95 at individual synapses may compensate for differences in synaptic density. Total levels of all glutamate receptor subunits tested were 30–80% lower in striatal lysates compared with hippocampal lysates (Fig. 1), and this was reflected in significantly lower levels of all subunits in both the extrasynaptic and synaptic fractions (Fig. 3). Lower striatal NMDAR subunit levels may contribute to the relative difficulty in inducing NMDAR-dependent LTP in the striatum compared with hippocampus (Calabresi et al. 1992). Also, differences in NMDAR expression may underlie the predominance of NMDAR-independent forms of LTD, such as endocannabinoid-mediated LTD, in striatum compared with hippocampus (Gerdeman et al. 2002; Lovinger 2010). While striatal levels of GluN2B were lower in both fractions, the difference was much more pronounced in the extrasynaptic fraction (∼69% lower striatal expression in the extrasynaptic vs. ∼30% lower striatal expression in the synaptic fraction).
Differential levels, distribution, and association of synaptic signaling proteins
Total levels of CaMKIIα, CaMKIIβ, and spinophilin were significantly (24–41%) lower in striatal compared with hippocampus lysates. In contrast, total levels of PP1 catalytic subunits are more than twofold higher in striatum compared with hippocampus (Fig. 2). Interestingly, CaMKII levels were similarly lower in striatal synaptic and extrasynaptic fractions compared with the corresponding hippocampal fractions, whereas the significantly lower levels of spinophilin are restricted to the extrasynaptic fraction (Fig. 4e). Although our fractionation paradigm does not distinguish between presynaptic and postsynaptic proteins, immunohistochemical, and electron microscopy studies suggest that spinophilin and CaMKII are highly enriched at postsynaptic sites (Ouimet et al. 1984; Muly et al. 2004). Together, these data suggest that the balance between postsynaptic CaMKII and PP1 signaling in synaptic and extrasynaptic fractions is quite different in these two brain regions.
To gain further insight into the nature of brain region selective differences in synaptic signaling, we also examined signaling protein complexes by co-immunoprecipitation. While this approach can fail to detect physiological protein-protein interactions or can detect interactions that are facilitated by tissue homogenization, depending on the extraction buffer and other conditions, we focused here on previously characterized interactions (see citations throughout). Initial studies using extracts solubilized in a low-ionic strength buffer containing Triton X-100 found that relative levels of CaMKIIα associated with spinophilin were almost twofold higher in striatum compared with hippocampus (Fig. 3a). However, there was no difference in the levels of PP1 catalytic subunit associated with spinophilin, despite the substantially higher levels of striatal PP1 (Fig. 3b). As predicted from these changes in the spinophilin complex, the amount of PP1 associated with CaMKII immune complexes was also about twofold higher in the striatum compared with the hippocampus (Fig. 3c). We then immunoprecipitated proteins from different subcellular fractions isolated in an isotonic KCl buffer (see 'Methods'). Interestingly, analysis of complexes isolated from the subcellular fractions revealed that there was a greater association of CaMKII and spinophilin in the extrasynaptic fraction compared with the synaptic fraction in both brain regions (Fig. 4g). Moreover, the enhanced striatal association of CaMKII is restricted to the extrasynaptic fraction (Fig. 4g). Unfortunately the low yields from isolated subcellular fractions precluded more complete analysis of these protein complexes for PP1 and other proteins.
The enhanced striatal association of CaMKII with spinophilin, and presumably PP1, in the extrasynaptic fraction may impact the dynamic control of phosphorylation of downstream targets of this complex, reminiscent of the coordinated targeting of protein kinase A and protein phosphatase 2B to AMPARs by AKAP79/150 (Logue and Scott 2010). However, it does not appear that the major CaMKIIα autophosphorylation site, Thr286, is a direct target for the PP1 that is associated with CaMKII via spinophilin because levels of Thr286 phosphorylation are significantly higher in the striatum than in the hippocampus (Fig. 2c). This is consistent with recent studies indicating that Thr286 may be protected from PP1-mediated dephosphorylation in PSDs (Mullasseril et al. 2007), perhaps because spinophilin can selectively inhibit PP1 activity toward certain substrates (Hsieh-Wilson et al. 1999; Terry-Lorenzo et al. 2000; Ragusa et al. 2010). Future studies will need to determine if differences in PP1 association with CaMKII modulate the phosphorylation of other sites on CaMKIIα or of other co-assembled proteins.
Mechanism of altered spinophilin/CaMKII interaction
CaMKII interacts both directly and indirectly with multiple domains in spinophilin (Baucum et al. 2012), making it difficult to uncover mechanism(s) underlying brain-region specific variations in the interaction. Although there was no difference in the ability of a C-terminal domain in spinophilin to bind CaMKII from hippocampal or striatal extracts (Fig. 3f), we found that N-terminal domains of spinophilin bound significantly more striatal CaMKII than hippocampal CaMKII (Fig. 3d and e). The increased interaction of striatal CaMKII with GSTSpN2 may be explained by higher levels of Thr286 phosphorylation (Fig. 2c), because autophosphorylation is required for a strong, direct interaction with this domain (Baucum et al. 2012). However, GSTSpN1 (containing residues 1–154) cannot directly bind purified CaMKII. Therefore, we speculate that binding of CaMKII in brain extracts to GSTSpN1 is mediated by other striatal proteins that can ‘bridge’ an interaction of CaMKII with GSTSpN1. One such candidate is myosin Va, which can bind to both CaMKII and GSTSpN1 (Costa et al. 1999; Baucum et al. 2010), and regulates GluA1 trafficking to dendritic spines, and LTP, in a CaMKII-dependent manner (Correia et al. 2008). Interestingly, we found that the levels of myosin Va associated with both CaMKII and spinophilin were significantly higher in extrasynaptic, but not synaptic, fractions isolated from striatum compared with hippocampus (Fig. 5b and c). Taken together, these data suggest that myosin Va may be involved in enhancing the interaction of CaMKII with spinophilin in the extrasynaptic fraction. However, additional intermediates may be involved. For instance, α-actinin can also bind both spinophilin and CaMKII (Walikonis et al. 2001; Baucum et al. 2010) and some isoforms are more highly expressed in striatum compared with hippocampus (Wyszynski et al. 1998). Future studies need to compare the broader composition of these complexes in the two brain regions.
Differential interactions of NMDAR GluN2B subunits with spinophilin and CaMKII
As a first step toward identifying physiological targets of the CaMKII-spinophilin complex, we investigated its association of NMDAR GluN2B subunits in extrasynaptic and synaptic fractions. CaMKII interaction with GluN2B is critical for proper synaptic targeting of CaMKII, hippocampal LTP, and memory consolidation (Strack and Colbran 1998; Barria and Malinow 2005; Bayer et al. 2006; Halt et al. 2012). Moreover, CaMKII also modulates NMDARs (Sessoms-Sikes et al. 2005; Gustin et al. 2011). Consistent with these findings, levels of GluN2B associated with CaMKII were higher in synaptic fractions compared with extrasynaptic fractions from both brain regions (Fig. 6a). Notably, while similar amounts of GluN2B associated with synaptic CaMKII in the two brain regions, the association of GluN2B with extrasynaptic CaMKII appears to be stronger in the striatum compared with hippocampus (Fig. 6 and Figure S3).
Biochemical studies have shown that the PDZ domain of spinophilin can associate with NMDAR GluN1 and GluN2 subunits in vitro (Kelker et al. 2007). Spinophilin has also been shown to facilitate PP1 regulation of NMDARs (Feng et al. 2000), but the association of native spinophilin with NMDARs in the brain has not been investigated. We demonstrate here for the first time that NMDAR GluN2B subunits are associated with spinophilin in the rodent brain. Similar to the association of CaMKII with spinophilin, the ratio of GluN2B to spinophilin within these complexes is substantially higher in the extrasynaptic fraction compared with the synaptic fraction in both striatum and hippocampus, despite the fact that spinophilin was relatively enriched in the total synaptic fraction (Figure S1). Moreover, more GluN2B was associated with extrasynaptic spinophilin in striatum compared with hippocampus (Fig. 6b and Figure S3g and h), correlating with the enhanced CaMKII association with striatal extrasynaptic spinophilin. The mechanism(s) underlying the association of NMDAR subunits with spinophilin requires further investigation.
Taken together, these data indicate that synaptic and extrasynaptic signaling complexes are differentially assembled in different cell types, perhaps contributing to cell-specific regulation, activity, and roles of synaptic compared with extrasynaptic NMDARs (Gladding and Raymond 2011; Kaufman et al. 2012; Papouin et al. 2012). Moreover, these differences may play a role in unique changes in striatal neurons in neurological disorders, such as Parkinson's and Huntington disease (HD). For example, the interaction of PP1 with spinophilin is increased following striatal dopamine depletion in a rat PD model (Brown et al. 2008). In addition, extrasynaptic GluN2B-containing NMDAR currents are selectively enhanced in cultured striatal MSNs isolated from a mouse model of HD compared with normal MSNs (Milnerwood et al. 2012). It will be interesting to investigate the impact of disease-related processes on the assembly of synaptic and extrasynaptic signaling complexes and their role in the resulting neurological deficits.
The authors thank Dr. Qin Wang (University of Alabama at Birmingham) and Drs. Brian Wadzinski, Gregg Stanwood, Danny Winder, and Brian Shonesy, and Ms. Emily Anderson (Vanderbilt University School of Medicine) for critical reading of the manuscript. The authors thank members of the Colbran Laboratory for helpful comments and suggestions. RJC was supported by NIH Grant R01-MH063232 and by a Hobbs Discovery Grant from the Vanderbilt-Kennedy Center. AJB was supported by Grant K01NS073700 from NINDS. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NINDS, the NIMH, or the NIH. The authors have no conflicts of interest to declare.