Prefrontal cortical functioning depends on dopaminergic neurotransmission, which in turn depends on a complex signal transduction pathway including protein phosphatase-1 (PP1). Targeted localization of PP1 by the scaffolding proteins, spinophilin and neurabin, is critical for dopaminergic modulation of glutamate neurotransmission. In this study, we report the preparation of an antiserum to neurabin, use it to study the subcellular localization of neurabin and compare that to our previous study of spinophilin, a closely related PP1 scaffold. Neurabin is found predominately in dendritic spines, but is also found in other compartments, including dendrites, axons, terminals and glia. This distribution contrasts with that of spinophilin in that neurabin is found in axon terminals where spinophilin is absent, and in parvalbumin-containing interneuron dendrites there is no significant neurabin though these dendrites contain substantial spinophilin. Within the dendritic spine compartment, however, the two proteins are similarly distributed. Both neurabin and spinophilin are concentrated in spines, and double-labeling reveals that they co-localize in most spines. Furthermore, post-embedding immunogold labeling demonstrates that within a spine, neurabin is distributed in the same pattern as spinophilin, concentrated in the postsynaptic density and the 100 nm just below. These results indicate that neurabin and spinophilin share important similarities and differences in their patterns of distribution. Varying patterns of scaffold localization may play an important role in determining the content and action of signal transduction pathways in different neuronal populations or compartments.
Optimal functioning of the primate prefrontal cortex (PFC) depends on the function of the D1 dopamine receptor (Sawaguchi and Goldman-Rakic, 1991; Williams and Goldman-Rakic, 1995; Müller et al., 1998), and its signal transduction pathway including protein phosphatase-1 (PP1). PP1 is important in effecting or modulating the actions of the D1 receptor on the AMPA (Yan et al., 1999; Snyder et al., 2000) and NMDA glutamate receptors (Snyder et al., 1998), the GABAA receptor (Flores-Hernández et al., 2000) and calcium currents (Surmeier et al., 1995). It is also necessary for one form of synaptic plasticity, long-term depression (LTD) (Mulkey et al., 1994). In the brain, PP1 is present in three different isoforms: PP1α, PP1β/δ and PP1γ1 (da Cruz e Silva et al., 1995). These isoforms differ in their regulation by protein kinase C (Takizawa et al., 1997), their expression patterns during cellular differentiation (Tognarini and Villa-Moruzzi, 1998), and their localization within the brain (Strack et al., 1999). As for other signal transduction proteins (Pawson and Scott, 1997), the localization of PP1 is tightly regulated by scaffolding proteins (Cohen, 2002). Appropriate localization of PP1 is required for proper modulation of glutamate gated currents (Westphal et al., 1999; Yan et al., 1999; Morishita et al., 2001), and the D1 receptor is specifically localized with respect to PP1 isoforms (Muly et al., 2001). Thus, the localization of PP1 may be important in defining the actions elicited by stimulation of the D1 or other receptors.
Two PP1 scaffolding proteins that have received particular scrutiny are spinophilin (Allen et al., 1997) and neurabin (Nakanishi et al., 1997). Both bind to PP1 and actin, and have a PDZ domain and coiled coil domains (Allen et al., 1997; Satoh et al., 1998; Hsieh-Wilson et al., 1999; Smith et al., 1999; Richman et al., 2001). Neurabin has a domain organization similar to that of spinophilin with the addition of a 278 amino acid region on the C-terminus (Nakanishi et al., 1997). In the shared regions, the two protein sequences have homology ranging from 43% to 86% in the different domains (Satoh et al., 1998). These proteins have been hypothesized to act as scaffolds which facilitate the interaction between PP1 and receptors, cytoskeleton and substrates, and as such, their presence or absence at different cellular sites may constrain the localization and/or action of PP1.
Both neurabin and spinophilin can be modulated by phosphorylation. The binding of PP1 to neurabin can be reduced by phosphorylation of serine 461 by PKA (McAvoy et al., 1999; Oliver et al., 2002). Spinophilin does not appear to be phosphorylated at or near its PP1 binding domain, rather it is phosphorylated by PKA at serine 94 and 177 in the actin binding domain, which causes it to dissociate from actin filaments (Hsieh-Wilson et al., 2003). These differences in phosphorylation patterns suggest that the two scaffold proteins participate in different functions in neurons, but this has been difficult to demonstrate. The peptides used to block PP1 binding to spinophilin in live cells (Yan et al., 1999) are also predicted to block PP1 binding to other PP1 regulatory subunits, including neurabin. Studies in spinophilin knock-out mice have shown disruption of PP1 modulation of AMPA- and NMDA-gated currents, as well as LTD (Feng et al., 2000), suggesting that neurabin cannot completely compensate for the lost spinophilin.
One way to understand the functional differences between spinophilin and neurabin may be through the analysis of their subcellular localization. Although the localization of spinophilin has been studied in detail in monkey PFC (Muly et al., 2004), there have been no studies of the subcellular localization of neurabin to date. In order to further characterize the roles that these two closely related scaffolding proteins play in neurotransmission, we produced and affinity purified an antiserum to neurabin to compare its subcellular localization with that previously reported for spinophilin in monkey PFC.
Materials and Methods
Neurabin residues 835–1095 were amplified by PCR from cDNA using the primers CAAGTCGACCATCTTTGAGAGAAGACCAT (+ strand) and GACGCGGCCGCTCACGGTTGCTCGGTTGTGGAAG (− strand). The PCR product was digested with SalI/NotI and ligated into the bacterial expression vector pGex-4T-2 (Pharmacia). This plasmid was transfected into bacterial strain BL21 (DE3) and protein expression induced with the addition of 0.1 mM IPTG. Bacteria were lysed by French press, and expressed protein (GST-neur-CT) was purified by affinity chromatograpy over glutathione-sepharose beads (Pharmacia). Protein was eluted with 20 mM reduced glutathione (Sigma, St Louis, MO) and dialyzed extensively against PBS. One guinea pig was immunized using 0.5 mg, with three boosts of 0.35 mg (Cocalico Biologicals, Reamstown, PA). Western blotting of total brain extracts identified positive bleeds. Antibodies were first affinity purified over a column of GST-neur-CT protein coupled to CH-sepharose (Pharmacia). Bound antibodies were eluted with 0.1 M glycine pH gradient 3.5–2.5. Fractions were neutralized by the addition of 1 M Tris base, and dialyzed extensively against PBS. Due to apparent cross reactivity of the preparation for spinophilin, a second neurabin GST fusion protein was prepared. Neurabin residues 921–1095 were amplified as above but with the + strand primer GGAGTCGACGAATTTCACCTTCAATGATGA, and purification of recombinant protein (GST-neur-SAM) followed the protocols above. Antibodies were re-purified over a second column consisting of GST-neur-SAM (921–1095), again coupled to activated sepharose. These antibodies were found to be specific for neurabin using Western blotting (see below) and immunofluorescence labeling of the hippocampus of wild type and neurabin knockout mice.
Tissue from two juvenile macaque monkeys, sacrificed as part of another study, was used for immunoblotting. The animals were sacrificed by pentobarbital overdose (100 mg/kg), the brains removed and blocks of various brain regions were frozen within ten min of the time of death. Samples of PFC were dounce homogenized in buffer containing 140 mM KCl, 10 mM glucose, 1.2 mM MgCl2, 10 mM HEPES, pH 7.4, with a cocktail of protease inhibitors added (1 mM phenylmethylsulfonyl fluoride, 10 mM benzamidine, 10 μg/ml aprotinin, 10 μg/ml leupeptin and 1 μg/ml pepstatin). After incubating for five min at room temperature, the homogenate was centrifuged at 500 g for 5 min at 4°C. The pellet was discarded as debris and the supernatant was assayed for protein concentration using a colorimetric assay (Bio-Rad Laboratories, Hercules, CA). A gel loading mixture was prepared with protein sample, water, sample buffer and reducing agent (Invitrogen Corp., Carlsbad, CA). The samples were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) on NuPage 4–12% Bis–Tris gels using MOPS buffer (Invitrogen). Each lane was loaded with 10 or 20 μg of protein sample and the gel was run for 50 min at 200 V. The gel was then transferred to PVDF membranes. The membranes were rinsed in 0.9% NaCl, 20 mM Tris, 0.2% Tween-20, pH 7.4 (TBS-T) and blocked in 5% non-fat powdered milk/0.25% fish gelatin in TBS-T. They were then probed with guinea pig anti-neurabin (used at 1:8000) and rabbit anti-spinophilin (Allen et al., 1997; used at 1:20 000) overnight at 4°C. The membranes were then rinsed three times in TBS-T and incubated in horseradish peroxidase (HRP)-conjugated secondary antisera for 90 min at room temperature (HRP–donkey anti-guinea pig IgG, 1:10 000, Jackson ImmunoResearch; HRP–goat anti-rabbit IgG, 1:10 000, Bio-Rad). The membranes were then rinsed with TBS-T and labeling was revealed by chemiluminescence using 1.25 mM luminol, 0.2 mM coumaric acid, 0.01% H2O2, 0.1 M Tris, pH 8.5 and Kodak Biomax film. The membranes were then stripped in 100 mM β-mercaptoethanol, 2% SDS, 62.5 mM Tris, pH 6.7 for 30 min at 50°C. Then they were reblocked and reprobed using mouse anti-tubulin βIII (used at 1:3000, Upstate) and HRP-goat anti-mouse IgG (used at 1:10 000, Bio-Rad) to confirm equal loading. A ladder of markers was used to estimate the molecular weight of the labeled bands (SeeBlue plus 2, Invitrogen).
Images of the Western blots were captured with a Spot RT color digital camera (Diagnostic Instruments, Inc., Sterling Heights, MI). Images in TIFF format were imported into an image processing program (Canvas 8, Deneba Software, Miami, FL). The contrast and brightness of the images was adjusted and labels were added. Pairs of lanes to be compared were adjusted in an identical fashion. The image was then printed on a Kodak Professional 8660 Thermal printer.
Animals and Preparation of Tissue for Immunohistochemisty
These experiments were performed on PFC (Walker's area 9) obtained from 9 young adult macaque monkeys. After deep anesthesia with an overdose of pentobarbital (100 mg/kg), the animals were perfused with a flush Tyrode's solution (137 mM NaCl, 2.7 mM KCl, 1 mM CaCl2, 0.5 mM MgCl2, 12 mM NaHCO3, 0.5 mM NaH2PO4, 5 mM glucose) in which 95% O2/5% CO2 was bubbled continuously, followed by 3 to 4 l of a fixative mixture of 4% paraformaldehyde/0.1–0.2% glutaraldehyde/0–0.2% picric acid in phosphate buffer (0.1 M, pH 7.4; PB). The brain was removed, blocked and post-fixed in 4% paraformaldehyde for 2–4 h. Coronal, 50 μm vibratome sections were cut and then stored frozen in 15% sucrose until immunohistochemical labeling was performed. The care of the animals and all anesthesia and euthanasia procedures in this study were performed according to the National Institutes for Health Guide for the Care and Use of Laboratory Animals, and were approved by the Institutional Animal Care and Use Committee of Emory University.
Pre-embedding Immunoperoxidase Labeling
Pre-embedding immunoperoxidase labeling with the guinea pig anti-neurabin antiserum described above (used at 1:4000, 200 ng/ml) was performed as described previously (Muly et al., 1998). Briefly, sections were thawed, incubated in blocking serum (3% normal goat serum, 1% bovine serum albumin, 0.1% glycine, 0.1% lysine in 0.01 M phosphate buffered saline, pH 7.4) for one h and then placed in the primary antiserum diluted in blocking serum. After 36 h at 4°C, the sections were removed from the primary antiserum, rinsed and placed in a 1:200 dilution of biotinylated donkey anti-guinea pig IgG (Jackson ImmunoResearch) for 1 h at room temperature. The sections were then rinsed, placed in ABC reagent for 1 h (Vector, Burlingame, CA) and processed to reveal peroxidase using 3,3′-diaminobenzidine (DAB) as the chromagen. Sections that were to be processed for electron microscopy were then osmicated, dehydrated and embedded in Durcupan resin (Electron Microscopy Sciences, Fort Washington, PA). Selected regions were then mounted on blocks and ultrathin sections were collected onto pioloform-coated slot grids and counterstained with uranyl acetate and lead citrate. Control sections, processed as above except for the omission of the primary immunoreagent, did not contain DAB label on electron microscopic examination.
Images of immunolabeled material were captured on a Leica DMRBE microscope using a Spot RT color digital camera (Diagnostic Instruments, Inc., Sterling Heights, MI). Images in TIFF format were imported into an image processing program (Canvas 8, Deneba Software, Miami). The contrast and brightness of the images was adjusted and labels were added. The images were then printed on a Kodak Professional 8660 Thermal printer.
In order to look for the presence of neurabin in cortical interneurons, or spinophilin-containing spines, double label experiments were performed. A pre-embedding immunogold/DAB protocol was used in which immunogold was used to label parvalbumin (PV) or spinophilin. Subsequently, neurabin was labeled with DAB. The methods used have been described previously (Muly et al., 1998). Briefly, sections were incubated overnight in a cocktail of primary immunoreagents (guinea pig anti-neurabin, 1:4000; and either mouse anti-PV, 1:10 000, Sigma; or rabbit anti-spinophilin, 1:10 000), and then incubated for 1 h in a cocktail of secondary antisera (biotinylated donkey anti-guinea pig, 1:200, Jackson ImmunoResearch; and either 1 nm gold-conjugated goat anti-mouse or 1 nm gold-conjugated goat anti-rabbit IgG, both used at 1:200 and obtained from Nanoprobes). The sections were then silver-intensified and gold toned. Following this, the sections were incubated in ABC reagent and reacted with DAB. All material from double label experiments was processed for electron microscopy as described above. Control sections, in which one of the two primary immunoreagents was omitted, showed no evidence either for nonspecific deposition of gold particles or for nonspecific deposition of DAB onto previously developed gold particles.
Post-embedding Immunogold Labeling
Tissue from the animals sacrificed above was prepared by high pressure freezing and freeze substitution for post-embedding labeling as described previously (Kieval et al., 2001; Muly et al., 2004). Briefly, 100 μm vibratome sections were cut and placed in an ascending series of cryoprotectant (final concentrations, 10% glycerol, 25% sucrose in 0.05 M phosphate buffer pH 7.4). Pieces of PFC areas taken from these sections were frozen in a high pressure freezer (Balzers, HPM-010) and then processed by freeze substitution in 0.5% uranyl acetate in methanol (Bal-Tec TTP010), infiltrated with Lowicryl HM-20 resin and polymerized with ultraviolet light at −40°C in a low temperature polymerization unit (Bal-Tec LTPU 010). Blocks were prepared and 60 nm thick sections were cut and collected on pioloform-coated 400 mesh gold grids.
For immunolabeling, all solutions except the antisera were filtered through a 0.2 μm syringe filter. Both primary and secondary antisera were diluted in solutions that had been syringe filtered. The grids were etched for 2 min with 0.1% Triton X-100 in 0.05 M Tris buffer (pH 7.6, TB) and treated with 0.1% sodium borohydride, 0.05 M glycine in TB for 10 min. The grids were rinsed in TB and blocked in a mixture of 1% normal goat serum and 2% human serum albumin (PE-BS) for 30 min. The grids were then incubated overnight at 4°C in the primary antiserum (1:400 guinea pig anti-neurabin) in PE-BS. After rinsing, the grids were incubated in goat anti-guinea pig IgG conjugated to 10 nm gold particles (1:100, produced by BBI, sold by Ted Pella, Inc., Redding, CA) with 5 mg/ml polyethylene glycol in PE-BS for 90 min. The grids were then rinsed and stained with uranyl acetate and lead citrate before being examined at the electron microscope (Zeiss EM10C).
Analysis of Material
The DAB-labeled material was analyzed as previously described (Muly et al., 2004). Blocks of tissue from cortical layers I, III and V of area 9 were made. Ultrathin sections from these blocks were examined using a Zeiss EM10C electron microscope. Regions of the grids containing neuropil were selected based on the presence of label and adequate ultrastructural preservation. We randomly selected fields of immunoreactive elements in the neuropil and images were collected at a magnification of 31 500 using a Dualvision cooled CCD camera (1300 × 1030 pixels), and Digital Micrograph software (version 3.6.5, Gatan, Inc., Pleasanton, CA). Images were collected from one block from each layer in three animals. A total of 253 micrographs representing 1545 μm2 were taken. On each micrograph, DAB-labeled profiles were identified and classified as spines, dendrites, terminals, axons, glia or unknown based on ultrastructural criteria (Peters et al., 1991). Profiles were identified as spines based on size (0.3–1.5 μm in diameter), presence of spine apparatus, absence of mitochondria or microtubules and, in some cases, the presence of asymmetric synaptic contacts. Dendrites were identified by their larger size (0.5 μm or greater in diameter), presence of microtubules, mitochondria and, in some cases, synaptic contacts. Axon terminals were characterized by the presence of numerous vesicles, mitochondria and occasionally a pre-synaptic specialization. Pre-terminal, unmyelinated axons were identified by their small size (0.1–0.3 μm in diameter), regular round shape and occasional presence of synaptic vesicles or neurofilaments. Glial profiles were identified based on their unusual shape, which appears to fill in the space between other, nearby profiles and a relatively clear cytoplasm which occasionally contained numerous filaments. Profiles that could not be clearly characterized based on these criteria were considered unknown profiles. The number of immunoreactive profiles was tabulated and the distributions in different layers compared with a Chi-square analysis.
Analysis of the double-labeled material was performed on blocks from layer III of area 9. We examined ultrathin sections from the surface of each block where both immunoperoxidase label for neurabin and immunogold label for either PV or spinophilin were visible. Electron micrographs of immunogold-containing profiles (immunoreactive for PV or spinophilin) were taken and these profiles were then examined for the presence of immunoperoxidase label (neurabin immunoreactivity).
The analysis of the post-embedding immunogold labeling was based on the analysis previously introduced and reported for spinophilin (Muly et al., 2004). Each trial of post-embedding labeling was examined and axospinous synapses were identified from one to three mesh squares. A digital image was made of each axospinous synapse that contained gold label in either the spine or presynaptic terminal. The image was imported into a computer graphics program (Canvas, Deneba Software, Miami) and the area of the spine and terminal profile on each micrograph was measured. The density of gold particles over the spines and terminals and the ratio of spine to terminal gold density were then calculated. For this study, a total of 48 grids, each considered a separate experiment, were analyzed. Most grids showed a spine to terminal gold label ratio of <5; however, there was a select group of eight grids which had ratios of >5. We further evaluated the localization of gold label in this group of grids with the best spine to terminal ratio. In these grids, the postsynaptic density (PSD) and spine apparatus were identified and the density of gold label in each compartment was determined. Finally, the spine was divided into bins based on distance from the PSD. The first bin was the PSD itself, the second bin was defined as the area between 1 and 50 nm deep to the PSD, and subsequent bins were additional 50 nm increments in from the PSD. For each bin, the total area, number of gold particles and density of gold label were determined.
Neurabin Antiserum Production and Characterization
Antibodies were raised in guinea pig against neurabin residues 835–1095. However, the affinity-purified preparation exhibited significant cross reactivity to spinophilin as judged by western blot analysis (data not shown). Although the primary sequence of neurabin 835–1095 is not similar to spinophilin, the region does contain a stretch of predicted coiled-coil structure. The C-terminus of spinophilin is similarly predicted to possess coiled-coil tertiary structure, and this may have accounted for the cross-reactivity. Antibodies were therefore re-purified over a second affinity column consisting of a bacterial fusion protein encoding the neurabin SAM domain (921–1095). These antibodies were found to be specific for neurabin by immunoprecipitation (data not shown), in stained tissue sections from wild type (Fig. 1B1) and neurabin knock-out mice (Fig. 1B2) and Western blotting (Fig. 1A).
In Western blots of macaque PFC, the antiserum to neurabin recognized a single band which migrated at an apparent molecular weight of 135 kDa, compared with the spinophilin antiserum, which recognized a single band that migrated at 97 kDa (Fig. 1A). The apparent molecular weights observed using NuPAGE gels and MOPS buffer are smaller than those observed using conventional Tris-glycine SDS–PAGE (∼180 kDa for neurabin and ∼135 for spinophilin, data not shown). Interestingly, the apparent molecular weights in the NuPAGE system are closer to those predicted for the two proteins based on their primary sequence (122.7 kDa for neurabin and 89.5 kDa for spinophilin). The altered migration patterns in the NuPAGE system likely involve an increased efficiency of SDS binding to spinophilin and neurabin. The intensity of the neurabin band appeared significantly weaker than the spinophilin labeling, corresponding to our impression from immunohistochemisty (see below).
Light Microscopic Distribution of Neurabin
Neurabin immunolabeling in macaque PFC was similar to that observed for spinophilin. The neuropil was labeled in all cortical layers and appeared slightly denser in superficial layer V and layer III (Fig. 1C). At higher magnification, this neuropil labeling was made up of small punctate elements (Fig. 1D). In deep layer VI, somatodendritic labeling of some cells is seen (Fig. 1D). These cells appear to be pyramidal neurons based on soma shape and the limited dendritic staining observed. Within the subcortical white matter, small numbers of labeled cells are observed. Overall, the cortical label was qualitatively lighter than that observed for spinophilin previously (Muly et al., 2004) and in sections run side by side.
Subcellular Distribution of Neurabin
We used immunoperoxidase electron microscopy to determine the subcellular distribution of the neuropil labeling. Dendritic spines were the most commonly encountered, and most densely labeled profiles observed in the PFC (Fig. 2A). Although less frequent, neurabin labeling was also observed in dendritic shafts that varied in size from large proximal dendrites to small distal dendrites (Fig. 2B). Labeled dendritic profiles occasionally received asymmetric synaptic contacts (Fig. 2C). In addition to dendritic structures, axon terminals, some of which formed asymmetric synapses, pre-terminal unmyelinated axons and glia were also labeled (Fig. 2D–G).
In order to facilitate comparison of the localization of neurabin with that of spinophilin, we quantified the relative distribution of neurabin-immunoreactive (IR) profiles in a total of 734 randomly collected immunoreactive neuropil elements on 253 images. These images were collected from three different cortical layers, layers I, III and V, in three monkeys. The distribution of labeled elements did not vary significantly between the three layers (χ2 = 15.59, P = 0.112) and, therefore, data from the different layers was summed and the distribution of labeled elements plotted in a histogram (Fig. 3). This distribution confirmed our qualitative impression of the material, that neurabin is overwhelmingly, but not exclusively, present in dendritic spines. Smaller numbers of labeled dendrites, terminals, axons and glia were observed. We compared the distribution of neurabin labeled profiles with that previously reported for spinophilin labeled profiles (Muly et al., 2004). The distribution of neurabin and spinophilin in the neuropil of PFC was significantly different (Fig. 3; χ2 = 114.79, P < 0.0001). Post hoc testing demonstrated that the difference in percentage of labeled profiles between the two proteins was significant for dendrites, terminals and axons. That is, a significantly smaller fraction of neurabin labeling was found in dendrites than for spinophilin, and a significantly larger fraction of neurabin labeling was found in terminals and axons than for spinophilin.
Co-localization of Neurabin and Parvalbumin
A previous finding from our own laboratory demonstrated that spinophilin is found, in part, in cortical interneurons (Muly et al., 2004). When parvalbumin (PV) labeling was used to identify a population of cortical interneurons, ∼20% of PV-IR dendritic profiles were also labeled for spinophilin in monkey PFC. In order to determine if the reduction in the percentage of dendritic labeling for neurabin compared with spinophilin represented a generalized decrease in dendritic localization or was selective for particular cell types, we performed a double-labeling experiment for PV and neurabin using pre-embedding immunogold to label PV, and immunoperoxidase staining to label neurabin (Fig. 4A). We examined 147 gold-labeled PV-IR dendrites in single ultrathin sections. Neurabin immunoperoxidase labeling was identified in only one of these profiles (0.7%). These results suggest that the relative decrease in the dendritic localization of neurabin is disproportionately the result of a total or near total absence of neurabin in the dendrites of at least one class of interneuron.
Co-localization of Neurabin and Spinophilin
Dendritic spines are the principal sites where both neurabin and spinophilin are located. This suggests that neurabin and spinophilin might be co-localized in individual spines. In order to test this, we performed a double-label immunoelectron microscopy experiment using pre-embedding immunogold to label spinophilin, and immunoperoxidase staining to label neurabin (Fig. 4B). We examined 164 gold-labeled spinophilin-IR spines in single ultrathin sections. Neurabin immunoperoxidase labeling was identified in 126 of these (76.8%). These results demonstrate that neurabin and spinophilin are co-localized in the majority of dendritic spines that contain either of these proteins. Since immunoperoxidase labeling is more sensitive and penetrates tissue sections better than pre-embedding immunogold labeling, these results further suggest that some spinophilin-IR spines do not contain neurabin. Whether or not some neurabin-IR spines lack spinophilin cannot be inferred from these data.
Intraspinous Localization of Neurabin
Finally, because neurabin and spinophilin are both concentrated in spines and are in many cases co-localized in the same spines, we studied the patterns of intraspinous localization of neurabin using post-embedding immunogold to compare with our recent findings on the distribution of spinophilin. As previously described for spinophilin, we used the pattern of labeling seen with the pre-embedding immunoperoxidase method to guide the analysis of the post-embedding immunogold material. For each grid stained with post-embedding gold, we imaged labeled axospinous synapses and determined the density of gold particles over spines (where most immunoperoxidase label was seen) and over axon terminals (where immunoperoxidase label was more rarely identified, see Fig. 3). The ratio of gold density in spines to that in terminals (STR) was calculated for each grid and this ratio was used to evaluate each grid for the quality of its labeling. A total of 49 grids were examined and, based on the STR, we identified two groups: 41 grids with a ratio of <5 (mean ± SE = 2.21 ± 0.20; range = 0.06–4.91) and eight grids with a ratio of >5 (10.52 ± 1.80; range = 5.56–18.1). By restricting our analysis to those grids with the STRs that were highest, and most closely approximated the ratio of label we saw in our DAB labeled material, we sought to avoid the problem of noisy label on some grids obscuring any differences in immunoreactivity distribution.
From these high STR grids, a total of 102 labeled axospinous synapses were identified. When this sample was taken as a whole, the density of gold label in the spines was 7.57 per μm2, compared with 0.98 gold particles per μm2 in the axon terminals, for a STR of 7.72. This ratio is reasonably close to the 6.61 ratio of labeled spines to terminals obtained from examination of the immunoperoxidase labeled material (see Fig. 3). Of these 102 axospinous synapses, 18 contained gold label only over the terminal; the other 84 had gold over the spine and in some cases the terminal. These 84 labeled spines were analyzed further. The gold particles were sometimes found in the post-synaptic density (PSD; Fig. 5B) but were more typically seen within the spine associated with electron-dense filamentous material (Fig. 5). Quantitative analyses were undertaken to determine in which spine compartments neurabin was localized. The gold-labeled spines were divided into three compartments: the PSD, the spine apparatus, and the rest of the spinoplasm according to criteria defined in our previous study (Muly et al., 2004). The densities of label in the total spines and each of the compartments is illustrated in Figure 6. The density of neurabin gold label in the spine apparatus was almost 40% lower than that seen in the spine as a whole. The PSD, on the other hand, showed an almost 90% increase in gold label density compared with the spine as a whole, suggesting a relative concentration of neurabin in the PSD of spines. The label density in the remainder of the spinoplasm was not substantially different from that measured in the spine as a whole. This analysis demonstrates that neurabin, like spinophilin, is relatively concentrated in the PSD and further suggests that neurabin may be relatively excluded from the spine apparatus.
Though the density of neurabin labeling is highest in the PSD, the absolute number of gold particles in the remainder of the spinoplasm is larger (26 in PSD, 104 in spinoplasm, 3 in spine apparatus). We have previously shown that spinophilin labeling in the spinoplasm is mainly found within 100 nm of the PSD. To compare neurabin localization with that of spinophilin, we performed a similar analysis for neurabin in which each spine was divided into ‘bins’ based on distance from the PSD. The first bin was the PSD itself, and the second and subsequent bins were defined as the region 50 nm deeper in the spine from the last bin. For the 84 spines containing label, the number of gold particles in each bin was determined, as was the total area in each bin, and the density of gold label for neurabin was calculated for each bin (Fig. 7). The density of neurabin label was highest in the PSD and then dropped by ∼20% for the 100 nm immediately subjacent to it. The density of neurabin label fell more sharply (by over 40%) further than 100 nm from the PSD. No neurabin label was found more than 350 nm from the PSD. This analysis demonstrates that neurabin is not homogeneously distributed within the spine, but like for spinophilin, the density of labeling is highest within 100 nm of the PSD.
This study is the first ultrastructural analysis that contrasts the localization of neurabin with that of the closely related PP1-binding protein, spinophilin, in the primate PFC. Neurabin is found predominately in dendritic spines, but is also located in dendrites, axons, axon terminals and glia. Double-labeling experiments demonstrate substantial co-localization of neurabin and spinophilin in dendritic spines, but, in contrast to spinophilin, reveal no significant neurabin localization in PV-containing interneurons. Post-embedding immunogold labeling of neurabin within dendritic spines revealed a pattern similar to that previously reported for spinophilin: one nanodomain consisting of the PSD and the subjacent 100 nm of spinoplasm, contains a high concentration of neurabin immunoreactivity; and a second consisting of the remainder of the spine, containing a lower concentration of neurabin immunoreactivity.
Neurabin and Spinophilin within Dendritic Spines
The results presented here demonstrate that neurabin, like spinophilin, is concentrated within dendritic spines. In fact, the two proteins co-localize in spines to a large extent; most spinophilin-IR spines are also labeled for neurabin. While the presence of single-labeled spinophilin-IR spines in our material suggests that some spines may contain only one of the two proteins, the limitations of the pre-embedding immunogold/immunoperoxidase method and of single section analysis preclude any definitive conclusion about whether other populations of spines that contain only spinophilin or neurabin exist. Based on the current results and our previous study of spinophilin, we suggest that the majority of pyramidal cell spines in monkey PFC contain both PP1-binding proteins.
The post-embedding immunogold labeling revealed that neurabin's intraspinous distribution is the same as that found for spinophilin. Specifically, both proteins are distributed in two nanodomains within the spine. High levels of neurabin and spinophilin immunoreactivity are observed in the PSD and the 100 nm subjacent to it, while lower levels are found in the remainder of the spine. As in our previous report on spinophilin, the data presented here suggest that the concentration of neurabin falls with increasing distance from the synapse to reach zero at ∼400 nm from the synapse, but the small area sampled that far from the PSD makes such a conclusion tentative.
While neurabin and spinophilin overlap to a great extent in and within spines, they appear to differ in their absolute amount per spine. MacMillan and colleagues (1999) purified PP1 binding proteins from rat forebrain and found that the molar ratio of spinophilin to neurabin was 7 to 1. The Western blots of monkey PFC reported here are consistent with these findings. A difference in the amount of these two proteins in monkey PFC is also supported by the qualitative differences in the relative intensity of immunoperoxidase staining for the two, and the overall reduced gold labeling signal observed in post-embedding experiments for neurabin compared with spinophilin. The lower level of neurabin in dendrites of interneurons likely underlies part of the difference in levels between these two PP1 binding proteins. However, the bulk of the immunoreactivity for both proteins is found in spines and the large difference in protein abundance must also be reflected in this compartment.
It appears then, that the majority of the spines of pyramidal cells in PFC contain spinophilin and a much smaller amount of neurabin and that the two proteins are distributed similarly within the spine. An important question for future study is the functional significance of having small amounts of neurabin mixed in with spinophilin in pyramidal cell spines. To date, the identified functional differences between the two scaffolding proteins relate to how phosphorylation modulates their function (McAvoy et al., 1999; Oliver et al., 2002; Hsieh-Wilson et al., 2003). Furthermore, spinophilin has been shown to interact directly with certain G-protein coupled receptors (Smith et al., 1999; Richman et al., 2001; Brady et al., 2003); such an interaction has yet to be established for neurabin. It does appear that neurabin is unable to compensate for the loss of spinophilin in a constitutive knock-out model (Feng et al., 2000), suggesting that they likely play unique roles in neuronal communication.
Finally, given the overlapping distribution of neurabin and spinophilin within spines, it is possible that the two proteins interact directly. Both proteins have C-terminal coiled-coil domains thought to mediate dimerization or oligomerization (Allen et al., 1997; Nakanishi et al., 1997). Heteromers between neurabin and spinophilin have been observed in immunoprecipitation experiments of rat brain extracts (Colbran et al., 1997; Oliver et al., 2002; Terry-Lorenzo et al., 2002). The extent of heteromerization was variable, and there were some differences observed between rat and mouse brain extracts (Terry-Lorenzo et al., 2002). The localization data presented here are consistent with possible heteromerization; however, the extraction used in the immunoprecipitation experiments involved detergent conditions that may have disrupted the original complexes. While these experiments show that neurabin and spinophilin have the capability to form heteromers, it remains to be seen if these exist in vivo.
Comparison of the Localization of Neurabin and Spinophilin in other Compartments
The pattern of localization reported here for neurabin shares many similarities with, and some important differences from that previously reported for spinophilin (Muly et al., 2004), which provides evidence for the common and individual functions of these PP1 scaffolding proteins. Within the neuropil, both proteins are found primarily in dendritic spines. However there are significant differences in their localization to other neuronal compartments. Neurabin labeling is less frequently seen in dendritic shafts and more common in pre-terminal axons and axon terminals, which do not contain significant spinophilin immunoreactivity. The relative reduction of neurabin labeling of dendrites is not uniform but affects some cell types preferentially. In particular, dendritic profiles of PV-containing interneurons, 20% of which co-localize with spinophilin, do not contain significant neurabin immunoreactivity. Spines, on the other hand, are the most commonly labeled compartment for both neurabin and spinophilin, and there is extensive co-localization of these two proteins in individual spines. Thus, different cortical cell types and different neuronal compartments share similarities but also display important differences in their content of PP1-binding proteins. While the significance of this finding depends on a more complete understanding of the localization of PP1 in the PFC, as well as of the functional differences between neurabin and spinophilin, these results provide evidence for site-specific differences in signal transduction protein targeting in the brain.
The authors gratefully acknowledge the excellent technical assistance of Marcelia Maddox and the comments of Yoland Smith on an earlier version of this manuscript. This work was supported by MH01994 (E.C.M.), MH40899 (P.G.), DA10044 (P.G.), RR00165 and the Michael Stern Parkinson's Research Foundation.
1Department of Psychiatry and Behavioral Sciences, Emory University, Atlanta, GA 30329, USA, 2Division of Neuroscience, Yerkes National Primate Research Center, Atlanta, GA 30329, USA, 3Department of Psychiatry, Yale University School of Medicine, New Haven, CT 06508, USA and 4Laboratory of Molecular and Cellular Neuroscience, The Rockefeller University, New York, NY 10021, USA