Morphine Administration Alters the Profile of Hippocampal Postsynaptic Density-associated Proteins

Numerous studies have shown that drugs of abuse induce changes in protein expression in the brain that are thought to play a role in synaptic plasticity. Drug-induced plasticity can be mediated by changes at the synapse and more specifically at the postsynaptic density (PSD), which receives and transduces synaptic information. To date, the majority of studies examining synaptic protein profiles have focused on identifying the synaptic proteome. Only a handful of studies have examined the changes in synaptic profile by drug administration. We applied a quantitative proteomics analysis technique with the cleavable ICAT reagent to quantitate relative changes in protein levels of the hippocampal PSD in response to morphine administration. We identified a total of 102 proteins in the mouse hippocampal PSD. The majority of these were signaling, trafficking, and cytoskeletal proteins involved in synaptic plasticity, learning, and memory. Among the proteins whose levels were found to be altered by morphine administration, clathrin levels were increased to the largest extent. Immunoblotting and electron microscopy studies showed that this increase was localized to the PSD. Morphine treatment was also found to lead to a local increase in two other components of the endocytic machinery, dynamin and AP-2, suggesting a critical involvement of the endocytic machinery in the modulatory effects of morphine. Because α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors are thought to undergo clathrin-mediated endocytosis, we examined the effect of morphine administration on the association of the AMPA receptor subunit, GluR1, with clathrin. We found a substantial decrease in the levels of GluR1 associated with clathrin. Taken together, these results suggest that, by causing a redistribution of endocytic proteins at the synapse, morphine modulates synaptic plasticity at hippocampal glutamatergic synapses.

Opiates are choice analgesics in the treatment of chronic pain. However, repeated opiate administration can lead to the development of tolerance, physical dependence, and addiction. Opiate addiction is thought to involve the brain reward circuit as well as brain regions involved in learning and memory, such as the hippocampus (1,2). There is accumulating evidence that opiates modulate synaptic transmission and plasticity in the hippocampus. For example, opiates have been shown to significantly alter glutamatergic transmission (3), neurogenesis (4), dendritic stability (5), and long term potentiation (6 -8). To date, however, the protein substrates involved in opiate-induced synaptic plasticity in the hippocampus are not well explored.
Recent studies have applied genomics and proteomics approaches to identify changes in gene and protein expression in the brain that may be involved in opiate addiction (9 -12). A number of reports have shown that repeated morphine administration alters the expression of proteins involved in receptor endocytosis, neurotransmission, energy metabolism, and protein degradation (11,(13)(14)(15). These effects could contribute to the morphine-induced neuronal changes that persist for a long time following the cessation of drug exposure (16). However, molecular and cellular mechanisms underlying these long lasting changes are still not fully understood.
Because of the complexity of the central nervous system, fractionation techniques that allow the enrichment of lower abundance proteins are usually necessary to achieve a better characterization of the cellular proteome. Within the synapse, the postsynaptic site contains a high concentration of proteins, including receptors and their intracellular signaling components that receive and transduce synaptic information. This postsynaptic electron-dense structure, named the postsynaptic density (PSD), 1 plays an important role in synaptic regulation and plasticity. Although considerable efforts have been put toward identifying the protein components of this fraction (17)(18)(19)(20), none have focused on examining changes in PSD proteins in the hippocampus following drug administration.
In this study, we examined the changes in expression levels of PSD-associated proteins in the mouse hippocampus upon repeated administration of morphine using quantitative proteomics. To measure the relative changes in the levels of proteins, we used ICAT followed by tandem mass spectrometric (LC-MS/MS) analysis. We found that the levels of 10 proteins were altered in both forward and reverse labeling experiments. Next we focused on clathrin heavy chain because it showed the largest increase. Interestingly this increase was found to be local at the PSD, and this was confirmed by Western blotting and electron microscopy. We also found that levels of adaptor protein-2 ␣1 (AP-2) and dynamin were increased locally at the PSD after morphine treatment. Because the clathrin⅐dynamin⅐AP-2 complex modulates the level of cell surface receptors including ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) at glutamatergic synapses (21), we investigated the effect of morphine on the association of the AMPA receptor subunit, GluR1, with clathrin. Morphine treatment led to a decrease in the levels of clathrin-associated GluR1, suggesting that this treatment may alter clathrin-mediated AMPA receptor trafficking. Altogether these studies propose a role for morphine in regulating endocytic proteins at the hippocampal PSD, which in turn would modulate synaptic transmission and plasticity at hippocampal synapses.

EXPERIMENTAL PROCEDURES
Animals and Morphine Treatment-Experimental protocols involving animals were approved by the Institutional Animal Care and Use Committee at Mount Sinai School of Medicine. Four-month-old male wild-type C57BL/6JBom mice (20 -25 g) were maintained on a 12-h light/dark cycle and were allowed to acclimatize to their environment for 1 week prior to drug administration. Morphine sulfate (Sigma) was prepared in 0.9% sterile isotonic saline. For each experiment, mice (n ϭ 5 per group) were injected intraperitoneally with saline or morphine; in the latter case four escalating doses of morphine (5,8,10, and 15 mg/kg) every 12 h for 48 h as described previously (22) were used. Animals were sacrificed 12 h following the last drug injection. The hippocampi from treated or control mice were dissected and stored at Ϫ80°C until use.
Subcellular Fractionation and PSD Isolation-The PSD fraction was isolated essentially as described previously (20). For each fractionation, hippocampi from five saline-treated or five morphine-treated mice were combined and homogenized in 3 ml of 0.32 M sucrose, 0.1 mM CaCl 2 containing protease (0.1 mM PMSF, 1 g/ml aprotinin, 1 g/ml leupeptin, 1 mM benzamidine, and 0.1 mM pepstatin) and phosphatase inhibitors (1 mM NaF and 1 mM ␤-glycerophosphate) from Sigma. The homogenate was brought to a final concentration of 1.25 M sucrose by adding 2 M sucrose and 0.1 mM CaCl 2 . The homogenate was then placed in an ultracentrifuge tube, overlaid with 1 M sucrose, and subjected to centrifugation at 100,000 ϫ g for 3 h at 4°C. The synaptosomal fraction was collected at the 1.25 M/1 M sucrose interface. To obtain the synaptic junctions, the synaptosomal fraction was diluted with 20 mM Tris-Cl, pH 6, 0.1 mM CaCl 2 containing 1% Triton X-100 (TX-100), mixed for 20 min at 4°C, and centrifuged at 40,000 ϫ g for 20 min at 4°C. The pellet containing the isolated synaptic junctions was collected. To separate presynaptic proteins from the PSD, the pellet was resuspended in 20 mM Tris-Cl, pH 8, 1% TX-100, 0.1 mM CaCl 2 . The mixture was again mixed for 20 min at 4°C and centrifuged at 40,000 ϫ g for 20 min at 4°C. The insoluble pellet containing the PSD fraction was collected and stored at Ϫ80°C until use.
ICAT Labeling and Peptide Separation-ICAT labeling was performed according to the manufacturer's guidelines (Applied Biosystems, Foster City, CA). In brief, PSD fractions were solubilized in 0.1% SDS, 200 mM Tris-Cl, pH 8.3. To reduce disulfide bonds, 5 mM tris(2-carboxyethyl)phosphine was added, and samples were boiled for 10 min. After cooling the samples to room temperature, samples from saline-treated animals were labeled with the light ( 12 C) ICAT reagent and from morphine-treated animals were labeled with the heavy ( 13 C) reagent for 2 h at 37°C (forward labeling). To avoid variability during the labeling procedure, reverse labeling was also performed with a separate set of PSD fractions obtained from an independent experiment. In this case, saline samples were labeled with the heavy ( 13 C) reagent, and morphine samples were labeled with the light ( 12 C) reagent. Following incubation with the ICAT reagent, saline and morphine samples were combined and digested overnight with trypsin at 37°C. Resulting peptides were separated by cationexchange chromatography using a POROS 50 HS (4.0 ϫ 15-mm) cation-exchange column. Peptides were eluted with 5 mM KH 2 PO 4 , pH 3, 25% acetonitrile (SCX A buffer) using a step gradient (from 40 to 600 mM KCl in SCX A buffer) to collect six fractions. The labeled peptides from each fraction were isolated using an avidin column (4.0 ϫ 15 mm) according to the manufacturer's instructions. After evaporation of the samples, the biotin tag was removed from the ICAT-labeled peptides by incubation with a cleaving reagent (provided by the manufacturer) for 2 h at 37°C, and resulting peptides were dried and stored at Ϫ80°C until use.
Liquid Chromatography and Mass Spectrometry-For peptide identification by MS/MS, the peptides were dissolved in 30 -50 l of HPLC sample solvents containing water:methanol:acetic acid:trifluoroacetic acid (70:30:0.5:0.01, v/v/v/v). Micro-HPLC-MS/MS analysis was conducted on a QSTAR XL mass spectrometer (Applied Biosystems) coupled with an on-line MicroPro-HPLC system (Eldex Laboratories, Napa, CA). 30 -50 l of tryptic peptide solution were injected into a Magic C 18 column (0.2 ϫ 150 mm, 5 m, 200 Å, Michrom BioResources, Auburn, CA) that had previously been equilibrated with 70% solvent A (0.5% acetic acid and 0.01% trifluoroacetic acid in water:methanol (95:5, v/v) and 30% solvent B (0.5% acetic acid and 0.01% trifluoroacetic acid in methanol:water (95:5, v/v)). Peptides were separated and eluted from the HPLC column with a linear gradient of 30 -70% solvent B in 100 min at a flow rate of 2.0 l/min. The eluted peptides were sprayed directly into the QSTAR XL mass spectrometer (2.9 kV). The QSTAR XL mass spectrometer was operated in a positive ion mode. The information-dependent acquisition method had a duty cycle of 9 s, selecting the three most intense ions between 300 and 2000 m/z with charge states of 2-5 and that exceeded five counts from the 1-s TOF-MS scan. Rolling collision energies were used to facilitate CID. Any ion within 100 ppm of a former target ion was excluded for 60 s to eliminate repeated fragmentation of the same peptide species.
ICAT Data Processing-Data acquired from the ICAT information dependent acquisition experiments were analyzed by Pro ICAT software version 1.0 Service Packs 2 and 3 (Applied Biosystems). Mass tolerances of 0.15 and 0.1 Da for parent ion spectra and fragment ion spectra, respectively, were used. The measured molecular masses of parent peptides and their MS/MS data were used to search the National Center for Biotechnology Information (NCBI) nonredundant DNA/protein sequence database (nr), which was specifically formatted for use with Pro ICAT, with a maximum of one missed cleavage specified. The identification algorithms of Pro ICAT software follow routines typical to those generally used for peptide fingerprinting (23) where the patterns of sequence ions derived from MS/MS peptide fragmentation are correlated with a sequence database in accordance with the mass and specified mass tolerance of the precursor ion (24). The quality of data interpretation is described by the protein confidence, which measures how distant the identifications are from the next likely peptide and protein candidate. Protein confidence is expressed in a unit called ProtScore, which is a transformation of percent confidence. To convert percent confidence to ProtScore, the following formula is used: ProtScore ϭ Ϫlog(1 Ϫ (percent confidence)/100).
This formula also determines the contribution made by a single peptide identification to the ProtScore. The maximal contribution of any peptide is limited. Peptides are considered to have no higher confidence than 99.00%. This means that each peptide can contribute no more than 2.0 to the ProtScore. This limitation is necessary due to the limit in accuracy in reporting peptide confidence. In this study protein identifications were made based on ProtScore values Ն2, representing protein confidence greater than 99%.
During data processing, Pro ICAT identifies the ICAT pairs and determines the form of the label. Analysis of the isotope clusters from the light and heavy forms of the peptides only considers those peaks with corresponding peaks in the other cluster during the ICAT ratio calculation. The ICAT ratios are calculated from the relative areas of the extracted precursor ion spectra. The protein ICAT ratios are calculated from the combination of the constituent peptides where each charge state is treated separately.
Immunocytochemistry for Electron Microscopy-Animals were sacrificed 12 h following the last drug injection. Deeply anesthetized mice were perfused with saline followed with cold fixative, containing 4% paraformaldehyde, 0.1% glutaraldehyde in 0.1 M phosphate buffer (PB, pH 7.4). Brains were removed and postfixed in 4% paraformaldehyde for 2 h. 250 m thick coronal Vibratome slices of hippocampal formation were treated successively with 0.5% osmium tetroxide, 1% uranyl acetate, dehydrated, and embedded in epoxy resin. The embedded slices were trimmed to the CA1 region of hippocampal stratum radiatum. Thin sections were mounted onto nickel grids, etched and immunostained with an anti-clathrin antibody (1:10, BD Biosciences, San Jose, CA) visualized with 10 nm goldlabeled anti-mouse secondary (EMS, Fort Washington, PA). Sections were examined with a Hitachi 7000 electron microscope. For quantitation, 25-28 fields from each animal were selected so as to contain well oriented synaptic clefts and digital images were recorded at 80,000 to 100,000 magnification. Enumeration of grains over each postsynaptic profile was performed on all images by an observer blinded to specimen identity, by counting particles at the postsynaptic density (1) as well as over the remainder of each post-synaptic profile (2). The ratio of (1) to the total (1 ϩ 2) was computed for each profile. Presynaptic labeling was ignored in this analysis.

RESULTS
To study the effect of morphine administration on the expression of PSD-associated proteins, mice were injected with escalating doses of morphine, and a fraction enriched in hippocampal PSD proteins was isolated by subcellular fractionation as described previously (20). This procedure allows for the isolation of synaptic junctions and separation of presynaptic proteins from the PSD, thus allowing enrichment of PSD proteins. We confirmed this by Western blotting with markers of PSD, such as PSD-95 and CaMKII. Next we used differential isotopic labeling with the ICAT reagent to quantitate differences in protein levels by mass spectrometry. The PSD proteins from saline-and morphine-treated animals were labeled using ICAT reagents and subjected to LC-MS/MS analysis. The acquired MS/MS spectra were analyzed by database searching and stringent data filtering (see "Experimental Procedures"). Protein identifications were made based on ProtScore values equal to or higher than 2.0 (representing protein confidence higher than 99%). Overall we identified 102 proteins, the majority representing signaling, trafficking, and cytoskeletal proteins (Table I). Furthermore these represented postsynaptic proteins involved in synaptic signaling (such as GluR1, Homer, PSD-95, CaMKII, and protein kinase C ␥) indicating that this fractionation protocol enriches proteins in the PSD. Quantitative analysis showed that of the 23 proteins that were altered by morphine treatment 10 proteins showed consistent changes in levels (in both forward and reverse labeling experiments). Upon averaging forward and reverse protein abundance ratios, we obtained a value of 1.10 with a standard deviation of 0.29. Therefore, changes were considered to be substantial when exhibiting a ratio of either Ͼ1.39 or Ͻ0.81 (see Table II). A typical spectrum is shown in Fig. 1; a robust increase in Na ϩ /K ϩ -ATPase ␣3 levels was observed when the studies were carried out by forward labeling (Fig. 1A, left  panel), and this ratio was reversed when the labeling was reversed (Fig. 1A, right panel). Fig. 1B shows the MS/MS analysis that identified the 591.85 m/z peak as a heavy ICATlabeled tryptic peptide C H ATILLQGK (where H represents heavy) of Na ϩ /K ϩ -ATPase ␣3. As a control, we found that the peaks corresponding to tubulin ␤5 were not changed in either forward (morphine/saline) or reverse (saline/morphine) labeling (Fig. 1C, left and right panels, respectively).
Among the proteins that showed consistent and substantial changes, clathrin heavy chain levels were the most dramatically increased (more than 2-fold) upon morphine treatment (Table II). Two other proteins that showed a substantial increase were Na ϩ /K ϩ -ATPase ␣3 and 14-3-3 (Table II). ICAT analysis also revealed decreases in several proteins, including  ␣3 (A and B) and tubulin (C) following ICAT-LC-MS/MS. A, PSD proteins isolated from the hippocampi of mice treated with saline and morphine were labeled with "light" (L) and "heavy" (H) ICAT reagents, respectively (forward labeling, left panel). A pair of peaks corresponding to triply charged ions with heavy/light ratio greater than 1 (heavy/light Ͼ 1) was detected. A reversed heavy/light ratio (heavy/light Ͻ 1) of this pair of peaks was detected in reverse labeling (right panel) where PSD proteins from salineand morphine-treated mice were labeled with heavy and light ICAT reagents, respectively. B, MS/MS analysis of 591.85 ion and database search revealed a heavy ICAT-labeled tryptic peptide corresponding to the amino acid sequence of C H ATILLQGK of Na ϩ /K ϩ -ATPase ␣3. Peptide fragment ions corresponding to peptide fragment of "y" ions were labeled as y n in the figure. C, the peak ratio of heavy/light-labeled peaks corresponding to tubulin ␤5 tryptic peptide was not changed in both forward (left panel) and reverse (right panel) labeling.
voltage-dependent anion channel 1, a protein enriched in the PSD that has been shown to be involved in learning and synaptic plasticity in the hippocampus (25). Morphine treatment also decreased the levels of glutamate-ammonia ligase and glutamate dehydrogenase 1 (Table II), enzymes that regulate glutamate levels.
Using Western blotting, we verified that the changes in their relative abundance upon morphine treatment were consistent with those observed by ICAT analysis. We found significant increases in levels of clathrin heavy chain, Na ϩ /K ϩ -ATPase ␣3, and 14-3-3 in the PSD fraction of morphine-as compared with saline-treated animals (Fig. 2). In contrast, tubulin levels showed no changes, consistent with the ICAT analysis data (see Table I and Fig. 1C). Interestingly the levels of these proteins in the total hippocampal homogenate were not altered by morphine. Thus, it appears that there is a local increase of these proteins at the PSD upon morphine treatment.
We further examined the effects of morphine on the local increase of clathrin using pre-embedding immunogold electron microscopic immunocytochemistry. For this, we focused on stratum radiatum of CA1 hippocampus and examined the ultrastructural localization of clathrin at dendritic synapses. To determine whether there were changes in the levels of clathrin at the PSD in response to morphine, the number of gold particles associated with the PSD versus the total number of particles in 25-28 postsynaptic profiles was compared in sections from saline-and morphine-treated mice (Fig. 3A). We found a significant increase in the proportion of PSD-associated clathrin immunogold labeling in morphine-treated mice (Fig. 3B), supporting our data from ICAT and Western blotting analysis and indicating a local increase of clathrin at the PSD.
We next analyzed the effect of morphine on the levels of two other key endocytic molecules: AP-2 and dynamin. Western blot analysis indicated that morphine treatment significantly increased the levels of dynamin and AP-2 in the PSD fraction (Fig. 4, A and B, respectively). Although these proteins were identified in the ICAT analysis, the changes seen upon morphine administration were found to be not statistically significant. However, Western blotting analysis, a more sensitive technique for detecting protein levels, revealed that their levels were significantly increased specifically at the PSD but not in the total hippocampal homogenate. This is consistent with the notion that morphine treatment leads to a recruitment of dynamin and AP-2, along with clathrin, to the PSD. These data indicate that morphine administration causes a redistribution of proteins involved in endocytosis at the postsynaptic membrane to the PSD.
To study whether morphine treatment also affects the posttranslational modification states (i.e. extent of phosphorylation) of these endocytic proteins, we subjected samples of hippocampal homogenate and PSD fractions from saline-or morphine-treated animals to 2-DE and performed Western blot analysis. We found that morphine treatment led to a substantial change in the post-translation modification state of dynamin (Fig. 4C, top panel); this change was observable only at the PSD fraction (and not in the total homogenate). Indeed morphine treatment caused a shift in the pI of dynamin isoforms to a lower pI. This is consistent with an increase in the extent of phosphorylation of dynamin. Phosphorylation of dynamin is thought to down-regulate the formation of clathrincoated pits, leading to a decrease in the extent of endocytosis and morphine-treated (MOR) animals were analyzed by Western blotting using antibodies against clathrin heavy chain (HC) (A), Na ϩ /K ϩ -ATPase ␣3 (B), 14-3-3 (C), and tubulin (to ensure equal protein loading and transfer). The levels of clathrin heavy chain, Na ϩ /K ϩ -ATPase ␣3, and 14-3-3 levels were normalized relative to tubulin levels. Data represent mean values Ϯ S.E. (**, p Ͻ 0.01 relative to saline-treated animals; t test; n ϭ 3). Levels of clathrin heavy chain, Na ϩ /K ϩ -ATPase, and 14-3-3 were significantly increased in the PSD fraction of morphine-versus saline-treated animals. Significant changes were not observed in the total homogenate. (26). To confirm the effects of morphine on the phosphorylation state of dynamin, membranes were stripped and blotted with an anti-tubulin antibody. The isoelectric point of tubulin was not altered by morphine treatment in either total homogenate or the PSD (Fig. 4C, bottom panel). We also found that morphine treatment did not alter the post-translational states of clathrin or AP-2 (data not shown). Thus it appears that morphine administration leads to increased phosphorylation of some but not all endocytic proteins.
We next examined the extent of association of clathrin with PSD-95 and Homer, two components of the PSD. The level of clathrin associated with PSD-95 and Homer was drastically increased upon morphine treatment (Fig. 5A, bottom panels). In contrast, the level of clathrin associated with AP-2 remained unaltered (Fig. 5A, middle panel). These results suggest that besides redistributing endocytic proteins to the PSD morphine treatment leads to the association of clathrin with components of the PSD; this could alter the extent of endocytosis of synaptic proteins including receptors. Because AMPA receptors are thought to undergo clathrin-dependent endocytosis (21), we next investigated the effect of morphine on the association of the GluR1 AMPA receptor subunit with clathrin. The level of clathrin associated with GluR1 was significantly decreased upon morphine treatment (Fig. 5B), although the total level of GluR1 was increased by this treatment (not shown). Taken together, these results show that the selective association of clathrin with endogenous GluR1 is significantly decreased upon morphine treatment.

DISCUSSION
In this study, we analyzed proteins enriched in the PSD fraction of mouse hippocampus. A number of studies have previously analyzed proteins in the PSD fraction obtained from whole brain (19,(27)(28)(29), forebrain (18,30,31), or cerebellum (30) using different proteomics techniques such as one-dimensional gel electrophoresis (19,27), 2-DE (18), multidimensional protein identification technology (MudPIT) (28,31), ICAT (18,30) or IMAC technology (29). When exploring the PSD, the majority of studies to date have used fractionation protocols that yield PSD fractions containing presynaptic proteins (32)(33)(34). To minimize presynaptic protein contam- ination in our samples, we used a recently developed subcellular fractionation protocol that allows the separation of synaptosomal preparations into two distinct fractions of proteins, a soluble fraction containing presynaptic particles and an insoluble fraction that corresponds to the PSD (20). Although this method allows for the selective enrichment of postsynaptic density proteins, the requirement of considerable amounts of starting material for the isolation of PSD makes this method less suitable for studies in smaller brain regions or specific brain nuclei.
We identified components of the mouse hippocampal PSD fraction using a proteomics approach that combines ICAT methodology with LC-MS/MS. The list of proteins generated in this study shows considerable overlap with other previously reported datasets of PSD proteins, indicating that our dataset consists of representative PSD proteins (18,19,31). However,

FIG. 4. Western blot analysis of hippocampal homogenate and PSD from saline-(SAL) or morphine-treated (MOR) animals.
Quantification of dynamin (A) and AP-2 (B) was performed relative to tubulin levels. Data represent mean values Ϯ S.E. (*, p Ͻ 0.05; **, p Ͻ 0.01 relative to saline-treated animals; t test; n ϭ 3). C, PSD fractions and total homogenate from mice treated with saline or morphine were separated by high resolution 2-DE and probed with antibodies to dynamin or tubulin. Morphine treatment induced an increase in the intensity of the spots at lower pI and a decrease at higher pI (indicated by arrowheads) in the PSD fraction (and not in total homogenate). the total number of proteins identified in our study is ϳ5-10fold lower than that identified in larger scale, non-brain regionspecific studies of the PSD (27,29). For example, using IMAC technology, Trinidad et al. (29) recently identified 1,264 unique proteins in the PSD fraction obtained from whole mouse brain, and Collins et al. (27) identified 620 PSD-associated proteins by performing a MS-based analysis of membrane-associated guanylate kinase signaling complexes from whole mouse brain.
The variability between different PSD datasets is likely due to differences in species/strains, brain regions, sample preparation, and method of analysis. For example, in a recent study Cheng et al. (30) compared the PSD protein profiles from rat forebrain and cerebellum using ICAT technology coupled with LC-MS/MS and found a number of proteins to exhibit statistically significant differences between these two brain regions, suggesting a marked heterogeneity of PSD profiles between brain regions. With the aim to further explore the PSD subproteome from specific brain regions, a recent report studied the PSD protein profile from rat hippocampal slices using two-dimensional LC-MS/MS (35). The list of proteins identified in the latter study overlaps significantly (more than 70 proteins) with our list of mouse hippocampal PSD proteins. When compared with another dataset of proteins identified by ICAT technology in the synaptosomal fraction (11), our dataset contains a higher proportion of proteins involved in signaling and scaffolding as well as receptors and channels. Together the datasets from these studies provide a database of hippocampal PSD proteins that will be useful for functional studies related to synaptic plasticity, learning, and memory.
Our dataset includes proteins that have been reported to be highly abundant in the PSD (such as CaMKII␣, PSD-95, and synaptic Ras) as determined using the absolute quantification (AQUA) strategy (28). Our dataset also includes proteins that are of moderate abundance (such as chapsyn-110, septin, and the GluR1 subunit of AMPA receptors) and proteins of low abundance in this fraction (such as Homer 2 and protein kinase C ␥), suggesting that this dataset covers a range of proteins regardless of their abundance in the PSD. The presence of these proteins, which are known to modulate synaptic signaling, at the PSD is in agreement with a critical role for the PSD as a scaffolding and signaling component of the synapse.
To avoid variability during the labeling procedure, we carried out independent forward and reverse labeling of PSD fractions and combined the results from these two distinct sets of studies. Of the total of 102 proteins, only 38 were identified in both forward and reverse experiments. This is likely due to the fact that current LC-MS/MS technology can only achieve ϳ30% reproducibility when samples with moderate complexity are repeatedly analyzed by this technique (36 -39). This is largely due to the fact that the MS/MS spectrometer fragments and consequently identifies only a small fraction of the peptides captured at the first stage of any given MS scan. In addition, there are some issues that can affect protein coverage when using ICAT technology. These include the absence of cysteine residues in some proteins, the low abundance of cysteine-labeled peptides, the low ionization efficiency during MS/MS, or the fact that only a subset of ions can be selected for MS/MS when many peptides are coeluted (30). We minimized these limitations by combining results from different experiments. Hippocampal synaptosomal fractions from morphine-or saline-treated animals were immunoprecipitated with GluR1 antibody. Clathrin heavy chain (upper) or GluR1 (middle) were detected in the immunoprecipitate by Western blotting. As a control, Western blotting was carried out on the synaptosomal fraction using antibody to clathrin (lower). IB, immunoblot; IP, immunoprecipitate; HC, heavy chain; SAL, saline; MOR, morphine.
We also report here the quantitative comparison of differentially expressed PSD proteins from mouse hippocampus upon morphine administration using ICAT technology. We found that several proteins were regulated in the morphinetreated animals; these represent trafficking proteins, signaling proteins, carriers/channels, and proteins involved in energy metabolism. Morphine-induced changes in these proteins have also been observed in other studies (11,13). In one of these studies, Prokai et al. (11) examined protein expression in synaptic membranes obtained from cortex using ICAT methodology. They found that chronic morphine exposure induced a decrease in the levels of the integral membrane proteins Na ϩ /K ϩ -ATPase ␣3 subunit and clathrin. We found that morphine administration induced an increase in the levels of Na ϩ /K ϩ -ATPase ␣3 subunit and clathrin at the PSD. This could be due to differences in the extent of fractionation; we used purified PSDs, whereas Prokai et al. (11) used intact synaptic membranes. Thus, the changes observed in the latter case would represent a result of morphine effects on both the presynaptic as well as PSD fraction. By separating presynaptic proteins from the postsynaptic components, we were able to specifically focus on changes in levels of these proteins at a distinct subcellular location (i.e. PSD).
In the present study, we showed that morphine treatment induced a dramatic increase in the levels of clathrin in the hippocampal PSD fraction and increased the association of clathrin with PSD-95 and Homer. This is, to our knowledge, the first report demonstrating that morphine treatment leads to increases in the levels of clathrin and its interactions with components of the PSD; previous studies reported clathrin to be localized in endocytic zones that are in close proximity but distinct from the PSD in this brain region (40,41). We also found that other endocytic molecules such as dynamin and AP-2 were increased at the PSD by morphine treatment. Although these proteins were identified by ICAT technology, they did not show statistically significant increases. This is likely due to the presence of more than one isoform of the same protein (caused by post-translational modification or the presence of different subunits), which cannot be distinguished at the peptide level when analyzed by mass spectrometry. The fact that we found that morphine treatment induced the recruitment of dynamin and AP-2 to the PSD suggests a possible role for the translocation of these proteins in the structure and maintenance of the PSD. These proteins are usually expressed in endocytic zones that are independent from the PSD (41). Recent studies have suggested that dynamin may localize at the PSD in association with components of the PSD, such as Homer, the mGluR5 receptor subunit, and shank (42), where it could participate in the development and modulation of synaptic plasticity.
In addition to increasing their levels at the PSD, morphine also altered the post-translational states of some endocytic proteins. It has been suggested that phosphorylation/dephosphorylation of endocytic proteins regulates their interactions, resulting in the regulation of clathrin-mediated endocytosis (43). This is supported by the findings that these proteins undergo dephosphorylation during the maturation of clathrinmediated endocytosis. Indeed the switch from the phosphorylated state of the endocytic proteins to the dephosphorylated state after nerve terminal depolarization is thought to trigger clathrin-mediated endocytosis (44). We found that morphine treatment led to an increase in the phosphorylation state of dynamin. This would inhibit its association with partner proteins and thus down-regulate clathrin-mediated endocytosis of synaptic vesicles (45).
In neuronal dendrites and spines, clathrin cycles between pools in the plasma membrane and cytosol. The plasma membrane pools constitute active endocytic zones that lie adjacent to but spatially segregated from the PSD (40). For a receptor or other membrane protein to be internalized, it must first be translocated to specialized endocytic areas. At these sites clathrin-dependent endocytosis occurs via the assembly of a coated pit. After the binding of coated pit components to the membrane, the coated pits invaginate and are pinched off from the membrane in a dynamin-dependent manner to form clathrin-coated vesicles (46). By modulating the levels and phosphorylation states of these endocytic proteins, morphine treatment could modulate the extent of clathrin-mediated endocytosis, and this could have an impact on the levels of receptors at the synapse.
Previous reports have shown that AMPA receptor endocytosis is a clathrin-and dynamin-dependent process that involves direct binding of the clathrin⅐dynamin⅐AP-2 complex to the receptors themselves (21,47,48). In this work, we showed that morphine treatment significantly decreased the specific association of clathrin with the AMPA receptor subunit, GluR1. Therefore, one could hypothesize that morphine treatment decreases the rate and extent of receptor endocytosis at the synapse by recruiting clathrin and other key endocytic proteins away from the endocytic zone to the PSD, thus disabling the association between clathrin and GluR1. Such a notion is in agreement with the findings from a recent study that found an increase in the proportion of immunogold particles for GluR1 on the synaptic plasma membrane in the amygdala of rats self-administering morphine (49). Taken together, these results support the idea that plasticity at excitatory synapses contributes to the development of drug addiction (50).
In summary, using a proteomics approach we specifically analyzed a PSD-enriched protein fraction from mouse hippocampus. This database along with other hippocampal PSD databases should be useful for further functional studies related to learning and memory. Using differential isotopic labeling we identified the components of the endocytic machinery that are substantially altered by morphine administration. These results suggest a critical involvement of the endocytic machinery in the modulatory effects of morphine at the PSD and provide further insight into mechanisms underlying synaptic changes elicited by drugs of abuse.