Distribution and Posttranslational Modification of Synaptic ERα in the Adult Female Rat Hippocampus

Abstract

Acute 17β-estradiol (E2) signaling in the brain is mediated by extranuclear estrogen receptors. Here we used biochemical methods to investigate the distribution, posttranslational modification, and E2 regulation of estrogen receptor-α (ERα) in synaptosomal fractions isolated by differential centrifugation from the adult female rat hippocampus. We find that ERα is concentrated presynaptically and is highly enriched with synaptic vesicles. Immunoisolation of vesicles using vesicle subtype-specific markers showed that ERα is associated with both glutamate and γ-aminobutyric acid-containing neurotransmitter vesicles as well as with some large dense core vesicles. Experiments using broad spectrum and residue-specific phosphatases indicated that a portion of ERα in synaptosomal fractions is phosphorylated at serine/threonine residues leading to a mobility shift in SDS-PAGE and creating a double band on Western blots. The phosphorylated form of ERα runs in the upper of the two bands and is particularly concentrated with synaptic vesicles. Finally, we used E2 with or without the acyl protein thioesterase 1 inhibitor, Palmostatin B, to show that 20 min of E2 treatment of hippocampal slices depletes ERα from the synaptosomal membrane by depalmitoylation. We found no evidence that E2 regulates phosphorylation of synaptosomal ERα on this time scale. These studies begin to fill the gap between detailed molecular characterization of extranuclear ERα in previous in vitro studies and acute E2 modulation of hippocampal synapses in the adult brain.

Acute effects of estrogens are mediated by extranuclear estrogen receptors (ERs). In neurons, 17β-estradiol (E2) acts through extranuclear ERs to activate intracellular signaling and alter neurophysiology within minutes (1–3). Such rapid E2 signaling in the brain has been implicated in learning and memory (4), affective behavior (5), reproductive function (6), energy homeostasis (7), and neuroprotection (8, 9). Understanding the distribution and functional specializations of extranuclear ERs in the brain is important in defining cellular mechanisms that link rapid E2 signaling with changes in physiology and behavior.

Extranuclear ERs include forms of the classical nuclear ERs, ERα and ERβ (10, 11), as well as G protein-coupled ERs, G protein-coupled ER-1 (also known as GPR30) (12) and Gq-mER (13). Of these, extranuclear ERα is the best understood. Studies in nonneuronal cells show that ERα is trafficked to the plasma membrane in a caveolin-dependent process (14) and is anchored there by palmitoylation (15). Membrane-associated ERα forms signaling complexes with other proteins such as IGF-I tyrosine kinase receptor and metabotropic glutamate receptor-1 (16, 17) through which E2 can rapidly activate both proximal Src and phosphatidylinositol 3-kinase (18) and distal kinases including ERK 1/2, MAPK, Akt, glycogen synthase kinase-3β, protein kinase C, and protein kinase A (8, 19–26). Despite tremendous progress in understanding extranuclear ERα in in vitro systems, much less is known about specializations of extranuclear ERα in the adult brain.

Electron microscopic immunocytochemistry has shown that extranuclear ERα is associated with synapses in the hypothalamus (27), hippocampus (28, 29), and prefrontal cortex (30). Milner et al. (28) reported that in hippocampal CA1, approximately 50% of extranuclear ERα labeling is present in axons, approximately 25% in dendritic spines, and approximately 25% in astrocytes. Hart et al. (29) found that 32% of γ-aminobutyric acid (GABA)ergic axonal varicosities in CA1 contain punctate ERα labeling and that within inhibitory boutons, ERα is associated with clusters of small clear vesicles. It is unknown, however, whether these small vesicles contain neurotransmitter or are a specialized population of endosomes that shuttle ERα to and from the plasma membrane. Ledoux et al. (31) showed that, in addition to ERα labeling on small clear vesicles, a subpopulation of neuropeptide-containing large dense core vesicles (LDCVs) is also ERα positive, suggesting that ERα is present on multiple types of secretory vesicles.

To begin to fill the gap between detailed molecular characterization of extranuclear ERα that has been done in cell culture and extranuclear ERα in the brain, we investigated the subcellular distribution and posttranslational modification of ERα in adult female rat hippocampal synaptosomes. Our results showed that extranuclear ERα is concentrated presynaptically and that a portion is associated with glutamatergic and GABAergic neurotransmitter vesicles as well as with LDCVs. We also found that the pool of ERα at or near synapses includes a phosphorylated form and that E2 treatment of hippocampal slices acutely dissociates ERα from synaptosomal membranes via depalmitoylation.

Materials and Methods

Animals

Animals were adult female Sprague Dawley rats (50–60 d old; Harlan, Indianapolis, IN). For initial characterization of ERα in tissue fractions, we compared gonadally intact, ovariectomized (OVX), or OVX estradiol-primed (OVX+E) rats. Because we found no differences, most studies were done with intact rats, except as noted. Ovariectomy was performed under ketamine (85 mg/kg, ip) and xylazine (13 mg/kg, ip) anesthesia using aseptic surgical procedures. For estradiol priming, rats were OVX and then 5 d later, each was given 2 injections (sc) of 10 μg of 17β-estradiol benzoate in 100 μl of sesame oil 24 h apart. Tissue was collected 2 d later. All animal procedures were performed in accordance with National Institutes of Health guidelines and were approved by the Northwestern University Animal Care and Use Committee.

Tissue fractionation by differential centrifugation

Tissue fractions were obtained from whole hippocampal homogenates or from hippocampal slices (see below). For whole hippocampal homogenates, rats were deeply anesthetized with sodium pentobarbital (80 mg/kg, ip) and perfused transcardially with ice-cold homogenization buffer containing (in millimoles): 5 HEPES-KOH (pH 7.2), 320 sucrose, 5 EDTA, 1 Na orthovanadate, 50 NaF, 10 Na pyrophosphate, 20 Na glycerophosphate, and 0.1 phenylmethylsulfonyl fluoride (PMSF), with 1 μg/ml leupeptin and 1 μg/ml aprotinin. Rats were decapitated and the hippocampus was dissected on ice. Hippocampi from two animals were pooled and homogenized in ice-cold homogenization buffer. All subsequent steps were carried out at 4 C. We followed the protocol described elsewhere (32) to obtain a highly enriched small synaptic vesicle (SV) fraction. Briefly, tissue homogenate was centrifuged at 1000 × g for 10 min to remove unbroken cells and nuclei. The postnuclear supernatant was centrifuged at 17,000 × g for 15 min, yielding a crude cytoplasmic fraction (S2) in the supernatant. The pellet was resuspended in homogenization buffer and centrifuged at 17,000 × g for 15 min, yielding washed synaptosomes (P2) in the pellet. Synaptosomes were then lysed in double-distilled H₂O with 4 mm HEPES (pH 7.4), 0.1 mm PMSF, 1 μg/ml leupeptin, and 1 μg/ml aprotinin and allowed to incubate for 30 min while rotating at 4 C. Lysed synaptosomes were centrifuged at 25,000 × g for 20 min, yielding heavy membranes in the pellet (LP1) and synaptosomal cytosol with synaptic vesicles in the supernatant (LS1). Synaptic vesicles were then pelleted at 160,000 × g for 2 h (LP2), leaving synaptosomal cytosol in the supernatant (LS2). Aliquots of each fraction were stored at −80 C.

In a separate experiment, lysed synaptosomes from P2 were loaded on a linear sucrose gradient (0.4 m-1.2 m sucrose in 20 mm HEPES-NaOH, pH 7.4; with 0.1 m PMSF; 1 μg/ml leupeptin; and 1 μg/ml aprotinin) and centrifuged at 110,000 × g for 2 h to separate synaptosomal organelles by sucrose velocity gradient fractionation (33). Thirteen fractions were collected from the top of the gradient and stored at −80 C.

Immunoisolation of defined vesicle populations

We used immunoisolation to investigate the association of ERα with defined vesicle populations. In one experiment, sucrose velocity gradient fractions 6–8 were pooled to yield an SV-enriched sample, and fractions 9–11 were pooled to yield an LDCV-enriched sample. Each of these samples was subjected to immunoisolation using antisynaptophysin antibodies coupled to tosylated magnetic beads (Dynal M-500; Invitrogen, Carlsbad, CA) as described elsewhere (34), with modifications. For each immunoisolation, 2 × 10⁷ (1.5 mg) beads were washed with 4 mm HEPES-NaOH (pH 7.4), pelleted in a magnetic device (Dynal MPC; Invitrogen), resuspended in 50 μl of 0.1 m borate buffer (0.1 m H₃BO₃-NaOH, pH 9.5), and incubated with 25 μg of goat antimouse IgG linker antibody (Millipore AP 124; Millipore, Bedford, MA) overnight at 37 C. Beads were then washed with wash buffer (4 mm HEPES-NaOH, pH 7.4; 0.1% BSA) and blocked with blocking buffer (0.2 m Tris-HCl, pH 8.5; 0.1% BSA) for 4 h at 37 C. Beads were then washed, resuspended in 50 μl of wash buffer, and incubated with 10 μg of mouse antisynaptophysin antibody (Millipore MAB5258) overnight at 4 C. Control beads were processed in parallel with immunobeads but were incubated with the primary antibody omitted. After this step, the beads were washed again, resuspended in incubation buffer (4 mm HEPES-NaOH, pH 7.4; 2 mm EDTA; 0.1% BSA) and 30–50 μl of either the SV-enriched or LDCV-enriched sample and incubated overnight at 4 C. Beads were pelleted and the supernatant was saved as the nonbound fraction. Pelleted beads were washed, resuspended in 50 μl of 4 mm HEPES-NaOH (pH 7.4), and 50 μl of Laemmli sample buffer (62.5 mm Tris-HCl, pH 6.8; 25% glycerol; 2% sodium dodecyl sulfate; 0.01% bromophenol blue; 5% β-mercaptoethanol), boiled for 5 min, and pelleted again in the magnet. The supernatant containing the bound fraction was stored at −20 C.

In a second experiment, crude synaptic vesicle fractions (LS1) were treated with dimethyl 3,3′-dithiobispropionimidate (DTBP; Thermo Scientific, Waltham, MA) to stabilize protein-protein interactions before immunoisolation. Samples were incubated with 0.25 mm DTBP for 30 min at room temperature, after which the reaction was quenched with 200 mm Tris buffer (pH 7.4), for 15 min. Samples were then centrifuged at 160,000 × g for 2 h, yielding a cross-linked LP2 fraction. Immunoisolation was then carried out as above, except that 4 mm HEPES was replaced with 0.1 m PBS (pH 7.4). After the blocking step, beads were incubated with 10 μg mouse antisynaptophysin, 15 μg mouse antivesicular γ-aminobutyric acid transporter (vGAT, Synaptic Systems 131011; Göttingen, Germany) or 15 μg mouse antivesicular glutamate transporter 1 (vGLUT1; Synaptic Systems 131511) overnight at 4 C. As a control, green fluorescent protein (GFP; 5 ng/μl; Vector Laboratories, Burlingame, CA) was added to LS1, subjected to crosslinking with DTBP as above, and then LP2 was probed for GFP (1:1000, Genscript A00886; Genscript, Piscataway, NJ).

Phosphatase treatments

To investigate whether the upper band of ERα detected in Western blots was a phosphorylated form, we carried out a series of dephosphorylation experiments. P2 and LP1 fractions (each 100 μg protein) were incubated either with 100 U calf intestine alkaline phosphatase (CIP; BioLabs, Lawrenceville, GA) or 2.5 U protein tyrosine phosphatase 1B (PTP1B; Abcam, Cambridge, MA) for 60 min at 37 C or with 1000 U Lambda phosphatase (BioLabs), 1 U protein phosphatase 2A (PP2A; Millipore), or 25 U protein phosphatase 1 (PP1; BioLabs) for 30 min at 30 C. Control reactions lacked the enzymes and were supplemented with equal volumes of the appropriate enzyme reaction buffers. After incubation with phosphatases, protein amounts were determined in each sample. Samples then were mixed with the equal volumes of Laemmli sample buffer, boiled for 5 min, and stored at −20 C.

Brain slice preparation and treatment

To prepare hippocampal slices, rats were deeply anesthetized with sodium pentobarbital (80 mg/kg, ip) and perfused transcardially with ice-cold oxygenated sucrose artificial cerebrospinal fluid (aCSF) containing (in millimoles): 75 NaCl, 75 sucrose, 25 NaHCO₃, 15 dextrose, 2.4 sodium pyruvate, 2 KCl, 1.3 ascorbic acid, 1.25 NaH₂PO₄, 2 MgCl₂, and 0.5 CaCl₂ (osmolarity 315, pH 7.4). Six to eight slices (400 μm) per hemisphere were cut into a bath of ice-cold oxygenated sucrose aCSF and then were transferred to oxygenated regular aCSF containing (in millimoles): 125 NaCl, 25 NaHCO₃, 10 dextrose, 2.5 KCl, 1.25 NaH₂PO₄, 1 MgCl₂, and 2 CaCl₂ (osmolarity 315, pH 7.4) at 35 C for 30 min. Slices then were incubated in regular aCSF containing 100 nm E2 (Sigma, St. Louis, MO) or vehicle (0.01% dimethylsulfoxide) for 20 min at room temperature. In some experiments, half the slices were pretreated with 10 μm Palmostatin B (EMD Biosciences, San Diego, CA) for 15 min before E2 or vehicle treatment. An equal number of slices from each animal was distributed to each treatment group for a total of 10–12 slices per sample. After treatment, hippocampi were dissected, homogenized, and subjected to fractionation.

Western blotting

Protein concentration was determined using the Bradford protein assay (Bio-Rad Laboratories, Hercules, CA). Samples containing equal total protein (or equal volume for immunoisolated samples) were separated on 10% SDS-PAGE gels and transferred to polyvinylidene difluoride membranes (GE Healthcare, Waukesha, WI). Membranes were blocked with 5% nonfat milk and then incubated with one of the following primary antibodies overnight at 4 C: rabbit polyclonal anti-ERα (1:1,000, Santa Cruz MC-20; Santa Cruz Biotechnology, Santa Cruz, CA), chicken polyclonal anti-ERβ (1:10,000, ck5912 described elsewhere (35), mouse monoclonal anti-Ca²⁺-dependent activator protein for secretion (CAPS)-1 (1:500, BD Biosciences 612138; BD Biosciences, Franklin Lakes, NJ), mouse monoclonal antisynaptophysin (1:3 million, Millipore MAB5258), mouse monoclonal antiflotillin 1 (1:1,000, BD Biosciences 610820), mouse monoclonal anti-Na⁺/K⁺ ATPase β-subunit (1:1,000, BD Biosciences 610914), rabbit polyclonal antipostsynaptic density (PSD)-95 (1:500, Cell Signaling 2507; Cell Signaling, Danvers, MA), rabbit polyclonal anti-vGAT (1:100,000, Sigma V5764), rabbit polyclonal anti-vGLUT1 (1:100,000, Abcam 18106, Abcam), rabbit polyclonal antipeptidylglycine α-amidating monooxygenase C terminus domain (PAM CD) (1:1,000, a generous gift from Drs. Richard Mains and Betty Eipper, University of Connecticut Health Center, Farmington, CT), or goat polyclonal anti-β-actin (1:2,000, Santa Cruz I-19R; Santa Cruz Biotechnology). Membranes were washed in PBS and then in 0.1% Tween 20 in PBS and incubated at room temperature for 1 h with horseradish peroxidase-conjugated antirabbit, antimouse, or antigoat IgG secondary antibodies (1:2,000, Vector Laboratories). Immunoreactivity was visualized using enhanced chemiluminescence (ECL Plus; GE Healthcare), films were scanned, and ODs were quantified using ImageJ software (National Institutes of Health, Bethesda, MD).

Quantitative analysis of Western blot data

Dephosphorylation/depalmitoylation experiments were run two to four times on independently collected protein samples, each with duplicate, triplicate, or quadruplicate reactions per run. For each run, all protein samples were loaded on the same gel for Western blot analysis. OD measurements were normalized to actin loading controls, averaged for each treatment group (e.g. across duplicates, triplicates, etc.) to obtain one value per treatment group per run. Treatment group means were then calculated with n as the number of independent runs of each experiment. For quantification of upper/lower band ratios of ERα, each band was normalized separately to its actin loading control, and the ratio was calculated from normalized values. Data were analyzed statistically using unpaired t tests for comparison of two groups or one-way ANOVA followed by Tukey post hoc tests for comparison of more than two groups. Data are reported as mean ± sem.

Results

Extranuclear ERα in hippocampal synaptosomal fractions

We investigated the distribution of ERα in synaptosomal fractions using differential centrifugation (Fig. 1A) followed by Western blotting. Initially, we used equal protein loaded per lane (4 μg) and probed Western blots for ERα, the SV marker synaptophysin, and the postsynaptic density marker PSD-95 (Fig. 1B). As expected, synaptophysin was enriched in synaptosomes (P2) and subsequent fractions expected to contain SVs and LDCVs (LS1, LP1, LP2). PSD-95 was present in crude synaptosomes (P2) and heavy membrane (LP1) fractions, both expected to contain postsynaptic densities, and was absent from SV-containing fractions (LS1, LP2). Overall, ERα levels were higher in the postnuclear cytoplasm (S2) than in whole synaptosomes (P2). Within synaptosomes, ERα concentrated in the crude synaptic vesicle fraction (LS1) and particularly in the enriched synaptic vesicle fraction (LP2). ERα levels were lower in the heavy synaptosomal membrane fraction (LP1), which, based on PSD-95, includes postsynaptic densities. To confirm ERα in P2 and subsequent fractions, we re-ran Western blots with variable protein loaded per lane (Fig. 1C). Doubling protein for P2, LP1, and LS2 to 8 μg per lane showed clearly that ERα is present in these fractions.

Fig. 1.

Distribution of ERα in hippocampal synaptosomal fractions. A, Schematic of hippocampal tissue fractionation by differential centrifugation. B, Equal amounts of protein (4 μg per lane) from fractions isolated by differential centrifugation were probed for ERα, the synaptic vesicle marker synaptophysin (Syn), and the postsynaptic density marker PSD-95. ERα was concentrated in crude cytoplasm (S2), synaptosomal cytosol (LS1), and synaptic vesicle (LP2) fractions. ERα did not concentrate with postsynaptic densities and migrated as a double band in synaptosomal fractions. C, To confirm ERα in P2 and subsequent SV-containing fractions, we varied protein loaded per lane. Doubling protein loaded for P2, LP1, and LS2 to 8 μg per lane showed that ERα is present in these fractions. The ERα signal was still strong in LP2, even with only 1 μg loaded.

As apparent in Fig. 1, B and C, anti-ERα MC-20 detected a double-band in P2 and subsequent fractions, i.e. in synaptosomes. In addition to the main band at approximately 66 kDa, a band that ran at approximately 70 kDa was present in synaptosomal fractions. This upper band was concentrated with SVs, being strongest in enriched SVs (LP2). The 70-kDa band was absent in postnuclear supernatant (not shown) and S2 fractions (Fig. 1, B and C) in which we detected only the 66-kDa band, although both bands were present in the nuclear fraction (not shown). That we detected ERα as a double band in synaptosomes suggested that synapse-associated ERα might be posttranslationally modified.

To investigate systemic E2 regulation of synaptosomal ERα, we performed the same differential centrifugation and Western blotting analysis on synaptosomal fractions isolated from OVX and OVX+E rats. There were no discernible differences in ERα levels or distribution in these fractions compared with gonadally intact animals (not shown). Therefore, subsequent studies were done with tissue from intact rats, except as noted.

Subcellular distribution of ERα in sucrose velocity gradient fractions of lysed synaptosomes

We used sucrose velocity gradient centrifugation to further investigate the distribution of ERα and other markers within synaptosomes. After synaptosome isolation, hypoosmotic lysis, and sucrose velocity gradient centrifugation (Fig. 2A), we collected 13 fractions and analyzed them by Western blotting (Fig. 2B). Postsynaptic density fragments migrated to the bottom of the gradient, evidenced by PSD-95 in fractions 10–13. The extrasynaptic plasma membrane marker Na⁺/K⁺ ATPase had a broader distribution but also toward the bottom of the gradient, in fractions 9–13. The lipid raft marker flotillin 1 was present in the lower half of the gradient, in fractions 7–13. The vesicle marker synaptophysin was present in fractions 6–12, peaking in fractions 7 and 8, suggesting that SVs may be present in fractions 6–8, whereas LDCVs may be present in fractions 9–12. To corroborate this idea, we examined the LDCV markers, CAPS-1 and PAM. PAM can be cleaved into several soluble fragments, but its C terminus contains a domain (CD) that remains stably anchored in the LDCV membrane (36). We found that PAM CD peaked in fractions 9–11, confirming LDCVs in these fractions. CAPS-1 was also detected in fractions 9–11 but was present in a greater quantity in fractions 2–4, consistent with a large fraction of the CAPS-1 being cytosolic (37).

Fig. 2.

Distribution of ERα in sucrose velocity gradient fractions. A, Schematic of sucrose velocity gradient fractionation. B, Isolated fractions were probed for ERα, ERβ-like protein (ck5912) and markers of LDCVs (CAPS-1, PAM CD), synaptic vesicles (Syn), lipid rafts (Flotillin 1), plasma membranes (Na⁺/K⁺ ATPase), PSD-95, and actin. ERα overlapped with synaptic vesicles in fractions 6 and 7 with a smaller peak in fraction 11, likely reflecting a subset of LDCVs. In contrast, ERβ-like protein was cytosolic and did not migrate with vesicle, membrane, or postsynaptic density markers.

As with simple differential centrifugation, the distribution of ERα overlapped with vesicle markers, peaking in fractions 6 and 7 and again in fraction 11, consistent with a portion of ERα being associated with SVs and with LDCVs. Interestingly, the distribution of ERβ-like protein (using ck5912) was very different from ERα. ck5912 recognizes ERβ, although like many anti-ERβ antisera, it also recognizes an approximately 50-kDa band in Western blots from ERβ knockout mice (35). Thus, although ck5912 and other anti-ERβ antisera cannot be used to determine where ERβ is located in tissues, they are useful for determining where ERβ is not. We found that ERβ-like protein was present in cytosolic fractions 1–3 but was not detectable in fractions containing plasma membranes, SVs, or LDCVs.

ERα is associated with glutamatergic and GABAergic neurotransmitter vesicles

That we found ERα in fractions containing SVs and LDCVs is consistent with previous observations of ERα-immunoreactive small vesicles (29) and LDCVs (31) in axonal boutons. We used immunoisolation with vesicle type-specific proteins, synaptophysin, vGAT (38), and vGLUT1 (39) to determine whether these are neurotransmitter vesicles.

To determine whether ERα is associated with SVs and/or LDCVs, first we pooled sucrose velocity gradient fractions 6–8 (enriched in SVs) and fractions 9–11 (enriched in LDCVs) and subjected each to immunoisolation using antisynaptophysin. Material immunoisolated from the SV-enriched fractions (Fig. 3A) contained a large amount of synaptophysin, as expected. Furthermore, this material also contained vGLUT1 and vGAT, confirming that we isolated whole vesicles, not just synaptophysin protein. Importantly, the immunoisolated SVs also contained ERα. Similarly, material immunoisolated from the LDCV-enriched fractions (Fig. 3B) contained abundant synaptophysin as well as PAM CD, confirming isolation of whole LDCVs. A small amount of ERα was also detectable in the immunoisolated LDCVs, indicating that ERα is associated with some LDCVs. In contrast to vesicle markers and ERα, the amount of PSD-95 pulled down by immunobeads was low and indistinguishable from that pulled down by control beads. Thus, these immunoisolation experiments provide good evidence that a portion of ERα is associated with synaptophysin-containing SVs and LDCVs.

Fig. 3.

ERα is associated with both glutamatergic and GABAergic neurotransmitter vesicles. A, Sucrose velocity gradient fractions 6–8 were pooled, yielding an SV-enriched fraction, and subjected to immunoisolation with antisynaptophysin. Western blots of material bound to control beads (CB) or immunobeads (IB) were probed for ERα, synaptophysin (Syn), vGAT, or vGLUT1. Synaptophysin-containing synaptic vesicles also contained vGLUT1, vGAT, and ERα. B, Sucrose velocity gradient fractions 9–11 were pooled, yielding an LDCV-enriched fraction, and subjected to immunoisolation with antisynaptophysin. Western blots of material bound to control beads or immunobeads were probed for ERα, synaptophysin, PAM CD, or PSD-95. Synaptophysin-containing LDCVs also contained PAM CD and ERα but not PSD-95. C, To control for the specificity of chemical cross-linking, recombinant GFP was added to LS1, cross-linked using DTBP, centrifuged to yield cross-linked LP2, subjected to immunoisolation using antisynaptophysin, anti-vGAT, or anti-vGLUT1 and then probed for GFP. No GFP was detected in any of the immunoisolated samples. D, DTBP-cross-linked LP2 was subjected to immunoisolation with anti-vGAT or anti-vGLUT1 and probed for synaptophysin, vGLUT1, vGAT, or ERα. Immunoisolated GABAergic and glutamatergic neurotransmitter vesicles contained ERα and synaptophysin.

We performed a second experiment to test more directly whether ERα-containing SVs are neurotransmitter vesicles. Crude synaptic vesicles (LS1) were treated with DTBP to stabilize protein-protein interactions, and the enriched synaptic vesicle fraction (LP2) was prepared from these samples and subjected to immunoisolation. As a control for the specificity of cross-linking, we first confirmed that DTBP did not create an association between GFP and vesicles when GFP was added to LS1 before cross-linking. No GFP was detected in vesicles immunoisolated with synaptophysin, vGAT, or vGLUT1 (Fig. 3C). Next we examined cross-linked vesicles immunoisolated using anti-vGAT and anti-vGLUT1 (Fig. 3D). As expected, vGAT immunobeads pulled down mostly vGAT-containing vesicles, and vGLUT1 immunobeads pulled down mostly vGLUT1-containing vesicles, indicating samples highly enriched in GABAergic or glutamatergic neurotransmitter vesicles, respectively. Importantly, ERα was associated with both GABAergic and glutamatergic vesicles, with a comparable amount present on both vesicle types. This experiment thus confirms that ERα can associate with both GABAergic and glutamatergic neurotransmitter vesicles.

ERα was detected as a double band on both types of vesicles, consistent with initial characterization of ERα in hippocampal synaptosomal fractions. The upper to lower band ratio of ERα on these vesicles was similar to P2, LP1, and LS1 but lower than in nonimmunoisolated LP2. This suggests that either the upper band is more likely to be lost during immunoisolation or that LP2 contains additional ERα that is enriched in the form that runs as the upper band.

The upper band of ERα in hippocampal synaptosomes contains a phosphorylated form

We hypothesized that upper ERα band in synaptosomes might contain posttranslationally modified ERα, such as by phosphorylation. Previous studies in COS-1 cells have shown that phosphorylation of ERα at serine 118 (Ser-118) produces a similar electrophoretic shift from 66 to 70 kDa (40).

To investigate whether the upper band of ERα contains a phosphorylated form, first we subjected P2 and LP1 fractions to dephosphorylation using the broad spectrum phosphatase, CIP, and measured the ratio of the upper to lower bands on Western blots (Fig. 4A). Compared with control samples treated with enzyme reaction buffer alone, CIP strongly decreased the upper band, by 72 ± 4% for P2 and by 67 ± 0.4% for LP1 (both n = 3, P < 0.01). Cotreatment with 10 mm sodium orthovanadate (SO), an inhibitor of CIP, significantly attenuated the CIP-induced decrease but did not completely block it (tested with LP1, P < 0.05 vs. both CIP alone and control; Fig. 4A). These results indicate that the 70-kDa ERα in synaptosomal and synaptic vesicle membrane fractions contains a phosphorylated form of the receptor. The intermediate effect of SO may be explained by the fact that SO inhibits CIP activity only partially (41). In contrast to results with CIP, treatment of P2 with another broad-spectrum phosphatase, lambda phosphatase, did not affect the upper to lower band ratio (n = 3, Fig. 4B), indicating a substrate preference for CIP.

Fig. 4.

The upper band of ERα in synaptosomal fractions contains a serine/threonine phosphorylated form. A, Treatment of P2 or LP1 fractions with the broad spectrum phosphatase, CIP (100 U), strongly decreased the upper to lower band ratio of ERα (each n = 3, P < 0.01 for P2, t test; P < 0.01 for LP1, one way ANOVA followed by Tukey post hoc test), indicative of a phosphorylated form of ERα in the upper band. Pretreatment of LP1 with 10 mm SO, an inhibitor of CIP, significantly attenuated the CIP-induced decrease (n = 3, P < 0.05, Tukey post hoc test). B, Treatment of P2 with another broad spectrum phosphatase, lambda phosphatase (lambda PP; 1000 U), had no effect on the upper to lower band ratio of ERα (n = 3). C, Treatment of P2 fractions with the serine/threonine phosphatase, PP2A (1 U), significantly decreased the upper to lower band ratio of ERα (n = 4, P < 0.05, one way ANOVA followed by Tukey post hoc test). Pretreatment with OA, an inhibitor of serine/threonine phosphatases, blocked the effect of PP2A (P < 0.05, Tukey post hoc test). D, Similar to results with PP2A, another serine/threonine phosphatase, PP1 (25 U), significantly decreased the upper to lower band ratio of ERα (n = 3, P < 0.05, one way ANOVA followed by Tukey post hoc test). Pretreatment with OA showed a strong trend toward blocking the effect of PP1 (P = 0.06, Tukey post hoc test). E, The combination of PP2A and PP1 decreased the upper band further than either PP2A or PP1 alone (n = 2, P < 0.01, one way ANOVA followed by Tukey post hoc test) and pretreatment with OA significantly attenuated the decrease (P < 0.01, Tukey post hoc test). F, Treatment with the tyrosine-specific phosphatase, PTP1B (2.5 U), had no effect on the upper to lower band ratio of ERα (n = 3). *, P < 0.05; **, P < 0.01.

To further investigate ERα phosphorylation, we treated P2 with the serine/threonine-specific phosphatases, PP1 and PP2A. PP1 and PP2A are heavily expressed in brain and known to target synaptic proteins (42–44). Treatment with either PP2A (Fig. 4C) or PP1 (Fig. 4D) significantly decreased the upper to lower band ratio, by 31 ± 5% for PP2A (n = 4, P < 0.05) and by 30 ± 9% for PP1 (n = 3, P < 0.05). In both cases, pretreatment with 1 μm okadaic acid (OA), a phosphatase inhibitor, attenuated the upper band decrease (for PP2A, P < 0.05 vs. PP2A alone; for PP1, P = 0.06 vs. PP1 alone). The combination of PP1 and PP2A decreased the top band farther than either phosphatase alone, by 57 ± 4% (n = 2, P < 0.01), and this was also inhibited by OA (P < 0.01 vs. PP2A+PP1). These results indicate that a portion of synaptosomal ERα is phosphorylated at serine/threonine residues. That the combination of PP1 and PP2A was less effective than CIP in reducing the upper band suggests the presence of other phosphorylated residues not affected by PP1 or PP2A. We tested for tyrosine phosphorylation by treating P2 with the tyrosine-specific phosphatase, PTP1B. PTP1B did not affect the upper to lower band ratio (n = 3, Fig. 4F), suggesting that the upper band does not contain phosphorylated tyrosine residues.

It should be noted that, in addition to decreasing the upper to lower band ratio, PP2A or PP1 decreased ERα levels overall, by approximately 20%. One possibility is that dephosphorylation decreases the stability of the receptor leading to its degradation, as has been described for nuclear ERα in breast cancer cells (45, 46).

Estradiol acutely decreases membrane associated ERα via depalmitoylation

Experiments with transfected COS-1 and MCF-7 cells show that E2 can acutely induce serine phosphorylation of ERα producing an upward shift in SDS-PAGE (40, 47). To investigate whether E2 acutely regulates the upper band of ERα in hippocampal tissue, we prepared acute hippocampal slices from OVX+E rats and treated them with vehicle or 100 nm E2 for 20 min, dissected the hippocampus, and obtained synaptosomal fractions. The most dramatic effects of E2 were to decrease overall ERα levels in LP1, which contains synaptosomal plasma membranes, by 62 ± 9% (n = 3, P < 0.01), and to increase ERα in LS1, which contains synaptosomal cytosol, by 42 ± 6% (n = 3, P < 0.05). There was no change in ERα levels in P2 (n = 3, P > 0.10), the fraction that contains whole synaptosomes. The opposing effects of E2 on LP1 and LS1 suggested that E2 acutely dissociated ERα from synaptosomal plasma membranes.

Previous studies in nonneuronal cells have shown that association of ERα with the plasma membrane is controlled, in part, by palmitoylation (48, 49). Thus, in a separate experiment, we tested whether the E2-induced decrease in ERα association with synaptosomal plasma membranes involves depalmitoylation using Palmostatin B, a potent and specific inhibitor of the depalmitoylating enzyme, acyl protein thioesterase 1 (50). Pretreatment of slices with 10 μm Palmostatin B had no effect on its own but completely blocked the effect of E2 to decrease ERα in LP1 (n = 4, P < 0.05, Fig. 5A); ERα levels in P2 were unaffected by Palmostatin B or E2 (n = 4, P > 0.10, Fig. 5B). These results indicate that E2 acutely depletes ERα from synaptosomal membranes by depalmitoylation.

Fig. 5.

Estradiol acutely decreases synaptosomal membrane-associated ERα via depalmitoylation. A, Hippocampal slices from OVX+E rats were pretreated with vehicle or Palmostatin B, an inhibitor of the depalmitoylating enzyme, acyl protein thioesterase 1, and then exposed to E2 or vehicle for 20 min followed by differential centrifugation to yield synaptosomal membranes (LP1). E2 significantly decreased total ERα levels in LP1, which was blocked by pretreatment with Palmostatin B (P = 0.01, one way ANOVA followed by Tukey post hoc test). Palmostatin B alone had no effect on ERα levels. Despite the lower total levels of ERα in E2-exposed slices, the upper to lower band ratio was unaffected by E2 or Palmostatin B. B, ERα levels in whole synaptosomes (P2) from the same slices as in A were unaffected by E2 or Palmostatin B, arguing against E2-induced degradation of ERα. *, P < 0.05.

Despite the acute E2-induced loss of ERα from synaptosomal membranes, the upper to lower band ratio of ERα was unaffected by E2 treatment (P > 0.10), indicating that E2 is unlikely to acutely regulate synaptosomal ERα phosphorylation at serine/threonine residues.

Discussion

Here we describe the distribution and posttranslational modification of ERα in adult female rat hippocampal synaptosomes. We found that ERα is present in synaptosomal cytosol and membrane fractions and is highly enriched with synaptic vesicles. Immunoisolation with vesicle subtype-specific markers showed that ERα is associated with both glutamatergic and GABAergic neurotransmitter vesicles as well as some LDCVs. Additionally, a portion of ERα in synaptosomal fractions is phosphorylated at serine/threonine residues, leading to a mobility shift in SDS-PAGE and creating a double band on Western blots. The phosphorylated form of ERα runs in the upper of the two bands and is concentrated with synaptic vesicles. Finally, we found that acute E2 treatment of hippocampal slices depletes ERα from synaptosomal plasma membranes by depalmitoylation but does not appear to affect the phosphorylation of the receptor.

ERα associated with vesicles

ERα was present in all synaptosomal fractions but was heterogeneously distributed, being most concentrated in the synaptic vesicle fraction and least concentrated in heavy synaptosomal membranes that include postsynaptic densities. Fractionation of whole synaptosomes by sucrose velocity gradient centrifugation showed two peaks of ERα, one that overlapped with the vesicle marker, synaptophysin, and another that overlapped with the LDCV markers, PAM CD and CAPS-1. The distribution of ERα was more discrete than vesicle markers, indicating that ERα-positive vesicles represent only a subset of all vesicles labeled with each marker.

Previously we found that ERα is associated with both small clear vesicles (29) and LDCVs (31) at inhibitory synapses. Others have found ERα associated with small clear vesicles at excitatory synapses (28). Here we show that at least a portion of the ERα-associated small vesicles are GABAergic and glutamatergic neurotransmitter vesicles, and we confirmed that ERα is associated with LDCVs. The presence of ERα on neurotransmitter/neuropeptide vesicles suggests that E2 could regulate these vesicles directly.

Estradiol rapidly modulates synaptic release of both glutamate (51) and GABA (52) in the hippocampus. Rapid E2 modulation of glutamate release is mediated by ERβ, whereas rapid modulation of GABA release is mediated by ERα. Electrophysiological studies show that ERα activation acutely suppresses synaptic inhibition, specifically in females, by mobilizing endocannabinoids to decrease the probability of GABA release. Endocannabinoids suppress GABA release through activation of type I cannabinoid receptors (CB1Rs) located on a subset of inhibitory presynaptic boutons (53). Interestingly, the inhibitory neurons that express CB1Rs are primarily cholecystokinin-expressing basket cells (54), which are also those that contain presynaptic ERα (29). It is therefore tempting to speculate that ERα associated with GABAergic vesicles is directly involved in rapid E2 modulation of inhibitory synaptic transmission. The functional significance of ERα localized to glutamatergic vesicles in the hippocampus is unclear, however. Further studies may reveal a role for ERα in modulating glutamatergic synaptic function in the hippocampus. This is an active area of investigation in our laboratory.

ERα phosphorylation at or near synapses

We found that a portion of synaptosomal and synaptic vesicle-associated ERα is phosphorylated at serine/threonine residues, leading to a double band on Western blots. Based on similarities with nonneuronal cells (40), phosphorylation at Ser-118 likely contributes to the upper band of ERα, although it does not account for it entirely. In nonneuronal cells, PP2A (55) and PP1 (47) completely dephosphorylate ERα at Ser-118. Because we found that CIP was more effective in reducing the upper band than PP2A and PP1 combined, this suggests that ERα at or near synapses may be phosphorylated at sites in addition to Ser-118.

Numerous studies in nonneuronal cells have investigated regulation of Ser-118 phosphorylation by E2 and by specific kinases and the functional consequences of ERα phosphorylation for genomic and nongenomic signaling. For example, Ser-118 phosphorylation promotes ERα interaction with coactivators and downstream signaling molecules such as cAMP response element-binding protein and steroid receptor coactivator-1 and enhances transcriptional activity (56–60). Similarly, phosphorylation at Ser-118 regulates ERα interactions with membrane-associated targets that could be involved in rapid nongenomic signaling. These targets include epithelial growth factor and IGF-I receptors, G protein-coupled receptors, and protein kinases including phosphatidylinositol 3-kinase, Akt, MAPK family members, protein kinase A, protein kinase C, and the nonreceptor tyrosine kinases Src and Shc (61–63).

We did not find differences in the overall levels of ERα or the upper to lower band ratio between synaptosomal fractions prepared from intact rats or ovariectomized rats with or without E2 priming. This suggests that synapse-associated ERα might be phosphorylated in a ligand-independent manner and/or that neurosteroid E2 (64) supports ERα phosphorylation at or near synapses.

Acute E2 regulation of ERα association with synaptosomal plasma membranes

We tested whether E2 acutely regulates phosphorylation of synaptosomal ERα as it does in nonneuronal cells. Contrary to our expectations, we found no evidence that E2 acutely alters ERα phosphorylation; the upper to lower band ratio in Western blots was unaffected by E2. Instead, we found that overall levels of ERα on synaptosomal plasma membranes were strongly decreased by E2 and that this decrease required depalmitoylation. Although we did not show that ERα itself is depalmitoylated, this is the most likely explanation for our results.

Palmitoylation is well known to influence the recruitment of proteins to membrane microdomains, caveolae, and lipid rafts and to influence signal transduction (65, 66). Studies in nonneuronal cells have shown that palmitoylation of ERα and interactions with caveolin-1 are crucial for its plasma membrane location and function (15, 49, 67–69). In cancer cell lines, E2 treatment (1 h) reduces ERα palmitoylation and its interaction with caveolin-1 and promotes nongenomic E2 signaling through ERK and Akt (48). Conversely, in HeLa cells, inhibition of palmitoyl-acyltransferases with the broad spectrum inhibitor, 2-bromo-hexadecanoic acid, prevents both ERα palmitoylation and E2-induced ERK activation (70). Interestingly, the lack of palmitoylation renders ERα more susceptible to E2-dependent degradation and blocks ERα phosphorylation at Ser-118 (in MCF-7 cells) (71).

That E2 acutely depletes ERα from synaptosomal membranes is reminiscent of previous findings but with some differences. For example, Dominguez and Micevych (72) showed in cultured hypothalamic neurons that E2 promotes ERα turnover at the plasma membrane. Specifically, E2 both increased insertion of ERα at the membrane reaching a peak at 30 min and also stimulated internalization of membrane-associated ERα with a comparable time course. Similar ERα internalization has been shown in cortical cultures (73). In this second study, however, parallel experiments in P10 cortical synaptosomes indicated that E2-induced ERα internalization may be mediated by β-arrestin, which directs receptors to endosomes. We found that, in slices from adult hippocampus, E2-induced removal of ERα from synaptosomal membranes requires depalmitoylation. The source(s) of these differences remain to be determined. On the one hand, differences might be due to the various cell types studied or to use of immature vs. adult brain tissue. Alternatively, depalmitoylation could also be involved in β-arrestin-mediated ERα internalization. In either case, depalmitoylation of ERα likely promotes its dissociation from caveolin-1 and from the membrane, which could lead to: 1) up- or down-regulation of ERα interactions with signaling molecules; 2) ERα dimerization, translocation to nucleus and later genomic effects [(67), e.g. in interneurons, which contain nuclear ERα] (74); and/or 3) later proteasomal degradation of ERα. Any or all of these could be related to E2 modulation of synaptic function.

In summary, we have found that ERα at or near hippocampal synapses is phosphorylated and subject to regulation by palmitoylation in a manner that controls its association with plasma membranes and likely its ability to rapidly activate nongenomic signaling. Understanding the regulation of membrane-associated ERα will be key to defining the molecular mechanisms that underlie acute E2 actions in the brain.