Estrogen receptors observed at extranuclear neuronal sites and in glia in the nucleus accumbens core and shell of the female rat: Evidence for localization to catecholaminergic and GABAergic neurons

Abstract

Estrogens affect dopamine-dependent diseases/behavior and have rapid effects on dopamine release and receptor availability in the nucleus accumbens (NAc). Low levels of nuclear estrogen receptor (ER) α and ERβ are seen in the NAc, which cannot account for the rapid effects of estrogens in this region. G-protein coupled ER 1 (GPER1) is observed at low levels in the NAc shell, which also likely does not account for the array of estrogens’ effects in this region. Prior studies demonstrated membrane-associated ERs in the dorsal striatum; these experiments extend those findings to the NAc core and shell. Single- and dual-immunolabeling electron microscopy determined whether ERα, ERβ, and GPER1 are at extranuclear sites in the NAc core and shell and whether ERα and GPER1 were localized to catecholaminergic or γ-aminobutyric acid-ergic (GABAergic) neurons. All three ERs are observed, almost exclusively, at extranuclear sites in the NAc, and similarly distributed in the core and shell. ERα, ERβ, and GPER1 are primarily in axons and axon terminals suggesting that estrogens affect transmission in the NAc via presynaptic mechanisms. About 10% of these receptors are found on glia. A small proportion of ERα and GPER1 are localized to catecholaminergic terminals, suggesting that binding at these ERs alters release of catecholamines, including dopamine. A larger proportion of ERα and GPER1 are localized to GABAergic dendrites and terminals, suggesting that estrogens alter GABAergic transmission to indirectly affect dopamine transmission in the NAc. Thus, the localization of ERs could account for the rapid effects of estrogen in the NAc.

1 |. INTRODUCTION

There is evidence that estrogens affect the progression of dopamine (DA)-related diseases, protecting against some symptoms of schizophrenia (Gogos et al., 2015; Kulkarni et al., 2012) and Parkinson’s disease (Bourque et al., 2019; Ragonese et al., 2004, 2006), and exacerbating the development of addiction to most drugs of abuse (Carroll et al., 2004; Maria et al., 2014; Perry et al., 2015). Estrogens also effect DA-dependent cognitive processes, including selective attention (Almey et al., 2013; Quinlan et al., 2010), reversal learning (Arad & Weiner, 2012), and memory system bias (Howes & Kapur, 2009; Madularu et al., 2014; Oades et al., 1985; Quinlan et al., 2008, 2013) among others (for review see Luine, 2014). Altered DA transmission in the dorsal striatum and the nucleus accumbens (NAc) is implicated in these diseases/disorders and cognitive processes (Gray et al., 1997; Howes & Kapur, 2009; Taghzouti et al., 1985), and 17β-estradiol (E2) rapidly increases both tonic and phasic DA release (Shams et al., 2016, 2018) as well as reduces calcium currents via a membrane receptor (Mermelstein et al., 1996) at least in the dorsal striatum. To understand how estrogens affect DA transmission in the NAc, a fuller understanding of estrogen receptor (ER) distribution in this region is required.

A previous study from our group used both light and electron microscopy to examine the distribution of ERα, ERβ, and G-protein coupled ER (GPER1) in the dorsal striatum, demonstrating that all three ERs are present at nonnuclear sites in this region (Almey et al., 2012). ERs in the dorsal striatum are not localized to DAergic neurons, but are localized to γ-aminobutyric acid-ergic (GABAergic) interneurons (Almey et al., 2016) and to a lesser extent, cholinergic interneurons (Almey et al., 2012). This study extends those findings by assessing the distribution of membrane-associated ERs (mERs) in the NAc core and shell, and by examining cells in which these mERs are present.

Estrogens modify DAergic transmission in the NAc at multiple stages. Systemic injections of E2 administered 48 h prior to testing result in significantly lower phasic DA release in the NAc, but an infusion of E2 into the NAc increases phasic DA release within 15 min (Thompson & Moss, 1994). This suggests that there are opposing genomic and nongenomic effects of E2 on DA release in the NAc. More recent work confirms this rapid effect of E2 on NAc DA, showing that systemic E2 administered 30 min prior to testing results in significantly higher cocaine-evoked DA release in the NAc (Yoest et al., 2018). E2 also rapidly increases the metabolism of DA in the NAc, indicated by increased levels of 3,4-dihydroxyphenylacetic acid (DOPAC) and homovanillic acid (HVA) within 30 min of E2 administration (Di Paolo et al., 1985). Some research has shown that E2 replacement administered to ovariectomized (OVX) rats attenuates DA reuptake in the NAc (Thompson, 1999), providing a potential explanation for the E2-induced increase in DA availability in the NAc. Additionally, ovariectomy results in decreases in D2 agonist and antagonist binding which is recovered by E2 replacement, suggesting that estrogens maintain levels of D2 receptors in this region (Landry et al., 2002; Le Saux et al., 2006). Some of the effects of E2 in the NAc occur over a long time scale (+24 h), suggesting that they are mediated by genomic effects of E2, while other effects of E2 occur rapidly (−30 min), demonstrating that E2 also has nongenomic effects in the NAc.

Previous experiments using light microscopy and in situ hybridization have examined the distribution of ERs in the NAc. Light microscopic and in situ hybridization studies observed low levels of immunolabeling for ERα and ERβ in the adult rodent NAc, exclusively at nuclear sites (Krentzel et al., 2021; Lorsch et al., 2018; Mitra et al., 2003; Quigley & Becker, 2021; Shughrue et al., 1998). GPER1 is also observed in the perikaryal cytoplasm in the adult rodent NAc at low levels (Hazell et al., 2009), presumably localized to cellular organelles and the plasma membrane, as has been previously observed (Funakoshi et al., 2006; Otto et al., 2008). Estrogens binding at nuclear Erα, ERβ, and GPER1 could be responsible for the estrogen response element–mediated genomic effects of estrogens in the NAc, and GPER1 could account for some of the nongenomic effects of estrogens in this region. However, the rapid nongenomic effects of estrogens in the NAc could also be mediated by mERα or mERβ. There is evidence that light microscopy may not be sufficient to observe mERα and mERβ (Almey et al., 2012); ultrastructural analysis would determine whether mERα, mERβ, and GPER1, are localized to the NAc, and if so, where and what types of cells these ERs are localized to.

These experiments used single-label immunoelectron microscopy to examine the distribution of ERα, ERβ, and GPER1 in the rat NAc. There is evidence suggesting functional differences between two subregions of the NAc, the core and the shell (Ito & Hayen, 2011; Saddoris et al., 2015), so we quantified ERs in these subregions separately to determine if ER distribution differs. Results show that ERα and GPER1 are frequently observed at nonnuclear sites in the NAc, and ERβ is also observed at nonnuclear sites albeit less so. Thus, dual-labeling experiments were also conducted to determine whether ERα and GPER1 are localized to catecholaminergic or GABAergic neurons in the NAc core and shell.

2 |. METHOD

2.1 |. Animals

Adult female (225–250 g; ~ 60 days old; n = 6 total) Sprague Dawley rats from Charles River Laboratories (Wilmington, MA) were pair-housed on a 12:12 light/dark cycle (lights on 06:00 – 18:00) with ad libitum access to food and water. All procedures were approved by the Weill Cornell Medicine Institutional Animal Care and Use Committee and were in accordance with the National Institutes of Health guidelines. The rats used in these experiments are the same as those used in our previous studies on mERs in the dorsal striatum (Almey et al., 2012) and a previous study by Williams and colleagues (Williams et al., 2011). After arrival, rats acclimated to the animal colony for a week, and then estrous cycle phase was determined using vaginal smear cytology (Turner & Bagnara, 1971; Williams et al., 2011). Only female rats with two consecutive, regular, 4–5 day estrous cycles were included in the study. Tissue from three rats in the diestrus 2 phase of the estrous cycle was analyzed for experiments 1, 2, and 3. Tissue from three rats in the estrus phase of the estrous cycle was analyzed for experiment 4. Results of vaginal smear cytology, used to determine estrous cycle phase, were verified by measuring uterine weights and plasma estradiol (E2) levels from blood samples collected during the perfusion procedure (Marcondes et al., 2002).

2.2 |. Antisera

ERα:

A rabbit polyclonal antiserum (AS409) produced against almost the full peptide for the native rat ERα (aa 61 through the carboxyl terminus) was supplied by S. Hayashi (Table 1). This antibody was previously tested for specificity and shown to recognize both ligand bound and unbound receptors (Alves et al., 1998; Okamura et al., 1992). This antiserum recognizes one major band migrating at ~67kD (the molecular weight of ERα) on immunoblots of uterine lysates (Milner et al., 2001). When tested on immunoblots of ERα fusion protein, the AS409 antibody recognized minor bands migrating at ~110 kDa (likely the ERα/fusion protein complex), one major band migrating at ~67 kDa, and minor bands migrating at ~41–45 kDa (the degradation products of ERα, following the purification of ERα from the fusion protein). Preadsorption of the antibody with purified ERα resulted in no detectable bands in any of these locations (Milner et al., 2001). Additionally, rat hippocampal section immunolabeled with AS409 antiserum preadsorbed with ERα fusion protein yielded no labeling of dendritic spines, terminals, and axons as well as almost no glial profiles in the CA1 region of the hippocampus examined by immuno-EM methods identical to those used in the present study (Milner et al., 2001).

TABLE 1.

Primary antibodies

Antibody name	Immunogen	Host	Antibody type	Dilution	Manufacturer	Catalog #, RRID
Estrogen receptor α	Native rat ERα peptide 61 to C-terminus	Rabbit	Polyclonal	1:10,000	S. Hayashi Okamura et al., 1992	RRID: AB_2314382
G-protein coupled estrogen receptor	Synthetic human GPER1 polypeptide (antigen sequence: CAVIPDSTEQSDVRFSSAV)	Rabbit	Polyclonal	1:1000	Filardo et al., 2000	RRID:AB_2631060
Estrogen receptor β	Native mouse ERβ peptide sequence 468–485 in the C-terminus	Rabbit	Polyclonal	1:2000	Zymed Laboratories, discontinued.	RRID: Z8P
γ-Aminobutyric acid	GABA-glutaraldehyde-hemocyanin conjugates	Rat	Polyclonal	1:2000	A.C.Towle (Lauder et al., 1986)	N/A
Tyrosine hydroxylase	Native rat tyrosine hydroxylase peptide (full length)	Mouse	Monoclonal	1:2000	Immunostar Inc.	Cat# 22941, RRID: AB_572268

GPER1:

A rabbit polyclonal antiserum generated against a synthetic peptide, CAVIPDSTEQSDVRFSSAV (Multiple Peptide Systems, San Diego, CA) derived from the C-terminus of the deduced sequence of the human GPER1 receptor (Revankar et al., 2005; Table 1). In western blots, this affinity-purified antibody specifically recognizes a 38-kD band that corresponds to the mature 351-amino acid GPER1 polypeptide and does not recognize either ERα or ERβ (Revankar et al., 2005). In brains fixed with 4% paraformaldehyde perfusion, immunoreactivity was greatly reduced when the antibody was preadsorbed with 10 mg/ml of purified C-terminal peptide (Filardo et al., 2000). Furthermore, we performed two additional controls to validate the specificity of the antibody in the hippocampus (Waters et al., 2015) and paraventricular nucleus of the hypothalamus (Contoreggi et al., 2021). The first control was preadsorption experiment. For this, a 1:2000 working dilution of the antibody was divided into two aliquots, and one was incubated with 40 mg of the antigenic peptide in 50 μl saline and the other with 50 μl saline. The solutions were incubated overnight at 4°C, spun at 10,000 rpm the following day, and the supernatant was collected. Acrolein/paraformaldehyde-fixed sections containing the hippocampus and PVN were processed for peroxidase immunocytochemistry using the two aliquots. In sections incubated with the antibody preadsorbed with the peptide, immunolabeling for GPER1 was completely abolished in the hippocampus (Waters et al., 2015) and PVN (Contoreggi et al., 2021).

ERß:

A rabbit polyclonal antiserum produced against a peptide sequence in the C-terminus (aa 468–485) of the mouse ERß protein was used (Z8P; discontinued; Zymed Laboratories, San Francisco, CA; Shughrue & Merchenthaler, 2001; Table 1). This antibody was tested for specificity using western blot analyses, which demonstrated a single band migrating at ~60 kDa. Preadsorption of the antibody with purified ERß resulted in no detectable band at this location. Additionally, this antibody exclusively labeled profiles containing ERß-mRNA according to in situ hybridization, and there was no labeling in brain sections from ERβ knockout mice (Creutz & Kritzer, 2002; Shughrue & Merchenthaler, 2001).

γ-Aminobutyric acid (GABA):

A rat polyclonal antiserum was provided by A. C. Towle, formerly affiliated with Department of Neurology and Neuroscience at Cornell University Medical College (Table 1). It was produced against GABA-glutaraldehyde-hemocyanin conjugates, and was tested for specificity using preabsorption with GABA-bovine serum albumin (BSA), eliminating GABA immunoreactivity (Lauder et al., 1986). Immunoreactivity of this antiserum is consistent with the specificity of other GABA-antisera (Lauder et al., 1986).

Tyrosine hydroxylase (TH):

A mouse monoclonal antiserum against the full length of the TH peptide in the rat (Immunostar, Inc., Hudson, WI; Table 1). This antibody has been characterized extensively in fixed rat brain (Pickel & Chan, 1990). The NAc has both DAergic and noradrenergic neurons (Kerfoot & Williams, 2011), so catecholaminergic profiles containing TH could be either DA or norepinephrine neurons.

2.3 |. Tissue preparation

Sodium pentobarbital was used to anesthetize rats (150 mg/kg, i.p.). All rats were perfused through the ascending aorta with 10–15 ml heparin (1000 U/ml) in saline, followed by 50 ml of 3.75% acrolein (Polysciences, Washington, PA) in 2% paraformaldehyde and 0.1 M phosphate buffer (PB; pH 7.4), and finally 200 ml of 2% paraformaldehyde in PB. Rats’ brains were removed, cut into four 5 mm blocks using a brain mold, and postfixed in 2% paraformaldehyde in PB for 30 min. The brains were sectioned coronally at 40 μm on a vibrating microtome (Vibratome; Leica) and stored in 30% sucrose and 30% ethylene glycol in PB at −80°C until immunohistochemical processing (Milner et al., 2011).

Tissue sections containing the NAc core and shell (Figure 1a) were rinsed in PB and coded with hole punches so that they could be pooled in single containers. Additionally, a section containing the ventromedial and arcuate nuclei of the hypothalamus or the supraoptic nucleus was included in analyses for ERα and ERβ, respectively. Abundant ERα/ERβ labeling was observed in these regions using light microscopy (Shughrue et al., 1998; Yaghmaie et al., 2010), so the success of immunolabeling could be confirmed at the light microscopic level in hypothalamic tissue if no labeling was observed in the NAc. Sections were incubated in 1% sodium borohydride in PB for 30 min to remove any active aldehydes. Tissue then was rinsed in PB, followed by 0.1M Tris-buffered saline (TBS; pH 7.6), and was incubated for 30 min in 1% BSA in TBS to reduce nonspecific labeling.

FIGURE 1.

Light microscopic examination of estrogen receptors (ERs) in the nucleus accumbens (NAc). a) Depiction of the region analyzed in electron microscopy experiments; the blue rectangle was considered the NAc core while the yellow was the NAc shell. b and c) Moderate levels of nuclear, but no extranuclear, labeling were observed for ERα. d and e) Very sparse nuclear labeling for ERβ was observed. f and g) Dense extranuclear labeling, but no nuclear labeling for G-protein coupled ER 1 (GPER1) was observed in the NAc. Black arrows, immunoreactive cells/nuclei. Dashed line indicates approximate delineation of the NAc core and shell. Scale bar b,d,f = 200 μm; c,e,g = 100 μm

2.4 |. Immunohistochemical labeling

Experiment 1.

Free floating tissue sections containing the NAc core and shell from three rats were processed for immunohistochemical localization of ERα and GPER1. Tissue sections from each rat were incubated in anti-rabbit ERα (1:10,000 dilution) or GPER1 (Bio-sciences, 1:1000 dilution) for 24 h at room temperature and for 4 days at 4°C in 0.1% BSA in TBS. Both ERs were visualized using the avidin-biotin complex (ABC) method (Milner et al., 2011). Briefly, the tissue was incubated in a 1:400 dilution of biotinylated donkey anti-rabbit immunoglobulin (IgG, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) in 0.5% BSA in TBS for 30 min. Tissue was then incubated in peroxidase-avidin complex (Vector, Burlingame, CA) for a further 30 min, and 3,3-diaminobenzidine (DAB; 0.22 mg/ml Aldrich, Milwaukee, WI) and 30% H₂O₂ (0.1 μl/ml) in TBS for 6 min.

Experiments 2 and 3.

Immunohistochemical localization of ERα or GPER1 and TH or GABA was carried out on tissue from three rats. Tissue sections were incubated in either ERα antisera (1:10,000 dilution) or GPER1 antisera (1:1000 dilution) for 24 h at room temperature and for 4 days at 4°C in 0.1% BSA in TBS. One day prior to processing either TH antisera (1:2000 dilution) or GABA antisera (1:2000 dilution) was added to the diluent.

For immunohistochemical localization this experiment used pre-embedding dual-labeling methods (Williams et al., 2011). The same ABC method described above for experiment 1 was used to visualize the ERs. TH and GABA were detected using silver enhanced immunogold. Briefly, tissue sections were incubated for 2 h in a 1:50 dilution donkey anti-rat or IgG conjugated to 1-nm colloidal gold particles (Electron Microscopy Sciences [EMS], Fort Washington, PA) in 0.001% gelatin and 0.08% BSA in 0.01M phosphate-buffered saline (PBS). Tissue sections then were rinsed in PBS, incubated in 1.25% glutaraldehyde in PBS for 10 min, rinsed again in PBS, followed by a brief wash in 0.2M sodium citrate (pH 7.4). A ~7 min incubation in a silver solution (IntenSE; GE Healthcare; discontinued) was used to enhance the conjugated gold particles.

Experiment 4.

After the completion of experiments 1–3, a fourth experiment was conducted to examine ERβ distribution in the NAc core and shell. Experiment 4 followed the procedures described for experiment 1, except that tissue was obtained from rats in the estrus phase of their cycle. Briefly, sections containing the NAc were incubated in anti-rabbit ERβ (1:2000 dilution) for 24 h at room temperature and for 4 days at 4°C in 0.1% BSA in TBS. Following procedures from the previous experiments, ERβ was visualized using the ABC method.

2.5 |. Tissue fixation and embedding for ultrastructural analysis

Following immunolabeling, tissue sections from all experiments were fixed for 60 min in 2% osmium tetroxide in PB, dehydrated through a graded series of ethanols and propylene oxide, and embedded in EMbed 812 (EMS) between two sheets of Aclar (Milner et al., 2011). Ultrathin sections (~70 nm) including the NAc core and shell were taken (Figure 1a) using a Leica UCT ultratome. The tissue was collected on copper grids (EMS) and was counterstained using Reynolds’ lead citrate and uranyl acetate. These grids were examined under a Philips CM10 electron microscope with an AMT digital camera.

2.6 |. Data analysis

The subcellular distributions of ERα, ERβ, and GPER1 alone, and ERα and GPER1 co-localized with either TH or GABA, were examined in the NAc core and shell. A profile was considered immunoreactive (IR) for immunogold labeling if it contained one or more gold particles. Two sections of 54 μm² were analyzed for the NAc shell and the NAc core, from either the right or left hemispheres, for each rat in all experiments. For quantification analyses ER-labeled profiles in each section were counted and categorized as: dendrites, dendritic spines, axons, axon terminals, or glia. The total number of labeled profiles in the two 55 μm² areas was calculated for each rat, and an average plus SEM was taken across the three rats. The number of each type of single or dual-labeled ER-containing profile was divided by the total number of profiles containing ER-IR to determine the percentage of each type of profile. Tissue selected for counting was taken from a depth of 0.2–1.5 μm from the plastic–tissue interface, and only samples that were thin sectioned evenly across the plastic tissue interface were included in these analyses. This study quantified the number of profiles containing immunolabeling for ERα, ERβ, and GPER1 profiles in the core and shell of the NAc, but no statistical analyses were used to compare the number of profiles expressing each type of ER.

The type of neuronal profile was determined using the description of ultrastructural morphology from Peters et al. (Peters et al., 1991). Dendrites were large profiles (usually between 1.0 and 2.0 μm) that contained regular microtubule arrays and were sometimes contacted by terminals. Dendritic spines were small (usually between 0.3 and 0.4 μm), sometimes contained a spine apparatus or budded from dendritic shafts and formed synaptic contacts with axon terminals. Axon profiles were less than 0.2 μm in diameter, contained a few small vesicles, and did not form synapses within the plane of section. Axon terminals had a cross-sectional diameter greater than 0.3 μm and contained numerous synaptic vesicles, and sometimes formed synapses with other neuronal profiles. Glial profiles were recognized by their conformation to the boundaries of other profiles and their lack of microtubules. Finally, soma were identified by their extremely large size, a lack of microtubules, and high numbers of cellular organelles. All sections were assessed for nuclear labeling, but soma were not included in the quantification analyses, since they typically occupy approximately half of the area counted for analysis, reducing the overall number of ER-IR profiles. Contact between neuronal profiles refers to symmetric and asymmetric synapses and appositions. Asymmetric synapses were identified by their thicker postsynaptic density, while symmetric synapses had thin, equal pre- and postsynaptic densities. Appositions were any contact between profiles that was not a synapse, as indicated by the absence of synaptic density in the plane of section.

2.7 |. Figure preparation

Light microscopy photomicrographs were adjusted for brightness and contrast to achieve uniformity between images using Microsoft PowerPoint 2010. Final electron photomicrographs were generated from digital images, and brightness and contrast were adjusted using GIMP 2.8, an open-source image editing software. Adjusted electron micrographs were assembled in Microsoft PowerPoint 2010, where additional changes were made to brightness and contrast to achieve a uniform appearance between images. Although images were adjusted, as described above, no alterations were made to immunolabeling.

3 |. RESULTS

3.1 |. Single labeling for ERs (Experiments 1 and 4)

By light microscopy, dense GPER1 and low levels of nuclear ERα or ERß are localized to the NAc core and shell.

Using light microscopy, moderate nuclear labeling for ERα (Figure 1b) and very low nuclear labeling for ERβ (Figure 1c) was observed in the in the NAc. In contrast to ERα or ERß, GPER1-IR was observed throughout the neuropil of the NAc, but there was no nuclear labeling for GPER1 (Figure 1d).

By EM, extranuclear ERα is observed in the NAc core and shell.

ERα-immunoreactivity was observed in all types of neuronal processes and glia in the NAc core and shell (Figure 2 and Tables 2 and 3). ERα immunoreactivity was discrete and was affiliated with the plasma membrane and/or clusters of small vesicles in axons (<0.2 μm in diameter). Semiquantitative analyses showed that 35% of ERα-IR profiles in the NAc core were axons, and 38% of ERα-IR profiles in the NAc shell were axons (Figure 2d). Analyses also demonstrated that 40% of ERα-IR profiles the NAc core and 44% of ERα-IR profiles in the NAc shell were in axon terminals (Figure 2a). Axon terminals in the NAc had cross-sectional diameters that were ~0.3–1.5 μm and contained numerous small synaptic vesicles (SSVs) and occasionally mitochondria. ERα immunoreactivity was commonly found in clusters of reaction product around SSV and was occasionally associated with mitochondrial and the plasma membrane, sometimes in proximity to synapses.

FIGURE 2.

Electron micrographs show examples of estrogen receptor α (Erα)-containing profiles in the nucleus accumbens (NAc) core and shell. ERα-immunoreactivity (IR) is observed in: a) a two axon terminals (TER), one that forms a synapse with an unlabeled dendritic spine (uSP), in the NAc core; b) at the membrane of a glial process (GL) in the NAc core; c) a dendrite (DEN), where it is associated with the membrane of a mitochondrion (mit), the cell membrane, and microtubules in the NAc shell; d) a dendritic spine (SP) that forms a synapse with an unlabeled terminal (uTER), and an axon (AX) in the NAc shell. Black arrow, immunoperoxidase for ERα. Scale bar, 500 nm

TABLE 2.

Estrogen receptor α (ERα), estrogen receptor β (ERβ), and G-protein coupled estrogen receptor (GPER1) distribution in neuronal profiles and glia in the nucleus accumbens (NAc) core

Receptor		ERα	ERß	GPER1
Dendrites	%	7.8	4.4	9.2
	# SEM	17.0 ±4.9	4.7 ±1.8	13.7 ±3.9
Spines	%	5.5	.9	2.0
	# SEM	12.0 ±5.0	1.0 ±0.6	3.0 ±1.5
Axons	%	35.1	49.1	40.2
	# SEM	76.0 ±10.1	52.0 ±11.8	60.0 ±7.0
Terminals	%	40.5	39.0	33.0
	# SEM	87.7 ±12.4	41.3 ±4.1	49.3 ±1.3
Glia	%	10.9	6.6	15.6
	# SEM	23.7 ±3.2	7.0 ±1.0	23.3 ±2.9
Total	%	100	100	100
	# SEM	216.3 ±18.2	106.0 ±9.1	149.3 ±9.9

Note: The percentage of total immunoreactive (IR) profiles and number of IR profiles, and the corresponding standard error (SEM), observed in ~6000 μm area of the nucleus accumbens (NAc) core, averaged across rats.

TABLE 3.

Estrogen receptor α (ERα), estrogen receptor β (ERβ), and G-protein coupled estrogen receptor (GPER1) distribution in neuronal profiles and glia in the nucleus accumbens (NAc) shell

Receptor		ERα	ERß	GPER1
Dendrites	%	5.8	8.1	6.7
	# SEM	12.3 ±2.4	8.0 ±3.1	10.3 ±0.9
Spines	%	3.3	1.7	3.9
	# SEM	7.0 ±4.0	1.7 ±0.3	6.0 ±2.5
Axons	%	37.9	45.6	41.7
	# SEM	81.0 ±4.0	45.0 ±7.6	63.7 ±9.6
Terminals	%	44.1	36.5	34.1
	# SEM	94.3 ±3.5	36.0 ±2.9	53 ±6.4
Glia	%	9.0	8.1	13.5
	# SEM	19.3 ±2.2	8.0 ±1.7	20.7 ±1.5
Total	%	100	100	100
	# SEM	214.0 ±2.3	98.7 ±9.5	152.7 ±1.3

Note: The percentage of total immunoreactive (IR) profiles and number of IR profiles, and the corresponding standard error (SEM), observed in ~6000 μm area of the nucleus accumbens shell, averaged across rats.

Peroxidase labeling for ERα was also observed at postsynaptic sites in the NAc. Dendritic shafts accounted for 7.8% of ERα-IR profiles in the NAc core, and 5.7% of ERα-IR profiles in the NAc shell (Figure 2c). Additionally, ERα immunoreactivity also was infrequently observed in dendritic spines; 5.5% of ERα-IR in the NAc core and 3.3% of ERα-IR in the NAc shell was localized to dendritic spines (Figure 2d). In the dendritic shafts, peroxidase reaction product was often affiliated with the mitochondrial and plasma membranes, and microtubules. In dendritic spines, immunolabeling for ERα sometimes accumulated in the spine head, and also was observed on the plasma membrane, particularly near the postsynaptic density. ERα was frequently observed near asymmetric synapses, where it was seen both pre- and postsynaptically. Occasionally, ERα-IR axon terminals synapsed onto ERα-IR dendrites. Immunolabeling for ERα was sometimes observed in perikaryal soma, associated with mitochondria or other cellular organelles and sometimes associated with the plasma membrane. Lastly, 10.8% of ERα-IR in the NAc core and 9.0% of ERα-IR in the NAc shell were observed in glial cells (Figure 2b), primarily at the plasma membranes.

By EM, extranuclear ERβ is observed in the NAc core and shell.

At the ultrastructural level ERß immunoreactivity was observed at extranuclear sites in all neuronal profiles and in glial cells in the NAc core and shell (Figure 3 and Tables 2 and 3). ERß immunoreactivity was most common in axons, where it constituted 49% of the total ERß-IR profiles in the NAc core, and 45% of total ERβ- IR profiles in the NAc shell (Figure 3c). In axons, immunolabeling was discrete and was localized to the plasma membrane or clusters of small vesicles. ERß immunoreactivity also was found in axon terminals, which accounted for 39% of the total ERβ-IR profiles in the NAc core and 36% of the total ERβ-IR profiles in the NAc shell (Figure 3a). Axon terminals containing ERß immunoreactivity contained numerous SSVs and occasional mitochondria, but did not contain dense core vesicles. ERß immunoreactivity was found in clusters of reaction product associated with SSVs and was sometimes affiliated with mitochondria and the plasma membrane.

FIGURE 3.

Electron micrographs show examples of estrogen receptor β (Erβ)-containing profiles in the nucleus accumbens (NAc) core and shell. a) ERβ-IR is observed in a terminal (TER), where it is localized to a mitochondrion (mit), small synaptic vesicles, and the plasma membrane. ERβ-IR is also associated with the membrane of a glial cell (GL) in the NAc shell; b) the membrane and microtubules of a dendrite (DEN) in the NAc core; c) an axon (AX). Black arrow, immunoperoxidase for ER. Scale bar, 500 nm

ERß immunoreactivity was infrequently observed at postsynaptic sites. Four percent of the total ERβ-IR profiles were dendrites in the NAc core, and 9% of ERβ-IR profiles were dendrites in the NAc shell (Figure 3b). ERβ immunoreactivity was almost never observed in dendritic spines, accounting for 0.9% of immunolabeling in the NAc core and 1.7% of immunolabeling in the NAc shell. In dendrites, immunoreactivity was typically associated with the plasma membrane or with mitochondria. ERß-IR perikarya rarely were observed in the NAc core or shell. ERß-IR terminals were not observed forming synapses with ERß-IR dendrites or spines. Lastly, ERß-IR glial cells were infrequently observed; 7% of the ERβ-IR profiles in the NAc core and 8% of the ERβ-IR profiles in the NAc shell were glia (Figure 3a). In glial cells, labeling was discrete and was localized primarily at the plasma membrane.

By EM, extrasynaptic GPER1 is observed in the NAc core and shell.

Immunoperoxidase labeling for GPER1 was also observed throughout both the NAc core and shell (Figure 4 and Tables 2 and 3). This labeling was associated with neurons and glia, and was found exclusively at extranuclear sites. Like ERα and ERß, most GPER1-IR profiles were presynaptic; GPER1-IR axons accounted for 40% of immunolabeling in the NAc core and 42% of immunolabeling in the NAc shell (Figure 4a,d). Axons containing GPER1-IR were small (<0.2 μm) and almost always unmyelinated. The labeling in axonal profiles was usually discrete, and often associated with the membrane and small clusters of vesicles. GPER1-IR axon terminals accounted for 33% of the total GPER1 in the NAc core and 34% of GPER1 in the NAc shell (Figure 4c). Axon terminals containing GPER1 immunoreactivity ranged from 0.3 to 1.5 μm, and contained numerous SSVs and occasionally mitochondria, where labeling was frequently observed. GPER1 immunoreactivity was also occasionally observed in proximity to synapses.

FIGURE 4.

Electron micrographs show examples of G-protein coupled ER 1 (GPER1)-containing profiles in the nucleus accumbens (NAc) core and shell. GPER1 immunoreactivity is observed in: a) an axon (AX) and associated with the membrane of a glial process (GL) in the NAc shell; b) a soma where it is associated with Golgi bodies. GPER1-immunoreactivity is also associated with the membrane of a glial cell (GL) in the NAc core; c) a dendrite (DEN), where it is associated with the membrane of a mitochondrion (mit), a dendritic spine (SP) and vesicles in an axon terminal (TER), in the NAc shell; d) in two axons in the NAc core. Black arrow, immunoperoxidase for ER. Scale bar, 500 nm

Low levels of GPER1 immunoreactivity were also observed in postsynaptic profiles. GPER1-IR dendritic shafts constituted 9% of total GPER1-IR profiles in the NAc core, and 7% of total GPER1-IR profiles in the NAc shell (Figure 4c). There were low levels of GPER1 immunoreactivity in dendritic spines: 2% of GPER1-IR profiles in the NAc core and 4% of GPER1-IR profiles in the NAc shell (Figure 4c). In the dendritic shafts, GPER1 was typically associated with the plasma and mitochondrial membranes, but also was affiliated with microtubules. In dendritic spines, GPER1 peroxidase reaction product accumulated in the spine head, and was associated with the plasma membrane. Although GPER1 immunoreactivity was observed both pre and postsynaptically, it was rare for GPR1-IR terminals to synapse onto GPER1-IR spines. Immunoperoxidase for GPER1 was observed in neuronal perikarya, where it was commonly associated with organelles, including mitochondria and Golgi bodies (Figure 4b). Finally, 16% and 14% of GPER1-IR profiles were observed in glia in the NAc core and NAc shell, respectively; the labeling in glial cells was discrete and was localized primarily to the plasma membrane (Figure 4a,b).

3.2 |. Experiment 2: Dual labeling for ERs and TH

In dual-labeled sections, immunoreactivity for both ERα and GPER1 was observed in similar proportions to that seen in experiment 1. There was higher total labeling for GPER1 in this dual-labeling study compared to the single-labeling experiment (Figure 5 and Tables 2–4). In agreement with previous studies (Sesack & Pickel, 1990), immunogold labeling for TH was observed throughout the NAc core and shell in axons and axon terminals. TH immunoreactivity most commonly was observed in terminals that were 0.3–1.5 μm in diameter and contained numerous closely packed round SSVs. TH-IR terminals sometimes formed symmetric synapses with dendrites and dendritic spines. TH immunolabeling was also infrequently observed in unmyelinated axons (0.1–0.2 μm diameter).

FIGURE 5.

Electron micrographs show examples of profiles containing of estrogen receptor α (Erα) or G-protein coupled ER 1 (GPER1) immunoreactivity and tyrosine hydroxylase (TH) immunoreactivity in the nucleus accumbens (NAc) core and shell. a) Immunoreactivity for ERα is associated with small synaptic vesicles and a mitochondrion in a catacholaminergic terminal (TER) that is adjacent to an unlabeled dendrite (uDEN). b) GPER1 immunoreactivity associated with synaptic vesicles close to the synapse in two catecholaminergic terminals (TER), and one noncatecholaminergic terminal (uTER). c) ERα-IR TER in close proximity to an uDEN. d) GPER1 immunoreactivity associated with synaptic vesicles in a TER, and an axon (AX). Black arrow, immunoperoxidase for ER; white arrow, immunogold for TH. Scale bar, 500 nm

TABLE 4.

Estrogen receptor α (ERα) and G-protein coupled estrogen receptor (GPER1) distribution in profiles containing tyrosine hydroxylase (TH) in the nucleus accumbens

		ERα or GPER-1 + TH in NAc core				ERα or GPER-1 + TH in NAc shell
Receptor		ERα	GPER-1	ERα + TH	GPER-1+TH	ERα	GPER-1	ERα + TH	GPER-1+TH
Dendrites	%	9.4	11.37	-	-	10.5	13.1	-	-
	# ±SEM	19.7 ±1.2	29.0 ±4.5	-	-	20.3 ±0.3	31.0 ±3.1	-	-
Spines	%	4.0	2.1	-	-	4.5	3.0	-	-
	# ±SEM	8.3 ±0.8	5.3 ±1.2	-	-	8.7 ± 0.7	7.0 ±1.5	-	-
Axons	%	38.3	35.4	1.25	6.6	38.1	40.4	7.2	2.8
	# ±SEM	80.0 ±9.7	90.3 ±2.9	1.0 ±0.8	6.0 ±1.6	74.0 ±9.3	95.7±5.8	5.3 ±0.7	2.7 ±1.2
Terminals	%	33.2	29.9	6.3	12.2	28.5	31.8	11.4	14.2
	# ±SEM	69.3 ±1.2	76.3 ±7.8	4.3 ±0.4	9.3 ±2.4	55.3 ±3.4	75.3 ±11.7	6.3 ±1.3	10.7 ±5.3
Glia	%	15.2	10.2	-	-	18.5	10.7	-	-
	# ±SEM	31.7 ±5.2	26.0 ±5.0	-	-	36.0 ±2.0	25.3 ±1.2	-	-
Total	%	100.0	100.0	2.6	6.5	100.0	100.0	6.0	5.7
	# ±SEM	209.0 ±15.0	255.0 ±7.6	5.3 ±0.4	16.7 ±4.9	194.3 ±9.7	236.7 ±13.5	11.7 ±1.6	13.7 ±6.9

The percentage of total immunoreactive (IR) profiles, the number of IR profiles, and the corresponding standard error (SEM), observed in ~6000 μm area of the nucleus accumbens (NAc) core and shell, averaged across rats.

There were low levels of colocalization between TH and both ERα and GPER1 in the NAc (see Table 4). In the NAc core, 1% of ERα-IR axons were also TH-IR and 6% of ERα-IR axon terminals contained TH immunoreactivity. In the NAc shell there were slightly higher levels of colocalization between ERα and TH: 7% of ERα-IR axons also contained TH immunoreactivity and 11% of ERα containing axon terminals were TH-IR (Figure 5a,c). Colocalization of GPER1 and TH was also observed in the NAc core and shell. In the NAc core, 7% of GPER1-IR axons also contained TH immunoreactivity and 12% of GPER1-IR axon terminals also contained TH immunoreactivity. In the NAc shell 2.8% of axons labeled for GPER1 were also TH-IR, and 14.2% of all GPER1-IR axon terminals were TH-IR (Figure 5b,d). These findings demonstrate that a small proportion of ERα- and GPER1-IR profiles in the NAc are catecholaminergic neurons.

3.3 |. Experiment 3: Dual labeling for ERs and GABA

The proportions of both ERα and GPER1 observed in the NAc core and shell were comparable to those observed in the single-label experiments, increasing confidence in these findings. Again, the total number of GPER1-IR profiles were higher in this experiment than in the single-label experiment (Tables 2, 3, and 5), but were comparable to the total number of GPER1-IR profiles observed in the dual-labeling experiment with TH (see Tables 4 and 5). Immunogold labeling for GABA-paralleled previous findings, with GABA immunoreactivity most commonly observed in terminals, dendrites, and perikarya of neurons in the NAc. GABA immunoreactivity was also occasionally observed in axons and dendritic spines, but this was infrequent.

TABLE 5.

Estrogen receptor α (ERα) and G-protein coupled estrogen receptor (GPER1) distribution in profiles containing γ-aminobutyric acid (GABA) in the nucleus accumbens (NAc)

		ERα or GPER-1 + GABA in NAc Core				ERα or GPER-1 + GABA in NAc Shell
Receptor		ERα	GPER-1	ERα + GABA	GPER-1+GABA	ERα	GPER-1	ERα + GABA	GPER-1+GABA
Dendrites	%	4.8	4.4	53.3	51.8	4.7	4.4	50.0	35.6
	# ±SEM	10.0 ±2.0	28.3 ±3.2	5.3 ±1.8	14.6 ±1.2	9.3 ±0.9	29.0 ±2.5	4.6 ±0.4	10.3 ±0.7
Spines	%	3.3	4.0	4.8	13.6	4.2	4.0	8.0	8.7
	# ±SEM	7.0 ±1.5	7.3 ±1.2	0.3 ±0.3	1.0 ±0.8	8.3 ± 0.3	7.7 ±0.7	0.7 ±0.8	0.7 ±0.7
Axons	%	37.5	36.8	3.4	3.2	39.1	36.8	1.7	1.3
	# ±SEM	79.0 ±2.9	94.0 ±1.5	2.7 ±1.7	3.0 ±1.6	77.0 ±3.2	75.7 ±2.3	1.3 ±0.4	1.0 ±0
Terminals	%	43.1	43.7	46.7	40.7	40.8	43.7	37.2	33.8
	# ±SEM	90.7 ±3.2	84.3 ±2.3	42.3 ±4.4	34.3 ±4.5	80.7 ±5.8	76.0 ±10.6	30.0 ±6.5	25.7 ±3.2
Glia	%	11.6	11.1	2.7	5.0	11.1	11.1	4.5	2.8
	# ±SEM	24.3 ±3.2	20.1 ±2.1	0.7 ±0.7	1.0 ±0.4	22.0 ±1.5	23.7 ±3.4	1.0 ±1.2	0.7 ±0.3
Total	%	100.0	100.0	24.4	23.3	100.0	100.0	19.1	18.1
	# ±SEM	210.3 ±5.6	237.0 ±5.2	51.3 ±2.2	55.3 ±0.8	197.6 ±7.3	212.0 ±7.0	37.7 ±4.1	38.3 ±3.5

The percentage of total immunoreactive (IR) profiles, the number of IR profiles, and the corresponding standard error (SEM), observed in ~6000 μm area of the NAc core and shell, averaged across rats.

There are moderate levels of colocalization between ERα and GABA immunoreactivity in the NAc. Colocalization between ERα and GABA-IR most commonly was observed in dendritic shafts in both the NAc core and shell. In the NAc core 53% of ERα-IR dendritic shafts were GABAergic, and in the NAc shell 50% of ERα-IR dendrites also contained GABA immunoreactivity (Table 5; Figure 6a). Colocalization between ERα and GABA was also frequently observed in axon terminals; 47% of ERα-IR terminals in the NAc core and 37% of ERα-IR terminals in the NAc shell also contained GABA-IR (Figure 6b). Lower levels of colocalization between ERα and GABA immunoreactivity were observed in axons and dendritic spines. Three percent of ERα-IR axons in the NAc core and 2% of ERα-IR axons in the NAc shell were GABAergic. Furthermore, 5% of ERα-IR dendritic spines in the NAc core and 8% of ERα-IR dendritic spines in the NAc shell also contained GABA immunoreactivity. Glial profiles containing immunolabeling for ERα and GABA were also observed infrequently, with 3% and 5% of ERα-IR glia also containing GABA immunoreactivity in the NAc core and shell, respectively. Additionally, ERα was occasionally observed in soma containing GABA immunoreactivity, although these profiles were not included in quantification analyses.

FIGURE 6.

Electron micrographs show examples of profiles containing of estrogen receptor α (Erα) or G-protein coupled ER 1-immunoreactivity (GPER1-IR) and GABA-IR in the nucleus accumbens (NAc) core and shell. a) ERα-IR is associated with microtubules and the plasma membrane of a γ-aminobutyric acid-ergic (GABAergic) dendrite (DEN). b) ERα-IR associated with synaptic vesicles and the membrane near a synapse in a GABAergic terminal (TER). c) GPER1-IR associated with synaptic vesicles in a GABAergic TER; d) a GABAergic dendrite (DEN) with a spine (SP). Black arrow, immunoperoxidase for ER; white arrow, immunogold for GABA. Scale bar, 500 nm

Levels of colocalization between GPER1 and GABA in the NAc core and shell was also observed. GPER1 immunoreactivity was most frequently colocalized with GABA immunoreactivity in dendritic shafts; 52% of GPER1-IR dendrites in the NAc core and 36% of GPER1-IR dendrites in the NAc shell were GABAergic (Table 5; Figure 6d). Colocalization was also observed in axon terminals, as 41% of GPER1-IR axon terminals in the NAc core and 34% of GPER1-IR terminals in the NAc shell also contained GABA immunoreactivity (Figure 6c). Lower levels of colocalization were observed in spines with 13% of GPER1-IR dendritic spines in the NAc core and 9% of the spines in the NAc shell also containing GABA immunoreactivity. Low levels of colocalization were observed between GABA and GPER1 in axons, with 3% of GPER1-IR axons in the NAc core and only 1% of GPER1-IR axons in the NAc shell, also containing GABA immunoreactivity. Finally, low levels of colocalization of GPER1 and GABA were also observed in glia in the NAc; 5% of GPER1-IR glia in the NAc core and 3% of GPER1-IR glia in the NAc shell also contained GABA immunoreactivity. GPER1 immunoreactivity was observed in GABAergic soma, but again these were not included in quantification analyses.

4 |. DISCUSSION

Ultrastructural analysis demonstrates that ERα, ERβ, and GPER1 are localized to extranuclear sites in the NAc core and shell of female rats. Although these receptors are observed at all types of neuronal profiles and in glia, the majority are observed at presynaptic sites. The distribution of the ERs in the NAc core and shell appear to be similar, suggesting that ERs do not differ between these subregions of the NAc. Additionally, these experiments indicate that a very low proportion of ERα and GPER1 are localized to catecholaminergic neurons, and a moderate proportion of ERα and GPER1 are localized to GABAergic neurons in the NAc.

4.1 |. Methodological considerations

Methodological considerations are discussed in detail in previous publications (Almey et al., 2012; Milner et al., 2001). Briefly, the immunolabelling methods used here lead to excellent preservation of cellular morphology allowing for discrete localization of antigens (Leranth & Pickel, 1989). All tissue sections were identical in size and were taken near the plastic tissue interface to prevent differences in antigen penetration from affecting the results of these experiments (Milner et al., 2011). Tissue sections were taken from similar but not identical regions of the NAc core and shell, so small differences in the proportion of profiles may be due to certain sections having a greater representation of specific profile types (i.e., axon tract). Immunoreactivity for ERα, ERβ, and GPER1 is discrete, and a very thin plane of section is examined, so a lack of ER immunoreactivity does not demonstrate that these profiles lack ERs. For the same reason, the probability of detecting both peroxidase and gold immunomarkers in the same plane of section, particularly for small profiles, is decreased. As a result, the semiquantitative analyses presented here are likely conservative, underestimating the number of ERs and the frequency with which these receptors are localized to profiles containing TH or GABA.

4.2 |. ERα is observed at extranuclear sites

ERα is the most frequently observed ER in the NAc, and results of the single- and dual-labeling experiments are similar, observing comparable levels of ERα-IR. Postsynaptic labeling is slightly lower in the NAc shell than core, but in general, these analyses indicate that ERα distribution is similar in these subregions of the NAc. ERα is observed in all types of neuronal profiles and in glial cells in both the NAc core and shell, differing from previous light microscopy experiments that observed relatively low levels of nuclear labeling for ERα in the NAc (Mitra et al., 2003; Shughrue et al., 1998). This discrepancy is likely due to the increased resolution of electron microscopy, allowing for detection of the mERα-IR profiles observed in this experiment.

Although ERα-IR is observed in all types of neuronal profiles, ERα is most commonly localized to presynaptic profiles, axons, and axon terminals. The presence of ERα in axons may simply reflect the transportation of this receptor to terminals but binding at receptors on axons can affect the transmission of action potentials and protein transport (Carr et al., 2010; Verdier et al., 2003). ERα in axon terminals is positioned to affect the transportation of vesicles to the synapse, which has been observed in hippocampal neurons (Hart et al., 2007). ERα in terminals is also positioned to affect the release of transmitter from terminal, providing a mechanism for the finding that E2 increases phasic DA release in the NAc (Thompson & Moss, 1994, 1997). ERα is also localized to dendrites and dendritic spines but is not frequently observed in these postsynaptic profiles. In addition to being localized to neurons, 10% of ERα is observed at the membrane of glial cells. Estrogens are involved in glial-mediated neuroprotection (Arevalo et al., 2010; Spence & Voskuhl, 2012), and these membrane-associated ERs provide a mechanism for this effect. These experiments observed ERα primarily at extranuclear sites, which corresponds to previous findings in the dorsal striatum (Almey et al., 2012).

4.3 |. ERβ is observed at extranuclear sites

These analyses found ERβ in all types of neuronal profiles and in glial cells in both the NAc core and shell. ERβ is observed approximately half as frequently as ERα in the NAc, suggesting that binding at ERβ would have less effect on transmission in the NAc. Our findings contrast those in a recent review, suggesting that ERβ is more prevalent in the NAc than ERα or GPER1 (Quigley et al., 2021). However, this review is based on studies by different research groups that looked at each ER independently in rats and mice using primarily light microscopic and/or molecular methods. A dual- or triple-label experiment assessing the density of these three ERs in the same tissue sections would help clarify the relative density of ERα, ERβ, and GPER1 in the NAc. Postsynaptic labeling for ERβ appears to be slightly lower in the NAc core than shell, although the difference is small and was not assessed using statistics. These experiments observe moderate levels of extranuclear ERβ-IR in the NAc, which differs from previous light microscopy findings that showed low levels of nuclear labeling for ERβ in this region (Mitra et al., 2003; Shughrue et al., 1998). Again, this discrepancy is likely due to the increased resolution of electron microscopy.

ERβ also is primarily observed presynaptically, in axons and axon terminals. Like ERα, ERβ in axons could affect the transmission of action potentials and protein transport (Carr et al., 2010; Verdier et al., 2003), and ERβ in axon terminals is positioned to affect the release of transmitter from terminal in the NAc. ERβ is also observed at postsynaptic profiles, including dendrites and dendritic spines, albeit at much lower levels. There are also glia in the NAc containing ERβ-IR, indicating that binding at these receptors could also contribute to estrogens’ effects on glial-mediated neurotransmission (Arevalo et al., 2010; Spence & Voskuhl, 2012). These experiments observed ERβ almost exclusively at extranuclear sites in the NAc, which corresponds to previous findings that estrogens rapidly alter transmission in the NAc (Thompson & Moss, 1994) and that binding at ERβ maintains D2 receptor levels in the NAc (Le Saux et al., 2006).

4.4 |. GPER1 is observed at extranuclear sites

GPER1 is also observed at all types of neuronal profiles in the NAc core and shell. To the best of our knowledge, this is the first study to report that GPER1 is prevalent in the NAc, as previous light microscopy experiments observe relatively low levels of GPER1 in this region (Hazell et al., 2009). These receptors are found at extranuclear sites using light microscopy, which was confirmed by ultrastructural analysis. The single- and dual-labeling studies show similar numbers of profiles containing GPER1-IR, suggesting that the distribution of GPER1 does not differ between subregions of the NAc. Although single- and dual-labeling studies for GPER1 observed a similar proportion of labeling in the different profile types, there is a discrepancy between the total number of labeled profiles observed. The single-labeling experiment observed ~25% less GPER1-IR profiles than the two dual-labeling studies. There are several potential causes for this discrepancy. It could result from the fact that sections in the single-labeling study may have been from slightly deeper in the tissue section reducing antibody permeability. Alternately, a difference in counterstaining of tissue could be responsible for the difference, resulting in darker background in the single-labeling study and greater difficulty observing peroxidase reaction product. Since the total number of profiles observed is very similar between the two dual-labeling studies, we believe that these studies are the most accurate representation of total GPER1 in the NAc, suggesting the GPER1 and ERα have similar prevalence in this region.

Like ERα and ERβ, GPER1 is observed primarily in presynaptic profiles in both the NAc core and shell. GPER1 in axons and axon terminals is positioned to affect the propagation of action potentials, protein transportation, and the release of neurotransmitters from terminals (Carr et al., 2010; Verdier et al., 2003). GPER1-IR is also observed in postsynaptic profiles in the NAc, including dendritic shafts and spines, but this occurred much less frequently. Glia in the NAc core and shell are IR for GPER1, suggesting that estrogens’ effects on glial-mediated neuroprotection could also occur through binding at this receptor (Arevalo et al., 2010; Spence & Voskuhl, 2012).

The single-labeling experiments clearly demonstrate that ERα, ERβ, and GPER1 are localized to extranuclear neuronal sites in the NAc. It appears that levels of these receptors do not differ between the core and shell subregions of the NAc. These data suggest similar levels of ERα and GPER1 in the NAc, while ERβ is less prevalent, occurring approximately half as frequently as the other ERs. All three ERs were most prevalent in presynaptic profiles of female rats, but were all observed in postsynaptic profiles, including dendritic spines. These postsynaptic receptors could contribute to the increased density and size of dendritic spines in the NAc of female, compared to male rats (Forlano & Woolley, 2010; Wissman et al., 2012). Paralleling findings in the dorsal striatum and hippocampus, ERα, ERβ, and GPER1 are commonly associated with the membranes of mitochondria in dendrites, terminals, and soma. Binding at ERs associated with mitochondria could contribute to estrogens’ effects on cellular metabolism observed in neurons and glia (Araújo et al., 2008; Klinge, 2017; Razmara et al., 2008). E2 infused into the NAc rapidly increases phasic DA release (Thompson & Moss, 1994) and increases D2 receptor binding without affecting D2 mRNA, suggesting that these changes in D2 receptor density occur via nongenomic mechanisms (Le Saux et al., 2006). Binding at the membrane-associated ERα, ERβ, or GPER1 observed in these experiments could cause these rapid E2-induced changes in transmission in the NAc.

4.5 |. Low levels of ERα and GPER1 are localized to catecholaminergic neurons in the NAc

The NAc contains predominantly GABAergic neurons; the majority of interneurons and all medium spiny projection neurons contain GABA, and the remaining interneurons a cholinergic (reviewed in Haber et al., 2012). There are also extrinsic inputs that synapse onto medium spiny neurons including glutamatergic projections from the cortex, thalamus, and brainstem and dopaminergic projections from the substantia nigra and ventral tegmental area (Haber et al., 2012). Thus, dual-labeling experiments next examined whether ERα and GPER-1 were localized to catecholaminergic axons and terminals to determine whether binding at ERs in the NAc could have direct effects on dopaminergic transmission. These experiments demonstrate that a low proportion of ERα-IR and GPER1-IR are localized to profiles containing TH. The highest level of colocalization between both ERα and GPER1 and TH is observed in catecholaminergic terminals; if these neurons are in fact DAergic, binding here could account for the rapid effects of estrogens on presynaptic DA transmission in the NAc (Davis et al., 2005; Thompson & Moss, 1994, 1997; Yoest, Quigley, et al., 2018). Alternately, these mERs could be working in conjunction with mERs on GABAergic profiles (see below) to affect DA transmission in the NAc.

4.6 |. Moderate levels of ERα and GPER1 are localized to GABAergic neurons in the NAc

These analyses demonstrate that a relatively large proportion of ERα and GPER1 are localized to GABAergic neurons in the NAc core and shell. There are similar levels of ERα-GABA colocalization and GPER1-GABA colocalization in the NAc core and shell, although colocalization of ERα-IR and GABA-IR is slightly lower in the NAc shell than in the NAc core. The greatest proportion of ERα and GPER1 colocalization with GABA is observed in dendritic shafts, as ~50% of ERα and GPER1-IR dendrites are GABAergic. A substantial proportion of colocalization is observed in axon terminals, as ~40% of ERα and GPER1-IR terminals are GABAergic. Low levels of colocalization between ERα/GPER1 and GABA are seen in axons and dendritic spines, likely due, in part, to the low levels of GABA immunoreactivity observed in these profiles (Pickel et al., 1988). Finally, ERα and GPER1 are infrequently localized to GABA-IR glial cells. These findings clearly demonstrate that some of the ERα and GPER1 in the NAc are localized to glia and GABAergic neurons. This parallels findings in the dorsal striatum, which showed ERα and GPER1 in GABAergic dendrites and terminals (Almey et al., 2016). Taken together, that previous study and the current experiments suggest similarities in the distribution of mERs in the dorsal striatum and the NAc (the ventral striatum).

Little research has examined the effect of estrogens on GABA in the NAc, but the presence of ERα and GPER1 in GABAergic neuronal profiles suggests that estrogens could alter GABA transmission. Indeed, elevated E2 following E2 administration to OVX rats or during the proestrus phase of the estrous cycle is associated with decreased excitability of medium spiny projection neurons in the NAc core (Alonso-Caraballo & Ferrario, 2019; Proaño et al., 2020; Proaño & Meitzen, 2020), that are presumably GABAergic (Haber et al., 2012). ERα and GPER1 on GABAergic dendrites and dendritic spines could contribute to this E2-indiced change in excitability. Additionally, ERα and GPER1 on GABAergic terminals have the potential to affect GABA release in the NAc. In the dorsal striatum, estrogens rapidly (<1 h) decrease extracellular GABA concentration (Hu et al., 2006; Schultz et al., 2009). The distribution of ERα and GPER1 appear to be similar in the dorsal striatum and NAc (Almey et al., 2012, 2016), so future studies may examine whether estrogens have a similar affect on GABA concentration in the NAc.

Further, changes in GABA transmission in the NAc affect can affect the DA availability. Decreasing GABA transmission by antagonizing GABA_A or GABA_B receptors increases phasic DA release (Melchior et al., 2015; Xi & Stein, 1998), while increasing GABA transmission by agonizing GABA_A or GABA_B receptors decreases phasic DA release (Brodnik et al., 2018; Pitman et al., 2014). Further modafinil and neurotensin –induced changes in DA release in the NAc are mediated by GABA (Ferraro et al., 1996; Tanganelli et al., 1994). These findings suggest that estrogens can alter GABA transmission in the NAc, so estrogens may indirectly increase DA availability in the NAc by reducing GABA. Additionally, TH terminals often synapse onto dendrites of spiny interneurons in the NAc, which are presumed to be GABAergic (Sesack & Pickel, 1990). The ERα and GPER1 seen in GABAergic dendrites in these experiments are ideally positioned to alter transmission at these synapses. This provides a second mechanism through which estrogens could alter DA transmission in the NAc.

Although approximately one third of ERα and GPER-1 are localized to GABAergic or catecholaminergic profiles, the remaining receptors observed in the dual-labeling studies are localized to an unknown neuron type. The NAc receives strong glutamatergic innervation, so it is possible that some of the presynaptic ERα, ERβ, and GPER1-IR observed here are localized to glutamatergic axons and terminals. A series of electrophysiological studies have demonstrated that during the proestrus phase of the cycle, when estradiol is high, or following application of estradiol to slice preparation, medium spiny neurons in the NAc core exhibit increased mEPSC amplitude (Krentzel et al., 2019), decreased mEPSC frequency (Krentzel et al., 2019; Proaño et al., 2020), and increased rheobase (Proaño et al., 2018, 2020). The presynaptic mERs observed here provide a potential mechanism for the rapid effects of menstrual cycle phase and estradiol administration on glutamatergic signaling in the NAc core (Krentzel et al., 2019; Tonn Eisinger et al., 2018; Yoest, Quigley, et al., 2018). Future research should examine whether ERα, ERβ, and GPER-1 are localized to glutamatergic neurons in the NAc.

5 |. CONCLUSION

This study is the first to observe and quantify mERs in the NAc, providing a mechanism for the rapid effects of estrogens on transmission in the NAc. All three ERs are predominantly localized to presynaptic sites where they are positioned to affect the transport and release of transmitters. These results provide a mechanism for ERα and ERβ effects in the NAc, where few nuclear receptors are observed, and demonstrate that GPER1 is present in this region at higher levels than previously thought. Low levels of ERα and GPER1 are localized to catecholaminergic neurons so estrogens could directly affect DA transmission. Higher levels of ERα and GPER1 are found in GABAergic neurons; estrogens could directly influence GABA transmission to indirectly affect DA transmission.

DATA AVAILABILITY STATEMENT

Raw data are available upon request.