Estrogen Receptor Beta Expression in the Mouse Forebrain: Age and Sex Differences

Abstract

Estrogen receptors regulate multiple brain functions including stress, sexual, and memory associated behaviors as well as control of neuroendocrine and autonomic function. During development, estrogen signaling is involved in programming adult sex differences in physiology and behavior. Expression of estrogen receptor alpha changes across development in a region specific fashion. By contrast, estrogen receptor beta (ERβ) is expressed in many brain regions, yet few studies have explored sex and developmental differences in its expression largely due to the absence of selective reagents for anatomical localization of the protein. In this study, we utilized bacterial artificial chromosome transgenic mice expressing ERβ identified by enhanced green fluorescent protein (EGFP) to compare expression levels and distribution of ERβ in the male and female mouse forebrain on the day of birth (P0), postnatal day 4 (P4) and P21. Using qualitative analysis, we mapped the distribution of ERβ–EGFP and found developmental alterations in ERβ expression within the cortex, hippocampus, and hypothalamic regions including the arcuate, ventromedial, and paraventricular nuclei. We also report a sex difference in ERβ in the bed nucleus of the stria terminalis with males showing greater expression at P4 and P21. Another sex difference was found in the anteroventral periventricular nucleus of P21, but not P0 or P4 mice, where ERβ-EGFP-ir cells were densely clustered near the 3^rd ventricle in females but not males. These developmental changes and sex differences in ERβ indicate a mechanism through which estrogens may differentially affect brain functions or program adult physiology at select times during development.

Introduction

The actions of estrogen in brain are mediated by estrogen receptors (ERs), including estrogen receptor alpha (ERα) and estrogen receptor beta (ERβ). In adulthood, activation of ERα has been shown to predominantly regulate reproductive neuroendocrine function and sexual behaviors (Lindzey et al., 1998; Wersinger et al., 1997), whereas ERβ has been shown to control many non-reproductive functions including learning and memory (Jacome et al., 2010), depressive- and anxiety-like behaviors (Weiser et al., 2009; Donner and Handa, 2009, Lund et al., 2005; Oyola et al., 2012), hypothalamic-pituitary-adrenal (HPA) axis reactivity (Weiser et al., 2009; Lund et al., 2006), and autonomic processes (Shih, 2009). Both ERα and ERβ are widely distributed in brain, particularly within regions that regulate the aforementioned functions (Merchenthaler et al., 2004; Mitra et al., 2003).

Although ERβ has been shown to be involved in the sexual differentiation of mouse behavior (Kudwa et al, 2005, 2006) and brain structure (Bodo et al, 2006), at present little is known about the developmental expression of this receptor. A recent study in rats indicates that ERβ mRNA expression in the forebrain changes greatly during the first few postnatal weeks of life (Cao and Patisaul, 2011). Furthermore, ERβ is expressed in a sexually dimorphic manner in the rat brain with females showing a greater expression in the anteroventral periventricular nucleus (AVPV; Orikasa et al., 2002) and the ventromedial hypothalamus (VMH; Ikeda et al., 2003). In mice, sexual dimorphisms in expression have also been reported with females having a greater number of ERβ-immunoreactive (ir) cells in the lateral septum, lateral amygdala, and endopyriform cortex (Milner et al., 2010). It has also been reported that female rats express considerably more ERβ in the caudal VMH than do males during early development (Ikeda et al, 2003). Nonetheless, little is known about possible changes in ERβ protein expression and the distribution of ERβ expressing neurons in the brains of mice during early postnatal development, nor have sex differences been explored during this time period. This is partially due to the lack of specific antibodies for rodent ERβ (Snyder et al., 2010) which have made these experiments in brain difficult to perform.

In the present study, we investigated the development of, and sex differences in ERβ expression using bacterial artificial chromosome (BAC) transgenic mice expressing ERβ (Esr2) which is identified by enhanced green fluorescent protein (EGFP; BAC-based Esr2-EGFP mice). Our findings indicate that ERβ expression changes dramatically during postnatal development in several regions throughout the forebrain and sex differences become apparent in discrete brain regions in an age dependent fashion.

Materials and Methods

Animals

Transgenic (BAC) Esr2-EGFP (ID169Gsat/Mmucd) mice were originated by Dr. Nathaniel Heintz of the gene expression nervous system atlas (GENSAT) project at Rockefeller University and obtained from the Mutant Mouse Regional Resource Center (MMRRC, UC Davis). This mouse line contains multiple copies of a modified BAC in which an EGFP reporter gene was inserted immediately upstream of the coding sequence of the Esr2 gene. As a result the EGFP coding sequence is placed under control of the Esr2 upstream promoter. Multiple copies of the BACs are inserted into the genome and as a result may allow detection of sites with low levels of endogenous expression. The native Esr2 gene in these mice is unaltered by this procedure and is reported to function normally (Milner et al., 2010). The generation of these mice and validation of ERβ specificity of EGFP expression has been previously described (Milner et al., 2010) and shown to be highly selective. Mice were originally generated on a FVB/N background and, at the University of Arizona, were subsequently crossed onto a C57BL/6 line for at least 5-6 generations by mating female heterozygous BAC Esr2-EGFP with C57BL/6 males obtained from Jackson Labs (Sacramento, CA). Offspring were weaned at 21 days of age and maintained under a 14/10 L/D cycle (lights on at 0600), with food and water available ad libitum. All offspring were genotyped at the time of euthanasia using the PCR protocol suggested by the MMRRC. All procedures were approved by the Arizona State University Institutional Animal Care and Use Committee, under subcontract from the University of Arizona College of Medicine-Phoenix, and were in accord with National Institutes of Health guidelines.

Perfusion and Tissue Processing

Mice were cryoanesthetized on ice and intracardially perfused either on the day of birth (P0) or on postnatal day 4 (P4) with 5 (P0), or 10 (P4) ml of phosphate-buffered 4% paraformaldehyde. P21 mice were administered a 100 mg/kg intraperitoneal overdose of Nembutal, and intracardially perfused with 15 ml of phosphate-buffered 4% paraformaldehyde. Brains were removed from the skull and placed in the same fixative at 4°C overnight. The following morning, brains were transferred into a 30% sucrose cryoprotectant solution, where they remained at 4° C until sectioning. For immunohistochemistry, brains were sectioned coronally at 35μm into 2 (P0 and P4) or 3 (P21) alternate series at −20°C using a Leica CM3050S cryostat (Leica, Buffalo Grove, IL). Tissues were placed in cryopreservative at 4° C until immunohistochemistry was performed. Tail clips were taken after anesthetization, prior to perfusion, and were subsequently used to identity Esr2-EGFP positive mice. Briefly, genotyping used a protocol provided by the MMRRC (http://mmrrc.ucdavis.edu/doc/GENSAT-EGFPGeno_Protocol.pdf) using the nucleotide sequences: CCT ACG GCG TGC AGT GCT TCA GC as the forward EGFP primer and CCT ACG GCG TGC AGT GCT TCA GC as the reverse primer to yield a 300 bp band indicating the presence of EGFP. PCR for actin was used as a control gene.

Immunohistochemistry

For visualization of ERβ-EGFP, sections were rinsed in tris-buffered saline (TBS; pH 7.6), incubated in 1% hydrogen peroxide and 0.4% Triton-X in TBS (TBS-TX) for 10 minutes, again rinsed in TBS, then incubated in 4% normal goat serum (NGS) in TBS-TX for 1 hour. After rinsing in TBS, tissue was incubated in primary antisera for GFP (1:10,000, rabbit, Torrey Pines BioLabs, Inc, TP401, Table 1) in 4% NGS and TBS-TX overnight at room temperature. This antisera was generated using E. Coli full length GFP as the immunogen and reacts with GFP and GFP variants EGFP and EBFP. Tissue was then rinsed in TBS and incubated for 1 hour in biotinylated goat-anti rabbit antisera in TBS-TX (1:1000; Vector Laboratories, Burlingame, CA) followed by rinses in TBS and a 1 hour incubation in avidin-biotin complex (ABC Elite kit, Vector Laboratories). Following rinses in TBS, tissue was then developed for visualization of ERβ-EGFP positive cells using diaminobenzidine as the chromogen. Subsequently, sections were rinsed in TBS and mounted on gelatin-coated slides. The following day, sections were dehydrated in ethanol, defatted in xylene, and coverslipped with Permount (Sigma, St. Louis, MO, USA). Wild type brain sections, used as negative controls to test for EGFP-ir, showed no labeling. An alternate series of sections from each mouse brain was slide mounted and counterstained with cresyl violet. These sections were used to match to adjacent sections used for immunohistochemistry and determine the area of the regions selected for quantification of ERβ-EGFP positive cells.

Primary Antibody used

Antigen	Immunogen	Source	Dilution
Green Fluorescent Protein (GFP)	E. coli-expressed full length GFP	Torrey Pines BioLabs Inc. (Houston, TX), rabbit polyclonal, #TP401	1:10,000

Microscopic Analysis of ERβ distribution

Analysis of ERβ-EGFP distribution and density was conducted on an Olympus IX81 microscope (Olympus, Center Valley, PA) equipped with a Hamamatsu Orca-ER digital camera (Hamamatsu Photonics, Hamamatsu City, Japan). Brain regions were identified using the mouse brain atlas of Franklin and Paxinos (2007). Two independent investigators, blinded to treatment group, determined immunoreactive cell density and intensity by assigning ratings based on the density and intensity of immunoreactive product within selected brain regions. Qualitative ratings were created by generating a score for each brain region in individual mice (N=4 per age/sex, except P4 females N=3). Density and intensity values were equally weighted and averaged to obtain an overall density/intensity score for each individual. Values were applied on the following scale. –, Absence of label; *, minimal labeling intensity and low density of cells within a given brain region; **, moderate staining intensity and moderate density of labeled cells within a region; ***, high labeling intensity and high density of cells within a region; ****, very high labeling intensity and very high density of labeled cells within a region. Group means and standard error of the mean (SEM) were then generated from these values and are shown in Figure 1. Photomicrographs used in figures were captured at 4, 10, and 20x and only manipulated for size, sharpness, contrast, and brightness in Adobe Photoshop 6.0 (Adobe Systems Inc,, San Jose, CA).

Figure 1. Distribution of ERβ-EGFP-ir cells in the developing mouse forebrain.

Intensity/density of immunoreactive signal in P0, P4, and P21 mouse forebrain was scored using the following scale. –, Absence of label; *, minimal labeling intensity and low density of cells within a given brain region; **, moderate staining intensity and moderate density of labeled cells within a region; ***, high labeling intensity and high density of cells within a region; ****, very high labeling intensity and very high density of labeled cells within a region. Density and intensity values were equally weighted and averaged to obtain an overall density/intensity score for each individual. (A-H) Regions which did not show sex differences. (A) Cortex, (B) septum and diagonal bands, (C) rostral hypothalamus, (D) medial hypothalamus, (E) caudal hypothalamus, (F) hippocampus and habenula, and (G-H) amygdala. (I-L) Brain regions which showed sex differences. (I) Bed nucleus of the stria terminalis (principal nucleus), (J) medial preoptic area, (K) anteroventral periventricular nucleus, and (L) dorsomedial nucleus. Med septum, medial septum; lat septum, lateral septum; hor diag band, horizontal diagonal band; lat diag band, lateral diagonal band; OVLT, organum vasculosum of lamina terminalis; SCN, suprachiasmatic nucleus; PVN, paraventricular hypothalamic nucleus; Peri-VN, periventricular hypothalamic nucleus; hyp, hypothalamus; VMHdm, ventromedial hypothalamus dorsomedial division; VMHc, ventromedial hypothalamus central division; VMHvl, ventromedial hypothalamus ventrolateral division; med eminence, median eminence; BNST, bed nucleus of the stria terminalis (principal nucleus); MPOA, medial preoptic area; AVPV, anteroventral periventricular nucleus; DMH, dorsomedial nucleus. N=4 per age/sex, except P4 females N=3. Shown as means ± SEM.

To confirm qualitative analyses, quantitative analyses were subsequently performed in selected brain regions (AVPV, arcuate nucleus, VMH, and bed nucleus of the stria terminalis principal nucleus (BNST)), which showed developmental and/or sex differences. For analysis of potential sex/developmental differences, images of each of these brain regions were captured on an Olympus IX81 microscope using a 4x objective The anatomical borders of the AVPV, arcuate nucleus, VMH, and principal nucleus of the BNST were identified in cresyl violet stained sections and traced using Slidebook software (Intelligent Imaging Innovations, Inc., Denver, CO, USA). These region traces were then transposed onto images of adjacent sections of the same mouse brain that were labeled for ERβ-EGFP. ERβ-EGFP positive cells were counted within these traces to obtain an estimate of ERβ-EGFP cells per mm². Quantifications were performed bilaterally within two atlas matched sections for the AVPV, arcuate nucleus, and VMH. The principal nucleus of the BNST appeared prominently within the P0 and P4 brain in only 1 section. Therefore, only one bilateral section per animal was analyzed for this nucleus. This method was utilized to verify our qualitative assessment by counting the number of cells in regions that appeared to differ in our visual analysis. These analyses are not designed to estimate the entire population of cells within a given anatomical region but rather provide an estimate of relative cell density within an identified portion of the brain region. Statistical analyses were performed by 2-way ANOVA with age (P0, P4, or P21) and sex (male or female) as factors (N=4 per age/sex, except P4 females N=3). All significant main effects or interactions were further analyzed using Students t-tests. Differences were considered significant when p<0.05 and all data are reported as means ± SEM.

Results

Distribution of ERβ in the developing male and female mouse forebrain

AVPV and Medial Preoptic area

A robust sex difference in ERβ-EGFP distribution was found in the P21 mouse AVPV with females, but not males, exhibiting a dense distribution of ERβ-EGFP in cells bordering the 3^rd ventricle. This sex difference was not present at earlier time points (P0, P4; Figure 1K, Figure 2). In the medial preoptic area (MPOA; Figure 1J) there was a very dense distribution of ERβ-EGFP-ir cells at P0, with a lesser but still dense distribution found at later time points. At P4 there appeared to be a transient sex difference in the density of ERβ-EGFP cells with females showing greater immunoreactivity. No sex differences were found in the MPOA at P0 or P21.

Figure 2. Expression of ERβ-EGFP in the AVPV.

At P0 (A-B) and P4 (C-D) both male and female mice display scattered ERβ-EGFP label in the AVPV. At P21 (E-H) a clear sex difference in the distribution of ERβ-EGFP is present. In females ERβ-EGFP-ir cells are densely distributed in the medial AVPV bordering the 3^rd ventricle, while in males few cells are found within this region. The medial AVPV in E-F is further magnified in G-H. Dashed lines indicate the anatomical borders of the AVPV as indicated in a mouse brain atlas (Franklin and Paxinos, 2007).

Cortex

The distribution of ERβ-EGFP-ir cells in cortical areas varied dramatically during postnatal development (Figure 3A-C; Figure 1A). At P0 (Figure 3A) immunolabeling is concentrated in the upper cortical plate, just beneath the marginal zone (MZ). ERβ-EGFP positive cells also showed a distinct labeling pattern within middle cortical layers as well as the deep cortical layer bordering dorsal to the corpus callosum. At P4 (Figure 3B), labeling remained strong in the upper cortical plate and the number of labeled cells within the middle cortical layers increased while immunolabeling near the corpus callosum decreased. At P21 (Figure 3C) ERβ-EGFP-ir cells were confined to deep cortical layers, with the greatest expression found in layer 5. ERβ-EGFP positive cells bordering the corpus callosum were fewer compared to P0 and P4.

Figure 3. ERβ-EGFP distribution in the cortex and paraventricular nucleus of the developing mouse brain.

(A-C) Representative images of the cerebral cortex. Note that ERβ-EGFP positive cells are confined to deep cortical layers at P21 (C), while at earlier time points labeling is also found in the upper cortical plate and middle cortical layers (A-B). (D-F) Representative image of the paraventricular nucleus (PVN) and surrounding area in P0, P4, and P21 mice. Arrows in D-F indicate the region directly surrounding the PVN which contains few ERβ-EGFP-ir cells at P0 (D) and 4 (E), while at P21 (F) numerous labeled cells are found. The PVN and surrounding region are further magnified in G-I. MZ, marginal zone; CC, corpus callosum; L5, Layer 5; L6, Layer 6.

Hypothalamus

The paraventricular nucleus of the hypothalamus showed extensive ERβ-EGFP-ir throughout postnatal development with the greatest labeling found at P21 (Figure 1D, Figure 3D-I). The region directly surrounding the PVN contained very few ERβ-EGFP cells in early postnatal development but contained numerous scattered cells by postnatal day 21 (Figure 3D-I). Similarly, the anterior hypothalamic area also showed the greatest ERβ-EGFP-ir at P21. In the caudal hypothalamus ERβ-EGFP was present within the dorsomedial and ventromedial hypothalamus as well as the arcuate nucleus (Figure 1E,L; Figure 4A-F). Labeling in the ventromedial nucleus was largely confined to the ventrolateral division in juvenile mice, while at earlier ages (P0/P4) labeling was also prominent in the central division. Overall, greater levels of ERβ-EGFP-ir were found in P0 and P4 VMH compared to P21. No sex differences were apparent in these regions. The perifornical region of P21 mice also showed extensive ERβ-EGFP immuno-labeling, which was not present in neonatal mice (Figure 4A-F).

Figure 4. ERβ-EGFP distribution in the caudal hypothalamus of the developing mouse brain.

(A-C) ERβ-EGFP-ir is present within the dorsomedial and ventromedial hypothalamus (VMH) as well as the arcuate nucleus. ERβ-EGFP-ir was greatest at P0 and P4 in the VMH, while in the arcuate nucleus ERβ-EGFP was greatest at P4 and P21. (D-F) Higher magnification images of the VMH and arcuate nucleus indicating the anatomical borders of these regions used for qualitative and quantitative analysis. All images were from male mice. VMHvl, ventromedial nucleus ventrolateral division; PF, perifornical region; DMH, dorsomedial hypothalamus; Arc, arcuate nucleus.

The arcuate nucleus showed widespread labeling at P4 and P21, with a much weaker expression at P0 (Figure 4A-F). No sex differences were found in ERβ-EGFP-ir in the arcuate nucleus. Ventral to the arcuate nucleus, ERβ-EGFP was also localized in the median eminence (Figure 4A-F). Expression in this region was greatest in P21 mice and was not sexually dimorphic. In the suprachiasmatic nucleus (SCN; Figure 1C; Figure 5A-C), ERβ-EGFP labeled cells were largely confined to the shell region with scattered ERβ-EGFP-ir located within the core division. No apparent sex differences were found in the SCN. Neonatal mice also showed more extensive ERβ-EGFP-ir in the SCN than did juveniles, particularly in the lateral shell area (Figure 5A-C).

Figure 5. ERβ-EGFP distribution in the suprachiasmatic nucleus (SCN) and bed nucleus of the stria terminalis (BNST).

(A-C) ERβ-EGFP-ir cells are highly expressed in the SCN of P0, P4, and P21 mice with labeling largely confined to the shell region. In neonates (P0-P4) greater expression is found in the lateral portions of the SCN shell compared to P21 mice. All SCN images were from female mice. In the BNST principal nucleus a sex difference was found in which males (D) show a greater expression of ERβ-EGFP-ir cells compared to females (E) at P21.

Hippocampus and habenula

ERβ-EGFP-ir cells were prominent in the CA1 region of P0 mice and expression was decreased by P4 but appeared to increase again slightly by P21 (Figure 1F; Figure 6A-C, K). In CA3 ERβ-EGFP levels were highest at birth and continued to decline through P21 at which point scattered cell labeling was found (Figure 1F; Figure 6A-C, J). No ERβ immunolabeling was found in neonatal (P0/P4) dentate gyrus, although at P21 scattered labeled cells were present particularly within the hilus region (Figure 1F; Figure 6A-C, L). The lateral habenula (Figure 1F) was comprised of dense and strongly labeled ERβ-EGFP positive cells at all time points examined with the greatest intensity found at P21. No immunoreactive cells were found within the medial habenula (Figure 1F).

Figure 6. ERβ-EGFP distribution in the developing mouse hippocampus, amygdala, and septum.

(A-C) Representative images of the hippocampus in P0, P4, and P21 mice. Note the higher levels ERβ-EGFP-ir in CA1 at P0, and in CA3 at P0 and P4 compared to P21. In contrast, no ERβ-EGFP immunolabeling is found in neonatal (P0/P4) dentate gyrus, while at P21 scattered labeled cells are present particularly within the hilus region. Representative images of the amygdala (D-F). Note the high levels of ERβ-EGFP-ir in the medial, basolateral, and cortical regions compared to the central and lateral divisions which are largely devoid of ERβ-EGFP. (G-I) The lateral septum shows extensive ERβ-EGFP immunoreactivity across all ages, while ERβ-EGFP-ir in the medial septum is greatest at P21. Higher magnification images of ERβ-EGFP-ir cells in hippocampal regions (J-L), amydala (M), and septum (N) from P21 mice are shown. All images were from male mice. Hil, hilus; La, lateral amygdala; Bla, basolateral amygdala; Bma, basomedial amygdala; Mea, medial amygdala; Ca, central amygdala; Coa, cortical amygdala; LS, lateral septum; MS, medial septum.

Amygdala and Bed Nucleus of the Stria Terminalis

In the amygdala (Figure 1G-H; Figure 6D-F, M), the highest levels of ERB-EGFP-ir were present in the medial amygdala, with the basolateral and cortical regions also showing considerable immunolabeling. By contrast, the central and lateral divisions were largely devoid of ERβ. ERβ-EGFP-ir levels were similar across ages and sex throughout amygdala subregions. A sex difference was found in the BNST (Figure 1I) with males showing greater ERβ-EGFP-ir at P4 and P21 (Figure 5D-F). This sex difference was not apparent at P0.

Septum, horizontal and vertical diagonal bands

The lateral septum contained perhaps the greatest expression of ERβ in the mouse brain (Figure 1B; Figure 6G-I, N). Lateral septum ERβ-EGFP levels were high at neonatal time points and remained high at P21. On the contrary, expression in the medial septum increased during development with little expression found at P0 and P4 and a moderate expression found at P21 (Figure 1B; Figure 6G-I, N). ERβ-EGFP was highly expressed in both the horizontal and vertical diagonal bands with no change across development. (Figure 1B).

Quantification of ERβ in select mouse brain regions

In order to provide a quantitative verification of qualitative observations, we counted ERβ-EGFP-ir cells within four selected brain regions (AVPV, arcuate nucleus, BNST and VMH). Two-way ANOVA of ERβ-EGFP-ir cells in the AVPV revealed a significant main effect of sex (F(1,17) = 21.91, p<0.001), age (F(2,17) = 7.02, p<0.01), and a significant age × sex interaction (F(2,17) = 12.27, p<0.001). Post hoc analyses revealed a significant sex difference (p<0.001) in the number of ERβ-EGFP positive cells in P21 mouse AVPV, indicating a greater density in females compared to males (Figure 7A).

Figure 7. Quantification of ERβ-EGFP in selected brain regions.

ERβ-EGFP positive cells were counted in brain regions of P0, 4, and 21 mice. (A) In the medial AVPV, females show a greater number of ERβ-EGFP-ir cells at P21, with no sex differences found at earlier time points. (B) The number of ERβ-EGFP-ir cells in the arcuate nucleus was decreased at P0 compared to P4 and P21, with no sex differences found. (C) In the VMH a greater number of ERβ-EGFP positive cells were found at P0 and P4 compared to P21, with no sex difference. (D) The number of ERβ-EGFP positive cells in the principal nucleus of the BNST increased with age and was greater in males than females at P4 and P21. * indicates p<0.05, ** indicates p<0.001. AVPV, anteroventral periventricular nucleus; VMH, ventromedial hypothalamus; BNST, bed nucleus of the stria terminalis.

Two-way ANOVA of ERβ-EGFP-ir cell counts in the arcuate nucleus revealed a significant main effect of age (F(2,17) = 16.15, p<0.001), indicating significantly fewer ERβ-EGFP positive cells at P0 compared to P4 and P21 (Figure 7B). No effect of sex, or an age × sex interaction were found. Two-way ANOVA of ERβ-EGFP-ir cells in the VMH revealed a significant main effect of age (F(2,17) = 93.34, p<0.001), indicating a significant decrease in ERβ-EGFP expressing cells at P21 compared to P0 and P4 (Figure 7C). No effect of sex, or an age × sex interaction were found. For ERβ-EGFP-ir cells the BNST, two-way ANOVA revealed a significant main effect of age (F(2,17) = 438.20, p<0.001), sex (F(1,17) = 10.02, p<0.01), and an age × sex interaction (F(1,17) = 4.41, p<0.05). Post hoc comparisons revealed a significantly greater density of ERβ-EGFP expressing cells in males than females at P4 (p<.05) and P21 (p<.05; Figure 7D).

Discussion

In this study, we used an Esr2-EGFP transgenic mouse line to demonstrate that ERβ expression changes considerably during development in many brain regions and in some, it also varies by sex. The Esr2-EGFP mouse line was chosen for these studies because of the controversial nature of current ERβ directed antibodies for use in immunocytochemistry (Snyder et al, 2010) in brain. The development of this mouse line and its utility for localization of ERβ in the brain of the adult has been described previously (Milner et al, 2010). In Milner et al, (2010), and in our studies, the immunocytochemical localization of EGFP recapitulated the known distribution of ERβ described in previous studies examining ERβ mRNA (Merchenthaler et al., 2004), estrogen binding in ERα knockout mice (Shughrue et al, 1999), as well as ERβ immunoreactivity using a previously available and well characterized antibody, Z8P (Shughrue and Merchenthaler, 2001; Perez et al., 2003). However, Milner et al. (2010) do indicate a small degree of non-correspondence between ERβ-EGFP-ir and ERβ-ir using the Z8P antibody. This indicates that within regions they examined (e.g. PVN, medial amygdala, CA3) and others examined in the current study there is always the potential for some false-positive and false-negative ERβ signal.

One of the most striking sex differences we report is ERβ expression in AVPV cells that are clustered near the 3^rd ventricle. ERβ-EGFP-ir cells in this region are more densely populated in females than males, similar to the sex difference in ERβ expression reported in the rat AVPV (Orikasa et al., 2002). We further observed that this sex difference appears at some time point after P4. In rats, a sex difference in ERβ has been reported as early as postnatal day 7 and is regulated by neonatal testicular secretions in males since neonatal castration of males abolishes this sex difference (Orikasa et al., 2002). It is important to note that the AVPV is larger and contains more cells in the female than the male rat (Bleier et al., 1982; Sumida et al., 1993) and sex differences in a number of cell phenotypes have been described in this area (Simerly et al., 1985; Simerly et al., 1988). Further, the AVPV contributes to the regulation of reproductive behaviors (Wiegand et al., 1978) and specifically, activation of ERβ within this region controls the estrogen induced surge of luteinizing hormone in female rats (Orikasa et al., 2002). Therefore, this sexually dimorphic structure in mice may regulate similar effects of estrogen on reproductive behaviors. At present, the phenotype of ERβ neurons in the mouse AVPV is unknown and a further description of these cells would be important to discern.

A sex difference in ERβ-EGFP-ir was also found in the BNST with male mice having a greater density of immunoreactive cells compared to females, a finding that is consistent with the sex difference in ERβ reported in adult rats (Zhang et al., 2002). We also report that this sex difference in mice becomes apparent at some point between P0 and 4 and persists into adolescence. The BNST is sexually dimorphic in mice, with males exhibiting a greater volume and cell number (Forger et al., 2004; Gotsiridze et al., 2007), and these sex differences appear during the second week postnatal in mice (Gotsiridze et al., 2007). Our demonstration of sex differences in ERβ expressing neurons in this region by P4 indicates that sex differences in this region may appear much earlier than previously reported. Our findings are consistent with other studies showing that males also have a greater number of androgen receptor (AR) expressing cells and a larger volume of the mouse BNST that is occupied by AR (Shah et al., 2004). Here we report a sexual dimorphism in ERβ expression in mice that is similar to that of AR. Furthermore, we have found that this dimorphism appears prior to the emergence of volumetric and cell number differences suggesting a possible mechanism through which androgens, after metabolism to estrogens, can activate a large number of ERβ-ir neurons to generate these sex differences. However, a recent study indicates that ERβ null mice show normal BNST morphology while ERα null mice have decreased neuron number and volume (Tsukahara et al., 2011) indicating that estrogen may act primarily through ERα to produce these sex differences. Alternatively, since androgen and estrogen treatments on the day of birth result in masculinization of the BNST (Hisasue et al., 2010) this indicates that alterations in ERβ expression at P4 and P21 may be a consequence of sexual differentiation of this nucleus, rather than a contributing factor.

In the present study, we did not detect sex differences in several regions where sexual dimorphisms in ERβ-ir had previously been reported in adult mice. These include the lateral amygdala, endopyriform cortex, lateral septum, and SCN (Milner et al., 2010; Vida et al., 2008). These results indicate that sex differences in ERβ within these regions may develop either during or after puberty or alternatively may be regulated by circulating adult gonadal hormone levels. In support of this possibility, Vida et al. (2008) report that the female mouse SCN contains a greater number of ERβ-ir cells and treatment of gonadectomized males and females with estradiol benzoate decreases ERβ-ir cell number. It is also possible that subtle differences exist in ERβ-ir within several brain regions which may not be detectable without further quantitative analysis.

Our studies show that ERβ expression patterns vary during postnatal development throughout the mouse forebrain. In general, the distribution of ERβ-EGFP in the P21 mouse brain appears quite similar to that described in the adult Esr2-EGFP mice (Milner et al., 2010) indicating that major developmental alterations in ERβ occur prior to P21. In the P0 hippocampus ERβ is highly expressed throughout CA1 and CA3 layers, while expression, particularly within the CA1 region, decreases by P4 and remains lower at P21. These results indicate that the neonatal hippocampus may be particularly responsive to estrogens. On the contrary, ERβ-EGFP-ir cells within the hilus region of the dentate gyrus are not present in the neonatal hippocampus indicating that estrogens that act via ERβ, may have a lesser effect on the development of this region. To our knowledge, the role of ERβ activation on the development of CA1 and CA3 mediated hippocampal function has not been explored to date. In contrast, ERα-ir cells appear to be more abundantly expressed throughout the developing rodent hippocampus, particularly within the dentate gyrus (Perez et al., 2003) indicating that this receptor, rather than ERβ, may be the primary mediator of estrogen effects on this hippocampal region.

Levels of estrogen receptors in the cortex change dramatically during development. Early studies, using binding assays showed a developmental rise in rat cortical ER that peaked between P7 and 10 (MacLusky et al, 1979). Extensive reorganization of the cortical ER system occurs in the early postnatal period (Shughrue et al, 1990) and it has been suggested that ER containing neurons may be migrating from deep to superficial layers during the first postnatal week. Because these earlier studies did not differentiate between ERα and ERβ, it is unknown if one of the other form of receptor contributes to the effect of estradiol on maturation and growth of the cortex (Torran-Allerand 1984), however, ERα levels have been shown to decrease in mouse cortex between 1 and 25 days of age (Prewitt and Wilson, 2007).

As the layers of cerebral cortex form during postnatal development it appears that the distributions of ERβ may also vary. At P21 ERβ-EGFP-ir cells are found almost exclusively in the 5^th and 6^th cortical layers, while in neonatal mice, cells are distributed within outer, middle, as well as inner cortical layers. Axons of ERβ-EGFP positive cells of the neonatal cortex are also strongly labeled and appear to span multiple layers. These developmental differences in ERβ expression likely reflect migration patterns of ERβ positive neurons during postnatal cortical development (Berry and Rogers, 1965; Rakic, 1974). A previous study (Wang et al., 2003), although yet to be replicated (Antal et al., 2008), showed altered migration of cortical neurons in ERβ null mice, raising the possibility that ERβ may be involved in normal development of the cortex.

Several hypothalamic nuclei also show developmental alterations in ERβ-ir, with some regions showing increased, and others decreased immunolabeling during development. In the PVN, ERβ-EGFP-ir is similar at neonatal time points but increases by P21, while other regions such as the arcuate nucleus show an increase in ERβ-EGFP between P0 and P4 with levels remaining consistent through P21. By contrast, the VMH decreases in ERβ expression between neonatal ages and adolescence. This is consistent with the studies of Ikeda et al (2003) who showed a similar age-related change in the rat caudal VMH. Of interest, Ikeda et al described a sex difference in ERβ levels in the rat VMH which we did not detect in the mouse. This difference may be species related. Development differences in ERβ expression throughout the hypothalamus indicate differences in regional sensitivities to estrogens. These findings further suggest that estrogen, acting via ERβ, may affect these regions differently depending on developmental age. ERβ-ir cells are also present in the median eminence during postnatal development with the greatest levels found at P21. This indicates that ERβ may be a key receptor through which estrogens may act to mediate functions of this region including release of regulatory hormones such as corticotropin releasing factor and gonadotropin releasing hormone. Furthermore, estrogens acting via ERβ may also regulate subtypes of neurons that occupy the internal zone of the median eminence (e.g. neuropeptide Y and β-endorphin positive cells) although further studies are needed to test this possibility.

Pharmacological activation of ERβ during the neonatal period has also been shown to alter some adult behaviors including anxiety, aggression, and sexual functions (Patisaul and Bateman, 2008; Sullivan et al., 2011). Such findings indicate that brain areas such as the hypothalamus and hippocampus, which regulate these behaviors, may be highly vulnerable to estrogen actions mediated by ERβ during the neonatal period. In particular, the hippocampus expresses high levels of ERβ early in the neonatal period with levels decreasing during postnatal development. This is consistent with earlier studies showing a transient increase in estrogen receptor in hippocampus (O'Keefe and Handa, 1990, O'Keefe et al, 1995). This is in part due to a transient increase in ERα levels (Solum and Handa, 2001) which peak near P10, but a similar contribution from ERβ is suggested by the present results. Such transient elevations in ERs may indicate a perinatal sensitive period and specifically, ERβ activation in the developing hippocampus could lead to long-term behavioral alterations. Further studies are needed to test effects of ERβ specific agonists, delivered at multiple developmental time periods, on hippocampus-associated behaviors.

Sexual differentiation of the brain occurs via exposure to testicular steroids, such as testosterone, early in life, which masculinize the developing brain leading to permanent changes in morphology and behavior. Traditionally, masculinization and defeminization of the rodent brain is believed to occur by metabolism of testosterone to estrogen by the aromatase enzyme and subsequent activation of estrogen receptors. Several studies suggest that activation of ERα is primarily responsible for sexual differentiation of the brain and behavior (Wersinger et al., 1997; Tsukahara et al., 2011), although ERβ also appears to contribute. Bodo et al. (2006) demonstrated that activation of ERβ is necessary for the sexual differentiation of the mouse AVPV. Furthermore, a study utilizing ERβ knockout mice indicates that ERβ is necessary for the defeminization of sexual behavior (Kudwa et al., 2005). Sex differences presented here in ERβ expression in the BNST and AVPV may be one mechanism through which estrogens can contribute to these sexual dimorphisms in brain and behavior. Furthermore, developmental changes in ERβ indicate that there may be sensitive periods during which ERβ activation can affect brain sexual differentiation.

Together, the present findings indicate both developmental and sex differences in the distribution of ERβ in the mouse forebrain. These findings contribute to our understanding of how estrogens may differentially affect the brain and behaviors throughout development in both male and female mice.