Breeding pairs of C57BL/6J mice (RRID:IMSR_JAX:000664) were obtained from Japan SLC Inc. and CLEA Japan to build and maintain our breeding colony. The breeding colony was periodically refreshed with new breeding pairs. Mice were raised under controlled conditions (12 h light/dark cycle, lights on at 8:00 AM, 23 ± 2°C; 55 ± 5% humidity, and ad libitum access to water and food). Mice were weaned at 4 weeks of age and housed in groups of four or five.

For assessing Moxd1 expression in the MPOA, five groups of mice were used: intact male, castrated male, neonatally-castrated male, intact female and ovariectomized female. Eight intact male and female mice were sacrificed and their brains were sampled as described below. Another six adult male mice were castrated, and 2 weeks later, their brains were sampled. Additionally, six ovariectomized female mice were sampled in the same manner. Six infant male mice were castrated on the day of birth and raised under normal conditions (Becker et al., 2005). Four mice from the intact male, neonatally-castrated male and intact female groups were also used for assessing Moxd1 expression in the BNSTpr and MePD. The same four intact male mice and female mice were used to confirm no sexual difference in the other areas. In addition, the same four males were used for assessing colocalization of Moxd1 and calbindin D. All mice were 10–20 weeks old when sampled. All procedures were carried out in accordance with the Guidelines for Animal Experiments of Toho University. All animal experimentation was approved by the Institutional Animal Care and Use Committee of Toho University (Approved protocol ID #15-52-254).

Mice were deeply anesthetized with sodium pentobarbital (50 mg/kg, i.p.) and then perfused transcardially with 4% (w/v) paraformaldehyde (PFA) in 0.01 M phosphate-buffered saline (PBS, pH 7.4). The brains were removed, postfixed in 4%PFA/PBS at 4°C overnight, followed by cryoprotection in 30% (w/v) sucrose in PBS for 2 days, embedded in Surgipath (FSC22, Leica Biosystems), and then stored at −80°C until cryosectioning. To assess the expression of genes of interest in the MPOA, 40 μm-thick serial coronal sections were prepared to cover the entire MPOA according to the mouse brain atlas (Franklin and Paxinos, 2007) and stored in a cyoprotectant solution (30% glycerol, 30% ethylene glycol, 0.05 M phosphate buffer) at −30°C until use. We used a set of every third sections from the serial sections to examine the expression of target molecules.

Candidate genes for an area marker of the SDN-POA were acquired from the Allen Gene Expression Atlas in the Allen Brain Atlas database¹. To investigate specific gene expressions in the MPOA, we used the following parameters. The coordinates of the region of interest were as follows: AP: 5.000–5.800 mm, DV: 5.000–6.600 mm, L: 5.800–6.200 mm. This coordinates contains the whole MPOA and a part of the BNST. Other parameters were not changed from default values. The search provided more than 500 candidate genes in descending order of fold-change together with ISH images. The authors visually examined the expressions of all candidate genes using the Allen brain atlas to find any genes that were strongly and specifically expressed in the MPOA. Our in silico search for MPOA marker genes selected nine genes that were localized to a specific subregion in the MPOA. We performed ISH of these genes in our mouse samples and confirmed the expression of the Moxd1 gene in the subregions of the MPOA.

One antisense and one sense riboprobes were used to assess the expression of Moxd1 mRNA (RefSeq ID: NM_021509.5, 422-2743 bp). cDNA fragments were amplified with polymerase chain reaction, inserted into the multi cloning site of a pGEM-T plasmid (A3600, Promega), and transformed into DH5α E. coli. The cDNA was verified by sequencing (Hokkaido system science, Japan). The plasmid DNA was extracted and purified after liquid culture, and the template cDNA was produced using polymerase chain reaction using the specific primers (5′-ATTTAGGTGACACTATAG-3′) and (5′-TAATACGACTCACTATAGGG-3′). Probes were transcribed by SP6 RNA polymerase (P1085, Promega) for antisense probe or T7 RNA polymerase (10881767001, Roche) for sense probe in the presence of digoxigenin-labeled UTP (Dig labeling mix; Roche Diagnostics, Basel, Switzerland) followed by precipitation with LiCl with ethanol.

Brain sections were processed for ISH as previously described (Tsuneoka et al., 2013, 2015) with some modifications. Briefly, the sections were washed with PBS containing 0.1% Tween-20 (PBT), postfixed with 4% PFA in PBS for 10 min at room temperature. Then, the sections were immersed in methanol containing 0.3% H₂O₂ for 10 min, followed by acetylation with 0.25% acetic anhydride in 0.1 M triethanolamine (pH 8.0). The hybridization solution contained 50% deionized formamide, 5× standard saline citrate (SSC, pH 7.0), 5 mM ethylene-diaminetetraacetic acid (pH 8.0), 0.2 mg/ml of yeast tRNA, 0.2% Tween-20, 0.2% sodium dodecyl sulfate, 10% dextran sulfate, and 0.1 mg/ml of heparin. The sections were prehybridized at 58°C in the mixture of the hybridization solution and PBT (1:1) for 30 min, immersed in the hybridization solution for 15 min, and then hybridized with the riboprobes (1 μg/ml) at 58°C for 16 h. After hybridization, the sections were washed twice with 2× SSC containing 50% deionized formamide at 58°C for 10 min, incubated with RNase A solution (20 μg/ml, R4642, Sigma) and avidin (0.1 μg/ml, 013-21013, Wako pure chemicals, Japan) at 37°C for 60 min, and rinsed twice in 2× SSC and four times in 0.2× SSC at 37°C (15 min each).

To visualize digoxigenin-labeled riboprobe, the sections were incubated in a peroxidase-conjugated anti-digoxigenin antiserum (1:1000, Roche Diagnostics, RRID:AB_514499) with biotin (0.5 μg/ml, 021-08712, Wako pure chemicals). After a 2-h incubation in the antibody solution at room temperature, the sections were washed and immersed in 0.1 M boric buffer (pH 8.5) containing 4 μM biotin-labeled tyramide, 4% dextran sulfate, 0.05 mg/ml iodophenol, and 0.003% H₂O₂ for 30 min. They were immersed in a cocktail of Alexa 647-conjugated streptavidin (1:10,000, Life Technologies, Grand Island, NY, USA) and Hoechst 33342 (1 μg/ml, Life Technologies). The sections were mounted on a glass slide with Gel/Mount. The sections were washed three times with PBS.

For double ISH for Moxd1 and Trh, an antisense riboprobe for Trh mRNA (RefSeq ID: NM_009426.3, 88–990 bp) was synthesized using SP6 RNA polymerase (P1085, Promega) in the presence of dinitrophenol-labeled UTP (NEL555001EA, PerkinElmer, Waltham, MA, USA). Double ISH was performed using hybridization solution containing both digoxigenin-labeled riboprobe for Moxd1 and dinitrophenol-labeled riboprobe for Trh mRNA (1 μg/ml for each). Digoxigenin-labeled Moxd1 riboprobe was visualized as described above. Dinitrophenol-labeled Trh riboprobe was visualized using a peroxidase-conjugated anti-dinitrophenyl antibody (1:2500, FP1129, PerkinElmer, RRID:AB_2629439), followed by the incubation in 0.1 M boric buffer (pH 8.5) containing 4 μM Alexa 568-labeled tyramide, 4% dextran sulfate, 0.05 mg/ml iodophenol and 0.003% H₂O₂ for 30 min.

To visualize Moxd1 mRNA and calbindin D, Moxd1-ISH performed sections were further processed for the immunostaining for calbindin D. Similarly, to visualize Moxd1 mRNA, Trh mRNA and oxytocin, sections that were performed for double ISH for Moxd1 and Trh mRNAs were further processed for the immunostaining for oxytocin-neurophysin I. To visualize Trh mRNA, calbindin D and oxytocin, Trh-ISH performed sections were processed for double immunostaining for calbindin D and oxytocin.

For immunostaining, sections were incubated in a mouse anti-calbindin D-28K antibody (1:1000, C9848, Sigma Aldrich, RRID:AB_476894) or in the cocktail of mouse anti-calbindin and goat anti-neurophysin I antibodies (1:2000, sc-7810, Santa Cruz, RRID:AB_650167) overnight. Then, the sections were washed, immersed in a cocktail of Alexa 568-conjugated anti-mouse IgG antibody (1:250, Thermoscientific, RRID:AB_2534013) and/or Alexa 488-conjugated anti-goat IgG antibody (1:500, Jackson Immunoresearch, RRID:AB_2340428). Immunostained sections were mounted on a glass slide with Gel/Mount.

To count Moxd1-positive cells in the SDN-POA, we used a set of three coronal sections 120 μm apart. To fix the position of the three sections, the most rostral section of the three sections was set to the level where the rostral end of the commissural part of the anterior commissure was located (approximately Bregma +0.22 mm). Thus, Figures 1B–D indicate the typical three sections that we used for cell counting. Each set of three sections from intact, castrated, or neonatally castrated males, or intact and ovariectomized females was used for single ISH and then processed for positive cell counting.

To evaluate the colocalization of Moxd1 and calbindin, four sets of MPOA sections from different male mice were randomly selected and then Moxd1- and calbindin-positive cells were counted and averaged in the SDN-POA, the border of which was determined by the distribution of the Moxd1 signals. For the other areas, every third coronal section from the whole brain except the olfactory bulb from different male mice (n = 4) were randomly selected and stained for Moxd1 ISH combined with IHC for calbindin. Identification of the brain areas was done according to the chemoarchitectonic atlases (Dong, 2008; Paxinos and Watson, 2010; Martínez-García et al., 2012) and our previous study (Tsuneoka et al., 2013, 2015).

Fluorescent microscopic images of brain sections were taken using a Nikon microscope Eclipse Ni under a 20× objective equipped with a confocal detection system A1R (Nikon Instruments Inc., Tokyo, Japan). To obtain fluorescent images, each channel was collected separately with single-wavelength excitation and then merged to produce a composite image. Each image was obtained as five-layer z-stack images, and the optical thickness of the sections was 1.0 μm. Experimental controls were prepared in which the primary antibodies or riboprobes were omitted from the reaction solution to confirm no detectable signal. For ISH-stained sections, some non-specific granule-like signals were observed but easily distinguished from cytoplasmic-specific staining. In addition, specific staining of antisense probes was confirmed by the staining using sense probes (data not shown).

Images were analyzed using ImageJ software (version 1.50i, NIH, USA, RRID:SCR_003070). Each image was binarized by the fixed threshold value and the cell was marked manually on the threshold image. Threshold was determined to be above background or nonspecific signals on the control sections, and the same threshold was used through analysis in all samples. Since all procedures of brain sampling, ISH and IHC were performed in an exactly same time course under controlled temperature, the fixed threshold work to evaluate positive cells of different mice. Cell counting was conducted in three sections containing the SDN-POA unilaterally, and the summed value from the three sections was applied to a statistical analysis. In the other brain regions, cell counting was conducted in one section which was selected to maximize counting area.

To calculate the proportion of double labeling, we separately marked singly labeled cells and doubly labeled cells of Moxd1 mRNA and calbindin on the threshold image. The double-labeling was judged when the signals were observed just around cell nuclei and overlapped each other.

To quantify the sex difference of the BNSTpr and MePD, ISH signal-positive area of the threshold image was measured instead of cell counting because the cell density was too high to separate each cell signals. All histological procedures were done without the observer knowing the sex of the samples.

Data were analyzed using Welch’s t-test or Welch’s one-way analysis of variance (ANOVA) followed by post hoc comparisons using Welch’s t-test to address unequal variances among groups if the data match the statistical premise of the tests. The P values of all multiple comparisons were adjusted appropriately using the Holm’s method, also called a sequential Bonferroni (Sokal and Rohlf, 1995). An adjusted P < 0.05 was regarded as statistically significant. All statistical analyses were conducted using software R 2. 14. 0. Quantitative data were presented as the mean ± SEM.

Moxd1 mRNA-positive cells observed in and around the MPOA were restricted to the SDN-POA and BNSTpr of adult mice. Moxd1 mRNA-positive cells were distributed in the entire SDN-POA and narrowly continuous with the BNSTpr (Figure 1). On the other hand, Moxd1 mRNA-positive cells were only a few in the remaining part of the MPOA.

Overall, the number of Moxd1-positive cells in the SDN-POA were significantly affected by sex and gonadectomy (F_(4,13.3) = 25.4, P < 0.001), although their distributions in the MPOA including the SDN-POA were not different (Figures 1, 2). The number of Moxd1-positive cells in the SDN-POA of male mice was significantly larger than that of female mice (Post hoc test, P < 0.001, Figures 1, 2A,D,F). To examine the effect of castration on the number of Moxd1-mRNA positive cells in the SDN-POA of male mice (Becker et al., 2005), we performed castration in the adult or neonatal period. Although the distributions of Moxd1 mRNA-positive cells of the castrated and neonatally castrated mice were similar to that of intact adult mice, the number of Moxd1 mRNA-positive cells of neonatally castrated mice was significantly smaller than those of castrated mice and intact mice (P < 0.01 for both) and similar to that of female mice (P = 0.544, Figure 2F). The number of Moxd1 mRNA-positive cells of intact males was similar to that of castrated males (P = 0.870). Additionally, we examined the effects of an ovariectomy in adult females on the number of Moxd1 mRNA-positive cells in the SDN-POA and found that the number was not significantly different between intact and ovariectomized female mice (P = 0.621; Figures 2D–F).

As previously reported (Orikasa and Sakuma, 2010), calbindin-ir cells were aggregated in the SDN-POA and broadly distributed in the MPOA (Figure 3). Double labeling of Moxd1 mRNA and calbindin protein confirmed that a large number of the Moxd1 mRNA-positive cells expressed calbindin (Figures 3A–G; n = 4, 70.4 ± 6.1%). In the SDN-POA, the population of Moxd1 mRNA-positive cells in calbindin-ir cells was relatively small (n = 4, 57.8 ± 2.6%).

Calbindin-ir and Moxd1 mRNA-negative cells were also found in the MPN outside of the SDN-POA, the ventral part of the MPOA (vMPOA) and the anterior commissural nucleus (ACN; Figure 3B). In contrast, the Moxd1 mRNA-positive and calbindin-negative cells were found only in the SDN-POA (Figure 3A). Most of oxytocin-neurophysin I-ir cells, which were localized mainly in the ACN and periventricular preoptic area (PVPOA), were immunopositive for calbindin and negative for Moxd1 (Figures 3H–K). Most of Trh-expressing cells, which were localized mainly in the ACN, MPN and vMPOA, were immunopositive for calbindin and negative for Moxd1 mRNA-positive cells (Figures 3K,O). No Moxd1 positive signal was found in oxytocin-neurophysin I-positive cells or Trh mRNA-positive cells in the MPOA (Figures 3L–O).

Next, we examined the expression of Moxd1 mRNA outside of the MPOA (Table 1). In the telencephalon, Moxd1 mRNA was most strongly expressed and densely clustered in two regions: the BNSTpr and MePD (Table 1). Furthermore, a male-biased sex difference in the expression of Moxd1 mRNA was recognized in the BNSTpr and MePD, and neonatal castration diminished male-dominant expressions in these areas (Figure 4). The area expressing Moxd1 mRNA in the BNSTpr of male mice was significantly larger than that of neonatally castrated male mice and that of female mice (male: 30074 ± 3022 μm², neonatally castrated male: 11764 ± 305 μm², female: 13498 ± 1150 μm², n = 4 for each, Welch’s ANOVA, F_(2,4.3) = 16.5, P = 0.010, Post hoc test, P < 0.05 for both, Figures 4A–D). Similar to the BNSTpr, the area expressing Moxd1 mRNA in the MePD of male mice was significantly larger than that of the neonatally castrated male mice and that of female mice (male: 37142 ± 2195 μm², neonatally castrated male: 12591 ± 1928 μm², female: 19221 ± 3344 μm², n = 4 for each, Welch’s ANOVA, F_(2,5.8) = 32.2, P < 0.001, Post hoc test, P < 0.05 for both, Figures 4E–H).

Calbindin was also strongly expressed in the BNSTpr. Double labeling of Moxd1 mRNA and calbindin protein confirmed that the majority of the Moxd1 mRNA-positive cells expressed calbindin in the BNSTpr (Figures 5A–C), as was observed in the SDA-POA (Figures 3A–F). The Moxd1 mRNA-positive cells were distributed uniformly and densely in the entire BNSTpr, whereas the calbindin-ir cells were concentrated in the center area of the BNSTpr (Figures 5A–C). In the remaining part of the BNST, a moderate number of Moxd1-positive cells were found without any difference between sexes and immunoreactivity for calbindin (Table 1). A uniform and dense distribution of the Moxd1-positive cells was found in the entire MePD, whereas a small number of the calbindin-ir cells were located in the center area of the MePD (Figures 5D–F). Moxd1 mRNA-expressing cells were found in the basolateral amygdala (BLA), basomedial amygdala (BMA), cortical amygdala (COA), piriform cortex and MEA outside of the MePD (Table 1, Figure 6). Most of the Moxd1 mRNA-positive cells in the amygdala were calbindin-positive, except for in the central amygdala (CEA; Figure 6). There was no apparent sex difference in Moxd1 mRNA expression outside of the MePD.

In the cerebral cortex (CTX), strong Moxd1 signals were found in the dorsal tenia tecta (DTT) and scattered throughout the cortical layers (Figure 7A). Many of the Moxd1 mRNA-positive cells in the DTT and cerebral cortical layer had also calbindin-ir signals (Figures 7A–F). Moxd1 mRNA was strongly expressed in the subplate layer of the CTX as reported previously (Figure 7D; Hoerder-Suabedissen et al., 2009), whereas the Moxd1-positive cells in the subplate were negative for calbindin (Figures 7D–F). Moxd1-positive cells were scattered in the olfactory tubercle, striatum, nucleus accumbens, hippocampus and dentate gyrus, without colocalization of calbindin (Table 1).

There were few cells expressing Moxd1 mRNA in the diencephalon and brainstem. In the cerebellar cortex, Moxd1 mRNA-positive cells were abundantly found in Purkinje cells, where intense calbindin-ir cells were distributed (Figures 7G–I). Intense Moxd1 signals were also found in the colloid plexus and dura.

Moxd1 is abundantly expressed in the SDN-POA, BNSTpr and MePD in a sexually dimorphic manner, where the cell number and volume show a male-biased sex difference which is determined by high testosterone level during the perinatal period (Gilmore et al., 2012; Campi et al., 2013; Moe et al., 2016a). Among these nuclei, BNSTpr and MePD are included in the medial extended amygdala, which serve male reproductive circuits via pheromonal inputs from accessary olfactory bulb and contain various sexually dimorphic structures (Newman, 1999; de Olmos and Heimer, 1999; Martínez-García et al., 2012). Neurocircuit, neurochemical and functional similarity were found between BNSTpr and MePD neurons in relation to the SDN-POA. The SDN-POA has some efferent and afferent connections to the BNSTpr and MePD (Simerly and Swanson, 1988; Maejima et al., 2015). MePD neurons send their axons most abundantly to the BNSTpr (Dong et al., 2001; Been and Petrulis, 2011). The MePD receives dense input from the vomeronasal organ via anterior and posterior accessory olfactory bulb to detect pheromones (Von Campenhausen and Mori, 2000; Mohedano-Moriano et al., 2007; Cádiz-Moretti et al., 2016), and the BNSTpr also receives inputs from anterior accessory olfactory bulb (Mohedano-Moriano et al., 2007, 2008), suggesting that odor/pheromones regulate the MePD neurons and BNSTpr, and subsequently SDN-POA neurons. In fact, female odor has been shown to induce Fos immunoreactivity in MePD neurons that project to the BNST and MPOA (Been and Petrulis, 2011). Male sexual behavior toward females induces fos expression in the MePD, BNSTpr and SDA-POA (Heeb and Yahr, 1996; Coolen et al., 1998). Rich expression of gonadal steroid receptors in these nuclei implies their role in sexual behaviors which are under the influence of gonadal steroids (Shughrue et al., 1997).

The SDN-POA was larger in volume in male than in female mammals (Gorski et al., 1978; a list in Campi et al., 2013). However, the sexual dimorphism of the SDN-POA in laboratory mice was not recognized or was very subtle in Nissl-stained sections (Young, 1982; Brown et al., 1999). Recently, calbindin has been used to visualize the SDN-POA in mice as well as rats, musk shrews and common marmosets in a sexually dimorphic pattern (Sickel and McCarthy, 2000; Edelmann et al., 2007; Bodo and Rissman, 2008; Orikasa and Sakuma, 2010; Jahan et al., 2015; Moe et al., 2016a,b). The sexual dimorphism of calbindin-ir cells in the MPOA depends on the presence of testosterone during the neonatal period and is independent of the gonadal steroid hormone level in adult animals (Sickel and McCarthy, 2000; Orikasa and Sakuma, 2010). Sexually dimorphic markers for the rat MPOA, such as the androgen receptor (McAbee and DonCarlos, 1998) and genes that are expressed in the SDN-POA-equivalent region (Simerly et al., 1986), have failed to show sexual dimorphism in mice (Jahan et al., 2015). So far, calbindin has been the only marker for the mouse SDN-POA.

However, calbindin-ir cells are not restricted to the SDN-POA and BNSTpr but exist abundantly in the MPOA, BNST and MEA outside of the SDN-POA, BNSTpr and MePD, respectively. Moreover, we found that the majority of Trh- and oxytocin-positive neurons in the MPOA also expressed calbindin (Figure 3), consistent with the rat MPOA (Arai et al., 1994). Therefore, optogenetic and pharmacogenetic studies using a Cre-driver mouse targeting calbindin may affect a variety of neurons surrounding the sexually dimorphic nuclei.

Unlike calbindin, Moxd1 is expressed more specific to sexually dimorphic nuclei and more uniformly expressed in the SDN-POA, BNSTpr and MePD. The majority of the Moxd1-positive cells in sexually dimorphic nuclei were calbindin-ir and not positive for Trh or oxytocin (Figure 3). Thus, together combined with a local injection of adeno-associated viral vector, the Moxd1 gene may enable us to manipulate neurons in sexually dimorphic nuclei to examine the effect on sexual behaviors and to elucidate the afferent and efferent connections of sexually dimorphic neurons (Schwarz et al., 2015; Lerner et al., 2016). Moxd1 encodes a monooxygenase, DBH-like 1, that is localized in the endoplasmic reticulum and predicted to hydroxylate a hydrophobic substrate based on its amino acid sequence similar to dopamine β-hydroxylase (Xin et al., 2004). Because the substrates of Moxd1 have not been identified, the biological roles of Moxd1 in sexually dimorphic neurons are not known. A frequent colocalization of Moxd1 and calbindin in sexually dimorphic nuclei implies that Moxd1 and calbindin work together to regulate intracellular calcium released from the smooth endoplasmic reticulum. Future studies to identify a regulatory network inducing the Moxd1 gene expression may elucidate the mechanism underlying the development of sexual dimorphic nuclei.