Androgen receptor deficiency leads to reduced aromatase expression in the ventromedial hypothalamus of male cichlids

Abstract

Social behaviors are regulated by sex steroid hormones such as androgens and estrogens. However, the specific molecular and neural processes modulated by steroid hormones to generate social behaviors remain to be elucidated. We investigated whether some actions of androgen signaling in the control of social behavior may occur through the regulation of estradiol synthesis in the highly social cichlid fish, Astatotilapia burtoni. Specifically, we examined the expression of cyp19a1, a brain-specific aromatase, in the brains of male A. burtoni lacking a functional ARα gene, which was recently found to be necessary for aggression in this species. We found that cyp19a1 expression is higher in wild-type males compared to ARα mutant males in the anterior tuberal nucleus, the putative fish homolog of the mammalian ventromedial hypothalamus (VMH), a brain region that is critical for aggression across taxa. Using in situ hybridization chain reaction (HCR), we determined that cyp19a1⁺ cells co-express ar1, the gene encoding ARα, throughout the brain, including in the ATn. We speculate that ARα may modulate cyp19a1 expression in the ATn to govern aggression in A. burtoni. These studies provide novel insights into the hormonal mechanisms of social behavior in teleosts and lay a foundation for future functional studies.

Graphical Abstract

Social behaviors are regulated by sex steroid hormones such as androgens and estrogens, but the specific molecular and neural processes that steroid hormones modulate to generate social behaviors need further investigation. Here, we investigate the expression of cyp19a1, a brain-specific aromatase, in an African cichlid lacking a functional ARα gene necessary for aggression. We find that cyp19a1 expression is lowered in males lacking ARα in the putative ventromedial hypothalamus homolog.

INTRODUCTION

Steroid hormones regulate physiology and behavior via ligand-dependent receptor activation ^1,2. Specifically, steroid hormones exert their effects via specific steroid receptors that when bound, will dimerize, and then translocate to the nucleus to alter the expression of suites of genes, thus altering cellular processes that give rise to modifications in biological functions. Additionally, steroid receptors can modulate cellular functions differently from modulating gene expression by rapidly affecting neural processes via ion channels and secondary messenger cascades³. Despite the known link between steroid hormones, steroid receptors, cellular processes, and behavior, it remains unclear which specific genes in the brain are modified by steroids to alter behavior. This is in stark contrast to our knowledge of other steroid-sensitive tissues like the prostate, where some of the most comprehensive data on steroid receptor controlled gene expression are available^4–6. A comprehensive understanding of the functions of steroid hormones in the regulation of the brain and behavior requires approaches that can elucidate the downstream genes controlled by steroid receptors.

Steroid receptors are expressed widely in the brain, including in areas of the brain known to control social behavior. Indeed, sex steroid receptors, including androgen receptors (AR) and estrogen receptors (ER), are expressed in regions of the social behavior network (SBN), a highly conserved, interconnected set of brain regions in vertebrates that respond to social cues and coordinate adaptive physiological and behavioral responses ^7–11. These regions include the preoptic area (POA), the anterior hypothalamus (AH), the ventromedial hypothalamus (VMH), the lateral septum (LS) and the bed nucleus of the stria terminalis (BNST) ¹⁰. Given its conserved functions, the role of the SBN and the functions of steroid hormone actions throughout the SBN have been extensively studied across diverse taxa to reveal fundamental insights into the neuroendocrine regulation of behavior.

One such species in which important insights have been gained on the hormonal regulation of behavior and SBN functions is the African cichlid fish, Astatotilapia burtoni. A. burtoni is a genetically tractable and highly social fish species making it an ideal organism to study the hormonal and neural mechanisms of social behavior¹². Male A. burtoni form a social hierarchy in which certain males are dominant (DOM) and others are subordinate (SUB). DOM males produce more territorial and reproductive behaviors and have higher levels of testosterone, 11-ketotestosterone, and estradiol compared to SUB males¹³. Androgen signaling has been shown to play a key role in controlling dominance behaviors in A. burtoni. For example, pharmacological antagonism of ARs in DOM male A. burtoni robustly reduces courtship behaviors ^14,15. Furthermore, recent discoveries in male A.burtoni have shown that different ARs control distinct traits of social dominance. For example, male A.burtoni lacking one (heterozygous/HET mutants) or two (homozygous null/knock out [KO] mutants) functional alleles encoding ARα, which is encoded by the ar1 gene, perform fewer DOM-typical aggressive and courtship displays compared to wild-type (WT) males. On the other hand, ARβ, which is encoded by the ar2 gene, heterozygous and KO mutants performed normal levels of DOM-typical aggressive and courtship behaviors. Therefore, androgenic signaling, particularly via ARα, appears to be necessary for dominance behavior. However, it is not known which genes and in which areas of the brain are controlled by androgenic signaling to modify these behaviors.

The conversion of androgens to estrogens by the enzyme aromatase is crucial for the activation of aggressive displays in male A. burtoni. O’Connell and Hofmann¹⁴ injected male A. burtoni with 17β-estradiol and observed an increase in aggression, while injection of an ER antagonist caused a reduction in aggression. In subsequent work, male A. burtoni were injected with the aromatase inhibitor fadrozole, which caused a significant decrease in aggression compared to males injected with just vehicle¹⁶. Hence, in male A. burtoni androgenic signaling via ARα, as well as the synthesis and action of estradiol, may be required for performing DOM-typical aggressive behavior. Additionally, in contrast to other vertebrate species, in which only one aromatase gene is present, there are two different copies of the aromatase gene in teleost species, one primarily expressed in the gonads called cyp19a1a and the other expressed primarily in the brain called cyp19a1^17–19. Previous observations in other teleost species also have shown that cyp19a1 expression seems to be primarily localized in glial cells in comparison to other vertebrate species^17,20–22.

Research in other species has shown that androgenic signaling enhances aromatase expression in the brain. Castrated male mice have lower aromatase expression in the diencephalon compared to intact male mice, while castrated male mice treated with testosterone—but not estradiol—show an increase in aromatase expression in the diencephalon to intact levels²³. The androgen, dehydroepiandrosterone (DHEA), increases aromatase expression in the POA and VMH of male song sparrows (Melospiza melodia)²⁴. In male rats, aromatase expression in the POA and mediobasal hypothalamus, which contains the VMH, was lower in castrated males compared to intact males. However, treatment with testosterone or dihydrotestosterone (DHT), but not estradiol, in castrated males restored aromatase expression to intact levels²⁵. Thus, the modulation of aromatase expression by androgens may be direct since aromatase-positive cell bodies in the hypothalamus are co-expressed with ARs in mice²⁶.

Given the above observations combined with the findings that both androgenic signaling and aromatase are required for dominance behaviors in A. burtoni, we tested whether ARα mutant A. burtoni possessed reduced cyp19a1 expression in the brain compared to WT males. We provided social opportunities to both ARα mutant and WT males to become socially DOM, since DOM males perform the highest levels of dominance behaviors. We analyzed expression of cyp19a1 in SBN brain regions involved in the production of aggressive and reproductive behaviors through in situ hybridization. Additionally, we used in situ hybridization chain reaction (HCR) to determine whether either AR (α or β) and cyp19a1 are expressed in the same cells in the brain. We hypothesized that 1) since ARα mutant males performed significantly fewer aggressive and courtship displays compared to WTs, they would have lower cyp19a1 expression compared to WTs in regions of the SBN; and 2) that ARα and cyp19a1 would be expressed in the same cells in brain regions in the SBN.

MATERIALS AND METHODS

Animal subjects and housing

The A. burtoni males from the ARα mutant line used in this study for detection of cyp19a1 expression were involved in another experiment in which behavior, body coloration, and gonad mass results have already been published, but for which brain gene expression had not been assessed²⁷. Fish originated from a stock from Lake Tanganyika, Africa and were kept in laboratory conditions simulating their natural habitat (25°C; 12h day:12h night cycle)^28,29 in accordance with Association for Assessment and Accreditation of Laboratory Animal Care standards. All experimental procedures were approved by the Stanford University Administrative Panel for Laboratory Animal Care (Protocol #9882) and by the University of Houston Institutional Animal Care and Use Committee (Protocol #202000001).

ARα mutant males

We used DOM male A. burtoni with a mutant ARα (accession number, NW_005179415) allele generated using CRISPR/Cas9 gene editing. The process used for generating these mutants is described in full detail in Alward et al²⁷. The Cas9 enzyme was directed towards a site within exon 1 upstream from the DNA binding and ligand binding domains. Specifically, Cas9 was directed towards two regions within exon 1 using two single guide RNAs (sgRNAs) targeting sequence ARα-A, 5′-ACTGTGGCGGATACTTCTCG-3′ and sequence ARα-B, 5′-GGTGCGCAAACT-GTGACGCG-3′, whose cut sites were separated by 178 bp. sgRNAs were designed to have minimum to no off-target sites. Furthermore, the ARα mutant fish included in these experiments are offspring from a stable line that has been maintained through several out-crossings to un-related WT fish, further minimizing off-target effects. G1 offspring from injected fish carried a frameshift deletion within exon 1 of ARα of 50-bp. We outcrossed this allele to unrelated WT fish and then intercrossed heterozygous mutants to obtain homozygous null ARα mutants (ARα^d50/d50, hereon referred to as ARα^KO), Het ARα mutants (ARα^d50/+, hereon referred to as ARα^Het), and ARα wild types (ARα^+/+, hereon referred to as ARα^WT).

Housing: ARα mutants

Male offspring from AR^Het × AR^Het matings were housed in stable DOM tanks, wherein five to eight size-matched males were housed in a 121-liter tank with 10 to 15 females and five potential mating sites that simulated by halved terra cotta pots²⁷. Housing the males used in this study in stable DOM tanks ensures 1) all fish came from a similar social environment and 2) were housed in an environment that reliably induces traits of social dominance, including elevated androgen levels^15,30–32. In the previously published work on the fish used in the present study, there were no differences in androgen levels between ARα^WT, ARα^Het, and ARα^KO males, and these values were well within the range typical of DOM males^15,27. Males of each genotype were housed randomly and blind to the experimenter, as PCR-based genotyping was not performed until the harvesting of tissue at the end of the experiment.

Males were housed in stable DOM tanks for 4–12 weeks before being moved to a dyad assay set up to test DOM-typical behaviors that were assessed in the previously published work²⁷. Upon removal from the stable DOM tank, fish underwent a pre-assay procedure that took approximately 45 seconds. Fish were photographed, a small (1–2 mm) piece of their caudal fin was clipped for subsequent DNA isolation and genotyping, and then they were immediately moved to a 30-liter tank where they were housed with three smaller females and one smaller stimulus male, allowing the focal fish to maintain dominance characteristics. A camcorder (Canon VIXIA HF R80) mounted on a tripod was placed in front of their tank and recordings were made on the day they were added to the tank, then for two subsequent days. At 2 pm on the day after the second day, DOM focal fish were removed, and tissue was harvested for physiological measurements and blood was collected. Brains from 15 males were used for the current experiment (ARα^WT n=5, ARα^Het n= 5, ARα^KO n= 5).

Housing: WT fish used for HCR

WT fish were housed for a 4–12-week period in an 80-gallon, mixed-sex tank allowing for naturalistic social interactions to occur. Males were given the opportunity to rise in dominance by providing three to five potential mating sites and free access to females within the tank. Once the males became DOM as measured by enhanced reproductive activity, territoriality, and brightness in coloration, they were taken from their tank, dissected, and had tissue and blood harvested for physiological measurements. Similarly, female WT fish from the same community tank were allowed equal access to natural social interactions and access to males. Females were selected for dissection when they were in a gravid reproductive state as measured by observation of a distended abdomen assumed to be filled with eggs. Once the females met these criteria, they too were dissected, their brain and blood collected, and their physiological measures taken. Brains from two males and two females (WT^male n= 2, WT^female n=2) were collected and used for HCR.

Tissue collection

Upon removal from their dyad assay tank, DOM males from the ARα^Het cross were euthanized using only rapid cervical transection and their brains and blood collected. DOM males and females used for HCR were removed from their home community tank and euthanized in an ice bath slurry before cervical transection then their brains and blood collected. We measured standard length (SL) in cm from the tip of the mouth to the base of the tail fin. Body mass (BM) in grams was measured and recorded, and gonads were extracted and weighed. Blood for each animal was collected from the dorsal artery using two to three heparin coated capillary tubes (VWR^®). The blood was then spun down at 8000rpm for 10 mins in a microcentrifuge (VWR^®), the plasma collected and then stored at −80°C for later use in hormone enzyme immunoassay.

Brains of fish from the ARα^Het cross were dissected and fixed in 4% paraformaldehyde (PFA) in a 20 mL glass vial with a pH of 7.4 for two hours before being cryoprotected using a sucrose gradient starting with 15% sucrose (in 1xPBS; Gibco) overnight and subsequently 30% sucrose (in 1xPBS; Gibco) for 1–2 days (or until they sunk to the bottom of the vial). Brains from WT fish used for HCR were fixed in 4% paraformaldehyde (PFA) with a pH of 7.0 for 24 hours before being cryoprotected using a sucrose gradient starting with 15% sucrose (in 1xPBS; Gibco) overnight and subsequently 30% sucrose (in 1xPBS; Gibco) for 1–2 days (or until they sunk to the bottom of the vial). Once all brains had sunk indicating absorption of the sucrose, they were embedded in Neg-50 (Epredia) and stored at −80°C until sectioning at 30 μm on a cryostat (Thermo Scientific, HM525NX).

Histology

In situ hybridization

In situ hybridization (ISH) of ARα mutant brains was performed to analyze the expression of the brain aromatase (cyp19a1) as previously described³³. Briefly, PCR was used to amplify a sequence from cyp19a1 using primers: cyp19a1 forward, 5’-AGATGATAATCGCAGCCCCC-3’; cyp19a1 reverse, 5’-TAATACGACTCACTATAGGGGAGTGACCAGGATGGCCTT-3’. The PCR product (456bp) was subsequently confirmed using sanger sequencing and transcribed in vitro using T7 RNA polymerase (Promega) and Digoxygenin (DIG) labeled dNTP’s (Roche), the resultant RNA cyp19a1 probe was then used for ISH.

Tissue for ISH was cryosectioned coronally at 30 μm (Thermo Scientific, HM525NX) at −20°C and mounted on Superfrost^® slides (VWR^®) and allowed to dry for two hours before storage in −80°C. To increase tissue adhesion, we coated slides with mounted tissue in a gelatin solution before starting the ISH. Specifically, gelatin (1 mg/ml; Thermo Scientific) was applied to each slide and placed on a slide warmer (VWR^®) for 45 minutes at 50°C until the liquid evaporated. Subsequently, the slides were dried in a desiccator overnight before continuing with ISH.

All steps of the ISH protocol were performed in a 40 ml coplin jar unless otherwise noted. ISH began by fixing the tissue in 4% PFA (Thermo Scientific) before adding Proteinase K (10 mg/ml; Life Technologies). After a second round of fixation with 4% PFA, sections were incubated in prehybridization buffer (50% formamide [Thermo Scientific], 5X SSC [Gibco], 0.1% Tween-20 [Thermo Scientific], 0.1% CHAPS [Thermo Scientific], 5 mM EDTA [Thermo Scientific]) for 2 hours at 62°C and subsequently incubated overnight at the same temperature in hybridization buffer (50% formamide, 5X SSC, 0.1% Tween-20, 0.1% CHAPS, 5 mM EDTA, 1 mg/ml torula yeast RNA [Life Technologies], 100 ug/ml Heparin [Life Technologies], 1x Denhardt’s solution [Sigma]) with 32 ng of DIG-labeled RNA probe. Next day, slides were washed with 50% formamide (Thermo Fisher) and 2X SSC (Gibco) at 62°C. This was followed by three washes with 2X SSC at 37°C and treatment with RNaseA (200 ng/ml; Qiagen) in 2X SSC at 37°C. The slides were then washed in Maleic acid Buffer (MABT; 100 mM Maleic Acid [Thermo Scientific], 150 mM NaCl, 0.1% Tween-20), and subsequently blocked with MABT plus 2% BSA for 1.5 hours. Anti-DIG antibody fragments conjugated with alkaline phosphatase (Roche; 1:5000) were diluted in MABT: 2% BSA (Invitrogen) and then incubated at 4°C overnight. On day 3, slides were washed with MABT and subsequently with alkaline phosphatase buffer (100 mM Tris pH 9.5 [Thermo Scientific], 50 mM MgCl₂ [VWR^®], 100 mM NaCl [VWR], 0.1% Tween-20, 5mM Levamisole [Tetramisole][Sigma]) with NBT (37.5 mg/ml; Roche) and BCIP (94 mg/ml; Roche) for 12–16 hours at room temperature. Slides were then washed in 1x PBS and cover slipped with Aquapolymount aqueous mounting media.

Hybridized Chain Reaction (HCR)

In situ HCR was used to visualize the AR genes (ar1 or ARα and ar2 or ARβ) and brain aromatase (cyp19a1) using a fluorescent labeled RNA probe that were custom designed (Molecular Instruments). The sequences used to design the ar1 and ar2 probes were identical to those used by Harbott and colleagues, in which they were validated³⁴. Tissue for HCR was cryosectioned at 30 μm at −20°C taking coronal sections that were mounted on Superfrost^® slides (VWR). The mounted slides were allowed to dry for 24 hrs at room temperature before being stored at −80°C before HCR. HCR was performed according to manufacturer protocols, recommendations, and reagents, unless otherwise noted³⁵. In summary, tissue slides were exposed to LED light for 60 mins to reduce autofluorescence. Subsequently, tissue was immersed in 0.2% Triton X-100 (in 1x PBS) (Thermo Fisher; Gibco) for 30 mins to increase signal detection. Next, tissue was dehydrated with serial concentrations of EtOH (Thermo Fisher) at 50%, 70%, and 100%, before adding Proteinase K (1:2000) (Thermo Fisher) in 1xPBS. HCR probes were then hybridized to tissue by adding 16 nMs of RNA probe in hybridization buffer for 12–16 hrs at 37°C. The HCR probes used against ARα and ARβ were engineered (Molecular Instruments) using identical binding sequences from previously published work in A. burtoni ³⁴. The HCR probe used against cyp19a1 was made custom and is propriety information from the manufacturer (Molecular Instruments) and had zero complimentary binding to the paralogous gene cyp19a1a. Following the hybridization step, the tissue plus probe was exposed to probe amplifiers containing different fluorescent fluorophores per gene of interest by adding 60 nMs of fluorescent amplifier in amplification solution and incubating it at room temperature in a dark chamber overnight. Slides were then washed in 5X SSCT (0.1%Tween-20) (Gibco; Thermo Fisher), and cover slipped using Prolong Gold Antifade Mountant with DNA stain DAPI (Invitrogen^™). The fluorophores for cyp19a1 and each of the ARs had excitation wavelengths in the GFP and CY5 range, respectively; thus, two series from the same brain were run in tandem allowing for the visualization of both cyp19a1 with ARα and cyp19a1 with ARβ. Due to the low signal intensity of both ARα and ARβ, we did not use autofluorescence quenching methods, which reduced signal intensity in optimization and validation of our HCR protocol. While this preserved signal intensity for reliable anatomical localization of ARα and ARβ signal, artifactual autofluorescence from what are believed to be blood vessels remained in many of the sections. They were verified to be artifacts by their presence in all wavelength filters and thus disregarded in all HCR runs.

Microscopy and image analysis

Microscopy: cyp19a1 in AR mutants using ISH

All images of stained cyp19a1 ISH brains were obtained using Nikon eclipse 80i microscope and MicroFire^™ camera set. Images were taken at 4x, 10x, and 20x magnification, but only images at 4x were used for the quantification of cyp19a1 signal. Consistent exposure settings were used for taking pictures across each of the magnifications as well as across different regions of the brain.

Microscopy: Co-expression of cyp19a1 and ARα and ARβ using HCR

All images of brains in which cyp19a1, ARα, and ARβ expression was detected using HCR were obtained using Nikon eclipse Ti2 microscope and DS-Qi2, Fi3 camera set. Images were taken at 10x and 20x magnification using the stitching function to get a complete picture of the coronal section of interest. Additionally, z-stacks were taken at 30x magnification in the specific regions of interest allowing for visualization of possible co-expression of ARα and cyp19a1 and ARβ and cyp19a1. Furthermore, the same exposure settings were consistently used for all the pictures taken to ensure consistency across all brain regions analyzed. Images depicting cyp19a1 and ARβ were color changed using NIS elements (Nikon) software so that ARβ appeared magenta to distinguish from images depicting ARα and cyp19a1.

Quantification of cyp19a1 expression

We quantified cyp19a1 expression in areas of the SBN. In the forebrain, we quantified expression of cyp19a1 in the following regions of the SBN: the POA; subcomissural nucleus of the ventral telencephalon (Vs), a putative partial homolog of the BNST; and ventral nucleus of the ventral telencephalon (Vv), a putative homolog of the LS. In the midbrain we looked at the ventral tuberal nucleus (VTn), a putative homolog of the AH, and the anterior tuberal nucleus (ATn), a putative homolog of the VMH^10,36. Figure 1 illustrates the different brain regions analyzed and their relative location along the brain.

FIGURE 1.

A. burtoni coronal brain regions analyzed for in situ hybridization and in situ hybridization chain reaction. (A) shows the Vv (homolog of the LS), (B) shows the Vs (partial homolog of the BNST), (C) depicts the preoptic area (POA), (D) shows the ATn (VMH homolog), and the VTn (AH homolog). Abbreviations: AH, anterior hypothalamus; ATn, anterior tuberal nucleus; BNST, bed nucleus of the stria terminalis; LS, lateral septum; POA, preoptic area; VMN, ventromedial nucleus; Vs, subcomissural nucleus of the ventral telencephalon; VTn, ventral tuberal nucleus; Vv, ventral nucleus of the ventral telencephalon.

For certain brain regions across genotypes, tissue folding and loss during the ISH meant some fish could not be included in the analysis of that brain region. For Het and KO brains the number of fish that could be used for quantification of expression in certain brain regions was as low as two or three for either genotype. Given that both AR^Het and AR^KO males do not differ on the behavioral phenotypes relevant to the proposed link to aromatase as shown by Alward et al²⁷ (See Figure S1 A and B), they were combined into a single group called AR mutants, abbreviated as AR^Mut. We confirmed via unpaired two-tailed t-tests that cyp19a1 expression in all brain regions for AR^Het versus AR^KO males did not differ (See Table S1). For the Vv, only two WT fish could have expression quantified reliably (final sample size Vv: AR^WT=2, AR^Mut=5; Figure S2 A and B), so this region was excluded from further analysis. Final sample sizes for analysis of cyp19a1 expression across brain regions are as follows: Vs: AR^WT=4, AR^Mut=6; POA: AR^WT=4, AR^Mut=7; VTN: AR^WT=4, AR^Mut=8; ATN: AR^WT=4, AR^Mut=9. The sample sizes used here are similar to recently published work measuring gene expression using in situ hybridization and other cellular phenotypes in the SBN of multiple teleost fishes including A. burtoni³⁷, medaka (Oryzias latipes)³⁸, and bluehead wrasse (Thalassoma bifasciatum)³⁹.

Analysis of brain regions was done with the free source program Image J (imagej.nih.gov/ij/; RRID: SCR_003070) which allowed for tabulation of expression in regions of interest (ROI) throughout the brain. The images were first converted to 8-bit black and white images and a ROI was drawn surrounding the specific brain region of interest. Once an ROI was selected, the background of the 8-bit images was removed by changing the image threshold using the auto-threshold function on Image J and consistently choosing the same threshold settings for mutant and WT brains. Subsequently, the resulting image was subtracted from the original 8-bit picture resulting in a .tiff image file with prominent staining and low/non-present background (hereby called the optimal image). The ROI was once again opened, and the threshold of the optimal image was once again adjusted using the threshold function on ImageJ. The thresholded particles within the ROI of the optimal image were then analyzed and numerical values were obtained detailing the intensity and area of each of the particles within the ROI using the analyze particles function in ImageJ. These values were then summed for each hemisphere of the ROI giving the resulting numerical values for each brain region.

Statistics

Statistical analyses were performed using PRISM software (GraphPad Prism version 9.0) to compare cyp19a1 expression in mutant and wild type brains. Specifically, unpaired two-tailed t-tests were performed to compare cyp19a1 signal intensity and area of intensity. If equality of variance or normality assumptions for parametric statistics were not met, log transforms were used to meet those assumptions. Effect sizes were calculated as Cohen’s d and are reported in Table S2. Differences were considered significant at p≤0.05.

RESULTS

ARα regulates cyp19a1 expression in the VMH of male A. burtoni

WT males showed significantly higher cyp19a1 expression intensity compared to mutants in the ATn (t(11)=2.5383, p=0.0309; Figure 2). There were no significant differences in cyp19a1 expression in the other brain regions analyzed. For instance, cyp19a1 expression intensity in Vs (t(8)=0.1154, p=0.9110; Figure S2 C and D), the POA (t(9)=0.0503, p=0.6910; Figure S2 E and F), and VTn (t(11)=0.0888, p=0.9303; Figure S2 G and H) did not differ between WT and mutants. Table S1 shows these results in table form and the effect sizes for each individual brain region. The Vv (which was excluded from further analysis due to the small final sample size in the WT group) shows a trending difference between the WT and mutant group. No significant differences between WT and mutants were detected in the area of intensity of cyp19a1 expression in all brain regions analyzed (Figure S3).

FIGURE 2.

(A) cyp19a1 expression (graphed as intensity) in the ATn (VMH) between WT and mutant males that were given the social opportunity to become DOM. Bars represent mean ± standard error of the mean (SEM). Stars in the WT group represent each individual fish analyzed. The mutant group represents AR^het fish as circles and AR^KO individuals as triangles. (B) Shows representative images of cyp19a1 expression in WT vs mutant brains using in situ hybridization in the ATn. Abbreviation: AR, androgen receptor; ATn, anterior tuberal nucleus; DOM, dominant; HET, heterozygous; KO, knock out; VMH, ventromedial hypothalamus; WT, wild type.

Both ar1 and ar2 co-express with cyp19a1 throughout the SBN

HCR visualization of cyp191a expression along with ar1 (ARα gene) or ar2 (ARβ gene) expression showed that ARα and ARβ are both expressed, in part, within cyp19a1⁺ cells. Indeed, both ar1 and ar2 are frequently found surrounding cyp19a1⁺ cells along the midline of the brain. This is true for both the male and female brains analyzed (See Figure 3 for male ATn, Figure S4 for these patterns in other relevant male regions and Figures S5 and S6 for female brains). In addition, similar to previous work, cyp19a1 expression is restricted to the ventricular areas of the brain in both males and females with very little expression in areas other than the ventricles⁴⁰. Both ar1 and ar2 also appear distributed around the cyp19a1⁺ cells, as well as in other regions of the brain away from the ventricles.

FIGURE 3.

HCR co-expression, ARs and cyp19a1 (A–C): cyp19a1 and ar1 (ARα) expression in the ATn of a DOM male. The dashed circles in the merged panel (A) depict where co-expression of cyp19a1 and ar1 are present. Note how ar1 signal seems to surround the cyp19a1⁺ cells with DAPI staining depicting the cell’s nucleus. (D–F) Same brain region but with cyp19a1 and ar2 (ARβ) co-expression. Note the similar expression levels between ar1 and ar2, as well as a similar expression pattern of ar2 surrounding cyp19a1⁺ cells. Abbreviation: AR, androgen receptor; ATn, anterior tuberal nucleus; HCR, in situ hybridization chain reaction.

Expression of ARα and ARβ followed similar patterns compared to previous work³⁴. ARα appeared to be mostly concentrated in the anterior part of the brain including the Vv, Vs, POA, ATn, and VTn (See Figure S4). Likewise, ARβ shows higher expression patterns around the anterior part of the brain before reducing around the VTn (Figure S4 S–X), and once again showed up in the posterior parts of the brain (not pictured). Furthermore, ARα and ARβ signal appeared to be sparser than cyp191a1, whose expression appeared much more densely.

DISCUSSION

In the current study we sought to determine if ARα modulates the expression of cyp19a1 in the male A. burtoni brain. Given that ARα and estrogens regulate aggression in A. burtoni^14,16,27 and androgen signaling enhances aromatase expression in other species^23,24, we hypothesized that male A. burtoni with mutant ARα would possess lower cyp19a1 expression in regions of the SBN. We found that males deficient in ARα have lower cyp19a1 expression specifically in the ATn, a putative homolog of the mammalian VMH. These differences in cyp19a1 expression are particularly interesting since the VMH is involved in the production of aggressive behavior across different species. Given the conserved functions of both aromatase and the VMH in controlling aggression^18,41–46, we speculate that ARα governs aggressive displays in male A. burtoni through the enhancement of cyp19a1 expression in the ATn, which leads to enhanced synthesis and actions of estrogens within this region ^7,10. Importantly, cyp19a1 expression levels did not differ between ARα mutants with one or two non-functional alleles suggesting that there is not dosage-related functions of the impacts of ARα’s influence on cyp19a1 expression in the ATn, which mirror the findings on the behavioral deficits in these ARα mutants²⁷. While perhaps surprising, these findings are similar to those by Mosher and colleagues showing that both mice possessing one or two non-functional alleles encoding an enzyme that synthesizes androgens have deficits in dominance behavior compared to WT mice⁴⁷. Our results thus imply there is no dosage compensation of ARα as it pertains to the modification of cyp19a1 expression in the ATn. Future inquiries into ar1 gene silencing, ARα subcellular actions, and ARα dimerization mechanisms may be fruitful in revealing the precise ways in which ARα controls cyp19a1 expression.

Outside of the ATn, no other brain regions exhibited altered cyp19a1 expression; however, the low sample size for the Vv precluded us from concluding whether ARα is or is not necessary for cyp19a1 expression in this region. Furthermore, it could be that ARα modulates cyp19a1 expression within additional SBN regions but only in very small cellular populations or subregions that we did not measure. It should also be noted that perhaps it is not surprising that only cyp19a1 expression within a single region was impacted because the full suite of androgenic control of brain processes is likely split between both ARα and ARβ. In other words, it could be that ARβ controls cyp19a1 expression in other regions in the brain and additionally that in other regions of the brain both AR paralogs are required for the modulation of cyp19a1 expression. Given the widespread co-expression between ar1 and cyp19a1 and ar2 and cyp19a1 in the majority of SBN regions that we have shown, it is reasonable to suppose that both AR paralogs control cyp19a1 expression throughout the SBN in a modular manner.

Additionally, our co-expression results are consistent with previous findings from Forlano et. al^48,49 and others ⁵⁰ that showed AR mRNA and aromatase mRNA/protein are produced in the same brain regions throughout the brain of the midshipman fish (Porichthys notatus), albeit not within the same cells. Compared to ar1 levels of overlapping expression of cyp19a1⁺ cells and ar2 appeared to be much lower Collectively, this suggests that both ARα and ARβ can act directly within cyp19a1⁺ cells to modulate cyp19a1 expression in A. burtoni. These results are also intriguing considering in A. burtoni cyp19a1 expression is exclusive to glial cells, including radial glia⁴⁰. It has been hypothesized that radial glia in A. burtoni do not act as progenitor cells but instead act as a scaffold in guiding newly generated cells to their final destination in the brain where they will integrate into a functional circuit⁵¹. Therefore, we postulate that both ARs may modulate neuroplasticity in the A. burtoni brain through actions in radial glia cells. As this pertains to findings in ARα mutants, we further propose that one way in which ARα modulates dominance is via alterations in neuroplasticity within or near the putative fish homolog of the VMH. These hypotheses have precedent based on a large body of work in birds, rodents, and fish showing that androgenic signaling is key for driving neurogenesis and neuron survival in the brain^52–56.

Extensive research suggests the VMH is critical for the production of aggressive and reproductive behaviors^45,46,57,58. In zebrafish and A. burtoni, it has been shown that ARs modulate aggression ^27,59. The present findings in A. burtoni lacking ARα are thus particularly intriguing considering the important role of the VMH and ARs in the production of behavior in other vertebrate species. Yet, available findings have not tested the role of AR signaling in the VMH in the control of male-typical aggression or aromatase expression in teleost or non-teleost vertebrate species. Instead, extensive evidence indicates estrogenic signaling within the VMH is critical for the production of aggression in both male and female mice ^45,46. Dugger et al⁶⁰ found that AR mutant male mice, which perform significantly fewer aggressive and mating acts compared to WT males, possessed a feminized VMH as measured by volume of this brain region. Additionally, androgens increase aromatase expression in male song sparrows²⁴ and male rats²⁵. Combined with our findings, it is reasonable to speculate that androgen signaling may serve a conserved function across vertebrate taxa to enhance aromatase expression in the VMH, which in turn drives male-typical aggression. This hypothesis is testable and warrants further investigation.

CONCLUSIONS

Overall, our findings showed that ARα deficiency reduces cyp19a1 expression in the A. burtoni VMH. This is an intriguing finding when considering the importance of the VMH in the production of aggressive behavior across different vertebrate species and suggests that appropriate levels of cyp19a1 may be an important component in the production of aggressive and reproductive behaviors in males of our species. However, many unknowns remain including the specific ways in which ARα indirectly influences cyp19a1 expression. Future studies should also aim to investigate questions such as how on a sub-cellular level ARα influences cyp19a1 expression, including whether it occurs via genomic and/or non-genomic mechanisms and if ARβ has different effects in the expression of cyp19a1 in regions of the SBN. Investigations into novel teleost steroid signaling genes will enrich our understanding of the molecular and neural mechanisms that contribute to the generation of behavior.