Abstract
Aggression is a crucial behavior that impacts access to limited resources in different environmental contexts. Androgens synthesized by the gonads promote aggression during the breeding season. However, aggression can be expressed during the non-breeding season, despite low androgen synthesis by the gonads. The brain can also synthesize steroids (“neurosteroids”), including androgens, which might promote aggression during the non-breeding season. Male song sparrows, Melospiza melodia, are territorial year-round and allow the study of seasonal changes in the steroid modulation of aggression. Here, we quantified steroids following a simulated territorial intrusion (STI) for 10 min in wild adult male song sparrows during the breeding and non-breeding seasons. Using liquid chromatography-tandem mass spectrometry (LC-MS/MS), we examined 11 steroids: pregnenolone, progesterone, corticosterone, dehydroepiandrosterone, androstenedione, testosterone, 5α-dihydrotestosterone, 17β-estradiol, 17α-estradiol, estriol, and estrone. Steroids were measured in blood and 10 microdissected brain regions that regulate social behavior. In both seasons, STI increased corticosterone in the blood and brain. In the breeding season, STI had no rapid effects on androgens or estrogens. Intriguingly, in the non-breeding season, STI increased testosterone and androstenedione in several behaviorally relevant regions, but not in the blood, where androgens remained non-detectable. Also in the non-breeding season, STI increased progesterone in the blood and specific brain regions. Overall, rapid socially modulated changes in brain steroid levels are more prominent during the non-breeding season. Brain steroid levels vary with season and social context in a region-specific manner and suggest a role for neuroandrogens in aggression during the non-breeding season.
Significance Statement
Steroids are important modulators of the brain and behavior. Steroids are secreted by glands into the bloodstream to exert their actions on target tissues. In addition, the brain itself can locally produce steroids (“neurosteroids”). We examined circulating and neural steroids across seasons and in response to an aggressive interaction in wild male song sparrows. Local androgen levels in the brain rapidly increase in response to an aggressive interaction in the non-breeding season, but not in the breeding season, indicating that neurosteroid production is dependent on the season and social context.
Introduction
Aggression is a social behavior that plays key roles in many life history stages (Nelson, 2005; Soma et al., 2008; Kennedy, 2022). In the breeding season, males often compete for territories and mates (Wingfield, 2005). Females are also aggressive when defending offspring, mates, or breeding sites (Rosvall et al., 2020). Beyond reproductive contexts, animals compete for territories that provide food and shelter during the non-breeding season (Rieger et al., 2022). Consequently, aggressive interactions have major impacts on reproductive success and winter survival (Arcese, 1989; Wingfield et al., 2006; Quintana et al., 2021; Rieger et al., 2022).
Social behaviors, including aggression, are regulated by a neural circuit known as the social decision-making network (SDMN; Newman, 1999; O’Connell and Hofmann, 2011; Lischinsky and Lin, 2020). The SDMN includes regions in the forebrain, midbrain, and hindbrain and regulates multiple social behaviors in all vertebrates (Goodson, 2005; Teles et al., 2015; Gossman et al., 2021; Kelly, 2022). Every node of the SDMN expresses sex steroid receptors, and sex steroids modulate the SDMN and social behaviors (Goodson and Kabelik, 2009; Trainor and Nelson, 2012; Tobiansky and Fuxjager, 2020).
Aggressive behavior is modulated by steroid hormones. In reproductive contexts, aggression depends on gonadal testosterone (T) as shown by castration and androgen replacement experiments (Vandenbergh, 1971; Balthazart, 1983; Moore and Marler, 1987; Wingfield and Hahn, 1994). Interestingly, during the non-breeding season, aggression persists in some species, despite low circulating androgen levels and even after castration (Wingfield, 1994; Jalabert et al., 2015). Non-breeding aggression might depend on androgens from a non-gonadal source (Soma et al., 2008, 2015; Jalabert et al., 2015, 2018; Munley et al., 2022; Zubizarreta et al., 2023). Therefore, the neuroendocrine mechanisms that support aggression change seasonally, and very little is known about this seasonal switch.
The brain itself can locally produce sex steroids, either de novo from cholesterol or from circulating precursors (Schmidt et al., 2008; Schlinger, 2015). The brain expresses all the necessary enzymes for sex steroid synthesis (Tsutsui, 2011; Hojo and Kawato, 2018). Brain-derived steroids, known as neurosteroids, were first characterized in rodents (Baulieu, 1991; Mellon et al., 2001; Hojo and Kawato, 2018) and later in other vertebrates such as birds (Schmidt et al., 2008; Tsutsui, 2011; Schlinger, 2015).
Song sparrows, Melospiza melodia, are territorial year-round and therefore an excellent animal model to study seasonal changes in the steroid modulation of aggression (Wingfield and Soma, 2002). In breeding males, plasma T levels are high and increase even further after an aggressive interaction, and androgen receptor antagonism reduces aggression (Wingfield, 1985; Sperry et al., 2010). Interestingly, non-breeding males aggressively defend territories just as fiercely as breeding males. However, in non-breeding males, the testes are fully regressed, systemic levels of androgens and estrogens are non-detectable (Jalabert et al., 2021), and castration does not reduce aggression (Wingfield, 1994). However, sex steroids still regulate non-breeding aggression. Inhibition of aromatase, which converts androgens into estrogens, decreases non-breeding aggression, but androgen receptor antagonism does not (Soma et al., 2000a,b; Sperry et al., 2010).
In song sparrows, steroidogenic enzyme activities in the brain change with season and social context. An enzyme necessary to produce androgens, 3β-hydroxysteroid dehydrogenase/Δ5–Δ4 isomerase (3β-HSD), is present in the forebrain, and its activity is higher during the non-breeding season than the breeding season. Moreover, forebrain 3β-HSD activity is rapidly upregulated by an aggressive interaction in the non-breeding season (Pradhan et al., 2010). In contrast, activity of 5α-reductase, which converts T to 5α-dihydrotestosterone (5α-DHT), in the forebrain is higher during the breeding season than the non-breeding season (Soma et al., 2003). Aromatase is expressed during both breeding and non-breeding seasons (Soma et al., 2003; Wacker et al., 2010). Together, these data provide evidence of neural steroidogenesis in a region-specific manner to support seasonal changes in social behavior.
Here, we studied the effects of a brief simulated territorial intrusion (STI) on aggressive behavior and steroids during the breeding and non-breeding seasons in free-living male song sparrows. We examined 11 steroids in whole blood (hereafter “blood”), plasma, and 10 microdissected brain regions using liquid chromatography-tandem mass spectrometry (LC-MS/MS).
Materials and Methods
Field procedures
Subjects were free-living adult male song sparrows in the breeding season (April 25 to May 8, 2019) and non-breeding season (Oct. 16 to Nov. 3, 2019). Field sites were located near Vancouver, British Columbia, Canada. In both seasons, subjects were randomly assigned to one of two treatment groups: STI or control (CON; n = 10 per group in breeding season; n = 9 per group in non-breeding season). STI subjects were exposed to conspecific song playback from a speaker and to a live caged conspecific male decoy placed in their territories for 10 min (duration determined in a pilot study). The song playback included songs from several individuals, with 10 repetitions per song at 10 s intervals (Wingfield, 1985; Maddison et al., 2012). For CON subjects, an empty cage and a silent speaker were placed in a subject’s territory for 10 min. During the CON or STI condition, we measured aggressive responses (Wingfield, 1985; Wingfield and Hahn, 1994) including song latency, flight latency, number of songs, number of flights, and time spent within 5 m of the decoy. Then a mist net (that was set up ahead of time) was rapidly unfurled, and subjects were captured using a mist net and conspecific song playback for a maximum of 5 min (breeding CON, 2.1 ± 0.6 min; breeding STI, 1.5 ± 0.5 min; non-breeding CON, 1.9 ± 0.6 min; non-breeding STI, 1.8 ± 0.8 min). The amount of playback used to catch subjects was similar in all four groups (F(3,34) = 0.71; p = 0.55). Immediately after capture, the subject was rapidly and deeply anesthetized with isoflurane and then rapidly decapitated. No more than 3 min elapsed between capture and killing to minimize the effects of handling (breeding CON, 2.5 ± 0.1 min; breeding STI, 2.5 ± 0.2 min; non-breeding CON, 2.8 ± 0.05 min; non-breeding STI, 2.6 ± 0.1 min). Handling duration was similar in all four groups (F(3,34) = 1.37; p = 0.27). The brain was immediately dissected from the skull and snap frozen on powdered dry ice. Trunk blood was collected in heparinized microhematocrit tubes (Fisher Scientific) that were kept on ice packs until returning to the laboratory (within 5 h). Once in the laboratory, blood was divided into two aliquots: one half of the blood sample was frozen, and the other half was centrifuged and plasma was collected. All samples were stored at −70°C.
All procedures were compiled with the Canadian Council on Animal Care, and protocols were approved by the Canadian Wildlife Service and the UBC Animal Care Committee.
Brain microdissection
The Palkovits punch technique (Palkovits, 1973) was used as before (Charlier et al., 2011; Heimovics et al., 2016; Jalabert et al., 2021, 2022). Microdissected brain tissue was collected from 10 brain regions (Fig. 1), including nodes of the SDMN (O’Connell and Hofmann, 2011): the caudal portion of the preoptic area (POA), anterior hypothalamus (AH), lateral septum (LS), bed nucleus of the stria terminalis (BNST), ventromedial hypothalamus (VMH), ventral tegmental area (VTA), central gray (CG), caudomedial nidopallium (NCM), nucleus taeniae of the amygdala (TnA; homolog of the mammalian medial amygdala), and cerebellum (Cb). Brains were sectioned in the coronal plane at 300 µm on a cryostat, and bilateral punches were collected. The same punch size (1 mm diameter) was used for all brain regions. Depending on the size of the brain region, a total of four or six punches were collected and tissue amount was 0.98 or 1.47 mg, respectively. Punches were expelled into cold 2 ml polypropylene tubes (Sarstedt) each containing five zirconium ceramic oxide beads (1.4 mm diameter) and stored at −70°C until processing.
Representative punch locations in the song sparrow brain. A–F, Coronal sections (300 µm) presented from rostral to caudal. Punches (1 mm diameter) are represented by circles (approximately to scale). Abbreviations: POA, caudal preoptic area; LS, lateral septum; BNST, bed nucleus of the stria terminalis; CoA, anterior commissure; AH, anterior hypothalamus; VMH, ventromedial hypothalamus; OM, tractus occipito-mesencephalicus; NIII, nervus oculomotorius; VTA, ventral tegmental area; PC, posterior commissure; CG, central gray; NCM, caudomedial nidopallium; TnA, nucleus taeniae of the amygdala; Cb, cerebellum. Adapted from Jalabert et al. (2021).
Reagents
High-performance liquid chromatography (HPLC)-grade acetonitrile, hexane, and methanol were from Fisher Chemical. Stock solutions of steroids were prepared in HPLC-grade methanol, and calibration curves and quality controls (QCs) were prepared in 50% methanol from certified reference standards from Cerilliant or made in-house. Deuterated internal standards (IS) of pregnenolone-d4, progesterone-d9, corticosterone-d8, dehydroepiandrosterone-d6, testosterone-d5, and 17β-estradiol-d4 (C/D/N Isotopes) stock solutions were prepared in methanol and further diluted with 50% methanol to a final working solution of 0.04 ng/ml for progesterone-d9 and testosterone-d5; 0.4 ng/ml for corticosterone-d8 and 17β-estradiol-d4; 1.2 ng/ml for dehydroepiandrosterone-d6; and 2 ng/ml for pregnenolone-d4.
Steroid extraction
Steroids were extracted from blood (5 µl, which weighs 5.12 mg), plasma (10 µl), and brain tissue (amount detailed above) using liquid–liquid extraction, as before (Jalabert et al., 2021). Compared with plasma, blood more accurately reflects circulating (systemic) levels of steroids, allowing for comparisons of systemic and local steroid levels. Plasma steroid concentrations generally overestimate systemic steroid concentrations by twofold (Taves et al., 2010, 2011; Hamden et al., 2021a; Jalabert et al., 2021, 2022). We included plasma to compare with previous studies that used plasma.
Briefly, five zirconium ceramic oxide beads were added to each vial. One milliliter of acetonitrile was added, and 50 µl of the deuterated IS was added (except for “double blanks”) to track recovery and matrix interference. Samples were homogenized using a bead mill homogenizer. Then samples were centrifuged at 16,100 × g for 5 min, and 1 ml of supernatant was transferred to a precleaned borosilicate glass culture tube. Then 0.5 ml of hexane was added, and samples were vortexed and centrifuged at 3,200 × g for 2 min. The hexane was removed and discarded, and the extracts were dried at 60°C for 45 min in a vacuum centrifuge. Samples were reconstituted with 55 µl of 25% methanol, transferred to 0.6 ml polypropylene microcentrifuge tubes, and centrifuged at 16,100 × g for 2 min. Then 50 µl of supernatant was transferred to an LC vial insert and stored at −20°C until injection.
Samples were processed alongside calibration curves, QCs, blanks, and double blanks. The lower limit of quantification (LLOQ) was determined as the lowest standard on the calibration curve in which the analyte peak signal was at a signal/noise ratio >10. The calibration curve ranged from 0.03 to 1,000 pg/tube for androstenedione and T (Fig. 2), 0.05 to 1,000 pg/tube for corticosterone, progesterone, and 5α-DHT; 0.1 to 1,000 pg/tube for estrone; 0.2 to 1,000 pg/tube for 17α-estradiol, 17β-estradiol, and estriol; 1 to 10,000 pg/tube for dehydroepiandrosterone; and 5 to 1,000 pg/tube for pregnenolone (Table 1). For assay assessment, QCs of two different quantities were used to cover a wide range of steroid concentrations. The low QCs contained 0.8pg of each steroid, except for dehydroepiandrosterone which had 8 pg. The high QCs contained 50 pg of each steroid, except for dehydroepiandrosterone which had 500 pg. In each of the five assays, three replicates of QCs were included at each quantity, totaling 15 low QCs and 15 high QCs. Precision was assessed using the coefficient of variation (CV) by comparing the replicates of the QCs within runs (intra-assay variation) and across runs (inter-assay variation; Table 2). Accuracy was determined by measuring the recovery of the QCs with a known concentration of steroid (Table 2). As shown in Table 2, all steroids demonstrated high accuracy and precision in the assays, which validates the conducted assays for the present work. Lastly, all blanks and double blanks were below the LLOQ.
Calibration curves of (A) androstenedione and (B) testosterone, with insets displaying the lowest standards on the curve. The area ratio is calculated by dividing an analyte peak area with its deuterated internal standard peak area in the same sample. Calibration curves ranged from 0.03 to 1,000 pg.
LC-MS/MS calibration curves and analyte retention times: linear regressions and lower limits of quantification (LLOQ)
LC-MS/MS assay quality controls (QCs): precision and accuracy
Steroid analysis by LC-MS/MS
Steroids were quantified using a Sciex QTRAP 6500 UHPLC-MS/MS system as previously described (Jalabert et al., 2021). Samples were transferred into a refrigerated autoinjector (15°C). Then, 45 µl of each sample was injected into a Nexera X2 UHPLC system (Shimadzu), passed through a KrudKatcher ULTRA HPLC In-Line Filter (Phenomenex) followed by an Agilent 120 HPH C18 guard column (2.1 mm) and separated on an Agilent 120 HPH C18 column (2.1 × 50 mm; 2.7 µm; at 40°C) using 0.1 mM ammonium fluoride in Milli-Q water as mobile phase A (MPA) and HPLC-grade methanol as mobile phase B (MPB). The flow rate was 0.4 ml/min. During loading, MPB was at 10% for 0.5 min, and from 0.6 to 4 min, the gradient profile was at 42% MPB, which was ramped to 60% MPB until 9.4 min. From 9.4 to 9.5 min, the gradient was 60–70% MPB, which was ramped to 98% MPB until 11.9 min and finally a column wash from 11.9 to 13.4 min at 98% MPB. The MPB was then returned to starting conditions of 10% MPB for 1 min. Total run time was 14.9 min. The needle was rinsed externally before and after each sample injection with 100% isopropanol.
We used two multiple reaction monitoring (MRM) transitions for each steroid and one MRM transition for each deuterated internal standard (Jalabert et al., 2021). The ion ratio between the two transitions (the quantifier and qualifier ions) was used as additional confirmation of analyte identity. Steroid concentrations were acquired on a Sciex 6500 Qtrap triple quadrupole tandem mass spectrometer (Sciex) in positive electrospray ionization mode for all steroids except estrogens, which were acquired in negative electrospray ionization mode. Subjects from different groups were included within and across assays. Chromatograms were analyzed blind to season and treatment. A sample was considered non-detectable if the signal-to-noise ratio for the quantifier transition was lower than 10.
Statistical analysis
Samples that fell below the LLOQ were considered missing not at random (MNAR; Wei et al., 2018a,b). Those samples have steroid concentrations somewhere between zero and the LLOQ. Their statistical analysis is challenging, because different methods introduce different biases, from artificially lowering the group mean (when setting missing values to 0) to increasing the group mean (when removing missing values from the dataset; Handelsman and Ly, 2019). As before, here MNAR values were imputed via quantile regression imputation of left-censored missing data using the MetImp web tool (Tobiansky et al., 2018, 2021; Wei et al., 2018a,b; Hamden et al., 2021a,b; Jalabert et al., 2021, 2022; Salehzadeh et al., 2022; Zubizarreta et al., 2023; Gray et al., 2024). This method uses the lowest observed value to set the upper truncation point (i.e., the LLOQ) and uses predictive information from detectable samples to impute non-detectable ones. The imputed values have a normal distribution within a reasonable range (zero to LLOQ) introducing minimal bias for statistical analysis (Xia et al., 2009; Wei et al., 2018a,b). Therefore, when 20% or more of samples were detectable in a brain region or blood, values were imputed. When <20% of samples in a brain region or blood were detectable, imputations cannot be performed, and therefore missing values were assigned a zero. Imputations were conducted consistently on all groups. Data were imputed for each season and treatment group independently.
Statistics were conducted using GraphPad Prism version 9.02 (GraphPad Software). When necessary, data were log transformed prior to analysis. For aggressive behaviors, the effects of STI and season were analyzed using a two-way ANOVA followed by Tukey’s multiple-comparison tests (corrected p values are reported). For steroid levels in blood, the effects of STI were examined using a t test. For steroid levels in brain regions, the effects of STI and region were analyzed using repeated mixed-measures two-way ANOVA with a between-subjects factor (STI) and a within-subjects factor (region). Analyses were followed by Sidak multiple-comparison tests (corrected p values are reported). When all values in a group were non-detectable in a brain region, 1 was added to each value of both groups and then data were log transformed. Then, the effects of STI in the brain region were tested using a Mann–Whitney test. Significance criterion was set at p ≤ 0.05. Graphs show the mean ± standard error of the mean (SEM) and are presented using the non-transformed data.
Results
Aggressive behavior in the breeding and non-breeding seasons
The STI increased aggressive behaviors in both breeding and non-breeding seasons (Figs. 3, 4). For song latency and flight latency, there was a significant main effect of STI but no significant main effect of season and no significant STI × season interaction. The STI decreased song latency (F(1,34) = 125.7; p < 0.0001; Fig. 3A) and flight latency (F(1,34) = 271.2; p < 0.0001; Fig. 3B).
Effect of a 10 min STI on aggressive responses of wild adult male song sparrows during the breeding (green) and non-breeding (blue) seasons. Bar graphs represent (A) song and (B) flight latencies, for control and STI subjects. Values are expressed as the mean ± SEM. n = 10 for breeding group, n = 9 for non-breeding group. ****p ≤ 0.0001.
Effect of a 10 min STI on aggressive responses of wild adult male song sparrows during the breeding (green) and non-breeding (blue) seasons. Bar graphs represent (A) number of songs, (B) number of flights, and (C) time in 5 m, for control and STI subjects. Values are expressed as the mean ± SEM. n = 10 for breeding group, n = 9 for non-breeding group. *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001.
For number of songs, there was a significant main effect of STI (F(1,34) = 141.7; p < 0.0001; Fig. 4A), a significant main effect of season (F(1,34) = 4.374; p = 0.04), but no significant interaction (F(1,34) = 2.842; p = 0.10). Subjects sung more during the breeding season than non-breeding season and STI increased the number of songs in both seasons.
For number of flights, there was a significant main effect of STI (F(1,34) = 503.8; p < 0.0001; Fig. 4B), a significant main effect of season (F(1,34) = 5.755; p = 0.02), and a significant STI x season interaction (F(1,34) = 10.35; p = 0.003). Post-hoc comparisons revealed that STI increased the number of flights in both seasons and that STI subjects made more flights during the non-breeding season than the breeding season (p = 0.002; Fig. 4B).
For time in 5 m, there was a significant main effect of STI (F(1,34) = 319.9, p < 0.0001; Fig. 4C), a significant main effect of season (F(1,34) = 6.121, p = 0.02), and a significant STI × season interaction (F(1,34) = 6.121; p = 0.02). Post hoc comparisons revealed that STI increased the time spent in 5 m in both seasons and that STI subjects spent more time in 5 m during the non-breeding season than those during the breeding season (p = 0.007; Fig. 4C).
Steroid levels in breeding males
In the breeding season, progesterone was detectable in blood and most brain regions, and progesterone levels were generally higher in blood than those in brain (Fig. 5A). STI did not affect progesterone levels in the blood (t(18) = 1.481; p = 0.16). In the brain, progesterone showed a trend for a main effect of STI (F(1,18) = 3.609; p = 0.07), a significant main effect of region (F(9,162) = 6.394; p < 0.0001), and a significant STI × region interaction (F(9,162) = 4.710; p < 0.0001). Post hoc comparisons revealed that STI increased progesterone levels in the VTA (p = 0.008) and NCM (p = 0.02) but tended to decrease progesterone levels in the BNST (p = 0.08). No effects were observed in other brain regions (all p > 0.1).
Effect of a 10 min STI on circulating and brain steroid levels of wild adult male song sparrows during the breeding season. Bar graphs represent concentrations of (A) progesterone and (B) corticosterone for control and STI subjects. Values are expressed as the mean ± SEM. n = 10 per group. *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001.
Corticosterone levels were generally greater in blood than those in brain (Fig. 5B). STI significantly increased corticosterone levels in the blood (t(18) = 4.934; p = 0.0001). In the brain, corticosterone showed a significant main effect of STI (F(1,18) = 12.22; p = 0.003) and a significant main effect of region (F(9,162) = 36.91; p < 0.0001) and a trend for a STI × region interaction (F(9,162) = 1.878; p = 0.06). STI increased corticosterone levels in every brain region.
Androstenedione levels in blood were not affected by STI (t(18) = 0.509; p = 0.62; Fig. 6A). Similarly, brain androstenedione levels did not show a significant main effect of STI (F(1,18) = 0.245; p = 0.63) but showed a significant main effect of region (F(9,162) = 32.03; p < 0.0001) and a trend for a STI × region interaction (F(9,162) = 1.804; p = 0.07).
Effect of a 10 min STI on circulating and brain androgen levels of wild adult male song sparrows during the breeding season. Bar graphs represent concentrations of (A) androstenedione, (B) testosterone, and (C) 5α-dihydrotestosterone for control and STI subjects. Values are expressed as the mean ± SEM. n = 10 per group.
For T, blood levels were generally higher than brain levels (Fig. 6B). In the blood, STI had no significant effect on T levels (t(18) = 0.584; p = 0.57). In the brain, there was a significant main effect of region (F(9,162) = 49.65; p < 0.0001) but no significant main effect of STI (F(1,18) = 0.036; p = 0.85) or an STI × region interaction (F(9,162) = 0.262; p = 0.98).
For 5α-DHT, blood levels were generally lower than brain levels (Fig. 6C). In the blood, STI had no significant effect on 5α-DHT levels (t(18) = 1.491; p = 0.15). In the brain, there was a significant main effect of region (F(9,162) = 36.53; p < 0.0001) but no significant main effect of STI (F(1,18) = 0.664; p = 0.43) or an STI × region interaction (F(9,162) = 0.603; p = 0.79).
Estrone and estradiol were not detectable in blood of breeding animals, regardless of treatment group (Fig. 7). In the brain, estrone showed a significant main effect of region (F(9,162) = 16.59; p < 0.0001) but no significant main effect of STI (F(1,18) = 0.4512; p = 0.51) or an STI × region interaction (F(9,162) = 0.503; p = 0.87). Similarly, brain 17β-estradiol showed a significant main effect of region (F(6,108) = 15.26; p < 0.0001) but no significant main effect of STI (F(1,18) = 0.915; p = 0.35) or an STI × region interaction (F(6,108) = 1.333; p = 0.25). 17β-Estradiol was non-detectable in VTA and CG in both groups.
Effect of a 10 min STI on circulating and brain estrogen levels of wild adult male song sparrows during the breeding season. Bar graphs represent concentrations of (A) estrone and (B) 17β-estradiol for control and STI subjects. Values are expressed as the mean ± SEM. n = 10 per group. nd, non-detectable.
Pregnenolone, dehydroepiandrosterone, 17α-estradiol, and estriol were non-detectable in blood and brain samples from breeding subjects.
Steroid levels in non-breeding males
In the non-breeding season, progesterone levels were generally higher in blood than those in brain (Fig. 8A). STI significantly increased the levels of progesterone in blood (t(16) = 3.404; p = 0.004). In the brain, STI increased progesterone levels in the AH (p = 0.03), VTA (p = 0.03), NCM (p < 0.0001), TnA (p < 0.0001), and Cb (p = 0.02). In contrast, STI decreased progesterone levels in the POA (p = 0.04) and did not affect progesterone levels in LS, BNST, VMH, and CG (all p > 0.1).
Effect of a 10 min STI on circulating and brain steroid levels of wild adult male song sparrows during the non-breeding season. Bar graphs represent concentrations of (A) progesterone and (B) corticosterone, for control and STI subjects. Values are expressed as the mean ± SEM. n = 9 per group. *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001. nd, non-detectable.
Corticosterone levels were generally greater in blood than those in brain (Fig. 8B). STI significantly increased corticosterone levels in the blood (t(16) = 5.510; p < 0.0001). In the brain, corticosterone showed a significant main effect of STI (F(1,16) = 30.80; p < 0.0001) and a significant main effect of region (F(9,144) = 42.57; p < 0.0001) but not an STI × region interaction (F(9,144) = 1.674; p = 0.10).
Androstenedione was non-detectable in blood and plasma of both CON and STI subjects (Table 3; Figs. 9E,G, 10A). In the brain, STI significantly increased androstenedione in BNST (p = 0.003), NCM (p < 0.0001; Fig. 9F,H), TnA (p < 0.0001), and Cb (p < 0.0001), decreased androstenedione in the VTA (p < 0.0001), and did not affect androstenedione in the VMH (p = 0.11; Fig. 10A). Androstenedione was non-detectable in nearly all samples of POA, AH, LS, and CG in both groups (Table 3).
Representative chromatograms for the two MRM transitions of androstenedione (left panel, quantifier m/z 287.2 → 109.1, qualifier m/z 287.2 → 97.2) and testosterone (right panel, quantifier m/z 289.0 → 97.0, qualifier m/z 289.0 → 109.1). Retention times for androstenedione (7.40 min) and testosterone (7.98 min) are indicated by the arrows. Samples were neat solution (in gray scale), (A, I) blank (no analyte) and (B, J) certified reference standard (0.1 pg), and wild male song sparrow blood and brain from breeding (green) and non-breeding (blue) seasons. C, Blood breeding control, (D) caudomedial nidopallium (NCM) breeding control, (E) blood non-breeding control, (F) NCM non-breeding control, (G) blood non-breeding STI (simulated territorial intrusion), (H) NCM non-breeding STI, (K) blood breeding control, (L) anterior hypothalamus (AH) breeding control, (M) blood non-breeding control, (N) AH non-breeding control, (O) blood non-breeding STI, (P) AH non-breeding STI. The ion ratio between quantifier and qualifier ions was similar between detectable samples and standards. For analyte quantification, a peak signal/noise ratio >10 was the criterion for detection. Blanks and non-breeding control samples were non-detectable for both analytes. Standards and breeding samples were detectable for both analytes. During the non-breeding season, both androstenedione and testosterone were detectable in brain samples but non-detectable in blood samples. The intensity was measured in counts per second (cps). n = 10 for breeding season and n = 9 for non-breeding season. The LLOQs were the lowest standard on the calibration curve in which the analyte peak had a signal/noise ratio >10.
Effect of a 10 min STI on circulating and brain androgen levels of wild adult male song sparrows during the non-breeding season. Bar graphs represent concentrations of (A) androstenedione and (B) testosterone for control and STI subjects. Values are expressed as the mean ± SEM. n = 9 per group. ***p ≤ 0.001, ****p ≤ 0.0001. nd, non-detectable.
Percent of samples with detectable androstenedione and testosterone in wild adult male song sparrows exposed to control or STI
T was non-detectable in blood and plasma of both CON and STI subjects (Table 3; Figs. 9M,O, 10B). STI increased T levels in POA (p < 0.0001), AH (p < 0.0001; Fig. 9N,P), and TnA (p < 0.0001). In contrast, STI tended to decrease T in the CG (p = 0.08) and had no effect on T in LS, BNST, NCM, and Cb (all p > 0.1; Fig. 10B). T was non-detectable in nearly all samples of VMH and VTA in both groups (Table 3).
Pregnenolone, dehydroepiandrosterone, 5α-DHT, estrone, 17α-estradiol, 17β-estradiol, and estriol were non-detectable in blood and brain samples from non-breeding subjects.
Discussion
Here, we demonstrate spatial specificity and rapid regulation of local steroid levels in the brain, particularly in the non-breeding season. In both seasons, an STI rapidly increases corticosterone in brain and circulation. However, in the non-breeding season only, STI induces rapid increases in T and androstenedione in behaviorally relevant regions, while both androgens remain non-detectable in the circulation. Also in the non-breeding season only, STI increases progesterone in the blood and five regions. In contrast, during the breeding season, androstenedione and T are not acutely modulated by STI, and progesterone increases in two regions only. This is the first report of rapid modulation of local steroid levels in wild songbirds. These data suggest that neuroandrogens are rapidly synthesized “on demand” when circulating androgens are non-detectable, to support the expression of non-breeding aggression.
Seasonality of aggression
Wild male song sparrows displayed robust aggressive behavior during the STI in both breeding and non-breeding seasons. STI subjects showed similar behavioral latencies in both seasons, as before (Newman and Soma, 2011). Control subjects sang during the breeding season only, and spontaneous singing (i.e., without an STI) is more frequent during the breeding season (Searcy, 1984; O’Loghlen and Beecher, 1999). The number of flights and time in 5 m were higher in non-breeding season. Overall, both breeding and non-breeding males are highly aggressive during the STI (Wingfield and Soma, 2002).
Effects of STI on steroid levels in breeding males
In control breeding subjects, blood and brain steroid levels in the present study were similar to the levels in our previous study using the same LC-MS/MS method (Jalabert et al., 2021). However, the present levels are lower than those in an earlier study using radioimmunoassays (Heimovics et al., 2016), likely due to the higher specificity of LC-MS/MS.
In the breeding season, there was no effect of STI on progesterone levels in blood and in most regions. However, the STI increased progesterone in VTA and NCM. The relationship between progesterone and breeding aggression has been studied primarily in females, and a negative relationship is observed. In female California mice, circulating progesterone decreases after an aggressive interaction (Davis and Marler, 2003). Progesterone administration inhibits aggression in female hamsters (Kohlert and Meisel, 2001), rats (Albert et al., 1992), and coucals (songbirds) (Goymann et al., 2008). The few studies in males are less consistent; in birds aggression either increases (Adreani et al., 2018) or decreases (Charlier et al., 2009) plasma progesterone. Progesterone administration promotes aggression in male lizards (Weiss and Moore, 2004). Progesterone action in specific regions might promote male aggression or song perception in the breeding season, but more work is required (Lischinsky and Lin, 2020).
STI increased corticosterone in both blood and brain during the breeding season, as previously observed (Newman and Soma, 2011). Aggressive challenges also increases plasma corticosterone in other avian species (Charlier et al., 2009; Landys et al., 2010), potentially aiding energy mobilization for the behavioral response (Romero, 2002). Interestingly, the impact of glucocorticoids on aggression varies depending on the duration of their action. In male rodents, acute inhibition of glucocorticoid synthesis decreases aggression (Mikics et al., 2004; Fish et al., 2005), and glucocorticoid administration rapidly (within 10 min) increases aggression (Hayden-Hixson and Ferris, 1991; Mikics et al., 2004). In contrast, chronic corticosterone administration inhibits aggression in male lizards (Tokarz, 1987; DeNardo and Licht, 1993) and male song sparrows (Wingfield and Silverin, 1986).
During the breeding season, androstenedione, T, and 5α-DHT levels in blood and brain were not rapidly affected by STI. Similarly, a short (<10 min) aggressive interaction did not increase plasma T levels in wild song sparrows (Wingfield and Wada, 1989) or captive song sparrows (Heimovics et al., 2016). However, circulating T increases when the challenge is longer (30+ min; Wingfield, 1985; Wingfield and Wada, 1989). Prolonged social challenges stimulate gonadal T secretion, which promotes long-term aggression (Oliveira et al., 2009).
In breeding males, estrone and 17β-estradiol were not detectable in blood, and brain estrogens were not rapidly affected by STI. Estrogens stimulate breeding aggression in many male vertebrates (Schlinger and Callard, 1990; Hayden-Hixson and Ferris, 1991; Ogawa et al., 1997; Scordalakes and Rissman, 2003). However, in breeding male song sparrows, aromatase inhibition and 17β-estradiol administration do not affect aggression (Soma et al., 2000a; Heimovics et al., 2015), and a short aggressive interaction does not affect plasma or brain 17β-estradiol (Heimovics et al., 2016).
Effects of STI on steroid levels in non-breeding males
In non-breeding subjects, the data reveal temporal and spatial specificity of steroid signals in the brain, and an aggressive interaction induces rapid and region-specific changes in progesterone, corticosterone, androstenedione and testosterone. These results suggest more fine-tuning of neural steroid levels in the non-breeding season than in the breeding season.
Progesterone could modulate non-breeding aggression by acting directly via brain progesterone receptors (PRs) or acting indirectly via local conversion to other steroids. PRs are present within the SDMN of the avian brain (Lea et al., 2001). Here, brain progesterone was upregulated by STI. Similarly, non-breeding juvenile male song sparrows rapidly increase progesterone levels in blood and specific brain regions after an STI, which might facilitate aggression during development (Gray et al., 2024). In male mice, PR+ neurons in the VMH promote aggression, independent of the gonads (Yang et al., 2017). In male non-breeding Siberian hamsters, aggression is positively correlated with progesterone levels in the LS, AH, and medial amygdala (homolog of the avian TnA; Munley et al., 2021). In addition, progesterone could serve as substrate for the synthesis of androgens, and the avian brain expresses the necessary steroidogenic enzymes (Tsutsui, 2011).
During the non-breeding season, a 10 min STI increased corticosterone in both blood and brain. In contrast, a 30 min STI did not affect plasma corticosterone levels of non-breeding song sparrows (Newman and Soma, 2011). The duration of the social stimulation is important, as glucocorticoids can be beneficial for short periods but deleterious for prolonged periods (Grissom and Bhatnagar, 2009). This is particularly relevant during the non-breeding season, given the environmental challenges. In Siberian hamsters, adrenalectomy decreases aggression, whereas adrenal demedullation has no effect, suggesting that adrenocortical steroids promote non-breeding aggression (Demas et al., 2004).
Importantly, the present data strongly suggest that the non-breeding brain synthesizes T and androstenedione from circulating or neural steroid precursors or de novo from cholesterol. This local androgen synthesis is rapidly increased in response to an aggressive challenge. Despite the lack of androgens in the circulation, CON subjects had T and androstenedione in some regions, which could be due to an increase in baseline brain 3β-HSD activity during the non-breeding season (Pradhan et al., 2010). While circulating T and androstenedione remained non-detectable in the STI subjects, neural T and androstenedione rapidly increased in a region-specific manner, consistent with the rapid increase in brain 3β-HSD activity after an STI (Pradhan et al., 2010). The increase in brain androgens is evident using our highly specific LC-MS/MS assay, as illustrated in the representative chromatograms (Fig. 9).
Estrogens are key regulators of non-breeding aggression in this species. Aromatase and ER are highly expressed throughout the SDMN in non-breeding song sparrows (Soma et al., 2003; Wacker et al., 2010). Aromatase inhibition decreases aggression in non-breeding song sparrows (Soma et al., 2000a, b), and this effect is reversed by 17β-estradiol replacement (Soma et al., 2000b). Moreover, 17β-estradiol administration rapidly increases aggression during the non-breeding season only (Heimovics et al., 2015), similar to mice (Trainor et al., 2007, 2008). Therefore, we looked for a rapid increase of neuroestrogens following an STI, by including several estrogens (17β-estradiol, estrone, 17α-estradiol, and estriol) in our steroid panel. Surprisingly, none of these estrogens were detected in brain samples of non-breeding STI subjects. One or more of these estrogens might be present but below the detection limit of the assay used. While the assay used here is highly sensitive, when combined with DMIS derivatization, the assay sensitivity is even better (Jalabert et al., 2022). However, DMIS derivatization increases the background noise and reduces sensitivity for androgens, which were the primary focus of this investigation. Ongoing studies are assessing a panel of 10 estrogens and estrogen metabolites using DMIS derivatization.
Neuroestrogens may have high spatiotemporal specificity, contributing to the challenge of their detection in our study. Aromatase is present in presynaptic terminals (Saldanha et al., 2011). If estrogens are produced only at specific synapses, then measuring estrogens in tissue punches will be very difficult. Further, brain aromatase is modulated by calcium-dependent phosphorylation within minutes (Balthazart et al., 2003). Neurally synthesized estrogens exert local actions via rapid membrane-initiated signaling, suggesting they may act more like neurotransmitters than hormones (Balthazart and Ball, 2006; Saldanha et al., 2011). Therefore, neuroestrogens could be locally synthesized to promote aggression and then rapidly converted to estrogen metabolites that were not examined here.
Seasonality of neuroendocrine mechanisms
Although territorial aggression is robust and similar during the STI in both seasons, the contexts are quite different. During each season, animals compete for different types of resources. Moreover, different environmental conditions cause distinct physiological states, and therefore, different neuroendocrine mechanisms regulate behaviors that are outwardly similar. Glucocorticoids are rapidly modulated in the circulation and brain during breeding. In contrast, rapid modulation of glucocorticoids, progesterone, and androgens is observed in the non-breeding season, particularly in the brain. The present data suggest that local steroid synthesis plays a key role in supporting aggression, especially in the non-breeding season.
The switch of neuroendocrine mechanisms is crucial in changing environments. Sustained high levels of circulating steroid hormones, such as androgens, carry major costs in winter. In birds, chronic T treatment decreases body mass and fat reserves (Ketterson et al., 1991; Wikelski et al., 1999) and suppresses the immune system (Casto et al., 2001). Therefore, although aggression plays a crucial role to secure resources during winter, such as food and shelter, it is important to maintain low levels of circulating androgens in the non-breeding season, when ambient temperatures are low, photoperiods are short, and food is scarce. Systemic androgens increase and remain high after prolonged social challenges during the breeding season (Wingfield et al., 1990). Here, we provide evidence for neural androgen synthesis that is rapid and transient after an acute challenge during the non-breeding season. Therefore, the challenge hypothesis may operate at a local level during the non-breeding season (Pradhan et al., 2010).
Conclusions
Steroids are differentially regulated in the blood and multiple brain regions in breeding and non-breeding adult male song sparrows. An aggressive challenge rapidly increases brain androgens in the non-breeding season only. These data, together with previous results, suggest that neuroandrogens are rapidly synthesized in a region-specific manner in response to aggressive social cues when circulating androgens are low. Thus, neuroendocrine mechanisms that support year-round aggression are season dependent and switch from gonadal to neural synthesis to avoid the costs of elevated systemic androgen levels.
Footnotes
This work was supported by Canadian Institutes of Health Research Operating Grant (133606) and Project Grant (426405) and Canada Foundation for Innovation Grant (32631; to K.K.S.), an Uruguay Graduate Scholarship from the Agencia Nacional de Investigación e Innovación and University of British Columbia Zoology Graduate Fellowship (to C.J.), as well as Natural Sciences and Engineering Research Council of Canada-Undergraduate Student Research Award (to S.L.G.). We thank Lucia Zubizarreta, Maria A. Shock, and Taylor J. Bootsma for help with field work and lab work. C.J. and S.L.G. currently work at Departamento de Neurofisiología Celular y Molecular, Instituto de Investigaciones Biológicas Clemente Estable, Montevideo, Uruguay, and Department of Psychology, University of Washington, Washington, respectively.
The authors declare no competing financial interests.
- Correspondence should be addressed to Kiran K. Soma at ksoma{at}psych.ubc.ca.
