The decision of male medaka to mate or fight depends on two complementary androgen signaling pathways

Abstract

Adult male animals typically court and attempt to mate with females, while attacking other males. Emerging evidence from mice indicates that neurons expressing the estrogen receptor ESR1 in behaviorally relevant brain regions play a central role in mediating these mutually exclusive behavioral responses to conspecifics. However, the findings in mice are unlikely to apply to vertebrates in general because, in many species other than rodents and some birds, androgens-rather than estrogens-have been implicated in male behaviors. Here, we report that male medaka (Oryzias latipes) lacking one of the two androgen receptor subtypes (Ara) are less aggressive toward other males and instead actively court them, while those lacking the other subtype (Arb) are less motivated to mate with females and conversely attack them. These findings indicate that, in male medaka, the Ara- and Arb-mediated androgen signaling pathways facilitate appropriate behavioral responses, while simultaneously suppressing inappropriate responses, to males and females, respectively. Notably, males lacking either receptor retain the ability to discriminate the sex of conspecifics, suggesting a defect in the subsequent decision-making process to mate or fight. We further show that Ara and Arb are expressed in intermingled but largely distinct populations of neurons, and stimulate the expression of different behaviorally relevant genes including galanin and vasotocin, respectively. Collectively, our results demonstrate that male teleosts make adaptive decisions to mate or fight as a result of the activation of one of two complementary androgen signaling pathways, depending on the sex of the conspecific that they encounter.

Keywords: aggression; androgen receptor; decision-making; mating; teleost.

Fig. 1.

Androgen/Ara signaling facilitates male aggression toward other males. (A) Schematic of the five types of aggressive acts. (B) Setup for testing aggressive behavior among grouped males. (C and D) Total number of each aggressive act observed among ara^+/+ males and among ara^−/− males in the tank. Results from both Δ326 (n = 6 per genotype; C) and Δ325 (n = 5 per genotype; D) ara-deficient lines are shown. (E and F) Total number of each aggressive act observed among arb^+/+ males and among arb^−/− males in the tank. Results from both Δ10 (n = 8 per genotype; E) and Δ11 (n = 6 per genotype; F) arb-deficient lines are shown. Statistical differences were calculated by the unpaired t test, with Welch’s correction where appropriate (C–F). Box plots show median with Tukey whiskers (C–F). *P < 0.05, **P < 0.01.

Fig. 2.

Androgen/Arb signaling facilitates the mating of males with females. (A) Schematic of a sequence of male-typical mating acts. (B) Setup for testing the mating behavior of ara^+/+ and ara^−/− males. (C and D) Latency of ara^+/+ and ara^−/− males to initiate each mating act toward the stimulus female. Results from both Δ326 (n = 12 per genotype; C) and Δ325 (n = 12 per genotype; D) ara-deficient lines are shown. (E) Setup for testing the mating behavior of arb^+/+ and arb^−/− males. (F and G) Latency of arb^+/+ and arb^−/− males to initiate each mating act toward the stimulus female. Results from both Δ10 (n = 23 and 24 for arb^+/+ and arb^−/−, respectively; F) and Δ11 (n = 12 for each genotype; G) arb-deficient lines are shown. (H) Setup for the additional mating behavior test using an esr2b-deficient female as the stimulus. (I) Latency of the focal arb^+/+ and arb^−/− males (Δ10 line; n = 24 per genotype) to initiate each mating act. (J) Number of each mating act performed. Statistical differences were calculated by the Gehan–Breslow–Wilcoxon test (C, D, F, G, and I) and unpaired t test (J). Box plots show median with Tukey whiskers (J). *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 3.

ara-deficient males frequently attempt to mate with other males. (A) Setup for testing the mating behavior of ara^+/+ and ara^−/− males toward other males. (B) Raster plots of behavioral responses of males to the stimulus male. Data from four representative males of each genotype (ara^+/+ and ara^−/−; Δ326 line) are shown. (C and D) Number of each mating act performed by ara^+/+ and ara^−/− males toward the stimulus male. Results from both Δ326 (n = 20 and 17 for ara^+/+ and ara^−/−, respectively; C) and Δ325 (n = 12 and 11 for ara^+/+ and ara^−/−, respectively; D) ara-deficient lines are shown. (E and F) Latency of males to initiate each mating act. Results from both Δ326 (E) and Δ325 (F) lines are shown. (G) Setup for the three-chamber test to assess the sex discrimination ability of ara^+/+ and ara^−/− males. (H) Heat maps depicting the time spent by the focal males (Δ326 line; n = 14 and 15 for ara^+/+ and ara^−/−, respectively) in each location of the test chamber. (I) Difference in time spent by the focal males on the stimulus female versus the stimulus male side. (J) Setup for the additional three-chamber test to assess the influence of removing the stimulus male. (K) Difference in time spent by the focal ara^+/+ and ara^−/− males (Δ326 line) on the stimulus female versus the stimulus male (n = 5) or empty (n = 6) side. Statistical differences were calculated by the unpaired t test, with Welch’s correction where appropriate (C, D, and comparisons between genotypes in I and K), Gehan–Breslow–Wilcoxon test (E and F), and one-sample t test (comparisons against the null hypothesis of no difference in I and K). Box plots show median with Tukey whiskers (C, D, I, and K). *P < 0.05, **P < 0.01, ***P < 0.001 between genotypes. ^†P < 0.05, ^††P < 0.01, ^†††P < 0.001 against the null hypothesis.

Fig. 4.

arb-deficient males attack females. (A) Setup for testing the aggressive behavior of arb^+/+ and arb^−/− males toward females. (B) Raster plots of behavioral responses of males to the stimulus female. Data from four representative males of each genotype (arb^+/+ and arb^−/−; Δ10 line) are shown. (C and D) Percentage of arb^+/+ and arb^−/− males exhibiting any aggressive act toward the stimulus female. Results from both Δ10 (n = 23 and 24 for arb^+/+ and arb^−/−, respectively; C) and Δ11 (n = 12 for each genotype; D) arb-deficient lines are shown. (E and F) Number of each aggressive act performed. Results from both Δ10 (E) and Δ11 (F) lines are shown. (G) Setup for the three-chamber test to assess the sex discrimination ability of arb^+/+ and arb^−/− males. (H) Heat maps depicting the time spent by the focal males (Δ10 line; n = 13 per genotype) in each location of the test chamber. (I) Difference in time spent by the focal males on the stimulus female versus the stimulus male side. (J) Scatter plot of the number of aggressive acts and wrapping attempts performed by arb^−/− males (Δ10 line; n = 24) toward the stimulus female. Each dot represents one focal male; the unbroken line represents the regression line. (K) Latency from the first mating act to the initiation of aggressive acts. Statistical differences were calculated by Fisher’s exact test (C and D), unpaired t test, with Welch’s correction where appropriate (E, F, and comparisons between genotypes in I), one-sample t test (comparisons against the null hypothesis of no difference in I), and pairwise Pearson correlation (J). Box plots show median with Tukey whiskers (E, F, and I). *P < 0.05, **P < 0.01, ***P < 0.001 between genotypes. ^†P < 0.05, ^††P < 0.01 against the null hypothesis.

Fig. 5.

Ara and Arb are expressed in different neurons and stimulate different behaviorally relevant genes. (A) Percentage of neurons expressing ara only, arb only, or both in each brain nucleus of wild-type males. (B) Total area of fos expression signal in brain nuclei (where ara and/or arb are expressed) of wild-type solitary males (n = 5), males that mated with females (n = 7), and males that attacked other males (n = 7 except for Vv/Vs/Vp, PPp/PMm/PMg, and VM, where n = 5, 6, and 6, respectively). (C and D) Total area (C) and representative images (D) of vt expression signal in the pNVT of males and females of ara^+/+ and ara^−/− fish (Δ326 line; n = 5 per sex per genotype). (E and F) Total area (E) and representative images (F) of vt expression signal in the pNVT of males and females of arb^+/+ and arb^−/− fish (Δ10 line; n = 5 per sex per genotype). (G and H) Total area (G) and representative images (H) of gal expression signal in the pPMp of males and females of ara^+/+ and ara^−/− fish (Δ326 line; n = 5 per sex per genotype, except n = 4 for ara^+/+ females). (I and J) Total area (I) and representative images (J) of gal expression signal in the pPMp of males and females of arb^+/+ and arb^−/− fish (Δ10 line; n = 5 per sex per genotype). (K) Setup for testing aggressive behavior among ara^−/− males treated with vehicle alone or Vt peptide. (L) Total number of each aggressive act observed among ara^−/− males in the tank (Δ326 line; n = 6 per treatment). All scale bar are 50 μm. For abbreviations of brain nuclei, see SI Appendix, Table S1. Statistical differences were calculated by Dunnett’s or Dunn’s post hoc test (B), unpaired t test with Bonferroni–Dunn correction (C, E, G, and I), and unpaired t test, with Welch’s correction where appropriate (L). Box plots show median with Tukey whiskers (B, C, E, G, I, and L). *P < 0.05, **P < 0.01, ***P < 0.001.