The doublesex gene regulates dimorphic sexual and aggressive behaviors in Drosophila

Significance

Animals display dimorphic sexual and aggressive behaviors. In Drosophila, the doublesex (dsx) gene encodes male-specific Dsx^M and female-specific Dsx^F transcription factors and controls sexual development and behaviors. In mammals, dsx-related Dmrt genes direct the differentiation of gonads that produce sex hormones, controlling both the organization and activation of neural circuits underlying sexually dimorphic behaviors. In this study, we find that sex-specific Dsx isoforms promote distinct sexual behaviors in both sexes, and oppositely regulate aggression to establish male-biased aggressiveness. We further show that Dsx functions both developmentally and acutely in sex-specific interneurons that control these behaviors. Our results reveal that a conserved sex-determination gene plays both organizational and activational roles in sexually dimorphic behaviors.

Abstract

Most animal species display dimorphic sexual behaviors and male-biased aggressiveness. Current models have focused on the male-specific product from the fruitless (fru^M) gene, which controls male courtship and male-specific aggression patterns in fruit flies, and describe a male-specific mechanism underlying sexually dimorphic behaviors. Here we show that the doublesex (dsx) gene, which expresses male-specific Dsx^M and female-specific Dsx^F transcription factors, functions in the nervous system to control both male and female sexual and aggressive behaviors. We find that Dsx is not only required in central brain neurons for male and female sexual behaviors, but also functions in approximately eight pairs of male-specific neurons to promote male aggressiveness and approximately two pairs of female-specific neurons to inhibit female aggressiveness. Dsx^F knockdown females fight more frequently, even with males. Our findings reveal crucial roles of dsx, which is broadly conserved from worms to humans, in a small number of neurons in both sexes to establish dimorphic sexual and aggressive behaviors.

A central goal in neuroscience is to understand how innate behaviors are generated at the molecular and neuronal levels, and in particular how sexually dimorphic behaviors are established. A substantial body of work has revealed sex-specific or sex-biased expression of genes that are responsible for sexually dimorphic behaviors from worms to humans (1, 2).

The fruit fly, Drosophila melanogaster, is an excellent model system with valuable genetic tools that can be used to study dimorphic behaviors (3). Notably, Drosophila shows robust reproductive and aggressive behaviors with both qualitative and quantitative aspects of sexual dimorphism, as well as well-characterized dimorphic circuits (4, 5). For example, male flies display male-specific courtship behaviors, while virgin females evaluate the quality of male courtship and decide to be receptive or not (6). Aggressive dimorphic behaviors can also be analyzed. Qualitative differences in fighting patterns are observed between males and females, and fighting is also quantitatively different in that males fight more frequently than females (7–9).

Studies over several decades have led to the identification of two terminal genes in the sex determination hierarchy: fruitless (fru), encoding male-specific Fru (Fru^M) proteins, and doublesex (dsx), encoding sex-specific Dsx^M and Dsx^F proteins (10–12). Together, the fru and dsx gene products control most aspects of sexual dimorphism (5, 13). Fru^M is largely necessary and sufficient for building the potential for innate courtship behavior into the nervous system (14–16). In the absence of Fru^M, males are unable to court without prior social experience, but can acquire some courtship behaviors in an experience- and Dsx^M-dependent manner (17, 18). Indeed, Fru^M and Dsx^M coordinate to generate a subset of male-specific P1 neurons that integrate multiple sensory cues and initiate courtship (19–23). Fru^M is responsible for the dimorphic projections of aSP-f neurons in males and aSP-g neurons in females that underlie sex-specific pheromone responses and sexual behaviors (24). Fru^M is also responsible for the sex-specific fighting patterns and dominance (25), and specific subsets of Fru^M neurons are involved in generating sexually dimorphic aggression patterns (26). Fru^M is also responsible for the development of male-specific aggression-promoting neurons that express the neuropeptide tachykinin (Tk) (9, 27). Moreover, Fru^M is expressed in a subset of male-specific P1 neurons and aSP2 neurons that promote intermale aggression (28, 29). These findings reveal the crucial roles of Fru^M in establishing male-specific sexual behaviors and aggression patterns, and perhaps male-biased aggression intensity.

In contrast to the well-established role of Fru^M in controlling sexually dimorphic behaviors, studies of Dsx^M and Dsx^F have thus far mainly focused on sexual differentiation, as dsx and its related Dmrt genes control most aspects of morphological dimorphism in flies and other animals (30). Consequently, how dsx functions in the nervous system to mediate sexually dimorphic behaviors has rarely been studied (31). Previous studies of dsx function in behavior either directly used dsx mutants or dominate alleles that affect dsx function in all cells and are often intersexual (18, 32), or indirectly from manipulation of fru or transformer (tra), which acts upstream of both fru and dsx (18, 22, 26, 33). A recent study used RNA interference (RNAi) to knock down dsx expression in adulthood to investigate its role in behavior, but the manipulation was still carried out in all dsx-expressing cells (34). Despite the relatively understudied role of dsx in the nervous system, neurons expressing dsx are found to control most dimorphic behaviors, including male courtship and aggression (18, 22, 29, 33, 35–37), female precopulatory and postmating behaviors (38–42), as well as female aggression (37, 43–45). Importantly, a subset of fru^M and dsx^M coexpressing P1 neurons specifically promote male courtship but not aggression, whereas another subset of fru^M-negative but dsx^M-positive pC1 neurons specifically promote male aggression but not courtship (37), implicating a potential role of Dsx^M in regulating male courtship and aggression. Thus, it is of significant interest to investigate the role of Dsx^M in males and Dsx^F in females in regulating dimorphic sexual and aggressive behaviors.

In this study, we knocked down dsx function in the male and female nervous systems, from pan-neuronal to specific knockdown in as few as two pairs of neurons, and revealed crucial roles of Dsx^M in promoting male courtship and Dsx^F in promoting female receptivity in central brain neurons. Surprisingly, we found that Dsx^M and Dsx^F function oppositely in a small number of sex-specific aggression-promoting neurons to promote male aggression and inhibit female aggression, respectively. These results reveal the function of sex-specific Dsx^M and Dsx^F proteins in promoting reproductive behaviors in both sexes, while biasing aggression in males and females to establish sexually dimorphic aggression levels.

Results

Central Brain Dsx^M Regulates Male Sexual and Aggressive Behaviors.

A previous study found that dsx mutant males show a low level of courtship toward females (32), suggesting that dsx may be crucial for male courtship. To validate this finding, we tested courtship in dsx-null mutant males using three deficiency lines (18). In addition, we generated a dsx^DBD allele, in which the first coding exon of dsx was replaced by the GAL4-DBD sequence (SI Appendix, Fig. S1A). The dsx^DBD allele is a null mutant, as evidenced by the loss of Dsx immunostaining as well as intersexual phenotypes in dsx^DBD/dsx^1649-9625 males and females (SI Appendix, Fig. S1 B–E). We found that dsx-null mutant males (dsx^683-7058/dsx^1649-9625, dsx^683-7058/dsx^M+R15, and dsx^DBD/dsx^1649-9625) indeed showed low levels of courtship toward wild-type virgin females (courtship indices [CIs] < 7%), whereas heterozygous control males courted vigorously (CIs > 70%) (Fig. 1A). Although these results confirm that courtship is affected, one important caveat of these experiments is that dsx-null mutant males are intersex and have developmental defects in sexual organs. Thus, a specific role for dsx in the central nervous system (CNS) cannot be determined yet.

Fig. 1.

Central brain Dsx^M regulates male sexual and aggressive behaviors. (A and B) Courtship indices for dsx mutant (A) and RNAi-mediated dsx^M knockdown males (B). Genotypes are indicated below. n = 24 for each. ***P < 0.001 and ns, not significant. Kruskal–Wallis test and post hoc Dunn’s multiple comparisons. (C and D) Expression pattern of dsx^GAL4 (C) and dsx^brain (Otd-Flp;tub > GAL80 >, dsx^GAL4) in male CNS (D). (Scale bars, 50 μm.) Representative of three samples each. (E–H) Anti-Dsx^M and anti-GFP signals in pupal (E and F) and adult (G and H) stages in control (UAS-myrGFP/+;dsx^brain/+) and dsx^M knockdown (UAS-myrGFP/+;dsx^brain/UAS-dsx^MRNAi) male brains. Yellow arrowheads indicate anti-Dsx^M signals in pC1 and pC2 neurons in control males (E and G), and white arrowheads indicate diminished anti-Dsx^M signals in the same brain region in dsx^M knockdown males (F and H). (Scale bars, 50 μm.) Representative of five samples each. (I) Courtship indices for control and dsx^M knockdown males. n = 23 for each. ***P < 0.001, Mann–Whitney U test. (J) The percentage of successful copulation within 30 min. n = 23 each. ***P < 0.001, χ² test. (K) An example image of intermale aggression. (L) Number of lunges per minute for control and dsx^M knockdown males. n = 24 for each. ***P < 0.001, Mann–Whitney U test. (M) Fighting latency of control and dsx^M knockdown males. n = 24 for each. ***P < 0.001, Mann–Whitney U test. Error bars indicate SEM.

To address if neuronal dsx contributes to dimorphic behaviors, we set out to knock down dsx function specifically in the nervous system (SI Appendix, Fig. S2). We used two RNAi lines as described previously, one targeting the male-specific exon (UAS-dsx^MRNAi) and the other the female-specific exon (UAS-dsx^FRNAi) of the dsx gene (34, 46). Their efficiency was validated using real-time qPCR (SI Appendix, Fig. S2 B and C) and by confirming the presence of intersexual phenotypes. To knock down dsx expression specifically in the nervous system, we then combined these validated RNAi tools with the pan-neuronal driver R57C10-GAL4, and found that Dsx^M expression in the brain was almost undetectable in RNAi-mediated knockdown males, except for several false-positive cells that are likely due to the imperfect specificity of this antibody (SI Appendix, Fig. S2 D and E). We also found that knocking down dsx^M in all dsx-expressing cells induced intersexual phenotypes like dsx mutant males, whereas knocking down dsx^M pan-neuronally did not affect external sexual traits (SI Appendix, Fig. S2H). The knockdown of neuronal Dsx^M expression strongly impaired male courtship, almost to the level of dsx-null mutant males (CI ∼13%, control CIs > 80%) (Fig. 1B). These results indicate that neuronal Dsx^M is crucial for male courtship.

We next sought to identify the crucial subset of neurons in which Dsx^M is required for male courtship. First, we targeted dsx neurons specifically in the brain, hereafter referred to as dsx^brain neurons, by using the brain-specific flippase recombinase (Otd-Flp) (9, 17) (Fig. 1C for dsx^GAL4 and Fig. 1D for dsx^brain [Otd-Flp;tub > GAL80 >, dsx^GAL4]). We further examined the efficiency of dsx RNAi-mediated knockdown in control (UAS-myrGFP/+;dsx^brain/+) and experimental (UAS-myrGFP/+;dsx^brain/UAS-dsx^MRNAi) males from different developmental stages (third-instar larval [L3], 48 h after pupal formation [48 h APF], 4- to 7-d-old adult) using both anti-GFP and anti-Dsx^M antibodies (Fig. 1 E–H and SI Appendix, Fig. S3 A and B). Whereas anti-Dsx^M signals mostly overlapped with the dsx^brain-driven GFP signals in all three developmental stages in control male brains (for pupal and adult expression, see Fig. 1 E and G; for larval expression, see SI Appendix, Fig. S3A), they were almost undetectable in UAS-myrGFP/+;dsx^brain/UAS-dsx^MRNAi male brains regardless of the developmental stage (for pupal and adult expression, see Fig. 1 F and H; for larval expression, see SI Appendix, Fig. S3B). Notably, the knockdown was not perfect in some samples, as we observed anti-Dsx^M signals in a small number of cells with weak or no GFP signal, likely due to variation in expression from the dsx^brain driver in individual flies (SI Appendix, Fig. S3C). Nevertheless, these results demonstrated effective knockdown of dsx^M in most, if not all, dsx^brain neurons throughout development.

We then tested the effect of knocking down dsx^M in dsx^brain neurons on male courtship and aggressive behaviors. We found that compared to the control, dsx^M knockdown males showed reduced courtship toward virgin females (Fig. 1I) (CI ∼23%, compared to ∼80% for control males) and less mating success (Fig. 1J) (∼20% mated within 30 min, compared to ∼90% for control males). To examine if Dsx^M function is also involved in male aggression, we tested intermale aggression in single-housed males (Fig. 1K). Knocking down dsx^M in dsx^brain neurons significantly reduced intermale lunging frequency (Fig. 1L) and increased the latency to initiate lunging behavior (Fig. 1M). Taken together, these results indicate that Dsx^M functions in the central brain to promote female-directed courtship and intermale aggression.

Dsx^M Expression in the Male-Specific P1^a Neurons Positively Correlates with Male Aggressiveness.

Central brain dsx^M-expressing neurons include as many as ∼300 pC1, pC2, and pCd neurons (47–49). To further investigate the function of Dsx^M in regulating male courtship and aggression in a smaller number of neurons, we used a splitGAL4 driver to target a subset of male-specific P1 neurons (P1^a neurons) (Fig. 2A) previously shown to be crucial for courtship initiation and aggression (21, 22, 29, 31, 50, 51). These P1^a neurons are a subset of dsx^M-expressing pC1 neurons (33). RNAi-mediated knockdown of dsx^M specifically in P1^a neurons significantly reduced male courtship toward virgin females (CI ∼36%) compared to two control lines (CIs > 60%) (Fig. 2B), and also reduced male mating success (∼60% mated in 30 min, compared to ∼90% for control males) (Fig. 2C), although the reduction was less than observed for males with dsx^M knockdown in all dsx^brain neurons (Fig. 1 I and J).

Fig. 2.

Dsx^M expression in P1^a neurons positively regulates male aggressiveness. (A) Expression pattern of the P1^a-GAL4 (R15A01-AD; R71G01-DBD) in male CNS. (Scale bars, 50 μm.) Representative of five samples. (B) Courtship indices for control and P1^a-dsx^M knockdown males. n = 24, 23, 23 from left to right. *P < 0.05, ***P < 0.001, Kruskal–Wallis test and post hoc Dunn’s multiple comparisons. (C) The percentage of successful copulation within 30 min. n = 24, 23, 23, respectively. *P < 0.05, χ² test. (D–G) Anti-Dsx^M and anti-GFP signals in pupal (D and E) and adult (F and G) stages in control (UAS-myrGFP/+;P1^a-GAL4/+) and P1^a-dsx^M knockdown (UAS-myrGFP/+;P1^a-GAL4/UAS-dsx^MRNAi) male brains. Yellow arrowheads indicate costaining of anti-Dsx^M and anti-GFP signals in control males (D and F), and white arrowheads indicate diminished anti-Dsx^M signals specifically in the GFP⁺ neurons in P1^a-dsx^M knockdown males (E and G). (Scale bars, 20 μm.) Representative of five samples each. (H and I) Lunging frequency (H) and fighting latency (I) for control and P1^a-dsx^M knockdown males, as well as males with dsx^M knockdown and ectopic dsx^F expression in P1^a neurons. Genotypes are indicated. n = 23, 24, 24 from left to right. **P < 0.01, ***P < 0.001; ns, not significant, Mann–Whitney U test. (J and K) Costaining using anti-Dsx^M and anti-Dsx^F antibodies in control P1^a-GAL4/+ (J) and experimental UAS-flag-dsx^F/+;P1^a-GAL4/UAS-dsx^MRNAi (K) male adult brains. White arrowheads indicate ectopic expression of Dsx^F and knockdown of Dsx^M. (Scale bars, 20 μm.) Representative of five samples each. (L and M) Lunging frequency (L) is enhanced and fighting latency (M) is reduced by overexpression of Dsx^M in P1^a neurons. n = 24 for each. ***P < 0.001, Mann–Whitney U test. (N) High-intensity aggression behavior (tussling) is induced with overexpression of Dsx^M in P1^a neurons. n = 48 for each. ***P < 0.001, χ² test. Error bars indicate SEM.

To validate the efficiency of dsx^M knockdown specifically in the P1^a neurons, we used anti-GFP to label P1^a neurons and anti-Dsx^M to simultaneously examine knockdown efficiency and specificity (Fig. 2 D–G and SI Appendix, Fig. S3 D–I). We found no overlap of anti-GFP and anti-Dsx^M signals in the L3 stage in control UAS-myrGFP/+;P1^a-GAL4/+ males, suggesting that P1^a-GAL4 drives expression in dsx^M-negative cells in the L3 stage (SI Appendix, Fig. S3D). Alternatively, these cells could be dsx^M-positive P1^a, but the Dsx^M expression was too weak to be detected by anti-Dsx^M antibody in this stage. In the pupal and adult stages, we observed well-overlapped anti-GFP and anti-Dsx^M signals in control UAS-myrGFP/+;P1^a-GAL4/+ males, indicating consistent GAL4 expression in P1^a neurons from pupal to adult stages (Fig. 2 D and F, yellow arrowheads, and SI Appendix, Fig. S3 F and H). Furthermore, anti-Dsx^M signals were almost diminished in the GFP⁺ P1^a neurons but present in other pC1 neurons in both the pupal and adult brains of UAS-myrGFP/+;P1^a-GAL4/UAS-dsx^MRNAi (Fig. 2 E and G, white arrowheads, and SI Appendix, Fig. S3 G and I), indicating effective and specific knockdown of dsx^M expression in P1^a neurons. Together, these results demonstrate the effectiveness and specificity of dsx^M knockdown in P1^a neurons, and the crucial role of Dsx^M in these male-specific neurons for courtship intensity and mating success.

In addition to courtship behavior, male-specific P1^a neurons also regulate aggressive behaviors (27, 29, 33). To test if Dsx^M in these P1^a neurons is also involved in male aggression, we observed intermale aggression in single-housed males from both control and RNAi-mediated knockdown lines. Knocking down dsx^M specifically in P1^a neurons significantly reduced intermale lunging frequency (Fig. 2H). We then tested if expression of the female isoform Dsx^F would rescue or aggravate the aggression phenotype in dsx^M knockdown males. We found similarly low levels of lunging frequency for dsx^M knockdown males with or without ectopic expression of dsx^F in P1^a neurons compared to control males (Fig. 2H). However, when compared to controls, males with dsx^M knockdown alone or together with Dsx^F expression did not show any significant difference in fighting latency (Fig. 2I). These results demonstrate that dsx^M knockdown in P1^a neurons reduces male aggression, but not as strongly as observed for dsx^M knockdown in all dsx^brain neurons, suggesting the additional involvement of non-P1^a dsx^brain neurons in regulating male aggression. Importantly, the reduction in male courtship and aggression in males with dsx^M knockdown in either dsx^brain or P1^a neurons was not due to locomotor defects as their walking speed was similar to control males (SI Appendix, Fig. S4 A and B).

To confirm the effectiveness of both dsx^M knockdown and Dsx^F expression, we performed costaining of Dsx^M and Dsx^F in UAS-flag-dsx^F/+;P1^a-GAL4/UAS-dsx^MRNAi male brains. We observed three to five pairs of anti-Dsx^F neurons within the pC1 cluster (ectopic expression of Dsx^F), in which there was no anti-Dsx^M signal (dsx^M knockdown) (Fig. 2 J and K). Note that the number of P1^a neurons was significantly smaller in males with ectopic Dsx^F expression, consistent with the involvement of Dsx^F in neural development (22, 48, 52) (for comparisons of neuronal numbers, see for example, Fig. 5).

Fig. 5.

Dsx isoforms promote development of aggression-promoting neurons in two sexes but oppositely modulate their function. (A) Expression pattern of P1^a-GAL4 (R15A01-AD; R71G01-DBD) in control (Left) and dsx^M knockdown males (Right). (B and C) Expression pattern of P1^a1-GAL4 (B, R15A01-AD;dsx^DBD) and P1^a2-GAL4 (C, R71G01-AD;dsx^DBD) in control (Left) and dsx mutant (dsx^DBD/dsx^1649-9625) males (Right). (Scale bars, 50 μm.) (D) Knocking down dsx^M and overexpressing dsx^F in P1^a neurons, but not knocking down dsx^M alone, affects cell numbers. n = 6, 5, 5 from left to right. **P < 0.01; ns, not significant, Mann–Whitney U test. (E) Systemic dsx^M knockout reduces the number of P1 neurons. n = 6, 7, 7, 6 from left to right. **P < 0.01, ***P < 0.001, Mann–Whitney U test. (F) A summary of Dsx^M function on the development of P1^a neurons and regulation of their aggression-promoting function. (G) Expression pattern of pC1^d-GAL4 (R26E01-AD;dsx^DBD) in control (Left) and dsx^F knockdown females (Right). (H) Expression pattern of pC1^d-GAL4 in control (Left) and dsx^F mutant (dsx^DBD/dsx^1649-9625) females (Right). (Scale bars, 50 μm.) (I) Systemic dsx^F knockout, but not restricted dsx^F knockdown, impairs development of pC1^d neurons. n = 7, 5, 5, 5 from left to right. ***P < 0.001, Mann–Whitney U test. (J) A summary of Dsx^F function on the development of pC1^d neurons and regulation of their aggression-promoting function. (K) Dsx^M and Dsx^F promote the development of male-specific P1^a in males and female-specific pC1^d neurons in females respectively, while oppositely regulate their aggression-promoting function. Error bars in D, E, and I indicate SEM.

We then tested if acute knockdown of dsx^M in P1^a neurons during adulthood would reduce male aggression. We utilized the temperature-sensitive GAL4 repressor, GAL80^ts, and raised tub-GAL80^ts/+;P1^a-GAL4/UAS-dsx^MRNAi males at 18 °C throughout development. Adult males were heat-shocked at 30 °C for 4 d before aggression tests (SI Appendix, Fig. S5A). We found that knocking down dsx^M expression specifically during adulthood did not affect male aggression (SI Appendix, Fig. S5 B and C), suggesting the crucial role of dsx^M during development of P1^a neurons.

To investigate whether the expression level of Dsx^M in P1^a neurons would positively correlate with male aggression, we overexpressed dsx^M specifically in P1^a neurons and tested aggression in group-housed males. We found that these males lunged much more frequently (Fig. 2L) and quickly (Fig. 2M) than control males. In fact, 17 of 48 males with Dsx^M overexpression displayed tussling (SI Appendix and Movie S1), a high-intensity fighting pattern rarely seen in control males (2 of 48) (Fig. 2N).

We further tested if acute overexpression of dsx^M in P1^a neurons during adulthood would enhance male aggression. We utilized the above strategy using GAL80^ts to restrict dsx^M overexpression during adulthood (SI Appendix, Fig. S5A). We observed strong anti-Myc signals in UAS-myc-dsx^M/+;P1^a-GAL4/tub-GAL80^ts but not in control males after heat shock (SI Appendix, Fig. S5D). Only weak anti-Myc signals were observed without heat shock (SI Appendix, Fig. S5 D–F), suggesting the successful but imperfect repression of GAL4 activity by GAL80^ts at 18 °C, consistent with previous findings (17). We found that heat-shocked males with Dsx^M overexpression initiated fighting much quicker than all three groups of control males, although the lunging frequency was not consistently increased compared to each control line (SI Appendix, Fig. S5 G and H). Furthermore, a fraction of heat-shocked males with Dsx^M overexpression displayed the high-intensity tussling behavior, while none of the control males did (SI Appendix, Fig. S5I). These results indicate that acute overexpression of Dsx^M in P1^a neurons during adulthood significantly enhances male aggression, although not as strongly as constitutive overexpression of Dsx^M. Together, the above results demonstrate that Dsx^M plays both developmental and acute roles in the male-specific P1^a neurons for male aggressiveness.

Central Brain Dsx^F Regulates Female Sexual and Aggressive Behaviors.

Since we found that Dsx^M plays a crucial role in male courtship and aggression, we next tested the role of Dsx^F in female behaviors. Knocking down neuronal Dsx^F expression using the pan-neuronal driver R57C10-GAL4 nearly eliminated female receptivity to courting males (∼2% mated in 30 min, controls > 75% in 30 min) (Fig. 3A). Dsx^F expression in the brain was reduced to an undetectable level in these RNAi-mediated virgin females (SI Appendix, Fig. S2 F and G). Furthermore, knocking down dsx^F in all dsx-expressing cells induced intersexual phenotypes, whereas knocking down dsx^F pan-neuronally did not affect external sexual traits (SI Appendix, Fig. S2I).

Fig. 3.

Central brain Dsx^F regulates sexual and aggressive behaviors in females. (A) Female receptivity by control and Dsx^F knockdown females, n = 48 for each. ***P < 0.001, χ² test. (B and C) Expression pattern of dsx^GAL4 (B) and dsx^brain-GAL4 (C) in female CNS. (Scale bars, 50 μm.) Representative of three samples each. (D–G) Anti-Dsx^F and anti-GFP signals in pupal (D and E) and adult (F and G) stages in control (UAS-myrGFP/+;dsx^brain/+) and dsx^F knockdown (UAS-myrGFP/+;dsx^brain/UAS-dsx^FRNAi) female brains. Yellow arrowheads indicate anti-Dsx^F signals in pC1 and pC2 neurons in control females (D and F), and white arrowheads indicate the diminished anti-Dsx^F signals in the same brain region in dsx^F knockdown females (E and G). White circles indicate GFP⁺ pC1, pC2, and pCd cell bodies. (Scale bars, 50 μm.) Representative of five samples each. (H) Knocking down dsx^F in dsx^brain neurons significantly increases numbers of pC1 and pC2, but not pCd neurons in females. n = 7 and 5 for control and dsx^F knockdown females, respectively. **P < 0.01; ns, not significant, Mann–Whitney U test. (I) Female receptivity for control UAS-dsx^FRNAi/+ (n = 96) and dsx^brain/UAS-dsx^FRNAi (n = 80) females. ***P < 0.001, χ² test. (J–L) Activation of dsx^brain neurons promotes receptivity in control females but does not restore receptivity in dsx^F knockdown females. For dsx^brain/+ females, n = 144 (22 °C) and 137 (29 °C); for dTrpA1/+; dsx^brain/+ females, n = 117 (22 °C), and 144 (29 °C); for UAS-dTrpA1/+;dsx^brain/UAS-dsx^FRNAi females, n = 80 (22 °C) and 96 (29 °C). **P < 0.01, ***P < 0.001; ns, not significant, χ² test. (M) An example image of headbutt aggression between two females. (N and O) Number of headbutts per minute (N) and fighting latency (O) for control and dsx^F knockdown females. n = 16 pairs for each. ***P < 0.001, Mann–Whitney U test. (P) A summary of how brain-specific dsx^F knockdown affects development and consequently sexual and aggressive behaviors in females. Error bars indicate SEM.

We then narrowed down the subsets of neurons in which Dsx^F is required for female receptivity by targeting dsx neurons specifically in the female brain (dsx^brain) using the same genotype described above (Fig. 3 B and C). The efficiency of RNAi in control (UAS-myrGFP/+;dsx^brain/+) and experimental (UAS-myrGFP/+;dsx^brain/UAS-dsx^FRNAi) females in the L3, 48 h APF, and adult stages was validated using anti-GFP and anti-Dsx^F antibodies (Fig. 3 D–G and SI Appendix, Fig. S6 A and B). We found that anti-Dsx^F signals largely overlapped with the anti-GFP signals in all the three developmental stages in control female brains (for pupal and adult expression, see Fig. 3 D and F; for larval expression, see SI Appendix, Fig. S6A); however, anti-Dsx^F signals were diminished in UAS-myrGFP/+;dsx^brain/UAS-dsx^FRNAi female brains regardless of the developmental stage (for pupal and adult expression, see Fig. 3 E and G; for larval expression see SI Appendix, Fig. S6B). Interestingly, we observed an increased number of GFP⁺ neurons in the dsx^F knockdown females, especially pC1 (9.6 neurons per hemisphere in control and 12.8 in dsx^F knockdown females) and pC2 (13.4 neurons per hemisphere in control and 32.6 in dsx^F knockdown females) neurons (Fig. 3H), consistent with the finding that Dsx^F generally promotes programmed cell death. Furthermore, females with Dsx^F knockdown in dsx^brain neurons were rarely receptive to courting males (∼14% mated in 30 min, ∼95% for controls in 30 min) (Fig. 3I). These results indicate that Dsx^F expression in dsx^brain neurons is crucial for female receptivity to courting males, probably through its role in the development of pC1 and pC2 neurons.

Since central brain dsx-expressing neurons promote female receptivity (38), we set out to test if artificial activation of dsx^brain neurons could restore female receptivity in Dsx^F knockdown females. Activation of dsx^brain neurons expressing the temperature-dependent cation channel dTrpA1 (53) at 29 °C significantly increased female receptivity (>80% mated in 6 min at 29 °C, and ∼40% mated in 6 min for control females) (Fig. 3 J and K); however, activation of dsx^brain neurons together with Dsx^F knockdown only partially increased female receptivity (∼20% at 29 °C and ∼2% at 22 °C in 30 min) (Fig. 3L) and was still far less than that in control females. These results suggest that the low level of female receptivity in Dsx^F knockdown females is not only an issue of neuronal activity but may also be due to organizational changes in the receptivity circuit (e.g., development of more pC1 and pC2 neurons, as shown in Fig. 3H).

We next tested aggressive behaviors, as dsx-expressing neurons are also involved in regulating female aggression (45). Surprisingly, we found that knocking down Dsx^F expression in dsx^brain neurons significantly enhanced the frequency of female aggression (Fig. 3 M and N) and also reduced fighting latency (Fig. 3O). These results indicate that Dsx^F in dsx^brain neurons promotes female receptivity but suppresses female aggression (Fig. 3P).

Dsx^F in Two Pairs of pC1^d Neurons Inhibits Female Aggression.

Two pairs of female-specific dsx-expressing pC1 neurons were previously shown to promote female aggression (43–45). To test if Dsx^F modulates female aggression in these pC1 neurons, we utilized the R26E01-GAL4 driver that labels the aggression-promoting pC1 neurons as well as many other neurons in females (45) (Fig. 4A). We then checked for overlapping expression with dsx^LexA using an intersectional strategy (21). Two pairs of female-specific pC1 neurons, hereafter referred to as pC1^d neurons according to previous studies (43, 44, 54), as well as some neurons in the abdominal ganglion were labeled (Fig. 4B). Females with Dsx^F knockdown in R26E01-GAL4 neurons showed an increase in aggressive behaviors, including high levels of headbutt frequency (Fig. 4C, SI Appendix, and Movie S2), and quicker initiation of the behavior (Fig. 4D).

Fig. 4.

Dsx^F in two pairs of pC1^d neurons suppresses female aggression. (A) Expression pattern of R26E01-GAL4 in the female brain. (Scale bars, 50 μm.) Representative of three samples. (B) Intersectional expression between R26E01-GAL4 and dsx^LexA (LexAop2-Flp/UAS > stop > myrGFP;dsx^LexA/R26E01-GAL4) in the female and male CNS. (Scale bars, 50 μm.) Representative of five samples. (C and D) Knocking down Dsx^F expression in R26E01-GAL4 neurons enhances female headbutt frequency (C) and reduces fighting latency (D). n = 16, 22, 16 pairs from left to right. ***P < 0.001; ns, not significant, Mann–Whitney U test. (E) Knocking down Dsx^F expression in R26E01-GAL4 neurons does not affect female receptivity. n = 96 for each. ns, not significant. χ² test. (F) An example image of a female fighting against a male. (G) Knocking down Dsx^F expression in R26E01-GAL4 neurons induces female fighting toward males with pan-neuronal knockdown of Dsx^M (R57C10-GAL4/UAS-dsx^MRNAi). n = 48 and 45, respectively. ***P < 0.001, unpaired Student’s t test. (H) Expression pattern of pC1^d-GAL4 (R26E01-AD;dsx^DBD) in female and male CNS. (Scale bars, 50 μm.) Representative of five samples. (I–L) Anti-Dsx^F and anti-GFP signals in pupal (I and J) and adult (K and L) stages in control (UAS-myrGFP/+;pC1^d-GAL4/+) and pC1^d-dsx^F knockdown (UAS-myrGFP/+;pC1^d-GAL4/UAS-dsx^FRNAi) female brains. Yellow arrowheads indicate costaining of anti-Dsx^F and anti-GFP signals in control females (I and K), and white arrowheads indicate diminished anti-Dsx^F signals specifically in the GFP⁺ neurons in pC1^d-dsx^F knockdown females (J and L). (Scale bars, 20 μm.) Representative of five samples each. (M and N) Headbutt frequency (M) and fighting latency (N) by control and pC1^d-dsx^F knockdown females, as well as females with dsx^F knockdown and ectopic dsx^M expression in pC1^d neurons. Genotypes are indicated. n = 16, 16, 28, 24 from left to right. ***P < 0.001; ns, not significant, Mann–Whitney U test. (O and P) Costaining of anti-Dsx^M and anti-Dsx^F signals in control pC1^d-GAL4/+ (O) and experimental UAS-myc-dsx^M/+;pC1^d-GAL4/UAS-dsx^FRNAi (P) female adult brains. White arrowheads indicate ectopic expression of Dsx^M and knockdown of Dsx^F. (Scale bars, 20 μm.) Representative of five samples each. Error bars indicate SEM.

As these females displayed strong aggression toward each other, we wondered if they would also fight with males. We found that Dsx^F knockdown females did not fight with courting wild-type males, and were equally receptive to them compared to control females (Fig. 4E). When paired with the Dsx^M knockdown males that we found rarely courted (Fig. 1B), Dsx^F knockdown females fought with the males more frequently than the control females did (1.1 headbutts vs. 0.1 headbutt per min) (Fig. 4 F and G, SI Appendix, and Movie S3). These results demonstrate that Dsx^F expression in two pairs of female-specific pC1^d neurons inhibits female aggression toward both females and males, but does not affect receptivity to effectively courting males. Alteration of receptivity and aggressive behaviors in females with Dsx^F knockdown in either dsx^brain or pC1^d neurons is not due to potential locomotor defect as these females displayed similar walking speed compared with control females (SI Appendix, Fig. S4 C and D).

To further confirm these findings, we utilized the splitGAL4 system (55) and obtained R26E01-AD (56) with above-mentioned dsx^DBD, and then successfully labeled the two pairs of female-specific pC1^d neurons in the R26E01-AD;dsx^DBD females (hereafter referred to as pC1^d-GAL4) (Fig. 4H). We also validated the efficiency of dsx^F RNAi in these pC1^d neurons in the L3, 48 h APF, and adult stages as we did above (Fig. 4 I–L and SI Appendix, Fig. S6 C–H). We found that the pC1^d-GAL4 showed no expression in the L3 stage in control UAS-myrGFP/+;pC1^d-GAL4/+ females (SI Appendix, Fig. S6 C and D), but labeled a pair of pC1^d neurons in the pupal stage (Fig. 4I and SI Appendix, Fig. S6E) and two pairs of pC1^d neurons in the adult stage (Fig. 4K and SI Appendix, Fig. S6G) that overlapped with anti-Dsx^F (Fig. 4 I and K, yellow arrowheads). In the RNAi-mediated UAS-myrGFP/+;pC1^d-GAL4/UAS-dsx^FRNAi females, anti-Dsx^F signals were specifically diminished in the GFP⁺ pC1^d neurons (Fig. 4 J and L, white arrowheads), indicating efficient and specific knockdown of dsx^F in pC1^d neurons. We then tested aggression in these RNAi-mediated knockdown females and found significantly enhanced headbutt frequency (Fig. 4M) and reduced latency to initiate fighting compared to two control lines (Fig. 4N).

We then asked if expression of the male isoform Dsx^M would rescue or enhance the aggression phenotype in dsx^F knockdown females. We found that dsx^F knockdown females with or without ectopic expression of dsx^M in pC1^d neurons showed a similarly high level of aggression (Fig. 4 M and N). Costaining of Dsx^M and Dsx^F in UAS-myc-dsx^M/+;pC1^d-GAL4/UAS-dsx^FRNAi female brains confirmed the effectiveness of both dsx^F knockdown and Dsx^M expression: two pairs of anti-Dsx^M neurons were observed within the pC1 cluster (ectopic expression of Dsx^M), in which there was no anti-Dsx^F signal (dsx^F knockdown) (Fig. 4 O and P). These results indicate that Dsx^F is crucial in the female-specific pC1^d neurons to inhibit aggression, whereas ectopic expression of Dsx^M in these neurons has no further effect, likely due to the already enhanced aggressiveness.

We further tested if acute knockdown of dsx^F in pC1^d neurons during adulthood using the above-mentioned GAL80^ts strategy would enhance female aggression (SI Appendix, Fig. S7A). We found that knocking down dsx^F expression in pC1^d neurons for 4 d during adulthood significantly enhanced female headbutt frequency (SI Appendix, Fig. S7B) and reduced fighting latency (SI Appendix, Fig. S7C). These results indicate that Dsx^F plays an acute role in pC1^d neurons to inhibit female aggression.

Dsx Isoforms Promote the Development of Aggression-Promoting Neurons in Both Sexes while Oppositely Regulate Their Functional Output.

The above results revealed an aggression-promoting role for Dsx^M in the male-specific P1^a neurons, and an aggression-inhibiting role for Dsx^F in the female-specific pC1^d neurons. Dsx is also crucial for neuronal development, as we found for females with dsx^F knockdown in dsx^brain neurons. We therefore tested if knocking down dsx in P1^a or pC1^d neurons would affect the development of these neurons (Fig. 5 A–E and G–I). However, no changes were observed in either neuronal number or apparent morphology of P1^a neurons in UAS-myrGFP/+;P1^a-GAL4/UAS-dsx^MRNAi males (7.8 neurons per hemisphere in control males and 8.6 in P1^a Dsx^M knockdown males) (Fig. 5 A and D). Notably, we only observed approximately four neurons per hemisphere in UAS-flag-dsx^F/+;P1^a-GAL4/UAS-dsx^MRNAi males with anti-Dsx^F expression (Figs. 2K and 5D). These results indicate that dsx^M knockdown with overexpression of dsx^F in P1^a neurons, but not dsx^M knockdown alone, impairs the development of these neurons.

To further examine if systemic dsx knockout would affect the development of P1^a neurons, we used the P1^a-splitGAL4 reagents (R15A01-AD and R71G01-DBD) and dsx^DBD to generate two additional splitGAL4 lines, P1^a1-splitGAL4 (R15A01-AD;dsx^DBD) and P1^a2-splitGAL4 (R71G01-AD;dsx^DBD), which could be easily assayed in the dsx mutant background (dsx^DBD/dsx^1649-9625). We found that systemic dsx knockout (dsx^DBD/dsx^1649-9625) significantly reduced the number of GAL4-labeled P1 neurons (for P1^a1-splitGAL4, 5.5 neurons per hemisphere in control males, and 3 neurons in dsx mutant males; for P1^a2-splitGAL4, 11.8 neurons per hemisphere in control males, and 9 neurons in dsx mutant males) (Fig. 5 B–E). These results indicate that systemic dsx^M knockout, but not restricted knockdown of expression, impairs the development of P1^a neurons. Taken together with our behavioral experiments described above, these findings show that Dsx^M promotes both the development of the male-specific P1^a neurons and their aggression-promoting function (Fig. 5F).

We next tested whether knocking down dsx^F in pC1^d neurons would affect the development of these neurons, for example by increasing the number of these pC1^d neurons as we observed with the dsx^brain driver (Fig. 5 G–I). However, knocking down dsx^F specifically in pC1^d neurons driven by the pC1^d-GAL4 did not change either the number or projection of these neurons (2.3 neurons per hemisphere in control females and 2.2 in pC1^d-dsx^F knockdown females) (Fig. 5 G and I). Simultaneous Dsx^F knockdown and Dsx^M expression in these pC1^d neurons also did not affect the cell number (Figs. 4O and 5I). Surprisingly, we found that systemic dsx knockout (dsx^DBD/dsx^1649-9625), as we found with P1^a neurons in males, severely impaired the development of pC1^d neurons (0.5 pC1^d neuron per hemisphere in the dsx mutant background, compared to ∼2 neurons in control females) (Fig. 5 H and I). Thus, systemic dsx^F knockout, but not restricted knockdown of dsx^F, affects the development of pC1^d neurons. Combined with the above behavioral experiments, these results reveal that Dsx^F promotes the development of female-specific pC1^d neurons while inhibiting their aggression-promoting function (Fig. 5J).

Taken together, our findings show that Dsx^M and Dsx^F promote the development of male-specific P1^a neurons and female-specific pC1^d neurons, respectively, but oppositely regulate the functional output of these aggression-promoting neurons (Fig. 5K).

Discussion

In this study, we investigated the function of sex-specific Dsx^M and Dsx^F proteins in the nervous system, and revealed crucial roles of Dsx^M in promoting male courtship and Dsx^F in promoting female receptivity. Most importantly, we found opposite regulatory roles for Dsx^M and Dsx^F in a subset of male- or female-specific aggression-promoting neurons, respectively. Whereas Dsx^M promotes male aggression, Dsx^F suppresses female aggression, indicating a potential mechanism underlying male-biased aggression across animal species.

For decades, the male-specific Fru^M was considered a master gene controlling most aspects of male courtship and male-specific aggressive pattern and dominance (14–16, 25, 26), whereas the roles of Dsx^M in males and Dsx^F in females were largely focused on sexual differentiation (12). Given that Dsx^M is expressed in ∼900 neurons that largely overlap with Fru^M in males, and Dsx^F is expressed in ∼700 neurons where there is no recognized or functional Fru protein, the potential involvement of Dsx^M and Dsx^F in these neurons in controlling male and female sexual and aggressive behaviors has been largely underestimated and rarely studied (47–49). Previous studies either used dsx mutant or dominant alleles that affect all dsx-expressing cells other than the nervous system (18, 32), manipulated tra and fru^M to infer dsx function indirectly (18, 22, 33), or focused on dsx-expressing neurons but not the function of the gene (37–40, 42–45, 48, 57).

Our study of the role of neuronal Dsx expression in controlling dimorphic sexual and aggressive behaviors is a crucial step forward in reevaluating the genetic basis underlying these behaviors. Importantly, our study has revealed that Dsx function in the nervous system is crucial for both male and female sexual and aggressive behaviors and is responsible for at least some dimorphic features. Knocking down dsx in the nervous system or specifically in dsx^brain neurons reduced courtship intensity and mating success by ∼80% in males that still have normal Fru^M function. Furthermore, males lacking Fru^M but retaining Dsx^M function could acquire courtship through social experience (17, 18, 58). These results indicate that neuronal Dsx^M is crucial, together with Fru^M, in establishing male courtship behavior. Unlike males that have both Fru^M and Dsx^M function, females have functional Dsx^F but no Fru^F, suggesting an even more fundamental role of Dsx^F in regulating female behaviors. Indeed, knocking down dsx^F pan-neuronally nearly abolished receptivity in virgin females (only 1 out 48 mated in 30 min), and knocking down dsx^F specifically in dsx^brain neurons reduced female receptivity by nearly 90%. Thus, while Fru^M is necessary for innate expression of male courtship, neuronal Dsx^M is crucial for courtship robustness and the experience-dependent courtship acquisition in the absence of Fru^M, and neuronal Dsx^F is necessary for female receptivity.

To date, the study of aggressive behaviors has focused on Fru^M, which has been found to regulate sex-specific fighting patterns (e.g., male lunge vs. female headbutt) and social dominance through fighting (25, 26). However, this role is limited to male-specific aggressiveness. Moreover, although subsets of aggression-promoting neurons including P1^a, Tk^Fru, and aSP2 are fru^M-positive (9, 27, 28), the function of Fru^M in P1^a and aSP2 neurons regarding aggression has not yet been determined. Our results showing that Dsx^M in the male-specific P1^a neurons promotes male aggression, while Dsx^F in the female-specific pC1^d neurons suppresses female aggression, provide a mechanistic model involving both sexes that can explain male-biased aggressiveness.

How would neuronal Dsx proteins regulate sexual and aggressive behaviors in both sexes? There are at least two possibilities: 1) Dsx regulates neuronal development (e.g., the number of neurons through cell proliferation or cell death), as suggested previously (22, 48, 49, 52), and fine neuronal projections (59); or 2) Dsx regulates neuronal physiology, (e.g., sensitivity or excitability), as recently found for Fru^M function (60, 61). Our results suggest both possibilities.

First, females with dsx^F knockdown in all dsx^brain neurons displayed reduced receptivity and increased aggression, probably due to Dsx^F function in neuronal development, as these females have an increased number of pC1 and pC2 neurons (Fig. 3H). The ectopically generated pC1 and pC2 neurons would affect the organization or capability of the dsx circuitry that is crucial for female behaviors (38–41). Indeed, while activation of dsx^brain neurons promotes receptivity in control females, activating these dsx^brain neurons (with more pC1 and pC2) fails to restore receptivity in dsx^F knockdown females.

Second, knocking down dsx^M specifically in P1^a neurons in males, or dsx^F specifically in pC1^d neurons in females, did not affect either the cell number or gross morphology of these neurons, but whether there is any fine dendritic change is not yet conclusive. Interestingly, we found fewer P1^a neurons in dsx mutant males, and pC1^d neurons were almost absent in dsx mutant females, indicating crucial roles for Dsx^M and Dsx^F in the development of P1^a neurons in males and pC1^d neurons in females, respectively. The discrepancy between systemic dsx knockout and restricted dsx knockdown on the development of P1^a/pC1^d neurons might be due to: 1) the late onset of dsx knockdown, as no faithful P1^a/pC1^d expression was observed in the larval stage; 2) dose-dependent mechanisms (knockout vs. knockdown); or 3) noncell autonomous mechanisms (restricted vs. systemic). Nevertheless, the decreased/increased aggression in P1^a-dsx^M knockdown/overexpression males and increased aggression in pC1^d-dsx^F knockdown females are potentially due to nondevelopmental mechanisms. Indeed, acutely overexpressing Dsx^M in P1^a neurons during adulthood also promoted male aggression. In addition, acutely knocking down Dsx^F in pC1^d neurons during adulthood enhanced female aggression. Thus, we propose that Dsx^M and Dsx^F might function to regulate neuronal physiology through their function as transcription factors during adulthood, in addition to their function in neuronal development. Future studies should reveal both developmental and nondevelopmental aspects of Dsx function in distinct subsets of dsx-expressing neurons in males and females. Intriguingly, the dsx-related Dmrt genes control the differentiation of gonads in mammals, which produce sex-specific hormones and play both organizational (developmental) and activational (nondevelopmental) roles in sexually dimorphic behaviors (2, 30, 62, 63). Thus, our findings reveal how a conserved sex determination gene functions cell autonomously in flies, or cell nonautonomously via hormones in mammals, to fulfill both organizational and activational functions in sexually dimorphic behaviors.

The male brain has ∼300 dsx^M neurons, whereas the female brain has only ∼60 dsx^F neurons (47, 48), and Dsx^F has been found to generally reduce cell number through programmed cell death (22, 48, 52). Interestingly, we found that Dsx^F promotes the development of approximately two pairs of female-specific pC1^d neurons, which seems contradictory to its function on other neurons. Thus, there could be a distinct mechanism by which Dsx^F regulates the development of pC1^d neurons, which awaits future investigation. The fact that Dsx^F promotes the development of the aggression-promoting pC1^d neurons while inhibiting their functional output is also intriguing. Such a mechanism does not simply shut down aggressive behaviors by inducing cell death of aggression-promoting neurons, but instead develops the neural substrates underlying the potential for female aggressive behaviors, while simultaneously modulating the intensity of aggression.

Materials and Methods

Fly Stocks.

Flies were maintained at 22 °C or 25 °C in a 12-h light/12-h dark cycle at ∼60% humidity. Canton-S (wtcs) were used as wild-type strains. dsx mutant lines used in Fig. 1A include dsx^683-7058, dsx^1649-9625, and dsx^M+R15, which were used as previously described (18). dsx^GAL4 (48), dsx^LexA (64), Otd-Flp (17), P1^a-GAL4 (w-; R15A01-AD attP40; R71G01-DBD attP2) (50), and UAS-dTrpA1 (53) were used as previously described. actin-GAL4 (BDSC_25374), R57C10-GAL4 (attP2, BDSC_39171), R26E01-GAL4 (attP2, BDSC_60510), R26E01-AD (attP40, BDSC_75740), R15A01-AD (attP40, BDSC_68837), R71G01-AD (attP40, BDSC_70798), R71G01-DBD (attP2, BDSC_69507), tub > GAL80> (BDSC_38881), tub-GAL80^ts (BDSC_7018), UAS-myrGFP attP40 (BDSC_32198), and UAS-myrGFP su(Hw)attP5 (BDSC_32199) were obtained from Bloomington Drosophila Stock Center. UAS-dsx^MRNAi (attP2) (34), UAS-dsx^FRNAi (attP2) (34), UAS-myc-dsx^M (attP40) (46), and UAS-flag-dsx^F (attP40) (46) lines were used as previously described and as also described below. dsx^DBD was generated in this study and described below in detail. Detailed information about fly stocks and other materials used in this study is listed as SI Appendix, Table S1.

Transgenic Flies.

dsx^DBD was generated by replacing the first coding exon of dsx with the GAL4-DBD fragment via CRISPR/Cas9-mediated homology-directed repair. UAS-dsx^MRNAi and UAS-dsx^FRNAi lines were recently generated by the Tsinghua Fly Center at Tsinghua University based on pVALIUM20 plasmids (65) and were used previously (34). UAS-flag-dsx^F and UAS-myc-dsx^M lines were generated previously in our laboratory (46).

Dsx^M and Dsx^F Antibodies.

Dsx^M and Dsx^F antibodies were generated in our laboratory with help from GenScript and were used previously (34). In brief, the rabbit polyclonal anti-Dsx^M antibody was generated using the male-specific Dsx fragment N′-ARVEINRTVA QIYYNYYTPM ALVNGAPMYL TYPSIEQGRY GAHFTHLPLT QICPPTPEPL ALSRSPSSPS GPSAVHNQKP SRPGSSNGTV HSAASPTMVT TMATTSSTPT LSRRQRSRSA TPTTPPPPPP AHSSSNGAYH HGHHLVSSTA AT-C′. The mouse monoclonal anti-Dsx^F antibody was generated using the peptide N′-PLGQDVFLDYCQKLLEKFRYPWELMPLMYVILKDADANIEEASRRIEEGQYVVNEYSRQHNLNIYDGGELRNTTRQCG-C′ (the underline marks the female-specific sequence).

Male Courtship Assay.

Newly enclosed virgin males were collected and single housed for 4 to 7 d until testing. Virgin wtcs females were collected and group-housed for 4 to 7 d as courtship targets. To measure courtship, small round two-layer chambers (diameter: 1 cm; height: 3 mm per layer) were used. Tester males and target females were loaded individually into the two-layer food contained chambers and were separated by a plastic transparent film until the courtship test for 30 min. CI, which is the percentage of observation time that a fly performs any courtship step in the first 10 min of the video, was used to measure courtship, and scored manually using the LifeSongX software (lifesong.bio.brandeis.edu/).

Female Receptivity Assay.

Virgin females were collected and group-housed for 4 to 7 d until the test. Newly enclosed wtcs virgin males were collected and group-housed for 4 to 7 d as courters unless otherwise described. Males and females were loaded and tested as described for the male courtship assay. Female receptivity was manually measured every 2 min as the cumulative percentage of females with successful copulation with males.

For dTrpA1-based neuronal activation experiments (Fig. 3 J–L), flies were raised at 22 °C until behavioral tests. Behavioral chambers containing flies were placed at control (22 °C) or experimental temperatures (29 °C) for 30 min before behavioral tests.

Male and Female Aggression Assays.

All aggression assays were carried out at 25 °C and ∼60% humidity between 11:00 AM and 5:00 PM. The aggression chamber is made up of four acrylic plates. The bottom plate has 24 wells, containing food substrates (regular food plus apple juice and yeast, diameter: 8 mm; depth: 3 mm). The second and third plates have 24 cylindrical arenas (diameter: 15 mm; height of each plate: 3 mm), in which two tester flies were gently aspirated and separated by a transparent film. Flies were allowed to recover for ∼60 min and video was recorded for ∼30 min after removing the transparent film. For M–M and F–F aggression, two males or females of the same-genotype were used. All aggression tests were performed at least twice for each genotype and were manually analyzed. For detailed information about the aggression assay, see SI Appendix, Materials and Methods.

Spontaneous Locomotion Assay.

Four- to 7-d-old males and females were raised using the same condition for aggression tests. Individual flies were loaded into round chambers (diameter: 2 cm; height: 3 mm) with regular food at the bottom and recorded for 24 h starting from 9:00 AM under constant light condition. The average walking velocity during the 24-h recording was quantified using the ZebraLab software system (ViewPoint Life Sciences), as previously described (50).

Quantitative Real-Time PCR.

qPCR was performed using a LightCycler 96 Real-Time PCR System (Roche). Total RNA was extracted from ∼30 adult flies using a commercial TRIZOL reagent (15596026, Invitrogen) and purified with DNA-free kit (AM1906, Ambion) according to the manufacturer’s protocol. First-strand cDNA was synthesized for each RNA sample using Prime Script reagent kit (18091050, Invitrogen). EvaGreen Dye (31000, Biotium) and High Fidelity PCR SuperMix (AS131-21, Transgen) were used to conduct qPCR. actin was used as control for normalization. The primers used were as follows:

• actin-forward: 5′- CAGGCGGTGCTTTCTCTCTA -3′,

• actin-reverse: 5′-AGCTGTAACCGCGCTCAGTA-3′;

• dsx^F-forward: 5′- TTCCGCTATCCTTGGGAGCT -3′;

• dsx^F-reverse: 5′- CATCCACATTGCCGCGTTGT -3′;

• dsx^M-forward: 5′-GAAGAGGCTTCCCGGCGAAT-3′;

• dsx^M-reverse: 5′- GGACAAATCTGTGTGAGCGG -3′.

Tissue Dissection, Staining, and Imaging.

Flies were dissected at three developmental stages (L3, 48 h APF, and 4- to 7-d-old adult). Larval gender was determined by taking advantage of the translucence of the cuticle and looking for the absence or presence of gonads. All flies were dissected in Schneider’s insect medium (Thermo Fisher Scientific) and fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 20 to 30 min at room temperature. After 4 × 15-min washing in PAT3 (0.5% Triton X-100, 0.5% bovine serum albumin in PBS), tissues were blocked in 3% normal goat serum (NGS) for 60 min at room temperature, then incubated in primary antibodies diluted in 3% NGS overnight at 4 °C, then washed four times in PAT3, and incubated in secondary antibodies diluted in 3% NGS for 1 to 2 d at 4 °C. Tissues were then washed four times thoroughly in PAT3 and mounted for imaging. Primary antibodies used: rabbit anti-GFP (Invitrogen A11122; 1:1,000), mouse anti-GFP (Invitrogen A11120; 1:1,000), mouse anti-Bruchpilot (Developmental Studies Hybridoma Bank nc82; 1:50), mouse anti-Myc (MBL, M047-3; 1:600), mouse anti-Dsx^F and rabbit anti-Dsx^M (see Materials and Methods for details of antibody generation; 1:400). Secondary antibodies used: donkey anti-rabbit IgG conjugated to Alexa 488 (Invitrogen, A21206; 1:500), donkey anti-mouse IgG conjugated to Alexa 488 (Invitrogen, A21202; 1:500), donkey anti-rabbit IgG conjugated to Alexa 555 (Invitrogen, A31572; 1:500), donkey anti-mouse IgG conjugated to Alexa 555 (Invitrogen, A31570; 1:500). Samples were imaged at 10× or 20× magnification on Zeiss 700 confocal microscope and processed with ImageJ software. For quantification of anti-Myc fluorescence intensity (SI Appendix, Fig. S5D), all control and experimental samples were imaged using the same parameter and analyzed using ImageJ.

For visualizing morphological appearances of flies (SI Appendix, Figs. S1 B–E and S2 H and I), 2- to 4-d-old adult males and females were frozen at −80 °C for 30 min and later imaged by a Nikon Shuttle pix P400RV stereoscopic microscope.

Statistics.

For all experiments, control and experimental flies were tested together under the same conditions, and data were collected from at least two independent experiments. For CNS imaging data, we dissected 5∼10 flies for each condition. For assaying courtship and aggression, which have a performance index for individual flies, we tested ∼20 flies. For assaying female receptivity, which used a single index for all samples (percentage mated successfully), we tested up to 100 pairs of flies. Statistical analyses were performed using GraphPad Prism 8 and are indicated in each figure legend. Data presented in this study were first verified for normal distribution by D’Agostino–Pearson normality test. If normally distributed, Student’s t test was used for pairwise comparisons, and one-way ANOVA was used for multiple group comparisons, followed by Tukey’s multiple comparisons. If not normally distributed, Mann–Whitney U test was used for pairwise comparisons, and Kruskal–Wallis test was used for comparisons among multiple groups, followed by Dunn’s multiple comparisons. For male mating success and female receptivity, χ² tests were performed to compare two different groups at the 30-min time point, if not otherwise mentioned.

Data, Materials, and Software Availability

All data generated or analyzed during this study are included in the main text and supporting information. This study does not involve new code or sequence data. Fly stocks and reagents used in this study are available from the corresponding author upon reasonable request.