A modular circuit coordinates the diversification of courtship strategies

Abstract

Mate recognition systems evolve rapidly to reinforce the reproductive boundaries between species, but the underlying neural mechanisms remain enigmatic. Here we leveraged the rapid coevolution of female pheromone production and male pheromone perception in Drosophila^1,2 to gain insight into how the architecture of mate recognition circuits facilitates their diversification. While in some Drosophila species females produce unique pheromones that act to arouse their conspecific males, the pheromones of most species are sexually monomorphic such that females possess no distinguishing chemosensory signatures that males can use for mate recognition³. We show that Drosophila yakuba males evolved the ability to use a sexually monomorphic pheromone, 7-tricosene, as an excitatory cue to promote courtship. By comparing key nodes in the pheromone circuits across multiple Drosophila species, we reveal that this sensory innovation arises from coordinated peripheral and central circuit adaptations: a distinct subpopulation of sensory neurons has acquired sensitivity to 7-tricosene and, in turn, selectively signals to a distinct subset of P1 neurons in the central brain to trigger courtship. Such a modular circuit organization, in which different sensory inputs can independently couple to parallel courtship control nodes, may facilitate the evolution of mate recognition systems by allowing novel sensory modalities to become linked to male arousal. Together, our findings suggest how peripheral and central circuit adaptations can be flexibly coordinated to underlie the rapid evolution of mate recognition strategies across species.

Fig. 1. Sexually ambiguous pheromones do not preclude courtship in the dark.

a, Phylogeny of 99 Drosophila species for which cuticular pheromones have been characterized (left) and primary sex pheromones of select species (right)^,. 7-T, 7-tricosene; cVA, cis-vaccenyl acetate; 5,25-HTCD, 5,25-hentriacontadiene; 7,11-ND, 7,11-nonacosadiene; 9-T, 9-tricosene (M. Khallaf, personal communication); 9,23-TTCD, 9,23-tritriacosadiene. b, Courtship as captured by the inter-fly distance (IFD) between a D. melanogaster (mel), D. erecta (ere), D. simulans (sim), D. yakuba (yak), D. eugracilis (eug) or D. ananassae (ana) male with a conspecific female in the dark. Left, IFD traces over time for a single representative pair as courtship proceeds. Middle, heatmaps for 12 pairs, aligned to courtship initiation for all species except D. eugracilis, for which videos were aligned to the time of first interaction. Red, dotted lines indicate the time of first interaction (D. eugracilis) or courtship initiation (all other species). Right, histograms of time as a function of IFD for the same 12 courting pairs. c, Average courtship bout length in the dark following courtship initiation for males paired with a conspecific female (mel/sim/yak n = 20; ere/eug/ana n = 12), D. yakuba males paired with a D. simulans female (n = 15) or D. simulans males paired with a D. yakuba female (n = 15). d, Left, average courtship bout length following courtship initiation for D. yakuba males paired with oe⁻ females mock perfumed (grey) or perfumed with the D. yakuba pheromone 7-T (black) in the dark (n = 10) and (right) histograms (as in b). Data points represent individual males; bars are median. Statistics: Kruskal–Wallis test (c) or unpaired Mann–Whitney (d). Letters denote statistically different groups (P < 0.05). ****P < 0.0001. Diagram in a adapted from ref. , Springer Nature Limited. Source Data

Fig. 2. P1 neurons of D. yakuba males share conspecific tuning pattern of dimorphic species.

a, P1 neurons labelled by 71G01>CD8::GFP (D. melanogaster, D. erecta), 71G01>GCaMP6s (D. simulans) or SplitP1>CD8::GFP (D. yakuba; Methods) expression, stained for GFP (green) and neuropil counterstain (magenta). Images were masked to remove glial fluorescence from ie1 marker and non-P1-specific labelling for clarity. Imaging location used for experiments in b is indicated by white, dotted box. b, Left, representative traces of P1 responses in the LPC in males of each species evoked in response to the taste of a conspecific or heterospecific female (black ticks indicate time of foreleg taps). Middle, averaged tap-evoked functional responses (∆F/F₀, black ticks) of the P1 neurons across all males. Right, average peak response (∆F/F₀) for each male evoked by a given female target (sample sizes: mel n = 10; ere n = 10; sim n = 8; yak n = 5). Data points represent individual males; error bars are mean ± s.e.m. Statistics: analysis of variance (ANOVA) with Tukey’s post hoc. Letters denote statistically different groups (P < 0.05). Source Data

Fig. 3. Altered pheromone sensitivity in D. yakuba sensory neurons.

a, Left, Ppk23⁺ sensory neurons in D. melanogaster and D. yakuba male foreleg tarsal segments, marked by CD8::GFP (green), distal end up. Middle, Ppk23⁺ sensory afferents in the first thoracic segment of the ventral nerve cord expressing CD8::GFP (green) with neuropil counterstain (magenta), anterior side up. Right, functional responses of Ppk23⁺ afferents evoked by indicated females (black ticks indicate foreleg taps). b, Average courtship bout length in the 10 min following courtship initiation for wild-type males paired with conspecific female (n = 18); D. melanogaster female (n = 20) or ppk23 mutant (n = 20 each) in the light. c,d, Average courtship bout length of ppk23 mutant D. melanogaster (n = 20 each) (c) or D. yakuba (n = 20 each) (d) males paired with conspecific females in the dark. e, Left, average functional responses (∆F/F₀) recorded from foreleg sensory afferents aligned to time of a tap and (right) average peak response (∆F/F₀) per male for a given female target. (sample sizes: yak ppk23>GCaMP n = 9; mel ppk23>GCaMP n = 7; yak ∆ppk23;ppk23>GCaMP n = 4; yak ∆ppk25;fru>GCaMP n = 8). f, Average courtship bout length of wild-type males paired with conspecific females in the dark (left) (n = 10 each) and ppk25 mutant males paired with D. melanogaster females in the light (right) (n = 10 each). g, Average courtship bout length before (Pre) or during (Stim) periods of optogenetic stimulation of Ppk23>CsChrimson or >CsChrimson control animals paired with a D. melanogaster female (n = 10 each). h, Model summarizing the inferred changes in the sensitivity of sensory populations and effect on P1 neuron activity. D. melanogaster and D. simulans diagrams on the basis of previous reports^,,,,,,. For behavioural tests (b–g), points represent individual males; bars are median. For imaging (e), shading is mean ± s.e.m., points are individual males and error bars are mean ± s.e.m. Statistics: ANOVA with Tukey’s post hoc (e), Mann–Whitney (b–d,f), Wilcoxon test (g). Letters denote statistically different groups (P < 0.05). **P < 0.01, ****P < 0.0001; NS, not significant. WT, wild type. Source Data

Fig. 4. Sensory specialization of Fru⁺ and Dsx⁺ P1 subpopulations in D. yakuba.

a,b, Left, template brain registrations of neurons labelled by intersection of Fru (a; Fru∩P1) or Dsx (b; Dsx∩P1) and the P1-driver 71G01 (Methods). Middle, averaged functional responses (∆F/F₀) aligned to tap of indicated female and average peak response (∆F/F₀) per male for a given female target. Right, average courtship bout length, total percentage time pursuing or total percentage time extending a unilateral wing towards a D. melanogaster female target before (Pre) or during (Stim) optogenetic stimulation of Fru∩P1>CsChrimson or Dsx∩P1>CsChrimson males and the fraction of flies engaging in these behaviours (sample sizes: Fru∩P1>GCaMP n = 6; Fru∩P1>CsChrimson n = 20; Dsx∩P1>GCaMP n = 10; Dsx∩P1>CsChrimson n = 20). c,d,e, Functional responses (as in a,b) of all Fru⁺ (c) or Dsx⁺ (d,e) neurons innervating the LPC in wild-type (c, n = 5; d, n = 8) or ∆ppk25 mutant (e, n = 8) males. f, Functional responses (as in b) of Dsx∩P1 to D. melanogaster mock-perfumed oe⁻ (oe⁻), 7,11-HD-perfumed (oe⁻ +7,11-HD), 7-T-perfumed (oe⁻ +7-T) or cVa-perfumed (oe⁻ +cVA). g, Courtship in the dark as captured by IFD of a D. yakuba male towards a conspecific female with constitutively silenced P1 subsets (Fru∩P1>Kir (middle), Dsx∩P1>Kir (bottom) or genetic control (top; 71G01-DBD; UAS-Kir)) (n = 6). Left, heatmaps for six pairs, aligned to courtship initiation (red, dotted line). Middle, histograms of time as a function of IFD for the same six courting pairs. Right, average courtship bout length in the 10 min following courtship initiation of each genotype (n = 25 each). For functional imaging (a–f), shading represents mean ± s.e.m. Points are individual males; error bars are mean ± s.e.m. For behavioural tests (a,b,e), points are individual males and bars are median. Statistics: ANOVA with Tukey’s post hoc (a–f), Wilcoxon test (a,b), Kruskal–Wallis test (g). Letters denote statistically different groups (P < 0.05). ****P < 0.0001. Source Data

Fig. 5. Subspecialization of P1 neuron subtypes.

a,b, Diagrams summarizing proposed sensory specializations of Fru∩P1 and Dsx∩P1 in D. melanogaster (a) and D. yakuba (b). Pheromone inputs inferred from functional imaging and behavioural data in Fig. 4 and Extended Data Fig. 6 or previously reported^,,, are indicated by solid arrows. Further hypothesized sensory inputs are indicated by dotted arrows. c, Model for the diversification of courtship behaviours by neural subspecialization. Independent retuning of sensory inputs to behaviourally redundant but molecularly distinct P1 subtypes may facilitate the rapid evolution of sensory signals that control a male’s sexual arousal and courtship.

Extended Data Fig. 1. Inter-fly distances below 8 mm approximate courtship in the dark.

a, Automated scoring of the samples plotted in Fig. 1b. (Left) percent time pairs spent at IFD < 8 mm and (Right) average bout length at IFD < 8 mm after first interaction (eug) or after courtship initiation (all other species). Each data point represents the individual pairs shown in the heatmaps in Fig. 1b (n = 12 each) b, (Top) Inter-fly distance (IFD) traces over time for a single representative pair of D. simulans or D. yakuba as courtship proceeds. Comparison of manually confirmed courtship bouts with those estimated by IFD thresholding revealed that an IFD threshold of 8 mm most accurately replicated manual scoring, resulting in the lowest incidence of false positives and negatives. (Bottom) Courtship bouts approximated by IFD < 8 mm thresholding compared to manual scoring by a blinded observer. Red dotted lines indicate initiation of courtship. c,d, Comparison between the percent time courting and the average courtship bout length scored by automated IFD < 8 mm thresholding versus manually scored by a blinded observer for the same courting pairs across species (c) or for D. yakuba male courtship of oenocyte-less females perfumed with the D. yakuba pheromone 7-T (d)(n = 12 each). Generally, IFD-thresholding overestimates the total amounts of courtship (due to incidental periods where flies are close but not courting) and underestimates the average courtship bout length (due to transient periods where the female moves away from the male but courtship continues), but captures the overall trends observed in samples manually scored by a blinded observer (Fig. 1c, d). The samples analyzed here in c are the same plotted in Fig. 1b and the samples analyzed in d are the same plotted in Fig. 1d (n = 10 each). All quantifications are of the 10 min period following courtship initiation. Data points represent individual males, bars are median. Statistics: Kruskal-Wallis tests (a). Letters denote statistically different groups (p < 0.05). Source Data

Extended Data Fig. 2. P1 neurons play a conserved role in promoting courtship across species.

a,b, Percent time courting prior to (Pre) or during (Stim) periods of optogenetic stimulation of D. erecta (a; n = 10 each) or D. yakuba (b; n = 9 each) males expressing CsChrimson in P1 neurons (71G01>CsChrimson (a) or splitP1>CsChrimson (b)) or control animals (> CsChrimson) paired with a D. melanogaster female. Data points represent individual males. Statistics: Wilcoxon test (a,b). Asterisks: **P < 0.01, ns, not significant. Source Data

Extended Data Fig. 3. Conserved anatomy of pheromone-detecting sensory neurons.

a, CRISPR/Cas9 targeting strategy for the genetic deletion of the ppk23 locus and validation by PCR and ie1-mCherry marker expression. b,c,d, Percent time courting in the 10 min following courtship initiation, courtship latency, and courtship bout length of ppk23 mutants relative to wildtype for D. melanogaster in the dark (b) and D. yakuba males in the light (c) or in the dark (d) (n = 20 each). Average Bout Lengths are re-plotted from Fig. 3 for reference. e, Representative bright field images and quantification showing expression of CD8::GFP (green) in the soma of the Ppk23+, Fru+, and Dsx+ sensory neurons in the foreleg tarsal segments of D. melanogaster and D. yakuba males. Images are distal end up. f, (Left) Average functional responses (∆F/F₀) aligned to time of a tap (as in Fig. 3) of indicated D. melanogaster (oe+ ctrl), mock-perfumed oenocyte-less (oe-), 7,11-heptacosadiene-perfumed (oe- +7,11-HD), 7-tricosene-perfumed (oe- +7-T), or cis-Vaccenyl acetate-perfumed (oe- +cVA) female of the foreleg sensory afferents and (Right) average peak response (∆F/F₀) per male for a given female target (n = 5). g, Functional responses of Dsx+ neurons in the LPC (as in Fig. 4d) in ppk23 mutant males (n = 6)). For behavioral tests (b-c), points represent individual males, bars are median. For anatomy (e), points represent individual forelegs, bars are mean ± SEM. For imaging (f,g), shading is mean ± SEM, points are individual males, error bars are mean ± SEM. Statistics: ANOVA with Tukey’s post-hoc (f,g) or unpaired Mann-Whitney (b-d). Letters denote statistically different groups (p < 0.05). Asterisks: **P < 0.01, ****P < 0.0001, ns, not significant. Source Data

Extended Data Fig. 4. Ppk25+ Sensory neurons promote courtship in the dark in D. yakuba.

a,b, CRISPR/Cas9 targeting strategy for the genetic deletion of the ppk25 locus and validation by PCR (a) and courtship in the dark (as in Extended Data Fig. 3) toward D. yakuba females (b). Average courtship bout lengths are re-plotted from Fig. 3 for reference (n = 20). c, Replotting of sensory responses in D. yakuba wildtype and ppk23 and ppk25 mutant males from Fig. 3e for statistical comparison. For behavioral tests (b), points represent individual males, bars are median. For imaging (c), points are individual males, error bars are mean ± SEM. Statistics: Unpaired Mann-Whitney (b) or ANOVA with Tukey’s post-hoc (c). Letters denote statistically different groups (p < 0.05). Asterisks: ****P < 0.0001, ns, not significant. Source Data

Extended Data Fig. 5. Differences in projection patterns of P1 subpopulations.

a, Diagram of the subdivision of the pC1 neurons by Dsx- (purple) and Fru-expression (red). Also noted is the subset of pC1 neurons captured by the P1-specific driver used in this study and the populations labeled through genetic intersection of this P1 driver with Dsx or Fru. Note that the labeled populations are not to scale, but precise neuron counts labeled in each intersection are provided in b-e. b-e, (Left) Unmasked images of Fru∩P1 and Dsx∩P1 neurons labeled by intersection of Fru (b,d) or Dsx (c,e) and the P1-driver 71G01 (Fru∩P1 and Dsx∩P1, respectively) in D. melanogaster and D. yakuba. Stained for GFP (green) and neuropil counterstain (magenta). (Right) Representative images of anti-Dsx (purple) and anti-Fru (red) staining used to count Dsx+ and Fru+ subsets and quantification of Dsx+ Fru∩P1 neurons and Fru+ Dsx∩P1 neurons within the Dsx+ pC1 cluster, based on anti-Dsx and anti-Fru immunostaining, respectively. Data points represent cell number in one hemisphere and both hemisphere counts are plotted separately (Sample sizes: mel Fru∩P1 n = 7; mel Dsx∩P1 n = 8; yak Fru∩P1 n = 7; yak Dsx∩P1 n = 5). f,g, Comparison of the projections of Fru∩P1 and Dsx∩P1 subpopulations across subtype (f) or across species (g). Heatmap (left) and skeletonized overlays (right) of the comparison groups. Regions of greatest difference across both subtype and species indicated (cyan triangle) and region of significant difference between Dsx∩P1 and Fru∩P1 in D. yakuba noted (yellow arrow). All images are dorsal side up. Source Data

Extended Data Fig. 6. Subspecialization of P1 types in D. melanogaster.

a,b, (Left) Maximum intensity projections of neurons labeled by intersection of Fru (a) or Dsx (b) and the P1-driver 71G01 (Fru∩P1 and Dsx∩P1, respectively) in D. melanogaster, registered to a common template brain (see methods). (Middle) Averaged functional responses (∆F/F₀) aligned to time of a tap of indicated conspecific or heterospecific female and average peak response (∆F/F₀) per male for a given female target (n = 10 each). (Right) Average courtship bout length, total percent time pursuing, or total percent time extending a unilateral wing toward a D. suzukii female target prior to (Pre) or during (Stim) optogenetic stimulation of Fru∩P1>CsChrimson or Dsx∩P1>CsChrimson males and the fraction of flies engaging in these behaviors over the course of the experiment. (n = 10 each) c, Functional responses of the or Dsx+ neurons innervating the LPC where P1 neurons reside with or without fru-Gal80. (Left) Diagram of the P1/pC1 neural cluster illustrating the nested nature of the Fru+ contingent within the broader Dsx+ population. Green highlighting represents which subsets express GCaMP in the different genetic backgrounds (all Dsx+Fru- and Dsx+Fru+ pC1 neurons in dsx>GCaMP flies, but only the Dsx+Fru- subset in the dsx>GCaMP; fru-Gal80 flies). Also, representative images Dsx+ neural processes and somata expressing GCaMP (green) and counter-stained for Fru (red). Note the denser projections and the presence of Fru+ GCaMP+ somata in the dsx>GCaMP animals, which is absent in animals expressing fru-Gal80. (Right) Averaged functional responses (∆F/F₀) aligned to time of a tap of indicated conspecific or heterospecific female and (Right) average peak response (∆F/F₀) per male for a given female target. (Sample sizes: dsx>GCaMP n = 6; dsx>GCaMP;fru-Gal80 n = 7). For functional imaging (a-c), shading represents mean ± SEM. Points are individual males, error bars are mean ± SEM. For behavioral tests (a,b), points are individual males. Statistics: ANOVA with Tukey’s post-hoc (a-c) or Wilcoxon test (a,b). Letters denote statistically different groups (p < 0.05). Asterisks: *P < 0.05, **P < 0.01, ***P < 0.001. Source Data

Extended Data Fig. 7. Activation of P1 subtypes.

a-d, (Top) Raster plots of female pursuit (gray) or unilateral wing extension behaviors (UWE; red) exhibited upon optogenetic stimulation of Fru∩P1>CsChrimson (a,c) or Dsx∩P1>CsChrimson (b,d) males and the proportion of flies engaging in these behaviors over the course of the experiment in D. yakuba (a,b; n = 20 each) or D. melanogaster (c,d; n = 10 each). Dotted lines and areas of red shading indicate stimulation periods. Source Data

Extended Data Fig. 8. Multimodal sensory integration by P1 facilitates sex discrimination in D. yakuba.

a, Percent time a D. melanogaster or D. yakuba male spends courting one target over the other when presented a choice between a conspecific female and male or a conspecific female and a heterospecific female that shares the same pheromone profile (D. simulans) (n = 10 each). b, CRISPR/Cas9 targeting strategy for the genetic deletion of the OR67d locus and validation by PCR. c, Average courtship bout lengths of a wildtype or OR67d mutant D. yakuba male directed toward another D. yakuba male (n = 20 each). d, Average courtship bout lengths of a D. yakuba male toward a D. yakuba female mock perfumed or perfumed with 7-T, 7,11-HD, or cVA (n = 10 each). e, (Left) Averaged functional responses (∆F/F₀) aligned to time of a tap of D. yakuba females perfumed as in c and average peak response (∆F/F₀) per male for a given female target (n = 7). f,g, Average courtship bout lengths of a wildtype or deafened (arista removed) D. yakuba male directed toward another D. yakuba male (f; n = 20 each) or a wildtype D. yakuba male toward a wildtype or muted (wings removed) D. yakuba male (g; intact n = 20, muted n = 18). h, Diagram summarizing the multisensory cues the D. yakuba males rely upon to mediate conspecific attraction and inhibit heterospecific and intrasexual courtship. For behavioral tests (a,c,d,f,g), points represent individual males, bars are median. For imaging (e), points are individual males, error bars are mean ± SEM. Statistics: Wilcoxon test (a,d), unpaired Mann-Whitney (c,f,g), or ANOVA with Tukey’s post-hoc (e). Letters denote statistically different groups (p < 0.05). Asterisks: *P < 0.05, ns, not significant. Source Data

Extended Data Fig. 9. Mating in the dark has arisen repeatedly in monomorphic species.

Phylogeny of Drosophila species (as in Fig. 1) with previous reports of various species’ capacity to mate in the dark indicated. Diagram in a adapted with permission from ref. , Springer Nature Limited.

Extended Data Fig. 10. Overall anatomical conservation of the Fru+, Dsx+, and ascending circuitry.

a, Unmasked splitP1 > CD8::GFP expression in the brains of D. melanogaster (left) and D. yakuba (right) males. Stained for GFP (cyan) and neuropil counterstain (gray). b, Dsx>CD8::GFP (magenta) expression in the brains of D. melanogaster (left) and D. yakuba (right) males. c, (Top) Fru>CD8::GFP (green) expression, as in (b). (Middle and bottom) Photoactivation labeling of mAL (red) and vAB3 (grey), respectively, using Fru>SPA-GFP. All images are dorsal side up.