Hox genes mediate the escalation of sexually antagonistic traits in water striders

Sexual conflict occurs when traits favoured in one sex impose fitness costs on the other sex. In the case of sexual conflict over mating rate, the sexes often undergo antagonistic coevolution and escalation of traits that enhance females' resistance to superfluous mating and traits that increase males' persistence. How this escalation in sexually antagonistic traits is established during ontogeny remains unclear. In the water strider Rhagovelia antilleana, male persistence traits consist of sex combs on the forelegs and multiple rows of spines and a thick femur in the rear legs. Female resistance traits consist of a prominent spike-like projection of the pronotum. RNAi knockdown against the Hox gene Sex Combs Reduced resulted in the reduction in both the sex comb in males and the pronotum projection in females. RNAi against the Hox gene Ultrabithorax resulted in the complete loss or reduction of all persistence traits in male rear legs. These results demonstrate that Hox genes can be involved in intra- and inter-locus sexual conflict and mediate escalation of sexually antagonistic traits.


Introduction
The evolutionary interests of males and females during mating interactions often diverge, leading to the coevolution of sexually antagonistic traits that are favoured in one sex at a fitness cost to the other [1,2]. Empirical and theoretical data on the coevolution of the sexes established sexual conflict as a major force in evolutionary change within and between lineages [1,3]. The consequences of sexually antagonistic selection are manifest in some spectacular shape changes in water striders, one of the most prominent model systems for the study of sexual antagonism in nature [4][5][6][7]. In many species, males are often favoured to mate repeatedly, but females pay increasing fitness costs for multiple mating [1,8]. The repeated evolution of grasping traits that allow males to overcome females' resistance is often matched by the evolution of anti-grasping traits in females that enhance their ability to resist [1,6,9,10]. Sexually antagonistic traits are highly variable in shape and can occur in any segment along the body axis. Examples involve the modification of antennae, forelegs or rear legs into grasping traits in the males, whereas females are known to match these with various anti-grasping traits such as erect abdominal spines [4,6,11].
Males and females share the same genome, and selection on sex-specific traits inevitably results in inter-locus or intra-locus sexual conflict [1,10,12,13]. Inter-locus sexual conflict occurs when males and females undergo different selective pressures at different loci, potentially leading to antagonistic coevolution [12,14]. On the other hand, intra-locus sexual conflict occurs when males and females undergo different selective pressures at the same locus, a process that is thought to limit the evolution of one sex owing to the change in the other [12,14]. Several studies have highlighted the role of developmental genes in shaping certain male-specific morphologies in insects, including a case of an antagonistic trait in water striders [6,15,16]. However, the developmental genetic dynamics underlying the escalation of sexually antagonistic traits in both sexes remain untested.
In the water strider Rhagovelia antilleana, both males and females have evolved antagonistic traits, which influence mating rate, along the body axis [5,17]. On the first thoracic segment, males develop sex combs located on their forelegs and winged females develop a prominent spike-like extension of the pronotum [5]. On the third thoracic segment, males have modified rear legs with rows of spines and large femurs, whereas females possess fewer spines and thinner femurs [5]. Because Hox genes are known to specify segment identity, we sought to test a possible role of these genes in the development of sexually antagonistic traits and whether these genes are involved in intra-and inter-locus sexual conflict. By inactivating Sex Combs Reduced (Scr) and Ultrabithorax (Ubx), known to control the identity of the first and third thoracic segments, respectively, we uncovered the importance of these two genes in sexual conflict and their role in the escalation of male and female sexually antagonistic traits.

Material and methods (a) Insect rearing
Laboratory populations of R. antilleana were kept in water tanks at 258C, 50% humidity, 14 h of daylight and fed daily on crickets. Styrofoam floaters were provided for adult females to lay eggs on. Eggs were regularly transferred to separate tanks to prevent cannibalism on the newly hatched nymphs.
(b) Cloning of R. antilleana Scr and Ubx Extraction of total RNA from a mix of 10 complete R. antilleana first to fifth instar nymphs was performed using Trizol. cDNA was synthesized using SuperScript III First-Strand kit according to the manufacturer's instructions (Invitrogen). Fragments of Scr and Ubx genes were amplified (electronic supplementary material, figures S1 and S2) by PCR using GoTaq G2 DNA polymerase (Promega) and the primers described in electronic supplementary material, table S1. PCR products were examined by electrophoresis, purified using PCR Minelute kit (Qiagen) and cloned into a pGEM-T vector kit (Promega). The sequences of R. antilleana Scr and Ubx can be retrieved in GenBank using the following accession numbers: MG999826 and MG999808, respectively, and in electronic supplementary material, figures S1 and S2.
(c) Nymphal RNA interference in R. antilleana RNAi procedures have been efficiently adapted in water striders [6,18,19]. To synthesize double-stranded RNA, a DNA template was produced using PCR and Scr or Ubx primers tagged with a T7 RNA polymerase promoter and the Scr or Ubx plasmids as PCR templates (electronic supplementary material, figures S1 and S2, and table S1). The resulting PCR products were purified using Minelute kit (Qiagen) followed by an in vitro transcription with T7 RNA polymerase þ (Ambion) to obtain the double-stranded RNA (dsRNA) corresponding to each gene. Both dsRNAs were purified using RNeasy Mini Kit (Qiagen), concentrated with speedvac and re-suspended in 1Â injection buffer [20] at a final concentration of 3 mg ml 21 . Yellow fluorescent protein (yfp) dsRNA was used as a control at 1.8 mg ml 21 concentration. We performed injections in R. antilleana first to third nymphal instars using a SteREO Discovery V8 (Zeiss), a Cell Tram Vario Oil Eppendorf injector and a Narishige micromanipulator under CO 2 anaesthesia. The number of injected nymphs, emerged adults and frequency of successful knockdown are presented in electronic supplementary material, table S2. royalsocietypublishing.org/journal/rsbl Biol. Lett. 15: 20180720 Merlin Compact Scanning Electron Microscope. Image analysis and measurements of the different body parts were performed using Zen software (Zeiss) and Fiji software [21]. For measurements of the teeth structures composing the sex combs, we measured the length of 20 teeth in three different males for both yfp control and Scr RNAi, which corresponds to an effective yfp control n ¼ 60, Scr RNAi n ¼ 60. Measurements of pronotum between control and Scr RNAi have been performed on: control n ¼ 12 (4 yfp þ 8 wild-type), Scr RNAi n ¼ 7. Then measurements and quantification of Ubx RNAi phenotypes have been performed with: yfp control males n ¼ 21, Ubx RNAi males n ¼ 9, yfp control females n ¼ 12, Ubx RNAi females n ¼ 10.

(e) Statistical analysis
Shapiro tests (to test for the normal distribution of data), Student's t-tests and Wilcoxon tests (depending whether variables followed normal distributions) were performed using RStudio  royalsocietypublishing.org/journal/rsbl Biol. Lett. 15: 20180720 version 1.0.153 for the statistical analysis of phenotypic quantification. R script used in this study [22] is available in the Dryad Digital Repository using the following link: http://dx.doi.org/ 10.5061/dryad.s76h0s6.

Results (a) Scr is required for both male persistence and female resistance traits
Secondary sexual traits in R. antilleana males and females start to appear late during the fifth nymphal instar ( figure 1a,b), and only become prominent in the adult ( figure 1c,d). In the adults, male forelegs are equipped with sex combs (figure 2a) [5], whereas female pronotum exhibits a prominent spike-like projection (figures 1c and 2c) [5]. The sex combs and the pronotum projection are both located on the first thoracic segment. Because this segment is under the control of the Hox gene Scr [16,23,24], we tested the role of Scr in the development of these structures. In males, Scr RNAi caused a notable reduction in the size of the teeth forming the comb (figure 2b,i). Interestingly, in females, Scr RNAi also resulted in a reduction in the size and disruption of the shape of the pronotum projection ( figure 2d,j). These results demonstrate that the same Hox gene, Scr, is involved in the development of both male persistence and female resistance traits that are located in its domain of action, i.e. the first thoracic segment.

(b) Ubx shapes another set of male persistence traits found in the rear legs
The rear legs of R. antilleana males are equipped with sets of large and small spines arranged in rows on the trochanter, femur and tibia (figure 2e) [5]. We therefore tested the role of the Hox gene Ultrabithorax, which is known to specify the identity of the third thoracic segment in insects [25][26][27][28][29][30]. Nymphal RNAi knockdown against Ubx resulted in a specific loss or reduction in all the armaments that otherwise develop on male rear legs (figure 2e,f) without any effect on the other segments of the legs (electronic supplementary material, figure S3). Specifically, the width of the femur was significantly reduced and the spines on both femur and tibia were lost or reduced such that the rear leg of the male now resembled that of the female (figure 2e,f,k,l). In females, Ubx RNAi also induced the loss of the small spines in the femur (figure 2g,h,k,l), suggesting that correlation of Ubx expression is associated with the slight modifications of female rear legs. These results indicate that Ubx mediates the development of male persistence traits located in the third thoracic segment in R. antilleana.

Discussion
We have reported that antagonistic coevolution of the sexes, here in Rhagovelia antilleana, can be developmentally controlled by the sex-specific action of the highly pleiotropic Hox genes.
Our results also uncover, for the first time, a case where the same gene, here Scr, controls the development of male persistence (sex comb) and female resistance ( pronotum projection) traits [5]. This indicates an ongoing intra-locus conflict at the Scr locus and suggests that sexual conflict over mating rate can be associated with intra-locus sexual conflict [1,14]. How this opposing role of Scr is mediated between the sexes of R. antilleana is unknown. Although both male and female traits are located in the first thoracic segments, the sex combs being in the leg and the projection being at the posterior portion of the pronotum indicate that these traits develop from distinct cell populations. It is therefore likely that alleles controlling the sex-specific regulation of Scr mediate its dimorphic expression and function, leading to sex-specific antagonistic traits. Ubx, on the other hand, controls the elaboration of male rear legs, which increases male persistence [5]. This suggests an ongoing inter-locus conflict between Ubx and Scr owing to sexually antagonistic coevolution between males and females of this species. Inter-locus conflict is expected to generate dimorphism and fuel arms races because of different loci under divergent selective pressures, while intra-locus conflict is expected to slow the evolution of the sexes because of the shared loci under divergent selective pressures [12,14]. Our results provide a new perspective on antagonistic coevolution because both inter-and intra-locus conflicts at the Scr and Ubx loci participate in sexual dimorphism.
How these evolutionarily constrained genes can shape sexspecific morphologies during development remains unknown. Hox genes are known to control a large number of downstream targets that can diverge greatly among insects [25,31]. It is likely that Ubx and Scr mediate the development of sexually antagonistic traits by interacting with sex-specific targets, as is the case for some fly sex-specific phenotypes [32,33]. The late development of sexually antagonistic traits may have favoured changes in Hox targets to accumulate without any pleiotropic effects, thus favouring rapid changes in Hox function. Other mechanisms could explain the sex-specific function of Hox genes in R. antilleana. Interactions with sex-determination gene isoforms, such as doublesex (dsx) [16,34,35], would allow the generation of sex-specific structures by differentially regulating Hox genes at the cellular level [16,32], which in turn could also differentially regulate dsx in both sexes [16,35]. These interactions and regulations might allow Hox transcription factors to be expressed in different cell populations within the same segment [16,32] and are consistent with sex-specific gene regulation participating in intra-locus sexual conflict [12,36]. Testing these hypotheses should further improve our understanding of the developmental genetic basis of antagonistic coevolution and the underlying role of Hox genes and their sex-specific regulation.
Data accessibility. Sequences of Rhagovelia antilleana Ultrabithorax and Sex combs reduced can be retrieved in GenBank under the following accession numbers: Ultrabithorax: MG999808; Sex combs reduced: MG999826. Data and R script [22] used for the analysis performed in this study are available on the Dryad Digital Repository: http:// dx.doi.org/10.5061/dryad.s76h0s6.