Genome editing reveals fitness effects of a gene for sexual dichromatism in Sulawesian fishes

Abstract

Sexual selection drives rapid phenotypic diversification of mating traits. However, we know little about the causative genes underlying divergence in sexually selected traits. Here, we investigate the genetic basis of male mating trait diversification in the medaka fishes (genus Oryzias) from Sulawesi, Indonesia. Using linkage mapping, transcriptome analysis, and genome editing, we identify csf1 as a causative gene for red pectoral fins that are unique to male Oryzias woworae. A cis-regulatory mutation enables androgen-induced expression of csf1 in male fins. csf1-knockout males have reduced red coloration and require longer for mating, suggesting that coloration can contribute to male reproductive success. Contrary to expectations, non-red males are more attractive to a predatory fish than are red males. Our results demonstrate that integrating genomics with genome editing enables us to identify causative genes underlying sexually selected traits and provides a new avenue for testing theories of sexual selection.

Fig. 1. Diversification of Adrianichthyidae in Sulawesi.

a A map of Sulawesi showing sampling sites of 17 endemic species. The map was provided by Thomas von Rintelen. b A maximum-likelihood phylogenetic tree based on 10,174 single orthologous genes extracted from whole-genome sequencing data. The scale bar indicates the substitution rate. All branches except where noted were supported by 100% maximum-likelihood bootstrap values. Open and closed arrowheads indicate the reference and focal species in this study, respectively. The major lineages of Sulawesi species are shown in different colors.

Fig. 2. Quantitative trait locus (QTL) mapping of body color divergence between O. woworae and O. celebensis.

a Representative images of adult males and females of the parental species. b Logarithm of the odds (LOD) scores plotted against the chromosomal position on each linkage group. For reflectance values, the different colors indicate the relative contributions of a particular spectral range (red, 605–700 nm; yellow, 550–625 nm; blue, 400–510 nm; violet, 300–415 nm) to the total brightness. The dashed lines indicate the genome-wide significance thresholds determined by 1000 permutation tests (α = 0.05). Significant QTLs are summarized in Supplementary Table 2. c, d Effect plots of the significant QTLs for the red pectoral fins. The X axis indicates the genotypes at each QTL: C/C, homozygous for the O. celebensis allele; W/W, homozygous for the O. woworae allele; C/W, heterozygote. Sample sizes are shown in the parenthesis. The Y axis indicates the percentage of fish with red pectoral fins (c) or the contribution of the red spectral range (605–700 nm) to the total brightness (d). d Each dot represents each individual. The blue and brown colors indicate males and females, respectively. In the box-plots, the center line indicates the median, box limits indicate the upper and lower quartiles, the whiskers indicate 1.5× interquartile range, and the points are outliers. Sample sizes are shown in the parentheses. Source data are provided as a Source Data file.

Fig. 3. Increased expression of csf1 in male pectoral fins by cis-regulatory mutations.

a, b RNA-seq of the pectoral fins. Minus-Average (MA) plots with the average gene expression levels on the X axis and log₂ fold-change in expression between the sexes in O. woworae (a) or between O. woworae males and O. asinua males (b) on the Y axis. Each dot represents a single gene with differentially expressed genes (DEGs) colored (FDR < 0.01). The gene names of DEGs within the red pectoral fin QTLs are also shown. c Venn diagram indicating overlap among inter-sex DEGs in O. woworae, interspecies DEGs in males, and genes within 95% Bayesian credible intervals of all red pectoral fin QTLs on LG7. Five genes were identified as strong candidate genes. d Comparison of csf1 expression levels among males and females of O. woworae and O. asinua, showing the highest expression in O. woworae males (n = 4 for each). Mean ± SD are shown. A dot indicates each individual. Expression levels of the other four candidates are shown in Supplementary Fig. 4. e qPCR analysis comparing the relative expression levels of csf1 in the seven different fin parts of O. woworae (red) and O. asinua (blue) males (n = 8 for each). The expression levels were normalized by the expression of an internal control gene, rpl13a. f GFP fluorescence in the pectoral (left) and caudal fins (right) in males (upper) and females (lower) of GFP knock-in O. woworae (csf1^Olhs:GFP). Expression in the lens is caused by the basal activity of the heat-shock promoter. g Microscopic images of pectoral fins in bright-field (left) and GFP fluorescent observation (right). Images are representative of three experiments. Scale bars indicate 200 µm. h Expression levels of csf1 in the pectoral fin of the F₂ family (n = 8 for each). The X axis indicates the genotype at the red pectoral fin QTL on LG7 (C/C, homozygous for the O. celebensis allele; W/W, homozygous for the O. woworae allele; C/W, heterozygote). Different letters above the boxes indicate significantly different groups (P < 0.05, post hoc test with the Tukey’s honest significant difference method; see Supplementary Table 4 for the statistical results). i Allele-specific expression analysis of csf1 in pectoral fins of the heterozygous fish (n = 8 for each). The ratio of the O. woworae allele to the O. celebensis allele was significantly higher in cDNA compared with that in genomic DNA. The P values are calculated separately for each sex using two-sided Wilcoxon’s signed rank test (male, V = 36, P = 0.007812; female, V = 28, P = 0.01562; *P < 0.05 and **P < 0.01). h, i the blue and brown colors indicate males and females, respectively. j Administration of methyltestosterone (MT) to O. woworae females increased csf1 expression in the pectoral fin compared with the control fish administered dimethyl sulfoxide (DMSO) (n = 12 in each group). The P value is calculated using two-sided Welch’s t test (t_12.233 = −5.4148, P = 0.00015, ***P < 0.001). The gray and green colors indicate fish with DMSO and fish with MT, respectively. e, h–j Each circle represents each individual, the center line indicates the median, box limits indicate the upper and lower quartiles, the whiskers indicate 1.5× interquartile range, and the points are outliers. Source data are provided as a Source Data file.

Fig. 4. Role of csf1 in red coloration and fitness components.

a, b Representative images of wild-type (csf1^+/+) (a) and knockout (csf1^−/−) O. woworae males (b). c, d Microscopic images of the pectoral fins of csf1^+/+ (c) and csf1^−/− fish (d). Scale bars indicate 200 µm. Images are representative of three experiments. e Comparison of mating latency between O. woworae with csf1^+/+ and csf1^−/−. The mating behaviors were observed in 46 pairs: 11 pairs of csf1^+/+ female and csf1^+/+ male; 11 pairs of csf1^+/+ female and csf1^−/− male; 12 pairs of csf1^−/− female and csf1^+/+ male; 12 pairs of csf1^−/− female and csf1^−/− male. *P < 0.05 and not significant (NS) according to two-sided post hoc analysis with Tukey’s multiplicity adjustment method in a generalized linear mixed model (GLMM). f Comparison of the behavioral preferences of male halfbeaks (Nomorhamphus cf. ebrardtii) for csf1^+/+ and csf1^−/− O. woworae males (n = 13 halfbeaks). The X axis indicates the csf1 genotype of O. woworae males and the Y axis indicates total time spent in each interaction zone by the halfbeaks during a 20-min behavioral test. The P-value (*P < 0.05) was calculated using the Wald test in a linear mixed model. e, f Each dot represents a single behavioral experiment. g No significant difference was found in the standard length between the csf1^+/+ and csf1^−/− O. woworae adults that were used for the mating behavioral assay (n = 12 for each). Each dot indicates each individual. NS, not significant by two-way ANOVA. h No significant difference in the clutch size was found between the csf1^+/+ and csf1^−/− O. woworae females (n = 9 for csf1^+/+ and n = 8 for csf1^−/−). g, h Each dot represents a single individual. NS, not significant according to the two-sided Mann–Whitney U test. e–h The red and gray colors indicate csf1 wild-type and knockout fish, respectively. In the box-plots, the center line indicates the median, box limits indicate the upper and lower quartiles, the whiskers indicate 1.5× interquartile range, and the points are outliers. Source data are provided as a Source Data file.