5α-Reductase underscores the development of pectoral fin breeding tubercles in zebrafish

Abstract

Breeding tubercles (BTs) are keratinized structures that appear on the pectoral fins of male zebrafish, but the evolutionary patterns and genetic mechanisms driving their development remain unclear. Through a survey of Danioninae species, we find BTs on pectoral fins are conserved within Rasborini and Danionini, but absent in Chedrini and outgroups. This survey also highlights pectoral-fin BTs in cypriniforms as a promising model system for investigating the evolution and development of secondary sexual traits in vertebrates. Comparative transcriptome analyses show BTs-bearing regions are enriched for genes involved in vasculature formation, keratin production, and steroid metabolism. Among these, we identify the localized enrichment of srd5a2b, one of the two zebrafish paralogs of steroid 5-α-reductase type 2, as a key candidate regulator. Experimental inhibition of steroid 5-α-reductase block BTs formation, while treatment with Dihydrotestosterone (DHT) partially restores it, indicating local androgen conversion is essential for BTs development. Cross-species comparisons confirm the conservation of this mechanism among species with pectoral fin BTs. Overall, localized androgen metabolism via the steroid 5-α-reductase pathway appears central to the development and evolution of this secondary sexual trait in zebrafish.

Transcriptomic analyses combined with pharmacological perturbations reveal that localized androgen metabolism through the 5α reductase pathway is essential for the development of breeding tubercles on zebrafish pectoral fins.

Introduction

Courtship and mating behaviors play a crucial role in successful breeding among many sexually reproducing organisms¹. These behaviors frequently correlate with the evolution of male secondary sexual traits (or sexually-selected exaggerated traits), exemplified by the peacock’s ornamental tail, the beetle’s enlarged horns and the deer’s antlers. Although biologists have long been fascinated by these secondary sexual traits, particularly since Darwin proposed sexual selection as a general explanation for their evolutionary origins, our understanding of the genetic/genomic basis of these traits is still limited^2–5.

In vertebrates, the development of these secondary sexual traits is typically associated with high levels of androgens (male hormones) in males but not females^6,7. Administering androgen-like substances can even induce their partial or fully development in females^8–11. Conversely, removing the androgen receptor (AR) gene can lead to the loss of these traits in males, as seen with the cockscomb in chickens¹² or the anal fin papillary processes in medaka fish¹³, further supporting the role of androgen signaling in their development. Despite this, the molecular/genetic mechanisms restricting these traits to specific anatomical locations remain not fully understood. For example, in female medaka, androgen treatment exclusively induces bony papillae on the posterior anal fin rays, but not on other fins or anterior anal fin rays¹⁴. Likewise, in some poeciliid fishes, androgens selectively develop the gonopodium (modified anal fin) and caudal sword in females^15,16.

The model organism zebrafish serves as another compelling model for studying such traits. Successful fertilization in zebrafish involves intricate courtship behaviors, with the male’s pectoral fins playing a crucial role in stimulating egg release through physical contact with the female^17–19. These male zebrafish possess breeding tubercles (BTs) along several central pectoral fin rays²⁰, which are essential for stimulating egg release. Studies have shown that males lacking these BTs have significantly reduced mating success¹⁹. Notably, administering androgen-like substances to female zebrafish can also induce the development of BTs on their pectoral fins, but not on other fin types²¹. This suggests the existence of inherent, sex-independent signaling cues within this specific pectoral fin region. While these cues likely respond to androgens and trigger the entire developmental program for BTs, the exact mechanisms are largely unknown. In addition, we do not know how different types of androgens (e.g., testosterone, dihydrotestosterone etc.) contribute differently to their formation.

On the other hand, the evolutionary history and taxonomic distribution of these pectoral fin BTs have been poorly understood, hindering a comprehensive understanding of this secondary sexual trait in zebrafish. In this study, we conducted a survey of pectoral fin BTs across species closely related to the zebrafish. Our findings suggest that this feature likely originated within a monophyletic group, which includes danionins and rasborins²². To further investigate the genetic basis of this trait, we used comparative transcriptomics and pharmaceutical methods. This approach allowed us to identify potential molecular pathways in the BTs-bearing region of zebrafish pectoral fins that respond to androgens. By integrating these molecular findings with phylogenetic analyses, we propose that the localized expression of a steroid-5α-reductase type 2 gene (srd5a2b) in this specific region may have driven the evolutionary origin of this secondary sexual trait in danionines and rasborines.

Results

Pectoral fin BTs are present in danionins and rasborins

BTs are keratinized epidermal projections found in a wide range of fish species^23,24. Their primary function is to facilitate contact between males and females during spawning, thus they are also called contact organs or nuptial tubercles in different literatures²³. In zebrafish, the pectoral fin BTs in males are the most obvious²⁰, but small tubercles are also found on the lower jaw and operculum¹⁹. However, systematic information on the distribution of pectoral fin BTs across danionins remains limited, although several studies have reported their occurrence in specific taxa such as Betadevario ramachandrani²⁵, two Danio species^26,27 and several Devario species²⁸. To address this knowledge gap, we leveraged the current Danioninae phylogeny^22,29 to map pectoral fin BTs occurrence across 3 major closely-related clades: Chedrini, Rasborini and Danionini (Fig. 1). All six danionin species show well-developed pectoral fin BTs in males but not females. Their morphology differed by species, with round, blunt tubercles in Devario interruptus and Devario kakhienensis, but sharp, conical ones in the other four species (Esomus caudiocellatus, Danio rerio, Danio margaritatus, and Devario chrysotaeniatus) (Fig. 1). We also found that the two rasborin species (Rasbora steineri and Trigonostigma heteromorpha) both have small, darkish BTs on the dorsal side of the males’ pectoral fins (Fig. 1). Notably, all pectoral fin BTs form continuous linear arrays along the fin rays. We only obtained a single Chedrini species, Opsarius pulchellus, which had no pectoral fin BTs in either sex (Fig. 1). Males of this species did, however, exhibit conspicuous BTs on their head and lower jaw, which supported their identification as males in addition to the presence of testis. Recent investigations of BTs in two other Chedrini species also supported their absence on the pectoral fins^30,31.

Fig. 1. Phylogenetic tree of included Danioninae and representative outgroup species based on Tang et al.²².

Images of male and female pectoral fins are placed adjacent to each species. Keratinized protrusions in the form of spines or cap-like structures (i.e., breeding tubercles, BTs) were observed on the pectoral fins of species within both Rasborini and Danionini tribes. In contrast, such structures were absent in the Chedrini tribe, which also belongs to Danioninae. Among more remotely related species, only Phoxinus phoxinus exhibited BT-like keratinized structures, while no BTs were detected in the remaining outgroup taxa. The Rasborini and Danionini lineages are highlighted with red boxes. Red arrowheads indicate BTs. It should be noted that pectoral-fin BTs are also present in many other cyprinid lineages, such as Cyprininae (e.g., carps and minnows). This phylogeny therefore suggests that the trait was either lost in Chedrini and several other cyprinid groups, or that it arose convergently in different lineages. Scale bars, 0.5 mm.

The distribution of pectoral-fin BTs outside Danioninae appears more complex but remains largely uncharacterized. This trait has been reported in numerous Cyprininae species²³, suggesting that it may be ancestral to this clade, although a comprehensive survey is still needed. In our examination of three leuciscine species, pectoral-fin BTs were detected only in Phoxinus phoxinus (Fig. 1; Supplementary Fig. 1). Based on this limited sampling, two evolutionary scenarios remain plausible: (1) pectoral-fin BTs originated earlier in the common ancestor of Cyprininae and Danioninae and were subsequently lost in multiple lineages; or (2) they arose independently in several groups, including the lineage comprising Rasborini and Danionini²².

This phylogenetic framework (Fig. 1) provides a foundation and working hypothesis for dissecting the genetic mechanisms underlying pectoral-fin BT formation in Rasborini and Danionini, using zebrafish as an experimentally trac Data model. Although the evolutionary history of this trait remains unresolved, our survey highlights pectoral-fin BTs as a promising model system for investigating how secondary sexual characters arise and diversify in vertebrates. Their discrete morphology, clear sexual dimorphism, and phylogenetically scattered distribution together offer a rare opportunity to link developmental genetics with macroevolutionary patterns.

Transcriptomic profiling of BTs-bearing vs. non-BTs regions in pectoral fins

To investigate whether the capacity to develop BTs on only a few central fin rays is determined by differential gene expression, we compared transcriptomes between the BTs-bearing and non-BTs regions of the pectoral fin (Fig. 2A), designated pf1 through pf10 (Table 1). These groups represent different biological contexts, including natural state, regeneration, and chemical induction; a comprehensive summary and schematic diagrams of these identifiers are provided in Supplementary Fig. 2C. The four regeneration groups were included to identify a mechanism that reflects a stable property of the fin’s positional identity rather than a transient developmental event. In addition, these data may also help validate the robustness of our transcriptome‑based screening approach.

Fig. 2. Differential gene expression between BTs-bearing and non-BTs regions and in situ hybridization of srd5a2b in zebrafish pectoral fins.

A The zebrafish pectoral fin was anatomically divided into the BTs-bearing region and the non-BTs region for downstream analyses. B Venn diagrams summarizing differentially expressed genes (DEGs) identified from five pairwise comparisons: pf1 vs pf2, pf3 vs pf4, pf5 vs pf6, pf7 vs pf8, and pf9 vs pf10.The left diagram shows genes consistently up-regulated in the BTs-bearing region, while the right diagram shows genes consistently down-regulated in the BTs-bearing region across all five comparisons. DEGs were defined using the criteria |log₂ fold change| > 1 and adjusted p < 0.05. C Volcano plots showing DEGs in two representative comparisons: pf1 vs pf2 and pf3 vs pf4.Labeled genes were selected from DEGs shared across all five pairwise comparisons, as identified in panel (B). D srd5a2b expression levels in pf1–pf4 tissues, based on RNA-seq and qPCR. Consistent trends were observed between the two methods. Data are presented as mean ± SEM. Statistical analysis was performed using unpaired two-tailed Student’s t test. Exact p-values are indicated. E Expression of srd5a2b in the zebrafish pectoral fin revealed by HCR-FISH. Left, dorsal view of a male pectoral fin showing srd5a2b expression (red); the white dashed line indicates the position of the transverse section. Right, transverse section showing srd5a2b expression, with dashed outlines marking the BTs and underlying bone. White arrowheads indicate BTs. Scale bars, 250 μm. Source data for this figure are provided as a Supplementary Data 4 file.

Table 1.

Summary of experimental groups and sample information for transcriptome analysis

Group ID	Label name	Physiological/experimental contextual description
pf1	Male- Proximal (BT+)	Proximal region of adult male pectoral fin (BT-bearing region)
pf2	Male- Distal (BT−)	Distal region of adult male pectoral fin (Non-BT bearing region)
pf3	Female- Proximal (BT−)	Proximal region of adult female pectoral fin (BT bearing region, competent)
pf4	Female- Distal (BT−)	Distal region of adult female pectoral fin (Non-BT bearing region)
pf5	Regen-Male- Proximal (BT+)	Proximal region of regenerating male pectoral fin (BT-bearing region)
pf6	Regen-Male- Distal (BT−)	Distal region of regenerating male pectoral fin (Non-BT bearing region)
pf7	Regen-Female- Proximal (BT−)	Proximal region of regenerating female pectoral fin (BT-bearing region, competent)
pf8	Regen-Female- Distal (BT−)	Distal region of regenerating female pectoral fin (Non-BT bearing region)
pf9	ETH-Female- Proximal (BT+)	Proximal region of female pectoral fin following Ethisterone (ETH) induction (BT-bearing region)
pf10	ETH-Female- Distal (BT−)	Distal region of female pectoral fin following Ethisterone (ETH) induction (Non-BT bearing region)

We first analyzed adult male pectoral fins with naturally developed BTs, designating the BTs-bearing proximal region as pf1 (Male-Proximal (BT+)) and the non-BTs distal region as pf2 (Male-Distal (BT−)) (Fig. 1, Supplementary Fig. 2A–C). Since androgen-like substances can induce BTs in female pectoral fins within the same region as males, we divided female fins into BTs-bearing proximal (pf3, Female-Proximal (BT−)) and non-BTs distal (pf4, Female-Distal (BT−)) regions based on this pattern (Supplementary Fig. 2C). In addition, we amputated pectoral fin rays in both sexes at the same proximal position to ensure complete BT removal in males. Once BTs had regenerated in the male fins, we collected samples from the BTs-bearing (pf5, Regen-Male-Proximal (BT+)) and non-BTs (pf6, Regen-Male-Distal (BT−)) regions of regenerates (Supplementary Fig. 2C). We also sampled the regenerated pectoral fins from females at the same stage, with pf7 (Regen-Female-Proximal (BT−)) and pf8 (Regen-Female-Distal (BT−)) corresponding approximately to pf5 and pf6 in males (Supplementary Fig. 2C). Lastly, to capture the induction phase, we treated females with Ethisterone to induce BTs¹⁹, collecting samples from the BTs-bearing (pf9, ETH-Female-Distal (BT+)) and non-BTs (pf10, ETH-Female-Proximal (BT-)) regions after 14 days (Supplementary Fig. 2C).

We performed bulk RNA-seq on 30 transcriptomes, with three biological replicates for each of the 10 sample types (pf1–pf10). Differential gene expression analysis was conducted on five paired groups: pf1 vs. pf2, pf3 vs. pf4, pf5 vs. pf6, pf7 vs. pf8, and pf9 vs. pf10. Differentially expressed genes (DEGs) were identified using a log2 (fold-change) threshold of >1 or <−1 and a padj < 0.05 (Supplementary Fig. 2D, Supplementary Data 1). Among these comparisons, the pf1/pf2 group exhibited the highest number of DEGs, with 1536 genes upregulated in pf1 and 765 in pf2. Conversely, the pf3/pf4 group showed the fewest DEGs, with 75 genes upregulated in pf3 and 68 in pf4 (Supplementary Fig. 2D, Supplementary Data 1). Interestingly, the Ethisterone-treated female group (pf9/pf10) exhibited a considerably lower number of DEGs compared to the pf1/pf2 group. Only 231 genes were upregulated in pf9 and 114 in pf10 (Supplementary Fig. 2D, Supplementary Data 1). Despite this difference in overall DEG count, we observed a notable overlap: 136 of the genes upregulated in pf9 were also found in the pf1 upregulated gene list, and 92 of the genes upregulated in pf10 were present in the pf2 upregulated gene list (Fig. 2B, Supplementary Data 1). This suggested that a core subset of transcriptional changes is conserved between the pf1/pf2 and pf9/pf10 groups, despite the disparity in the total number of differentially expressed genes. This shared transcriptional signature underscores the common biological processes underlying BT development across different conditions.

To assess the quality of our transcriptome data, we also analyzed the DEGs in the pf9 vs. pf3 and pf1 vs. pf3 comparisons, expecting these comparisons to be rich in genes related to the formation of BTs. We found 1659 upregulated genes in pf1, significantly more than the 779 genes upregulated in pf9 (Supplementary Fig. 3A). This difference highlights the inherent physiological distinctions in gene expression between male and female pectoral fins. By identifying the genes that were upregulated in both pf1 and pf9, we found a set of 434 shared DEGs (Supplementary Fig. 3B). Functional analysis of these genes showed a strong enrichment in pathways associated with erythrocyte homeostasis, TP53-mediated transcriptional regulation, keratinization, and steroid metabolism (Supplementary Fig. 3C, D). The direct relevance of keratinization and steroid metabolism to the development of BTs is well-established. Additionally, the enrichment of genes involved in vascularization and TP53 signaling provides further support to the biological relevance of our transcriptomic findings, as previous studies have confirmed the importance of these processes in BTs formation^21,32.

To identify potential local molecular cues triggering BT formation in males and in Ethisterone-treated females, we analyzed shared DEGs across five BTs-bearing regions (pf1, pf3, pf5, pf7, pf9) compared to non-BTs regions (pf2, pf4, pf6, pf8, pf10). This analysis revealed 19 genes with significantly higher expression and 27 with lower expression in all BTs-bearing regions compared to non-BTs regions (Fig. 2B, C, Supplementary Fig. 4A, Supplementary Data 1). Through functional enrichment analysis of the 19 genes, we found the vascular network development pathway (Supplementary Fig. 4B), which aligned with the observed increase in vascular networks in the pectoral fins of dihydrotestosterone (DHT)-treated females (Supplementary Fig. 5)²¹. This suggests that the regions where BTs develop are characterized by a specialized molecular signature with pro-angiogenic potential. This localized transcriptional environment likely provides a permissive context for the vascular remodeling that accompanies BT formation, which may be further augmented or recruited by androgen signaling.

5α-reductase inhibitors suppress BT formation in male zebrafish

Among the 19 genes consistently upregulated in all BTs-bearing regions, srd5a2b emerged as the most significant in both pf1/pf2 and pf3/pf4 comparisons (Fig. 2B, Supplementary Data 2). qPCR validation confirmed significantly higher srd5a2b expression in pf1 and pf3 versus their corresponding non-BTs regions (pf1/pf2: p = 0.01827; pf3/pf4: p = 0.0000066; Fig. 2D). Hybridization Chain Reaction Fluorescence In Situ Hybridization (HCR-FISH) performed on adult male pectoral fins localized srd5a2b expression predominantly along the fin rays with BTs (Fig. 2E). Transverse sections revealed that the srd5a2b expressing cells are positioned right below the BTs and lined externally to the bony fin rays (Fig. 2E). To determine whether this spatial pattern is conserved across sexes, as suggested by the transcriptome analysis above, we further examined srd5a2b expression in female pectoral fins. HCR-FISH detected srd5a2b expression in locations where BTs can be induced by 11‑KT treatment, although 11‑KT appears to drive a more robust expression pattern (Supplementary Fig. 6). Srd5a2 encodes steroid 5α-reductase 2, an enzyme responsible for converting testosterone (T) to dihydrotestosterone (DHT)³³. Zebrafish has two copies as srd5a2a and srd5a2b due to teleost genome duplication. DHT is a more potent androgen (through binding AR) than T and is essential for developing male external genitalia, prostate, and facial/body hair in humans³⁴. Interestingly, DHT is generally considered a paracrine hormone that is synthesized and acts primarily in target tissues, rather than functioning as a circulating hormone^35,36. This localized action makes srd5a2b a strong candidate for mediating the development of region-specific sexually dimorphic traits.

To investigate the role of 5α-reductase in zebrafish BTs formation, we treated adult males with finasteride (FIN) at two concentrations (15 μM and 20 μM) (Supplementary Fig. 7A). FIN is a clinically approved 5α-reductase inhibitor, widely used to treat androgen-dependent conditions such as male pattern baldness and benign prostatic hyperplasia³⁷. While FIN’s effects in mammals are well-characterized, its impact on fish androgen signaling remains largely unexplored³⁸. After 28 days of treatment, we observed that over 50% of specimens exhibited phenotypes of BT regression (Supplementary Fig. 7B–E). To quantify these effects, we defined “mild regression” as a reduction in tubercle size or number while some structures remain visible, and “severe regression” as the complete or near-complete loss of detectable tubercles. Specifically, in the 20 μM treatment group, 5 out of 14 fish showed severe regression, characterized by barely detectable BTs (Supplementary Fig. 7E), while the remaining affected individuals exhibited mild regression (Supplementary Fig. 7D). To the contrast, none of the control group specimens showed any detectable reduction of BTs (Supplementary Fig. 7B). To further validate this result, we treated adult males with dutasteride (DUT), a more potent 5α-reductase inhibitor (Fig. 3A). At a concentration of 0.4 μM, DUT induced clear regression of BTs in 6 out of 8 specimens after 21 days, whereas the DMSO control group maintained normal BTs morphology without detectable changes (Fig. 3B). To identify the gene expression changes underlying the reduction of BTs following DUT treatment, we performed RNA-seq on pectoral fins from DUT- and DMSO-treated groups and detected DEGs between them (Supplementary Fig. 7F, G). This analysis revealed 68 downregulated and 21 upregulated genes in DUT-treated versus DMSO-treated samples (Fig. 3F, Supplementary Data 1). Functional enrichment analysis of downregulated genes identified functional terms directly linked to DUT treatment or BT formation, including “steroid metabolism” and “cornified envelope assembly” (Supplementary Fig. 7H). Among these, several genes are known associated with keratinocytes, such as transglutaminases (tgm1l2–4), keratins (krt94/95), and sulfotransferases (sult2st3)^39,40. qPCR validation confirmed downregulation of sult2st3, tgm1l2–4, krt95, and klf5b in the DUT-treated group (Fig. 3F). Collectively, these results demonstrated that 5α-reductase inhibitors (FIN/DUT) can suppress pectoral fin BTs formation, likely through downregulation of genes critical for keratinocyte development and metabolism.

Fig. 3. Effects of 5α-reductase inhibition on BTs development and gene expression in male zebrafish following DUT exposure.

A Schematic illustration of androgen metabolism and experimental design. Testosterone is converted to dihydrotestosterone (DHT) via 5α-reductase, a process that can be inhibited by finasteride (FIN) or dutasteride (DUT). Male zebrafish were exposed to FIN- or DUT-containing water for 21 days to assess the effects of 5α-reductase inhibition on BTs. B Quantification of BTs regression severity following DUT exposure. Male zebrafish were treated with DUT, and pectoral fins were classified into three categories based on the degree of BTs regression: normal, mild regression (partial reduction), and severe regression (near-complete loss). The percentage of individuals in each category is shown. C–E Representative images showing the three BTs phenotypic categories following DUT treatment: C normal, D mild regression, and E severe regression. Red arrowheads indicate BTs. Scale bars, 0.5 mm. F Left, heatmap of DEGs identified by RNA-seq analysis of pectoral fins from DMSO-treated and DUT-treated groups. DEGs were defined using the criteria |log₂ fold change| > 1 and adjusted p < 0.05; Right, qPCR validation of selected DEGs, showing expression patterns consistent with the transcriptomic results. Data are presented as mean ± SEM. Statistical significance was assessed using an unpaired two-tailed Student’s t test, and exact p-values are indicated. Phenotypic frequencies (n/N) are indicated in the lower-left corners of representative panels. Source data for this figure are provided as a Supplementary Data 4 file.

5α-reductase inhibition altered male androgen levels

Previous studies have revealed that teleosts primarily utilize 11-ketotestosterone (11KT) as the functional ligand for AR, whereas humans rely on T or DHT as the dominant androgens^41,42. On the other hand, the roles of Srd5a2 and DHT in fish and amphibians remain largely unexplored⁴³. To assess the impact of 5α-reductase inhibition on zebrafish androgen levels, we first characterized its role within the broader steroidogenic pathways based on previous studies^44,45. This revealed that Srd5a2 is likely involved in the conversion of T to DHT, 11-hydroxytestosterone (11OH-T) to 11-hydroxydihydrotestosterone (11OH-DHT), and 11KT to 11-ketodihydrotestosterone (11K-DHT) (Fig. 4A).

Fig. 4. Summarized steroidogenic pathways and the effects of dutasteride on androgen levels and steroidogenic gene expressions in male zebrafish.

A Proposed steroidogenesis pathway in teleost fish. Enzymes hypothesized to mediate each step are indicated above the arrows. The androgen androstenedione (A4) is converted to testosterone (T) via Hsd17b3. A fraction of A4 and T undergo 11β-hydroxylation catalyzed by Cyp11c1. These hydroxylated steroids are further oxidized to their 11-keto forms, 11KA4 and 11KT, by Hsd11b2. Additionally, 11KA4 can be converted to 11KT through Hsd17b3 activity. The synthesis of dihydrotestosterone (DHT), 11OHDHT, and 11KDHT is dependent on the enzymatic activity of 5α-reductase. Serum concentrations of B 11-keto-testosterone (11-KT), C testosterone (T), and D dihydrotestosterone (DHT) in DMSO- and DUT-treated male zebrafish. E–G qPCR analysis of gene expression in testes from DMSO-treated and DUT-treated male zebrafish. E Expression levels of steroidogenic enzyme genes cyp11c1 and cyp17a1. F Expression levels of hydroxysteroid dehydrogenases hsd11b2 and hsd17b3. G Expression levels of steroid 5α-reductase genes srd5a2a and srd5a2b. Data are presented as mean ± SEM. Statistical analysis was performed using unpaired two-tailed Student’s t test. Exact p-values are indicated. Source data for this figure are provided as a Supplementary Data 4 file.

After 21 days of DUT or DMSO treatment in male zebrafish, we attempted to quantify 11KT, T and DHT in pectoral fin homogenates using the enzyme-linked immunosorbent assay (ELISA). However, no detectable levels of these steroids were found in either group, likely due to the technical limitations of ELISA in low-concentration tissue samples. We therefore analyzed blood samples and measured peripheral concentrations of these hormones. Compared to the DMSO control group, the DUT-treated group showed significantly higher circulating levels of 11KT and T but lower DHT (Fig. 4B–D). The elevated circulating T and 11KT levels observed in DUT-treated fish could likely resulted from a synergistic mechanism: the passive accumulation of precursor substrates due to the pharmacological blockade of 5α-reduction, and a robust compensatory response of the steroidogenic machinery (Fig. 4A). With our current approach, we cannot determine the extent to which the former contributes to the observed increase. Supporting the latter, our qPCR analysis revealed a broad upregulation of key enzymes, including cyp11c1, cyp17a1, hsd11b2, and hsd17b3, within the testes (Fig. 4E, F). This finding aligns with previous studies in zebrafish, where disabling cyp11c1 or cyp17a1 resulted in lower T and 11KT levels^46,47. Similarly, inactivating hsd11b2 has been shown to cause infertility in male zebrafish⁴⁸. Notably, the significant transcriptional induction of srd5a2a, but not srd5a2b, further suggests a targeted feedback mechanism attempting to restore 5α-reductase activity (Fig. 4G).

The Srd5a2 enzyme and its products mediate pectoral fin BTs development

The induction of pectoral fin BTs in female zebrafish following T and DHT treatments is well-established²¹. Our study extended this understanding by demonstrating that 11KT treatment at 0.5 μM for 14 days similarly stimulated BTs formation in female zebrafish (Supplementary Fig. 8B). However, we observed that inhibitor-treated males exhibited reduced BTs despite possessing elevated circulating levels of T and 11KT (Fig. 4B, C). This apparent paradox suggested that T and 11KT may not directly mediate BT formation but rather serve as substrates for the Srd5a2 enzyme to synthesize more potent downstream effectors.

To discern which specific androgen may directly mediates BT development, we conducted a series of rescue-like experiments in female zebrafish using T, DHT, and 11KT in combination with the 5α-reductase inhibitor FIN. We utilized FIN (20 μM) to analyze both morphological and molecular responses. In control groups without the inhibitor, 11KT, DHT, and T, all successfully induced BT formation. All samples in the 11KT group developed well‑formed BTs, whereas several samples in the T and DHT groups exhibited less‑developed structures (Supplementary Fig. 8A–D). The addition of FIN in combination with 11 KT, DHT, or T reduced BT prominence across all groups, with the T + DHT group showing the least severe regression by the percentage of samples (Supplementary Fig. 8B, D). This pattern was corroborated by qPCR analysis of keratinocyte markers (krt95 and tgm1l3), which were significantly downregulated in the T + FIN and 11 KT + FIN groups, whereas the DHT + FIN group maintained relatively stable expression levels (Supplementary Fig. 8E). These results indicate that DHT can at least partially bypass 5α‑reductase inhibition, supporting the idea that 5α‑reduced androgens act as the primary functional effectors.

However, 11KT is the predominant androgen in teleosts, and its conversion to 11K‑DHT is likely the most potent androgenic signal driving BTs formation. This may partially explain the stronger induction effect of 11KT compared with DHT and T. Because 11K‑DHT is not currently available to us, we were unable to directly test its function. Little is known about this conversion in vivo, and clarifying this pathway will be an important direction for future investigation.

Differential srd5a2b expression may underscore the origin of pectoral fin BTs

Our initial transcriptome analysis identified srd5a2b as a potential candidate gene for the local development of BTs, which showed higher expression in BTs-bearing regions compared to non-BTs-bearing regions in both male and female zebrafish. Given the challenges of performing a localized inhibition of the Srd5a2 enzyme, we opted to study other closely related species that also develop BTs on their male pectoral fins (Fig. 1). We aligned the srd5a2b gene sequences of zebrafish and R. ocellatus, and designed qPCR primers targeting their conserved regions for analysis in six species: R. steineri, D. interruptus, D. kakhienensis, E. caudiocellatus, D. margaritatus, and D. chrysotaeniatus. In all these taxa possessing pectoral fin BTs, srd5a2b expression was significantly higher in BTs-bearing regions than in non-BTs regions (pf1 > pf2 in males, pf3 > pf4 in females, Fig. 5A, B). To the contrast, this srd5a2b expression pattern was reversed in the two outgroup species (R. ocellatus and O. pulchellus) without pectoral fin BTs, suggesting that this expression pattern originated in Rasborini and Danionini (Fig. 1).

Fig. 5. Differential expression of srd5a2b in pectoral fins of selected fish species and its association with drug-induced changes in BTs.

A, B Relative expression levels of srd5a2b in the pectoral fins of multiple fish species based on qPCR. Note that these expression levels should only be compared within each species. A Comparison between pf1 and pf2 groups. B Comparison between pf3 and pf4 groups. Red boxes highlight B. pulchellus and R. ocellatus, which have no pectoral fin BTs and exhibit distinct srd5a2b expression patterns compared to other species. C–G Schematic diagrams of chemical treatment experiments for E. caudiocellatus and R. ocellatus, respectively. C, D Morphological changes of pectoral fin BTs following treatment. DUT could also inhibit BTs formation in E. caudiocellatus. E Quantification of BTs phenotypic severity under treatment conditions. F, G 11KT could not trigger BTs formation in R. ocellatus. Red arrowheads indicate BTs. Scale bars, 0.5 mm. Phenotypic frequencies (n/N) are indicated in the lower-left corners of representative panels. Source data for this figure are provided as a Supplementary Data 4 file.

Finally, we treated E. caudiocellatus (a species with pectoral fin BTs) with DUT (0.4 μM for 21 days), which repressed BT formation in all samples (10/10; Fig. 5C–E). In contrast, 11KT treatment (0.5 μM for 14 days) failed to induce pectoral fin BTs in either male or female R. ocellatus, supporting that this trait is a fixed secondary sexual characteristic in Rasborini and Danionini (Fig. 5F, G). However, our taxonomic survey of this trait remains limited in this fish group, and future studies are needed to confirm the distribution pattern.

Discussions

Pectoral-fin BTs as a model for studying the evolution of secondary sexual traits

Our comparative analysis within Danioninae shows that pectoral-fin BTs are male-specific in Danionini and Rasborini, but absent in Chedrini. Similar structures have been documented across many cyprinids and loaches²³, suggesting that pectoral-fin BTs may have originated in a common cypriniform ancestor⁴⁹ and were subsequently lost in multiple lineages. Thus, the trait could represent either an innovation restricted to Danionini and Rasborini or a secondary loss in Chedrini. Notably, the precise position of pectoral-fin BTs varies among cyprinids—for example, they are concentrated on the first pectoral-fin ray in Carassius and Cyprinus^23,50—highlighting the need for a broader survey of their morphology and distribution to resolve their evolutionary history. Despite these uncertainties, the striking morphological diversity and phylogenetic patchiness of pectoral-fin BTs underscore their potential as a powerful model system for uncovering how genetic and genomic changes drive the evolution of secondary sexual traits in vertebrates.

Our findings also establish another model system for investigating the origin of sexual dimorphism and evolutionary innovation, complementing the well-studied poeciliid fishes (e.g., Xiphophorus and Gambusia) known for their swordtails and gonopodial anal fins^15,16,51. The precise and localized nature of these BTs—forming linear arrays on specific fin rays—prompts a fundamental question: what molecular mechanisms restrict the formation of this androgen-dependent trait to a specific anatomical region?

Localized Srd5a2b activity potentially mediates BTs formation in zebrafish

Androgen signaling, primarily through the AR, is the main driver of secondary sexual characteristic development in vertebrates⁵². However, different androgen hormones have distinct functions and are not always interchangeable. In humans, testosterone (T) and dihydrotestosterone (DHT) are the two primary androgens. The conversion of T to DHT through the Srd5a2 enzyme in specific tissues, such as the prostate gland and external genitalia, is essential for their proper development^36,38. This conversion is critical because DHT has a higher binding affinity and bioactivity potency for the AR than T. Therefore, this localized expression pattern of Srd5a2 provides a potential mechanism for the development of secondary sexual traits in different anatomical locations.

Indeed, our study identified srd5a2b as a potential localizing factor of BTs in the pectoral fins of zebrafish. Through a comparative analysis of the genes expressed in both BTs-bearing and non-BTs regions, we consistently found that srd5a2b expression was significantly higher in the regions with BTs, in both male and female fish. The inclusion of regenerated fin groups (pf5–8) provides key evidence for the stability of positional control in BT development. Regenerated male fins (pf5) closely recapitulated the transcriptional profile of intact fins (pf1), with srd5a2b consistently exhibiting a proximally restricted expression pattern. This demonstrates that the cues directing srd5a2b expression are not transient developmental artifacts but are embedded within the fin’s positional memory, remaining intact after complete amputation and regrowth. The faithful re‑establishment of this pattern during regeneration indicates that the spatial restriction of srd5a2b is an inherent property of the fin’s positional identity, reinforcing its role as a localizing factor that ensures secondary sexual traits are restored to their precise anatomical positions during tissue repair.

Notably, our HCR-FISH analysis localized srd5a2b expression specifically to the innermost epithelial layer directly attached to the fin rays. Although this expression domain is spatially more restricted than the entire BT‑forming region, its localized enrichment suggests that it functions as a local enzymatic “source” for producing DHT or 11K‑DHT. As lipophilic steroids, these androgens synthesized in basal epidermal cells can readily diffuse into adjacent epithelial layers, thereby activating AR signaling across a broader spatial range. It is also important to note that the precise BT‑forming region varies among individuals, and our HCR data cannot be used to infer that BTs will form exclusively where srd5a2b is expressed. Thus, the correspondence between expression and BT position remains qualitative rather than spatially exact.

We further demonstrated that inhibiting Srd5a2 with FIN or DUT effectively suppressed BTs formation in male zebrafish, even when peripheral levels of T and 11KT were elevated, confirming the essential role of Srd5a2 in the formation of BTs. The observation that DHT-treated females with Srd5a2 inhibition still developed BTs, while T-treated females do not, further reinforces the central role of DHT in this process. On the other hand, DUT could attenuate the inductive effects of both 11KT and DHT on BTs in females, indicating that 11K-DHT may be the primary androgen responsible for BTs development, warranting further investigation.

Our findings challenge the idea that T or 11KT are the sole drivers of pectoral fin BTs formation in zebrafish. Instead, our data suggests that the enzyme Srd5a2, which is localized in the mesenchymal tissues of the pectoral fin, converts T and possibly 11KT into the more potent androgens, DHT and 11K-DHT. This localized synthesis of a more potent androgen likely serves as the critical trigger for the developmental cascade that leads to BT formation.

The effects we observed in zebrafish mirror previous studies in other species. For instance, DHT treatment has been shown to induce secondary sexual characteristics in female fathead minnows⁵³, while administration of DUT reduced nuptial tubercles in male fathead minnows⁵⁴. These effects are similar to what we observed in zebrafish treated with DHT and FIN/DUT, though the secondary sexual characteristics manifest as nuptial tubercles on the head of fathead minnows rather than BTs on the pectoral fins of zebrafish. This suggests that similar mechanisms may have evolved in different cypriniform species to form these keratinocyte-related secondary sexual traits in various anatomical locations. In contrast, exposure to finasteride does not elicit any effects on the papillary processes (secondary sexual characteristics) of the anal fins in medaka⁵⁵, suggesting a different signal is involved in the formation of secondary sexual characteristics. Nevertheless, as Srd5a2 primarily acts locally through paracrine or intracrine mechanisms³⁵, it could be a key candidate for explaining the localized development of secondary sexual traits in vertebrates. The evolutionary recruitment of srd5a2b into the pectoral fin BT program likely required the emergence of regulatory inputs. We hypothesize that lineage‑specific cis‑regulatory elements (CREs) or enhancers near the srd5a2b locus evolved to respond to pre‑existing positional cues in the pectoral fin, thereby driving its localized expression. Comparative genomic analyses aimed at identifying conserved non‑coding elements (CNEs) across Danioninae species, together with functional assays such as ATAC‑seq or transgenic reporter lines, will be essential for pinpointing the molecular “switch” that enabled this androgen‑dependent trait. Nevertheless, Srd5a2 expression alone is insufficient for BT formation, as our DHT‑treatment experiments induce BTs only in the pectoral fins and not in other fins. Further investigation in this area will help clarify the molecular and genetic basis of tissue heterogeneity and deepen our understanding of the evolvability of secondary sexual traits.

In this study, the functional requirement for Srd5a2 was assessed through systemic pharmacological inhibition. While this approach effectively demonstrates the acute necessity of 5‑α‑reduction for breeding tubercle (BT) maintenance in adults, we acknowledge that pharmacological manipulation has inherent limitations, including potential systemic or off‑target effects. Future work should incorporate CRISPR/Cas9‑mediated knockouts to investigate the roles of 5α‑reductases in zebrafish and other teleosts, an area that has not yet been systematically explored³⁸. Given that zebrafish possess two srd5a2 paralogs and one srd5a1 gene, strong compensatory mechanisms are likely. It may therefore be necessary to disrupt all three genes to overcome compensation and fully elucidate the physiological and developmental functions of 5α‑reductases in teleosts.

Materials and methods

Animals

A total of 15 fish species were utilized in this study for comparative analysis. These included zebrafish (Danio rerio, AB strain) and 1AP4 other teleost species: Danio margaritatus, Devario kakhienensis, Devario interruptus, Devario chrysotaeniatus, Esomus caudiocellatus, Rasbora steineri, Rasbora heteromorpha, Opsarius pulchellus, Rhodeus ocellatus, Pseudorasbora parva, Phoxinus phoxinus, Gyrinocheilus aymonieri, Makunaima pittieri, and Oryzias sinensis. All non-zebrafish specimens were obtained from local commercial aquarium suppliers in Xiamen, China. All fish were maintained in automated recirculating freshwater systems with a 14 h light/10 h dark cycle and fed three times daily. Most species were housed at a standard temperature of 28.5 °C. Notably, the cold-water species Phoxinus phoxinus was maintained in a dedicated cooling-controlled system at 16 °C to ensure physiological health. All animal procedures were conducted in accordance with the institutional guidelines and were approved by the Institutional Animal Care and Use Committee (IACUC) of Xiamen University. We have complied with all relevant ethical regulations for animal use.

Anesthesia and surgical procedures

For all surgical procedures and terminal experiments, zebrafish were anesthetized via immersion in a buffered Tricaine methane sulfonate solution (MS-222, Sigma-Aldrich). A stock solution was prepared by dissolving 400 mg of Tricaine powder in 97.9 mL of double-distilled water, buffered with 2.1 mL of 1 M Tris (pH 9.0) to achieve a stable pH of 7.0. The working anesthesia solution was prepared by diluting the stock at a ratio of 3:50 in system water (final concentration: ~226 mg/L).

Anesthetic depth was monitored by observing the loss of uprighting reflex, slowed opercular movement, and lack of response to tactile stimuli (tail pinch). Throughout the procedures, fish were kept moist and handled with care to minimize stress. No surgical procedures were performed until a deep plane of anesthesia was confirmed. After surgery, fish were transferred to fresh system water for recovery and closely monitored for the resumption of normal swimming behavior.

For the treatment, 4-month-old healthy zebrafish (both males and females) were randomly selected and allocated to treatment or control groups. The experimental unit was an individual fish. To minimize confounders, all fish were handled by the same experimenter under consistent environmental conditions. No animals were excluded from the analysis.

Sample sizes were not predetermined by statistical methods but were based on our previous experience and established protocols in zebrafish developmental studies. Due to random allocation and natural variation in sex ratios, the number of males and females in each group was not identical; however, all available individuals were included in the analysis.

RNA-seq analysis

Pectoral fins of anesthetized zebrafish were dissected according to the sampling scheme illustrated in Fig. 2, Supplementary Fig. 1. Pectoral fins of anesthetized zebrafish were excised following the sampling scheme outlined in Fig. 2, Supplementary Fig. 1B, C and subdivided into ten anatomical regions (pf1–pf10). In addition, whole pectoral fins from the DUT and DMSO treatment groups were collected. Each of the twelve sample groups included three biological replicates. Following dissection, fin tissues were rinsed briefly with 1× PBS prepared in DEPC-treated water and immediately immersed in RNAlater™ Stabilization Solution (Invitrogen). Samples were kept at −20 °C and transported to Novogene Co., Ltd. for eukaryotic transcriptome library preparation and sequencing.

Total RNA was extracted following standard protocols using TRizol™ (Invitrogen). RNA integrity and concentration were assessed using the Agilent 2100 Bioanalyzer in conjunction with the RNA Nano 6000 Assay Kit (Agilent Technologies, CA, USA), ensuring that the RNA samples met the quality requirements for library preparation.

Poly(A)+ mRNA was enriched from total RNA using magnetic beads conjugated with oligo(dT). The purified mRNA was then fragmented under high-temperature conditions in a divalent cation-containing buffer. These fragments served as templates for first-strand cDNA synthesis using M-MuLV reverse transcriptase (RNaseH-free). Second-strand cDNA synthesis was subsequently performed using DNA Polymerase I, dNTPs, and followed by RNase H incubation to remove the RNA template. The resulting double-stranded cDNA underwent end repair, 3’ adenylation, and adapter ligation using Illumina sequencing adapters. AMPure XP beads (Beckman Coulter, Beverly, USA) were used for fragment size selection to obtain library fragments ~370–420 bp in length (including adapters). The size-selected libraries were amplified by PCR and purified again using AMPure XP beads. The preliminary concentration of each library was measured using a Qubit 2.0 fluorometer. The insert size was verified using the Agilent 2100 Bioanalyzer. The effective concentration of the library was further quantified by qRT-PCR. Only libraries with an effective concentration above 2 nM were used for sequencing.

Qualified libraries were pooled according to effective concentrations and loaded onto the Illumina NovaSeq 6000 platform. Sequencing was performed using the Illumina Sequencing-by-Synthesis (SBS) technology in paired-end 150 bp (PE150) mode.

RNA-seq data were analyzed using the community-curated workflow nf-core/rnaseq (v3.18.0)⁵⁶, which provides a standardized and reproducible pipeline for quality control, alignment, and quantification of transcriptomic data Specifically, raw read quality was assessed with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/), followed by adapter trimming and low-quality base removal using Trim Galore (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). Cleaned RNA-seq reads were aligned to the zebrafish reference genome (Ensembl GRCz11) using STAR⁵⁷, followed by transcript quantification with Salmon⁵⁸ based on the Ensembl annotation file (GTF: Danio_rerio.GRCz11.114.gtf), yielding TPM values and expected counts.

Differential gene expression analysis was conducted using the community-curated workflow nf-core/differentialabundance pipeline (v1.4.0)⁵⁶. For RNA-seq data, DESeq2⁵⁹ was used by default to fit a negative binomial generalized linear model, estimating statistical significance (adjusted p-value) and effect size (log2 fold change) of gene expression differences. The p values were adjusted using the Benjamini & Hochberg method. adjusted p-value (padj) ≤0.05 and absolute log2(foldchange) ≥1 were set as the thresholds for significantly differential expression justification.

To investigate the potential association between differentially expressed genes (DEGs), functional annotation of DEGs was performed using the Metascape database⁶⁰. Pathways with a -log10(p-value) >2 were considered to be more relevant to BTs formation. To identify genes that were consistently upregulated across multiple treatment groups, a Venn diagram analysis was conducted using TBtools-II⁶¹. Gene expression heatmaps and volcano plots were generated using the OmicStudio platform⁶² for visualization of DEGs.

Hybridization chain reaction fluorescence in situ hybridization (HCR-FISH)

The HCR-FISH protocol was adapted from Choi et al.⁶³. Probe sets (Part A and Part B) targeting the srd5a2b transcript (ENSDART00000148402.2) and DNA fluorescent hairpins (H1 and H2) were synthesized by Sangon Biotech (Shanghai) Co., Ltd. The probes carried initiator sequences complementary to the fluorescent hairpins, which trigger a hybridization chain reaction to amplify the fluorescence signal. The sequences of the probes and hairpins are listed in Supplementary Data 3.

Adult zebrafish pectoral fins were dissected in cold 1× PBS and grouped accordingly. The samples were fixed overnight at 4 °C in a fixative containing 4% paraformaldehyde (PFA) and 0.1% Triton X-100 in 1× PBS. After fixation, samples were washed three times with 1× PBS containing 0.1% Triton X-100 (PTr), each wash lasting 2 min. Tissue dehydration was then performed using a graded methanol series (20%, 40%, 60%, 80%, and 100% methanol in PTr), with each step lasting 15 minutes. Samples were stored at –20 °C in 100% methanol until HCR-FISH was performed.

On the day of HCR-FISH, samples were rehydrated through a reverse methanol series (80%, 60%, 40%, and 20% methanol in PTr), each for 15 min. Following rehydration, samples were washed twice in PTr for at least 10 min each. Tissue digestion was then carried out by incubating the fins in 1 mL of PTr containing 7.5 µg/mL Proteinase K at 37 °C for 40 min. The digestion was quenched by adding 10 µL of 10% glycine, mixing gently, and incubating for 5 min. After removing the digestion solution, samples were rinsed twice in PTr for 1 min each and then fixed in PFA fixative for 1 h at room temperature. Following fixation, samples were rinsed in PTr and then incubated in 500 µL of bleaching solution (25 µL deionized formamide, 25 µL 30% H₂O₂, 5 µL 20× SSC, and 445 µL DEPC-treated water) for 1 h at room temperature to remove pigmentation. After bleaching, samples were permeabilized in a permeabilization buffer (PTr with 1% DMSO and 1% Triton X-100) for 3–5 h to enhance probe penetration. Samples were then washed in 5× SSC containing 0.1% Triton X-100 (SSCT) for 5 min, followed by incubation in pre-warmed hybridization buffer (HB, as described in Choi et al.⁶³) at 37 °C for 2–3 h. The hybridization solution was prepared by adding 10 µL of each 2 µM probe stock solution to 500 µL of pre-warmed HB. The pre-hybridization buffer was then replaced with this probe solution, and samples were incubated overnight at 37 °C.

After hybridization, samples were washed with 500 µL of probe washing buffer (PWB, from Choi et al.⁶³) 2–3 times, each for 20 min, followed by thorough washing with 5× SSCT (3–4 times, at least 1 h each). Next, amplification was performed by incubating samples in pre-warmed amplification buffer (AB, from Choi et al.⁶³) at 37 °C for at least 1 h. Fluorescent hairpins were prepared by taking 5 µL each of 3 µM hairpin H1 and H2 stock solutions, transferring them into RNase-free tubes, heating at 95 °C for 90 s, and then snap-cooling on ice for 30 min. The cooled hairpins were mixed together and added to 490 µL of AB. This mixture was then applied to the samples and incubated overnight in the dark at room temperature.

Following amplification, samples were washed with 5× SSCT at room temperature three times (30 min each), followed by 4–5 washes at 4 °C (1 h each). Prior to nuclear staining, tissues were rinsed twice in PTr (10 min each). DAPI staining was performed at room temperature using a 0.01 µg/mL DAPI solution for 10–20 min, followed by 4–5 washes in PTr, each lasting 15 min. Finally, samples were embedded in low-melting-point agarose and imaged using confocal microscopy.

qRT-PCR

Adult zebrafish were euthanized, and the pectoral fins and gonads were dissected. Tissues were rinsed with 1× PBS prepared with DEPC-treated water. Subsequently, 500 μL of RNA isolater Total RNA Extraction Reagent (Vazyme Biotech Co., Ltd.) was added to the tissue. The samples were homogenized on ice, followed by the addition of an additional 500 μL of RNA isolater. After incubation at room temperature for 5 min, 200 μL of chloroform was added, and the samples were vigorously vortexed. Phase separation was performed by centrifugation at 15,000 × g for 15 min at 4 °C. The aqueous phase was carefully transferred to a new tube, mixed with an equal volume of isopropanol, and incubated at –80°C for 30 minutes. RNA was pelleted by centrifugation at 15,000 × g for 15 min at 4 °C. The RNA pellet was washed with 70% ethanol (prepared with DEPC-treated water), air-dried, resuspended in nuclease-free water, and quantified before storage at –80 °C.

Total RNA was extracted from fin tissue and used for first-strand cDNA synthesis with the TransScript® First-Strand cDNA Synthesis SuperMix (TransGen Biotech, China), according to the manufacturer’s instructions. Quantitative real-time PCR (qRT-PCR) was performed using the CFX96 Touch Real-Time PCR Detection System (Bio-Rad, USA) and the PerfectStart® Green qPCR SuperMix (TransGen Biotech, China). The qRT-PCR cycling conditions were as follows: initial denaturation at 95 °C for 3 min, followed by 40 cycles of denaturation at 95 °C for 20 s, annealing at 60 °C for 20 s, and extension at 72 °C for 20 s. A final extension was performed at 72 °C for 5 min. Each gene was analyzed in technical triplicates for each sample.Relative mRNA expression levels were calculated using the 2^–ΔΔCt method, based on the Ct values of the target genes normalized to that of β-actin. All primers were synthesized by Sangon Biotech (Shanghai) Co., Ltd., and the specific sequences of gene-specific primers are listed in Supplementary Data 3.

Chemical treatments

Ethisterone (CAS No. 434-03-7, 97.01% purity) was purchased from MedChemExpress (China) and prepared as a 2.5 mg/mL stock solution in absolute ethanol. Finasteride (CAS No. 98319-26-7, 98% purity) was obtained from Macklin (China) and dissolved in analytical-grade dimethyl sulfoxide (DMSO) to prepare a 200 mM stock solution. 11-Keto testosterone (CAS No. 564-35-2, 97% purity) was purchased from Acmec (China), and a 100 mM stock solution was prepared in analytical-grade DMSO. Dihydrotestosterone (CAS No. 521-18-6, >98.00% purity; 100 mM), testosterone (CAS No. 58-22-0, >98.00% purity; 34.67 mM), and dutasteride (CAS No. 164656-23-9, >99.50% purity; 10 mM) were all obtained from Glpbio (USA), and their respective stock solutions were prepared in analytical-grade DMSO.

All chemical agents were dissolved according to the manufacturer’s instructions using either DMSO or ethanol and were stored long-term at –80 °C, with working aliquots maintained at –20 °C for routine use. Healthy adult zebrafish of both sexes (≥4 months old) were exposed to the drugs by immersion in 400 mL of system water containing the appropriate concentration of each compound. Water was renewed daily, and fresh drug solutions were re-added accordingly. Specific treatment concentrations and exposure durations are detailed in the main text.

Imaging

Zebrafish and other fish species were anesthetized using Tricaine, and pectoral fins were excised using corneal scissors. Bright-field images were captured using a stereomicroscope (T2-3M18) equipped with ImageView software (SHENZHEN AOSVI OPTICAL INSTRUMENT CO., LTD). For confocal imaging, fluorescent signals from HCR-FISH samples were acquired using a TCS SP8 DLS confocal microscope and LAS X LS software (Leica Microsystems). All images were processed using ImageJ (NIH) and Adobe Illustrator.

Measurement of sex steroid levels

Male zebrafish were divided into a DMSO control group and a Dutasteride-treated group, and subjected to immersion exposure for 28 days. Following treatment, fish were anesthetized, and blood collection was performed as described by Babaei et al.⁶⁴. Briefly, anesthetized zebrafish were positioned tail side down in a 0.5 mL microcentrifuge tube (punctured and smoothed), after making a diagonal incision at the caudal peduncle using a razor blade. The 0.5 mL tube was then placed into a 1.5 mL collection tube containing anticoagulant and centrifuged at 40 × g for 5 min at 11 °C. Fish were then removed and a second incision was made posterior to the initial wound to remove clotted blood. The centrifugation step was repeated, and the pooled blood was collected from the bottom tube. Whole blood was centrifuged at 13,700 × g for 15 minutes at 4 °C to separate blood cells from serum. The serum was carefully collected and stored at –80 °C for subsequent analysis.

Zebrafish serum samples were used to determine the concentrations of three steroid hormones: dihydrotestosterone (DHT), 11-ketotestosterone (11-KT), and testosterone (T). Hormone levels were quantified using commercially available ELISA kits: DHT (Dihydrotestosterone) ELISA Kit (FineTest), 11-Keto Testosterone ELISA Kit, and Testosterone ELISA Kit (Cayman Chemical). All assays were performed strictly according to the manufacturers’ instructions. Steroid hormone concentrations were expressed as nanograms per milliliter of serum (ng/mL).

Statistics and reproducibility

To ensure reproducibility, all experiments in this study were performed in at least three independent biological replicates. All bar graphs presented in this study represent the mean ± standard error of the mean (SEM). Statistical significance was assessed using the multiple t-tests function in GraphPad Prism 8. A p-value < 0.05 was considered statistically significant, while p-values > 0.05 were interpreted as not significant (NS). For all representative images, the phenotypic frequency is indicated in the lower-left corner as (n/N), where n represents the number of fish displaying the phenotype and N represents the total number of fish in the group.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Z.X. and Q.Q. conceived and designed the study. Z.X. performed the majority of the experiments, analyzed the data, and prepared the figures. X.Z. assisted with the experiments. C.Z., B.T., Z.D., C.W., S.L., L.H. and G.L. provided suggestions for image analysis and experimental assistance. Z.X. and Q.Q. wrote the manuscript. All authors reviewed and approved the final version of the manuscript.

Peer review

Peer review information

Communications Biology thanks Aubrey Converse, Akinori Kawamura and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: John Mulley and Johannes Stortz. A peer review file is available.

Data availability

All generated datasets are publicly available in the GenBank database under BioProject accession number PRJNA1307184.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-10140-z.