RNA binding protein RBM20 regulates turtle temperature-dependent sex determination by repressing the splicing of Wt1 KTS

Abstract

Accumulating correlative studies suggest a potential role of alternative splicing in vertebrate sex determination; however, the underlying mechanism remains elusive. Herein, we show that RNA binding motif protein 20 (RBM20) exhibits temperature-dependent sexually dimorphic expression in early embryonic gonads of the red-eared slider turtle Trachemys scripta, preceding the onset of gonadal differentiation. Knockdown of Rbm20 at male-producing temperature induces male-to-female sex reversal, whereas Rbm20 overexpression at female-producing temperature triggers male development. RBM20 represses the splicing of Wilms tumor suppressor WT1 at KTS sites, thereby governing the balance of +KTS and −KTS isoforms between different temperatures. In addition, functional analysis reveals that the Wt1 +KTS variant is both necessary and sufficient to drive male development in T. scripta. This study demonstrates that RBM20 activates the male pathway by repressing the Wt1 KTS splicing in T. scripta, thus establishing causality and a direct genetic link between alternative splicing mechanism and vertebrate sex determination.

Splicing regulator RBM20 activates the male pathway by repressing the splicing of Wt1 KTS, within a turtle TSD system.

INTRODUCTION

Alternative splicing (AS) is a ubiquitous pre-mRNA processing mechanism regulated by RNA binding splicing factors during gene expression of eukaryotes. This posttranscriptional process plays a crucial role in development by generating multiple transcripts that encode protein isoforms with diverse functions (1–3). Sex determination is the first critical event in gonadal development, directing the differentiation of bipotential gonads into either testes or ovaries. Classical theory posits that sex determination in invertebrates, such as most insects, is controlled by AS of the same genes (4–8), whereas in vertebrates, it is typically regulated by the differential expression of different genes (9–22). Accumulating research indicates that sexually dimorphic RNA splicing events of sex differentiation–related genes also occur in the embryonic gonads of vertebrates (23). Notably, the Wilms tumor suppressor gene (Wt1), essential for early gonadal development, produces two isoforms (+KTS and −KTS) via AS at the exon 9–10 junction, inserting or omitting three amino acids (KTS) between zinc fingers 3 and 4. Functional studies show that +KTS supports Sry activation and testis development, whereas −KTS promotes ovarian differentiation using knockout and transgenic mouse models (24–27). In addition, nonmammalian vertebrates, including reptiles, fish, and birds, show splicing events associated with gonadal differentiation and development (23). However, it remains unclear how the differences in these splicing forms between the sexes are regulated, specifically, which upstream RNA splicing factors are involved. Direct genetic evidence linking AS regulatory mechanisms to vertebrate sex determination is still lacking.

Temperature-dependent sex determination (TSD), where the ambient temperature during embryogenesis determines gonadal sex, is widely observed in reptiles, including the red-eared slider turtle (Trachemys scripta elegans, T. scripta) (28–30). Recently, spliceoform differences in two Jumonji family genes, Jmjd3 and Jarid2, have been reported in the embryonic gonads of three TSD reptilian species (Pogona vitticeps, Alligator mississippiensis, and T. scripta) (31, 32). Intriguingly, previous transcriptome-wide analysis of T. scripta embryonic gonads reveals a splicing factor, RNA binding motif protein 20 (RBM20), exhibiting low temperature and male-specific expression during the temperature-sensitive period (33). Rbm20 encodes an RNA binding protein with two critical functional domains: RNA-recognition motif 1 (RRM1) and a serine/arginine-rich (SR-rich) region, both essential for regulating pre-mRNA splicing (34). These observations imply that the splicing factor–mediated AS regulatory mechanism might be present and functions on the reptilian TSD system. Although Rbm20 has been extensively studied in the context of human cardiac diseases, particularly hereditary cardiomyopathy (35–39), whether and how it functions on reptilian TSD remains largely unexplored. Therefore, functional identification of Rbm20 and its direct downstream AS pathway that influences T. scripta gonadal differentiation is of great importance for the in-depth understanding of the AS regulatory mechanism of vertebrate sex determination.

In this study, we reveal that Rbm20 is both necessary and sufficient to initiate male development by repressing the splicing of Wt1 KTS in T. scripta. This finding underscores the critical role of RNA splicing in regulating turtle TSD and provides the first direct genetic evidence linking AS mechanism to vertebrate sex determination.

RESULTS

Rbm20 exhibits temperature-dependent and sexually dimorphic expression in early T. scripta gonads

To examine the expression profile of Rbm20 throughout the temperature-sensitive period at male-producing temperature (MPT; 26°C) and female-producing temperature (FPT; 32°C), we analyzed previously published T. scripta gonadal transcriptomes using a publicly available chromosome-level reference genome (33, 40). RNA sequencing (RNA-seq) and quantitative reverse transcription polymerase chain reaction (qRT-PCR) analyses revealed that MPT-specific expression of Rbm20 began in the T. scripta gonad at stage 15, preceding the onset of gonadal sex determination. This sexually dimorphic expression profile persisted in gonads through both the temperature-sensitive and sex-determination periods (Fig. 1, A and B).

Fig. 1. Temperature-dependent and sexually dimorphic expression of Rbm20 in T. scripta embryonic gonads.

(A and B) Transcript expression levels of Rbm20 in embryonic gonads of different developmental stages, determined by RNA-seq (A) and qRT-PCR (B). Rbm20 exhibited a highly MPT-specific expression pattern during temperature-sensitive periods (stages 15 to 19), covering the critical stages of sex determination (stages 15 to 17). (C) Coimmunofluorescence of RBM20 and GATA4 in embryonic gonads of stage 15. RBM20 protein was localized in somatic cells of seminiferous cords of MPT gonads. Scale bars, 100 μm. (D and E) Time-course response of Rbm20 expression to temperature shifts from either MPT-to-FPT (D) or FPT-to-MPT (E) in vivo. After eggs were shifted at stage 14, Rbm20 expression in gonads responded rapidly to new temperatures, with significant expression changes occurring as early as at stage 15. Data in (A), (B), (D), and (E) are shown as means ± SD. N ≥ 3. **P < 0.01; ***P < 0.001; n.s., not significant.

Immunofluorescence showed that RBM20 protein was detected in gonadal somatic cells (marked by GATA4) of seminiferous cords (Fig. 1C), implying that RBM20 functions in somatic cells to regulate the sexual development of T. scripta. We next examined the responses of Rbm20 expression to temperature shifts during the temperature-sensitive window. In gonads shifted from MPT to FPT, or from FPT to MPT at stage 14, significant changes in Rbm20 expression were evident by stage 15, preceding gonadal sex determination (Fig. 1, D and E). These expression profiles suggest that Rbm20 is an early responder to temperature, with the potential to act as a key regulator of male gonadal development at 26°C.

Knockdown of Rbm20 induces male-to-female sex reversal of MPT embryos

To investigate the functional role of Rbm20 in male sex determination and sexual differentiation of T. scripta, we established the Rbm20-deficient [Rbm20–knockdown (KD)] 26°C turtle models by introducing lentiviral vectors (LVs) carrying Rbm20-specific short hairpin RNAs (shRNAs) (Rbm20-shRNA#1, #2) in ovo at stage 14, before the onset of sex determination (fig. S1A). qRT-PCR revealed significant Rbm20 reduction in MPT gonads from stage 16 embryos after lentivirus injection, with the two Rbm20-shRNA (#1, #2) treated groups showing expression levels similar to control FPT gonads (NC-shRNA) (fig. S1B). Phenotypes of Rbm20-KD MPT gonads were subsequently assessed through gonadal histology, immunofluorescence, and qRT-PCR. The embryo survival rate at stage 25 ranged from 29.67% (89/300) to 50.67% (152/300), with green fluorescent protein (GFP)–positive gonads observed in 64.86 to 69.39% (48/74 to 34/49) of sampled embryos (table S1). Control MPT and FPT gonads treated with NC-shRNA developed normally. At stage 25, control MPT gonads were short and cylindrical, whereas control FPT gonads were long and flat. In Rbm20-KD MPT embryos (shRNA#1, #2), gonads became elongated and flattened, resembling female morphology. Hematoxylin and eosin (H&E) staining showed Rbm20-KD MPT embryos (shRNA#1, #2) were completely feminized, characterized by a thickened cortex with massive primordial germ cells and a highly degenerated medulla lacking sex cords (Fig. 2A). At 3 months and 1 year posthatching, Rbm20-KD MPT gonads (shRNA#2) displayed a phenotype similar to control FPT gonads, with abundant follicles (Fig. 2B and fig. S2A). VASA staining showed Rbm20-KD MPT gonads had germ cells (marked by VASA) localized exclusively to the thickened cortex, resembling control FPT gonads at stages 21 and 25 (Fig. 2A and fig. S2B).

Fig. 2. Knockdown of Rbm20 at MPT leads to male-to-female sex reversal.

(A and B) Representative images (top) of the GMCs from control and Rbm20-KD turtles at stage 25 and 3 months (3M) posthatching. The Rbm20-KD MPT gonads became elongated and flat, and an obvious ovary-like structure appeared, whereas the inner structure of Rbm20 mutant gonads exhibited the thickened cortex occupied by follicles and absent testis cords in the degenerated medulla, as well as the female-typical distribution pattern of germ cells, evidenced by H&E staining (middle) and VASA immunostaining (bottom) of gonadal sections. Gd, gonad (outlined by yellow dotted lines). Cor, cortex; Med, medulla. Scale bars, 1 mm (top) and 100 μm (middle and bottom). (C) qRT-PCR of male and female marker genes in gonads. It showed significant down-regulation of Sox9, Amh, and Dmrt1 and remarkable up-regulation of Foxl2, Cyp19a1, and R-spondin1 in Rbm20-KD MPT gonads of stage 25. Data are shown as means ± SD. N ≥ 3. *P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant. (D) Double immunofluorescence of SOX9&FOXL2 and AMH&AROM (aromatase) in gonadal sections. SOX9 and AMH protein expression almost disappeared, accompanied by the ectopic activation of FOXL2 and AROM in Rbm20-KD MPT gonads. Scale bars, 100 μm. (E) Phenotypic sex ratios of gonads from control and two KD groups at stage 25 and 3M. Phenotype of gonads was assessed by gonadal histology and immunofluorescence.

Gene expression analysis confirmed activation of the female developmental pathway in Rbm20-KD MPT gonads. mRNA levels of male markers (Sox9, Amh, and Dmrt1) were significantly reduced, whereas female markers (Foxl2, Cyp19a1, and R-spondin1) were up-regulation in Rbm20-KD MPT gonads (shRNA#1, #2) at stage 25 and 3 months and 1 year posthatching (Fig. 2C and fig. S2C). Immunofluorescence staining showed that SOX9 and AMH were expressed in Sertoli cells of control MPT gonads but absent in control FPT gonads. FOXL2 and AROM (AROMATASE, the protein coded by Cyp19a1 gene) were expressed in granulosa cells of control FPT gonads but undetected in control MPT gonads. In Rbm20-KD MPT gonads, SOX9 and AMH expressions were reduced, whereas FOXL2 and AROM were induced in gonadal somatic cells (Fig. 2D and fig. S2, D and E). Statistical analysis revealed that 91.67% (44/48) and 88.10% (37/42) of MPT gonads from stage 25 embryos with Rbm20-shRNA#1 and Rbm20-shRNA#2 treatment exhibited male-to-female sex reversal, respectively. At 3 months and 1 year posthatching, the reversal ratio in the Rbm20-shRNA#2 KD group was 92.59% (25/27) and 94.12% (16/17) (Fig. 2E and fig. S3A). These findings indicate that Rbm20 is essential for male sex determination in T. scripta, with its loss leading to male-to-female sex reversal.

Rbm20 overexpression results in female-to-male sex reversal of FPT embryos

The ectopic activation of Rbm20 in FPT embryos was conducted to assess its role in initiating the male development of T. scripta. Rbm20-overexpressing [Rbm20–overexpression (OE)] FPT embryos with over threefold up-regulation were generated by injecting an LV carrying the Rbm20 open reading frame (ORF) into stage 14 turtle eggs incubated at 32°C (fig. S1, A and B). The survival and transfection rates were comparable across FPT and MPT groups, with GFP expression detected in 66.67 to 69.57% (30/45 to 32/46) of sampled gonads at stage 25 (table S2). In Rbm20-OE FPT embryos, gonads exhibited cylindrical structures longer than control MPT gonads but distinct from FPT gonads. Overexpressing Rbm20 led to well-organized medullae with seminiferous cord-like structures, a reduced cortex (testis-like), and in some cases, intersexual gonads (ovotestis) with remaining germ cells in the cortex (Fig. 3A). At 3 months and 1 year posthatching, these gonads fully developed into testes, with no ovotestes observed (Fig. 3B and fig. S4A). As observed at stage 25, two germ cell distribution patterns were also present in Rbm20-OE gonads at stage 21. In some FPT gonads, germ cells were restricted to the medullary region (testis-like), whereas others displayed germ cells in both the cortex and medulla (fig. S4B).

Fig. 3. Overexpression of Rbm20 at FPT results in male-to-female sex reversal.

(A and B) Representative images (top) of the GMCs from control and Rbm20-OE turtles at stage 25 and 3 months (3M) posthatching. The Rbm20-OE FPT gonads displayed approximately cylindrical surface without follicles, and the inner structure exhibited an organized medulla with apparent testis cords and cortex of varying degrees of degeneration as well as the male-typical distribution of germ cells, determined by H&E staining (middle) and VASA immunostaining (bottom). Gd, gonad; Cor, cortex; Med, medulla. Scale bars, 1 mm (top) and 100 μm (middle and bottom). (C) qRT-PCR of male and female genes in gonads, showing an obvious reduction in Foxl2, Cyp19a1, and R-spondin1 and an increase in Sox9, Amh, and Dmrt1 in Rbm20-OE FPT gonads. Data are shown as means ± SD. N ≥ 3. *P < 0.05; **P < 0.01; ***P < 0.001. (D) Double immunofluorescence of SOX9&FOXL2 and AMH&AROM (aromatase) in gonadal sections. The SOX9 and AMH protein was induced to robustly express in Sertoli cells of the medullary sex cords, whereas FOXL2 and AROM were almost undetectable in completely sex-reversed gonads (testis). Male and female proteins were existed simultaneously in those incompletely sex-reversed gonads (ovotestes). Scale bars, 100 μm. (E) Phenotypic sex ratios of gonads from control and overexpression groups at stage 25 and 3M. Phenotype of gonads was assessed by gonadal histology and immunofluorescence.

Rbm20-OE FPT gonads showed up-regulated male marker genes (Sox9, Amh, and Dmrt1) and down-regulated female markers (Foxl2, Cyp19a1, and R-spondin1) at stage 25 and 3 months and 1 year posthatching (Fig. 3C and fig. S4C). Immunofluorescence confirmed reduction of FOXL2 and AROM and ectopic activation of SOX9 and AMH in masculinized gonads, localized to Sertoli cells in testis cords, resembling control MPT gonads (Fig. 3D and fig. S4, D and E). Ovotestes showed simultaneous expression of SOX9, FOLX2, AMH, and AROM (Fig. 3D). In the Rbm20-OE group, 75.00% (27/36, 22 testes; 5 ovotestes) of FPT embryos underwent sex reversal at stage 25, and 78.57% (22/28, 22 testes; 0 ovotestes) and 86.67% (13/15, 13 testes; 0 ovotestes) developed testes by 3 months and 1 year posthatching, respectively (Fig. 3E and fig. S3B). These findings demonstrate that Rbm20 is sufficient to initiate the male developmental pathway in T. scripta.

RBM20 regulates the balance of the relative abundance of Wt1 +KTS and −KTS isoforms

To investigate genes regulated by RBM20 during sex determination and their splicing differences between sexes in T. scripta, we performed gonadal RNA-seq at stage 16, just after the onset of sexually dimorphic expression of Rbm20 (stage 15). There are significant differences in splicing variants of 61 genes (34 in MPT and 27 in FPT), including Kdm6b and Jarid2, previously reported in TSD turtles (Fig. 4A). One key gene that stood out was Wt1, known to regulate gonadal development. Wt1 (+/−KTS) isoforms are selectively spliced on the basis of the insertion or omission of the three amino acids KTS sequence at the end of exon 8 (fig. S5A). In MPT gonads, the +KTS isoform was more abundant, whereas FPT gonads showed higher −KTS splicing (Fig. 4A). qRT-PCR confirmed that the splicing of Wt1 KTS is temperature dependent, with higher +KTS levels in MPT gonads and higher −KTS levels in FPT gonads during the sex determination period (stages 16 and 17) (Fig. 4B). This resulted in a higher +/−KTS ratio in MPT gonads and −/+KTS ratio in FPT gonads (Fig. 4C). In temperature-shift experiments, changes in Wt1 splicing were observed at stage 16, later than the Rbm20 response at stage 15 (fig. S5B).

Fig. 4. RMB20 regulates the alternative splicing of Wt1 KTS.

(A) Differential splicing analysis for transcripts between MPT and FPT gonads of stage 16 by RNA-seq. In MPT gonads, the splicing level of Wt1 +KTS was significantly higher than that of −KTS whereas just the reverse in FPT gonads. (B and C) qRT-PCR of Wt1 +/−KTS in embryonic gonads. The AS of Wt1 KTS is temperature dependent, with a higher ratio of +/−KTS in MPT gonads and −/+KTS in FPT gonads at stages 16 and 17. (D) Responses of gene isoforms expression changes to KD of Rbm20 in MPT gonads by RNA-seq. Rapid down-regulation of Wt1 +KTS was detected at stage 16. (E to G) RT-PCR and qRT-PCR of Wt1 +/−KTS in embryonic gonads of stage 16 after treatment with Rbm20 KD or OE. The relative expression abundance of Wt1 +/−KTS was reduced in response to Rbm20 KD, which was increased in Rbm20-OE FPT gonads. Data in (B) and (F) are shown as means ± SD. N ≥ 3. *P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant. (H) RIP-PCR analysis of RBM20 in MPT gonads of stage 16, showing that RBM20 could directly bind to the pre-mRNA of Wt1.

To confirm that RBM20 regulates Wt1 KTS splicing, we examined splicing levels in Rbm20-deficient gonads. RNA-seq analysis revealed that +KTS splicing was down-regulated in Rbm20-KD MPT gonads, but no significant changes in the other 60 gene isoforms (Fig. 4D). RT-PCR and qRT-PCR further showed that Rbm20-KD MPT gonads had reduced +KTS levels and increased −KTS levels, whereas Rbm20-OE in FPT gonads led to up-regulation of +KTS levels and down-regulation of −KTS levels at stage 16 (Fig. 4, E and F). The +/−KTS ratio in Rbm20-KD MPT gonads was reduced to levels similar to control FPT gonads, whereas in Rbm20-OE FPT gonads, the ratio increased above that of control MPT gonads (Fig. 4G). By stage 25 and 3 months posthatching, qRT-PCR of +KTS showed significant down-regulation in MPT gonads with LV-Rbm20-shRNA (#1 or #2) treatment and obvious up-regulation in FPT gonads with LV-Rbm20-OE treatment (fig. S6). In addition, RNA immunoprecipitation–PCR (RIP-PCR) and RNA pull-down confirmed that RBM20 directly binds to the RNA sequence of Wt1 (Fig. 4H and fig. S7). These results demonstrate that RBM20 directly regulates the relative expression of Wt1 +/−KTS isoforms by repressing the splicing of the +KTS variant, identifying Wt1 as a key target gene of RBM20.

Wt1 +KTS is both indispensable and sufficient for male development

To clarify that Rbm20 induces male development by regulating Wt1 KTS splicing, we simulated the relative expression of +/−KTS in MPT and FPT gonads by constructing LVs carrying +KTS-specific shRNA (for KD) and +KTS ORF (for OE) (fig. S8A). These vectors were injected into embryos at stage 15, and infection efficiency was analyzed at stage 16 by qRT-PCR. As expected, the +KTS expression was down-regulated in the KD group and up-regulated in the OE group, whereas −KTS splicing remained unchanged (fig. S8B). The changes led to a reduced +/−KTS ratio in MPT gonads (similar to FPT gonads) and an elevated ratio in FPT gonads (similar to MPT gonads) (fig. S8C). Sex reversal was confirmed by histological analysis, immunofluorescence, and qRT-PCR of marker genes at stage 23. GFP expression was observed in 60.81 to 65.00% (45/74 to 26/40) of sampled gonads (table S3). MPT gonads with Wt1(+KTS)-KD displayed male-to-female sex reversal, with a flat outline, a well-defined cortex populated with two to three layers of germ cells, and a degraded medullary region lacking seminiferous cords. In contrast, FPT gonads with Wt1 (+KTS)-OE showed female-to-male sex reversal, with a highly developed medulla full of seminiferous cords and a degenerated cortex resembling MPT gonads. By 3 months posthatching, the changes in the morphology and structure of the treated gonads became more visible (fig. S9A). Changes in the distribution of VASA-positive germ cells at stage 23 further confirmed the sex reversal (Fig. 5, A and B).

Fig. 5. Wt1 +KTS is both necessary and sufficient to drive male development in T. scripta.

(A and B) Immunofluorescence of T. scripta embryonic gonads of stage 23 after treatment with Wt1 +KTS KD or OE (A: whole mount; B: transection). KD of Wt1 +KTS at MPT led to male-to-female sex reversal, evidenced by female-type distribution of VASA-positive germ cells, induction of FOXL2 and Aromatase (AROM), and disappearance of SOX9 and AMH proteins. OE of Wt1 +KTS at FPT promoted male development with epitopic expression of SOX9 and AMH. Scale bars, 200 μm (A) and 100 μm (B). (C) qRT-PCR of Sox9, Amh, Foxl2, and Cyp19a1 in stage 23 gonads. qRT-PCR results were consistent with immunofluorescence analysis. Data are shown as means ± SD. N ≥ 3. *P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant. (D) Phenotypic sex ratios of gonads from control and (Wt1 +KTS)-KD/OE groups at stage 23. Phenotype of gonads was assessed by immunofluorescence. (E) Summary of epigenetic and splicing pathways regulating TSD in red-eared slider turtles.

Gene expression analysis at stage 23 and 3 months posthatching showed that Wt1 (+KTS)-KD MPT gonads had decreased Sox9 and Amh levels and increased Foxl2 and Cyp19a1 levels. Conversely, Wt1 (+KTS)-OE FPT gonads showed up-regulation of Sox9 and Amh, with down-regulation of Foxl2 and Cyp19a1 (Fig. 5C and fig. S9B). Double immunofluorescence staining for whole mount and transection of gonads both confirmed the ectopic presence of the FOXL2 and AROM protein and the absence of SOX9 and AMH expression in (+KTS)-KD MPT gonads. In FPT gonads with Wt1 (+KTS)-OE, the four proteins showed the male gonadal pattern (Fig. 5, A and B, and fig. S9C). The sex reversal ratios induced by Wt1 +KTS KD and OE were 93.33% (42/45, 3 testes; 42 ovaries) and 83.33% (30/36, 30 testes; 6 ovaries) at stage 23, respectively (Fig. 5D). By 3 months posthatching, the sex reversal ratios of KD and OE groups was 94.44% (17/18, 1 testes; 17 ovaries) and 88.24% (15/17, 15 testes; 2 ovaries) at stage 23, respectively (fig. S9D). These findings provide strong evidence that proper Wt1 +KTS expression is both indispensable and sufficient for early testicular differentiation in T. scripta.

DISCUSSION

The central role of AS in sex determination has been well documented in insects (such as Drosophila melanogaster), in which primary sex is controlled by a short cascade of sex-specifically spliced genes (Sxl, Tra, and Dsx) (6–8, 41, 42). However, the mechanism governing the formation of different transcript isoforms of sex-related genes [Wt1 (24–27), Nr5a (43), Dmrt1 (44–47), etc.] remains unknown, leaving a gap in the evidence supporting the importance of AS in vertebrate sex determination. Here, we functionally identify RBM20 as a critical regulator of TSD in T. scripta. Previous research reveals that RBM20 functions as a splicing regulator in cardiac development by modulating the exon skipping of titin mRNA (35, 48, 49). This study provides the first evidence of its involvement in gonadal development. RBM20 exhibited sexually dimorphic and temperature-sensitive expression patterns in early embryonic gonads, with a rapid response to temperature shifts during the critical window of sex determination. Loss-of-functional and gain-of-functional assays demonstrated that RBM20 is both necessary and sufficient for early testicular development. This strongly indicates RBM20 as a pivotal factor in initiating the male pathway in T. scripta, expanding its functional repertoire beyond the cardiovascular system.

Our findings highlight a conserved role of Wt1 AS in vertebrate sex determination. In T. scripta, the Wt1 splice variants, +KTS and −KTS, exhibited temperature-dependent and sex-dependent expression patterns. A higher level of the +KTS isoform was observed in MPT gonads, whereas the −KTS isoform was predominant in FPT gonads, which was also found in another TSD turtle, the snapping turtle Chelydra serpentina (50). Functional studies confirmed that Wt1 +KTS is indispensable for the initiation of male development in T. scripta. This aligns with findings in mammals, where Wt1 +KTS regulates Sry expression, essential for male sex determination, and Wt1 −KTS supports ovarian development (24, 25, 27).

RBM20 directly regulates the AS of Wt1 by modulating the production of its KTS isoforms. RIP and RNA pull-down assays demonstrated a direct interaction between RBM20 protein and Wt1 RNA, indicating that RBM20 binds Wt1 transcripts to influence splicing outcomes. Rbm20 KD at MPT led to a reduction in the +KTS isoform and an increase in the −KTS isoform, whereas a rise in the +/−KTS ratio was observed in FPT gonads overexpressing Rbm20. It is speculated that RBM20 protein binds to sites on Wt1 RNA that are typically recognized by splicing activators for the KTS isoform, consequently preventing the activation of KTS splicing. These findings suggest a crucial link between RBM20 activity and Wt1 splicing, providing a mechanistic basis for how temperature cues translate into differential gonadal outcomes. We propose a working model for the role of RBM20 in T. scripta sex determination. At MPT, high expression of RBM20 represses the AS of Wt1, favoring the retention of +KTS isoform, which may promote Dmrt1 activation and testicular differentiation by stabilizing transcriptional complexes or enhancing chromatin accessibility (51). In contrast, at FPT, reduced RBM20 expression allows other splicing factors to enhance KTS skipping, resulting in increased −KTS levels, which may contribute to ovarian differentiation either by repressing Dmrt1 or activating pro-ovarian genes such as Foxl2 (Fig. 5E). This mechanism differs from RBM20’s established role in cardiac titin splicing, where it primarily mediates exon skipping (48, 49). In gonads, we hypothesize that RBM20 acts as a splicing repressor by inhibiting the binding of activators to the KTS site, reflecting a context-specific regulatory function that remains to be tested.

We previously identified a thermosensitive epigenetic pathway in T. scripta, in which phosphorylated STAT3 (pSTAT3) represses Kdm6b transcription at FPT, thereby inhibiting the male pathway while directly activating Foxl2 to initiate ovarian development (52–54). In contrast, at MPT, KDM6B activates male sex determination by removing H3K27 trimethylation at the Dmrt1 promoter, thereby promoting Dmrt1 transcription (55, 56). As Dmrt1 is necessary and sufficient to initiate male development in this species, its regulation represents a key node in the sex determination network. Together with this study, these findings reveal that epigenetic (KDM6B), splicing (RBM20–Wt1), and transcriptional (pSTAT3–Foxl2) pathways converge on Dmrt1 as the central decision node of TSD. Supporting this model, KD of RBM20 or Wt1 +KTS at MPT rapidly reduces Dmrt1 expression (fig. S10), demonstrating that the Wt1 +KTS isoform, produced through RBM20-mediated splicing, is required for the male pathway activation. At FPT, pSTAT3-mediated activation of Foxl2 initiates the ovarian program by up-regulating Cyp19a1 (aromatase) and suppressing Dmrt1 and Sox9 (53, 54). These parallel but interconnected pathways establish a multilayered, nonhierarchical, bistable regulatory network (57–59), in which perturbation of RBM20, WT1, or KDM6B destabilizes the balance and triggers sex reversal. We propose a model in which splicing, epigenetic, and transcriptional regulation act in parallel but converge on Dmrt1 to canalize gonadal fate (Fig. 5E). Notably, Wt1 +KTS may also recruit or stabilize KDM6B-containing chromatin complexes at the Dmrt1 locus, suggesting a direct mechanistic link between splicing and epigenetic regulation—a key hypothesis for future testing.

This study reveals that the splicing factor RBM20 activates the male pathway of a turtle TSD system by repressing the splicing of Wt1 KTS, thus establishing causality and a direct genetic link between the AS mechanism and vertebrate sex determination. Identification of how temperature regulates Rbm20 expression and which splicing factors interact with RBM20 to control Wt1 isoform production may uncover key components of this regulatory network, which will enhance our understanding of AS in TSD and its evolutionary conservation.

MATERIALS AND METHODS

Turtle eggs incubation and tissue collection

Freshly laid T. scripta eggs were obtained from the Hanshou turtle farm (Hunan, China). Fertilized eggs were randomized in plastic boxes with moist vermiculite and placed in egg incubators at 26°C (MPT, producing all males) or 32°C (FPT, producing all females), with humidity maintained at 70 to 80%. Embryos were staged according to criteria established by Greenbaum (60). At stage 14, 100 eggs were shifted from an incubator set at 26°C to one set at 32°C, and vice versa. Embryos were carefully removed from their eggshells at various developmental stages using forceps. The naked embryos and turtles after hatching (3 months and 1 year) were decapitated and placed in phosphate-buffered saline (PBS) for pure gonads and/or gonad-mesonephros complexes (GMCs) collection. Gonads were thoroughly broken up and immersed in TRIzol reagent (Invitrogen, USA) for total RNA isolation. GMCs were fixed in 4% paraformaldehyde (PFA) overnight at 4°C, dehydrated through 50% ethanol, and then stored in 70% ethanol at 4°C until paraffin embedding and sectioning. All the animal experiments were carried out according to a protocol approved by the Experimental Animal Ethics Committee of Zhejiang Wanli University, China (approval number: 20190507001).

Construction of lentivirus expression vector systems

Preparation of the shRNA vectors and the generation of lentivirus (LV) were carried out as previously described (56). In brief, two shRNAs targeting turtle Rbm20 and one shRNA targeting Wt1 +KTS mRNA were designed and subsequently ligated into pGP-U6 (GenePharma, China) to produce pGP-U6-Rbm20-shRNA and pGP-U6-(Wt1 +KTS)-shRNA. The generated shRNA constructs were digested with AgeI-EcoRI and inserted into the EcoRI site of the pGLV-U6-GFP vector (GenePharma), producing the recombinant vectors pGLV-GFP-Rbm20-shRNA (termed LV-Rbm20-KD) and pGLV-GFP-(Wt1 +KTS)-shRNA [termed LV-(Wt1 +KTS)-KD]. The negative control vector pGLV-GFP-NC-shRNA (termed LV-NC-shRNA) was prepared by inserting a nonsense shRNA. The sequences of the shRNAs are as follows: Rbm20-shRNA#1 (5′-GGCCGAGTAGTACACATCTGC-3′); Rbm20-shRNA#2 (5′-GGTCACAGTGGATGAAGTG GG-3′); (Wt1 +KTS)-shRNA(5′-GTATGTCCATTTTG TTCACTT-3′) (fig. S6A). Human embryonic kidney (HEK) 293T cells were transfected with optimized packaging plasmids (pGag/Pol, pRev, and pVSV-G) along with the constructed LVs by Lipofectamine 2000 (Invitrogen, USA). Seventy-two hours posttransfection, the lentivirus was harvested by centrifugation (3000 rpm, 15 min, and 4°C) and then filtered through a 0.45-μm filter (Millipore, USA). A viral titer of ~4 × 10⁸ infectious units/ml was obtained. The lentivirus was stored at −80°C before the infection of turtle embryos.

The construction methods of the two overexpression vectors were carried out following a previous study (56). In brief, the full-length ORF sequences of Rbm20 [3762 base pairs (bp)] and Wt1 +KTS transcript (1254 bp) were PCR amplified from the testis cDNA of adult turtles. Subsequently, the PCR products were digested with EcoRI and cloned to pGLV-EF1a-GFP (LV-4, GenePharma), respectively. The recombinant vectors pGLV-GFP-Rbm20 and pGLV-GFP-(Wt1 +KTS) were named LV-Rbm20-OE and LV-(Wt1 +KTS)-OE. The empty vector pGLV-GFP-empty was constructed as a negative control (LV-empty). High-quality proviral DNA was used to transfect 293T cells. Virus propagation was carried out as described above.

Infection of turtle embryos

Turtle eggs were sterilized by swabbing with alcohol and then microinjected with the high-titer virus (at least 1 × 10⁸ infectious units/ml, ~5 μl per egg) using a fine metal Hamilton needle (diameter: 0.5 mm). The LV-Rbm20-shRNA and LV-Rbm20-OE viruses were injected into the region adjacent to the embryo (extraembryonic membranes) at stage 14 (Rbm20 began to exhibit a highly MPT-specific expression at stage 15) at MPT and FPT, respectively. The LV-(Wt1 +KTS)-shRNA (under MPT) and LV-(Wt1 +KTS)-OE (under FPT) virus were injected at stage 15. A total of 300 eggs were treated in each group. Control eggs at MPT and FPT were injected with LV-NC-shRNA (KD experiment) or LV-empty (OE experiment), respectively. Following injection, the eggshell punctures were sealed with parafilm to prevent contamination. Embryos were sampled and dissected for RNA-seq (stage 16) and sex reversal analysis (stage 23 or 25, 3 months and/or 1 year after hatching). For RNA-seq, each biological replicate consisted of 20 pairs of gonads pooled per group, with three biological replicates. For sex reversal analysis, one gonad from each pair in embryos was examined by histology and immunostaining, and another one was used for qRT-PCR.

Gonadal RNA-seq and differential alternative splice analysis

Gonadal transcriptomes from T. scripta at stages 15 to 18 and 21 were obtained from the NCBI under Bioproject PRJNA331105. Total RNA was extracted from FPT, MPT, and Rbm20-KD MPT gonads at stage 16 using TRIzol reagent (Invitrogen). Three biological replicates were performed per group. RNA-seq was performed on an Illumina HiSeq 2000 platform, generating 150-bp paired-end reads. Each RNA-seq library yielded ~52 million to 68 million high-quality reads, as summarized in table S4.

Gene expression profiling was performed using the assembled T. scripta genome and its accompanying gene annotation (GCF_013100865.1_CAS_Tse_1.0) from NCBI. RNA-seq data were aligned to this reference genome using HISAT2 (version 2.2.1) with default parameters. The resulting SAM files were converted to BAM files and sorted using Samtools (version 1.6) to prepare for downstream processing. Transcript abundances were calculated by StringTie (version 2.2.1) and imported into R (version 4.2.1) using the tximport package (version 1.26.0). Transcripts were summarized to the gene level using the tx2gene feature of tximport. To account for variations across samples, FPKMs (fragments per kilobase of transcript per million mapped reads) were normalized using a scaling method based on TMM (trimmed mean of M values), which assumes that most genes have similar expression levels across samples. Differential expression analysis of RNA isoforms between the MPT and Rbm20-KD groups was conducted using Ballgown (version 4.2.0), with a false discovery rate (FDR) threshold of 5%.

We performed differential splicing analysis for transcripts and local splicing events following the SUPPA2 tutorial (https://github.com/comprna/SUPPA/wiki/SUPPA2-tutorial). Briefly, RNA-seq data were aligned to the reference genome using Salmon (version 1.9.0) for transcript quantification. SUPPA2 (version 2.3) was then used to identify AS differences. For each AS event, the splicing level was quantified using the Percentage of Spliced-In (PSI). Differential splicing between MPT and FPT groups was evaluated by calculating the mean PSI difference (ΔPSI) and determining statistical significance using P values derived from independent t tests comparing PSI values between the groups.

qRT-PCR and statistical analyses

Gonads in each group were microdissected from the mesonephros at the indicated time points and harvested for total RNA extraction using TRIzol reagent (Invitrogen). cDNA synthesis was performed according to the manufacturer’s protocol of the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, USA). Gene transcript levels in embryonic gonads of all treated and control groups were examined by qPCR. The qPCR reaction was carried out using SYBR PrimeScript II (Takara, Japan) in a Bio-Rad iCycler system. Gapdh was used as a reference gene. After normalization with Gapdh, relative RNA levels were calculated by the comparative threshold cycle (Ct) method. The primer sequences for qPCR are listed in table S5.

For embryonic gonads, each biological replicate comprised RNA pooled from 10 gonads. For posthatching gonads (3 months and 1 year), each group included three gonads. Three biological replicates per group were performed, and each RNA sample was analyzed in triplicate to ensure technical reliability. All data are presented as means ± SD and analyzed by the one-way Duncan test and analysis of variance (ANOVA) with SPSS software. For all analyses, a P value of <0.05 was regarded as statistically significant (*P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant).

Immunofluorescence

Immunofluorescence was performed as previously described (56). Briefly, GMCs were fixed in 4% PFA overnight at 4°C, dehydrated in graded ethanol, and then embedded in paraffin wax and sectioned. Paraffin sections (5 to 6 μm) were deparaffinized and rehydrated before immersion in 10 mM sodium citrate buffer for 30 min at a sub-boiling temperature (96° to 99°C) for antigen retrieval. After blocking for 1 hour in blocking solution at room temperature, sections were incubated with primary antibodies overnight at 4°C, followed by washing (three times, 20 min each) and then secondary antibodies incubation (2 hours). The primary antibodies used in this analysis included rabbit anti-RBM20 (1:50, Thermo Fisher Scientific, AB_2646373), rabbit anti-SOX9 (1:500, Millipore, AB_2239761), rabbit anti-AMH (1:500, privately produced), goat anti-FOXL2 (1:300, privately produced), mouse anti-AROMATASE (AROM) (1:500, privately produced), rabbit anti-VASA (1:500, Abcam, AB_443012), mouse anti-CTNNB1 (1:500, Sigma-Aldrich, AB_94544), and mouse anti-GATA4 (1:300, Santa Cruz, AB_627667). Primary antibodies were detected using secondary antibodies Alexa Fluor 488/594 donkey anti-rabbit immunoglobulin G (IgG) (AB_2535792/AB_141637), Alexa Fluor 488/594 donkey anti-mouse IgG (AB_141607/AB_2535789), and Alexa Fluor 488 donkey anti-goat IgG (AB_2762838) (all diluted at 1:250, Invitrogen). Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) (286 nM, Sigma-Aldrich) and then washed with PBS (three times, 5 min each time). Fluorescence signals were observed under a confocal microscope (A1 Plus, Nikon).

RNA immunoprecipitation

RIP assay was conducted using the RIP kit (BersinBio, China) in accordance with the manufacturer’s instructions. In brief, 30 pairs of fresh gonads from MPT (26°C) embryos of stage 16 were pooled in a single tube for each group (Input, RBM20 IP, and IgG negative control). Gonadal cells were lysed, and DNA was removed. RBM20 antibody (1:50) and IgG antibody were mixed thoroughly into cell lysates of the IP and IgG groups, respectively, and incubated overnight at 4°C. Then, equal amounts of preequilibrated A/G magnetic beads were added and incubation continued for 1 hour. The magnetic beads were rinsed thoroughly, and RNA was extracted from RBM20 IP, IgG groups, and the Input group. The RNA was reverse-transcribed and subsequently validated by gel electrophoresis and qPCR. The Wt1 primer sequences are listed in table S1.

RNA pull-down and Western blot

The RNA pull-down assay was performed using an RNA pull-down kit (BersinBio). Briefly, 30 pairs of fresh gonads from MPT (26°C) embryos of stage 16 were pooled in a single tube. The gonadal tissues were fully lysed to extract proteins, and the lysate was divided into three groups (Input, RPD, NC, at a ratio of 1:5:5). Biotinylated RNA probes for Wt1 (+KTS) and its complementary strand (300 pmol each) were synthesized separately and incubated with streptavidin magnetic beads. The probe-bound beads were then incubated with the protein lysates of the RPD and NC group (4°C, overnight). After thoroughly washing the beads, proteins were eluted by adding protein lysis buffer and heating at 95°C.

The eluted proteins were identified using Western blot with the rabbit anti-RBM20 (1:100) and β-actin (1:5000, Proteintech Group, AB_2923704) antibodies. Equal amounts of protein solutions were separated on a 10% SDS–polyacrylamide gel electrophoresis gel and transferred onto a polyvinylidene difluoride membrane. The membrane was incubated in 5% bovine serum albumin at room temperature for 1 hour and subsequently with the primary antibody overnight at 4°C. Antibody recognition was detected with goat anti-rabbit IgG–horseradish peroxidase (1:5000, Proteintech Group). Chemiluminescent signals were detected using an enhanced chemiluminescence detection kit (NCM Biotech, China), and images were captured using a gel imaging system (ProteinSimple, USA).

Supplementary Materials

This PDF file includes:

Figs. S1 to S10

Tables S1 to S5