Temperature fluctuations and estrone sulfate affect gene expression via different mechanisms to promote female development in a species with temperature-dependent sex determination

Conditions during embryonic development can have strong and lasting effects on offspring phenotype. The degree to which developmental conditions vary depends upon several factors including the local environment, parity mode and maternal contributions to the embryo. Embryos of most oviparous species complete the majority of their development outside of the female and may be subjected to both abiotic and biotic forces that can alter their developmental trajectories. For example, abiotic factors such as environmental temperature can affect the rate of embryonic development, survival and sexual differentiation (Crews et al., 1994; Booth, 2006; Turriago et al., 2015; Lambert et al., 2018). Biotic factors such as maternal steroids can also affect offspring development and behavior (Carere and Balhazart, 2007), including begging behavior, offspring growth and success, and immune function and survival (Groothuis et al., 2005; Ruuskanen, 2015; Groothuis et al., 2019; Mouton and Duckworth, 2021). Many traits are subject to the effects of both abiotic and biotic factors; for example, metabolic rates in birds can be affected by both incubation temperature (DuRant et al., 2012) and testosterone exposure (Nilsson et al., 2011). It remains to be determined whether abiotic and biotic factors can elicit their effects through similar mechanisms or whether these effects arise independently. Work in reptiles with temperature-dependent sex determination (TSD) may provide some insights into how abiotic and biotic factors might interact at a mechanistic level to affect offspring phenotype. Because reptiles with TSD have a single phenotypic trait (e.g. sex) that is sensitive to both temperature and steroids, they provide a unique opportunity to investigate the mechanisms by which abiotic and biotic factors act and potentially interact to affect hatchling phenotype.

For many species with TSD, exposure to constant, warm incubation temperatures induces ovary development, and exposure to constant, cool temperatures induces testis development (Bull, 1985; Wibbels and Crews, 1995; Merchant-Larios et al., 1997). In the red-eared slider turtle (Trachemys scripta), eggs incubated at 26°C for the entirety of development will produce 100% male hatchlings and eggs incubated at 31°C for the entirety of development will produce 100% female hatchlings (Wibbels and Crews, 1995). In natural nests, T. scripta embryos are regularly exposed to both male- and female-producing temperatures during development (Carter et al., 2017) and recent research demonstrates that transient bouts of heat exposure are sufficient to induce ovary formation. For example, exposure to 29.5±3°C for only 8 days (from a baseline of 27±3°C) can induce a 50:50 hatchling sex ratio (Carter et al., 2018). The timing of exposure is critical; sex ratios are not affected by temperature early in the incubation period before the gonads have developed or by temperature late in the incubation period after gonadal fate becomes committed (Wibbels et al., 1991a; Breitenbach et al., 2020). The stage of development where gonadal fate is affected by incubation temperature is termed the ‘temperature-sensitive period’ (TSP) (Wibbels et al., 1991a). These findings highlight how under more natural incubation conditions, warm temperatures can induce ovary formation during relatively short periods of development.

Investigation into how warm incubation temperatures induce ovary formation led to the discovery that exogenous estradiol can override the effects of male-producing temperatures on developing embryos and result in female-biased sex ratios (Fleming and Crews, 2001; Murdock and Wibbels, 2006; Barske and Capel, 2010; Capel, 2017; Ge et al., 2017, 2018). Most of these studies were conducted using constant incubation temperatures, with eggs being treated with estradiol at the beginning of the TSP (Wibbels and Crews, 1995; Ge et al., 2018). While these studies were critical for identifying a role for estradiol in ovary formation, they did not reflect natural conditions with regards to embryonic exposure to maternal estrogens. Maternally derived steroids, including estrogens, are present in the yolk of many oviparous vertebrates, and these endogenous compounds have been found to affect hatchling phenotype (Bowden et al., 2000; Lovern and Wade, 2001, 2003; Paitz and Bowden, 2013; Carter et al., 2017). In T. scripta eggs, maternal estrogens exhibit seasonal variation, where late season eggs have both more estradiol (Bowden et al., 2002; Carter et al., 2017) and more estrone sulfate (Paitz and Bowden, 2013). Once development begins, levels of estradiol in the yolk decline rapidly (Paitz and Bowden, 2009) as estradiol is metabolized to estrone sulfate, resulting in increasing levels of estrone sulfate in the egg (Paitz and Bowden, 2013). This results in eggs having low levels of estradiol and high levels of estrone sulfate after the first third of development, which is when exogenous estradiol manipulations occur in most studies (Wibbels and Crews, 1995; Ge et al., 2018). Importantly, when estrone sulfate is applied either immediately after oviposition or after the first third of development, ovary development can be induced (Paitz and Bowden, 2013). Further, embryos from late season eggs, which have naturally higher concentrations of maternal estradiol and estrone sulfate, are more likely to develop into female hatchlings (Bowden et al., 2000; Carter et al., 2017). These findings demonstrate that estrone sulfate is capable of inducing ovary formation in a similar manner to estradiol and that both are present as maternally derived compounds.

While prior work has shown that both warm incubation temperatures and elevated maternal estrogens are capable of inducing ovarian development in embryos, the question of how, mechanistically, temperature and estrogens influence gonadal development remains. Much recent work has focused on a series of conserved genes that have been identified as playing a role in the sex-determining process. At present, we have a greater understanding of how temperature directly induces the male developmental pathway, a process that involves temperature-sensitive epigenetic regulation of genes necessary for testis development. Kdm6b is a histone demethylase which activates Dmrt1 and triggers the male sex-determining pathway (Ge et al., 2017; Ge et al., 2018). Kdm6b is temperature sensitive as it exhibits temperature-responsive intron retention, where an intron is retained at cool, male-producing temperatures (Deveson et al., 2017; Marroquín-Flores et al., 2021). While the functional consequence of intron retention in Kdm6b remains an open question, intron retention occurs at the temperature that promotes its expression, suggesting that intron retention does not impair Kdm6b function (Deveson et al., 2017; Marroquín-Flores et al., 2021). The temperature response of Kdm6b is likely regulated by CDC-like kinases (CLKs) that activate RNA binding proteins (via phosphorylation) at cool temperatures to promote intron retention (Haltenhof et al., 2020). We have previously shown that removal of the retained intron of Kdm6b occurs rapidly in response to a fluctuating female-producing temperature and results in an overall drop in expression (Marroquín-Flores et al., 2021). Thus, the effect of incubation temperature on the expression of Kdm6b appears to result from temperature-sensitive splicing that regulates epigenetic processes such as histone demethylation. At warmer, female-producing temperatures, FoxL2 and Cyp19A1 exhibit increased expression (Govoroun et al., 2004; Hudson et al., 2005; Pannetier et al., 2006; Batista et al., 2007; Bowden and Paitz, 2021), but whether this is due to a direct effect of warm temperature or simply the absence of antagonistic genes such as Dmrt1 remains unknown. It is known that exogenous estradiol, when applied at the beginning of the TSP, also effects the expression of many of the genes involved in gonadal differentiation (Ramsey and Crews, 2007; Matsumoto et al., 2013). Estradiol suppresses Kdm6b, Dmrt1 and Sox9 expression (Fleming and Crews, 2001; Murdock and Wibbels, 2006; Barske and Capel, 2010; Ge et al., 2018) and induces FoxL2 and Cyp19A1 (Ramsey and Crews, 2007; Matsumoto et al., 2013). Despite being abundant throughout development and capable of inducing ovary formation, the effect of estrone sulfate on the expression of genes involved in sex determination has not been explored, nor has the effect of estrone sulfate on Kdm6b splicing.

While estrogens and female-producing temperatures can both induce ovarian development, tissues produced under varying feminizing conditions can differ in gene expression (Ramsey and Crews, 2007; Barske and Capel, 2010; Canesini et al., 2018) and in morphology (Díaz-Hernández et al., 2015; Canesini et al., 2018), suggesting that developing gonads respond differently to these two stimuli. In order to determine whether gene expression and intron retention events are affected by female-producing conditions, we used a fluctuating incubation environment where embryos are transiently exposed to both male- and female-producing temperatures throughout development, mimicking conditions embryos may experience in wild populations (Carter et al., 2017, 2018; Breitenbach et al., 2020). In order to determine whether gene expression and intron retention events are affected by estrogens, we treated eggs with exogenous estrone sulfate and incubated them under male-producing temperatures. We hypothesize that the expression of sex-determining genes will differ in response to the feminizing effects of female-producing temperatures and exogenous estrone sulfate. Our approach enables us to decouple the effects of steroid metabolites and naturalistic incubation temperatures on gene expression and hatchling phenotype.

Trachemys scripta (Thunberg 1792) eggs were purchased separately from Concordia Turtle Farm, LLC (Jonesville, LA, USA) in May 2020 and May 2021 for two experiments. In both years, eggs were excavated the day they were laid and shipped overnight to the laboratory, where they were placed in moist vermiculite upon receipt and placed into temperature-controlled incubators within 36 h of being laid. Clutch identity was not available for shipped eggs. Egg collection was approved by the Louisiana Department of Agriculture and Forestry. All hatchling work was carried out in accordance with methods approved by the Illinois State University Institutional Animal and Care Use Committee (IACUC).

In 2020, 142 eggs were randomly sorted into one of three treatment groups to control for clutch identity (Fig. 1). The first treatment was to test for the effect of heat exposure on gene expression. The second treatment was to test for the effect of estrone sulfate on gene expression. The third treatment was a male-producing condition to serve as a control. Prior to incubation, eggs assigned to the estrone sulfate treatment were dosed topically with a 10 µl bolus of 70% ethanol containing 10 µg of estrone sulfate; all other eggs were given a 10 µl bolus of 70% ethanol as a control. All eggs were incubated for the first 24 days at a male-producing temperature of 25.0±3°C (IPP 110 Plus, Memmert GmbH+Co.KG, Schwabach, Germany). After incubation day 24, the eggs in the heat exposure group were shifted to a female-producing temperature of 29.5±3°C (IPP 400, Memmert GmbH+Co.KG) and remained at this female-producing temperature for the remainder of the study. All other eggs remained at the male-producing temperature.

Eggs in all treatment groups were sampled and staged (Greenbaum, 2002) at four points between incubation days 24 and 44. Sampling was designed to capture gene expression across the TSP (Greenbaum stages 15–21). Incubation temperature can affect the rate of embryonic development (Stubbs and Mitchell, 2018); thus, eggs were sampled based on developmental stage, not incubation day. Embryos were staged by a series of stage-specific indicators such as embryo size, eye disk development and shell ridge development (Stubbs and Mitchell, 2018). Embryonic gonads were dissected and placed into 1 ml of Trizol for RNA isolation (Fig. 1). Fifty-nine of the eggs not used for embryonic studies continued to incubate until hatching to determine sex ratios. Approximately 6 weeks post-hatching, hatchlings were euthanized and sexed by macroscopic examination of the gonads.

The expression of several genes involved in testis and ovary development was quantified using qPCR, including specific transcripts of Kdm6b that contain or lack an intron (Marroquín-Flores et al., 2021). Embryonic gonads stored in Trizol were homogenized and extracted using 2-propanol (Fisher Chemical) and chloroform (Marroquín-Flores et al., 2021). cDNA was synthesized to a standardized concentration of 40 ng µl⁻¹ using a Maxima First Strand cDNA Synthesis Kit (Thermo Scientific), following the manufacturer's protocol. PowerUp SYBR Green Master Mix (Applied Biosystems) was used for RT-qPCR to capture changes in gene expression for Kdm6b, Dmrt1, Sox9, FoxL2 and Cyp19A1. Three primers were used to capture the expression of Kdm6b: one set of primers, Kdm6bGe, was non-discriminating and amplified all Kdm6b transcripts (Ge et al., 2018); the other two primer pairs, Kdm6b(+IR) and Kdm6b(−IR), were designed to discern the intron-containing transcript and the spliced transcript, respectively (Marroquín-Flores et al., 2021). For each primer pair, gene expression was normalized using the housekeeping gene Gapdh, and relative expression was calculated using the ΔΔCT method (Rao et al., 2013). RNA was also extracted from hatchling gonads, as described above, to verify that developed gonads that were morphologically determined to be ovaries or testes were expressing genes consistent with those tissues. cDNA was synthesized for 8 hatchlings from each treatment to a standard concentration of 40 ng µl⁻¹ and RT-qPCR was used to determine the expression of Dmrt1 and FoxL2 in the gonads of hatchlings from each treatment (Fig. S1). All RT-qPCR reactions were performed in triplicate.

A follow up study was conducted to examine the effects of more physiologically relevant doses of estrone sulfate on gene expression. In 2021, 87 eggs were randomly sorted into three treatment groups, where eggs were dosed topically with a 10 μl bolus of 70% ethanol containing 150 ng of estrone sulfate, a 10 μl bolus containing 75 ng estrone sulfate, or a 10 μl bolus of 70% ethanol as a control. All eggs were incubated under a male-producing temperature of 25.0±3°C for the duration of the study. Eggs were sampled every 3 days, starting on incubation day 35 and ending on incubation day 47, to capture changes in gene expression at later stages of development (stages 18–21), when downstream testis- and ovarian-typical genes become upregulated. Embryos were staged at the time of sampling and gonads were dissected and placed into 1 ml of Trizol for RNA isolation. Tissues were prepared for RT-qPCR to capture changes in Kdm6b, Dmrt1, Sox9, FoxL2 and Cyp19A1, as described above. No eggs were allowed to hatch in 2021.

All statistical tests were conducted using R (http://www.R-project.org/). For study 1, embryos within the four sampling intervals were grouped by stage to establish four stage ranges for analysis. Most studies that establish markers for developmental stages have been conducted at constant temperature. Under fluctuating temperatures, many embryos exhibit traits associated with two stages, despite being sampled on the same day and under the same incubation conditions. For example, an embryo may exhibit serration of the digital plate, but no obvious digits (indicative of stage 17) while also having distinctive scutes and a lower eyelid (indicative of stage 18) (Greenbaum, 2002). Additionally, other embryos sampled on the same day and under the same incubation conditions may exhibit traits consistent with a single developmental stage. To account for these differences, embryos were grouped into the following stage ranges for analysis: stages 15–16, stages 17–18, stage 19 and stages 20–21. Gene expression was analyzed using a generalized linear model, which allowed us to specify distributions for data that were not normally distributed. For all analyses, data were transformed to better fit the assumptions of the model. To determine the effects of the treatment on gene expression, treatment and stage of development were used as fixed effects and estimated marginal means (R Package: emmeans) were used for post hoc comparisons among groups. Two individuals in the estrone sulfate-treated group and two individuals in the female-producing temperature group were underdeveloped at the time of sampling and were removed from the analysis. For study 2, gene expression was analyzed using a generalized linear model and data were transformed to better fit the assumptions of the model. Treatment (male-producing temperature, 75 ng estrone sulfate or 150 ng estrone sulfate) and sampling day were used as fixed effects and estimated marginal means were used for post hoc comparisons.

Both warm temperatures and estrone sulfate affected the pattern of gene expression but their specific effects differed. As previously demonstrated, Kdm6b expression was significantly lower when embryos experienced warm temperatures (Marroquín-Flores et al., 2021), but embryos treated with estrone sulfate did not exhibit reduced Kdm6b expression. When examining non-discriminate Kdm6b expression using the primers from Ge et al. (2018), we found that temperature had a significant effect on Kdm6b expression (χ²=14.3733, d.f.=2, P<0.001; Fig. 2A). Embryos that experienced the shift to female-producing temperature had significantly lower expression when compared with embryos at male-producing temperature (P<0.01) and those in the estrone sulfate treatment (P<0.01; Fig. 2A). There was also some variability in non-discriminate Kdm6b expression between stages (χ²=9.3689, d.f.=2, P<0.05; Fig. 2A), but after correction for multiple contrasts, the pattern was only marginally significant. For the intron-containing transcript of Kdm6b [the Kdm6b(+IR) transcript], we found an interaction effect between treatment and stage of development (χ²=28.263, d.f.=6, P<0.001). Expression of the Kdm6b(+IR) transcript was not responsive to the exogenous estrone sulfate but was responsive to incubation temperature by stages 17–18 (Fig. 2B). Embryos that experienced the shift to female-producing temperature had significantly lower expression of the Kdm6b(+IR) transcript compared with embryos at male-producing temperature (P<0.001) and embryos in the estrone sulfate treatment (P<0.001) in stages 17-–21 (Fig. 2B). Expression of the spliced Kdm6b(−IR) transcript lacking the intron did not respond to the treatment (χ²=1.4782, d.f.=2, P=0.4774) and did not change across stages of development (χ²=6.3443, d.f.=3, P=0.096; Fig. 2C).

Unlike Kdm6b, all other genes involved in gonadal differentiation responded to both the female-producing temperature treatment and treatment with estrone sulfate. For Dmrt1, we identified a significant interaction effect between treatment and stage of development (χ²=15.432, d.f.=6, P<0.05; Fig. 3). Across all stages of development, Dmrt1 expression was significantly lower in estrone sulfate-treated embryos when compared with that in embryos in the male-producing temperature treatment (P<0.001) and embryos in the female-producing temperature treatment (P<0.001). Dmrt1 expression was significantly reduced in embryos in the female-producing temperature treatment compared with that in embryos in the male-producing temperature treatment by stages 20–21 of development (P<0.001). Dmrt1 expression also increased in male-producing temperature-treated embryos between stages 15–16 and stages 17–18 (P<0.001). For Sox9, we also identified a significant interaction effect between treatment and stage of development (χ²=59.618, d.f.=6, P<0.001; Fig. S2). When compared with expression at the male-producing temperature, Sox9 expression was significantly lower in response to the female-producing temperature (P<0.001) by stages 20–21 and in response to the estrone sulfate treatment (P<0.001) by stages 19–21. Sox9 expression also increased in embryos in the male-producing temperature treatment between stage 19 and stages 20–21 of development (P<0.001).

FoxL2 expression was affected by both treatment and developmental stage (χ²=17.596, d.f.=6, P<0.01; Fig. S3A). FoxL2 expression was significantly higher in response to female-producing temperature (P<0.001) and treatment with estrone sulfate (P<0.01) by stages 20–21 of development, when compared with that of embryos incubated under the male-producing temperature. FoxL2 expression also increased in embryos in the female-producing temperature treatment between stage 19 and stages 20–21 of development (P<0.0001). Cyp19A1 expression was affected by both treatment and developmental stage (χ²=17.959, d.f.=6, P<0.01; Fig. S3B), where expression was significantly higher in response to female-producing temperature (P<0.001) and treatment with estrone sulfate (P<0.001) by stages 20–21 of development, compared with that in embryos incubated under the male-producing temperature. As evidence that our treatments resulted in the expected sex ratios, 94% of the embryos exposed to warm temperatures developed ovaries (15/16), 100% of the embryos treated with estrone sulfate developed ovaries (18/18), and 0% of the embryos incubated under cool temperatures developed ovaries (0/25).

When applied at lower doses meant to mimic levels found in late season eggs, estrone sulfate did not affect gene expression. There was no effect of estrone sulfate treatment on non-discriminate Kdm6b expression (χ²=1.41114, d.f.=2, P=0.4938; Fig. S4A). When isolating the intron-containing transcript Kdm6b(+IR), we found no effect of estrone sulfate treatment on expression (χ²=0.4893, d.f.=2, P=0.783; Fig. S4B). For the spliced Kdm6b(−IR) transcript lacking the intron, we saw a small effect of estrone sulfate treatment on expression, but this was only marginally significant after correction for multiple contrasts (χ²=8.0162, d.f.=2, P<0.05; Fig. S4C).

Dmrt1 expression was not affected by estrone sulfate treatment (χ²=1.001, d.f.=2, P=0.6063), but did change across the sampling period (χ²=76.363, d.f.=4, P<0.001), increasing between incubation day 38 and day 41 (P<0.001; Fig. S5A). Similarly, Sox9 expression was not affected by estrone sulfate treatment (χ²=1.018, d.f.=2, P=0.6012), but was by sampling period (χ²=53.054, d.f.=4, P<0.001), where Sox9 expression increased between incubation day 38 and day 41 (P<0.001; Fig. S5B). FoxL2 expression was also not affected by estrone sulfate treatment (χ²=1.7296, d.f.=2, P=0.4211) or sampling period (χ²=1.8090, d.f.=4, P=0.7708; Fig. S6A). Cyp19A1 expression was not affected by estrone sulfate treatment (χ²=2.8355, d.f.=2, P=0.2423; Fig. S6B), but was by incubation day (χ²=2.8355, d.f.=4, P<0.001). However, the effect of incubation day was eliminated when corrected for multiple comparisons.

In this study, we characterized the expression of sex-determining genes in response to naturalistic incubation temperatures and an abundant estrogen metabolite to decouple the effects of thermal and hormonal stimuli on gene expression and hatchling phenotype. We found that patterns of gene expression and intron retention differed in response to the feminizing effects of warm temperature and exogenous hormones. When applied at doses that mimic maternally derived levels, we found that estrone sulfate, in isolation, did not induce similar changes in gene expression. Our findings highlight the complexity of processes organizing development and the necessity of using experimental designs that mimic environmental conditions.

Temperature-sensitive intron retention has recently been identified as an important regulator of gene expression in species with TSD (Deveson et al., 2017; Georges and Holleley, 2018). We used multiple primers to characterize changes in Kdm6b expression and intron retention in response to temperature and hormone manipulations. We found that both warm temperature and estrone sulfate induced ovary development, but the mechanisms underlying this induction differ. We showed that exposure to female-producing temperature can largely eliminate intron retention in Kdm6b during the stages of development when sex is determined (Fig. 2B) and that this decrease in intron retention corresponds to downregulation of Kdm6b (Fig. 2A). However, estrone sulfate did not have an effect on Kdm6b expression or intron retention (Fig. 2), which is notable because other work has demonstrated that Kdm6b expression is sensitive to the effects of estradiol (Ge et al., 2018). Several factors could contribute to the differences between the effects of estradiol and estrone sulfate on Kdm6b. Estrone and estradiol are the products of the aromatizable androgens androstenedione and testosterone, respectively (Crews et al., 1996). Estrone sulfate is an estrone metabolite that can influence sexual development and is abundant in the yolk of T. scripta eggs during the thermal sensitive period of development (Paitz and Bowden, 2013), but the mechanism by which estrone sulfate induces femininization in T. scripta embryos is still unknown. It has been proposed that sulfonation enables uptake of estrogen sulfates by the embryo, which can then be converted back into estrogens to induce developmental changes (Pasqualini et al., 1986; Paitz et al., 2012, 2017; Paitz and Bowden, 2013), but this has not been experimentally tested. Our current and prior findings suggest that Kdm6b responds to changes in incubation temperature (Marroquín-Flores et al., 2021). Estrone sulfate, however, did not affect expression or splicing of the Kdm6b transcript.

Unlike Kdm6b, Dmrt1 and Sox9 exhibited a similar response to warm temperature and estrone sulfate. We found that Dmrt1 was downregulated in response to both estrone sulfate and female-producing temperature. Dmrt1 exhibited a particularly strong response to estrone sulfate, where expression was lower following estrone sulfate treatment than in response to female-producing temperature (Fig. 3). In response to male-producing temperatures, Dmrt1 expression increased at later stages of development, consistent with canalization to the male pathway (Fig. 3; Shoemaker-Daly et al., 2010, Ge et al., 2017, Breitenbach et al., 2020). Sox9 exhibited similar responses to the female-producing temperature and estrone sulfate treatments, which supports prior work on the effect of temperature and estrogens on Sox9 expression (Fig. S2; Shoemaker-Daly et al., 2010; Matsumoto et al., 2013; Breitenbach et al., 2020). When we looked at genes involved in ovarian development, both FoxL2 and Cyp19 responded to the estrone sulfate and female-producing temperature treatments (Fig. S3). While prior work has demonstrated that FoxL2 and Cyp19 expression is sensitive to the effects of estradiol (Matsumoto and Crews, 2012; Matsumoto et al., 2013), we showed that both FoxL2 and Cyp19 are also sensitive to estrone sulfate. Estrone sulfate induced the expression of FoxL2 and Cyp19 in eggs incubated under male-producing temperatures to levels similar to those in eggs incubated under female-producing temperatures. Similar to prior work, our fluctuating warm temperature treatment also resulted in increased expression of FoxL2 and Cyp19 at later stages of development (Willingham et al., 2000; Murdock and Wibbels, 2003; Matsumoto and Crews, 2012; Breitenbach et al., 2020). These findings suggest that warm temperature exposure and estrone sulfate can both induce ovarian differentiation by impacting downstream gene expression.

While our data show that high doses of estrone sulfate can alter gene expression and induce ovarian differentiation, doses that more closely mimic natural concentrations of estrone sulfate did not affect the expression of any of the genes studied. Trachemys scripta eggs contain a variety of maternal estrogens, including both estradiol and estrone sulfate (Paitz and Bowden, 2010, 2013; Carter et al., 2018). Temperature and estrogens exhibit a synergism, where late season eggs have higher concentrations of estradiol and estrone sulfate and are more likely to produce female hatchlings at intermediate temperatures (Wibbels et al., 1991b; Paitz and Bowden, 2010, 2013; Carter et al., 2018). In our second study, we evaluated the effects of estrone sulfate by itself at physiologically relevant concentrations and did not see an effect on gene expression. It is worth noting that the endogenous levels of estrone sulfate in treated eggs were not quantified, and it is possible that the amount of estrone sulfate applied exogenously may represent only a fraction of what the embryo received. It is also possible that lower doses of estrone sulfate may have reduced potency and be less likely to induce changes at the molecular level (Crews et al., 1991; Crews et al., 1996; Sheehan et al., 1999). Additionally, exogenous estrogens have different dosage effects when they are applied individually and in combination, even when used at low concentrations (Bergeron et al., 1999). Applied exogenously, 150 ng of estrone can induce female development in 30% of T. scripta embryos, but when applied in conjunction with 75 ng of estradiol or 10 ng of estriol, these estrogens can induce 40% and 50% of embryos to develop as female, respectively (Bergeron et al., 1999). It is likely that the seasonal shift in sex ratios is mediated by a synergism between multiple maternally derived estrogens, rather than in response to a single estrogen.

In this study, we showed that estrone sulfate can impact embryonic development by affecting the expression of sex-determining genes. These results are further supported by our hatchling data, which show that both female-producing temperature and estrone sulfate can lead to gonads that are morphologically female and that hatchling gonads exhibit patterns of gene expression consistent with ovarian differentiation (Fig. S1). Few studies have examined gene expression in response to maternal estrogens or in response to naturalistic incubation temperatures, and to the best of our knowledge, this study is the first to characterize the expression of sex-determining genes in response to an estrogen metabolite. In T. scripta, estradiol is present at the time of oviposition but declines prior to gonadal differentiation and is converted into various estrogen sulfates (Paitz and Bowden, 2009, 2011, 2013; Paitz et al., 2012). The decline of estradiol early in development corresponds to increases in estrone sulfate during the TSP, which leads to the development of ovaries (Paitz and Bowden, 2009, 2013; Paitz et al., 2012). While the exact mechanism has not yet been identified, steroid sulfates appear to play an important role in moderating the effects of maternal steroids on embryonic development (Paitz and Bowden, 2008, 2013). Our data further support this by showing that estrone sulfate can elicit its effects on the embryo by impacting the expression of sex-determining genes in ways that are similar to the effects of estradiol.

Taken together, our findings suggest that development can be differentially regulated by biotic and abiotic conditions. We found that Kdm6b responds directly to changes in incubation temperature through temperature-responsive intron retention, which enables Kdm6b to regulate the expression of Dmrt1 via epigenetic modification. In both T. scripta and the American alligator (Alligator mississippiensis), Kdm6b exhibits temperature-responsive intron retention, but intron retention in Kdm6b leads to different phenotypic outcomes in these species (Deveson et al., 2017). Our findings may provide insight into mechanisms regulating gonadal development, given our results that estrogens can also regulate Dmrt1, independent of the effects of Kdm6b. This shows that Dmrt1 can be impacted by both changes in incubation temperature and maternal steroids, which may explain seasonal variability in hatchling sex ratios (Bowden et al., 2000; Carter et al., 2017). Our findings highlight the value of using environmentally relevant incubation and hormone treatments. While the use of naturalistic laboratory incubations has become more common, TSD is most frequently studied using constant incubation temperatures. However, we know that incubating embryos experience a range of temperatures that regularly fluctuate between male and female conditions during development (Carter et al., 2017). The range of temperatures that embryos experience and the duration of exposure can differentially impact phenotypic characteristics such as body size, immune function, gene expression and sex (Les et al., 2007, 2009; Bowden et al., 2014; Carter et al., 2017, 2017; Breitenbach et al., 2020). The interplay between incubation temperature and maternal estrogens (Wibbels et al., 1991b; Crews et al., 1994; Bergeron et al., 1999) underlies the complexity of development in species with TSD. Investigations using treatments that more closely mimic environmental conditions allow us to more accurately characterize the mechanisms organizing gonadal development in species with TSD and should be the focus of future work.

We would like to thank Dr Anthony Breitenbach for assistance with egg sampling and hatchling husbandry and Dr Ben Sadd for guidance in coding and statistics.