Introduction

Sex determination in mammals is chromosomally controlled with males and females carrying the XY and XX sex chromosomes, respectively. In most mammals, the Y chromosomal Sry gene triggers the fate of the bipotential gonad to develop into a testis rather than into an ovary resulting in male development (Sinclair et al. 1990; Koopman et al. 1991). The Sry gene evolved before the divergence of marsupials and placental mammals around 144–168 million years ago, and thus is absent in monotremes such as in platypus (Wallis et al. 2007, 2008). The SRY protein is the founder member of the SOX (SOX = SRY-related high mobility group (HMG)-Box) family of transcription factors which share at least 50% homology within their DNA-binding and bending HMG domain (Bowles et al. 2000; Schepers et al. 2002). Despite its pivotal role in mammalian male sex determination, a handful of rodent species have been identified who apparently determine sex in the absence of Sry and a Y chromosome. These include the mole voles Ellobius lutescens and Ellobius tancrei (Matthey 1958; Just et al. 1995; Vogel et al. 1998) and the spiny rats Tokudaia osimensis and Tokudaia tokunoshimensis (Soullier et al. 1998). The loss of the Y chromosome in these two genera can be considered as independent events because Ellobius and Tokudaia belong to different subfamilies, to Arvecolinae and Murinae, respectively. Their sex-determining mechanisms, i.e. which gene acts as the sex-determining switch, remain elusive. However, it has been demonstrated recently that males of both Sry-negative Tokudaia species have additional copies of Cbx2, a gene acting upstream of Sry (Katoh-Fukui et al. 1998), suggesting that CBX2 might be involved in male sex determination in Tokudaia (Kuroiwa et al. 2011). In E. lutescens, several genes have been excluded as the sex-determining switch genes namely Dax1, Sf1, Foxl2/Pisrt1, Sox9, Sox3 and Dmrt1 (reviewed in Just et al. 2007).

The genus Ellobius contains at least five species, namely Ellobius fuscocapillus, E. lutescens, the sister species E. tancrei and Ellobius talpinus, and Ellobius alaicus (Just et al. 2007; Romanenko et al. 2007). The only species of this genus with an XY/XX karyotype and for which the presence of the Sry gene could be demonstrated is E. fuscocapillus (Just et al. 1995). E. lutescens has an odd number of chromosomes, with both sexes having the karyotype 2n = 17,X (Matthey 1958). The sister species E. tancrei and E. talpinus have an even number of chromosomes ranging from 2n = 32,XX to 2n = 54,XX and a constant number of 2n = 54, respectively, in both sexes (Kolomiets et al. 1991; Romanenko et al. 2007). Since the Sry gene is present in E. fuscocapillus, it can be assumed that the common ancestor of these Ellobius species also possessed Sry, but that Sry was lost shortly after or during the speciation of Ellobius.

Evolution of new species can be driven by various genetic mechanisms, including gene mutations and gene duplications resulting in novel protein functions, gene loss and mutations in cis-regulatory elements altering gene regulation. A recent breakthrough in the field of mammalian sex determination was the finding that in the mouse, SRY up-regulates the Sertoli cell expressed gene Sox9 through direct binding to a 1.3-kb cis-regulatory element, termed TESCO (testis-specific enhancer of Sox9 core), which is located 13 kb upstream of Sox9 (Sekido and Lovell-Badge 2008). Like Sry, Sox9 is both required and sufficient for male sex determination, for example, XX transgenic mice over-expressing Sox9 develop as infertile males (Foster et al. 1994; Wagner et al. 1994; Vidal et al. 2001; Chaboissier et al. 2004; Barrionuevo et al. 2006). It is therefore possible that Sox9 is the only important target for SRY in the developing testis. The discovery of the testis-specific enhancer of Sox9 now opens the possibility to screen Sry-negative mammals for sequence variations in TESCO. This analysis might reveal clues as to which transcription factors bind to TESCO in those species and might shed light on the fundamentals of SRY–TESCO interaction. The genus Ellobius is a bona fide model for such analyses since closely related species exist with Sry (E. fuscocapillus) and without Sry (E. lutescens).

Materials and methods

Cloning of Ellobius TESCO sequences

The 2.2-kb fragment of E. lutescens was amplified by polymerase chain reaction (PCR) using the human forward primer hSox9TE2F: 5′-TTAGCAGAAATCAGCTGTAATA-3′ and the mouse reverse primer mSox9TE1R: 5′-CCTTTAGGGGTAAAAACC-3′ (Fig. 1a). PCR reactions were performed in a total volume of 25 μl, using 50 ng DNA sample per reaction. Cycling conditions were 95°C for 3 min, and 35 cycles at 95°C (30 s), 46°C (30 s) and 72°C (3 min). The evolutionarily conserved region (ECRs; 200 bp genomic fragments) of E. lutescens, E. tancrei, E. talpinus and E. fuscocapillus were amplified by PCR using human forward primer hSox9TE2F: 5′-TTAGCAGAAATCAGCTGTAATA-3′ and the E. lutescens reverse primer eSox9TE3R: 5′-CCTCCCTGTTGTTGGTAGCTGCC-3′ (Fig. 1a). Cycling conditions were 95°C for 3 min, and 35 cycles at 95°C (30 s), 57°C (30 s) and 72°C (1 min). The 2.2-kb E. lutescens and the 200-bp E. fuscocapillus fragments were cloned into the pGEM®T Easy vector (Invitrogen) according to the manufacturer's instructions.

Fig. 1
figure 1

Amplification of Ellobius TESCO sequences by genomic PCR. a Amplification of a 2.2-kb genomic fragment from Ellobius lutescens by PCR using human primer hSox9TE2F and mouse primer mSox9TE1R. Nucleotides which are conserved in at least two species are in red. b Amplification of a 200-bp genomic fragment from Ellobius fuscocapillus by PCR using the human primer hSox9TE2F and the E. lutescens primer eSox9TE3R

Full size image

DNA sequence analysis

Four pGEM®T Easy clones containing the 2.2-kb E. lutescens fragment and two pGEM®T Easy clones containing the 200-bp E. fuscocapillus fragment were sequenced with standard vector primers Sp6 and T7, using an Applied Biosystems 3130xl Genetic Analyzer fitted with an 80-cm array to generate read lengths of approximately 1,000 bases. The four pGEM®T Easy clones containing the 2.2-kb E. lutescens fragment were also sequenced with the internal primers NestF: 5′-AGCAAGGCAGGACTCAGACA-3′ and NestR: 5′-ATCCGGTCCAGCATTCACCT-3′. The ECRs (200 bp PCR fragment) from five E. lutescens, two E. tancrei and three E. talpinus individuals were amplified and sequenced using primers hSox9TE2F and eSox9TE3R.

Identification of transcription factor binding sites

To identify putative transcription factor binding sites within the ECRs of E. lutescens, E. tancrei and E. fuscocapillus, we used the online program MatInspector (www.genomatix.de) with core and matrix similarity set to 1.00 and optimized, respectively. The MatInspector library at the time of analysis was the Matrix Family Library version 8.4 (June 2011).

Site-directed mutagenesis

The TESCO-E1b-Luc reporter was constructed by cloning the mouse 1.3-kb testis-specific enhancer of Sox9 (TESCO) (Sekido and Lovell-Badge 2008) by PCR into the E1b-luciferase reporter (Bernard et al. 2008). The TESCO-Δ14-E1b-Luc construct was generated by site-directed mutagenesis (Stratagene Kit) of the TESCO-E1b-Luc construct using primers SOXDELMF: 5′-CACAAAATAACAATGCCTTCTGGCTAAGAAAGAGAAGACTCC-3′ and SOXDELMR: 5′-GGAGTCTTCTCTTTCTTAGCCAGAAGGCATTGTTATTTTGTG-3′ according to the manufacturer's instructions.

In vitro luciferase assays

Human embryonic kidney carcinoma (HEK293T) cells were cultured at 37°C with 5% CO2 in Dulbecco's modified Eagle's medium (DMEM), high glucose, GlutaMAX media (Invitrogen) containing 10% foetal bovine serum, 1% sodium pyruvate and 1% penicillin-streptomycin. For in vitro luciferase assays, 30,000 cells were seeded into each well of a 96-well tissue culture plate 24 h prior to transfection. Cells in each well were co-transfected with the reporter construct TESCO-Δ14-E1b-Luc (10 ng), TESCO-E1b-Luc (10 ng) or the empty vector E1b-Luc (8 ng), together with 40 ng of each of the expression constructs pCDNA3-SF1, pCDNA3-SRY or pCDNA3-SOX9. pRL-TK-Renilla (Promega; 1 ng) was included as an internal control. Total DNA amount was made up to 100 ng per well using the plasmids pCDNA3 and pUC. The cells were transfected with FuGENE6 Transfection Reagent (Roche) following the manufacturer's instructions. Cell lysis was performed 46 h after transfection, and firefly and Renilla luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega). Firefly luciferase activity (Luc) was normalised against that of Renilla luciferase (Ren). Five independent assays were performed, each in triplicate. Paired t tests were performed for statistical analysis.

Results

Cloning of the ECRs of E. lutescens, E. tancrei, E. talpinus and E. fuscocapillus

Previously, we identified an ECR of 180 bp within the 1.3-kb mouse TESCO sequence present in mammals, birds, reptiles and amphibians (Bagheri-Fam et al. 2010). The ECR contains five highly conserved modules (ECRi-v) with putative binding sites for SOX, T-cell factor (TCF)/lymphoid enhancer factor-1 (LEF), GATA and other transcription factors. These modules were spared from sequence variation throughout evolution, implying that they might be important for Sox9 regulation in the testis. Indeed, module ECRii contains the SOX site R5 (AACAAT), one of three SOX sites important for SRY/SOX9-mediated activation of TESCO in vitro and in vivo in the mouse (Sekido and Lovell-Badge 2008). We aimed to clone and sequence the ECR from E. lutescens (Sry-negative) and E. fuscocapillus (Sry-positive), which might reveal clues how Sry was lost in E. lutescens. Since genomic sequences from Ellobius are currently not available, we designed primer pairs within human and mouse TES for genomic regions highly conserved across species. Using the primer pair hSox9TE2F/mSox9TE1R, we were able to amplify a 2.2-kb genomic fragment from E. lutescens by PCR (Fig. 1a), which was subsequently cloned into the pGEM®T Easy vector for sequencing. The 2.2-kb genomic fragment contains the ECR and includes all but the first 0.2 kb of mouse TESCO. Comparison between the E. lutescens sequence and mouse TESCO using the online tool CLUSTAL (http://www.ch.embnet.org/software/ClustalW.html) revealed 75% sequence identity over 1,135 bp (Fig. S1). Over the same region, the evolutionarily more closely related species mouse and rat and the evolutionarily more distantly related species mouse and human show 83% and 60% sequence identity, respectively (data not shown). This demonstrates that TESCO is highly conserved in E. lutescens despite the loss of Sry. We were not able to amplify the corresponding 2.2-kb genomic fragment of E. fuscocapillus using primer pair hSox9TE2F/mSox9TE1R, possibly due to primer mismatches. However, a 200-bp genomic fragment containing the E. fuscocapillus ECR was successfully amplified by PCR (Fig. 1b) using the human primer hSox9TE2F and the E. lutescens specific primer eSox9TE3R. Using the same primer pair, we also amplified the ECRs from the Sry-negative species E. tancrei and of the closely related E. talpinus which were identical in sequence (Fig. 2).

Fig. 2
figure 2

Sequence alignment of the Ellobius ECRs with those of other vertebrates. ECRi to ECRv (highlighted in yellow) are five highly conserved modules which were described in Bagheri-Fam et al. (2010). Transcription factors predicted to bind to these modules and the SOX site R5 are shown (Sekido and Lovell-Badge 2008; Bagheri-Fam et al. 2010). Sequence variations in Ellobius of nucleotides conserved across vertebrates (asterisks) are highlighted in cyan. Note that the highly conserved module ECRiii is absent in all Ellobius species analyzed due to a 14-bp deletion

Full size image

Identification of a 14-bp deletion (Δ14) within the ECR of Ellobius

The Ellobius ECR sequences spanning from ECRi to ECRv were aligned with those of human, mouse, rat, opossum, platypus, chicken, lizard and frog (Bagheri-Fam et al. 2010) (Fig. 2). The ECRs of E. lutescens/E. fuscocapillus and of E. tancrei/E. fuscocapillus share a sequence identity of 94% and 97%, respectively, while the mouse/rat and mouse/human ECR sequences are 88% and 81% identical, respectively (Table 1). This shows that the ECRs between the Sry-negative Ellobius species and the Sry-positive E. fuscocapillus are highly conserved with no major sequence rearrangements (Fig. 2).

Table 1 DNA sequence identities (percent) calculated between the different mammalian ECRs
Full size table

Notably, all Ellobius species carry a large deletion within the ECR (14 bp when compared to mouse/rat (Δ14) and 15 bp when compared to other vertebrates), which removes most of the evolutionarily conserved module ECRiii including its SOX/TCF site, as well as additional 3′-flanking sequences (Fig. 2, highlighted in cyan). In the remainder of module ECRiii, E. lutescens shows a unique T to C change. The 14-bp deletion in Ellobius is an intriguing finding since no deletions in module ECRiii are found in the ECR sequences of 41 vertebrate species (including 37 mammalian species) obtained from the nucleotide collection (nr/nt), high throughput genomic sequences (htgs) and whole-genome shotgun reads (wgs) databases at the National Centre of Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov). Moreover, only two of the 37 mammalian species show sequence variation in ECRiii, namely the Chinese hamster (CTTTCAG to CTTTGGA) and the bottlenose dolphin (CTTTCAG to CTTTTAG). Another noteworthy sequence variation present in all Ellobius species is a C to A change in module ECRv.

Only E. fuscocapillus shows an A to G change in module ECRii, predicted to disrupt the SOX site R5 (Fig. 2, highlighted in cyan) which is important for SRY and SOX9-mediated activation of TESCO in the mouse (Sekido and Lovell-Badge 2008). In all Ellobius species, no sequence variation was found in modules ECRi (SOX/TCF site) and ECRiv (GATA site; Fig. 2).

Sequence variation in the Ellobius ECRs alter the prediction of potential transcription factor binding sites

To investigate whether the Ellobius sequence variations in modules ECRii, ECRiii and ECRv influence the prediction of potential transcription factor binding sites, we used the online program MatInspector (www.genomatix.de) with core and matrix similarity set to 1.00 and optimized, respectively. As mentioned above, Δ14 removes most of ECRiii in all Ellobius species analyzed, and E. lutescens contains an additional T to C change (Fig. 2). Due to these sequence changes, MatInspector predicted the generation of new transcription factor binding sites in the ECRs of E. lutescens and E. fuscocapillus, namely an E-twenty-six (ETS) site in E. lutescens and a binding site for Heat transcription factors in E. fuscocapillus (Fig. 3, marked in red). The single nucleotide change (A to G) in module ECRii of E. fuscocapillus leads to loss of the predicted SOX site R5 (Fig. 3, marked in blue), while new predicted binding sites are generated for MYBL and HZIP (Fig. 3, marked in red). No transcription factor binding sites were predicted for module ECRv (data not shown).

Fig. 3
figure 3

Putative transcription factor binding sites within the ECRs of human, mouse, Ellobius lutescens, Ellobius tancrei and Ellobius fuscocapillus. Transcription factor binding sites were predicted by the online program MatInspector (www.genomatix.de) with core and matrix similarity set to 1.00 and optimized, respectively. Transcription factor binding sites (1) marked by asterisks were identified by visual inspection (Bagheri-Fam et al. 2010), (2) highlighted in blue are located within the highly conserved modules ECRi to ECRiv and (3) highlighted in red are generated in Ellobius due to sequence changes in the highly conserved modules ECRii and ECRiii. Δ14 represents the 14-bp deletion found in all Ellobius species analyzed

Full size image

Δ14 increases basal activity and SF1-mediated activation of mouse TESCO in HEK293T cells

Previously, we have shown that in the presence of SF1, both SRY and SOX9 can synergistically activate mouse TESCO (Sekido and Lovell-Badge 2008). Since Δ14 removes a highly conserved potential SOX/TCF binding site (Fig. 2), we speculated that Δ14 might alter SRY and/or SOX9-mediated activation of mouse TESCO. To test this, we introduced Δ14 into mouse TESCO (TESCO-Δ14) by site-directed mutagenesis and cloned both TESCO and TESCO-Δ14 into a luciferase reporter vector. Human SF1, SRY and SOX9 expression plasmids were co-transfected with either the TESCO or TESCO-Δ14 reporter construct into the human embryonic kidney 293 (HEK293T) cell line (n = 5). We found that SF1 alone and SF1 together with SRY activated TESCO-Δ14 significantly higher (~1.4-fold) than wild-type TESCO (Fig. 4a). In contrast, SOX9 together with SF1 activated TESCO and TESCO-Δ14 to a similar extent (Fig. 4a). We also noted that basal activity of TESCO-Δ14 was significantly higher (~2.5-fold) than that of wild-type TESCO in HEK293T cells (Fig. 4b). Taken together, these data indicate that Δ14 leads to increased activity of TESCO.

Fig. 4
figure 4

Δ14 increases SF1 and SF1-SRY-mediated activation and basal activity of mouse TESCO in HEK293T cells. a HEK293T cells were transiently co-transfected with E1b promoter-driven luciferase reporter constructs containing either intact TESCO or TESCO containing the 14-bp deletion found in Ellobius (TESCO-Δ14). The empty E1b-Luc vector was used as a negative control. Expression plasmids bearing human SF1, SRY and SOX9 were also co-transfected into the cells. b HEK293T cells were transiently co-transfected with E1b promoter-driven luciferase reporter constructs containing either intact TESCO or TESCO containing the 14-bp deletion found in Ellobius (TESCO-Δ14). The empty E1b-Luc vector was used as a negative control. Means of normalised luciferase readings and standard error of the mean from five independent assays are shown. Paired t tests were performed for statistical analysis. *P < 0.05; **P < 0.01

Full size image

Discussion

Through a PCR-based approach using degenerate primers, we were able to clone 1.1 kb of E. lutescens TESCO (including the entire ECR) which shares 75% sequence identity with mouse TESCO. This shows that TESCO is present in this species with high sequence conservation despite the loss of Sry. We did not expect a complete loss of TESCO in E. lutescens since TESCO is controlled by additional important transcription factors including SF1, SOX9 (SOX9 auto-regulation) and FOXL2 (SOX9 repression) (Sekido and Lovell-Badge 2008; Uhlenhaut et al. 2009). Moreover, TESCO sequences are conserved in non-mammalian species including chicken, lizard and frog (Bagheri-Fam et al. 2010) which all lack Sry indicating that TESCO might be important for testicular Sox9 expression independent of the sex-determining switch mechanism. It is thus possible that Sox9 expression in E. lutescens is still regulated via TESCO by factors such as SOX9, SF1 and FOXL2 and potentially also by the new sex-determining switch gene which remains to be identified.

We could also clone the ECR of E. tancrei (which also lacks Sry) and of the Sry-positive species E. fuscocapillus spanning all evolutionarily conserved modules (ECRi-ECRv) which allowed us to compare the Ellobius ECR sequences with each other and to other vertebrate ECRs. All three Ellobius species carry a 14-bp deletion (Δ14) removing module ECRiii and additional 3′-flanking sequences. As a result, a putative SOX/TCF site is removed in all three Ellobius species, and new transcription factor binding sites are created, an ETS binding site in E. lutescens and a Heat protein binding site in E. fuscocapillus. The only sequence difference between the Sry-positive species E. fuscocapillus and the two Sry-negative Ellobius species within the highly conserved ECR modules is the A to G change in module ECRiii predicted to abolish SRY binding to the bona fide target site R5 (Sekido and Lovell-Badge 2008).

The presence of Δ14 in all Ellobius species is an intriguing finding since it shows that this deletion occurred before the speciation of Ellobius and thus before the loss of Sry (Fig. 5). So, could there be a link between Δ14 and the evolution of a Sry-independent sex-determining mechanism in the genus Ellobius? Our in vitro data show that Δ14 increased both basal activity and SF1-mediated activation of mouse TESCO. One potential explanation for this observation is that Δ14 removes a predicted TCF binding site which might mediate repression of TESCO. In the XY mouse gonad, over-expression of beta-catenin (which associates with TCF transcription factors to regulate target genes) leads to loss of SOX9 expression and male to female gonadal sex reversal (Maatouk et al. 2008) possibly by reducing SF1 binding to TESCO (Bernard et al. 2011). Sox9 expression is initiated by SF1 in both sexes and is then up-regulated and maintained in XY gonads by SRY and SOX9, respectively, eventually leading to testis development (Sekido and Lovell-Badge 2008). Based on our functional assay, one could speculate that Δ14 might have raised Sox9 expression in the gonads of the common ancestor of the Ellobius species above a threshold level allowing SOX9 to maintain its own expression. This would lead to female-to-male sex reversal in XX individuals and thus to a destabilization of the XY/XX sex-determining mechanism. Moreover, in XY individuals, the expression of Sox9 and thus male development would become less dependent on or even independent of Sry (Fig. 5). This effect could be exacerbated through the inhibitory action of SOX9 on Sry transcription (Chaboissier et al. 2004; Barrionuevo et al. 2006). In support for such a scenario, XX transgenic mice over-expressing SOX9 develop as infertile males demonstrating that SOX9 is sufficient to induce maleness in the absence of Sry (Vidal et al. 2001). Evolutionary pressure for survival of such a proto-Ellobius species with predominantly male individuals could have resulted in speciation during which E. lutescens/E. tancrei and E. fuscocapillus independently stabilized their sex determination mechanisms, utilizing Sry-independent and Sry-dependent approaches, respectively (Fig. 5). In the former case (E. lutescens/E. tancrei), a female sex-determining gene might have evolved to down-regulate Sox9 expression specifically in female gonads. In the latter case, E. fuscocapillus could have prevented the loss of Sry by restoring Sox9 expression to normal levels in XX and XY gonads such as by loss of the important SRY/SOX binding site R5 (Figs. 2 and 3) thereby weakening TESCO activity. In such a scenario, the dependence of Sox9 expression and male development on Sry would be restored (Fig. 5).

Fig. 5
figure 5

Putative model for the evolution of a Sry-independent sex-determining mechanism in Ellobius lutescens and Ellobius tancrei. A 14-bp deletion (Δ14) occurred in the common ancestor of E. lutescens, E. tancrei and Ellobius fuscocapillus. Δ14 might have increased Sox9 expression in XX gonads thereby destabilizing the sex-determining mechanism in Ellobius. Subsequent evolutionary pressure for survival of this species could have resulted in speciation during which E. fuscocapillus and the common ancestor of E. lutescens and E. tancrei independently stabilized their sex determination mechanisms, utilizing Sry-dependent and Sry-independent approaches, respectively

Full size image

There are various possible new sex-determining mechanisms to replace Sry in Ellobius. For example, in the Sry-negative Tokudaia species, additional copies of Cbx2 might trigger male sex determination (Kuroiwa et al. 2011). A potential link between a mutation in the testis-specific enhancer of Sox9 and the evolution of a Sry-independent sex-determining mechanism is an intriguing hypothesis which remains to be tested.