Mechanism of asymmetric ovarian development in chick embryos

The vertebrate body plan is organized by three orthogonal axes: the anterior-posterior, dorsal-ventral and left-right (L-R) axes. A complex series of genetic interactions controls proper L-R axis development, which leads to later L-R asymmetry of the visceral organs. This process is evolutionarily conserved in vertebrates and relies upon genes encoding transcription factors and growth factors (Levin,2005; Raya and Belmonte,2006). PITX2, a member of the conserved bicoid-type homeobox gene family (Gage et al.,1999), has been implicated in the establishment of L-R asymmetry through its expression in the left lateral plate mesoderm. Indeed, Pitx2-knockout mice exhibit abnormal visceral organ asymmetry(Lin et al., 1999; Lu et al., 1999).

In mammals, the gonads develop bilaterally through the orchestrated action of a number of genes (Ross and Capel,2005; Swain and Lovell-Badge,1999). Many of these genes have been identified through the study of knockout mice and patients suffering from gonadal developmental abnormalities. However, no gonad defects identified so far have exhibited clear laterality, and no genes involved in gonad development are expressed with L-R asymmetry. Unlike mammals, most female birds develop ovaries only on the left side, while males develop bilateral testes. During early developmental stages before sexual differentiation, chick embryonic gonads show no obvious morphological L-R asymmetry and consist of two components, the cortex and medulla (Smith and Sinclair,2004). After sexual differentiation, testicular development occurs bilaterally in the male (genetically ZZ). The testicular cords appear in the medulla where Sertoli and Leydig cells differentiate, while the cortex regresses and eventually disappears. In female birds (genetically ZW), the left cortex proliferates and develops into the ovary, whereas the right cortex disappears (Smith and Sinclair,2004). Studies of chick gonad development implicate estrogen in asymmetric ovarian development. Indeed, estrogen receptor α(ERα) is expressed in the left but not the right cortex(Nakabayashi et al., 1998) of both sexes (Andrews et al.,1997). However, it has not been known how asymmetric ERα expression is induced and whether or not asymmetric estrogen signaling is related to asymmetric ovarian development.

Retinoic acid (RA) plays a variety of roles throughout development(Niederreither et al., 1999; Sakai et al., 2001) and functions by binding to the nuclear receptors RAR and RXR to regulate gene expression. The synthesis and distribution of RA is controlled by the coordinated expression of genes encoding RA-synthesizing retinaldehyde dehydrogenases (RALDH1, RALDH2 and RALDH3)(Zhao et al., 1996; Niederreither et al., 1997; Swindell et al., 1999; Mic et al., 2000; Niederreither et al., 2002)and RA-metabolizing cytochrome P450 proteins (CYP26A1, CYP26B1 and CYP26C1) (Fujii et al.,1997; Swindell et al.,1999; MacLean et al.,2001; Tahayato et al.,2003; Reijntjes et al.,2004). Because RA is involved in cell growth and differentiation,cell-cycle accelerator and suppressor genes have been studied as potential RA targets (Chen and Ross, 2004; Dragnev et al., 2004; Guidoboni et al., 2005).

Here, we investigate the roles of RALDH2, PITX2, ERα and Ad4BP/SF-1 (Ad4-binding protein/steroidogenic factor 1) in L-R asymmetric ovarian development in chicks. We demonstrate that a genetic cascade including these factors differentially regulates cyclin D1 gene expression in the left and right cortices and eventually causes asymmetric cell proliferation.

Chick cDNA was supplied as follows: Ad4BP/SF-1 from Dr Sutou(Kudo and Sutou, 1997); RALDH2 and CYP26A1 from Dr Eichele(Swindell et al., 1999); RALDH1 and RALDH3 from Dr Noda(Suzuki et al., 2000); RARβ from Dr Nohno (Nohno et al., 1991); CYP26B1 (clone ID: ChEST239e5), RXRα (clone ID: ChEST142D11), RXRβ (clone ID:ChEST102B17), RXRγ (clone ID: ChEST248B14), LHX9(clone ID: ChEST664O12) and cyclin D1 (clone ID: ChEST223D23) from the Medical Research Centre Geneservice (Cambridge, UK); and RARα (clone ID: pgp1n.pk009.m12) from the Delaware Biotechnology Institute, University of Delaware (Newark, DE). cDNA encoding mouse Pitx2c was supplied by Dr Semina(Semina et al., 1996). Chick cDNA for PITX2a was amplified by 5′-RACE (Invitrogen, Carlsbad,CA) with the primer shown in Table 1. Chick cDNA for PITX2c, RARγ and ERα was amplified with primers listed in Table 1. As the probe for PITX2a contains a region common to PITX2b, the probe detects both PITX2a and PITX2b. These cDNAs were cloned into pCR II-TOPO vector (Invitrogen) and digoxigenin-labeled probes were prepared according to the manufacturer's instructions (Roche Diagnostics, Penzberg,Germany). Whole-mount in situ hybridization was performed as described(Yoshioka et al., 1998). After staining, the embryos were embedded in 2% gellan gum followed by sectioning with a microslicer (Dosaka EM, Kyoto, Japan).

Differentiation stages are defined according to Hamburger and Hamilton(Hamburger and Hamilton,1951). Three embryos at stage 27 [embryonic day 5-5.5 (E5-5.5)]and 29 (E6-6.5) were fixed with 4% paraformaldehyde (PFA)/PBS and cross-sectioned. Five sections containing the gonad were randomly selected for each embryo and were labeled with 4′,6-diamidino-2-phenylindole (DAPI)and mouse anti-cytokeratin monoclonal antibody (1:100; Lab Vision Corp., part of Thermo Fisher Scientific, Fremont, CA) for 1 hour at 37°C to visualize the cortical region (Oreal et al.,1998). To detect cytokeratin, Alexa-Fluor-555-labeled goat anti-mouse antiserum (1:200; Molecular Probes, Willow Creek Road, Eugene, OR)was used for secondary labeling. DAPI-positive nuclei in the cortical(cytokeratin-positive) and medullary (cytokeratin-negative) regions were counted and summed across all five sections for each embryo. The mean and standard deviation were determined using data from all three embryos. The relative cell numbers are presented as described in the figure legends. Bromodeoxyuridine (BrdU)-labeling of chick embryos and detection of the BrdU-labeled cells were basically performed as described elsewhere(Yoshioka et al., 2005). In brief, 1 μl BrdU solution (10 mg/ml in H₂O) was injected into untreated or manipulated (as described below) embryos through the vitelline vein at stage 27 or 29. After 1 hour incubation, the embryos were fixed with 4% PFA/PBS and cross-sectioned. Three embryos were used for each experiment. Five sections containing the gonad were randomly selected for each embryo and were subjected to immunofluorescence. The sections were incubated with sheep anti-BrdU antiserum (1:300; Exalpha Biological, Watertown, MA) and mouse anti-cytokeratin monoclonal antibody for 1 hour at 37°C. Alexa-Fluor-488-labeled donkey anti-sheep antiserum (1:200; Molecular Probes)and Alexa-Fluor-555-labeled goat anti-mouse antiserum were used for secondary detection. The number of BrdU-positive and total DAPI-stained cells in the cortical and medullary regions was determined. Statistical analysis and calculation of relative cell numbers are described above.

Manipulation of chick embryos of both sexes was performed as described previously (Yoshioka et al.,2005). To construct Pitx2-en, the repressor domain of Drosophila engrailed (amino acids 1-296)(Jaynes and O'Farrell, 1991)was fused to the carboxy-terminus of Pitx2c. cDNAs for Pitx2,Pitx2-en, Ad4BP/SF-1 and alkaline phosphatase (AP) were cloned into the avian viral vector RCASBP (A) and transfected into chick embryonic fibroblasts to generate virus-producing cells. Pitx2, Ad4BP/SF-1 and AP-expressing cells were implanted into the right lateral plate mesoderm of stage 11-12 embryos at the level of the presumptive gonad, while the Pitx2-en and AP-expressing cells were implanted into the left presumptive gonad (see Fig. S1A in the supplementary material). After being incubated for 72 (stage 27) or 96 hours (stage 29), the embryos were fixed with 4% PFA-PBS and subjected to whole-mount in situ hybridization or the cell proliferation assay described above. AG1-X2 ion-exchange resin beads(150-200 μm diameter; BioRad Laboratories, Hercules, CA) were soaked in 10μM all-trans-RA (Wako Pure Chemical Industries, Osaka, Japan) or 12.7 mM retinoid antagonist AGN 193109 (Allergan, Irvine, CA)(Mercader et al., 2000)dissolved in dimethylsulphoxide (DMSO). The beads were washed with PBS, and implanted into the coelomic epithelium of the left or right presumptive gonad region at stage 18-19 (see Fig. S1B in the supplementary material). After 48(stage 27) or 72 hours'(stage 29) incubation, the embryos were subjected to whole-mount in situ hybridization or the cell proliferation assay.

The sex of chick embryos was determined by the presence or absence of the W chromosome. To detect the W chromosome, the W-specific XhoI repeat was amplified by PCR with genomic DNA as a template. β-actin was amplified as a control (Kent et al.,1996).

To better understand the detailed morphological changes that occur in the chick gonad before sexual differentiation, we prepared cross sections at stage 27 and stage 29 (see Fig. S2A-D in the supplementary material). No morphological differences were observed between the right and left cortices at stage 27, with both cortices composed of one or two layers of columnar epithelia. By stage 29, the left cortex became thicker with several more layers than the right cortex. The number of left cortical cells increased by 2.1-fold during this time, but no change in the number of right cortical cells was evident (Fig. 1A). This asymmetric event was observed in both female (ZW) and male (ZZ). By contrast,the number of cells in the medulla increased approximately four-fold bilaterally (Fig. 1B).

As differences in cell number can arise from differential cell proliferation and/or cell death, we examined cell proliferation by BrdU incorporation in embryos at stage 27 (Fig. 1C) and 29 (Fig. 1D). BrdU-positive proliferating cells were visualized by immunofluorescence (Fig. 1C,D,green) and the cortical region of the developing gonad was visualized by anti-cytokeratin staining (Fig. 1C,D, red) (Oreal et al.,1998). The number of BrdU-positive cells in the left cortices of ZW and ZZ embryos were 1.8-fold and 1.6-fold higher, respectively, than in the right cortices at stage 27 (Fig. 1E), and this bias persisted through stage 29. By contrast, there was no significant asymmetry observed in the BrdU incorporation in the medullae during the same time period (Fig. 1F). Next, we examined whether apoptotic cell death occurs asymmetrically in the developing gonad. Only a few TUNEL-positive cells were detected on both sides of the cortex and medulla, and their numbers did not increase between stages 27 and 29 in either sex (see Fig. S2G,H in the supplementary material). Taken together, these data indicate that increased proliferation in the left cortex, rather than increased apoptosis in the right cortex, is primarily responsible for the observed asymmetric cortical development.

Although the gonad is still structurally symmetric at stage 27, we hypothesized that some genes are expressed asymmetrically at this stage. As RA signaling has been implicated in cell proliferation, we examined the expression of genes involved in RA signaling (RALDH1, RALDH2, RALDH3,CYP26A1, CYP26B1, RARα, RARβ, RARγ, RXRα, RXRβ and RXRγ). In addition,we examined the expression of genes implicated in establishing the L-R asymmetric body plan (PITX2a, PITX2b and PITX2c). Interestingly, RALDH2 and CYP26A1 were expressed asymmetrically in a mutually exclusive fashion at stage 27. RALDH2was expressed in the right cortex (Fig. 1G), whereas CYP26A1 was expressed in the left cortex and in both sides of the medullae (Fig. 1H). RARα (Fig. 1I) and RXRα (data not shown) expression was stronger in the right cortex than in the left, and was not detected in the medullae, suggesting that the right cortex is activated by RA signaling. PITX2c was expressed only in the left cortex, showing a good correlation with a previous report (Fig. 1J) (Logan et al.,1998). As there is genetic evidence that Ad4BP/SF-1 and LHX9 are essential for gonad development(Birk et al., 2000; Luo et al., 1994), we also examined their expression. The pattern of Ad4BP/SF-1 expression was similar to that of CYP26A1 (Fig. 1K), whereas LHX9 was expressed symmetrically(Fig. 1L). Interestingly,asymmetric gene expression occurred only in the cortex in both sexes. The other genes examined were not expressed in the gonad at stage 27.

We next wished to determine the functional relationship between the asymmetrically expressed genes. The complementary expression of RALDH2 and CYP26A1 suggested that a higher concentration of RA is present in the developing right cortex. As RA signaling is probably related to the absence of PITX2 and Ad4BP/SF-1, we hypothesized that RA downregulates PITX2 and Ad4BP/SF-1 in the right cortex, while in the left cortex, inhibition of RA signaling leads to the upregulation of a variety of genes. To test this hypothesis, we implanted beads soaked in RA or an RA antagonist (RAA) in the region of the developing left or right gonad, respectively, at stage 19 and the treated embryos were then incubated to stage 27. In the left gonad implanted with the RA beads, Ad4BP/SF-1 expression was substantially reduced in the cortex but persisted in the medulla (Fig. 2A, arrowhead; number of affected male embryos/total number of male embryos manipulated (male)=12/14; number of affected female embryos/total number of male embryos manipulated (female)=13/15). However, the expression of PITX2c in the left cortex was not downregulated by RA(Fig. 2B; male=0/11,female=0/12). Conversely, implantation of the RAA beads increased the expression of Ad4BP/SF-1 in the right cortex(Fig. 2C, arrowhead;male=12/15, female=13/16), but the RAA beads did not affect the expression of PITX2c in the right cortex (Fig. 2D; male=0/10, female=0/11). Implantation of DMSO beads had no effect on expression of either Ad4BP/SF-1(Fig. 2E; male=0/12,female=0/14) or PITX2c (Fig. 2F; male=0/11, female=0/11).

As the L-R cortical asymmetry seen at stages 27-29 is caused by increased cell proliferation rather than cell death, we examined whether RA signaling affected cortical cell proliferation. After BrdU was injected into bead-implanted embryos, we counted the number of BrdU-positive cells at stage 27. Interestingly, the number of BrdU-positive cells decreased in the left cortex implanted with RA beads compared with DMSO-treated embryos(Fig. 2G,H,O), but there was no corresponding effect in the medulla (Fig. 2P). Conversely, the number of BrdU-positive cells increased in the right cortex implanted with the RAA beads(Fig. 2I,J,Q), with no effect on the medullary cell number (Fig. 2R). We next examined the morphologic effects of bead implantation. By stage 29, the thickness of the left cortex, as visualized by cytokeratin immunostaining, was reduced in the RA-bead-implanted embryos(Fig. 2K,L), while the right cortex thickened following RAA bead implantation(Fig. 2M,N).

Our data indicate that exogenous RA can suppress the expression of Ad4BP/SF-1 but not PITX2c, suggesting that PITX2 inhibits the expression of RALDH2 in the left cortex. To test this hypothesis,we prepared expression plasmids encoding either wild-type mouse Pitx2 or a dominant-negative form, Pitx2-En. The transcriptional activities of Pitx2 and Pitx2-En were examined using luciferase reporter gene assays. As expected,Pitx2 induced robust luciferase activity, whereas Pitx2-En suppressed transcription driven by Pitx2 (see Fig. S3 in the supplementary material).

We implanted cells producing either Pitx2- or Pitx2-en-encoding viruses at stage 11-12 into the presumptive right or left gonad, respectively. Because the viruses were replication competent,transgene expression expanded throughout the gonad region(Fig. 3A,D,G). Operated chick embryos were then incubated to stage 27, and the expression of RALDH2and Ad4BP/SF-1 was examined. Exogenous expression of Pitx2led to reduced expression of RALDH2 in the right cortex(Fig. 3B, arrowhead;male=11/14, female=12/16), and increased expression of Ad4BP/SF-1(Fig. 3C, arrowhead;male=10/14, female=12/15). By contrast, expression of Pitx2-enincreased expression of RALDH2 in the left cortex(Fig. 3E, arrowhead; male=8/11,female=10/13), and reduced expression of Ad4BP/SF-1(Fig. 3F, arrowhead; male=8/10,female=12/15). A control virus encoding AP did not affect expression of either RALDH2 (Fig. 3H; male=0/13, female=0/15) or Ad4BP/SF-1(Fig. 3I; male=0/14,female=0/15).

Pitx2 is required for cell-type-specific proliferation during cardiac outflow tract (Kioussi et al.,2002) and pituitary gland(Kioussi et al., 2002; Zhu and Rosenfeld, 2004)development. Thus, we examined the effect of Pitx2 and Pitx2-en on BrdU incorporation in the developing gonad. When cells producing Pitx2-encoding viruses were implanted into the right presumptive gonad region, the number of BrdU-positive cells increased by 2.0-fold in the right cortex at stage 27(Fig. 3J,R), but proliferation in the medulla was not affected (Fig. 3S). By contrast, the number of BrdU-positive cells decreased in left cortices implanted with Pitx2-en virus-producing cells(Fig. 3L,T), whereas the medulla was not affected (Fig. 3U). Exogenous AP expression did not affect cell proliferation (Fig. 3K,M,R-U). Structurally, right cortices implanted with Pitx2 virus-producing cells thickened by stage 29 (Fig. 3N,O), whereas Pitx2-en-implanted left cortices were thinner (Fig. 3P,Q).

Likewise, we examined the effect of exogenous Ad4BP/SF-1expression on the right gonad cortex (Fig. 4A). Expression of PITX2c nor RALDH2 was unaffected following implantation of Ad4BP/SF-1 virus-producing cells(data not shown). However, the number of BrdU-positive cells in the right cortex increased by 1.8-fold (Fig. 4B,C,F) and the cortical layer thickened(Fig. 4D,E). Proliferation of the right medulla was unaffected (Fig. 4G).

Our data indicate that PITX2, RA and Ad4BP/SF-1 are involved in asymmetric L-R cortical cell proliferation. In order to determine the mechanism by which these factors act, we examined their effect on the expression of the cell cycle regulators cyclin D1 and D2. At stage 27, cyclin D1 was expressed asymmetrically in the left cortex and bilaterally in the medulla (Fig. 5A), whereas cyclin D2 was transiently expressed on both sides of the cortex and disappeared by stage 29 (data not shown). As the expression of cyclin D1 overlapped with that of CYP26A1 and Ad4BP/SF-1, we examined whether production of RA, PITX2 and Ad4BP/SF-1 leads to asymmetric cyclin D1 expression. RA and RAA beads were implanted as described. At stage 27, the expression of cyclin D1 was significantly downregulated in the RA-bead-implanted left cortices (Fig. 5B, arrowhead; male=5/8, female=6/10). Expression in the medulla was unaffected. By contrast, implantation of RAA beads in the right cortex stimulated expression of cyclin D1 (Fig. 5C, arrowhead; male=5/7, female=7/10). Embryos implanted with DMSO beads (Fig. 5D; male=0/9,female=0/9) resembled untreated embryos. When Pitx2 virus-producing cells were implanted, cyclin D1 was upregulated in the right cortex(Fig. 5E, arrowhead; male=5/8,female=6/8), but forced expression of Pitx2-en reduced the expression of cyclin D1 in the left cortex (Fig. 5F, arrowhead; male=5/9, female=5/10). When cells producing Ad4BP/SF-1-encoding virus were implanted, the expression of cyclin D1 was upregulated in the right cortex (Fig. 5G, arrowhead; male=6/8, female=7/9). Finally, implantation of cells producing a control AP-virus did not affect the expression of cyclin D1 (Fig. 5H; male=0/7,female=0/9). These results indicate that cyclin D1 is negatively regulated by RA and positively regulated by PITX2 and Ad4BP/SF-1 in the developing gonad,although it remains to clarify whether the regulations are direct or indirect.

As estrogen signaling has been implicated in asymmetric ovarian development in chicks (Nakabayashi et al.,1998; Romanoff,1960), we examined the expression of ERα at stage 27. As shown in Fig. 6A, ERα was expressed asymmetrically in the left but not right gonadal cortex, and symmetrically in both sides of the medulla(Andrews et al., 1997; Smith et al., 1997). This expression was identical in males and females (data not shown). When RA beads were implanted in the left developing gonad, the expression of ERα was significantly reduced in the left cortex(Fig. 6B, arrowhead; male=5/6,female=8/9). Conversely, RAA-bead-implantation in the right developing gonad increased the expression of ERα in the right cortex(Fig. 6C, arrowhead; male=5/7,female=7/9). Implantation of DMSO beads had no effect on the expression of ERα (Fig. 6D;male=0/6, female=0/7). Subsequently, we examined the effect of Ad4BP/SF-1 on ERα expression by implanting cells with Ad4BP/SF-1-encoding viruses in the right cortex. There was no change in cortical ERα expression in animals implanted with Ad4BP/SF-1-expressing cells (Fig. 6E; male=0/6, female=0/7) or control AP-expressing cells(Fig. 6F; male=0/5,female=0/5). These data suggest that asymmetric RA signaling in the developing chick gonad regulates asymmetric ERα expression, whereas Ad4BP/SF-1 is not involved in ERα expression. We further investigated whether or not estrogen regulates expression of RALDH2,PITX2 and Ad4BP/SF-1 by implanting estrogen-coated beads, but found no effect of estrogen on the expression of these genes (data not shown).

Likewise, we examined the effect of RA on the expression of CYP26A1. RA-bead-implantation in the left developing gonad(Fig. 6G, arrowhead; male=4/5,female=5/5) increased and expanded CYP26A1 expression in the left medulla, while RAA-bead-implantation in the right developing gonad decreased CYP26A1 expression in the left medulla(Fig. 6H, arrowhead; male=5/5,female=5/5). Implantation of DMSO beads into the left gonad had no effect(Fig. 6I; male=0/4,female=0/4). The effect of Ad4BP/SF-1 was also examined by implanting Ad4BP/SF-1-expressing cells into the right developing gonad. As was the case for control AP-expressing cells(Fig. 6K; male=0/5,female=0/4), Ad4BP/SF-1-expressing cells had no effect on the expression of CYP26A1 (Fig. 6J; male=0/7, female=0/7). Taken together, these results show that asymmetric RA signaling in the developing gonad regulates asymmetric expression of CYP26A1, whereas Ad4BP/SF-1 does not.

RA signaling has been shown to activate CYP26 gene expression during the development of several tissues(Swindell et al., 1999; Moreno and Kintner, 2004; Yashiro et al., 2004). Here,we show that CYP26A1 expression is induced by RA signaling in the developing gonad. Interestingly, however, CYP26A1 is not expressed in the right cortex. As the right cortex produces RA via RALDH2 and expresses RARα (Fig. 1I)and RXRα (data not shown), RA signaling is likely to be active there. Therefore, suppression of CYP26A1 expression in the right cortex must be mediated by an unknown mechanism independent of RA signaling.

Although chick embryos form only a single ovary on the left-hand side, the gonad primordia initially develop symmetrically(Romanoff, 1960). Just before sex differentiation, the gonad cortex begins to develop morphological asymmetry, with the right cortex becoming thinner and the left cortex thickening. Given that the cortex contributes significantly to ovarian development, it seemed likely that the right gonad loses the ability to develop into an ovary at this stage. Thus, we focused on this step and investigated the molecular mechanism by which L-R asymmetry is induced in the gonadal cortices.

As summarized in Fig. 7,RA-synthesizing RALDH2 is expressed asymmetrically in the right cortex, while RA-metabolizing CYP26A1 is expressed in the left cortex. Likewise, PITX2c, Ad4BP/SF-1 and ERα show asymmetric expression at the left cortex. Interestingly, the asymmetric expression of these genes occurs only in the cortex, suggesting that the initial step in L-R asymmetric ovarian development is dominated by differential expression of genes in the cortex. To investigate the functional relationship between these L-R asymmetric factors, we performed various experiments to test how forced gene expression or chemical treatment affects laterality of the developing gonads. Our data indicate that in the developing left cortex, PITX2 suppresses RALDH2 expression, leading to reduced RA signaling and increased Ad4BP/SF-1 and ERαexpression. Subsequently, Ad4BP/SF-1 upregulates cyclin D1 expression and promotes cell proliferation. In the right cortex, lack of PITX2 results in upregulation of RALDH2, which in turn activates RA synthesis. RA then suppresses cell proliferation through downregulation of Ad4BP/SF-1and cyclin D1 (Fig. 7). Our current studies successfully demonstrate a functional relationship between the above components, although it still remains to be determined whether the regulation is direct or indirect.

It would be interesting to examine the later development of gonads implanted with Pitx2, Pitx2-en and Ad4BP/SF-1-expressing cells and RA and RAA beads. However, the operated embryos die at around stage 30, when L-R asymmetric ovarian development and gonadal sex differentiation begin, so we were not able to examine gonadal development in these animals. As operated embryos die even when control AP-expressing cells and DMSO beads are implanted, it appears that the embryos die from physical damage resulting from the operation itself. Novel techniques that cause less damage to the embryo are therefore necessary to address this question.

In the present study, we show that the genes involved in the initial L-R asymmetric development of the gonads are expressed similarly in both sexes,and initial morphogenetic events occur independent of sex(Fig. 7). Indeed, the medulla develops bilaterally, whereas the cortex develops only on the left in both sexes. This left-sided cortical development is thought to be important for left-sided ovarian development. Thereafter, the gonads display sexually dimorphic development, with bilateral testicular development in the male and left-sided ovarian development in the female. As estrogen activates development of the gonadal cortex into the ovary(Scheib, 1983), and estrogen is synthesized only in the female, left-sided ovarian development is achieved in the female but not the male. By contrast, as it is the medulla that contributes most to development of the testis, the testis develops bilaterally in males.

During the establishment of L-R asymmetry in ovarian development, cell proliferation appears to be accelerated in the left cortex and decelerated in the right cortex. Our data demonstrate that Ad4BP/SF-1 activates cyclin D1 transcription, which probably accelerates cell proliferation. Interestingly,LRH1 (NR5A2), which is highly homologous to Ad4BP/SF-1, activates cyclin D1 gene transcription through its direct interaction with β-catenin(Botrugno et al., 2004). Like LRH1, Ad4BP/SF-1 can interact directly with β-catenin to synergistically activate transcription (Mizusaki et al.,2003). Therefore, it is possible that Ad4BP/SF-1 functions in a similar manner to activate cyclin D1 gene transcription. Moreover, LRH1 activates cyclin E1 gene transcription by binding directly to its promoter. As the nucleotide sequence recognized by LRH1 is almost identical to that bound by Ad4BP/SF-1, Ad4BP/SF-1 may also regulate cyclin E1 gene transcription. In addition to the possible activation of cyclin D1 by Ad4BP/SF-1, Pitx2 also directly upregulates mouse cyclin D1 transcription(Kioussi et al., 2002; Zhu and Rosenfeld, 2004). The chick cyclin D1 promoter contains a putative PITX2-binding site, suggesting that PITX2 may also directly activate chick cyclin D1 gene expression. Moreover, the synergistic activation of LHβ transcription by Pitx2 and Ad4BP/SF-1 (Tremblay et al.,2000) suggests that cyclin D1 gene transcription might be synergistically regulated by these two factors. However, more work is necessary to determine the precise mechanism by which these factors interact.

The deceleration of cell proliferation in the right cortex appears to be mediated by inhibition of cyclin D1 expression by RA. Consistent with our present observations, other labs have shown that RA inhibits cyclin D1, cyclin E1 and cyclin-dependent kinase 6 in cultured cell lines(Balasubramanian et al., 2004; Chen and Ross, 2004; Sah et al., 2002). In addition to downregulating these cell cycle accelerators, RA stimulates the expression of several cell cycle inhibitors, including cyclin-dependent kinase inhibitor, p27Kip1 and p21Cip1(Balasubramanian et al., 2004; Buzzard et al., 2003; Chen and Ross, 2004; Guidoboni et al., 2005). This is further supported by studies of Raldh2 (also known as Aldh1a2 - Mouse Genome Informatics) knockout mice(Lai et al., 2003), which exhibit reduced expression of p27Kip1 (Cdkn1b) and p21Cip1 (Cdkn1a), and increased proliferation of endothelial cells. These findings are consistent with our current observations in the developing gonad.

Interestingly, a novel role for RA was recently observed in the mouse embryonic gonad. Typically, germ cells enter meiosis in the embryonic ovary,whereas meiosis is retarded in the embryonic testis. Increased Cyp26b1 expression in the male gonad leads to decreased RA relative to the female gonad, and the different levels of RA appear to regulate the sex-specific timing of meiotic initiation in germ cells(Bowles et al., 2006; Koubova et al., 2006). We did not observe sex-specific expression of RALDH2 or CYP26 in the chick gonad, but the meiotic regulation of these genes by RA could occur at a later stage, when sexual differentiation has already occurred.

Previous work has shown that in ovo inhibition of estrogen synthesis by aromatase inhibitor induces bilateral female-to-male sex reversal(Elbrecht and Smith, 1992; Hudson et al., 2005; Smith et al., 2003; Vaillant et al., 2001a; Vaillant et al., 2001b). By contrast, although in ovo estrogen administration induces partial male-to-female sex reversal (ovotestis), it occurs only in the left gonad(Nakabayashi et al., 1998; Romanoff, 1960). This differential effect of estrogen has been thought to be due to the preformed differential potential of the left and right gonadal cortices(Mittwoch, 1998). Based on these observations, it was proposed that the differential potential of the cortices is established independently of estrogen signaling(Nakabayashi et al., 1998). Our present study demonstrates that RA signaling triggers two events in the right cortex: loss of responsiveness to estrogen through suppression of ERα expression, and inhibition of cell proliferation through suppression of Ad4BP/SF-1 and thus cyclin D1 expression. As the left cortex retains both responsiveness to estrogen and the ability to undergo cell proliferation, it is not surprising that exogenous estrogen induces ovotestis only on the left side.

L-R asymmetric ovarian development is interesting from both an evolutionary and developmental standpoint. During L-R axis formation, L-R asymmetry at the node is transmitted to the lateral plate mesoderm by upregulation of PITX2 expression on the left side, which is thought to result in asymmetric visceral organ development. Although the correlation between the L-R asymmetry of the body plan and ovarian development remains to be investigated, we have shown that PITX2, RA signaling, Ad4BP/SF-1, estrogen signaling and cyclin D1 are involved in the process of asymmetric ovarian development in the chick.

We thank Drs Kudo and Sutou (Ito Ham Co., Japan) for the chick Ad4BP/SF-1 cDNA, Dr Eichele (Baylor College, Texas) for the chick RALDH2 and CYP26A1 cDNAs, Dr Noda (National Institute for Basic Biology, Japan) for chick RALDH1 and RALDH3 cDNAs, Dr Nohno (Kawasaki medical school, Japan) for the chick RARβ cDNA,Dr Murray (Iowa University, USA) for the mouse Pitx2 cDNA. This work was supported in part by Grants-in-Aid for the Ministry of Education, Culture,Sports and Technology, and Japan Science and Technology Corporation. This work was supported in part by Grants-in-Aid from the Ministry of Education,Culture, Sports, Science and Technology of Japan, and Grants-in-Aid for Scientific Research on Priority Areas, from the Japan Science and Technology Corporation, and Sumitomo Foundation.