Msx2 promotes vaginal epithelial differentiation, Wolffian duct regression, and dampens the vaginal response to diethylstilbestrol

Abstract

In utero exposure to diethylstilbestrol (DES) leads to patterning defects in the female reproductive tract (FRT) and a propensity to the development of vaginal adenocarcinomas in humans. In the mouse, DES treatment similarly induces a plethora of FRT developmental defects, including stratification of uterine epithelium and presence of glandular tissue in cervix and vagina. Uterine abnormalities are associated with repression of the homeobox gene Msx2, and DES leads to an altered uterine response in Msx2-mutants including a dilated uterine lumen. Here we investigate the role of Msx2 in normal vaginal development and in FRT response to DES. During vaginal development, Msx2 is required for transforming growth factor β2 (Tgfβ2) and Tgfβ3 expression and for proper vaginal epithelial differentiation. Moreover, Msx2 is involved in caudal Wolffian duct regression by promoting apoptosis. Consistently, neonatal DES exposure represses Msx2 expression in the Wolffian duct epithelium, and inhibits its apoptosis and subsequent regression. Intriguingly, although DES treatment also represses Msx2 expression in the vaginal epithelium, a much more severe DES-induced vaginal phenotype was observed in Msx2-mutant mice, including a complete failure of Müllerian vaginal epithelial stratification and a severely dilated vaginal lumen, accompanied by loss of p63 and water channel protein expression. These results demonstrate a critical role for Msx2 in counteracting the effect of DES on FRT patterning, and suggest that the response to DES may be highly variable depending on the genotype of an individual.

INTRODUCTION

Development of the female reproductive tract (FRT) is complex and requires concerted regulation of cell-fate decision, cell proliferation, and differentiation (1, 2). Most of the FRT, including the oviduct, uterus, cervix, and upper part of the vagina (Müllerian vagina), derives from the Müllerian duct. This duct’s embryonic origin is the intermediate mesoderm (3). Müllerian duct epithelium adopts different cell fates along the anteroposterior axis as development proceeds. Oviductal and uterine epithelia are composed of simple (unstratified) columnar epithelial cells that are morphologically distinct, whereas the vagina is lined with stratified squamous epithelium. The cervix is the transition zone where both types of epithelia are present. During FRT development, the caudal portion of the Wolffian duct fuses with the Müllerian duct, and together they migrate downward to form the anterior part of vagina (4). As development proceeds, the Wolffian duct regresses because of a lack of androgen to support its development, and its caudal remnant does not disappear completely until P8 (5). Little is known, however, about the cellular and molecular mechanism of Wolffian duct degeneration.

Several transcription factors and growth factors, including Lim1, Pax2, Emx2, and Wnt4, have been implicated in the initial formation of the Müllerian duct during mammalian embryogenesis (6–9). Mutation in any one of these genes results in either failure of Müllerian duct formation or arrested development. Hox and Wnt genes play important roles in determining segmental identity along the anteroposterior axis of the developing Müllerian duct, as mutations in these genes lead to regional patterning defects along the FRT (10–17). p63, a p53 homologue, is expressed in epithelia fated to become stratified and functions as a molecular switch to induce stratification (18). p63-deficient mice fail to develop stratified epithelia and epithelial appendages, whereas ectopic expression of one p63 isoform TAp63 in single-layered lung epithelium leads to its stratification (19, 20). Similarly p63 is required in the FRT to induce cervical and vaginal epithelial stratification in response to stromal signals and loss of p63 expression leads to the presence of uterine glandular cells in the cervicovaginal region (adenosis) (21).

Exposure of the developing Müllerian duct to the synthetic estrogen diethylstilbestrol (DES) perturbs this normal cell-fate decision process by inducing regional stratification of the uterine epithelium and vaginal adenosis (22). This change in cell fate has been linked to alterations in gene expression involved in normal cell-fate decision in the FRT, such as Hox, Wnt, and p63 genes (12, 13, 21, 23). Particularly DES represses p63 expression in the developing vaginal fornix which is thought to be the cause of vaginal adenosis (21). In addition, DES-treated FRT display delayed oviductal development, paraovarian cysts, and abnormal patterning of the uterus and cervix (24, 25). In the vagina, neonatal DES exposure prevents caudal Wolffian duct regression and leads to persistent vaginal epithelial proliferation and cornification even after removal of endogenous estrogen (22, 26, 27). DES exerts most of its teratogenic effects on FRT patterning through binding to estrogen receptor α; most of these defects disappear in estrogen receptor α knockout mice (28). Consequently, DES fails to repress Hoxa-10, Hoxa-11, and Wnt7a expression in the uterus of the estrogen receptor α knockout mouse (28).

In the developing uterus, we have recently shown that neonatal DES treatment induces premature abnormal differentiation characterized by the expression of an array of differentiation markers (29). In addition, DES represses Msx2 expression in developing uterine epithelium, and in Msx2-mutant mice, DES induces a dramatically enlarged uterine lumen phenotype accompanied by abnormal uterine epithelial cell morphology (29). In this study, we show that Msx2 functions to promote normal vaginal epithelial differentiation and caudal Wolffian duct degeneration. Furthermore, Msx2 appears to play a critical role in counteracting the effect of DES on the FRT because DES induces much more severe vaginal phenotypes in Msx2-mutants, such as the failure of epithelial stratification and abnormal water trafficking. Our results implicate Msx2 in vaginal development and identify it as a key molecule whose presence can modulate the severity of the DES response.

RESULTS

DES Represses Msx2 Expression in Vaginal and Degenerating Wolffian Duct Epithelia

We have shown recently that Msx2 expression is repressed by neonatal DES exposure in uterine epithelium (29). To determine the effect of DES on Msx2 expression in the developing vagina, in situ hybridization was performed on P5 transverse vaginal sections. As shown in Figure 1, Msx2 expression was detected in the vaginal epithelium (A) and DES treatment significantly reduced Msx2 expression (B). Strong Msx2 expression was also detected in the degenerating Wolffian duct epithelium (Fig. 1C), which is significantly downregulated by DES-treatment as well (Fig. 1D). All in situ hybridization experiments presented were repeated on at least two individual samples and same results were obtained. Quantification by real-time RT-PCR with RNA extracted from the whole Müllerian vagina showed that the Msx2 transcript level decreased 40% in DES-treated mice compared with that in oil-injected controls (Fig. 1E). However, the repression fold is likely underestimated in the epithelial compartment: as DES treatment increased the number of Msx2-expressing epithelial cells in the vagina, a 40% decrease in the tissue must reflect lower levels of Msx2 expression on a per cell basis.

Fig. 1.

DES represses Msx2 expression in developing vagina. In situ hybridization revealed high Msx2 expression in P5 wild-type vaginal epithelium (A) and Wolffian duct remnant, both outlined by dashed lines (C). In all figures, in situ or immunohistochemistry signals are shown in red and nuclei are blue. Msx2 expression is repressed by DES in both tissue layers (B and D). Quantification by real-time RT-PCR showed a 40% reduction in Msx2 expression by DES (E; *, P < 0.05). Bars: 50μm in A and B; 20μm in C and D. ve, vaginal epithelium; Wd, Wolffian duct.

Msx2 Promotes Caudal Wolffian Duct Regression

Since developmental exposure to DES results in persistent Wolffian duct remnant in females (22) and that DES potently represses Msx2 expression in the degenerating Wolffian duct epithelium, it is possible that loss of Msx2 leads to persistence of Wolffian duct remnants. To test this hypothesis, we examined Wolffian duct regression in Msx2−/− females. Histological examination revealed paired ductal remnants located dorsally to the vaginal lumen in 88.2% adult Msx2-mutants (n = 17; Fig. 2B, arrow) whereas only 12.5% of wild type controls contain ductal remnants (n=16, data not shown). These are likely Wolffian duct remnants because they express Pax2, a marker for Wolffian and Müllerian duct epithelium (Fig. 2A). They are not Müllerian duct derivatives because on serial sections, these ducts are never connected to the vaginal epithelium and they are also present in some P5 wild-type vagina (Fig. 2C).

Fig. 2.

Msx2 promotes caudal Wolffian duct regression. (A) In situ hybridization shows Pax2 expression in P5 remnant demonstrating its intermediate mesodermal origin. (B) A typical Wolffian duct remnant in 2-month-old Msx2-mutant vagina (arrow) is shown. (C) TUNEL assays show that at P5, apoptotic cells are present in degenerating caudal Wolffian duct remnant but no apoptotic cells can be detected in DES-exposed Wolffian duct epithelium (inset). (D) At P5, no apoptotic cells were detected in Msx2-mutant Wolffian duct remnant. (E) Immunohistochemistry shows several active-Caspase-3 positive cells in wild type Wolffian duct remnant but not in Msx2 mutant (F). Similarly, active-Caspase-9 positive cells are only detected in wild type but not Msx2 mutant Wolffian duct remnant (G, H). Bars: 50μm in B; 20μm in A, C–H. ve, vaginal epithelium; s, stroma.

Since Msx2 was implicated in promoting apoptosis in the developing rhombomeres and in P19 cell aggregates (30, 31), we reasoned that it might play a similar role in duct regression. To determine whether apoptosis contributes to the normal involution of the Wolffian duct in females and whether loss of Msx2 perturbs this process, we performed TUNEL assay on both wild-type and Msx2−/− vaginal sections. Wild type Wolffian duct remnants exhibited a consistent apoptotic rate of 3.22±0.07% (mean±S.D. per section, n=3) and 3.16±0.20% (n=3) on P2 and P5, respectively (data not shown and Fig. 2C). This apoptosis was completely repressed by DES-treatment (Fig. 2C inset). In Msx2−/−, we observed 1.63±0.06% (n=3) apoptotic cells in P2 remnants, which was not statistically significant compared to that of wild type (p=0.48). However, only 0.28±0.01% (n=3) TUNEL positive cells were detected in P5 Msx2−/− Wolffian duct remnants, a significant reduction compared to wild type (p<0.05, Fig. 2D). This suggests that the molecular mechanism underlying DES-induced persistent Wolffian duct remnant could be through repression of Msx2 expression.

To investigate how loss of Msx2 expression leads to this repression of apoptosis, we examined active caspase-3 and active caspase-9 activity in P5 wild type and Msx2 mutant Wolffian ducts. These two proteins are critical cysteine proteases that carry out essential steps during apoptosis (32). Both activated caspase-3 and caspase-9 were detected in wild type degenerating Wolffian ducts but not in those of Msx2 mutants (Fig. 2E-H). To test whether this block in apoptosis resulted from an upregulation of anti-apoptotic proteins, we examined the expression of Bcl-2 and Birc1a, two known inhibitors of apoptosis, by immunohistochemistry and in situ hybridization, respectively (33, 34). However, no detectable changes in the expression of these molecules were observed (data not shown). These results indicate that Msx2 is involved in caudal Wolffian duct regression partially by promoting or maintaining apoptosis, and they suggest that Wolffian duct remnants observed in DES-treated wild-type mice are likely due to a loss of Msx2 expression in these structures.

Failure of Vaginal Epithelial Stratification in DES-Treated Msx2-Mutants

During dissection, we noticed that DES-treated Msx2−/− females showed significantly swollen vaginas and uterine horns at P5 compared with wild-type controls (Fig. 3G, H). Fluid was expelled out of those swollen reproductive tracts during dissection. Histological analyses on these reproductive tracts revealed that P5 wild type vagina contains three morphologically distinct epithelial layers: basal, intermediate and superficial layers (Fig. 3A). In contrast, Msx2−/− vagina lacked the superficial layer of the vaginal epithelium and cells that make up the basal and intermediate layers appeared histologically abnormal. Their nuclei are enlarged and round-shaped compared to elongated nuclei typically found in wild type vaginal epithelium (Fig. 3B). In addition, we examined the expression of Tgfβ proteins in the vaginal epithelium for their well-established role in regulating cellular differentiation (35). Tgfβ2 protein was localized to the basal epithelial layer, and its expression decreased as cells exited the cell cycle and differentiated into the intermediate layer (Fig. 3C, arrows). Tgfβ3 expression was barely detectable in the basal and superficial layers but was strong in the intermediate layer (Fig. 3E, arrows). Consistent with morphological observations, expression of both Tgfβ2 and Tgfβ3 proteins was severely compromised in Msx2-mutant vaginal epithelium (Fig. 3D, F). These results were further confirmed by RT-PCR (data not shown). Thus, expression of Tgfβ2 and Tgfβ3 in the vaginal epithelium requires Msx2 function, and altered Tgfβ signaling may contribute to the differentiation defects in Msx2-mutant vagina. In wild-type mice, DES treatment increased vaginal epithelial thickness at P5, mainly by enlarging the squamous cell population residing in the intermediate layer (Fig. 3I). Remarkably, DES-treated Msx2-deficient vaginal epithelium failed to stratify: only simple columnar epithelium formed (Fig. 3J). In addition, the mutant vaginal lumen was extremely dilated, and a severe reduction in vaginal wall thickness was also observed (Fig. 3J, inset). At P15, the anterior vaginal epithelium of DES-treated Msx2−/− was still not stratified (Fig. 3K, L). Failure of vaginal stratification was observed in 21 of 25 DES-treated Msx2-mutants. Variations in the efficacy of DES administration could have contributed to the incomplete penetrance of this phenotype. This phenotype was confined to the Müllerian vagina because the posterior region of the mutant vagina which is derived from the urogenital sinus properly stratified (Fig. 3L inset). Even though Msx2−/− vaginal epithelia are structurally abnormal, they can properly respond to endogenous cycling hormones evidenced by cytological analysis of the vaginal smear (Supplementary Fig.1). In addition, Msx2−/− females have near normal fertility with an average litter size of around 9 in CD-1 background compared to around 13 in wild type mice, indicating that Msx2−/− females are capable of responding to endogenous estrogen.

Fig. 3.

Msx2 promotes vaginal differentiation and dampens FRT responsiveness to DES. Hematoxylin and eosin staining shows that at P5, wild-type vaginal epithelium is well organized and consists of superficial (su; arrow), intermediate (in), and basal (b) layers (A). At P5, Msx2-mutant vagina is missing the superficial layer while the basal and intermediate layers are composed of cells with round-shaped nuclei (B). Consistent with this morphological anomaly, Tgfβ-2 and -3 which are normally expressed in basal and intermediate layers, respectively in wild type vagina (C, E) are no longer detected in that of Msx2 mutant (D, F). (G, H) Gross views of the entire reproductive tract of DES-treated wild type and Msx2-mutant show a severely swollen vagina in Msx2 mutant. (I) DES treatment in wild-type mice results in an increase of cell population mainly in the intermediate layer and augmented vaginal epithelium thickness (arrows). (J) Msx2−/−vaginal epithelium becomes simple columnar and the vaginal lumen is severely dilated (arrows and inset). This simple columnar cell morphology is maintained at least to P15 in Msx2 mutant (L, arrow) compared to well differentiated and stratified vaginal epithelium in DES-treated wild type mice (K ). The lower part of the mutant vagina derived from the urogenital sinus, however, is well differentiated (L, inset). Bars: 20 μm in A–F, I-L; 100 μm in L inset.

Since mutation in p63 results in the failure of epithelial stratification both in the skin and in the vagina (19, 21), it is possible that failure of vaginal epithelial stratification in DES-treated Msx2−/− mice results from loss of p63 expression. We tested this hypothesis by examining p63 expression by immunohistochemistry. p63 was expressed in the basal layer of P5 vaginal epithelium and neonatal DES exposure repressed its expression in a fraction of cells in the developing vaginal fornix (21) (Fig. 4A, B, inset, arrows). Most cervicovaginal epithelial cells recover from this repression 2 days after stoppage of DES treatment, but some cells do not recover and remain simple columnar (21). These incipient cells form ectopic glandular tissues later in the adult cervix and vagina (21). In Msx2−/− oil-treated vagina, although most regions displayed normal p63 expression, some basal cells did not express p63 (Fig. 4C, arrows). DES treatment, however, completely abolished p63 expression in all four Msx2−/− vagina examined (Fig. 4D). At P15, p63 was still absent in most of the simple columnar cells in DES-treated Msx2−/− vaginal epithelium, whereas some clusters of p63-expressing cells were observed (Fig. 4D, inset). This repression was observed as early as 6 hours after the first DES injection as revealed by in situ hybridization, demonstrating that this regulation occurred at the transcription level (Fig. 4E, F).

Fig. 4.

DES treatment in Msx2−/− abolishes p63 expression. (A) In wild type mice, immunohistochemistry shows p63 expression in the basal layer of vaginal epithelium. (B) DES-treatment does not affect p63 expression in the majority of basal cells except in some regions of the vaginal fornix where it represses p63 expression (inset, arrows). (C) p63 expression was observed in most basal cells in Msx2 mutant vagina except for a few (arrows). (D) In vast contrast, no p63 positive cells can be detected in DES-treated Msx2−/− vaginal epithelium. The repression persisted at least to P15 in the simple columnar vaginal epithelial cells of DES-treated Msx2−/−, when most Müllerian vaginal cells still did not express p63 with the exception of some p63 positive clusters (inset). (E) At P1, in situ hybridization shows p63 transcripts in the posterior region of the vaginal epithelium of oil-treated control Msx2−/− (arrows). (F) Six hours after DES administration, p63 expression in the basal layer is completely abolished. Bars: 50 μm in A–D; 200 μm in E and F.

In mature epidermis, p63 functions to maintain the proliferative protential of basal keratinocytes (20). Thus we reasoned that loss of p63 expression in DES-treated Msx2−/− vagina could also affect normal cell cycle progression. To assess vaginal epithelial proliferation, neonatal wild-type and Msx2−/− pups were injected daily with oil or DES and BrdU 1 hour prior to sacrifice. The vaginas were harvested at different time-points as indicated and processed for immunohistochemistry. At P1, the average number of BrdU-positive cells in wild type vaginal epithelium per section was increased approximately 40% in DES-injected mice compared with oil-injected controls (n = 6) (Fig. 5), which presumably allowed the increase in the total cell number in P5 vaginal epithelium. In contrast, DES-treatment significantly reduced BrdU-positive cell number in Msx2−/− vaginal epithelium whereas normal amount of BrdU-positive cells was detected in Msx2−/− oil controls (Fig. 5). From P2 onwards, a reduced number of proliferating epithelial cells was observed in both DES-treated wild type and Msx2−/−vaginas (Fig. 5). Therefore, DES exerts a different effect on P1 vaginal epithelial cell proliferation in the absence of Msx2 which may be due to the loss of p63 expression.

Fig. 5.

DES elicits a different mitogenic response in Msx2−/− vaginal epithelium. At P1, 6 hours after the first DES injection, a 40% increase in cell proliferation was observed in wild-type vaginal epithelium. On the contrary, DES treatment represses proliferation in vaginal epithelial cells in Msx2−/− by 60%. Starting from P2 and extend to P5, DES-treated wild type vaginal epithelium shows a reduced number of proliferating cells. In Msx2−/−, DES continues to repress vaginal epithelial cell proliferation.

Abnormal Regulation of Aquaporins by DES in Msx2-Mutant Vagina

The abnormal water trafficking phenotype in DES-treated Msx2 mutant vagina may result from mis-regulation of water channel proteins, termed aquaporins. These proteins are normally found as tetramers and transport water, small molecules such as glycerol, or both across the cell membrane (36). So far 13 aquaporins expressed in various organs have been identified in mammals (37). Our microarray results have previously shown that Aquaporin3 (Aqp3) expression is induced by DES in the neonatal uterus (29). In this study, we examined the expression of all 12 functional murine aquaporins in oil- and DES-treated vaginas using RT-PCR. Three of the 12 aquaporins were expressed at significant levels in both wild-type and in Msx2−/− vagina on P5, and their expression was unaffected by DES treatment (Fig. 6A). On the other hand, expression of Aqp3, Aqp4, and Aqp8, was induced by DES in P5 wild-type vagina (Fig. 6A). The robust induction of Aqp3 and Aqp4 by DES, however, was lost in P5 Msx2−/− vagina, whereas that of Aqp8 was minimally affected by Msx2 mutation (Fig. 6A). These results were further confirmed by real-time RT-PCR assays (data not shown). Because Aqp8 mutants do not have obvious defects in water transport, we focused our subsequent studies on Aqp3 and Aqp4 (38).

Fig. 6.

Aqp3 and Aqp4 fail to be induced by DES in Msx2-mutant vagina. (A) RT-PCR experiment reveals that Aqp1, Aqp5, and Aqp11 are expressed at high levels in both wild-type and Msx2-mutant vagina, and their expression not affected by DES. Expression of Aqp8 is dramatically upregulated by DES in both wild type and Msx2 mutant vaginas. On the other hand, expression of Aqp3 and Aqp4 is readily induced by DES in wild-type vagina but not in Msx2−/−. Immunohistochemistry shows that Aqp3 is weakly expressed in the plasma membrane of intermediate layer cells of wild-type (B) and Msx2-mutant vagina (D). DES exposure increases Aqp3 expression level in wild-type (C) but not in Msx2−/− vagina (E). Aqp4 is normally detected in the basolateral membrane of the superficial layer cells, and its expression is enhanced by DES (F and G). In Msx2−/− vaginal epithelium, no Aqp4 expression is detected in either oil or DES-treated mice (H). Bars: 50μm.

To examine the tissue distribution of Aqp3 and Aqp4 in vagina, we performed immunohistochemistry on P5 vaginal sections. Low Aqp3 expression was detected in the cell membrane of the intermediate layer of vaginal epithelium (Fig. 6B). DES increased the cell population of this tissue layer and at the same time upregulated Aqp3 expression in these cells (Fig. 6C). Similar basal Aqp3 expression was observed in Msx2−/− vagina, but its expression failed to be upregulated by DES, possibly as a consequence of the failure of vaginal stratification (Fig. 6D, E). Expression of Aqp4 was detected in the cell membrane of the superficial layer of wild-type vagina and was upregulated by DES (Fig. 6F, G). However, no Aqp4 expression could be detected in Msx2-mutant vagina treated with either oil or DES (Fig. 6H, I). Loss of Aqp4 expression likely results from the differentiation defects in Msx2−/− vagina because the superficial layer where Aqp4 is normally expressed is missing (Fig. 3B). In situ hybridization experiments showed similar results (data not shown). Together, these results suggest that the water trafficking defect in DES-treated Msx2-mutant vagina may be caused by failure of Aqp3 and Aqp4 induction by DES, resulted from cell-fate changes in the DES-treated Msx2−/− vaginal epithelium. An equally attractive explanation for the water trafficking defect is that the mutant vagina fails to stratify and therefore is defective in barrier function, leading to excessive water loss. These two hypotheses, however, are not mutually exclusive.

DISCUSSION

Msx2 in Normal Vaginal Development

The Msx2 homeodomain protein plays important roles during the development of multiple organs (39). Its function is required during terminal differentiation of epithelial structures, such as hair follicle and tooth (40, 41). In this study, we found that in the absence of Msx2, vaginal epithelial differentiation is abnormal. We determined that the superficial layer is missing through both histological examination and the absence of Aqp4, which is normally expressed in this tissue layer. The basal and intermediate layers of the mutant vagina are histologically abnormal and do not express Tgfβ2 and Tgfβ3, respectively. The Tgfβ signaling pathway plays fundamental roles in a variety of cellular processes, including growth inhibition, migration, adhesion, and differentiation (35). More specifically, Tgfβ2 has been shown to be involved in hair follicle morphogenesis and corneal development, and Tgfβ3 in secondary palate formation (42–47). Although the role of Tgfβs in vaginal epithelial development is currently unclear, expression of all three Tgfβs has been shown to be regulated by perinatal DES exposure in both the uterus and vagina (48). The tissue-specific expression of Tgfβs in vaginal epithelium suggests that they also play distinct roles in mediating vaginal epithelial proliferation and differentiation. Tgfβ signaling appears to negatively regulate Msx2 expression both in the interdigital death zone and in the developing suture (49, 50). Therefore it is possible that a negative feedback loop exists between Msx2 and Tgfβ signaling during organogenesis. However, we do not know at present whether alteration in Tgfβ signaling directly results in abnormal vaginal differentiation or it is a secondary effect to the vaginal anomalies in Msx2−/− mice. Eliminating Tgfβ signaling specifically from the developing vaginal epithelium should be able to address this question.

Another interesting phenotype in Msx2-mutant is the persistent Wolffian duct remnants. The failure of the caudal Wolffian duct to degenerate completely in Msx2-mutant females indicates that Msx2 normally participates in this process. Previous studies have shown that overexpression of Msx2 in the developing eye and even-numbered rhombomeres leads to apoptosis in the presumptive neural epithelium, and cranial neural crest cells, respectively (51, 52). Here we show that Msx2 could play a cell-autonomous role because it is expressed in the caudal Wolffian duct during regression. Previous studies showed that perinatal DES exposure led to persistence of these ducts into adulthood through an unknown mechanism (22). Our results suggest that one likely mechanism is through the repression of Msx2 expression and subsequent apoptosis in Wolffian duct remnants. We observed activated Caspase-3 and Caspase-9 in wild type degenerating Wolffian duct but not in Msx2 mutants indicating that this regression involves the mitochondrial apoptotic pathway. On the other hand, we did not detect changes in Bcl-2 and Birc1a expression in Msx2 mutants suggesting that Msx2 may promote apoptosis by affecting the expression of other molecules in the apoptotic pathway. It is noteworthy that the cranial Wolffian duct in Msx2-mutant females degenerates normally, indicating that differences exist during regression of different regions of the Wolffian duct along the anteroposterior axis. The caudal Wolffian duct is surrounded by vaginal stroma which may provide a survival factor for the ductal remnant in the absence of Msx2, whereas in wild-type vagina, this effect may be overwhelmed by the proapoptotic function of Msx2. We stress that processes other than apoptosis may be involved in normal Wolffian duct regression and Msx2 could play a role in those processes as well.

Msx2 Modulates FRT Response to DES

For more than two decades neonatal DES exposure in humans and mice has been known to lead to a plethora of reproductive tract developmental abnormalities. Nevertheless the molecular mechanism underlying this perturbation remains unclear. Moreover, response to DES varies among different mouse strains; this indicates the importance of gene-environment interaction during organogenesis (53, 54). Thus identifying genetic loci that modify DES-induced phenotypes is important because these genes are likely to be crucial for normal FRT patterning as well. Recently, Wnt7a and Msx2 knockout mice have shown an abnormal uterine response to DES exposure (23, 29). We show that the loss of Msx2 renders the FRT highly responsive to DES exposure resulting in much more severe vaginal developmental defects compared with those in wild-type controls. These early phenotypes include a failure of vaginal stratification and a severely dilated vaginal lumen. One possible explanation is that Msx2 mutation invokes a global failure of hormonal responsiveness such that the mutant FRT cannot respond to DES treatment. However, our data do not support this scenario. First, if this were true, DES treated Msx2-mutant vaginal epithelium should resemble oil-treated control which is not the case. Second, Msx2−/− mice exhibit normal fertility and estrous cycle indicating that both the mutant uterus and vagina can properly respond to endogenous estrogen. Therefore, these results indicate a specific role for Msx2 in FRT responsiveness to DES exposure. In Msx2-mutant vagina, DES treatment leads to a complete abolishment of p63 expression in the epithelium. Such a loss of p63 expression should be sufficient to explain this phenotype because p63−/− vaginal epithelium fails to stratify (55). In this scenario, Msx2 functions to counteract DES repression on p63 expression. Thus in wild type vagina, DES can repress p63 expression in some cells through partial repression of Msx2 whereas in the absence of this counteracting factor, DES completely inhibits p63 expression in all epithelial cells of the mutant Müllerian vagina. At present the molecular mechanism for this complex regulation of p63 is unclear. One possibility is that Msx2 may participate in the formation of a transcriptional complex which positively regulates p63 expression and competes with a repressive complex activated by DES. The fact that p63 expression is lost in a fraction of Msx2−/− vaginal epithelial cells supports this hypothesis. As a consequence of loss of p63 expression, the mutant epithelium remains unstratified. This unstratified vaginal epithelium will likely remain as simple columnar epithelium and eventually develop into vaginal adenosis.

In addition to the failure of vaginal stratification, Msx2-mutant vaginal lumen is dramatically dilated as a result of abnormal water trafficking. This phenotype may be caused by the combination of two defects: loss of barrier function for the vaginal epithelium and failure of water transport from the vaginal lumen back to the stroma. The barrier function of the vagina is provided by the upper layers of epithelium (56). Since these layers are absent in DES-treated Msx2−/− vagina, the barrier function is presumably lost which results in excessive water leakage into the lumen. On the other hand, water resorption from the lumen back into vaginal tissue may also be defective. Two water channel proteins, Aqp3 and Aqp4, fail to be induced by DES in mutant vaginal epithelium. In the kidney, both proteins are detected in the basolateral membrane of the collecting tubule epithelium, consistent with their roles in water resorption (57). Aqp3 and Aqp4 double mutants are polyuric and have defects in water transport from the collecting duct lumen (58). The induction of these proteins by DES in the vaginal epithelium could be in response to fluid pressure in the vaginal lumen as a result of water imbibition, a part of the physiological response to estrogen stimulation. This induction by DES fails in Msx2-mutant vagina probably because the tissue layers that express these two aquaporins are missing. The absence of these two aquaporins in DES-treated Msx2-mutant vagina could lead to a water transport defect from the lumen back into the tissue for the lymphatic system to recycle, thus result in excessive fluid trapped in the vaginal lumen. This trapped water has no where to go because the vaginal opening does not occur until around P35 (59). Therefore our data suggest that failure of epithelial stratification is the primary defect in DES-treated Msx2-mutant vagina, and the abnormal water trafficking defect may result from a cell-fate change which leads to both a loss of barrier function and a lack of water channel protein expression. We propose that DES causes reproductive malformations by altering genetic pathways that govern normal FRT morphogenesis. For example, Wnt, Hox, and Msx genes are important regulators of FRT development. By partially repressing these genes, DES affects FRT development. When these genes are absent, DES can freely exert its teratogenic effects on FRT development. In this sense, Msx2 appears to counteract the effect of DES on FRT development because DES can elicit a much more dramatic FRT phenotype in mice lacking Msx2. Our data also suggest the possibility that protective mechanisms may exist to dampen the effects of any environmental hormones that might influence development or physiology and people with genetic defects may be more sensitive to exogenous hormones.

MATERIALS AND METHODS

Mice Maintenance and DES Administration

Mice were handled according to National Institutes of Health guidelines and in compliance with an animal protocol approved by Washington University. Msx2−/− mice were generated by gene targeting as previously described and were maintained on CD1 genetic background (39). CD1 mice were obtained from Charles River Laboratories, Inc. (Wilmington, MA). DES (Sigma-Aldrich, St. Louis, MO) was dissolved in ethanol at 10 mg/ml and further diluted 1:100 in corn oil. Female pups were injected from P1 to P5 subcutaneously with 20 μl of either corn oil or DES (2 μg/pup/day). Mice were sacrificed 6 hours after the last injection, and the FRT was removed and subjected to either RNA-extraction or fixation.

In Situ Hybridization

Immediately after dissection, vaginal tissues were rinsed in phosphate-buffered saline (pH 7.6), fixed overnight in 4% paraformaldehyde at 4°C, dehydrated, and embedded in paraffin following routine protocol. The template for making p63 anti-sense probe was generated by PCR using gene-specific primers (Supplementary Table 1) and vaginal cDNA, followed by subsequent cloning of the PCR product into pCR4-TOPO vector (Invitrogen, CA). The construct was sequence verified and linearized with NotI and transcribed with T3 RNA polymerases in the presence of ³⁵S-rUTP. Msx2 cRNA probe was generated as previously described (60). Pax2 cRNA probe was generated by digesting a Pax2 cDNA clone with BamHI and transcribed in the presence of ³⁵S-UTP with T3 RNA polymerases. In situ hybridization was carried out on 10-μm sections as described previously (61).

Reverse Transcription-Polymerase Chain Reaction (RT-PCR) and Real-Time RT-PCR

Total RNA was extracted from whole Müllerian vaginal tissue using RNA STAT-60 reagent (Tel-Test, Inc., Friendswood, Texas) following the manufacture’s instructions. cDNA was produced using SuperScriptII reverse transcriptase (Invitrogen) in the presence of oligo-dT primers. Primers for Aqp1-9 were designed according to Offenberg et. al. (62), others were designed using Primer3 online software (Whitehead Institute for Biomedical Research) (Supplementary Table. 1). PCR was performed in 20 μl reactions using gene-specific primers listed in Supplementay Table1. The results were visualized and analyzed in ethidium bromide-stained 1% agarose gels. SYBR green-based real-time RT-PCR was performed in 25 μl reaction volume with Applied Biosystems 7300 Real-Time PCR systems. PCR conditions were as follows: 2 minutes each at 50°C and 95°C followed by 40 cycles of 15 sec at 95°C, 30 sec at 60°C and 30 sec at 72°C. To ensure PCR specificity, negative controls included no template, no primers and using RNA as templates. The comparative threshold cycle (Ct) values were determined using the Applied Biosystems software. The relative amount of PCR products was obtained by subtracting the Ct of β-actin from the Cts of genes of interest. The difference between the Cts of DES-treated samples and oil-treated controls were assigned as ΔΔCt and the fold change was calculated as 2^{− ΔΔCt}. All PCR products were subsequently cloned into pCR4-TOPO vector (Invitrogen) and sequences verified.

Immunohistochemistry

Bouin’s (Biopharm Inc. Hatfield, AR)-fixed 5μm vaginal sections were rehydrated and rinsed in phosphate-buffered saline (PBS), followed by boiling in Trilogy (Cell Marque, Hot Springs, AR) for 15 min. After washed in PBS, slides were blocked in 3% normal goat serum (Invitrogen) and 1% BSA in PBS overnight at 4°C. Rabbit polyclonal antibodies were diluted as following concentrations in blocking solution and incubated at room temperature for 2 hours: anti–Aqp3 (1:1000, Chemicon, Temecula, CA), anti-Aqp4 (1:1000, Chemicon), anti-p63 (1:50, Santa Cruz Biotechnology, Santa Cruz, CA), anti-Tgfβ2 (1:100, Santa Cruz), anti-Tgfβ3 (1:100, Santa Cruz), anti-active-Caspase-3 (1:20, Chemicon) and anti-active-Caspase-9 (1:100, Cell Signaling, Danvers, MA). After three PBS washes, Alexa Fluor^® dye-labeled secondary goat anti-rabbit IgG (Invitrogen) was diluted 1:2000 and applied to the slides for 1 hour at room temperature. Slides were washed, mounted with Vectashield containing DAPI (Vector Laboratories, Inc., Burlingame, CA), and examined under a fluorescent microscope.

Bromodeoxyuridine (BrdU) Incorporation and TUNEL Assay

Mice were injected subcutaneously with 1.5 ml/100 g body weight of BrdU labeling reagent 1 hour before sacrifice (Roche Diagnostic Corp., Indianapolis, IN). Vaginal tissues were harvested and fixed in Carnoy’s fixative (10% acedic acid, 30 % chloroform and 60 % Ethyl Alcohol). BrdU assay was carried out with a 5-Bromo-2'-deoxyuridine Labeling and Detection Kit II (Roche Diagnostic Corp.), according to the manufacturer’s instructions. TUNEL assay was performed on paraformaldehyde-fixed sections with the In Situ Cell Death Kit (Roche Diagnostic Corp.), according to the manufacturer’s instructions. Apoptotic rate of degenerating Wolffian ducts was obtained by dividing the number of TUNEL positive cells by the total number of ductal epithelial cells in 6 vaginal sections from three different individuals of both wild type and Msx2−/− at each time point.