Homeodomain Transcription Factor Msx2 Regulates Uterine Progenitor Cell Response to Diethylstilbestrol

The fate of mouse uterine epithelial progenitor cells is determined between postnatal days 5 to 7. Around this critical time window, exposure to an endocrine disruptor, diethylstilbestrol (DES), can profoundly alter uterine cytodifferentiation. We have shown previously that a homeo domain transcription factor MSX-2 plays an important role in DES-responsiveness in the female reproductive tract (FRT). Mutant FRTs exhibited a much more severe phenotype when treated with DES, accompanied by gene expression changes that are dependent on Msx2. To better understand the role that MSX-2 plays in uterine response to DES, we performed global gene expression profiling experiment in mice lacking Msx2. By comparing this result to our previously published microarray data performed on wild-type mice, we extracted common and differentially regulated genes in the two genotypes. In so doing, we identified potential downstream targets of MSX-2, as well as genes whose regulation by DES is modulated through MSX-2. Discovery of these genes will lead to a better understanding of how DES, and possibly other endocrine disruptors, affects reproductive organ development.


Introduction
Developmental exposure to endocrine disrupting chemicals (EDCs) can have catastrophic consequences in humans and animals, including malformations of various organs and tissues and increased risk for certain diseases, such as infertility, obesity and cancers [1].For example, in humans, in utero exposure to diethylstilbestrol (DES), a synthetic estrogen commonly prescribed to pregnant women for the prevention or treatment of miscarriage in the 40's to 70's, led to patterning defects in the female reproductive tract (FRT) and clear cell carcinomas of the cervix and vagina in affected individuals [2].Similar defects were found in laboratory animal models, including uterine metaplasia, loss of uterotubal junction and vaginal adenosis.The postnatal mouse DES model was established by Dr. John McLachlan's group [3,4], and was widely adopted as the experimental model system to better understand the molecular mechanism underlying the observed DES effects.This model involves subcutaneous injection of DES from postnatal day 1 to day 5.The murine female reproductive tract develops mainly from the Müllerian duct.Classical tissue recombination experiments demonstrated that Müllerian epithelial cell fate is determined between postnatal days 5-7 in the mouse by signals from the underlying mesenchymal cells [5].More specifically, it was shown recently that BMP4/Activin-A signaling is activated in the vaginal mesenchyme to promote epithelial stratification [6].Thus before postnatal day 5, the Müllerian epithelial cells are pluripotent and are sensitive to EDC exposure.It was shown that DES exerts its function mainly through estrogen receptor α, as most if not all DES effects were absent in ERα knockouts [7].We have focused our previous work on understanding the immediate gene expression and cellular changes induced by DES exposure [8][9][10].Our rationale is that gene expression changes in early development will have a long-lasting effect on organ function in adults.We showed that neonatal DES exposure in mice caused immediate gene expression changes, especially in the uterine epithelium, and the affected genes included many transcription factors, signaling pathway components, and growth factors [9].Intriguingly, peroxisome proliferator-activated receptor gamma (Pparγ) and a group of genes critical to lipid metabolism, showed significant changes in expression level in the uterine epithelium, which led to increased lipid droplets accumulation in merely five days after DES exposure in the same tissue layer.In addition, DES exposure changed glucose metabolism, water/small molecule trafficking, proliferation as well as apoptosis profiles of the uterine epithelial cells.
MSX-2 homeodomain transcription factor is required during organogenesis by mediating epithelial-mesenchymal interactions.Msx2 is highly expressed in a variety of tissues during embryogenesis, and is critical for the terminal differentiation of many tissue types and organs, including the limb, bone, hair follicle, nail, and tooth [11][12][13][14][15].In wild type neonates, Msx2 is highly expressed in the uterine as well as vaginal epithelium, and its expression is significantly reduced in both tissues upon DES exposure [8,10].Moreover, Msx2 mutant reproductive tract appears to be a sensitized background for DES exposure and the FRTs in these mice exhibited a much more severe patterning defects in response to DES compared to wild type controls [10].These data led us to hypothesize that DES may elicit its cellular effects through down-regulation of a master regulator Msx2, which in turn led to gene expression changes of an array of down-stream targets via transcription activation/repression.To further understand the role of Msx2 in modulating FRT response to DES, here we performed global gene expression profiling experiment in vehicle-or DEStreated Msx2-null mice and compared this result to our previously published microarray data on wild-type animals.The differential gene regulation by DES in wild-type and Msx2-/-mice revealed many genes whose expression and/or regulation by DES depends on Msx2 in the developing uterus.

Neonatal DES Exposure
In an attempt to identify potential MSX-2 targets that mediate the DES effects, we performed cDNA microarray using uterine RNA harvested from vehicle (oil)-or DES-treated Msx2-/-neonatal mice.Among the 46628 annotated genes included on the Affymetrix microarray chips, 14025 were expressed in uterine tissues, with or without DES treatment.When comparing data set treated with  DES to that of the vehicle-treated control group, we identified 1459 annotated genes that were differentially regulated by DES exposure in Msx2-/-uterus with a p-value less than 0.05 (data not shown).These genes were referred to as Differentially Regulated Genes (DRGs) hereinafter.Gene ontology analysis revealed that these DRGs were involved in a plethora of cellular processes, including development, response to stimulus, and anatomical structure development (Figure 1A).Since Msx2 is normally highly expressed in the uterine epithelium, its targets would presumably reside in the same tissue layer.To test this hypothesis, we compared the expression of these DRGs to our previous microarray data conducted on wild type neonatal uteri.Since the two microarrays were performed on different platforms, only 1254 out of 1459 DRGs overlapped between the two studies.Nevertheless, more than 80% (1025 out of 1254) of those DRGs were detected in the wild-type uterine epithelium, indicating that Msx2 is indeed functioning as a transcriptional regulator in the uterine epithelium.To focus on potential transcriptional targets of Msx2, we carefully excluded DRGs that also showed mesenchymal expression from our previous tissue-specific microarray data (GEO# GSE37969, [9], and only compared our current array data to that obtained using wild-type uterine epithelial RNA. A total of 212 genes showed similar expression change trends and fold-changes in the uterine epithelium (Table 1, Group I), indicating that their regulation by DES is not mediated through MSX-2 transcriptional activation/suppression.We plotted fold changes of these genes in both wild-type and Msx2-/-uteri on a star glyph, and the two sets of curves fit almost perfectly (Figure 1B).On the other hand, expression of 198 genes were regulated either in the opposite direction by DES or that their fold changes were changed dramatically (equal or greater than four-fold) in the two different genotypes (Table 1, Groups II-III).When plotted the log-scaled fold-changes of these genes on a star glyph, noticeable differences in the positions of peaks were observed (Figure 1C).These genes, or at least a subset of them, could be MSX-2 targets.

Validation of microarray results
To validate our microarray findings, quantitative real-time RT-PCR (qRT-PCR) was performed on RNA samples extracted from isolated control (Msx2+/-) and Msx2-/-neonatal uterine epithelia, exposed to either oil or DES.As a positive control, we first tested some Group I DRGs that showed similar regulation by DES in both genotypes.Wnt7a, a ligand for the canonical WNT/βcatenin pathway that is critical for many cellular processes, is highly expressed in the developing uterine epithelium and is responsible for uterine gland formation [16,17].We and others have previously shown that its expression was strongly suppressed by DES administration [8,18,19].In Msx2-/-uterus, a similar effect was observed by qRT-PCR (Figure 2).Expression of Aqp3, a gene encoding one of the aquaporin water trafficking proteins, was induced dramatically by DES in the wild type array.A similar regulation was observed in the Msx2-/-microarray and confirmed by qRT-PCR (Figure 2).Similar results were obtained from another group I gene, Rasa4 (Figure 2).Rpl7, a housekeeping gene whose expression level was not affected by DES in either genotype, was used in the qRT-PCR to determine relative transcript level of each DRG, employing the comparative Ct method.
Next, we classified the 198 DRGs that showed differential regulation by DES in Msx2-/-uterus into two major categories, as shown in Table 1 (Group II-III).Group II contained genes whose expression was either up-or down-regulated by more than twofold by DES in the control uterus, but showed different regulatory patterns in the absence of Msx2.Group II was further divided in to three subgroups: Group IIa comprised of 114 DRGs whose DESregulation was either lost or attenuated in Msx2-/-by fourfold or more; Group IIb included 33 DRGs whose DES-regulation was enhanced in Msx2-/-; and Group IIc contained 24 DRGs that showed opposite regulation by DES in the two genotypes.Group III consisted of 27 DRGs that showed expression changes by more than twofold when exposed to DES only in Msx2-/-, but not in wild-type.We selected a few genes from each group to verify gene expression changes by qRT-PCR.As shown in Figure 2, H2afy from Group IIa, a gene encoding a histone H2A variant, was induced markedly by DES in the Msx2+/-control animals, but to a lesser extent in Msx2-/-uterus (Figure 2).Similarly, up-regulation of Prkg1 was only detected in DES-treated control but not in mutant uterus.Another examples of Group IIa genes was Krt17 (Krt1-17), a keratin family member.Expression of Krt17 was decreased more than 6-fold upon DES treatment.In the Msx2-/-, however, this decrease was attenuated (Figure 2).These results indicate that MSX-2 plays an obligatory role in the DES regulation of Group IIa genes under normal circumstances, and in its absence, the DES-induced gene expression changes are abolished.It is noteworthy that the majority of Group IIa DRGs (92 out of 114) were down-regulated by DES in the wild-type, but this down-regulation was diminished in Msx2-/-mice.As clearly visualized in Fig. 1C, these genes were repressed by DES in the wild-type (blue peaks pointing inwards), while the repression was no longer detectable or much reduced in Msx2-/-uteri (corresponding inconspicuous red peaks).These results support the notion that MSX-2 acts as a transcriptional suppressor and suggest that it is involved in the DES-suppression of these genes.
An example of Group IIb genes was Avil, encoding the protein advillin that may be involved in the morphogenesis of microvilli.Microarray data showed an almost tenfold increase in expression by DES in Msx2-/-when compared to controls.By qRT-PCR we found the induction by DES in Msx2-/-was actually 107-times more in the mutant (Figure 2).Such discrepancy was not unusual due to the difference in nature of the two assay systems.Two other group IIb genes, Prap1 and Sftpd, showed similar trend in expression (Figure 2).These results suggest that in the normal uterus, Msx2

Comparison of DRGs by DES between wild-type and Msx2-/-uterus.
Table 1: List of DRGs found in Msx2-/-uterus that showed similar (group I) or differential (groups II and III) DES regulation compared to wild-type.Each DRG is presented as "gene name, fold-change in wild-type (fold-change in Msx2-/-).antagonizes transcriptional activation of Group IIb genes by DES, and in its absence, this effect is augmented.Group IIc genes showed a more complicated regulation, where DES exhibited opposite regulation on gene expression in the control and Msx2-/-uterus.For example, Fbp2, a gene encoding a gluconeogenesis regulatory enzyme, was up-regulated in the control uterus upon DES exposure, but was down-regulated by DES in Msx2-/-uterus (Figure 2).Similar results were obtained for Pdzrn3, a gene encoding a PDZ domain containing E3 ubiquitin ligase.It is likely that Msx2 is involved in the DES regulation of these genes, but the exact mechanisms remains unclear at present.Group III genes showed no apparent expression changes in the control uterus by DES, but were either up-or downregulated by more than twofold in the Msx2 -/-when exposed to DES.Expression of two such genes, Slc39a8 and Kcnn4, were validated by qRT-PCR (Figure 2).These genes are similar to Group IIb genes in the sense that MSX-2 antagonizes DES-regulation, and DES effect on gene expression changes was only observed in Msx2-/-uterus.

Msx2 Is Dispensable for the DES-induced Changes in Adipogenesis and Lipid Metabolism in the Uterine Epithelium
As mentioned in the introduction, we have previously reported that neonatal DES exposure altered lipid metabolism in the uterine epithelium through upregulation of a master transcription factor of adipogenesis, PPARγ [9].To investigate whether Msx2 plays a role in this process, we analyzed expression of Pparγ in the control and Msx2-/-uterus.As shown in Figure 3A, Pparγ is upregulated by DES, regardless of Msx2 status, indicating that this process is not mediated through Msx2.Many Pparγ downstream targets including Acsl1 and Arntl, and genes involved in lipid metabolism and trafficking (e.g.Agpat2, Slc2a1), also showed similar regulation by DES in Msx2-/-uterus (Figure 3A and Table 1).We next examined frozen uterine sections stained with Oil red O, a lysochrome diazo dye that binds neutral lipids, for signs of lipid metabolism changes.As we have previously reported, administration of DES caused increased lipid deposition exclusively in the uterine epithelium (Figure 3B, C).Similar results were obtained in Msx2-/-uteri (Figure 3D, E).These results collectively demonstrated that the DES-induced lipid metabolism changes in the uterine epithelium through Pparγ upregulation is Msx2-independent.

Discussion
Since the discovery of the adverse effects of DES on human health, numerous studies in model animals have been conducted trying to uncover the underlying molecular mechanisms.The neonatal DES mouse model is widely used, because DES-treated animals exhibited a variety of developmental abnormalities and pathologies recapitulating the human syndrome including uterine metaplasia and vaginal adenosis.We have previously conducted global gene profiling studies, and found that neonatal DES treatment in mice caused a series cellular changes including proliferation, apoptosis, metabolic as well as other physiological changes through modulation of uterine gene expression [8,9].We have also discovered that the transcription factor MSX-2, can modulate the responsiveness of uterus and vagina to DES [10].To better understand how Msx2 modulates DES response in the uterus, here we performed microarray analyses using RNA extracted from oil-or DES-treated Msx2-/-uteri, and compare this result to that of our previous array data performed on wild-type animals.By comparing the two data sets, we uncovered three groups of genes in general: group I contains genes that are similarly regulated by DES both in wild type and in Msx2-/-uteri.Thus Msx2 is dispensable for the regulation of this group of 212 genes by DES.Group II consists of 171 genes whose regulation by DES is altered (in either direction) in Msx2 mutant uteri.Thus Msx2 plays important roles in modulating their regulation by DES.Group III consists of 27 genes that are not regulated by DES in wild type uterus, but are now DES targets in Msx2 mutant uteri.These genes are likely not MSX-2 transcription targets but rather exhibit altered DES responsiveness due to secondary changes, such as cell fate change.We further subdivided group II genes into three categories: genes whose regulations by DES are either lost or attenuated in Msx2 mutant uterus; those whose regulations are enhanced by Msx2 mutation; and those oppositely regulated by DES in the two genotypes.RT-PCR was used to validate representative genes from each subgroup.Interestingly, we observed that most DES-induced genes showed a greater response in Msx2-/-uterus, evidenced by the red peaks in the star glyph (Figure 1C).On the other hand, DES-repressed genes lost this repressive regulation in Msx2 mutant uterus, evidenced by the blue peaks (Figure 1C).These results are in agreement with previous studies suggesting that Msx2 mainly functions as a transcriptional repressor [20].In this scenario, DES can induce target gene expression to a greater degree due to the removal of a repressor that normally counteracts this induction.On the other hand, repression of certain gene expression by DES may require MSX-2 participation, as its absence leads to failure of repression by DES.
We previously showed that the regulation of Aqp3 by DES was abolished in Msx2 mutant vaginal epithelium [10].Surprisingly, here we found that Aqp3 was similarly induced in both wild-type and Msx2 mutant uterus.These seemingly contradictory results can be reconciled by the fact that the intermediate cell layers expressing Aqp3 in the developing vagina are missing in Msx2 mutants, and thus failure of induction by DES in Msx2 mutant vagina is secondary to a defective patterning event.In this sense, Aqp3 is likely not a target of Msx2 either in the uterus or in the vagina.
We showed recently that lipid metabolism in the neonatal uterus was dramatically affected by DES exposure [9].This change is mediated through the upregulation of a transcription factor PPARγ, which in turn led to gene expression changes in an array of genes involved in lipid trafficking and metabolism.By analyzing the expression of those genes in DES-treated Msx2-/-uterus, we found that Msx2 is not involved in the DES-regulation of this process.Pparγ was similarly induced in Msx2-/-as in controls, so were many Pparγ downstream genes.Consistently, lipid droplet accumulation in the uterine epithelium was also observed in the Msx2-/-uterus upon DES exposure.Interestingly, it was previously reported that in myofibroblasts, Msx2 stimulates osteogenesis but suppress adipogenesis by activating Osx transcription and concomitantly suppressing Pparγ expression [21].Our results that Pparγ showed no change in basal level in the absence of Msx2 indicate that Msx2 is unlikely to be a transcriptional suppressor of Pparγ, at least in the uterus.In addition, DES increases Pparγ, and alters lipid metabolism in the uterus in an Msx2-independent way.

Mice
All mice were housed in the animal facility at Washington University with controlled light/dark cycles and handled in accordance with National Institutes of Health guidelines.All procedures were approved by the Washington University Institutional Animal and Use Committee.Msx2 mutants were generated and described previously [11], and breeding between female Msx2+/-and male Msx2-/-were used to obtain control and homozygous pups.DES was prepared and injected from P1 to P5 as described previously [8].Uteri from female pups were harvested for fixation or RNA extraction 24 hours after the last injection.

cDNA Microarray and Data Analysis
Oil-or DES-treated P5 Msx2-/-mice were harvested for RNA preparation.Uterine tissues from 3-4 animals of the same treatment group were pooled together and RNA isolated to make one sample.Total RNA was cleaned using RNeasy Kit (Qiagen, Valencia, CA) and submitted to the Genome Technology Access Center (GTAC) at Washington University School of Medicine for microarray analysis on Affymetrix Genechip.A total of three biological replicates were used for microarray analysis.Normalized data were provided by the GTAC and detailed analysis protocol was available upon request.Subsequent analyses were performed and star glyphs generated in Microsoft Excel 2007.DRGs that showed high expression in the wild-type uterine mesenchymal tissue from our previous microarray data [9], either with or without DES treatment, were excluded from the current analysis to better focus on Msx2-dependent gene regulation.
Uterine Epithelium Isolation, RNA Extraction and Quantitative Reverse Transcription-Polymerase Chain Reaction (qRT-PCR) Uterine epithelium isolation was performed as described previously [9].Briefly, dissected neonatal uterine horns were cut into 3-4 mm long segments and incubated on a rotating platform in calcium-free, magnesium-free Hank's balanced salt solution containing 1% trypsin for 1 hour at 4°C.Enzymatic digestion was stopped by adding an equal volume of 5 mg/ml bovine serum albumin after the incubation, and epithelial tissues were extruded from the uterine horns by applying gentle pressure along each segment.Epithelial tissues from individual animals of the same genotype/treatment were transferred to fresh tubes and pooled together.Low speed centrifugation was applied to collect tissue pellets [22].Total RNA was isolated from the cell/tissue pellets with RNA Stat-60 (Tel-Test, Inc., Friendswood, TX) following manufacture's instruction.Primer design, reverse transcription and qRT-PCR were performed as previously described [23,24].Primer sequences used were as follows:

Figure 1 :
Figure 1: Differentially regulated genes by DES in Msx2-/-uterus.(A) Pie chart showing gene ontological classification of the 1459 DRGs.(B) Star glyph showing the fold changes in gene expression by DES of common DRGs between control (blue) and Msx2-/-(red).Each peak represents the Log2 value of fold change by DES of one gene, note that the peaks of wild-type and Msx2-/-for each given gene almost always overlap.The list of these common DRGs is in Table 1(Group I).(C) Star glyph showing the fold changes of the unique DRGs listed in Table 1 (Groups II-III).Note the drastic changes in the positions of wild-type peak and Msx2-/-peak for each gene.

Figure 2 :
Figure 2: Quantitative real-time RT-PCR validations of the microarray data in isolated neonatal uterine epithelium.Relative transcript level on logarithmic scale determined by qRT-PCR of two or three genes representing each group were shown.Two numbers under each gene name denote foldchanges by DES in control and Msx2-/-, with and without parentheses, respectively.Three biological replicates were included in each treatment group.

Future
investigations on the function of these MSX-2 target candidates are needed to further evaluate their function(s) during normal uterine development as well as during DES-induced pathogenesis of the female reproductive tracts.Nevertheless, identification of these genes may shed light on how Msx2 regulate terminal differentiation of the uterus, and possibly other Msx2expressing tissues.

Figure 3 :
Figure 3: DES alters uterine lipid metabolism in an Msx2-independent manner.(A) RT-PCR showing that activation of the adipogenic marker, Pparγ, as well as its downstream targets, is independent of Msx2.No RT controls were amplification of housekeeping Rpl7 genes using RNA as templates.(B-E) Oil red O staining showing DES treatment resulted in lipid droplets deposition in the basal-lateral side of uterine epithelium (arrows) in both control (C) and Msx2-/-(E).The basement membranes of oil-treated control (B) and Msx2-/-(D) were outlined with dotted lines.UE, uterine epithelium; UM, uterine mesenchyme.Scale bars: 50µm in B-E.

Table 1
(Group I).(C) Star glyph showing the fold changes of the unique DRGs listed in Table 1 (Groups II-III).Note the drastic changes in the positions of wild-type peak and Msx2-/-peak for each gene.