A repressive role of enhancer of zeste homolog 2 in 11β-hydroxysteroid dehydrogenase type 2 expression in the human placenta

The expression of 11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2), which acts as a placental glucocorticoid barrier, is silenced in cytotrophoblasts but substantially up-regulated during syncytialization. However, the repressive mechanism of 11β-HSD2 expression before syncytialization and how this repression is lifted during syncytialization remain mostly unresolved. Here we found that enhancer of zeste homolog 2 (EZH2) accounts for the silence of 11β-HSD2 expression via trimethylation of histone H3 lysine 27 at the promoter of the 11β-HSD2 gene. Further studies revealed that, upon syncytialization, human chorionic gonadotropin reduced the phosphorylation of retinoblastoma protein (pRB) via activation of the cAMP/PKA pathway, which sequesters E2F transcription factor 1 (E2F1), the transcription factor for EZH2 expression. As a result of inactivation of the pRB-E2F1-EZH2 pathway, the repressive marker trimethylation of histone H3 lysine 27 at the 11β-HSD2 promoter is removed, which leads to the robust expression of 11β-HSD2 during syncytialization.

In the human placenta, syncytialization is a differentiating process whereby mononuclear cytotrophoblasts fuse into multinuclear syncytiotrophoblasts (1,2). The syncytialized trophoblasts, by covering the placental villi that immerse in the maternal blood, form a nutrient exchange interface between mother and fetus as well as a frontline defense for the fetus against harmful substances from the maternal side (1). One such defensive role of syncytiotrophoblasts is to provide a barrier against maternal glucocorticoids that is about 10-fold higher than that on the fetal side (3). Although appropriate amounts of glucocorticoids facilitate the maturation of fetal organs, accumulating evidence has indicated that excessive glucocorticoids not only compromise fetal growth but also program the development of chronic diseases during the postnatal period (3,4). Therefore, this placental glucocorticoid barrier is critical for the normal development of the fetus. The establish-ment of the placental glucocorticoid barrier depends on the robust induction of 11␤-hydroxysteroid dehydrogenase type 2 (11␤-HSD2), 2 an exclusive oxidase catalyzing the conversion of biologically active cortisol into inactive cortisone, during syncytialization (4 -6). Intriguingly, as the precursors for syncytiotrophoblasts, cytotrophoblasts express very little 11␤-HSD2 (6,7). It remains obscure which strategy is adopted by cytotrophoblasts to suppress 11␤-HSD2 expression and how this suppression is lifted during syncytialization. Elucidating these issues will provide fundamental insights into the molecular mechanism underlying the establishment of the placental glucocorticoid barrier.
11␤-HSD2 is encoded by the HSD11B2 gene, which is rich in CpG islands (8). The tissue-specific expression of 11␤-HSD2 is associated with the methylation state of these CpG islands, particularly in the promoter region (8). In tissues that scarcely express 11␤-HSD2, HSD11B2 promoter is highly methylated, whereas in tissues that express 11␤-HSD2 abundantly such as the mineralocorticoid target organs, a low methylation rate has been revealed in the HSD11B2 promote (8). Likewise, the CpG islands in the HSD11B2 promoter in human placental tissue have also been found to have a low methylation frequency (8), which is supported by our findings that a low methylation rate of the HSD11B2 promoter is maintained throughout syncytialization (6), suggesting that cytotrophoblasts may employ an alternative mechanism rather than DNA methylation in the suppression of 11␤-HSD2 expression prior to syncytialization.
In addition to DNA methylation, epigenetic modifications of histones are also crucial in the regulation of gene expression by affecting chromosome functions. Among histone marks that repress gene transcription, trimethylation of histone H3 lysine 27 (H3K27me3) is a well recognized gene silencing mark (9). In addition, trimethylation of H3K27 is conducted by the enhancer of zeste homolog 2 (EZH2) and its cofactors suppressor of zeste 12 homolog (SUZ12) and embryonic ectoderm development (EED) (9). Of interest, ChIP sequencing revealed that HSD11B2 is among the genes that are immunoprecipitated by an antibody against EZH2 in several cell lines (10 -12). Our preliminary data revealed that syncytialization caused a dramatic decrease in EZH2 expression and a robust increase in 11␤-HSD2 expression reciprocally in human trophoblasts. In light of the prominent role of EZH2-mediated methylation of H3K27 in gene silencing and the fact that CpG domains commonly recruit EZH2 (13,14), it is feasible to propose that 11␤-HSD2 expression might be repressed via EZH2-mediated trimethylation of H3K27 in the cytotrophoblasts and that this repressive mark might be removed as a consequence of downregulation of EZH2 expression by the factors produced during syncytialization. Here we tested this hypothesis by using human villous tissues obtained from both early and term pregnancies.

Changes in EZH2 and 11␤-HSD2 abundance during syncytialization
Initially, immunofluorescent staining was conducted to examine whether EZH2 in the villous tissue manifested a unique distribution pattern. The identity of syncytiotrophoblasts was illustrated with staining for the ␤ subunit of hCG (␤-hCG), a well recognized marker for syncytiotrophoblasts Figure 1. Distribution patterns of EZH2 in relation to 11␤-HSD2 in human placental villous tissue and in trophoblasts before and after syncytialization. A and B, representative images showing intense staining of EZH2 (red) in the nuclei of the cytotrophoblast layer and weak staining of EZH2 in the nuclei of the syncytial layer of human chorionic villi in both early (A) and term (B) pregnancies. The syncytial layer was stained for ␤-hCG (green). Nuclei were counterstained blue with DAPI. n ϭ 5. C and D, representative images showing a reciprocal distribution pattern of EZH2 and 11␤-HSD2 in human chorionic villi in both early (C) and term (D) pregnancies. EZH2 staining (red) is detected mainly in cytotrophoblasts, whereas 11␤-HSD2 (green) is found in the syncytial layer. Nuclei were counterstained blue with DAPI. n ϭ 5. E, representative images showing strong EZH2 staining and weak 11␤-HSD2 staining in trophoblasts before syncytialization (2 h) and weak EZH2 staining and strong 11␤-HSD2 staining in trophoblasts after syncytialization (72 h). The cells were counterstained with hematoxylin. n ϭ 5. (2). In the villous tissue obtained from both early (Fig. 1A) and term ( Fig. 1B) pregnancies, EZH2 was localized dominantly in the nuclei of cytotrophoblasts, and weak staining for EZH2 was observed in the nuclei of syncytiotrophoblasts and interstitial cells, suggesting that EZH2 functions mainly in the nuclei of cytotrophoblasts. Further examination of the distribution of EZH2 in relation to 11␤-HSD2 in villous tissue collected from both early (Fig. 1C) and term (Fig. 1D) pregnancies revealed that the cytotrophoblast layer with little staining for 11␤-HSD2 was intensely stained for EZH2, whereas the syncytial layer carrying intense staining for 11␤-HSD2 was faintly stained for EZH2.

EZH2 represses placental 11␤-HSD2 expression
To observe the dynamic change of EZH2 expression in relation to 11␤-HSD2 expression during syncytialization, we employed primary human cytotrophoblasts, which are capable of fusing spontaneously into large syncytial clumps in vitro (15). Immunocytochemical staining revealed that strong EZH2 staining was detected in the nuclei of trophoblasts before syncytialization (2 h) and became rather weak after syncytialization (72 h). On the contrary, the 11␤-HSD2 staining signal, which was observed sparsely in cytotrophoblasts, became rather intense in syncytiotrophoblasts (Fig. 1E). Further studies demonstrated that the abundance of EZH2 mRNA and protein was dramatically decreased in trophoblasts after 24-h incubation, which was accompanied by a concurrent increase in 11␤-HSD2 mRNA and protein abundance as well as a simultaneous reduction in H3K27me3 and H3K27me2 (Fig. 2, A-C). A time course study within the first 24 h of cell culture revealed that the change in EZH2 abundance occurred ahead of the changes in H3K27me3 and 11␤-HSD2 abundance (Fig. 2, D and E). These data suggest that a negative correlation between EZH2 and 11␤-HSD2 expression exists in trophoblasts and that the down-regulation of EZH2 expression is very likely the causative factor for the up-regulation of 11␤-HSD2 expression by removing H3K27me3 from the HSD11B2 promoter during syncytialization.

The role of EZH2 in the regulation of 11␤-HSD2 expression in human placental trophoblasts
To examine the role of EZH2 in the regulation of 11␤-HSD2 expression in trophoblasts, we first determined by ChIP assays whether EZH2 is enriched at the HSD11B2 promoter in human cytotrophoblasts. We found that the enrichment of EZH2 as well as SUZ12 and EED, the cofactors for EZH2, was significantly more abundant at the HSD11B2 promoter in trophoblasts before syncytialization compared with that in syncytiotrophoblasts. The changes in H3K27me3 enrichment at the HSD11B2 promoter manifested a pattern similar to that of EZH2 during syncytialization. Studies of the enrichment of these proteins at the housekeeping gene GAPDH were performed in parallel as a negative control (Fig. 3A).
To further clarify the role of EZH2 in the regulation of 11␤-HSD2 expression, 2 h after plating, cytotrophoblasts were treated with EPZ005687 (EPZ) or GSK343 (GSK), the specific antagonists for EZH2 (16,17), for 48 h. EPZ or GSK treatment increased the abundance of both 11␤-HSD2 mRNA and protein significantly (Fig. 3B). Consistently, siRNA-mediated knockdown of EZH2 also increased the , and H3K27me3 (n ϭ 5) at the promoter of HSD11B2 (left) and low enrichment of these proteins at the promoter of the housekeeping gene GAPDH (right). This enrichment at the promoter of HSD11B2 but not at the promoter of GAPDH was reduced significantly after syncytialization of trophoblasts. Preimmune IgG (n ϭ 5) served as the negative control. Top panel, the location of CpG islands, the putative EZH2 binding region, and the spanning region of primers used for ChIP in the HSD11B2 promoter. TSS, transcription start site. B, treatment with EPZ (20 M) or GSK (10 M), inhibitors for EHZ2, for 48 h significantly up-regulated the amounts of 11␤-HSD2 mRNA and protein in trophoblasts. n ϭ 4. Ctr, control. C and D, siRNA-mediated knockdown of EZH2 (siEZH2) significantly up-regulated the abundance of 11␤-HSD2 mRNA (n ϭ 6) and protein (n ϭ 4). Randomly scrambled siRNA served as a negative control (Nc). E and F, overexpressing EZH2 for 48 h decreased the abundance of 11␤-HSD2 mRNA (n ϭ 5) and protein (n ϭ 6). The empty vector GV230 served as a control. G, ChIP assays showed that the enrichment of both EZH2 and H3K27me3 at the HSD11B2 promoter was increased by overexpression of EZH2 for 48 h.

EZH2 represses placental 11␤-HSD2 expression
amounts of 11␤-HSD2 mRNA and protein in trophoblasts (Fig. 3, C and D), whereas overexpressing EZH2 for 48 h significantly decreased 11␤-HSD2 mRNA and protein abundance in trophoblasts (Fig. 3, E and F). The ChIP assay showed that overexpression of EZH2 significantly enhanced the enrichment of EZH2 and H3K27me3 at the HSD11B2 promoter in trophoblasts (Fig. 3G). Accordingly, EPZ005687 treatment decreased the enrichment of H3K27me3 significantly, but not EZH2 per se, at the HSD11B2 promoter (Fig. 3H), whereas the enrichment of H3K27 acetylation (H3K27ac), a histone marker associated with the up-regulation of HSD11B2 transcription during syncytialization (5), was increased significantly at the HSD11B2 promoter (Fig.  3H). As removal of the methyl group is a prerequisite for histone acetylation (18), we assume that the reduction in H3K27me3 enrichment at the HSD11B2 promoter may predispose H3K27 to acetylation during syncytialization. Altogether, these data indicate that EZH2-mediated H3K27 methylation is responsible for the repression of 11␤-HSD2 expression before syncytialization and that this repression is lifted via down-regulation of EZH2 expression during syncytialization.

Involvement of the pRB-E2F1 pathway in the regulation of placental EZH2 expression
Previous studies in tumor cells have demonstrated that EZH2 is under the transcriptional regulation of E2F transcription factor 1 (E2F1) (19). The transcriptional activity of E2F1 is suppressed when it is bound to retinoblastoma protein (pRB). When pRB is phosphorylated, E2F1 becomes free and regains its transcriptive activity (19,20). Consistent with the pattern of EZH2 localization in villous tissue, stronger staining for phosphorylated pRB (p-pRB) was observed in cytotrophoblasts compared with syncytiotrophoblasts in early pregnancy (Fig.  4A). The time course study within the first 8 h showed that the reduction in p-pRB abundance occurred before the reduction in EZH2 abundance, although the abundance of E2F1 was barely changed (Fig. 4B). These data suggest that the amount of E2F1 bound to pRB is increased, which may account for the reduced transcriptional activity of E2F1 and the subsequent attenuation of EZH2 expression during syncytialization.
The role of E2F1 in the regulation of EZH2 expression was further examined in trophoblasts. A ChIP assay showed that the enrichment of E2F1 at the EZH2 promoter was more abundant before syncytialization than after syncytialization (Fig. 4C). Knockdown of E2F1 expression by siRNA significantly reduced EZH2 abundance but increased 11␤-HSD2 abundance in trophoblasts (Fig. 4D). These data indicate that E2F1 is the transcription factor that stimulates EZH2 expression in cytotrophoblasts and that, during syncytialization, E2F1 is sequestered by pRB, resulting in decreased expression of EZH2.

The human CG-activated cAMP/PKA pathway is responsible for the removal of repression on 11␤-HSD2 expression during syncytialization
It is well recognized that the hCG-activated cAMP/PKA pathway is pivotal for syncytialization (2,21). Moreover, the cAMP/PKA pathway can attenuate pRB phosphorylation in some tumor cells (22,23). Therefore, we hypothesized that the hCG-activated cAMP/PKA pathway might be the upstream signal that directs the down-regulation of EZH2 expression during syncytialization. We found that treatment with dibutyryl cyclic AMP (Bt 2 cAMP), an analogue of cAMP, or forskolin, an activator for adenylyl cyclase, decreased the amounts of phosphorylated pRB and EZH2 while increasing 11␤-HSD2 abundance (Fig. 5A). Consistently, neutralization of hCG with ␤-hCG antibody or inhibition of PKA with H89 significantly increased the abundance of p-pRB and EZH2 while decreasing 11␤-HSD2 abundance (Fig. 5B). Despite an unaltered abundance of total E2F1 by manipulation of the hCG-cAMP-PKA pathway (supplemental Fig. S1), Bt 2 cAMP treatment significantly reduced the enrichment of E2F1 at the EZH2 promoter (Fig. 5C), which is possibly due to the reduction in the amount of active E2F1. These data suggest that the hCG-activated cAMP-PKA pathway is the upstream signal that causes the down-regulation of EZH2 expression via inactivation of the pRB-E2F1 pathway during syncytialization.

Discussion
This study demonstrates that EZH2 plays a repressive role in the regulation of 11␤-HSD2 expression before syncytialization by way of trimethylating H3K27. We believe that the transcription factor E2F1 drives the abundant expression of EZH2 in cytotrophoblasts before syncytialization. Upon syncytialization, the marked production of hCG leads to a reduction in the phosphorylation of pRB via activation of the cAMP-PKA pathway, which decreases the level of free E2F1 and diminishes E2F1 transcriptional activity. As a result, EZH2 and H3K27me3 abundance are reduced, rendering H3K27 available for acetylation, which transforms the chromatin from a compact structure into a loose form, allowing access of the HSD11B2 promoter to the transcription factors, thus driving the expression of 11␤-HSD2 (Fig. 6). We have demonstrated previously that P300 acetylates H3K27 and that Sp1 is a transcription factor that regulates 11␤-HSD2 expression in syncytiotrophoblasts (5,6).
Our finding that the pRB-E2F1 pathway maintains EZH2 expression in cytotrophoblasts is similar to what is found in tumor cells, which adopt this signaling pathway to maintain proliferation (19,20). Indeed, cytotrophoblasts resemble tumor cells in many aspects, and they employ similar molecular pathways in the regulation of proliferation, migration, and invasion (24). In tumor cells, exaggerated EZH2 expression is under the tight regulation of pRB-E2F1 to promote proliferation (16,19). In human trophoblasts, when cytotrophoblasts undergo syncytialization, the cell cycle is arrested, resulting in the loss of such

EZH2 represses placental 11␤-HSD2 expression
tumor-like properties (24,25), which may explain the compromised pRB-E2F1 signaling pathway and the consequent down-regulation of EZH2 expression during syncytialization. Although we are unclear about the exact genes pertinent to trophoblast proliferation that are regulated by EZH2, this study does suggest that HSD11B2 is one of the genes that is repressed by EZH2 before syncytialization. Glucocorticoids have been reported to regulate cell proliferation in a cell context-dependent manner with anti-or pro-proliferative actions (26,27). Whether lifting EZH2-mediated repression of 11␤-HSD2 expression is related to the loss of trophoblast proliferative after syncytialization remains unclear, but one role is certainly related to the establishment of a placental glucocorticoid barrier.
In this study, we demonstrate that the cAMP-PKA pathway activated by hCG is responsible for lifting the pRB-E2F1-EZH2 pathway-mediated repression of 11␤-HSD2 expression upon syncytialization. Of interest, in tumor cells, a cascade of reactions caused by cAMP signals has also been shown to be engaged in the inhibition of the expression of cyclin-dependent kinases that directly phosphorylate pRB (22,23,28,29). A reduction in cyclin-dependent kinase expression decreases pRB phosphorylation, thereby restraining E2F1 from its transcriptional activity, resulting in the down-regulation of EZH2. It appears that cAMP-PKA pathway-mediated EZH2 repression may be common in both tumor and trophoblast cells, although the upstream signal to the cAMP-PKA pathway may vary. Here we found that hCG is the upstream signal to this pathway in placental trophoblasts.
Among the placental hormones that utilize the cAMP-PKA pathway in their signaling transduction, hCG is not only the first hormone secreted by trophoblasts during pregnancy but also the most abundant hormone secreted by trophoblasts upon syncytialization (1,21). Here we found that neutralizing endogenous hCG during syncytialization increased the abundance of phosphorylated pRB and EZH2. These effects are consistent with our previous findings that hCG neutralization reduces 11␤-HSD2 abundance during syncytialization (30). All of these properties and effects of hCG can well explain the reciprocal dramatic down-regulation of EZH2 and up-regulation of 11␤-HSD2 upon syncytialization.
Despite the fact that the CpG islands in placental HSD11B2 are indeed subject to the methylation modifications under adverse conditions, such as prenatal maternal anxiety (31,32) and intrauterine growth restriction (4,33), the findings in this study as well as in previous studies (5, 6) have provided convincing evidence for the participation of histone marks in the regulation of 11␤-HSD2 expression during syncytialization. In addition to the decreases in H3K27 trimethylation, these histone marks also include increases in H3K9 and H3K27 acetylation and decreases in H3K9 dimethylation (6). Our findings are consistent with a previous study that also revealed a stronger signal for H3K27me3 in human cytotrophoblasts compared with syncytiotrophoblasts (34). Nonetheless, knockout of EZH2 or its cofactors has indeed been shown to impair the development of mouse trophoblasts (35). Taking all of these findings into consideration, we believe that histone modifications rather than DNA methylation play a dominant role in the regulation of Figure 6. A schematic illustrating the mechanism that underpins the lifting of EZH2-mediated repression on 11␤-HSD2 expression during syncytialization. Before syncytialization, the abundant free E2F1 as a result of pRB phosphorylation leads to the abundant expression of EZH2. By interacting with EED and SUZ12, EZH2 trimethylates H3K27 at the HSD11B2 promoter, resulting in the repression of 11␤-HSD2 expression. Upon syncytialization, the dramatic production of hCG activates the cAMP-PKA pathway, which leads to inactivation of the pRB-E2F1 pathway by decreasing the phosphorylation of pRB. A subsequent reduction in EZH2 expression results in decreased H3K27 methylation, rendering its acetylation at the HSD11B2 promoter, which transforms the chromatin from a compact structure into a loose form, allowing access of the HSD11B2 promoter to the transcription factors, thus driving the expression of 11␤-HSD2.

EZH2 represses placental 11␤-HSD2 expression
11␤-HSD2 expression during syncytialization. This notion is in line with a report that the global expression changes of genes pertinent to the differentiation of mouse trophoblast stem cells are largely attributed to histone methylation rather than DNA methylation (36).
In conclusion, this study has revealed that EZH2-mediated H3K27 methylation represses 11␤-HSD2 expression before syncytialization and that activation of the hCG-cAMP-PKA pathway upon syncytialization lifts this repression through inactivation of the pRB-E2F1-EZH2 pathway.

Collection of human placental villous tissues
Human placental villous tissues were obtained from the first trimester of gestation (6 -8 weeks, n ϭ 5) after painless induced abortions and from term (38 -40 weeks, n ϭ 5) pregnancies after elective cesarean section with written informed consent under a protocol approved by the Ethics Committee of Ren Ji Hospital, School of Medicine, Shanghai Jiao Tong University. In the case of early gestation, women with histories of spontaneous abortion and those who used contraceptives were excluded from this study. In the case of term pregnancy, women with gestational complications such as preeclampsia, fetal growth restriction, and gestational diabetes were excluded from the study. The villous tissues were fixed with formalin for immunofluorescent staining or minced for cytotrophoblast isolation.

Immunochemical staining
Immunofluorescent staining was performed on paraffin-embedded sections as described previously (37). Briefly, after deparaffination and rehydration, the sections were boiled in 10 mM sodium citrate buffer for antigen retrieval and permeabilized with PBS containing 0.4% Triton X-100. After blocking and incubation with the primary antibodies (supplemental Table S1), the sections were incubated with Alexa Fluor 488or Alexa Fluor 594-labeled secondary antibodies (Proteintech, Rosemont, IL). Nuclei were stained with DAPI (1 g/ml). The signals were examined under a fluorescence microscope (Zeiss).
Immunocytochemical staining was conducted on cultured trophoblasts with an avidin-biotin-peroxidase method using a Vectastain kit (Vector Laboratories, Burlingame, CA) as described previously (5). Briefly, the cells were fixed with 4% paraformaldehyde and then permeabilized. The endogenous peroxidase activity was quenched by incubation with 3% H 2 O 2 . After blocking and incubation with the primary antibodies (supplemental Table S1), the cells were incubated with a biotinylated secondary antibody and then with a peroxidase/ streptavidin solution. The peroxidase activity was developed by incubation with 3-amino-9-ethylcarbazole (Vector Laboratories) to produce a red-colored product. Before microscopic examination, the cells were counterstained with hematoxylin and mounted.

Isolation of human placental cytotrophoblast cells
Cytotrophoblast cells were isolated from human term placenta using a modified Kliman method as reported previously (38). Briefly, the minced tissue was digested with 0.125% trypsin (Sigma) and 0.03% DNase I (Sigma) in DMEM (Gibco). After purification using a 5-65% Percoll (GE Healthcare) gradient, cytotrophoblast cells were collected and cultured in DMEM containing 10% FBS (Biological Industries) at 37°C in 5% CO 2 and 95% air.

Extraction of RNA and quantitative real-time PCR
Total RNA was extracted using an RNA extraction kit (Omega Bio-Tek, Norcross, GA) as described previously (39). After evaluating the RNA concentration, mRNA was reversetranscribed to cDNA using the PrimeScript RT kit (TaKaRa, Dalian, China), and cDNA was used for quantitative real-time PCR using SYBR Premix (TaKaRa). The primer sequences are listed in supplemental Table S2. GAPDH was routinely used for normalization. The relative mRNA abundance was quantified using the 2 Ϫ⌬⌬Ct method.

Extraction of protein and analysis with Western blotting
Whole-cell lysate protein was prepared using ice-cold radioimmune precipitation assay buffer (Active Motif, Carlsbad, CA) containing a complete protease inhibitor mixture (Roche) and phosphatase inhibitor (Active Motif). Western blotting was conducted, and the data were analyzed as described previously (37,39). Briefly, after determination of the protein concentration with a Bradford assay, about 20 g of protein was electrophoresed in SDS-polyacrylamide gel and transferred to a nitrocellulose membrane (Merck Millipore). After blocking and incubation with primary antibodies ( supplemental Table S1), the membranes were incubated with horseradish peroxidase-EZH2 represses placental 11␤-HSD2 expression conjugated secondary antibodies (Proteintech). The band with peroxidase activity was detect by a chemiluminescence detection system (Merck Millipore) and visualized using a G-Box chemiluminescence image capture system (Syngene, Cambridge, UK). Densities were analyzed using Gel-Pro Application. The abundance of methylated H3K27 and phosphorylated pRB was expressed as the ratio of their band densities over total H3 and pRB, respectively, whereas EZH2, 11␤-HSD2, and E2F1 abundance was expressed as the ratio of their band densities over GAPDH.

Chromatin immunoprecipitation assay
The chromatin immunoprecipitation assay was carried out as described previously (5,39). Briefly, after cross-linking the proteins at chromatin DNA using 1% formaldehyde, the cells were lysed with 1% sodium dodecyl sulfate lysis buffer and sonicated to shear chromatin DNA. Sheared DNA was immunoprecipitated with antibodies (supplemental Table S1) and protein A-agarose magnetic beads (Merck Millipore). After washing the beads, reverse cross-linking, and removing the contamination of RNA and protein, the immunoprecipitated DNA was extracted using a DNA purification kit (Cwbiotech, Beijing, China) for further quantitative real-time PCR analysis. Similar amounts of sheared DNA without precipitation served as inputs. The primers for quantitative real-time PCR ( supplemental Table S2) were designed to amplify the region between Ϫ843 to Ϫ978 bp of the HSD11B2 promoter, which spans the putative EZH2 binding domain (Fig. 3A), and the region between Ϫ648 and Ϫ754 bp of the EZH2 promoter, which bears the putative E2F1 binding sites (Fig. 4C). The primers amplifying the region between Ϫ1240 to Ϫ1436 bp in the GAPDH promoter were used as a negative control. The data retrieved from quantitative real-time PCR were analyzed using the following equation to indicate the abundance of the proteins enriched at the respective promoters: percent input ϭ 100% ϫ 2 (C[T] input sampleϪC[T] IP sample) .

Statistical analysis
All data are expressed as mean Ϯ S.E. The number for each experiment indicates repeated experiments using trophoblast cells from different placentas. After examination of normal distribution, paired Student's t test or one-way analysis of variance followed by Newman-Keuls multiple comparisons test was used where appropriate to assess the differences. Significance was set at p Ͻ 0.05.
Author contributions-R. Z. and K. S. designed the study and wrote the manuscript. R. Z., X. L., and W. W. performed the experiments. W. L. and H. Y. collected the clinical samples and information.