Interferon tau gene (IFNT) is expressed only by mononuclear trophectoderm cells in ruminant ungulates. To our knowledge, its epigenetic regulation and interaction with trophectoderm lineage-specific caudal-related homeobox 2 transcription factor (CDX2) have not been characterized. Herein, we studied differences in chromatin structures and transcription of endogenous bovine IFNT in bovine trophoblast BT-1 and CT-1 cells and in nontrophoblast MDBK cells. Transcripts from endogenous IFNT and CDX2 genes were found in BT-1 and CT-1 cells but not in MDBK cells. Chromatin immunoprecipitation study revealed that CDX2 binding sites exist in proximal upstream regions of IFNT (IFN-tau-c1). Endogenous IFNT transcription in BT-1 cells was increased with CDX2 overexpression but was reduced with short interfering RNA specific for the CDX2 transcript. In chromatin immunoprecipitation studies, histone H3K18 acetylation of IFNT was higher in CT-1 cells than in MDBK cells, while histone H3K9 methylation was lower in CT-1 cells than in nontrophoblast cells. In MDBK cells (but not in CT-1 cells), histone deacetylases were bound to IFNT, which was reversed with trichostatin A treatment; treatment with trichostatin A and CDX2 then increased IFNT mRNA levels that resulted from abundant CDX2 mRNA expression. These data provide evidence that significant increase in endogenous IFNT transcription in MDBK cells (which do not normally express IFNT) can be induced through CDX2 overexpression and high H3K18 acetylation, but lowering of H3K9 methylation could also be required for the degree of IFNT transcription seen in trophoblast cells.
Interferon tau (IFNT), secreted by mononuclear trophectoderm cells of peri-implantation conceptuses into the uterine lumen, is the major cytokine implicated in the process of maternal recognition of pregnancy in ruminant ungulates [1, 2]. IFNT decreases endometrial oxytocin and estrogen receptors, which attenuates episodic prostaglandin F2α secretion, resulting in prevention of luteolysis [3, 4]. IFNT secretion exhibits temporal and spatial limits in that its production is restricted to trophectoderm cells during the peri-implantation period [5, 6]. In ewes, IFNT production begins on Day 8 of pregnancy (estrus is Day 0), and the quantity of its production seems to parallel the degree of trophoblast elongation [7, 8]. Its secretion peaks on Day 16, just before attachment of the conceptus to the uterine epithelium, and declines subsequently [8, 9]. By Day 22, when placenta formation is initiated, IFNT is no longer detected . Bovine and caprine conceptuses exhibit a similar cell-specific and temporal pattern of IFNT expression [10–12].
Although IFNT mRNA and genes were discovered more than two decades ago, the molecular mechanisms by which IFNT expression is regulated in a temporal and spatial manner are not well understood. By analyzing the 5′ upstream regions of IFNT, transcription factors such as activating protein 1 (AP1, official symbol JUN) and DNA-binding ETS domain ETS2 were found to be responsible for increased IFNT transcription [13–16]. In addition to trophoblasts and placenta, however, transcription factors such as JUN and ETS2 are expressed in a wide variety of tissues and cell types [17, 18]. Moreover, the transcription coactivator cAMP-response element binding protein (CREB)-binding protein (CREBBP), which has histone acetyltransferase (HAT) function, was shown to regulate ovine IFNT expression [19, 20]. It is still unclear why IFNT is expressed only in trophoblast and how its production is initiated and terminated within a short period of development. Therefore, it is thought that the unique IFNT expression could be controlled not only by transcription factors but also by chromatin modification.
Epigenetic alterations such as variation in DNA methylation and covalent histone modification regulate gene expression by altering chromatin conformation. DNA methylation usually occurs at cytosine residues within CG dinucleotides or CNG trinucleotides and generally opposes transcription [21, 22]. Changes in the degree of DNA methylation in the upstream sequences of ovine IFNT could be one of the major mechanisms leading to downregulation of its expression and possibly its silencing in nonconceptus tissues . In addition to DNA methylation, histone posttranslational modifications such as acetylation and methylation are correlated with positive or negative transcriptional status of various genes . Acetylation of lysine residues is controlled by specific HATs and histone deacetylases (HDACs). Four lysine residues (K9, K14, K18, and K56) of histone H3 (H3K9, H3K14, H3K18, and H3K56, respectively) and four lysine residues (K5, K8, K13, and K16) of histone H4 (H4K5, H4K8, H4K13, and H4K16, respectively) are acetylated by specific HATs . It has been demonstrated that histone acetylation/deacetylation alters chromosome structure, which in turn affects accession of transcription factors to DNA sequences. Histone posttranslational modifications are also modified through histone methylation. Five lysine residues (K4, K9, K27, K36, and K79) of histone H3 (H3K4, H3K9, H3K27, H3K36, and H3K79, respectively) and K20 of histone H4 (H4K20) are methylated by the relevant histone methyltransferases, are demethylated by histone demethylases, and are associated with repression and activation of transcription . It is generally believed that methylated H3K4, H3K36, and H3K79 are associated with active transcription, while methylated H3K9, H3K27, and H4K20 are associated with a transcriptionally inactive state . Moreover, it was demonstrated that H3K9 methylation is mechanistically linked to DNA methylation . H3K9 methylation is considered crucial for heterochromatin assembly, resulting in specific binding of heterochromatin protein 1 (HP1, official symbol CBX5) to methylated H3K9 [27, 28], whereas H3K4 methylation occurs preferentially within transcriptionally competent chromatin . However, modification of the chromatin structure of IFNT, 5′ upstream regions, and open reading frame (ORF) that are associated with the degree of IFNT transcription has not been studied in much detail.
It was shown that caudal-related homeobox 2 transcription factor (CDX2) is expressed in ovine and bovine trophoblasts during the conceptus elongation period [30, 31]. In fact, overexpression of CDX2 along with JUN and ETS2 in human choriocarcinoma JEG3 cells was effective in increasing the degree of transcription of an ovine IFNT reporter construct . These results suggest that CDX2 could be a key element that determines trophoblast cell-specific activation of IFNT expression. In this study, we examined the importance of CDX2 and chromatin structures in endogenous IFNT transcription in bovine trophoblast-derived BT-1  and CT-1  cells and in nontrophoblast bovine kidney MDBK cells. Our findings should lead to a better understanding of how trophectoderm expression of a conceptus gene is regulated during the establishment of pregnancy.
MATERIALS AND METHODS
Cell Cultures and Their Treatment
Bovine trophoblast BT-1  and CT-1  cells (kindly provided by Dr. A. Ealy, University of Florida, Gainesville, FL) were established from in vitro matured and fertilized blastocysts and were cultured without feeder cells. BT-1 cells were cultured on plastic plates coated with type I collagen (Nitta Gelatin, Osaka, Japan) in Dulbecco modified Eagle (DME)/F12 medium (Invitrogen, Carlsbad, CA) supplemented with 10% v/v fetal bovine serum (FBS; JRH Biosciences, Lenexa, KS) and antibiotics/antimycotic solution (Invitrogen) at 37°C in air with 5% CO2. Cells were mechanically dissociated with a pipette, and the clumps were plated on plastic plates in culture medium as already described. CT-1 cells were maintained at 37°C in air with 5% CO2 in DME medium containing 10% v/v FBS supplemented with 4.5 g/L of d-glucose, nonessential amino acids, 2 mM glutamine, 2 mM sodium pyruvate, 55 μM β-mercaptoethanol, and antibiotic/antimycotic solution (all from Invitrogen). CT-1 cells were scraped from the plates and passed through a small-bore needle to generate small clumps of cells before reseeding. MDBK cells, a bovine kidney-derived epithelial cell line (American Type Culture Collection, Rockville, MD), were grown in DME medium supplemented with 10% v/v FBS and antibiotics at 37°C in 5% CO2 in air. Bovine ovarian cumulus granulosa (oCG) cells were obtained from ovarian follicles that had been collected at a local abattoir, and ear-derived fibroblast (EF) cells were obtained from biopsied ear skin of 4-mo-old Japanese black bulls. Both cells were cultured in DME medium containing 5% v/v FBS (JRH Biosciences) and antibiotics (Invitrogen) at 37°C in air with 5% CO2. Each cell type was serum starved for 24 h before treatment. In addition to bovine cells, human choriocarcinoma JEG3 cells (American Type Culture Collection) were cultured in DME medium supplemented with 10% FBS (JRH Biosciences) and antibiotics (Invitrogen) as described previously .
PCR Analysis and Real-Time PCR Analysis
Total RNA was isolated from cultured cells of each type using Isogen (Nippon Gene, Tokyo, Japan) according to the protocol provided by the manufacturer. Isolated RNA (250 ng) was converted to cDNA using SuperScript II (Invitrogen) and oligo(dT) primers in a 20-μl reaction volume, and the resulting cDNA (RT template) was stored at 4°C until use. The cDNA reaction mixture was diluted 1:10 using DNase and RNase-free molecular biology grade water, and 3 μl was taken for each amplification reaction. The PCR was performed using 0.5 U of ExTaq polymerase (Takara Biomedicals, Tokyo, Japan), 1× ExTaq buffer, 0.2 μM of the oligonucleotide primers listed in Table 1, and 0.2 mM deoxyribonucleotide triphosphate in a final volume of 20 μl. The thermal profile for PCR was at 95°C for 10 min, followed by 28 cycles of 95°C for 10 sec, 60°C for 30 sec, and 72°C for 30 sec. The PCR products were separated on a 1.5% agarose gel containing ethidium bromide and were visualized under UV light.
Reverse-transcribed cDNA (3 μl) was subjected to real-time PCR amplification using 0.5 U of ExTaq HS polymerase (Takara Biomedicals), 1× ExTaq HS buffer, 0.2 μM of the oligonucleotide primers listed in Table 1, 0.2 mM deoxyribonucleotide triphosphate, SYBR green (SYBR Green I Nucleic Acid Gel stain; Takara Biomedicals) as a fluorescence intercalater, and Rox reference dye (Invitrogen) in a final volume of 20 μl. The PCR amplification was performed on an 7900HT real-time PCR system (Applied Biosystems, Foster City, CA). The thermal profile for real-time PCR was 95°C for 10 min, followed by 40 cycles of 95°C for 10 sec, 60°C for 20 sec, and 72°C for 40 sec. Average cycle threshold (Ct) values for IFNT, CDX2, ETS2, JUN, and CREBBP were calculated and normalized to Ct values for ACTB . Each run was completed with a melting curve analysis to confirm the specificity of amplification and the absence of primer dimers.
Plasmid Construction, Short Interfering RNAs, and Transfection
A mouse Cdx2 expression plasmid in pRC-cytomegalovirus was kindly provided by Dr. EunRan Suh, University of Pennsylvania School of Medicine, Philadelphia, PA. Cdx2 cDNA was subcloned into the pSG5 plasmid (Invitrogen), resulting in construction of the pSG5-Cdx2 plasmid. Constructs containing each of 14 CDX2 sites of mutated bovine IFNT were prepared by an inverse PCR procedure as previously described  with appropriate primers that included mutation; the CDX2 recognition site was mutated from 5′-TTTACTG-3′ to 5′-TTTTGTG-3′. Nucleotide structures of these constructs were confirmed by DNA sequencing. Transient transfection and luciferase activity measurements were performed in JEG3 cells as previously described . CDX2 short interfering RNAs (siRNAs) (whose nucleotide structures were designed using the siDirect program; RNAi Co., Ltd., Tokyo, Japan), were prepared commercially (Sigma-Aldrich, St. Louis, MO). The nucleotide sequence of bovine CDX2 (XM_871005) was used to design three different siRNAs for CDX2 coding regions, while an unrelated sequence of EGFP (EU056363) was used as a negative control (Table 1).
To evaluate the effects of mouse Cdx2 overexpression or its knockdown on the abundance of bovine IFNT mRNA, the pSG5-Cdx2 or CDX2 siRNA was transfected into BT-1 cells using Lipofectamine 2000 reagents (Invitrogen) according to the procedure recommended by the manufacturer. The procedure for the transfection was essentially as previously described . Concentrations of each siRNA had been determined before the experiment. The pSG5-Cdx2 siRNA (2 μg [seeded on a six-well dish]) and CDX2 siRNA (50 nM [seeded on a 24-well dish) were transfected into BT-1 cells for 48 h, from which tRNA was extracted and reverse transcribed to cDNA. The yielded cDNAs were subjected to real-time PCR analysis with the primers listed in Table 2 for determination of bovine IFNT and CDX2 mRNA levels. Average Ct values for bovine IFNT and CDX2 were calculated and normalized to Ct values for ACTB.
Chromatin Immunoprecipitation Assay
Chromatin immunoprecipitation (ChIP) assays were performed using the ChIP assay kit (Upstate Biotechnology, Lake Placid, NY) following the protocol provided by the manufacturer. Briefly, bovine cells (∼5 × 106 cells/10-cm dish) were treated with or without 200 nM trichostatin A (TSA; Wako, Tokyo, Japan). Twenty-four hours after initiation of TSA treatment, the cells were transfected with pSG5 only or with pSG5-Cdx2 (4 μg) and cultured for 48 h, resulting in 72 h of total TSA treatment. Potential protein-DNA complexes in these cells were cross-linked with 1% formaldehyde for 20 min, washed with chilled PBS, resuspended in 200 μl of SDS lysis buffer, and sonicated six times for 10 sec each at 60% maximum setting of the sonicator (Handy Sonic UR-20P; Tomy Seiko Co., Ltd., Tokyo, Japan). The supernatant of sonicated cells was diluted 10-fold to a total volume of 2 ml, and 1% (20 μl) of diluted lysates was used for total genomic DNA as input DNA control. To reduce nonspecific ChIP signals, the remaining cell lysates were subjected to an immunoclearing procedure to which 75 μl of protein A agarose/salmon sperm DNA (50% slurry) (Upstate Biotechnology) was added, and sample tubes were agitated at 4°C for 30 min. Immunoprecipitation was then performed overnight at 4°C with rabbit polyclonal antihistone H3 acetyl K18 (H3K18ac) antibody (2 μg; ab1191), rabbit polyclonal antihistone H3 monomethyl K9 (H3K9me) antibody (5 μg; ab9045), rabbit polyclonal antihistone H3 dimethyl K9 (H3K9me2) antibody (5 μg; ab1220), rabbit polyclonal antihistone H3 trimethyl K4 (H3K4me3) antibody (4 μg; ab1012), mouse monoclonal antiHDAC1 antibody (2.5 μg; ab31263), or mouse monoclonal antiHDAC2 antibody (2.5 μg; ab51832 [all from Ab Chem, Inc., Ville St-Laurent, QC, Canada]) or with mouse monoclonal antihuman CDX2 antibody (4 μg; BioGenex, San Ramon, CA). Normal rabbit IgG (Santa Cruz Biotechnology, Santa Cruz, CA) or normal mouse IgG (Cell Signaling Technology, Inc., Danvers, MA) instead of antibody was used as a negative control. Using the ChIP assay protocol provided by the manufacturer, concentrations of each antibody were examined and validated for these assays. Precipitates were washed sequentially for 5 min each in low-salt, high-salt, and lithium chloride immune complex wash buffers and were finally washed twice with Tris/edetic acid buffer. Histone complexes were then eluted from the antibody by freshly prepared elution buffer (1% SDS, 0.1 M NaHCO3, and 10 mM dithiothreitol). Protein-DNA cross-links (including the input DNA control samples) were reversed by 5 M NaCl at 65°C for 4 h, followed by proteinase K (Invitrogen) digestion at 45°C for 2 h. The DNA fragments were extracted with phenol and chloroform, precipitated with ethanol and 0.3 M sodium acetate (pH 8.2), and resuspended in 30 μl of Tris/edetic acid buffer, from which 3 μl was used for PCR (35 cycles) with the primers listed in Table 2. The PCR products were separated on a 1.5% agarose gel containing ethidium bromide and were visualized under UV light. The band intensity was determined using NIH Image analyzer (National Institutes of Health, Bethesda, MD), and each value was normalized against the relevant input bands.
All quantitative data were subjected to least squares ANOVA using SAS software (SAS Institute, Cary, NC). The model used in the least squares ANOVA included treatment and replicate as sources of variation. The SEMs in the figures were derived from this analysis. When a significant effect of treatment was detected (P < 0.05), the data were analyzed using Dunnett test for multiple comparisons. In these analyses, treatment was considered an independent source of variation, and replicate was considered a dependent source.
Abundance of Bovine IFNT, CDX2, and Other Transcription Factor mRNAs in Various Bovine Cell Lines
Bovine IFNT, CDX2, ETS2, JUN, and CREBBP mRNA expressions in BT-1, CT-1, MDBK, EF, and oCG bovine cell lines are shown in Figure 1. The results indicated that expression of IFNT and CDX2 mRNA was found in trophoblast BT-1 and CT-1 cells but not in nontrophoblast MDBK, EF, or oCG cells.
Binding of CDX2 to the Upstream Region of Bovine IFNT
ChIP assay was performed to determine the CDX2 binding site on bovine IFNT (IFN-tau-c1) in BT-1 cells (Fig. 2A). In the upstream region of bovine IFNT, CDX2 binding sites in region 2 (−240 to −47 bp) and region 3 (−755 to −560 bp) were identified in BT-1 cells, whereas no CDX2 binding site was found in the ORF region (314–520 bp). In addition, CDX2 overexpression seemed to increase the ChIP signal, particularly the CDX2 binding site in region 3. These results indicated that the CDX2 binding sites on bovine IFNT seemed to be functional in trophoblast BT-1 cells.
Region 2 of bovine IFNT represents CDX2-rich regions, having at least 14 potential CDX2 binding sites (Supplemental Fig. S1 available at www.biolreprod.org). One point mutation was introduced to each of these sites. In transient transfection experiments in CT-1 cells, all constructs with mutated CDX2 sites had reduced luciferase activity (one half to one fourth of wild-type constructs [results not shown]). Together with CDX2 binding by ChIP assay, these data indicated that each of 14 CDX2 sites was functional.
Endogenous Bovine IFNT mRNA Levels Following Changes in CDX2 Expression in BT-1 Cells
It was previously shown that a −654-bp ovine IFNT construct was transactivated more than 30-fold in JEG3 cells by cotransfection of Jun, Ets2, and Cdx2 . In lieu of heterologous transfection studies [13–16, 18], levels of endogenous bovine IFNT transcription were studied in BT-1 cells that had been transfected with the pSG5-Cdx2 expression construct. The results revealed that overexpression of Cdx2 in BT-1 cells increased bovine IFNT mRNA levels more than 5-fold (P < 0.01) compared with control cells, into which the pSG5 plasmid without Cdx2 (mock) had been transfected (Fig. 2B [left]).
To determine whether downregulation of CDX2 reduces endogenous bovine IFNT transcription, three different siRNAs for bovine CDX2 mRNA were designed and synthesized (Table 1). BT-1 cells were then transiently transfected with one of these CDX2 siRNAs or with an EGFP siRNA as a negative control. Although two of these siRNAs (siRNA 1 and siRNA 2) reduced CDX2 mRNA (P < 0.05 and P < 0.01, respectively), CDX2 siRNA 2 more effectively reduced the amount of CDX2 mRNA than CDX2 siRNA 1, which in turn resulted in the reduction of endogenous bovine IFNT mRNA (Fig. 2B [right]). The effect of the CDX2 siRNAs was limited to CDX2 mRNA and bovine IFNT mRNA expression, as ETS2, JUN, CREBBP, and POU5F1 mRNA levels were unaffected by this treatment (data not shown). Figure 2B illustrates that the transcription of bovine IFNT was regulated by the degree of CDX2 transcription.
Histone Methylation and Acetylation of Bovine IFNT in CT-1 and MDBK Cells
During experiments, we noticed that BT-1 cells with more than 300 passages were losing or had lost endogenous IFNT transcription and/or production. Instead of BT-1 cells, CT-1 cells were therefore used for the remaining experiments, as CT-1 cells at that point gave similar (if not the same) results compared with those initially found in BT-1 cells.
To elucidate whether the status of histone H3 methylation and acetylation determines the degree of bovine IFNT transcription, the ChIP assay was executed in bovine trophoblast and nontrophoblast cells using various histone antibodies. DNA-protein complexes were treated with antibodies against H3K18ac, H3K9me, H3K9me2, H3K4me3, or rabbit IgG (as a negative control) (Fig. 3). The degree of H3K18 acetylation in region 2 of bovine IFNT was higher in trophoblast CT-1 cells than in nontrophoblast MDBK cells (control). The degree of H3K9 dimethylation in region 1 (314–520 bp) and region 2 (−240 to −47 bp) was higher in MDBK cells (control) than in CT-1 cells. In addition, H3K4 methylation seemed to be associated with bovine IFNT expression in CT-1 cells but not in MDBK cells (control). The status of histone H3 methylation and acetylation in oCG and EF cells was the same as that in MDBK, and the status of these in BT-1 cells was also the same as that in CT-1 cells (data not shown). To determine the effect of TSA treatment on bovine IFNT transcription, changes in the status of histone acetylation and methylation on bovine IFNT in MDBK cells were studied following treatment with 200 nM TSA. In TSA treatment, the degree of H3K18 acetylation in region 1 and region 2 of bovine IFNT and of H3K4 trimethylation in region 1 seemed to increase in MDBK cells (TSA treatment), while H3K9 dimethylation was unchanged. These data suggest that H3K18 acetylation and H3K4 methylation of bovine IFNT are important in its expression, while H3K9 dimethylation in the upstream region where the CDX2 binding site is located, as well as in the ORF region, may be associated with suppression of bovine IFNT transcription in nontrophoblast cells.
HDAC1 and HDAC2 Binding to Bovine IFNT in CT-1 and MDBK Cells
Because methylated H3K9 recruits repressing effector proteins, including HDAC complexes [24, 27, 28], the presence and/or binding of HDACs on bovine IFNT was studied using antibodies against HDAC1 and HDAC2 (Fig. 4). In CT-1 cells, a minute ChIP signal of HDAC1 was detected on the ORF (region 1), while HDAC2 was not expressed in either region. In MDBK cells, ChIP signals of HDAC1 and HDAC2 were detected in both regions. In addition, TSA treatment effectively removed HDAC1/HDAC2 from both regions in MDBK cells. It seems that the removal of HDAC1/HDAC2 by treatment with TSA to MDBK cells results in increased acetylation of H3K18, as shown in Figure 3. Therefore, in nontrophoblast cell lines in which IFNT is not transcribed, the upstream and ORF regions seem have recruited HDACs as negative modifiers.
Effects of CDX2 and TSA Treatment on CDX2 Binding to Bovine IFNT and the Degree of Its Transcription
Because the difference in bovine IFNT transcription between trophoblast cells and nontrophoblast cells results from a lack of CDX2 expression (Fig. 1) and its binding to region 2 (Fig. 2A), we assessed binding of CDX2 to the upstream region of bovine IFNT in MDBK cells when treated with the pSG5-Cdx2 construct (4 μg) and with 200 nM TSA (Fig. 5A). Binding of CDX2 to upstream region 2 in bovine IFNT was found following pSG5-Cdx2 transfection and TSA treatment.
Because no CDX2 expression was found in nontrophoblast cell lines (Fig. 1) and because TSA treatment seemed to increase the degree of H3K18 acetylation (Fig. 3), endogenous bovine IFNT mRNA levels were determined following TSA treatment or TSA treatment plus Cdx2 overexpression in MDBK cells (Fig. 5B). Although some increases in endogenous bovine IFNT transcripts in MDBK cells were found throughout Cdx2 overexpression and TSA treatment, the greatest increase in endogenous bovine IFNT transcripts and Cdx2 mRNA from the expression construct was found when MDBK cells were treated with Cdx2 plus 200 nM TSA. The increase was minimal in Cdx2 mRNA in MDBK cells that had been treated only with TSA. These results indicate that bovine IFNT transcription in nontrophoblast cell lines, in which IFNT expression is normally kept silent, could be induced by TSA treatment, but sufficient induction requires both Cdx2 overexpression and TSA treatment.
IFNT is thought to be hypomethylated during the zygotic stage, suggesting that its transcription could be initiated at any time. However, IFNT expression can first be detected at the blastocyst stage. Ezashi et al.  reported that ETS2-induced transactivation of bovine IFNT (IFNT1) promoter is repressed by POU5F1 through direct interaction of ETS2 with POU5F1. Disappearance of POU5F1 in trophectoderm by Day 10 of pregnancy allows ETS2 to assume a positive regulatory role in bovine IFNT transcription. However, Imakawa et al.  reported that, although ETS2 and JUN activated ovine IFNT transcription, overexpression of Cdx2 plus Ets2 and Jun was most effective in transactivation of an ovine IFNT construct. It was shown in the present study that active CDX2 binding sites are present in the proximal upstream region of bovine IFNT. It seems that CDX2 expression parallels that of IFNT during the peri-implantation period in bovine and ovine trophoblast cells (Sakurai and K. Imakawa, unpublished results). However, it is well known that CDX2 is not unique to trophectoderm cells, as it is expressed in a number of cell types such as intestinal epithelial cells. CDX2 is activated by phosphorylation via the mitogen-activated protein kinase pathway, resulting in modulation of its transactivating activity . The phosphorylated form of CDX2 is localized in proliferating cells of the crypt in the gut, while the nonphosphorylated form is present in more differentiated cells. The presence of phosphorylated CDX2 has been detected in the blastcyst . Although genes regulated by CDX2 in the blastcyst have not been well characterized, it is probable that phosphorylated CDX2 may regulate IFNT expression during the period of conceptus elongation.
In bovine IFNT analyzed in this study, there are CpG sites between −900 and −527 bp but no such sites between −526 and −7 bp in its upstream region, whereas CpG sites are well distributed between −900 and 0 bp in the upstream region of ovine IFNT (Supplemental Fig. S2). Results from our laboratory  revealed that the methylation status of ovine IFNT during active transcription was different from that during other periods when gene transcription was silenced. In fact, the upstream region of ovine IFNT in Day 14 sheep conceptuses, which had the highest levels of ovine IFNT mRNAs, was hypomethylated at CpG dinucleotides located in the upstream region (−1000 to −7 bp) of ovine IFNT (M88773). In contrast, ovine uterine endometrium and peripheral blood mononuclear cells, which do not express ovine IFNT, were hypermethylated at these CpG sites . Five CpG sites found in the upstream region (−526 to −7 bp) of ovine IFNT are initially methylated as ovine IFNT transcription declines . Therefore, it is likely that methylation to the 5′ flanking region of IFNT is one of the molecular mechanisms by which this gene is kept silent in tissues or cell types other than the peri-implantation conceptus.
In addition to DNA methylation, histone posttranslational modifications such as histone acetylation and methylation are involved in activation and/or repression of transcription. It has become apparent that gene transcription is regulated in a dynamic fashion rather than by on-and-off switches . In this study, active transcription in bovine IFNT was found in CT-1 and BT-1 cells in which the degree of H3K18 acetylation was kept at high levels. Broad histone acetylation leads to partial decondensation of chromosomal domains, enabling various transcription factors to bind to DNA sequences. A DNA-bound activator in the promoter region then recruits positive modifiers such as CBP/p300, and DNA-bound RNA polymerase recruits histone methyltransferases at the ORF, resulting in increased H3K4 methylation to loosen chromatin structure and further promote transcription. In fact, H3K4 methylation on bovine IFNT was found in promoter and ORF regions of CT-1 cells, whereas it was found in the ORF region of TSA-treated MDBK cells but not in the TSA-nontreated MDBK cells (Fig. 3). These data suggest that, in addition to H3K18 acetylation, H3K4 methylation could also be associated with activation of bovine IFNT transcription in trophoblast cells.
In the “off” state of gene transcription, the DNA-bound repressor(s) at the repressor site recruits negative modifiers such as HDACs that remove acetyl groups from histones, resulting in condensation of chromosomal domains. Increased histone H3K9 methylation, especially H3K9 dimethylation, also recruits repressing effector proteins such as HP1 and HDAC complexes that stabilize nucleosomes during transcription attenuation or repression . Binding of HDACs in the upstream region of bovine IFNT suggests that histone deacetylation and possibly HDAC complex formation are involved in the silencing of IFNT expression in nontrophoblast cells. In TSA-treated MDBK cells, H3K18 acetylation and H3K4 trimethylation were increased, while H3K9 dimethylation was unchanged in upstream and ORF regions of bovine IFNT. Despite the fact that H3K18 acetylation was increased in MDBK cells that had been treated with TSA, bovine IFNT expression was not induced. Although histone acetylation loosens chromatin structure, this opening may not be tightly correlated with active transcription. Even in the euchromatin state, gene transcription is repressed where histone methyltransferases is recruited through interaction with CBX5 by tumor suppressor retinoblastoma protein RB1 . However, endogenous bovine IFNT transcription was highest when TSA was applied and Cdx2 was overexpressed in MDBK cells, resulting from maximal Cdx2 mRNA expression (Fig. 5B). It was recently demonstrated that HDAC levels in the DRA promoter are reduced to 50% of control levels and remained stable following TSA treatment . These authors indicated that reduced HDAC in the promoter region of the DRA gene following TSA administration results from intranuclear mobility of HDAC1 and HDAC2. These data suggest that sufficient removal of HDACs by TSA may have promoted abundant expression of transfected Cdx2, which facilitates formation of a positive modifier complex and the transcriptional coactivator CREBBP, which also exhibits HAT activity. In our previous study , overexpression of Crebbp increased the degree of ovine IFNT transcription, further supporting the notion that HAT activity is involved in IFNT transcription. These findings suggest that induction of endogenous IFNT transcription in bovine trophoblast cells results from partial decondensation of chromosomal domains by histone acetylation and sufficient CDX2 expression.
In conclusion, because 14 potential CDX2 binding sites were found in region 2 (−240 to −47 bp) of bovine IFNT and because TSA treatment and Cdx2 overexpression were most effective in CDX2 binding in region 2 of bovine IFNT in nontrophoblast cells, we propose that active bovine IFNT transcription in trophoblast BT-1 and CT-1 cell lines results from the absence of HDACs and from sufficient CDX2 binding in region 2. Further experiments are required to elucidate whether this molecular mechanism is sufficient to explain in utero regulation of IFNT in bovine or ovine species.
The authors thank Dr. Lutz Weber, Oak Ridge Institute for Science and Education, Oak Ridge, TN, for his critical reading of the manuscript. The authors also thank Drs. Alan Ealy (University of Florida, Gainesville, FL) and EunRan Suh (University of Pennsylvania School of Medicine, Philadelphia, PA) for generously providing bovine trophoblast CT-1 cells and the CDX2 expression plasmid, respectively.
Oligonucleotide primer sequences for ChIP assay and nucleotide sequences for CDX2 siRNA.
Oligonucleotide primer sequences used for RT-PCR analysis.