CCAAT-enhancer-binding Protein β (C/EBPβ) and Downstream Human Placental Growth Hormone Genes Are Targets for Dysregulation in Pregnancies Complicated by Maternal Obesity*

Background: Effect of maternal obesity on metabolically important human placental growth hormone (CS and GH-V) gene expression has not been studied. Results: CS, GH-V, and C/EBPβ, a factor that binds CS-related enhancer chromatin, are decreased by maternal obesity. Conclusion: C/EBPβ is a target for maternal obesity that negatively affects placental CS/GH-V gene expression. Significance: Genes regulated by C/EBPβ, a placenta development factor, are targeted in pregnancies complicated by obesity. Human chorionic somatomammotropin (CS) and placental growth hormone variant (GH-V) act as metabolic adaptors in response to maternal insulin resistance, which occurs in “normal” pregnancy. Maternal obesity can exacerbate this “resistance,” suggesting that CS, GH-V, or transcription factors that regulate their production might be targets. The human CS genes, hCS-A and hCS-B, flank the GH-V gene. A significant decrease in pre-term placental CS/GH-V RNA levels was observed in transgenic mice containing the CS/GH-V genes in a model of high fat diet (HFD)-induced maternal obesity. Similarly, a decrease in CS/GH-V RNA levels was detected in term placentas from obese (body mass index (BMI) ≥ 35 kg/m2) versus lean (BMI 20–25 kg/m2) women. A specific decrease in transcription factor CCAAT-enhancer-binding protein β (C/EBPβ) RNA levels was also seen with obesity; C/EBPβ is required for mouse placenta development and is expressed, like CS and GH-V, in syncytiotrophoblasts. Binding of C/EBPβ to the CS gene downstream enhancer regions, which by virtue of their position distally flank the GH-V gene, was reduced in placenta chromatin from mice on a HFD and in obese women; a corresponding decrease in RNA polymerase II associated with CS/GH-V promoters was also observed. Detection of decreased endogenous CS/GH-V RNA levels in human placental tumor cells treated with C/EBPβ siRNA is consistent with a direct effect. These data provide evidence for CS/GH-V dysregulation in acute HFD-induced obesity in mouse pregnancy and chronic obesity in human pregnancy and implicate C/EBPβ, a factor associated with CS regulation and placental development.

The prevalence of obesity is increasing in women of reproductive age such that it has been suggested that more than one in five pregnant women are obese (1)(2)(3). Normal human pregnancy is characterized as a metabolic stress involving a series of metabolic changes promoted by insulin resistance. The demand for increased maternal insulin in normal pregnancy due to insulin resistance is related to weight gain and placental hormone production as well as increased food intake and fetal burden (4,5). To compensate, there are increases in insulin secretion per pancreatic ␤-cell and in ␤-cell proliferation such that ␤-cell mass increases 50% in pregnancy (5,6). Chorionic somatomammotropin (CS) 2 and placental growth hormone variant (GH-V) are major stimuli of ␤-cell proliferation in pregnancy (5,7,8). CS/GH-V are also involved in shifting the maternal energy metabolism from carbohydrate to lipid substrates, thereby reducing insulin-mediated utilization of glucose. Thus, they contribute to fetal growth and development by mobilizing the mother's nutritional resources, primarily glucose, and making them available to the fetus (9,10). Maternal obesity may exacerbate insulin resistance associated with pregnancy by affecting CS/GH-V levels. Thus, the combination of obesity and decreased insulin sensitivity increases the long term risk for metabolic syndrome and associated problems. Furthermore, genetic variation associated with the CS/GH-V genes does not cause, but can contribute substantially to, development of metabolic syndrome (11). Thus, if these metabolic effects are related to the dysregulation of CS/GH-V hormones as a result of genetic variation, it is possible that similar CS/GH-V gene dysregulation will be observed as a result of physiological/ pathophysiological perturbation resulting from maternal obesity and its associated health complications.
CS and GH-V are members of the human growth hormone (GH) family and are synthesized and secreted by syncytiotrophoblasts on the maternal side of placenta (12). CS is the product of either of two genes hCS-A and hCS-B, which flank a single GH-V gene (hGH-V); a third CS gene (hCS-L) is a pseudogene (13). The placental CS/GH-V genes are all located at a single locus on chromosome 17 together with the pituitary GH gene, hGH-N (13). Efficient CS/GH-V production is closely related to villus syncytiotrophoblast development and placental mass during pregnancy (12, 14 -17). Activation and expression of the human GH/CS genes has been linked to a set of remote regulatory elements associated with five nuclease hypersensitive sites (I-V). These sites are found in the loci of the CD79b and SCN4A genes that lie upstream and adjacent to the GH/CS locus on chromosome 17. Hypersensitive sites III and V comprise the pituitary GH locus control region (LCR) (18 -21), which permits the site of integration-independent and appropriate pituitary-specific hGH-N expression (20,21). It is unclear, however, whether sequences included in the LCR alone are sufficient for appropriate placental expression of CS/GH-V. There is evidence to suggest cooperation between remote GH LCR sequences and DNA elements more proximal to the CS/GH-V genes that might collectively act to regulate individual promoters. Candidates for highly conserved and more proximal regulatory regions include 5Ј-enhancer/repressor P sequences, located ϳ2 kb upstream of each of the placental GH/CS genes (22)(23)(24) and efficient 3Ј-enhancer regions located 2 kb downstream of the CS genes, which by virtue of their position also flank, albeit distally, hGH-V (25)(26)(27)(28)(29)(30)(31)(32)(33)(34). The placenta-specific enhancer activity was localized to a 126-base pair (bp) fragment (25,26), and these 3Ј-enhancer regions were shown to contain hyperacetylated histone H3 and H4 in primary human placental and choriocarcinoma (BeWo) cells in culture (35). These data imply an open chromatin configuration with higher levels of accessibility to transcription factors. Two nuclease-protected sites were identified within the 126-bp CS 3Ј-enhancer regions with placenta nuclear protein (25,26). One was characterized as a transcription enhancer factor 1(TEF-1) or TEF-5-binding site (25)(26)(27)(28)(29)(30)(31)(32)(33), and subsequently, a consensus binding site for the CCAAT-enhancer-binding protein (C/EBP) family and associated enhancer activity was identified (33) that corresponds to the second nuclease protected site within the 126-bp 3Ј-enhancer regions (25,26). Furthermore, C/EBP␤ was shown to associate with the CS 3Ј-enhancer regions in human term placenta chromatin in situ, suggesting a possible role in vivo (33). C/EBP␤ levels increase in human term placenta and like CS and GH-V are also linked to villous syncytiotrophoblast versus cytotrophoblasts (37). A physiological role for C/EBP␤ in placenta morphogenesis is suggested based on genetic deletion of C/EBP family members in mice (38 -40). Most important in the context of the current study, C/EBP␤ is implicated in adipogenesis/glucose metabolism in the context of obesity (41)(42)(43). Together these observations suggest C/EBP␤ as a strong candidate to be targeted by obesity and in turn to modulate CS gene expression during pregnancy.
Although useful, both human choriocarcinoma cell lines and primary term placenta cell cultures are limited in their ability to allow testing of CS/GH-V gene regulation during pregnancy (44 -46). By contrast, murine systems provide a model to study events during pregnancy but are limited by structural differences between the CS genes in primates and non-primates (47) as well as the absence of the GH-V gene in non-primates (48). Thus, in vivo CS/GH-V gene regulation under physiological or pathophysiological conditions, such as maternal obesity, has not been well studied. We generated "humanized" hGH/CS transgenic (TG) CD-1 mice containing all five GH/CS genes under the control of the GH/CS LCR, which includes GH LCR, P, and 3Ј-enhancer related sequences (18). The GH/CS LCR directs placenta-specific expression of CS/GH-V in the labyrinthine layer, which represents the region of maternal-fetal exchange in the mouse placenta and corresponds in function to the villous syncytium in the human placenta (18,21).
Here we have used a combination of hGH/CS-TG mouse studies, human placenta samples, and choriocarcinoma cells to provide evidence for the first time that the CS/GH-V genes can be regulated during pregnancy. Specifically, C/EBP␤ and CCAAT elements are components of composite CS enhancer regions that are targeted in pregnancies complicated by maternal obesity.

EXPERIMENTAL PROCEDURES
Mice and Diet-All procedures involving animals, their tissues, and cells conform to the Guide for the Care and Use of Laboratory Animals published by the United States National Institutes of Health (NIH Publication, 8th Ed., 2011) and was approved by the animal Protocol Management and Review Committee at the University of Manitoba. Animals were individually housed with ad libitum access to food and water in an environmentally controlled room maintained on a 12-h light/ dark cycle. Four-week-old female hGH/CS-TG mice (49,50) and non-TG CD-1 mice were assigned at random to either a low fat diet (LFD) (20% protein, 10% total fat; Research Diets, New Brunswick, NJ) or high fat diet (HFD) (20% protein, 60% total fat; Research Diets) group. After 4 weeks of dietary intervention, mice were allowed to breed overnight in a 1:2 (male:female) ratio. Mating was confirmed by inspection for the presence of a vaginal mucous plug the next morning and if so was designated gestation day (GD) 0.5. Body weights were recorded weekly before and through gestation. Daily food intake was estimated by weighing remaining food at the end of each week and was used to calculate average daily caloric intake. Mice were maintained on their respective diets until time of euthanization and assay. Placentas and litter size (and GD 18.5 pup weight) were weighed and recorded.
Glucose Tolerance Test (GTT)-After 4 weeks of feeding mice with LFD or HFD, at GD Ϫ1 (before pregnancy) and at GD 16.5 mice were food-deprived overnight for 16 h with only water available. A base-line blood sample was taken via a small tail nick for determination of blood glucose by a handheld glucose meter (OneTouch Ultramini, Lifescan, Inc). A GTT was performed using 2 g/kg of intraperitoneal glucose (Sigma). Tailvein blood glucose level was monitored 15, 30, 45, 60, 90, and 120 min post-glucose injection.
Human Study Subjects-The human study was performed after approval of the Health Research Ethics Board at the University of Manitoba. Signed informed consent forms were obtained from all participants. Our study group consisted of healthy lean (body mass index (BMI) 20 -25 kg/m 2 ; n ϭ 9) and obese (BMI Ͼ35 kg/m 2 ; n ϭ 5) women with no gestational complications, including preeclampsia or gestational diabetes. BMI was assessed based on the pre-pregnancy weight. The basic clinical characteristics of the study population are summarized in Table 1.
Detection of RNA by Reverse Transcription and Quantitative Polymerase Chain Reaction (qPCR)-Total RNA was obtained from mouse placental tissue enriched for labyrinthine cells, the major site of CS and GH-V production as reported previously (18,51). Human placentas were collected on ice, and tissue samples (10 mm 3 ) were removed and immediately frozen (Ϫ80°C) after delivery for storage or processing. An effort was made to take a villous syncytiotrophoblast-rich sample from the same region of the chorion. The RNeasy Plus Mini kit was used for all RNA extractions (Qiagen). One g of total RNA was reverse-transcribed by the QuantiTect Reverse Transcription kit (Qiagen) according to the manufacturer's instructions. qPCR using specific primers ( Table 2) was done as described previously (49). ϪRT controls were performed using the same PCR primers and thermal cycle conditions as a control for the presence of genomic DNA. Specific amplifications were identified by a single peak in the melting curve and a single band in the final PCR product visualized in an agarose gel. The gene expression level in each sample (absolute quantification) was calculated from the standard curve and normalized to mouse/human glyceraldehyde 3-phosphate dehydrogenase (GAPDH) expression as appropriate. Tests were normally run in duplicate on six independent samples unless otherwise indicated.
Electrophoretic Mobility Shift Assay (EMSA)-EMSA was performed using nuclear extracts from human choriocarcinoma JEG-3 cells overexpressing C/EBP␤ and a 32 P-labeled oligonucleotides corresponding to wild type and mutated sequences from the 126-bp CS enhancer regions (25,26). JEG-3 cells were grown in monolayer in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 10% (v/v) fetal bovine serum and antibiotics (10 IU/ml penicillin, 10 mg/ml streptomycin). Cells were incubated in humidified atmosphere of 95% air and 5% CO 2 . Transfection of JEG-3 cells was done with TransIT-LT1 transfectin reagent (Mirus, Madison, WI) according to the manufacturer's protocol. Briefly, cells were seeded at 1 ϫ 10 6 cells per well in 100-mm plates 1 day before transfection. Cells were transfected with a total of 10 g of expression plasmids, pcDNA 3.1 (empty vector as negative control), or pCMV-C/EBP␤ (a kind gift from Dr. L. Laimins, North-western University, Chicago, IL) (52). Cells were harvested at 72 h post-transfection, and nuclear protein was isolated using EpiSeeker Nuclear Extraction kit (Abcam). EMSA and competition was done with oligonucleotides or specific antibodies to C/EBP␤ (sc-150; Santa Cruz Biotechnology, Inc.). Briefly, 8 g of nuclear extract was incubated with EMSA buffer containing 2 g of poly-dI⅐dC for 5 min. Radiolabeled oligonucleotide probes (1 ng) were then added, and the reactions were incubated for a further 20 min at room temperature. In competition experiments, 50-and 100-fold molar excesses of unlabeled oligonucleotide duplexes or antibody were added and preincubated on ice for 10 min before adding radiolabeled probes. The DNA-protein complexes were resolved in non-denaturing 5% (w/v) polyacrylamide gels and visualized by autoradiography.
Chromatin Immunoprecipitation (ChIP) on Chip Tiling Array Assays and ChIP-qPCR-"ChIP on chip" tiling array assays were performed by Active Motif (Carlsbad, CA). The assay and methodologies have been reported previously (33). Briefly, chromatin from placental labyrinth tissue was isolated, and subsequently immunoprecipitated DNA was amplified using a GenomePlex WGA kit (Sigma). The amplified DNA was fragmented and labeled using the DNA terminal labeling kit from Affymetrix (Santa Clara, CA) and then hybridized to Affymetrix GeneChip Tiling arrays. Arrays were analyzed using Affymetrix TAS software. Thresholds were selected and analyzed using Genpathway software (proprietary) that provides comprehensive information on genomic annotation, peak metrics, and sample comparisons for all peaks (intervals). To verify peaks, the results for ChIP DNA were normalized against existing input DNA array data (Active Motif database).
ChIP assays were performed according to the protocol from Millipore (EZ-Magna ChIP A/G chromatin immunoprecipita-  tion kit). Placental tissues were cross-linked with 1% formaldehyde before lysis. Chromatin was fragmented by sonication (10 times in 10-s pulses), then insoluble material was removed by centrifugation, and the DNA content was measured by spectrophotometry. Soluble chromatin was immunoprecipitated with 4 g of antibodies against C/EBP␤ (Santa Cruz) or RNA polymerase II (Santa Cruz, sc-899 and Covance, MMS-134R) overnight at 4°C as well as with normal rabbit immunoglobulins (Millipore) as a control. Protein A/G magnetic beads (Millipore) were added to immunoprecipitation reactions along with the antibodies and rotated for 24 h at 4°C. The magnetic beadsantibody complexes were subjected to a series of washes before elution. The eluted antibody complexes were reverse-crosslinked at 65°C overnight, and DNA was isolated using the QIAquick PCR purification kit (Qiagen). Quantitative PCR was performed in a 7500 Fast Real Time PCR system (AB Applied Biosystems) under conditions standardized for each primer set (Table 3). Dissociation curves were analyzed as a means to ensure the quality of amplicons and to monitor primer dimers.
Final PCR products were visualized as a single band in an aga-rose gel. ChIP enrichment was determined based on a percent input method as described previously (50). C/EBP␤ RNA Interference-mediated Knockdown in JEG-3 Cells-Cellular transfection to knockdown C/EBP␤ expression was performed using TransIT-TKO Transfection Reagent according to the manufacturer's protocol (Mirus). C/EBP␤ small interfering RNAs (siRNA; pools of five target-specific 19 -25-nucleotides siRNAs) were purchased from Santa Cruz Biotechnology, Inc. (sc-29229). The knockdown of the C/EBP␤ gene was performed in 24-well plates at cell density of 1 ϫ 10 5 cells/well using 50 nM siRNA and scrambled siRNA (Qiagen, siRNA negative control-1027280).
Whole Cell Protein Extraction and Detection-For detection of C/EBP␤ protein, 10 g of whole cell protein was analyzed by protein immunoblotting as previously described (50). Proteins were separated by 12.5% sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE), transferred to polyvinylidene fluoride membrane, and immunoblotted with two independent antibodies specific to C/EBP␤ (Santa Cruz, sc-150, and Millipore, 04-1153). ␤-Tubulin (sc-9104, Santa Cruz) was assessed as control for protein loading. The proteins were visualized using horseradish peroxidaseconjugated anti-immunoglobulin G (IgG) secondary antibody and ECL plus immunoblotting detection reagents (Thermo Fisher Scientific Inc., Nepean, ON, Canada).
Statistical Analysis-Statistical analysis was performed using GraphPad Instat software. For single comparisons, paired t tests were applied as appropriate. For multiple comparisons, one-way analysis of variance was used with the Tukey-Kramer or Dunnet's post-test as appropriate. A value of p Ͻ 0.05 was considered statistically significant and is represented in figures as: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. FIGURE 1. A high saturated fat diet leads to significant weight gain accompanied by glucose intolerance in mice during pregnancy. A, shown is daily food and caloric intake. B, shown is the effect of a 4-week LFD or HFD on female hGH/CS-TG mouse body weight gain before and during pregnancy. Values are the mean Ϯ S.E. (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001, n ϭ 9). LFD ϭ black-filled; HFD ϭ white-filled. C, shown are differences in gonadal white adipose tissue (fat pad) between the pregnant mice on a LFD or HFD at GD 18.5 (*, p Ͻ 0.05, n ϭ 6). D, the effect of a HFD on glucose homeostasis before (GD Ϫ1) and in pregnancy (GD 16.5) was assessed by a GTT. Mice were fasted 16 h overnight and injected with 2 g/kg body weight of glucose intraperitoneal followed by blood glucose measurement for 2 h. E, glucose excursions were also measured as area under the curve (AUC) (mM glucose ϫ min) and expressed as mean Ϯ S.E. (*, p Ͻ 0.05, **, p Ͻ 0.01, n ϭ 6).

Region
Primer sequence

RESULTS
A HFD Leads to Significant Weight Gain before and during Pregnancy and Is Accompanied by Glucose Intolerance in hGH/ CS-TG Mice-To study the effects of maternal obesity on human placental CS/GH-V levels, hGH/CS-TG (CD-1) mice were placed on a LFD or HFD. Mice fed the HFD had a greater daily calorie intake than those fed the LFD (Fig. 1A). The obesogenic high fat feeding for 4 weeks before and during pregnancy resulted in a profound increase in body weight. A significant increase in body weight gain in transgenic mice fed the HFD was evident after only 1 week (5.2-fold, p Ͻ 0.01, n ϭ 9), and this trend continued throughout the dietary protocol and pregnancy (1.5-fold at GD 18.5, p Ͻ 0.05, n ϭ 9) (Fig. 1B). The total weight of the maternal gonadal fat pad was increased 3.3fold in hGH/CS-TG mice fed with the HFD (p Ͻ 0.05, n ϭ 6) (Fig. 1C). In conjunction with the changes in body weight, glucose clearance was assessed by a GTT at GD Ϫ1 (after 4 weeks of HFD and before pregnancy) and at GD 16.5. Mice maintained on the HFD showed normal fasting blood glucose levels compared with the LFD group (data not shown) but were glucose-intolerant and insulin-resistant before and during pregnancy as demonstrated by impaired clearance of an intraperitoneal glucose challenge (p Ͻ 0.05, n ϭ 6) (Fig. 1D). Glucose excursions were also measured as area under the curve (AUC), which was significantly higher in mice on the HFD relative to the control LFD group (p Ͻ 0.05, n ϭ 6) (Fig. 1E). The same trend in weight gain seen in hGH/CS-TG mice was detected in wild type, non-TG, and CD1 mice 2 weeks after initiation of the HFD (2.4-fold, p Ͻ 0.05, n ϭ 6) ( Fig. 2A). A GTT was also performed on non-TG CD1 mice, and similar significant impaired glucose-load clearances as seen in TG mice before and during pregnancy were also observed (p Ͻ 0.05, n ϭ 6) (Fig. 2B). Placentas and litter size weight were not affected by the HFDinduced obesity in hGH/CS-TG and non-TG CD1 mice, and they were comparable to the LFD group (data not shown).

CS-A, CS-B, and GH-V RNA Levels Are Reduced by High Fat Diet-induced Obesity and Glucose Intolerance in hGH/CS-TG
Mice in Pregnancy-To assess whether the CS/GH-V genes are targets for maternal obesity, CS-A, CS-B, and GH-V RNA levels in placental labyrinth RNA from hGH/CS-TG mice on a LFD or HFD were examined by qPCR at GD 18.5. The HFD, initiated before and during pregnancy, resulted in an apparent decrease in CS-A (30%, n ϭ 9), CS-B (60%, n ϭ 9), and GH-V (40%, n ϭ  Induction of a diabetes-like state is supported by an increase in adipokine (placental) leptin transcripts (n ϭ 8-10). C, RNA was isolated from term placenta samples and assessed by qPCR for CS-A, CS-B, and GH-V relative to GAPDH transcripts in placentas from obese (BMI Ͼ35 kg/m 2 , n ϭ 5) (black-filled) compared with lean (BMI 20-25 kg/m 2 , n ϭ 9) women (white-filled). Values were calculated from a standard curve (absolute quantification). The results are expressed as mean percentage change mean ϮS.E. relative to LFD, which is arbitrarily set to 100%. Significance was assessed by t test (*, p Ͻ 0.05; **, p Ͻ 0.01). 9) RNA levels (Fig. 3A). Induction of a diabetes-like state was also supported by an increase in adipokine (placental) leptin transcripts (3.4-fold, n ϭ 6) (Fig. 3B), which is a characteristic of pregnancies complicated with diabetes that require insulin therapy in humans (53). Mouse placental lactogen (PL) I and PL-II expression were also investigated, as they play a major role in maternal metabolic adaptation in the mouse by regulating pancreatic ␤-cell expansion in pregnancy (54,55); glucose intolerance and diabetes develops when prolactin receptors, for which PL-I/II are ligands, are lacking in pregnancy (56). Maternal obesity in hGH/CS-TG mice resulted in an apparent 55% decrease in PL-I RNA levels (p Ͻ 0.05, n ϭ 9) but no significant change in PL-II (Fig. 3B). The same pattern of expression for PL-I and PL-II RNA was also observed in wild type CD1 mice at GD 18.5 (p Ͻ 0.05, n ϭ 6) (data not shown).

CS-A, CS-B, and GH-V RNA Levels Are Decreased in Fullterm Human Placenta Samples from Pregnancies Complicated by Maternal
Obesity-Human term placenta CS/GH-V RNA levels were assessed relative to GAPDH in pregnancies complicated by extreme maternal obesity by qPCR. There was a significant decrease in CS-A (88%), CS-B (93%), and GH-V (81%) RNA expression in placentas from obese (BMI Ͼ 35 kg/m 2 ) compared with lean (BMI 20 -25 kg/m 2 ) women (p Ͻ 0.01, n ϭ 5-9) (Fig. 3C).

Placental C/EBP␤ RNA Levels and Protein Binding to CS Enhancer Regions Are Decreased in Obese hGH/CS-TG and Pregnant
Women-RNA levels for CS enhancer binding proteins C/EBP␣, C/EBP␤, TEF-1, and TEF-5 were assessed in placenta samples from both hGH/CS-TG mice fed a HFD and obese individuals. A significant 20% decrease in C/EBP␤ RNA levels was detected in the placental labyrinth samples from TG mice on the HFD (p Ͻ 0.01, n ϭ 9); however, no change in C/EBP␣, TEF-1, and TEF-5 transcripts was observed (n ϭ 9) (Fig. 4A). Similarly, a 62% decrease in C/EBP␤ RNA levels (p Ͻ 0.01, n ϭ 5) was also detected in human full-term placenta obtained from obese women, but again no significant effect on C/EBP␣, TEF-1, and TEF-5 transcripts was seen (Fig. 4B). Maternal hepatic C/EBP␣ and C/EBP␤ transcript levels from hGH/CS-TG mice fed a HFD were also assessed. A significant 2.5-fold increase in C/EBP␤ RNA levels was detected in the liver samples from TG mice on a HFD compare with the LFD group (p Ͻ 0.001, n ϭ 4); however, no change in C/EBP␣ transcripts was observed (n ϭ 4) (Fig. 4C).
The association of C/EBP␤ protein with the equivalent nuclease-protected sequences in CS-A and CS-B gene 3Ј-enhancer regions (25,26) (Fig. 5A) and identified as a component of placenta CS enhancer activity was assessed via EMSA in vitro. Elevated levels and a specific complex were detected between putative CS-A and CS-B 3Ј-enhancer sequences and nuclear protein from JEG-3 cells overexpressing C/EBP␤ versus "control" JEG-3 cells (Fig. 5B). To confirm C/EBP␤ binding, the EMSA was repeated with C/EBP␤ antibodies, and a "supershifted" band was detected (open arrowhead, Fig. 5B). The above-mentioned band was also effectively competed with a 50and 100-fold molar excess of a well characterized C/EBP␤ DNA element (positive control) (57). By contrast, oligonucleotides containing nucleotides 61-86 of the CS-A and CS-B enhancer regions in which the C/EBP sequences were mutated were not efficient competitors (Fig. 5C). To assess the impact of maternal obesity on in vivo association of C/EBP␤ protein with CS gene enhancer regions, isolated placental labyrinth chromatin from hGH/CS-TG mice fed a LFD versus HFD was assessed using a tiling array. The reason for use of the ChIP on chip tiling array assay as opposed to ChIP-qPCR was the extensive sequence similarities between the enhancer regions of the three CS genes including hCS-A, hCS-B, and pseudo-hCS-L (overall Ͼ95%) (Fig. 5A). Levels of C/EBP␤ associated with the CS-A, CS-B, and pseudo-CS-L enhancer regions were significantly reduced in placental chromatin from TG mice fed with the HFD (p Ͻ 0.001, in a pool of three samples) (Fig. 6, A and B). The C/EBP␤ peak intervals spanned nucleotides 59300535 to 59303012, which includes the full 241-bp enhancer regions (33); nucleotide numbering corresponds to human chromosome 17 sequences as included in the University of California Santa Cruz Genome Browser, Human 2004. This pattern of C/EBP␤ association with the enhancer regions in chromatin of TG mice was comparable to our previous observation seen in human term placenta in situ (33). The decrease in C/EBP␤ binding associated with mice fed the HFD versus LFD was further validated by ChIP-qPCR using specific primers that amplify the three enhancer regions of the CS-A, CS-B and pseudo-CS-L genes (Fig. 6C). The association of C/EBP␤ with the CS FIGURE 4. A significant decrease in transcription factor C/EBP␤ RNA levels, but not C/EBP␣, TEF-1, and TEF-5, is detected in the placenta with maternal obesity. Levels of C/EBP␣, C/EBP␤, TEF-1, and TEF-5 in total RNA from placental labyrinth from hGH/CS-TG mice at GD18.5 on a LFD (black-filled) or a HFD (white-filled) (n ϭ 9) (A) human full term placenta obtained from lean (n ϭ 9) (black-filled) and obese (n ϭ 5) (white-filled) women (B) or maternal liver from hGH/CS-TG mice at GD18.5 on a LFD (black-filled) or a HFD (white-filled) (n ϭ 4) (C) were assessed relative to GAPDH transcripts by qPCR. Values were calculated from a standard curve (absolute quantification). The results are expressed as mean percentage change (mean Ϯ S.E.) relative to LFD, which is arbitrarily set to 100%. Significance was assessed by t test (**, p Ͻ 0.01, ***, p Ͻ 0.001).

Regulation of Human Placental GH Genes in Obesity
enhancer regions was also assessed in isolated term placenta chromatin from obese versus lean women by ChIP-qPCR (Fig.  6D). A significant 94% reduction in C/EBP␤ associated with the CS enhancers was observed in placentas from obese women (p Ͻ 0.001, in a pool of 5 samples).
A loss of enhancer activity is expected to decrease transcription. Association of RNA polymerase II (pol II) with the CS-A, CS-B, and GH-V promoter regions was assessed as an indication of transcriptional status of the CS/GH-V promoters by ChIP-qPCR assay. Samples of placenta chromatin from hGH/ CS-TG mice on a LFD versus HFD as well as from pregnancies complicated by chronic maternal obesity were investigated. There was a significant 40 -60% reduction in RNA pol II associated with the upstream flanking promoter regions of the CS/GH-V genes (p Ͻ 0.01, n ϭ 3) (Fig. 7, A and B).
Interference with Human C/EBP␤ Production Decreases Human CS/GH-V RNA Levels-Human choriocarcinoma JEG-3 cells were used to determine whether a decrease in human C/EBP␤ production will affect endogenous human CS-A and CS-B RNA levels. JEG-3 cells were treated with siRNAs to specifically knock down C/EBP␤, and subsequently CS/GH-V as well as C/EBP␤ RNA levels were assessed at 72 h  (57) and mutations of the wild type C/EBP sites from CS-A and CS-B at the underlined sequences (from AAACG to gacta) were used with their respective wild type radiolabeled counterpart and assessed by EMSA. A specific DNA-protein complex is indicated (closed arrowhead). by qPCR. A significant 55% (p Ͻ 0.01) reduction in C/EBP␤ RNA was accompanied by an 85% (p Ͻ 0.05) reduction in C/EBP␤ protein levels (Fig. 8, A[en]C), and this was associated with significant decreases in CS-A (66%, p Ͻ 0.01), CS-B (77%, p Ͻ 0.001), and GH-V (83%, p Ͻ 0.05) RNA levels (Fig. 8C).

DISCUSSION
Expression of the metabolic homeostatic placental CS and GH-V hormone genes was decreased in two models of maternal obesity in pregnancy, specifically, acute HFD-induced obesity in mouse pregnancy and chronic obesity in human pregnancy. C/EBP␤ binds to the downstream CS gene enhancer regions, which although distal, would also flank the GH-V gene, and stimulates CS/GH-V promoter activity. Maternal obesity resulted in decreased association of C/EBP␤ with the CS enhancer regions as well as a corresponding reduction in RNA pol II binding at CS and GH-V gene promoters. Further evidence for a direct effect of C/EBP␤ was provided through knock down of endogenous human C/EBP␤, which decreased CS and GH-V RNA levels in human placenta tumor cells. These data suggest CS/GH-V gene expression can be modified during pregnancy through the effects on C/EBP accessibility and also supports the use of humanized mice as an additional and complementary tool to study CS/GH-V gene control during pregnancy at the molecular level.
The C/EBP family consists of six highly related members (␣, ␤, ␥, ␦, ⑀, and ) that are widely involved in controlling proliferation and differentiation in various cell types (for review, see Ref. 58). C/EBP␤ is expressed abundantly in the placental labyrinth as well as in adipose tissue and liver (59). C/EBP␤ is known to play a role in mouse placental development and embryonic outcome (38 -40). C/EBP␤ expression in white adi-FIGURE 6. Effect of maternal obesity on C/EBP␤ binding to CS 3-enhancer regions. A, the association of C/EBP␤ protein with CS 3Ј-enhancer regions in labyrinth chromatin isolated from hGH/CS-TG mice fed a low versus high fat diet was assessed by ChIP on chip tiling array. Arrays were analyzed using Affymetrix TAS software. Thresholds were selected (Ͼ1.5) and were analyzed using Genpathway software (proprietary) that provides comprehensive information on genomic annotation, peak metrics, and sample comparisons for all peaks (intervals). Gray arrows under the map of the GH/CS locus indicate the positions of the five GH/CS genes. White filled arrows indicate the positions of CS 3Ј-enhancer regions. B, to verify peaks, the results for ChIP DNA were normalized against existing input DNA array data. Intervals on the CS-A (2130 nucleotides, 59323740-59325870) and CS-B (2477 nucleotides, 59300535-59303012) genes that include enhancer sequences were assessed. The equivalent CS-L interval (1045 nucleotides, 59338177-59339222) was also included. Significant reductions in C/EBP␤ association with the CS-A, CS-B, and pseudo-CS-L downstream enhancers in the hGH/CS-TG mice fed with a HFD were observed (p Ͻ 0.001, in a pool of 3 samples). C, tiling array results were also validated by ChIP-qPCR using an anti-C/EBP␤ antibody and primers (bold sequence in Fig. 5A) to amplify the equivalent CS-A, CS-B, and pseudo-CS-L 3Ј-enhancer regions. The binding events data were normalized to the amount of input chromatin used for the C/EBP␤ immunoprecipitation reaction and the untranscribed region on chromosome 6 (Untr6) in the mouse, used as a negative control. D, ChIP-qPCR was also used to assess the binding of C/EBP␤ to the CS 3Ј-enhancer regions in human term placenta chromatin from lean (black-filled) and obese (white-filled) women. Negative controls represent primer pairs for the well characterized untranscribed human genomic region Untr12. The results are expressed as relative mean -fold change, mean Ϯ S.E. compared with the control value, which is arbitrarily set to 1.0. Significant differences were assessed by t test and are indicated by: **, p Ͻ 0.01 and ***, p Ͻ 0.001, n ϭ 3-5.
pose tissue and liver is implicated in adipogenesis/glucose metabolism in the context of obesity (42,43). The absence of C/EBP␤ protects mice from HFD-induced obesity, suggesting its active involvement in the pathogenesis of obesity (43,60). C/EBP␤ is also known to play a role in obese type 2 diabetic individuals (61). Consistent with a role in obesity, we observed a significant up-regulation of hepatic C/EBP␤ in hGH/CS-TG mice on the HFD versus LFD. We show for the first time, however, that placental C/EBP␤ is a target for maternal obesity and that its expression is significantly reduced in placentas from pregnancies complicated by obesity from two species, human and mouse. The basis for the differential expression of C/EBP␤ in placenta versus liver is not yet clear but likely reflects differences in functional and possibly tissue-specific targets linked with different physiological outcomes.
The human CS genes are expressed at high levels in placental villus syncytiotrophoblasts throughout pregnancy (12). The specific sequence requirements for efficient in vivo CS expression have not as yet been determined. Placental enhancer activity (300 -400-fold) was detected about 2 kb downstream of the CS genes and has been linked to a 126-bp region containing two sites of DNA-protein interaction, as detected in both CS-A and CS-B sequences with placental cell protein (25,26). TEF was shown to bind one of these sites (27,34,36); however, the identity of the protein, detected in placental and non-placental cell extracts capable of binding the second site, referred to as FP2 for CS-A and DF3 for CS-B 3Ј-enhancer sequences, was not defined (25,26). Subsequently, we identified putative C/EBP sites in these sequences using transcription factor databases (33). We also showed that C/EBP overexpression significantly stimulated the expression of hybrid reporter genes driven by a minimal (496 bp) CS promoter and containing either CS-A or CS-B 3Ј-enhancer regions in JEG-3 cells (33). Our data now confirm C/EBP␤ binding to nucleotides 61-86 of the CS-A or CS-B 3Ј-enhancer regions in vitro, which correspond to the previously identified "footprint" regions FP2 and DF3, which is linked to placental enhancer (25,26). This is also consistent with our previous observation that C/EBP␤ associates with the CS 3Ј-enhancer regions in normal human term placenta chromatin in situ (33). Furthermore our finding is consistent with a significant role for C/EBP␤ as a physiological regulator of the CS genes as well as a target for maternal obesity in pregnancy; decreases in C/EBP␤ RNA, protein binding, and CS RNA levels were all detected in placentas from hGH/CS-TG mice and human full term placenta associated with acute HFD-induced mouse and/or chronic human obesity in pregnancy. The decrease in C/EBP␤ association with CS 3Ј-enhancer regions was also accompanied by a corresponding decrease in RNA pol II levels at the CS-A and CS-B promoters, suggesting a direct effect of C/EBP␤ on CS gene expression. A direct effect of C/EBP␤ was also supported by the decrease in endogenous CS-A and CS-B RNA levels observed with siRNA knockdown of human C/EBP␤ in JEG-3 cells.
The decrease in C/EBP␤ with maternal obesity in mouse or human pregnancies as well as through knockdown in JEG-3 cells resulted in decreases in GH-V RNA levels and RNA pol II association with the GH-V promoter. Unlike the CS genes, the GH-V gene has no conserved downstream enhancer region or equivalent CCAAT-binding site (assessed by using available transcription factor databases). This then raises a question as to the mechanism by which GH-V gene expression is down-regulated in response to a loss of C/EBP␤. An explanation may lie in the position of hGH-V between hCS-A and hCS-B in the GH/CS gene locus, which would result in the GH-V gene being flanked at both ends by two potent placental CS 3Ј-enhancer and CS promoter regions. An assessment of hyperacetylated histone H3 and H4 in human syncytiotrophoblast versus cytotrophoblast cultures indicated that the CS and GH-V genes reside in a region of hyperacetylated chromatin centered between hCS-A and hCS-B, presumably reflecting increased accessibility (35). In addition, examples of transcriptional regulation by enhancers from an adjacent gene have been described (62). The decrease in RNA pol II levels detected at the GH-V as well as the CS-A and CS-B promoters in human and murine placental samples with maternal obesity is consistent with this possibil-FIGURE 7. RNA polymerase II (pol II) levels are reduced at the CS/GH-V promoters with maternal obesity. ChIP was performed with an anti-RNA pol II antibody to assess changes in association of the transcriptional machinery with CS and GH-V promoter regions in placental chromatin from hGH/ CS-TG mice with acute HFD-induced maternal obesity (A) and human pregnancies complicated by more chronic maternal obesity (B). The binding events data were normalized to the amount of input chromatin used for the RNA polymerase II immunoprecipitation reaction and the control untranscribed regions of mouse (Untr6) and human (Untr12) as appropriate. The results are expressed as relative mean -fold change, mean Ϯ S.E. compared with the control value, which is arbitrarily set to 1.0. Significant differences are indicated by: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001, n ϭ 3-5.
ity. The mechanism of enhancer-promoter looping is a well characterized model for regulation of a neighboring gene by an enhancer (63,64). In this case hGH-V expression, like hCS-A and hCS-B, might be influenced by a reduction in C/EBP␤ binding at the downstream CS 3Ј-enhancers and, as a result, a corresponding decrease in CS enhancer activity. Relatively long distances and looping out of intervening sequences are required to bring enhancers and target promoters into close proximity. This allows delivery and interaction of bound "enhancer" and "basal promoter" proteins, thereby activating and controlling gene expression (65). Physical and functional interaction between C/EBP␤ and chromatin remodeling complexes has been described, which would be consistent with this role for the CS and potentially GH-V genes (66).
C/EBP␤ production in placenta is localized to syncytiotrophoblasts in both human and non-human primates, which is consistent with a role in CS/GH-V gene regulation (67). Remarkably, the villous cytotrophoblasts are negative for C/EBP␤ immunostaining at all stages of gestation (37,68,69), and as the syncytial layer derives from the fusion of cytotrophoblasts, this might predict a role for C/EBP␤ in cytotrophoblast differentiation into syncytiotrophoblast (70). By extension, a decrease in C/EBP␤ availability resulting from maternal obesity might be expected to preserve cytotrophoblasts and potentially an immature placental structure with compromised function. Histopathological assessments of placentas complicated by obesity are variable and, as such, are inconclusive. In a recent study no substantial difference in placental maturity and the degree of terminal villi formation was observed in placentas from obese compared with lean women; placental immaturity was characterized as larger chorionic villi and more centered blood vessels (71). By comparison, in an earlier study, severe placental villous immaturity as defined by delayed villous maturation was detected in some obese women (BMI Ͼ 30 kg/m 2 ) who showed signs of mild glucose intolerance but did not meet the criteria for gestational diabetes (72). Placentas from obese non-human primates also showed signs of decreased placental syncytiotrophoblast amplification factor and thicker villi, which is consistent with structural impairment of the syncytiotrophoblast microvilli surface (73). Thus, the possibility that maternal obesity may interfere with the transition of cytotrophoblasts to syncytiotrophoblast and, as result, indirectly affect CS/GH-V gene expression cannot be ruled out (70). Of course, the possible indirect and direct effects of C/EBP␤ on CS gene expression are not mutually exclusive.
Although the mechanism by which the CS/GH-V genes are regulated by C/EBP␤ appears conserved, the degree of response observed in the "acute" mouse model was reduced relative to the "chronic" human model of maternal obesity. Possible explanations may relate to the nature of the obese state and exposure time to the obesity-associated endocrine complications, species differences including related to placental structure as well as FIGURE 8. Knockdown of human C/EBP␤ in JEG-3 cells results in a decrease in endogenous CS and GH-V RNA levels. JEG-3 cells were treated with siRNAs for C/EBP␤ or a "scrambled sequence" as a negative control for 72 h. After siRNA transfection, C/EBP␤ RNA levels were assessed relative to GAPDH transcripts by qPCR (A), and protein (ϳ45 kDa) levels in whole cell lysate (10 g) were examined by SDS-PAGE and immunoblotting (IB) (B). Two independent antibodies were used and detected the same pattern by chemiluminescence; a representative blot using Millipore, 04-1153 C/EBP␤ antibody is shown. ␤-Tubulin (ϳ55 kDa) was used as loading control. C, C/EBP␤ and ␤-tubulin protein levels were quantified using Image J software. D, CS/GH-V RNA levels were assessed at 72 h relative to GAPDH transcripts by qPCR post C/EBP␤ siRNA transfection. The results are expressed as mean percentage change, mean Ϯ S.E., relative to the scrambled sequence control siRNA, which is arbitrarily set to 100%. Significance was assessed by t test, and significant differences are indicated by: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001, n ϭ 4. redundancy related to the C/EBP family. The differences in the time course of obesity, acute versus chronic, may provide variable opportunity for metabolic derangements, like inflammation, to influence placental gene expression (71,73). The mouse and human placenta serve the same function but have different cellular composition and organization. In mice, a bilayer of syncytiotrophoblasts separates fetal capillaries from the maternal side, whereas in humans this separation involves only one layer of syncytiotrophoblasts (74,75). Moreover, the properties of these cell types may be different; the major placental lactogen (PL-II) in mouse is expressed predominantly by trophoblasts (74 -76), whereas the major source of placental lactogen (CS) in the human placenta is syncytiotrophoblasts (12). CS expression is localized to cells positioned adjacent to the maternal-fetal interface and throughout the labyrinth in the hGH/CS-TG mice (18,21). This is consistent with CS expression in both syncytiotrophoblasts as well as trophoblast cells (18). There is also evidence for redundant functions of C/EBP␣ and C/EBP␤ in placentas based on null mutations in mice, although C/EBP␤ is most abundant in human and mouse placenta (37,38,68,69). Despite these differences, the data presented are consistent with a negative impact of maternal obesity, whether short or long term, on the expression of C/EBP␤ and its downstream targets, which would include the CS genes.
Three placenta tissue/cell systems were used to study the effects of maternal obesity and/or C/EBP␤ on human CS/GH-V gene regulation. Multiple human choriocarcinoma cell lines as well as primary term placental cultures have helped increase our understanding of how placental members of GH gene family are controlled by various metabolic and endocrine factors. They have proven to be invaluable tools and models for the study of the cellular and molecular aspects of human placental trophoblasts (77,78). The humanized hGH/CS-TG mouse complements these cell systems by modeling pregnancy and allowing assessment of CS/GH-V gene expression in the presence of an intact endocrine system, with sampling as pregnancy progresses; this includes pregnancies with complications, such as the HFD-induced maternal obesity and insulin-resistant state as used here.