Oxytocin receptor is regulated by Peg3

Mouse Peg3 encodes a DNA-binding protein involved in the milk letdown process. In the current study, we tested whether PEG3 controls the expression of the oxytocin receptor gene. According to the results, PEG3 directly binds to a genomic region within the 3rd exon of Oxtr, which contains a DNA-binding motif for PEG3. In nursing female mice, removal of PEG3 resulted in the increased expression of Oxtr in mammary epithelial cells and also in the hypothalamus. This suggests a repressor role of PEG3 in the expression of Oxtr in these tissues. Overall, this study suggests that Peg3 may function as a direct transcriptional regulator for Oxtr expression that acts to moderate the milk letdown process.


Introduction
Peg3 (Paternally Expressed Gene 3) is an imprinted gene localized in proximal mouse chromosome 7/human chromosome 19q13.4 [1][2][3]. This gene encodes a DNA-binding protein with 12 C2H2 zinc finger motifs known to bind to a large number of genomic targets [4][5][6][7]. The list of known downstream genes includes Pgm2l1, H19, Msl1 and Msl3. PEG3 also contains a KRAB-A domain, thus is predicted to be a transcriptional repressor through its interaction with KAP1 [8]. This predicted repression function has been further confirmed through the observation that the expression levels of its downstream genes are usually up-regulated in the tissues of the mouse mutant models targeting Peg3 [6,7]. According to the results from mouse knockout (KO) models, Peg3 is also involved in controlling fetal growth rates and maternalcaring behaviors [9][10][11][12][13]. However, the detailed mechanism by which Peg3 is involved in these biological pathways is not well understood.
In murine Peg3-KO studies, both nursing females and pups tend to have a problem in milk provision, subsequently causing reduced growth rates in the pups lacking Peg3 [9][10][11][12][13]. In placental mammals, milk letdown is mediated through oxytocin circuitry involving the peptide hormone oxytocin (Oxt) and its receptor, oxytocin receptor (Oxtr) [14,15]. Oxytocin is produced in the supraoptic nucleus (SON) and paraventricular nucleus (PVN) of the hypothalamus. Axonal projections of these nuclei lead to the posterior pituitary gland, where oxytocin is released into the bloodstream, and exerts its effects through binding to oxytocin receptors present in the target tissues, including uterus, ovary and mammary gland [14,15]. In a prior mutant model of Peg3, reduced numbers of Oxt-expressing neurons were initially observed, a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 which was then thought to be a main cause for the milk provision problem [9]. However, this has been later argued by two independent studies reporting that the mutations did not cause any change in the numbers of Oxt-immunoreactive neurons in the hypothalamus [13,16,17]. Nevertheless, subsequent experiments confirmed again the presence of the milk provision problem in several Peg3-KO models, and further substantiated that this defect was more pronounced through the conditional mutation of Peg3 in the mammary gland than in the hypothalamus [12]. This further suggests that Peg3 might play more significant roles in the mammary gland than in the hypothalamus for the milk provision process [12]. Consistent with this, Peg3 is expressed not only in the hypothalamus but also in the other tissues, including placenta, uterus, testis, ovary, and mammary gland [1][2][3].
Oxtr is also known to be widely expressed in various tissues, including the ovary, uterus, brain and mammary gland, whereas Oxt is mainly expressed in the hypothalamus. Yet, conditional KO of Peg3 in the mammary gland exhibited similar defects as those observed from the earlier Peg3-KO models, suggesting that Peg3 may be involved in the oxytocin circuitry not only through Oxt in the hypothalamus but also through Oxtr in the mammary gland. This possibility was tested in the current study by performing a series of experiments. The results suggest that PEG3 may function as a transcriptional repressor for the expression of Oxtr.

PEG3 binding to the mouse Oxtr locus
The predicted protein function of PEG3 has been previously tested through performing several ChIP-seq (Chromatin ImmunoPrecipitation-sequencing) experiments [4][5][6]. For these experiments, we time-mated C57BL/6J females with Peg3 CoKO/+ males, and the subsequent 14.5-dpc (day postcoitum) embryos were used for preparing a set of mouse embryonic fibroblast (MEF) cells, Peg3 +/+ (WT) and Peg3 +/CoKO (KO) [6]. The chromatin prepared from this set of MEFs were individually immunoprecipitated with anti-PEG3 antibody, and subsequently analyzed with Next Generation Sequencing. This survey identified an initial set of 16 downstream genes, which have been recently published [6]. We further inspected the output by manually scanning the entire set of chromosomes for potential downstream genes. This series of manual inspections identified an additional set of potential genomic targets, including the Oxtr locus (Fig 1A). At the Oxtr locus, a ChIP-seq peak was found within the 3rd exon, which spans a 225-bp genomic region (chr6:112,488,870-112,489,904 in mm10). It is also relevant to note that we were unable to find any significant ChIP-seq peak around the Oxt locus (S1 File).
Potential binding of PEG3 to the identified region of the Oxtr locus was further tested through performing independent ChIP experiments (Fig 1B). Several sets of the chromatin from MEFs and brains of WT and KO were immunoprecipitated again with the antibody against PEG3. As shown in Fig 1B, the levels of enrichment by the antibody were higher in WT than those from KO-MEF cells, confirming the binding of PEG3 to the target region of the Oxtr locus. This was also the case for the brain, which displayed higher levels of enrichment in the WT than in the KO samples. Thus, this series of ChIP experiments confirmed that the identified region is indeed bound by PEG3.
According to previous studies, PEG3 is known to bind to genomic regions with the following consensus DNA motif: 5'-N-N-G-G-[C/G]-N-[C/G-]-T-3' with N indicating any base [5]. Thus, we scanned the 20-kb genomic region of the Oxtr locus to identify potential genomic regions with this motif. This manual scanning also focused on the genomic regions with potential regulatory function based on their histone modification and DNA methylation profiles. Oxtr is known to be expressed in cerebellum and uterus, thus we selected three genomic regions that are protected from DNA methylation in these tissues. From these regions indicated with blue bars in Fig 2A, we have identified 5 potential binding sites, BS1 through 5. PEG3 binding to these potential binding sites was subsequently tested through a series of EMSAs (Electrophoretic Mobility Shift Assays) (Fig 2B). For this series of analyses, we used a previously characterized binding site from the Pgm2l1 locus as a probe [4,5], and nuclear extracts prepared from mouse brains. Binding of PEG3 to the P 32 -labeled Pgm2l1 oligonucleotide duplex was competed against each candidate DNA-binding site. The results indicated that two candidate sites, BS2 and BS5, did not compete at all, suggesting that these two regions may not contain a DNA-binding site for PEG3. In contrast, the three remaining binding sites, BS1, BS3, and BS4, competed well against the Pgm2l1 probe, suggesting that these regions may have high-affinity DNA-binding sites for PEG3. In particular, BS4 is located within the target region that had already been identified through independent ChIP-seq experiments (Fig 1). Thus, this series of analyses confirmed that the Oxtr locus has several potential binding sites for PEG3, and that one of these binding sites is indeed bound by PEG3 in vivo.

Mutational effects of Peg3 and Oxtr on animal survival and growth rates
Potential involvement of Peg3 in the expression and function of Oxtr was further tested through crossing two mutant mouse lines, Oxtr Venus/+ and Peg3 CoKO/+ (Fig 3A). Both mutant alleles were generated using similar knock-in (KI) and knock-out (KO) strategies [11,18]. The 3rd exon of Oxtr makes up the majority of the coding region (amino acid position 1-306 of the 388 amino-acid-long ORF), which has been replaced by an exogenous cassette expressing the reporter construct Venus. Thus, the mutant allele expresses the reporter protein instead of the endogenous gene product, oxytocin receptor. In the case of the Peg3 locus, another cassette . The P 32 -radiolabeled oligonucleotide duplex probe derived from the Pgm2l1 locus was mixed with mouse brain nuclear extracts (lane 2), and further competed against a set of 5 potential binding sites (lane 3-7). The relative ratio of the labeled Pgm2l1 probe to BS1-BS5 oligonucleotide duplexes was 1 to 200. The sequences of the oligonucleotides used for this assay are shown on right. The potential binding sites are indicated with either underlines or red fonts. The sites with red font indicate the confirmed DNA-binding sites through EMSA, whereas the sites with underlines indicate the predicted but unconfirmed sites. The bold-typed nucleotide bases within each site represent the critical bases for PEG3 binding.
https://doi.org/10.1371/journal.pone.0202476.g002 expressing β-Galactosidase (β-Gal) and Neomycin Resistance (NeoR) was inserted into the 5th intron of Peg3. In this mutant model, two sets of Poly(A) signals included in the expression cassette cause transcriptional truncation for the endogenous gene, thus resulting in the removal of PEG3 protein in the entire body [11,12]. Since Peg3 is expressed only from the paternal allele as an imprinted gene, paternal transmission of the mutant allele derives its phenotypic outcomes, such as reduced growth rates and defects in maternal-caring behaviors.
As a feasibility test, male heterozygotes for Peg3 CoKO/+ were bred with female heterozygotes for Oxtr Venus/+ , deriving 10 litters of 78 pups (Fig 3B). The average litter size, 7.8, appeared to be normal given the genetic background of the two mutant strains, C57BL/6J. The observed frequency of the four genotypes among the pups were also overall similar to those predicted based on mendelian inheritance pattern. Although the frequency of the pups with double heterozygote, Oxtr Venus/+ ; Peg3 +/CoKO , was slightly smaller than the expected, 16 versus 19.5, there was no statistical significance. Thus, this suggests that the pups with the four genotypes are all viable without any major phenotypic consequence. According to previous studies, the paternal transmission of Peg3 CoKO/+ has been shown to cause reduced body size among the mutant pups [9][10][11][12][13]. Consistent with this, we also observed reduced body weights among the two groups of the pups, the first group (Oxtr +/+ ; Peg3 +/CoKO ) and the second group (Oxtr Venus/+ ; Peg3 +/CoKO ). However, we have not observed any weight difference between these two genotypes, indicating that the mutation on Oxtr may not have any additional effect on the reduced growth rates of the animals that had already been impacted by the mutation on Peg3.

Mutational effects of Peg3 on the expression of Oxtr in the mammary gland
The females generated from the breeding experiments were also used to examine the mutational effects of Peg3 on the expression levels and patterns of Oxtr (Fig 4 and Fig 5). For this series of analyses, we focused on the two major organs that are involved in milk letdown, the mammary gland and hypothalamus [14,15]. The tissues were also harvested from the dams at one day postpartum, since this time-point should be critical for establishing the milk letdown process in the dams. We set up several sets of breeding pairs with each housing one wild-type male along with two female littermates with the following genotypes: Oxtr Venus/+ ; Peg3 +/+ and Oxtr Venus/+ ; Peg3 +/CoKO . These two types of females will be simply referred to as 'WT' and 'KO' hereafter based on the genotype of Peg3, respectively. At one day postpartum, the mammary glands were harvested from the two females, WT and KO, and the harvested tissues were fixed, and then used for inspecting the spatial expression pattern of Oxtr (Fig 4A). Overall, we observed a higher density of Venus-positive (Venus + ) cells lining the mammary ducts as well as an increase in Venus + cells along the basal region of cross-sectioned mammary ductal structures in KO than in WT samples (Fig 4A). To further follow up this observation, another set of the mammary glands were harvested from one-day-postpartum female littermates with the following genotypes: Oxtr +/+ ; Peg3 +/+ (WT) and Oxtr +/+ ; Peg3 +/CoKO (KO). These mammary glands were used for further isolating mammary epithelial cells (MECs), which were used for isolating total RNA and subsequent cDNA synthesis. This set of cDNA was then used for another series of expression analyses involving qRT-PCR (Fig 4B). As expected, the expression levels of Peg3 were significantly down-regulated in the KO sample compared to the WT sample. This was also the case for another imprinted gene Zim1 (Zinc finger gene 1 imprinted) and Ghrh (Growth hormone-releasing hormone). In the case of Esr1 (Estrogen receptor alpha), there was no difference between WT and KO. In contrast, the expression levels of both Oxtr and Oxt were up-regulated in the KO sample, 1.4-fold and 2.1-fold, respectively. The observed up-regulation of Oxtr appeared to be consistent with the greater density of Venus + signals detected from the KO sample (Fig 4A). Thus, this series of analyses concluded that the expression levels of Oxtr were up-regulated in the mammary gland of the KO sample, particularly in the basal region of mammary epithelial cells.

Mutational effects of Peg3 on the expression of Oxtr in the hypothalamus
We also performed a similar series of expression analyses using the hypothalamus that had been harvested from the same sets of the females with the two genotypes, WT and KO (Fig 5). For this series of analyses, the harvested brains were first fixed, sectioned and finally immunostained with the following antibodies: anti-OXT and anti-PEG3 antibodies. The expression of Oxtr was monitored again through the signal of the reporter Venus. As expected, the expression of Peg3 was not detectable at all throughout the entire area of the hypothalamus in the KO sample. On the other hand, the spatial expression patterns of Oxtr were quite different between WT and KO samples. In the WT sample, the expression of Oxtr was mainly detected from the neuronal cells that are localized immediately adjacent to the 3rd ventricle column. In contrast, Total RNA was isolated from the mammary epithelial cells of female littermates with the following genotypes: Oxtr +/+ ; Peg3 +/+ (WT) and Oxtr +/+ ; Peg3 +/CoKO (KO). These RNA were used for cDNA synthesis, which were subsequently used for qRT-PCR analyses. The expression levels of each gene were compared between WT and KO, and the observed differences were presented with statistical significance ( Ã , p <0.01 and ÃÃ , p < 0.001).
https://doi.org/10.1371/journal.pone.0202476.g004 the expression of Oxtr in the KO sample was detected from the neuronal cells that are localized in the much broader region of the hypothalamus, suggesting that the mutation on Peg3 may result in an ectopic expression of Oxtr within the hypothalamus. A series of sections derived from the different areas of the hypothalamus were also inspected, showing similar ectopic expression of Oxtr (S2 File). In the case of Oxt, we have not detected any significant difference between WT and KO samples in terms of its spatial expression pattern (S3 File).
We also performed a series of qRT-PCR to measure the expression level differences of several genes between Oxtr +/+ ; Peg3 +/+ (WT) and Oxtr +/+ ; Peg3 +/CoKO (KO) samples (Fig 5B). According to the results, the expression levels of Peg3 were not detectable in KO, consistent with the results from previous studies [11,12]. The expression levels of the three genes, including Zim1, Ghrh, and Esr1, showed no major difference between WT and KO samples. On the other hand, the expression levels of Oxtr and Oxt were similarly up-regulated in the KO sample Total RNA was isolated from the hypothalamus of female littermates with the following genotypes: Oxtr +/+ ; Peg3 +/+ (WT) and Oxtr +/+ ; Peg3 +/CoKO (KO). These RNA were used for cDNA synthesis, which were then used for qRT-PCR analyses. The expression levels of each gene were compared between WT and KO, and the observed differences were presented with statistical significance ( Ã , p <0.01 and ÃÃ , p < 0.001).
https://doi.org/10.1371/journal.pone.0202476.g005 compared to those of the WT sample. The observed up-regulation of Oxtr appeared to agree with the ectopic expression pattern detected through the immunostaining (Fig 5A). This series of analyses were repeated three times using three biological replicates. Thus, three independent trials with two different methods confirmed similar up-regulation of Oxtr in the KO sample. Overall, this series of analyses demonstrated that the mutation on Peg3 results in the up-regulation of Oxtr in the hypothalamus, which is also similar to the outcome observed in the mammary gland.

Discussion
In the current study, we characterized the in vivo binding of PEG3 to a genomic target that is located within the 3rd exon of the mouse Oxtr locus. Also, the expression analyses using a combination of two mutant mouse models, Venus and CoKO, further demonstrated that removal of PEG3 resulted in the up-regulation of Oxtr in mammary epithelial cells and in the hypothalamus. Thus, this suggests that Peg3 may play roles as a transcriptional regulator for the expression of Oxtr.
The results presented here demonstrated that PEG3 may directly control the expression of Oxtr, as a DNA-binding protein (Fig 1 and Fig 2). This also agrees with the previous studies that PEG3 functions as a DNA-binding protein controlling the expression of several downstream genes, including Pgm2l1, H19, Msl1 and Msl3 [4][5][6][7]. PEG3 has been predicted to be a transcriptional repressor, since the known downstream genes tend to be up-regulated in the mutant samples lacking PEG3 [6,7]. This turned out to be the case for Oxtr. In the mammary gland, the KO sample displayed 1.4-fold up-regulation of Oxtr based on the results from qRT-PCR, and also exhibited much more dense signals within the location of mammary epithelial cells (Fig 4A). In the hypothalamus, the overall expression levels of Oxtr were also 2-fold up-regulated in the KO sample. Interestingly, the immunostaining showed that the neuronal cells localized in the broader region adjacent to the 3rd ventricle ectopically expressed Oxtr in the KO sample. A similar observation was also observed from the previous study that several placenta-specific gene families are ectopically de-repressed in the hypothalamus of the mutant mice [11]. In that regard, it is also interesting to note that Peg3 is known to be expressed more preferentially in the vasopressinergic neurons than in the oxytocinergic neurons in the hypothalamus of rats [18]. If this is the case, the observed up-regulation of Oxt might also be an outcome of the ectopic expression in the KO model, although we have not been able to confirm this possibility (S3 File). Thus, PEG3 may play a transcriptional repressor role not only for the cells that already express Oxtr, such as the epithelial cells in the mammary gland, but also for the cells that do not usually express Oxtr, as seen in the neuronal cells in the hypothalamus. Overall, the current study clearly demonstrated that PEG3 functions as a transcriptional repressor for Oxtr, and further suggests that this repressor function of Peg3 may be cell type-and context-dependent.
Several mutant models targeting Peg3 are known to have defects in the milk letdown process, and these defects have been, so far, believed to be associated with a loss of OXT protein levels [9][10][11][12]. Yet, the current study indicated that the mutant animals lacking PEG3 have increased levels of Oxtr and Oxt (Fig 4 and Fig 5). It is somewhat enigmatic how the up-regulated oxytocin circuitry might cause defects in the milk letdown process. Two scenarios are possible. First, the relative amount of OXTR and OXT to the other proteins may be optimized and tightly controlled as part of some signaling pathways. In this case, the excessive amount of OXTR could have a diluting effect on the downstream proteins that are involved in these signaling pathways, thus causing a defect in the milk letdown process. Second, it is also feasible to predict that the hormone OXT might be produced just sufficient for the cells participating in the oxytocin circuitry. In this situation, if many unrelated cells happen to also express Oxtr due to its de-repression, this might cause a similar diluting effect on the downstream proteins involved in the signaling pathways. This is feasible, since Oxtr was found to be ectopically expressed in the neuronal cells within the hypothalamus of the KO sample (Fig 5A). Overall, although very speculative at the moment, it should be very interesting to investigate these possibilities in the near future. On a separate note, it is relevant to note that the expression of Oxt and Oxtr in other tissues, such as the gastrointestinal tract, may also be associated with maternal care defects and gastrointestinal motility in rat models [19]. Although the current study has mainly focused on the hypothalamus and mammary gland, potential connection of the oxytocin circuitry to the enteric nervous system may also offer some insight into the defects observed in Peg3-deficient mice.

Ethics statement
All the experiments related to mice were performed in accordance with National Institutes of Health guidelines for care and use of animals, and also approved by the Louisiana State University Institutional Animal Care and Use Committee (IACUC), protocol #16-060.

Chromatin ImmunoPrecipitation (ChIP)
Chromatins were prepared from MEF and adult brains, according to the method previously described [6]. In brief, the homogenized samples were first cross-linked with 1% formaldehyde for 20 minutes, and then lysed with buffer containing protease inhibitor cocktail (Cat. No. 539131, Millipore). The subsequent nuclei were fractionated with sonication to derive a pool of DNA fragments size-ranging from 300 to 500 bp in length. The prepared chromatin was immunoprecipitated with a commercial anti-PEG3 antibody (Cat. No. ab99252, abcam). The immunoprecipitated DNA was dissolved in 100 μl of TE for PCR analyses.

Electrophoretic Mobility Shift Assay (EMSA)
EMSA was performed using a gel shift assay system kit (Cat. No. E3053, Promega). This series of assays used mouse brain nuclear extract (Cat. No. 36053, Active Motif). For the competition assays, competitor oligonucleotide duplexes (1.74 pico mole, 200X) were first incubated with mouse brain nuclear extract (2.72 μg) at room temperature for 10 minutes. Later, the P 32labeled Pgm2l1 oligonucleotide duplex probe (8.7 fento mole, 1X) was added and incubated at room temperature for additional 20 minutes. The reaction mixture was separated on a 5% TBE gel (Cat. No. 456-5014, Bio-Rad), and exposed to a film for 2 to 6 hours at -80˚C.

Isolation of mammary epithelial cells
Female dams were sacrificed using CO 2 asphyxiation at one day postpartum. Mammary glands were extracted without lymph nodes, and placed in a 50 mL conical tube with 10 mL of DMEM/F12 1:1 media. The harvested mammary glands were transferred to petri dishes and minced to a size of~1 mm 3 under a laminar flow hood. The minced tissue was transferred to a new 50 mL conical tube containing 10 mL of DMEM/F12 1:1 supplemented with Collagenase (300 U/mL) and Hyaluronidase (100 U/mL), and placed on a 37˚C shaking incubator at 200 RPM for 2 hours. The mixture was spun down at 2000 RPM for 5 minutes, and the subsequent pellet was recovered from the supernatant containing the lipid layer. The pellet was again resuspended in 10 mL of HBSS. Red blood and stromal cells were depleted with 3 rounds of pulse centrifugations at 1500 RPM for 7 seconds. After the three rounds of wash, the final pellet was resuspended in 10 mL of HBSS, partitioned into two 5 mL aliquots, and spun down again at 2000 RPM for 5 minutes. These two fractions were used for RNA isolation and subsequent cDNA synthesis.

Quantitative qRT-PCR analyses
Total RNA was isolated from the mammary epithelial cells and the hypothalamus of adult mice using a commercial kit (Trizol, Invitrogen). The total RNA was reverse-transcribed using the M-MuLV kit (Invitrogen), and the subsequent cDNA was used as a template for quantitative real-time PCR. This analysis was performed with the iQ SYBR green supermix (Bio-Rad) using the ViiA™ 7 Real-Time PCR System (Life Technologies). All qRT-PCR reactions were carried out for 40 cycles under standard PCR conditions. The analyses of the results from qRT-PCR were described previously [21]. Statistical significance of potential difference of expression levels of a given gene between two samples was tested with Mann-Whitney U test. The information regarding individual primer sequences is available through previous studies [11].
To enhance the signal of Venus, immunocytochemical localization of Venus with antigreen fluorescent protein (GFP) antibody (ab13970; abcam, Cambridge, UK) was conducted in the brain sections. The free-floating brain slices were incubated with the primary antibodies against GFP at dilutions of 1:10,000 in PBS containing 0.5% Triton X-100 (PBST) with continuous gentle agitation at 4˚C for overnight. The brain sections were subsequently incubated with a secondary antibody (goat anti-chicken) conjugated with Alexa Fluor 488 (Jackson ImmunoResearch, West Grove, PA) at room temperature for 4 hours. The sections were also incubated with anti-PEG3 at dilutions of 1:10,000 (custom-made antibody, [6]) and monoclonal anti-OT-neurophysin antibody (PS38 provided by H. Gainer, NIH) at 1:500 overnight at 4˚C, and labeled with AffiniPure Goat Anti-Rabbit IgG Alexa Fluor647 and AfiiniPure Goat Anti-Mouse IgG DyLight549 respectively (Jackson ImmunoResearch, West Grove, PA) at 1:250 dilution at room temperature for 2-4 hours. The sections were mounted in polyvinyl alcohol (PVA) with anti-fading agent 1,4-Diazabicyclo[2.2.2]octane (DABCO) that consists of 4.8 g PVA, 12 g glycerol, 12 ml dH 2 O, 24 ml 0.2 M Tris-HCl, and 1.25 g DABCO. Fluorescence microscopic images (1280 x 1024) were acquired digitally (Eclipse 80i equipped with DS-QiMc, Nikon, Tokyo, Japan).