Tissue-specific changes in Srebf1 and Srebf2 expression and DNA methylation with perinatal phthalate exposure

Abstract Perinatal exposure to endocrine disrupting chemicals negatively impacts health, but the mechanism by which such toxicants damage long-term reproductive and metabolic function is unknown. Lipid metabolism plays a pivotal role in steroid hormone synthesis as well as energy utilization and storage; thus, aberrant lipid regulation may contribute to phthalate-driven health impairments. In order to test this hypothesis, we specifically examined epigenetic disruptions in lipid metabolism pathways after perinatal phthalate exposure. During gestation and lactation, pregnant Long–Evans rat dams were fed environmentally relevant doses of phthalate mixture: 0 (CON), 200 (LO), or 1000 (HI) µg/kg body weight/day. On PND90, male offspring in the LO and HI groups had higher body weights than CON rats. Gene expression of lipid metabolism pathways was altered in testis and adipose tissue of males exposed to the HI phthalate dosage. Specifically, Srebf1 was downregulated in testis and Srebf2 was upregulated in adipose tissue. In testis of HI rats, DNA methylation was increased at two loci and reduced at one other site surrounding Srebf1 transcription start site. In adipose tissue of HI rats, we observed increased DNA methylation at one region within the first intron of Srebf2. Computational analysis revealed several potential transcriptional regulator binding sites, suggesting functional relevance of the identified differentially methylated CpGs. Overall, we show that perinatal phthalate exposure affects lipid metabolism gene expression in a tissue-specific manner possibly through altering DNA methylation of Srebf1 and Srebf2.


Introduction
Phthalates are environmental toxicants that have endocrine disrupting effects. Phthalates can be found in many household plastic and cosmetic products [1,2] and as contaminants in foods such as poultry and dairy [3]. Due to phthalate influence on hormone balance, epidemiological studies have extensively examined the relationship between phthalates and male reproductive health. Phthalate exposure in adulthood has been associated with reduced sperm quality and testosterone levels [4][5][6], while prenatal phthalate exposure results in shorter anogenital distance in infants [7]. Several animal models have confirmed and expanded upon these observations. Early-life exposure has been of particular interest due to heightened sensitivity during the critical period and the potential to impact long-term health outcomes. In addition to validating findings in human studies [8,9], rodent models have shown that perinatal phthalate exposure disrupts Leydig [10,11] and Sertoli [12,13] cell populations as well as induces aberrant DNA methylation [14,15] and histone modifications [10,16] in the testis.
Additionally, studies have quantified metabolic outcomes after phthalate exposure. While some have suggested a positive association between early-life phthalate exposure and body mass index (BMI) [17,18], others have found no correlation [19] or a negative correlation between perinatal phthalates, BMI, waist circumference, and body fat percentage [20,21]. Similar inconsistencies have been reported in animal models. Indeed, early-life phthalate exposure in rodents has been shown to have positive [22,23], negative [24][25][26], and no relationship [27,28] with body weight. There remains no clear consensus as to whether early-life phthalate exposure produces sustained metabolic complications. Thus, understanding the possible mechanisms by which phthalates affect reproduction and metabolism is critical in clarifying the longterm health implications of environmental exposure.
Proper metabolic and reproductive function involves appropriate regulation of lipid synthesis, catabolism, and storage. Obesity is characterized by excessive fat storage and hyperlipidemia. As for reproductive health, not only does adiposity impact endocrine function, but cholesterol itself is also a precursor of androgen and estrogen. Previous experimentation in animal and in vitro models suggests that phthalates may induce lipid accumulation by upregulating expression of fatty acid synthesis and lipid transport genes [28][29][30][31][32][33][34]. Such metabolic processes are highly regulated and involve a number of transcription factors. Given their widespread regulatory role in lipid metabolism, sterol regulatory element binding proteins (SREBPs) have been hypothesized to underlie phthalate-induced metabolic dysregulation [35,36]. When cellular sterol levels are low, SREBP cleavage-activating protein (SCAP) escorts SREBP from the endoplasmic reticulum to the Golgi apparatus, where it is cleaved and released by Site-1 protease and Site-2 protease [37]. SREBP then translocates to the nucleus and can bind sterol response elements to induce transcription of genes such as fatty acid synthase (Fasn), acetyl CoA carboxylase (Acc), and HMG-CoA synthase (Hmgcs) [38,39]. SREBP1 upregulates lipogenesis and triglyceride synthesis while SREBP2 promotes cholesterol uptake and synthesis. Thus, SREBPs are critical in responding to dietary conditions and maintaining metabolic health across tissues.
We sought to uncover the impact of perinatal phthalate exposure on lipid metabolism in key metabolic and reproductive tissues. We hypothesized that aberrant DNA methylation and elevated expression of SREBP genes may underlie the metabolic and reproductive abnormalities caused by phthalates. In the current experiment, we asked whether perinatal phthalate exposure impacts lipid metabolism in adult offspring. Long-Evans rat dams consumed environmentally relevant doses of phthalates during gestation and lactation. Male offspring were sacrificed at postnatal day (PND) 90 and gene expression and DNA methylation were measured in testis, adipose, and liver. Overall we show that phthalate exposure drives tissue-specific changes in gene expression and potentially operates through an epigenetic mechanism.

Body Weight and Composition
Pregnant Long-Evans rat dams were fed one of three phthalate dosages from gestational day 2 until PND10: 0 [control (CON)], 200 [low (LO)], or 1000 [high (HI)] mg/kg of body weight/day. Body weights of male offspring were measured at PND10 and 90. At PND10, the LO group had significantly lower body weights than controls (P ¼ 0.015; Fig. 1A). However, at PND90, this trend was reversed, as both LO and HI animals had higher body weights than control (P ¼ 0.0080 and P ¼ 0.070, respectively; Fig 1B). Additionally, body fat and lean mass were measured on PND90. LO rats had slightly lower and HI rats had slightly higher average body fat percentage than CON, however neither of these were significant (P ¼ 0.29 and P ¼ 0.31, respectively; Fig. 1C).

Gene Expression
On PND90, animals were sacrificed and testis, gonadal adipose, and liver were removed for analysis. Due to the endocrine disrupting properties of phthalates and the important role of lipids in hormone synthesis, we measured genes involved in tissue-specific as well as general lipid metabolic processes. In testis, we measured cholesterol and hormone synthesis, lipogenesis, and transcriptional regulation genes. In phthalateexposed animals, we found a decrease in Niemann-Pick intracellular cholesterol transporter 1 (Npc1; P ¼ 0.042), fatty acid synthase (Fasn; P ¼ 0.012), and Srebf1 (P ¼ 0.027; Table 1). These differences were only observed in the HI group. Moreover, we saw a slight decrease in glycerol-3-phosphate acyltransferase (Gpam; P ¼ 0.076) and stearoyl-CoA desaturase (Scd; P ¼ 0.084). In the LO group, there was a non-significant increase in Srebf2 expression (P ¼ 0.065).
In adipose tissue, we measured an increase in Fasn (P ¼ 0.040) and Srebf2 (P ¼ 0.031; Table 2). We found slight increases in expression of cholesterol-regulating gene, insulin induced gene 1 (Insig1; P ¼ 0.061) as well as Scd (P ¼ 0.073). In liver, only minimal gene expression differences were observed (Table 3). Scd was upregulated in HI group, but this was not statistically significant (P ¼ 0.066).

DNA Methylation
Next, we tested the hypothesis that differences in gene expression may be caused by phthalate-mediated alterations in DNA methylation. We identified several differentially expressed lipid metabolism genes, but due to their potential for widespread transcriptional regulation, we only measured DNA methylation associated with Srebf1 and Srebf2. MSP was performed for several sites surrounding the transcription start site (TSS) of the two genes. The HI phthalate had reduced testicular Srebf1 expression, whereas in adipose and liver, Srebf1 expression was unchanged ( Fig. 2A). The sequence of Srebf1 includes one cytosine phosphate guanine dinucleotides (CpG) island straddling the TSS (Fig. 2B). We measured DNA methylation of three CpG loci located in islands and one locus in either shore. In testis, we detected increased DNA methylation in the HI animals within the upstream and downstream shores (P ¼ 0.061 and P ¼ 0.048, respectively; Fig. 2C). Additionally, there was a decrease in methylation in the HI group in the island CpGs located 295 bp downstream of the TSS (P ¼ 0.080). There were not DNA methylation differences in adipose tissue (Fig. 2D). In liver, there was increased DNA methylation in the LO group in the island CpGs 295 bp downstream of the TSS (P ¼ 0.011; Fig. 2E).
DNA methylation changes around Srebf2 were minimal. Gene expression in adipose tissue was significantly increased in the HI group. In testis, there was a slight increase in the LO group (Fig. 3A). Surrounding the Srebf2 TSS, there is one large CpG island (Fig. 3B). Another CpG island is located $500 bp downstream. In testis, there was a non-significant decrease in DNA methylation in the LO group at CpGs located 78 bp   upstream of the TSS (P ¼ 0.058; Fig. 3A). The same locus was differentially methylated in the liver, where it was hypermethylated in the HI group (P ¼ 0.0021; Fig. 3E). In adipose, there was increased DNA methylation at an intronic site located 368 bp downstream of the TSS and within the CpG island (P ¼ 0.048; Fig. 3D).

Transcription Factor Prediction
Finally, we attempted to predict whether the identified differentially methylated regions might impact transcription and expression of Srebf1 and Srebf2. We performed a bioinformatics analysis of the differentially methylated regions using PROMO. First, we looked at three regions around Srebf1 that were differentially methylated in the testis. Within the site 130-100 bp upstream of Srebf1, we found potential binding sites for SMAD3, ELK1, E2F, DP1, and STAT4 (Fig. 4A). Within the region 280-310 bp downstream of the Srebf1 TSS, there was a potential binding site for nuclear factor I/CAAT box transcription factor (NFI/CTF; Fig. 4B). Within the region 355-400 bp downstream of the Srebf1 TSS, we identified putative binding sites for VDR, MYB, and POU3F2 (Fig. 4C). In Srebf2, we examined the region 355-395 bp downstream of the TSS and found sites for ELK1, VDR, and MYB binding (Fig. 4D).

Discussion
In this paper, we have demonstrated the long-lasting effect of endocrine disruptors on lipid metabolism. During the perinatal period, Long-Evans rats were administered a mixture of phthalates at concentrations observed in human populations. Phthalates altered early postnatal and adolescent body weight but did not alter fat deposition. Gene expression analysis at PND90 revealed minor alterations in hepatic lipid metabolism genes, but significant changes in cholesterol and fatty acid synthesis in gonadal adipose tissue. Finally, high doses of perinatal phthalates significantly altered cholesterol trafficking and fatty acid metabolism genes in testis. In particular the transcriptional regulator Srebf1 was downregulated, possibly through hypermethylation of its promoter and coding region. Early-life phthalate exposure might interfere with adipose and testicular fatty acid and cholesterol synthesis through epigenetic mechanisms, and thus alter reproductive function.

Phthalates and Lipid Metabolism
Lipid metabolism is an integral part of reproductive and general metabolic health. Through highly regulated pathways, fatty acids, triglycerides, and cholesterol are synthesized in response to environmental and homeostatic stimuli in order to meet metabolic demands (Fig. 5). Such processes are transcriptionally regulated by factors including SREBP1, which upregulates fatty acid and triglyceride synthesis; SREBP2, which promotes cholesterol uptake and synthesis; and LXR, which opposes the effects of SREBP2 and activates SREBP1 (Fig. 5). In testis, cholesterol plays a particularly critical role as a precursor of testosterone (Fig. 5). Previous investigation has suggested that exposure to individual phthalates impacts lipid metabolism. In Caenorhabditis elegans, life-long exposure to phthalates DEHP and DEP increased fat accumulation along with survival and reproductive function, possibly related to the upregulation of fatty acid synthesis (fasn-1, pod-2, fat-5, elo-2, acs-6, sbp-1) and lipid transport genes (vit-2, -4, -5, and -6) [29]. In zebrafish, water-exposure to diisononyl phthalate (DiNP) activated orexigenic signals in the brain and steatohepatitis in adult animals [30]. In human cells, exposure to monoethylhexyl phthalate (MEHP) affected adipocyte size, possibly via increased lipolysis, glucose uptake, and oxygen consumption [31]. In placental cells MEHP produced lipidome-wide changes with specific alterations in glycerolipids and glycerophospholipids and promoted triacylglycerols accumulation [34]. Interestingly, prenatal DEHP exposure in rats was shown to target adrenal gland peroxisome proliferatoractivated receptor (PPAR) and cholesterol biosynthesis pathways [33], which rendered offspring more susceptibility to second-hit stressor challenge [32]. Similarly, in utero DEHP exposure has been shown to potentiate the effect of a postnatal high fat-diet [28]. In our study, we used a phthalate mixture to mimic human environmental exposure. However just as individual phthalates induce transcriptional changes in lipid metabolism genes in different tissues, we also observed specific adipose tissue and testicular gene expression changes in cholesterol and fatty acid synthesis pathways.
Lipid metabolism genes that are affected by phthalates vary among tissues, and depend on the window of exposure. In our paper, prenatal exposure to low or high doses of phthalates produced no significant effects in liver. This finding contradicts previous studies that have shown phthalate-mediated changes in hepatic morphology and Ppara gene expression [40,41]. This difference might be due to the low phthalate dose used in our study (1000Â lower). Furthermore, the lack of gene expression changes in liver might not be reflective of the ability of the liver to handle fatty acids and cholesterol, given that the animals were exposed to neither a high fat-diet nor a second stressor later in life.
In adipose tissue, we showed a significant increase in cholesterol, fatty acid, and lipid metabolism-related genes after phthalate treatment. A trend toward increased Insig1 expression might speak to altered cholesterol signaling, which is vital for the production of hormones within adipose tissue. Importantly, Srebf2 (Srebp2) was significantly upregulated by a high dose of phthalates, suggesting higher cholesterol synthesis as its main target genes include Hmgcr and Hmgcs, key enzymes for cholesterol biosynthesis [37,42]. Moreover, increases in Fasn and Scd indicate a prominent effect on lipid synthesis with high phthalate doses, possibly affecting the ability of adipocytes to accumulate fat.
Lastly, a high dosage of perinatal phthalates decreased expression of genes involved with cholesterol, fatty acid, and triglyceride metabolism, namely Npc1, Fasn, Gpam, and Srebf1. Dysregulation of Srebf1 (Srepb1) is known to affect both fatty acid and cholesterol metabolism via the two known isoforms of SREBP protein 1a and 1c [43]. Indeed, Npc1, Fasn, Gpam are transcriptional targets of Srebp1 [38,44,45]. This suggests that the phthalate-induced decrease in expression of these genes may be a result of decreased SREBP1 binding. Previous studies have shown that exposure to DiBP (600 mg/kg BW/day) caused dysregulation of testicular fatty acid and cholesterol-metabolism genes, which impacts steroidogenesis and reproductive function [41]. Studies have shown that androgen production does not require the accumulation of SREBP2 protein [46], thus highlighting the effect of Srebf1 in regulating fatty acid and cholesterol metabolism, and in turn modulating the hormone production of testis and reproductive function.

Phthalates and DNA Methylation
In this study, we provide evidence that early-life phthalate exposure induces different DNA methylation patterns around the Srebf1 and Srebf2 promoters. Several others have highlighted the epigenetic implications of phthalates. In rat testis, maternal phthalate exposure resulted in hypermethylation of steroidogenic factor-1 (SF-1) and specific protein-1 (Sp-1) [14] and hypomethylation of the mineralocorticoid receptor [47]. Phthalates also induced decreased global DNA hydroxymethylation [48] and increased global DNA methylation [49,50]. In adipose tissue, we have previously observed decreased methylation in the adipogenesis gene, frizzled 1 (Fzd1), and the triglyceride cleaving enzyme, lipoprotein lipase (Lpl) [51]. Phthalate administration in rodent models also resulted in hepatic hypomethylation of the oncogene, c-Myc [52,53], while another study reported no change in global DNA methylation [54]. Our results add to the literature by demonstrating that phthalates alter DNA methylation in a tissue-specific manner.
Our findings support the idea that DNA methylation in the promoter prevents transcription factor binding and induces a closed chromatin confirmation in order to repress transcription. We found three regions around the Srebf1 TSS that were differentially methylated in the testis. Two of the three were located in CpG shores and were hypermethylated after phthalate exposure. This finding is consistent with previous literature showing that CpG shores are more vulnerable to differential methylation than CpG islands [55,56]. Additionally, Srebf1 gene expression was downregulated in testis; thus, an increase in DNA methylation follows the canonical assumption that DNA methylation blocks transcription. Transcription factor analysis identified SMAD3, ELK1, E2F, DP1, and STAT4 binding sites within the upstream shore (130-100 bp upstream of Srebf1 TSS). Each of these factors has been shown to activate transcription [57][58][59][60][61]. Within the downstream shore (355-4000 bp downstream of the Srebf1 TSS), there was a potential binding site for VDR, MYB, and POU3F2, which have also been demonstrated to have activational properties [62][63][64].
We identified two differentially methylated CpG loci that showed a positive association between DNA methylation and gene expression. First, in testis we observed decreased DNA methylation at the CpG site 295 bp downstream of the Srebf1 TSS. This decrease was associated with decreased gene expression. Upon examination of the transcription factor binding sites, we saw that relatively few factors bind to the region. NFI/CTF may activate or repress transcription depending on recruitment of other cofactors and chromatin modifiers [65]. However, the same locus was highly methylated in the liver, despite no change in gene expression. Thus, it may be the case that DNA methylation at this site is not critical for the transcriptional regulation of Srebf1. Secondly, in adipose tissue we observed increased Srebf2 expression accompanied by increased DNA methylation at the region 355-395 bp downstream of the TSS. Binding sites for ELK1, VDR, and MYB were found. Although much of the literature points to the activating role of these transcription factors, there is evidence that ELK1 and MYB display repressive characteristics in certain contexts [66][67][68]. It is possible that increased methylation could increase transcription by inhibiting repressor binding.

Conclusion
Here we demonstrate the tissue-specific implications of perinatal phthalate exposure. We are among the first to investigate phthalate-induced DNA methylation across tissues; however, our experiments should be viewed in light of a few limitations. First, our sample size was relatively small (n ¼ 8-9/group). This might be one reason that we found several genes and CpGs that were slightly but not statistically altered with phthalate treatment. Our data can be used to generate hypotheses and should be replicated in a larger cohort. Secondly, while we measured body weight and body composition as metabolic outcomes, we did not quantify reproductive outcomes beyond changes in gene expression. Future experimentation should focus on discovering associations between DNA methylation at the identified loci and physiological outcomes such as sperm count and quality as well as other fertility measures. Additionally, we only looked at a limited number of CpG sites using MSP.
To gain a more comprehensive picture of phthalate-mediated DNA methylation, genome-wide methods such as bisulfite sequencing or methylated DNA immunoprecipitation sequencing (MeDIP-seq) may be utilized. Finally, future experiments can confirm the impact of DNA methylation and TF binding within the identified loci. We measured differential methylation at CpGs within the Srebf1 and Srebf2 promoters, but it is unclear whether these differences directly impact transcription or whether they are consequences of gene expression changes. Moreover, we report several computationally predicted TF binding sites around Srebf1 and Srebf2, however these findings should be validated for transcriptional relevance through appropriate in vitro assays.
Overall, we show that perinatal phthalate treatment induces distinct gene expression changes in lipid metabolism pathways in testis and adipose tissue but not in the liver. Importantly, expression and DNA methylation of Srebf1 and Srebf2 were altered by phthalates. We propose that perinatal phthalate exposure alters DNA methylation around lipid-related transcription factors, changes the expression of their gene targets, and alters lipid metabolism. This study provides insight into the mechanism by which phthalates might disrupt reproductive and general metabolic outcomes.

Animals
Three-month old Long-Evans hooded female rats (n ¼ 8-9 per group) were obtained from Harlan Laboratories (Indianapolis, IN). Before breeding, rats were housed in same-sex pairs and were given ad libitum access to a low phytoestrogen diet (Harlan 2020X; Taklad Diets, Madison, WI). During breeding, dams were placed in suspended wire-bottom cages, and once a sperm plug was detected, dams were removed and individually housed. From gestational day 2 until PND10, dams received one of three phthalate treatments, 0 (CON), 200 (LO), or 1000 mg/kg body weight/day (HI).
The phthalate mixture was designed to reflect exposure levels found in pregnant women in Champaign-Urbana, Illinois (unpublished data) as well as exposure levels in the general US population [69]. The mixture consisted of $35% diethyl (DEP), 21% bis(2-ethylhexyl) (DEHP), 15% dibutyl (DBP), 15% diisononyl (DiNP), 8% diisobutyl (DiBP), and 5% benzyl butyl (BBP) phthalate. The phthalates were suspended in tocopherol-stripped corn oil at a concentration of 0, 0.6, or 3 mg phthalates/ml for the 0, 200, or 1000 mg phthalates/kg body weight/day doses, respectively. The phthalate mixture was administered orally on top of half a cookie (Newman's Own organic alphabet cookie, vanilla flavor). Dams were acclimatized for 2 days with half a cookie topped with tocopherol-stripped corn oil, and from gestational day 2 through PND10, dams voluntarily ate the half cookie and phthalate mixture. Additionally, dams were given ad libitum access a control diet (15.8% kcal fat, 63.9% kcal carbohydrate, 20.3% kcal protein; D10012G, Research Diets Inc., New Brunswick, NJ).
On PND25, one male offspring from each dam was randomly selected and moved to its own individual cage (CON: n ¼ 8, LO: n ¼ 9, HI: n ¼ 9). Offspring continued on the D10012G diet and received no phthalates for the remaining duration of the experiment. On PND90, body composition was measured using the EchoMRI-700 Body Composition Analyzer (Echo Medical Systems, Houston, TX). The rats were then sacrificed and liver, testis, and gonadal adipose tissue was dissected out. All tissues were snap-frozen in liquid nitrogen and stored at À80 C for genetic analysis. All procedures were approved by the University of Illinois Institutional Care and Use Committee and adhere to the National Institutes of Health guidelines on the ethical use of animals.

Gene Expression
Frozen tissue was ground in liquid nitrogen and placed in a 1.5 ml tube containing Trizol. Before loading the adipose sample into the column, the tube was spun and the lipid layer was removed. Total RNA was extracted using Direct-zol RNA MiniPrep Plus (Zymo Research, Irvine, CA) with in-column DNase I treatment to eliminate DNA contamination. cDNA synthesis and qRT-PCR were performed as previously described [70]. Primers were designed (VectorNTI software; Life Technologies, Grand Island, NY), validated for specificity (Basic Local Alignment Search Tool; BLAST; https://blast.ncbi.nlm.nih.gov/Blast.cgi), and synthesized (IDT; Coralville, IA). All mRNA primer sequences and amplification efficiencies can be found in Supplementary Table S1.

DNA Methylation
Genomic DNA was isolated (ZR Genomic DNA Tissue MiniPrep kit; Zymo Research) and bisulfite converted (EZ DNA Methylation Gold Kit; Zymo Research). The EZ DNA Methylation Gold Kit was chosen for its high bisulfite conversion efficiency, which has repeatedly been reported to range between 99 and 100% [71][72][73][74]. Thus, we can confidently assume that methylation differences are likely not artifacts of different conversion efficiencies. To perform the conversion reaction, samples were heated to 98 C for 10 min, 64 C for 2.5 h, and held at 4 C. Samples were desulphonated and diluted to 5 ng/ll for methylation specific PCR (MSP) using Sybr Green (Quanta Biosciences, Beverly, MA). Quantitative real-time PCR was performed using the StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, CA). CT values were quantified based on a serially diluted standard curve and percent methylation was calculated by dividing quantity of methylated DNA by the sum of methylated and unmethylated DNA. Methylation was normalized to the CON group.
Methylated and unmethylated primer pairs were manually designed (Primer Express 3.0.1 software's Primer Probe Test Tool; Life Technologies). The T m for each primer was between 68 and 70 C and the CG content was between 30 and 80%. Primers were validated in the IDT OligoAnalyzer to minimize dimers and hairpin loops. Unmethylated DNA was measured using both unmethylated forward and reverse primers. Methylated DNA was measured by pairing one methylated primer with the corresponding unmethylated primer. All MSP primer sequences and amplification efficiencies can be found in Supplementary Table S2.

Transcription Factor and Statistical Analysis
Computational prediction of transcription factor binding was performed with PROMO version 3.0.2 [75,76]. Identified transcription factor sequences were validated with the Jaspar Database [77]. One way ANOVA followed by pairwise comparisons between CON vs LO and CON vs HI was used to determine statistical significance for body weight, body composition, gene expression, and DNA methylation. Statistical analysis was performed in R version 3.3.