Objective
Data from animal models show that in utero exposure to a maternal high-fat diet (HFD) renders susceptibility of these offspring to the adult onset of metabolic syndrome. We and others have previously shown that epigenetic modifications to histones may serve as a molecular memory of the in utero exposure, rendering the risk of adult disease. Because mice heterozygous for the Glut4 gene (insulin sensitive glucose transporter) born to wild-type (WT) mothers demonstrate exacterbated metabolic syndrome when exposed to an HFD in utero, we sought to analyze the genome-wide epigenetic changes that occur in the fetal liver in susceptible offspring.
Study Design
WT and Glut4 +/− (G4 +/− ) offspring of WT mothers that were exposed either to a control or an HFD in utero were studied. Immunoblotting was used to measure hepatic histone modifications of fetal and 5-week animals. Chromatin immunoprecipitation (ChIP) followed by hybridization to chip arrays (ChIP-on-chip) was used to detect genome-wide changes of histone modifications with HFD exposure.
Results
We found that levels of hepatic H3K14ac and H3K9me3 significantly increased with HFD exposure in WT and G4 +/− fetal and 5-week offspring. Pathway analysis of our ChIP-on-chip data revealed differential H3K14ac and H3K9me3 enrichment along pathways that regulate lipid metabolism, specifically in the promoter regions of Pparg, Ppara, Rxra, and Rora .
Conclusion
We conclude that HFD exposure in utero is associated with functional alterations to fetal hepatic histone modifications in both WT and G4 +/− offspring, some of which persist up to 5 weeks of age.
According to the developmental origins of health and disease hypothesis, the in utero experience can have a profound effect on the individual. Studies have suggested that the effects of a suboptimal intrauterine milieu can persist into adulthood. In utero exposure to either a maternal low-protein diet, caloric restriction, or a maternal high-fat diet (HFD) is associated with an increased susceptibility to the adult onset of metabolic syndrome. In the current era of obesity, studies concerning how a mother’s HFD may influence the health of her offspring are of increasing relevance. HFD consumption during pregnancy is associated with gestational diabetes mellitus. Animal models of in utero HFD exposure have shown that offspring are more susceptible to fatty liver in early life and increased adiposity, diabetes mellitus, and cardiovascular disease in adulthood. The question therefore remains, how can the memory of an exposure only experienced during gestation be maintained over the lifetime of the individual?
The possibility that epigenetic modifications contribute to this memory is intriguing. Epigenetic modifications constitute changes to the local chromatin structure that do not change the underlying DNA sequence. The addition or removal of posttranslational histone modifications alongside changes in DNA methylation patterns are potential mechanisms that could contribute to the memory of an in utero exposure. Some histone modifications are enriched within the promoters of transcriptionally active genes, such as acetylation of lysine 14 of histone H3 (H3K14ac) ; other modifications such as trimethylation of lysine 9 of histone H3 (H3K9me3) are enriched in promoters of repressed genes as well as within heterochromatin. It is also well established that activating and repressive marks are not mutually exclusive, even within the same promoter. During embryogenesis, chromatin domains that contain both repressive and activating motifs cluster throughout the genome. Our previous work in a nonhuman primate model has demonstrated that HFD exposure in utero alters the fetal hepatic histone code. Specifically, hepatic H3K14ac is increased in the HFD-exposed fetal animals. H3K14 appears particularly sensitive to the intrauterine milieu because it is also modified in a rat model of nutrient restriction in skeletal muscle. How these modifications are established and their relationship to the susceptibility to the adult onset of disease remains to be investigated fully.
Not every animal exposed to an HFD is equally susceptible, begging the question as to whether the offspring genotype may serve as a modifier of the in utero environment. We have studied offspring heterozygous for the Glut4 gene (G4 +/− ) from wild-type mothers that were fed an HFD. Glut4 haploinsufficiency results in peripheral insulin resistance, altered lipid metabolism, and type 2 diabetes mellitus. It has been shown that exposure to an HFD during critical periods of development leads to development of metabolic syndrome such as increased adiposity, impaired glucose tolerance, and insulin insensitivity in G4 +/− and wild-type offspring. Interestingly, genotype-dependent differences were observed, which suggests that haploinsufficiency of Glut4 may result in a different metabolic remodeling in response to the HFD. Thus, it is possible that an interaction between the in utero environment (exposure to a maternal HFD) and the offspring genotype (specifically a heterozygous deletion of Glut4 ) may lead to different epigenetic changes that either protect against or increase the risk of the development of metabolic disease.
Based on our findings of an altered hepatic epigenome in a nonhuman primate model of maternal HFD consumption, we sought to determine (1) whether an increase in hepatic acetylation is observed similarly in a murine model of maternal HFD exposure and (2) whether offspring that are susceptible genetically to metabolic syndrome (the G4 +/− offspring) have a similarly altered hepatic epigenome with HFD exposure in order to study whether diet × genotype interactions may also contribute to offspring disease susceptibility. Because the paternal germline is the source of the Glut4 haploinsufficiency (and Glut4 is not expressed in the liver until late postnatal life ), any intrauterine genotypic contribution would be attributable only to altered placental glucose and/or nutrient uptake in G4 +/− offspring.
In this study, we found that fetal hepatic H3K14ac and H3K9me3 increase with maternal HFD exposure. These alterations were also observed in animals at 5 weeks of age. Chromatin immunoprecipitation (ChIP) followed by hybridization to a promoter array chip (ChIP-on-chip) was used to determine which promoters on a genome-wide scale show differential enrichment for H3K14ac and H3K9me3 in response to maternal diet. We observed that these modifications are enriched predominantly among those gene promoters that relegate to lipid metabolism networks. We conclude that HFD exposure in fetal life is associated with significant alterations of distinct histone modifications that render enriched occupancy in the promoters of genes that regulate lipid metabolism.
Materials and Methods
Murine model
All animal procedures were done in accordance with approved institutional review board protocols from both Baylor College of Medicine and Albert Einstein College of Medicine as previously described. Wild-type CD1 female mice were maintained on a control breeding chow (PicoLab Mouse Diet #5058; 9% fat, 20% protein, 53% carbohydrate; Lab Diet, Brentwood, MO) or high-fat (Product #F3282; 35.5% fat as lard, 20% protein, 36.3% carbohydrate; Bio-Serv, Frenchtown, NJ) diet 2 weeks before mating with G4 +/− males throughout gestation and lactation. Offspring were weaned onto a low-fat diet (Pico Lab Mouse Diet #5053; 4.5% fat, 20% protein, 54.8% carbohydrate; Lab Diet) at postnatal day 21. The animals that were used in this study were male wild-type and G4 +/− offspring. For fetal tissue, pregnant mice were killed at embryonic day 18.5. Fetuses were killed by decapitation, and the organs were harvested immediately and snap frozen with liquid nitrogen.
Experimental methods can be found in the Appendix ( supplemental information ).
Results
Hepatic H3K14 acetylation and H3K9 trimethylation increase with HFD exposure in utero and during lactation
Immunoblotting was used to determine whether hepatic histone modifications are altered in either wild-type or G4 +/− offspring with HFD exposure ( Figure 1 , A). In fetal liver at embryonic day 18.5, H3K14ac is increased in both the wild-type (3.6-fold; P = .002) and G4 +/− (3.0-fold; P = .002) offspring with HFD exposure ( Figure 1 , B). H3K9me3 is also increased with HFD exposure in both the wild-type (5.7-fold; P = .007) and G4 +/− (4.6-fold; P = .047) fetal offspring ( Figure 1 , C).
Immunoblotting was performed similarly on livers from 5-week-old animals that were exposed in utero to a maternal control or HFD and weaned onto a low-fat diet 2 weeks before the tissue was harvested. In these animals, both H3K14ac and H3K9me3 are significantly increased in HFD-exposed animals compared with control diet in both wild-type and G4 +/− offspring ( Figure 1 , D and E). Relative levels of fetal hepatic H3K27me3, H4K20me3, H3K9ac, H3K18ac, and H3K4me3 were found to be unchanged with HFD exposure in the fetal wild-type animals, so they were not further assessed ( Supplementary Table 1 ).
Hepatic gene expression of histone-modifying enzymes is reduced in wild-type 5-week-old offspring with HFD exposure during lactation
Because of the observed increase in histone acetylation in the fetal and 5-week-old animals, we hypothesized that expression levels of histone acetyltransferases would be altered. GCN5 is a histone acetyltransferase of histone H3. HDAC1, HDAC3, and SIRT1 deacetylate histone H3K14. Hepatic messenger RNA levels of these genes were measured in both fetal and 5-week-old animals in the wild-type and G4 +/− offspring. Although we failed to observe a significant alteration in expression in the fetal animals ( Figure 2 , A) at 5 weeks of age, Sirt1 expression was decreased significantly when compared with the control diet-exposed cohort in both wild-type and G4 +/− animals ( Figure 2 , B). In wild-type offspring, gene expression of Gcn5 , Hdac1, and Hdac3 were also significantly reduced in 5-week-old offspring that experienced both prenatal and postnatal HFD exposure ( Figure 2 , B).
ChIP-on-chip reveals that global H3K14ac and H3K9me3 are both enriched at the transcription start site, regardless of genotype or diet exposure in the fetal liver
To determine the localization of H3K14ac and H3K9me3, ChIP-on-chip was performed on the fetal liver from wild-type and G4 +/− offspring that were exposed to either control or HFD in utero ( Figure 3 , A). Enrichment (as measured by log 2 immunoprecipitation [IP]/input) of these modifications was plotted throughout the array as a means of internal validation. Consistent with previous epigenome-wide characterizations, we observed robust enrichment in a broad region surrounding the transcription start site (TSS) in each of the 8 groups that were studied ( Figure 3 , B and C).
H3K14ac and H3K9me3 are differentially enriched in genes that are involved in lipid metabolism in the fetal liver
To determine promoter-specific changes of H3K14ac and H3K9me3 in the fetal animals, gene lists were generated from the ChIP-on-chip data that demonstrated differential enrichment over input ( Supplementary Tables 2-9 ). We then determined which genes had differential enrichment between the control and HFD groups ( Supplementary Tables 10-13 ). Of interest were the genes that were altered significantly for both H3K14ac and H3K9me3: 454 genes in the wild-type animals and 755 genes in the G4 +/− animals ( Figure 3 , D). To determine whether there is an offspring genotype effect on histone modification localization, the generated gene lists were analyzed for overlap for each modification between wild-type and G4 +/− offspring ( Figure 3 , E). When comparing genes that were enriched differentially between control and HFD for H3K14ac, 8% of genes (427/5182 total) were enriched similarly between wild-type and G4 +/− liver; 10% of the genes (566/5588 total) that were enriched for H3K9me3 were found in both wild-type and G4 +/− liver.
For the genomic regions that were enriched for H3K14ac or H3K9me3 between control and HFD-exposed animals ( Supplementary Tables 10-13 ), HOMER, the motif discovery software, was used to determine which transcription factor binding sites were significantly (Benjamini q-value < 0.05) represented in the dataset. In the wild-type animals, there were no known motifs that were enriched in either the H3K14ac or H3K9me3 datasets. However, analysis of the G4 +/− offspring revealed 9 significantly enriched motifs ( Figure 4 ). H3K14ac is enriched differentially in regions that contain Gata1, 2, and 4 and Myf5 binding motifs. H3K9me3 is enriched differentially in regions that contain GFY, E2F, E2F4, and RUNX-AML binding motifs.
Genes important for lipid metabolism are enriched differentially in H3K14ac and H3K9me3 in fetal livers of wild-type and G4 ± offspring
The generated gene lists were analyzed with ingenuity pathway analysis (IPA) to determine which biologic networks are represented differentially. For each group, lipid metabolism was the top network to be identified ( Supplementary Table 14 ). In each analysis, 4 genes that were involved in lipid metabolism consistently emerged as central convergence nodes for the differentially represented pathways: Pparg , Ppara , Rora , and Rxra . Enrichment of H4K14ac and H3K9me3 within gene-specific promoters was interrogated with quantitative polymerase chain reaction (qPCR) on ChIP’ed DNA with the use of primers that were proximal to the TSS ( Supplementary Table 15 ). Indeed, both H3K14ac and H3K9me3 were enriched among both wild-type and G4 +/− fetal liver after in utero HFD exposure compared with control diet ( Figure 5 , A-D).
We hypothesized that, if there were an enrichment of these modifications with HFD exposure, this enrichment should correlate with significant alterations in gene-specific transcription. Pparg , Ppara , Rora, and Rxra expression was quantified by qPCR ( Figure 5 , E and F). Pparg was increased in the G4 +/− HFD-exposed fetuses compared with the control diet group. Expression levels of Ppara , Rora, and Rxra were not altered in HFD-exposed fetuses in either the wild-type or G4 +/− groups.
Promoter enrichment of H3K14ac and H3K9me3 is reversed at 5 weeks of age, despite HFD exposure during lactation
We performed qPCR on H3K14ac and H3K9me3 ChIP’ed DNA using primers proximal to the TSS. At 5 weeks of age, both modifications are enriched with maternal HFD exposure in the Pparg promoter, but only among G4 +/− offspring ( Figure 6 , A). In HFD wild-type animals, there was a significant decrease in H3K14ac in the Ppara promoter ( Figure 6 , B). Similarly, both H3K14ac and H3K9me3 were decreased in the Rora promoter with HFD exposure ( Figure 6 , C). There were no significant changes in the Rxra promoter ( Figure 6 , D). However, messenger RNA expression analysis demonstrated that Ppara expression was decreased significantly after maternal HFD exposure in both the wild-type and G4 +/− 5-week-old offspring ( Figure 6 , E and F); messenger RNA expression of Pparg and Rxra were decreased significantly only in the G4 +/− offspring ( Figure 6 , F).
