A maternal high-fat diet modulates fetal SIRT1 histone and protein deacetylase activity in nonhuman primates

Abstract

In nonhuman primates, we previously demonstrated that a maternal high-fat diet (MHFD) induces fetal nonalcoholic fatty liver disease (NAFLD) and alters the fetal metabolome. These changes are accompanied by altered acetylation of histone H3 (H3K14ac). However, the mechanism behind this alteration in acetylation remains unknown. As SIRT1 is both a lysine deacetylase and a crucial sensor of cellular metabolism, we hypothesized that SIRT1 may be involved in fetal epigenomic alterations. Here we show that in utero exposure to a MHFD, but not maternal obesity per se, increases fetal H3K14ac with concomitant decreased SIRT1 expression and diminished in vitro protein and histone deacetylase activity. MHFD increased H3K14ac and DBC1-SIRT1 complex formation in fetal livers, both of which were abrogated with diet reversal despite persistent maternal obesity. Moreover, MHFD was associated with altered expression of known downstream effectors deregulated in NAFLD and modulated by SIRT1 (e.g., PPARΑ, PPARG, SREBF1, CYP7A1, FASN, and SCD). Finally, ex vivo purified SIRT1 retains deacetylase activity on an H3K14ac peptide substrate with preferential activity toward acetylated histone H3; mutagenesis of the catalytic domain of SIRT1 (H363Y) abrogates H3K14ac deacetylation. Our data implicate SIRT1 as a likely molecular mediator of the fetal epigenome and metabolome under MHFD conditions.—Suter, M. A., Chen, A., Burdine, M. S., Choudhury, M., Harris, R. A., Lane, R. H., Friedman, J. E., Grove, K. L., Tackett, A. J., Aagaard, K. M. A maternal high-fat diet modulates fetal SIRT1 histone and protein deacetylase activity in nonhuman primates.

The sirtuins are a highly conserved family of proteins implicated as critical mediators, both as energy sensors and transcriptional effectors by controlling the acetylation state of histones. Mammalian SIRT proteins (homologs of Saccharomyces mating type cassette regulator Sir2p) are well-characterized NAD⁺-dependent protein deacetylases that are evolutionarily well conserved (1, 2). In mammals, SIRT1 is viewed as a broad integrator of mammalian physiology and mediates cellular differentiation, metabolism, tolerance to oxidative stress, cell survival, apoptosis, and circadian responses (3–9). To execute such vast physiological processes, the sirtuins modulate equally broad molecular and cellular functions. SIRT1, SIRT6, and SIRT7 are enriched in the nucleoplasm, nucleoli, and heterochromatin, while SIRT1 and SIRT2 shuttle between the nucleus and cytoplasm. SIRT3, SIRT4, and SIRT5 are mitochondrial and play fundamental roles in the regulation of energy metabolism and insulin secretion and regulation of critical metabolic pathways, such as the urea cycle (reviewed in ref. 10).

Of the 7 SIRT proteins homologous to Sir2 in mammals, all are known to differentially deacetylate transcription factors (i.e., p53 or the circadian regulator PER2) and additionally serve as regulators of RNA polymerase transcribed genes (3, 7–9). As the first acetylated protein described was a histone (11), it is not surprising that the sirtuins have also been ascribed crucial roles as histone deacetylases (HDACs; ref. 12). In recent years, studies in yeast have definitively demonstrated that in addition to functioning as nonhistone protein deacetylases, Sir2 functions as a highly efficient and modification-specific HDAC (1). These observations have been generally extended toward mammalian regulation of chromatin (1), with one report suggesting that the ability of SIRT1 to silence tumor suppressor genes with known 5′ CpG islands is related to its functioning as an H4K16 and H3K9 deacetylase (13).

There are several lines of evidence to suggest that SIRT1 has crucial functions in regulating and integrating hepatic and circadian metabolism. The first described nonhistone target for SIRT1 was p53, which is deacetylated and repressed on DNA damage or oxidative stress, resulting in impaired apoptosis (14, 15). Second, SIRT1 is a circadian deacetylase that counteracts the core CLOCK histone acetyltransferase (HAT) activity by regulating the amplitude of circadian clock-controlled gene expression (9, 16). We have previously demonstrated in our Japanese macaque nonhuman primate model of chronic maternal high-fat diet (MHFD) exposure that the fetal hepatic circadian machinery (namely NPAS2) is perturbed in association with alterations in H3K14 occupancy of the NPAS2 promoter (17). Third, transgenic knockout mice demonstrate that SIRT1 regulates hepatic lipid metabolism, thereby mediating nonalcoholic fatty liver disease (NAFLD; refs. 5, 18–20). SIRT1 liver-specific deletion in a conditional-knockout model renders weight gain and hepatic steatosis in response to a high-fat high-cholesterol diet (18). Of interest to our work, moderate overexpression of a Sirt1 transgene under its endogenous promoter renders protection against high-fat diet (HFD)-induced NAFLD in adult mice in association with Ppargc1a up-regulation (5). However, the molecular pathogenesis of these mechanisms is not entirely clear. Because SIRT1 lacks a DNA-binding domain, its actions are coordinated initially at the level of recruitment to regions of interest through DNA binding factors and likely via deacetylation of coactivators such as PPARGC1A (18, 20). It is equally probable that SIRT1 may directly modify promoter occupancy in reprogrammed genes of interest via alterations in the histone code. However, direct evidence of SIRT1 as a potent HDAC in a biologically relevant mammalian system has been lacking.

Based on the literature and prior findings in our nonhuman primate model of MHFD exposure (17, 21–29), we speculated that SIRT1 may modulate fetal NAFLD and disruption of the peripheral circadian circuitry as a result of its ability to function both as a histone and a protein deacetylase. To determine whether SIRT1 is in fact a potent HDAC with biological function, we employed a set of biochemical and molecular tools. We show here that in utero exposure to MHFD, but not maternal obesity per se, increases fetal acetylation of lysine 14 of histone H3 (H3K14ac) with concomitant decreased SIRT1 expression and is accompanied by diminished in vitro activity. Deleted in breast cancer 1 (DBC1) association with SIRT1 increases under HFD conditions. Similarly, genes associated with SIRT1 were observed to increase with HFD exposure (i.e., PPARG, ref. 30; SREBF1, ref. 31; PPARA, ref. 30; and CYP7A1, ref. 32). Employing mass spectrometry, we further demonstrated that purified SIRT1 retains deacetylase activity on an H3K14ac peptide substrate. In support of its role as an HDAC, mutagenesis of the catalytic domain of SIRT1 (H363Y) abrogates H3K14ac deacetylation. Taken together, our data imply that SIRT1 functions as a likely molecular mediator of the fetal epigenome in response to MHFD but not obesity.

MATERIALS AND METHODS

Experimental design

All animal procedures were done in accordance with approval from the institutional animal care and use committees of both Baylor College of Medicine and the Oregon National Primate Research Center (ONPRC), as described previously (17, 33). Young adult female Japanese macaques were bred and socially housed in indoor/outdoor enclosures in groups of 5–9 with 1–3 males/group. The animals were fed either a control diet (standard monkey chow, 14% calories from fat; Lab Diet Monkey Diet no. 5052; Test Diet, Richmond, IN, USA) or an HFD (32% calories from fat; Custom Diet 5A1F; Test Diet). The MHFD was supplemented with calorically dense treats. Before our studies, the animals were fed a control diet. During the fifth year of this study, diet-reversal animals consumed control-diet monkey chow during breeding and throughout gestation. Animals were allowed to breed naturally. Pregnancies were terminated by cesarean delivery at early third trimester at gestational day 130 (G130) by ONPRC veterinarians. Immediately after cesarean delivery, the fetuses were delivered to necropsy, deeply anesthetized (sodium pentobarbital, >30 mg/kg i.v.), and then exsanguinated. All studies reported herein employed snap-frozen fetal liver from control-diet-, diet-reversal-, or HFD-fed dams.

Quantitative PCR (qPCR)

Gene expression analysis of liver was performed using fetal control-diet- HFD-, and diet-reversal-exposed animals (n=6/group). All genes were first cloned from Japanese macaque mRNA and sequenced. The sequence was used to design primers and probes for expression analysis. RNA was extracted using the NucleoSpin kit (Machery Nagel, Düren, Germany) and converted to cDNA using the high-capacity cDNA reverse transcription kit (Applied Biosystems, Carlsbad, CA, USA). Real-time qPCR was performed for each sample in quadruplicate and repeated 3 times. Data are presented as a fold change in expression compared with control-diet-exposed animals. The ΔΔC_T method was used for analysis (34).

Site-directed mutagenesis

A pair of primers was designed to generate a histidine to tyrosine mutation at residue 363 of SIRT1, as described previously (14). The sequences of the primer pair were as follows: sense strand 5′-ATAATTCAGTGTACTGGTTCCTTTGCA-3′ and antisense strand 5′-TGCAAAGGAACCAGTACACTGAATTAT-3′. The plasmid construct pcDNA3.1/His-SIRT1 was used as the template, and the H363Y mutation was generated using the Quick Change site-directed mutagenesis kit (Stratagene, Santa Clara, CA, USA). The mutagenesis procedure was performed according to the manufacturer's instructions. The H363Y mutation was confirmed by DNA sequencing, and no other mutations were detected.

Cell culture and transfection

Cos-1 cells (American Type Culture Collection, Manassas, VA, USA) were cultured in a humidified tissue culture incubator with 5% CO₂ at 37°C. Cos-1 cells were maintained in DMEM (Invitrogen, Grand Island, NY, USA) supplemented with 10% FBS (Invitrogen). The constructs pCDNA3.1/His-SIRT1, pcDNA3.1/His-H363Y, and empty vector pcDNA3.1/His were transfected into Cos-1 cells by Lipofectamine (Invitrogen) according to the manufacturer's instructions. The cells were collected 48 h after transfection. Several groups of cells were incubated with 100 μM of resveratrol for 16 h before collection.

Western blot analysis

Western blot was done as described previously (21). The anti-SIRT1 rabbit antibody was purchased from Abcam (Cambridge, MA, USA). DBC1 association with Sirt1 was assayed in 20 μg of immunoprecipitated Sirt1 samples separated by SDS-PAGE (10% gels) and transferred onto a PVDF membrane. Western blot analysis was carried out using DBC1 antibody (cat no. A300-432A; Bethyl Laboratories, Montgomery, TX, USA) at a dilution of 1:1000.

SIRT1 activity assay

The activity of SIRT1 from Cos-1 cells or liver tissues of Japanese macaques was tested with a Biomol SIRT1 fluorescence assay kit (AK-555; Biomol, Farmingdale, NY, USA) according to the manufacturer's instructions. The fluorophore was excited with 360-nm light, and the emitted light (460 nm) was detected on a fluorometric plate reader.

HDAC activity assay

HDAC activity was measured with a Biomol HDAC Colorimetric Assay Kit (AK-501) according to the manufacturer's instructions. Fetal macaque liver or Cos-1 cell extracts were incubated with 500 μM of deacetylase substrate for 45 min at 37°C. The samples were incubated with 50 μl of developer II for 15 min at room temperature. Optical density values of samples were read at 405 nm in a microtiter plate reader.

Mass spectrometry

To clearly delineate SIRT1-mediated histone deacetylation of H3K14ac, MALDI mass spectrometry was performed. Acetylated peptides were incubated with nuclear extracts of transfected and nontransfected Cos-1 cells for 45 min at 37°C. The human recombinant Sirt1 was used for SIRT1 positive control (AK-555; Biomol). Peptide samples were crystallized with 2,5-dihydroxybenzoic acid, and mass spectra were collected with a PerkinElmer MALDI-prOTOF mass spectrometer (PerkinElmer, Wellesley, MA, USA; refs. 35, 36). All deacetylation products were verified by tandem mass spectrometry with a Thermo LTQ-XL mass spectrometer (Thermo Scientific, Waltham, MA, USA). Comparison studies used H3K14ac, H3K9ac, and H4K5acK8acK12acK16ac peptides. Quantitative analysis of deacetylation was performed by isotopic labeling of nonacetylated lysines with d6-acetic anhydride, as described previously (35, 37, 38).

Statistical analysis

An independent-sample 2-tailed t test was performed for expression analysis and activities in fetal liver of Japanese macaques, comparing the livers from the HFD and diet-reversal groups with those from the control-diet group.

RESULTS

H3K14 acetylation is altered with maternal diet exposure

We previously observed that the fetal hepatic epigenome is preferentially hyperacetylated on H3K14 in response to MHFD exposure (17, 21). However, we had not definitively distinguished whether this observed increased acetylation was due to maternal diet exposure or maternal obesity. To distinguish between the two scenarios, we investigated the level of fetal hepatic H3K14ac in diet-reversal-exposed animals. Briefly, Japanese macaque (Macaca fuscata) dams were fed either a control diet (14% fat) or a high-fat breeding diet (32% fat) for successive gestations. A cohort of the obese dams fed the HFD were switched to the control diet just before breeding (diet reversal) and thereafter maintained on the control diet throughout gestation. Ergo, the dams were obese but consuming a control diet. This diet-reversal cohort allowed us to distinguish between the effects of MHFD vs. maternal obesity on the developing fetal histone code and reprogrammed gene expression (22).

Histones were extracted from fetal tissue in the early third trimester, e.g., G130, for measurement of differential acetylation. We found that in the diet-reversal animals, H3K14ac levels were similar to those seen in control-diet-exposed animals (Fig. 1A). We hypothesized that there were two possible reasons for the increased acetylation, namely increased activation of a HAT or repression or inhibition of an HDAC (21). We observed that significant increases in H3K14 acetylation were accompanied by decreased HDAC activity, as measured in a class I–III HDAC in vitro assay (Fig. 1A, B). Interestingly, reversal of the maternal diet with persistent obesity significantly partially abrogated these affects (Fig. 1B). We also found a significant increase in GCN5 expression with HFD exposure (Fig. 1C). GCN5 expression was similar to that of controls in the diet-reversal animals, revealing a similar pattern as seen with H3K14 acetylation (Fig. 1C).

Figure 1.

Fetal hepatic histone H3K14 acetylation, HDAC activity, and HAT expression are altered by virtue of maternal diet exposure. We have previously reported that H3K14 acetylation is increased with exposure to an MHFD. A) Here we show that H3K14ac levels return to those of control-diet levels with diet reversal exposure (n=6/group). A representative Western blot is shown at bottom. B) Using a commercially available kit to measure class III HDAC activity, we found that activity is decreased with HFD exposure. Levels remain decreased in diet-reversal animals (n=6/group). C) Using qPCR, we measured GCN5 expression levels in fetal hepatic tissue. We found increased GCN5 levels that correspond with increased H3K14 acetylation levels in HFD-exposed animals. GCN5 expression is similar to controls with diet reversal (n=10 control, 6 HFD, and 6 reversal). *P < 0.05.

Fetal hepatic SIRT1 gene expression, protein level, and activity are deregulated following MHFD exposure

We hypothesized that the increased H3K14ac levels seen with MHFD exposure would have many levels of regulation. As we observed that GCN5 is increased in HFD exposed liver (Fig. 1C), and it is known that the HAT GCN5 and the HDAC SIRT1 are mutual antagonists of the nonhistone protein PGC1a (reviewed in ref. 39), we sought to determine whether SIRT1 is altered similarly with HFD exposure. We found that Sirt1 mRNA expression levels were significantly decreased in the fetal livers from HFD-exposed animals relative to control (Fig. 2A); these findings extended to protein translation (Fig. 2B). Among the dietary reversal cohort offspring, hepatic Sirt1 transcript and protein levels were no longer significantly different from control (Fig. 2A, B).

Figure 2.

Fetal hepatic SIRT1 expression, protein, and activity decrease with HFD exposure. A, B) SIRT1 mRNA expression levels (A) and SIRT1 protein levels (B) decrease with HFD exposure. A representative SIRT1 Western blot is included at bottom. C) SIRT1 activity was measured using a commercially available kit (see Materials and Methods) using p53 acetylated substrate (n=9 control, 6 HFD, and 7 reversal for A–C). D) DBC1-Sirt1 interaction in HFD-fed primate liver. Equivalent amounts of liver protein homogenates from primates fed either control diet or an HFD or exposed to diet reversal were subjected to immunoprecipitation with Sirt1, followed by Western blot analysis with DBC1 (see Materials and Methods). Fold change relative to control-diet-fed animals expressed relative to IgG input. E) Representative Western blot for DBC1 following anti-SIRT1 immunoprecipitation demonstrates significant increased fetal DBC1 in association with in utero HFD exposure. *P < 0.05.

Because SIRT1 is a known protein deacetylase, we next sought to determine whether SIRT1-specific deacetylase activity was significantly altered in our primate model. To test SIRT1 activity, we utilized the well-characterized in vitro Fleur de Lys assay (40–42). As shown in Fig. 2C, in this in vitro assay employing the well-known target of SIRT1 (namely p53) as the substrate (6, 43), fetal liver extracts from dams fed the control diet retained significantly higher levels of p53 deacetylase activity compared with HFD exposure; this was due to the maternal diet and not obesity, as diet reversal abrogated this effect (Fig. 2C). We further demonstrated that association of the DBC1 gene product with fetal SIRT1 was increased following MHFD exposure in utero but not maternal obesity per se (Fig. 2D, E). Interestingly, the DBC1 gene product negatively regulates SIRT1 activity via binding to its active site and thereby inhibiting SIRT1-substrate (e.g., p53) interactions (44, 45). Moreover, DBC1 also plays a key role in modulating hepatic cellular steatosis by its regulation of SIRT1 activity (46).

Fetal hepatic SIRT3 protein levels are not altered with HFD exposure

SIRT3 is a mitochondrial NAD⁺-dependent protein deacetylase that aids in the adaption to food withdrawal and calorie restriction (47). During times of food withdrawal, SIRT3 is up-regulated to maintain homeostasis following nutrient deprivation and is involved in activation of the fatty acid oxidation pathway. However, with an HFD challenge, hepatic SIRT3 expression is down-regulated in adult mice, leading to mitochondrial hyperacetylation (48). To determine whether a similar alteration was occurring in our model system, we measured SIRT3 protein levels by Western blot. We found that Sirt3 levels were similar between control-diet-fed (1.05±0.236) and HFD-exposed (0.928±0.117; P>0.05) animals. Ergo, the decrease in Sirt1 expression, protein, and activity we observe in the fetal liver following in utero HFD exposure is not ubiquitous to all sirtuins, nor could a change in SIRT3 account for our observed increase in H3K14ac.

Downstream targets of SIRT1 are altered with MHFD exposure

Because SIRT1 is involved in many cellular processes, its down-regulation with MHFD exposure could potentially lead to many alterations in the expression of downstream effectors. SIRT1 has a role in activating PPARGC1A, a transcriptional coactivator essential for nutrient and energy homeostasis by deacetylation (30, 49). It interacts functionally with nuclear receptors, such as PPARG and PPARA, to drive transcription of genes important in the regulation of glucose and lipids (30). We have reported that PPARGC1A expression is increased (49), so we sought to investigate changes in its interacting partners PPARG and PPARΑ. We found that expression of both genes increases with HFD exposure compared with control-diet exposure (Fig. 3A, B). However, while PPARG does not differ from control diet in the diet-reversal group, the expression levels of PPARΑ remain elevated under conditions of maternal obesity (Fig. 3B vs. A).

Figure 3.

SIRT1-associated genes are altered with HFD exposure. A–D) We found that PPARG (A), PPARA (B), SREBF1 (C), and CYP7A1 (D) are all increased with HFD exposure in macaque fetal liver. E) FASN. F) SCD. Except for PPARA (B), whose expression levels remain elevated, all genes show levels similar to control-diet-exposed animals with diet reversal (n=10 control, 6 HFD, and 6 reversal). *P < 0.05.

Armed with these findings, we investigated expression of a number of genes known to be integral to lipid metabolism and to serve as SIRT1-mediated targets. Similarly, we observed several such downstream mediators of SIRT1 to be altered by virtue of HFD exposure but not maternal obesity. Specifically, SREBF1 is a target of SIRT1 (50), and SREBF1 expression is increased with MHFD exposure (Fig. 3C). SIRT1 represses expression of CYP7A1 through interaction with the orphan nuclear receptor NR0B2 (32). In HFD-exposed animals, CYP7A1 expression is increased (Fig. 3D) The SIRT1-responsive genes fatty acid synthase (FASN; ref. 51) and stearoyl-CoA desaturase (SCD; refs. 52, 53) also increase with MHFD exposure (Fig. 3E, F).

Purified SIRT1 deacetylates lysine 14 of histone H3, as evidenced by mass spectrometry, and the catalytic domain of SIRT1 is necessary for this activity

Although the data presented in Fig. 2 suggest that the MHFD was modulating SIRT1 protein and HDAC activity, they directly ascribed neither substrate nor deacetylase specificity. To do so, we generated a series of expression vectors that overexpressed either endogenous (wild type) human SIRT1 (huSIRT1) or a catalytically inactive, dominant-negative inhibitor of SIRT1, human SIRT1 with histidine 363 mutated to tyrosine (huSIRT1H363Y; ref. 14). We used these vectors to transfect Cos-1 green monkey kidney epithelial cells, which lack endogenous Sirt1. As shown in Fig. 4A, Cos-1 cells have a low endogenous level of Sirt1 (lane 1, pcDNA3.1 vector alone) compared with transfected huSIRT1 (lane 2) or huSIRT1H363Y (lane 3). We then purified SIRT1 or SIRT1H363Y and subjected the enzymes to a deacetylase assay, using acetylated peptide substrates and measuring deacetylation by mass spectrometry.

Figure 4.

Catalytically active SIRT1 deacetylates H3K14ac and has preference for acetylated histone H3. A) Cos-1 cells were used for expression and purification of huSIRT1. Cos-1 cells have a low level of expression of endogenous SIRT1, as seen with empty vector (pcDNA3.1), but levels increase with expression of huSIRT1 or huSIRT1H363Y. B) Mass spectrometric analysis identifies H3K14ac as a SIRT1 substrate. Spectrum lanes 1 and 2 shows no-peptide and no-lysate controls, while lane 3 shows that H3K14ac peptide, plus commercially available purified recombinant huSIRT1, demonstrates a shift of −42 Da, indicative of complete deacetylation of K14 by SIRT1. Cos cell lysate (lane 4) without transfected Sirt1 lacks detectable H3K14 deacetylase activity (no −42 Da shift). However, with transfection of huSirt1 (lane 5) we observe again a −42 Da shift specific to H3K14 deacetylation, and addition of resveratrol (a SIRT1 agonist) to the Cos lysate (lane 6) preserves deacetylation. Transfection with dominant-negative huSirt1 H363Y into Cos cells without (lane 7) or with (lane 8) resveratrol abrogates this observed deacetylation of H3K14ac. C) Quantitative mass spectrometry shows H3K14ac and H3K9ac as preferential substrates for SIRT1 relative to a tetraacetylated H4 (K5K8K12K16) peptide.

Mass spectometric results demonstrate that purified huSIRT1 demonstrates specific activity against a K14ac peptide, with a loss of acetyl group (−42 Da; Fig. 4B). Spectrum lane 1 in Fig. 4B shows the no-peptide control, while lane 2 shows the H3K14ac peptide for reference. H3K14ac peptide, plus commercially available purified recombinant huSIRT1, demonstrates a shift of −42 Da, indicative of complete deacetylation of K14 by SIRT1 (Fig. 4B, lane 3). As anticipated, Cos cell lysate (Fig. 4B, lane 4) without transfected Sirt1 lacks detectable H3K14 deacetylase activity, as no −42 Da shift is observed. However, with transfection of huSirt1 (Fig. 4B, lane 5), we observe again a −42 Da shift specific to H3K14 deacetylation, and addition of resveratrol (a SIRT1 agonist) to the Cos lysate (Fig. 4B, lane 6) preserves deacetylation. Transfection with dominant-negative huSirt1 H363Y into Cos cells without (Fig. 4B, lane 7) or with (Fig. 4B, lane 8) resveratrol abrogates this observed deacetylation of H3K14ac, suggesting catalytic domain specificity of SIRT1-directed histone deacetylation. Finally, we used quantitative mass spectrometry in delineating the relative histone substrate specificity for SIRT1. As shown in Fig. 4C, preferential HDAC activity for SIRT1 is observed for H3K14 and H3K9 acetyl modifications, over any of the assayed H4 acetyl motifs (K5, K8, K12, and K16). We conclude that SIRT1 has H3K14 and K9 preferential deacetylase activity.

DISCUSSION

Alterations that occur in fetal life due to an adverse in utero environment may help to set the stage for the adult onset of metabolic disease. The role of epigenetic modifications, such as histone modifications and DNA methylation, in establishing a molecular memory of this in utero exposure is still poorly understood. We have previously reported in our Japanese macaque model of maternal excess nutrition that fetal hepatic H3K14ac is increased with MHFD exposure (21). Here we sought to determine whether this change in acetylation is maternal diet or maternal obesity dependent and the molecular mechanisms behind the increase.

We found that fetal animals exposed to maternal diet reversal demonstrated H3K14 acetylation similar to chow-fed controls (Fig. 1A), leading us to conclude that the increased acetylation is associated with HFD exposure and not maternal obesity per se. An increase in acetylation can be due to an increase in abundance or activity of the enzyme responsible for acetylation or to a similar decrease of the enzymes responsible for removing the acetyl group. We found that both scenarios likely contribute to H3K14 acetylation status in fetal liver. An assay for class III HDAC activity reveals that HDAC activity is decreased in fetal livers from mothers fed an HFD and remains decreased with diet reversal (Fig. 1B). Expression of GCN5 (the HAT responsible for H3K14 acetylation) is increased in HFD-exposed fetal liver, corresponding with the increase in K14 acetylation (Fig. 1C). Similar to the trend with H3K14ac, GCN5 expression is no different from control-diet-exposed animals with diet reversal, suggesting that GCN5 may play an important role in diet-induced H3K14 acetylation.

We next sought to determine the role of SIRT1 in H3K14ac in our model system for two reasons. First, GCN5 and SIRT1 have been shown to be mutually antagonistic in regulating of PPARGC1A activity (39). Second, SIRT1 has a very well established role in the coordination of cellular metabolism and the sensing of depleted or excess nutrient availability through its dependence on cellular availability of NAD⁺. SIRT1 is known to be increased with caloric restriction; however, data on SIRT1 levels with nutritional excess are lacking.

We found that fetal hepatic Sirt1 transcription, translation, and deacetylase activity are decreased with MHFD exposure (Fig. 2A–C). Moreover, this is HFD specific, as transcript, protein, and activity levels are similar among offspring of control-diet- and diet-reversal-group (obese) dams. This decrease in SIRT1 parallels the increase in H3K14 acetylation observed in these same offspring. In addition, DBC1, a negative regulator of HDAC3 as well as SIRT1 (54), was bound to SIRT1 under MHFD conditions, and this association decreased on diet reversal. Although DBC1 levels do not show a significant change in mice fed an HFD, both PKA and AMPK can acutely activate SIRT1 by inducing dissociation of SIRT1 from DBC1 (54), suggesting that under MHFD conditions, histone H3 hyperacetylation at lysine 14 is dependent on both GCN5 and SIRT1. However, as a decrease in SIRT3 protein is not observed, we conclude that the histone deacetylation is not ubiquitous to the sirtuin family and is likely SIRT1 specific.

Because SIRT1 is involved in cellular metabolism, DNA damage repair, and circadian rhythms, we sought to determine whether any well-characterized SIRT1 effectors are similarly altered in our model. Expression of 6 known SIRT1-associated genes (PPARA, PPARG, SREBF1, FASN, SCD, and CYP7A1) was increased in association with MHFD exposure (Fig. 3A–D). Except for PPARG, which remained elevated in the diet-reversal group, offspring of obese dams fed a control diet manifested levels of SIRT1 effectors akin to those of the lean control-fed dams. Taken together, these findings suggest that an MHFD (but not maternal obesity per se) modifies the fetal epigenome at H3K14 in association with modulation of expression and activity of Sirt1. Moreover, this likely results in further functional modifications of downstream molecular effectors, notably PPARG, SREBF1, and CYP7A1.

Finally, we sought to determine whether SIRT1 has activity specific for H3K14ac. To answer this question, we subjected histone H3 peptides acetylated at K14 to an in vitro deacetylation reaction and measured deacetylation by mass spectrometry. We found that SIRT1 indeed has H3K14-specific deacetylase activity. Cos-1 lysate alone, which has little endogenous Sirt1, did not deacetylate the H3K14 peptide unless transfected with catalytically active huSirt1 construct. Expressing a catalytically null mutant of Sirt1 in Cos-1 cells did not increase deacetylase activity, even in the presence of a potent resveratrol agonist. While this is not specific to H3K14, it appears specific to H3 histone motifs, as H3K14 and H3K9 are both deacetylated, while H4 deacetylation is not observed. This is consistent with our prior published data, demonstrating significant fetal H3K14 acetyltion and modest H3K9 acetylation with MHFD exposure (17, 21).

In summary, we have shown that HFD exposure, but not maternal obesity per se, modifies the fetal hepatic histone code. Specifically, we report that H3K14ac is only increased with MHFD exposure and not by exposure to maternal obesity (diet-reversal cohort). We demonstrate that the increase in expression of the HAT GCN5, concomitant with a decrease in expression and activity of the HDAC SIRT1, likely contributes to observed increased levels of acetylation. This is likely of considerable relevance, as it has been previously shown that GCN5 and SIRT1 work on the same nonhistone substrate and have opposing effects.

In this study, we significantly extend these findings to histone substrates and H3K14 in particular. We further characterize the specificity of SIRT1 in vitro and demonstrate that catalytically active SIRT1 is a histone H3K14 deacetylase with a preference for acetylated H3 rather than H4. Based on these findings, we speculate that although SIRT1 is lauded as a molecular mediator of caloric restriction, SIRT1 likely plays a broader role in regulating energy metabolism and the cellular responses to excess as well as limited nutrition in the developing fetus. Given that overexpression of SIRT1 protects against HFD-induced nonalcoholic fatty liver in adult mice, we further specultate that functional alterations in SIRT1 play a critical role in the development of MHFD-induced nonalcoholic fatty liver and inflammation in the primate fetus. When considered in its entirety, this expanding body of literature lends credence to the notion that SIRT1 histone and protein deacetylase activity is a likely critical molecular mediator linking nutrition to cellular metabolism and the development of fatty liver.