Hepatic Methionine Homeostasis Is Conserved in C57BL/6N Mice on High-Fat Diet Despite Major Changes in Hepatic One-Carbon Metabolism

Obesity is an underlying risk factor in the development of cardiovascular disease, dyslipidemia and non-alcoholic fatty liver disease (NAFLD). Increased hepatic lipid accumulation is a hallmark in the progression of NAFLD and impairments in liver phosphatidylcholine (PC) metabolism may be central to the pathogenesis. Hepatic PC biosynthesis, which is linked to the one-carbon (C1) metabolism by phosphatidylethanolamine N-methyltransferase, is known to be important for hepatic lipid export by VLDL particles. Here, we assessed the influence of a high-fat (HF) diet and NAFLD status in mice on hepatic methyl-group expenditure and C1-metabolism by analyzing changes in gene expression, protein levels, metabolite concentrations, and nuclear epigenetic processes. In livers from HF diet induced obese mice a significant downregulation of cystathionine β-synthase (CBS) and an increased betaine-homocysteine methyltransferase (BHMT) expression were observed. Experiments in vitro, using hepatoma cells stimulated with peroxisome proliferator activated receptor alpha (PPARα) agonist WY14,643, revealed a significantly reduced Cbs mRNA expression. Moreover, metabolite measurements identified decreased hepatic cystathionine and L-α-amino-n-butyrate concentrations as part of the transsulfuration pathway and reduced hepatic betaine concentrations, but no metabolite changes in the methionine cycle in HF diet fed mice compared to controls. Furthermore, we detected diminished hepatic gene expression of de novo DNA methyltransferase 3b but no effects on hepatic global genomic DNA methylation or hepatic DNA methylation in the Cbs promoter region upon HF diet. Our data suggest that HF diet induces a PPARα-mediated downregulation of key enzymes in the hepatic transsulfuration pathway and upregulates BHMT expression in mice to accommodate to enhanced dietary fat processing while preserving the essential amino acid methionine.


Introduction
The epidemic occurrence of obesity in the general population has caused an increase in the incidence of obesity-associated diseases.NAFLD is an incremental liver dysfunction that is associated with obesity [1] and induces a wide range of hepatic alterations starting with steatosis and non-alcoholic steatohepatitis that can progress to cirrhosis and hepatocellular carcinoma [2].In general, diet-induced obesity (DIO) in mice generated by feeding animals a HF diet causes hyperglycemia, hyperinsulinemia, reduced glucose tolerance and hepatic triacylglycerol (TG) accumulation [3,4].Previously, we have shown that feeding C57BL/6N mice a beef tallow based HF diet resulted in significant changes in hepatic and intestinal phospholipid (PL) and cholesterol contents, as well as changes in PC signature indicative for a) an increased PC synthesis via the CDP-choline pathway, b) an increased phosphatidylethanolamine (PE) methylation pathway activity in the liver and c) alterations in membrane PL remodeling [5].The observed higher levels of PC species with longer carbon chains found in the liver could originate most likely from an increased activity of the PE methylation pathway in hepatocytes [6].Changes of specific PC levels upon HF diet may modulate the activation state of the nuclear receptor PPARa, which is a prime candidate promoting fatty acid oxidation, lipid transport and ketogenesis in liver and intestine.Diacyl-phosphatidylcholine PC.aa (16:0/18:1) was recently identified as a natural ligand and activator of PPARa [7].
Biosynthesis and turnover of PC are important in the formation of VLDL particles and lipid export from hepatocytes which, when disturbed, promotes the accumulation of lipid droplets in hepatocytes causing steatosis [8,9,10].Hepatic PC biosynthesis is mainly dependent on dietary choline supply via the CDP-choline pathway, which accounts for approximately 70% of hepatic PC biosynthesis, whereas the remaining 30% is synthesized by the methylation of PE via phosphatidylethanolamine N-methyltransferase (PEMT) [9].This second pathway is also known to be required for VLDL secretion [11,12].Interestingly, PEMT-deficient mice (Pemt 2/2 ) fed a HF diet are protected from DIO due to disturbed de novo choline biosynthesis (PE methylation pathway), thus linking PC biosynthesis to the development of DIO [13].Furthermore, dietary choline supplementation of Pemt 2/2 mice reversed the protective effect suggesting that choline is essential for systemic lipid metabolism and distribution [13].Pemt polymorphisms resulting in altered PEMT activities have also been associated with the susceptibility for NAFLD in humans [14].Moreover, an enhanced secretion of PC derived from the hepatic PE methylation pathway has been observed in mice fed a high-fat/high-cholesterol diet [15].This may indicate increased demands of PL secreted into bile for the assembly of micelles [16] required in intestinal fat absorption.
The C1-metabolism is the principal pathway providing the methyl-donor S-adenosyl-methionine (SAM) in the methionine cycle necessary for numerous transmethylation reactions (Fig. 1).PEMT transfers three methyl-groups from SAM to PE thus linking PC biosynthesis and C1-metabolism in the liver [8,17].SAMdependent transmethylation leads to the synthesis of S-adenosylhomocysteine (SAH) that is hydrolyzed to homocysteine (Hcy).Hcy is either remethylated to methionine via folate-dependent processes in the folate cycle and by choline oxidation processes associated with the sarcosine pathway or converted via the transsulfuration pathway to cystathionine which can be catabolized to cysteine [8].Cysteine can be used for glutathione synthesis or is further metabolized (Fig. 1) to either taurine as main metabolic end product or used up for sulfate production [8].Genetic studies in Pemt 2/2 mice and CTP:phosphocholine cytidyltransferase 1a gene knockout mice (CTa 2/2 ) deficient for the hepatic PEMT pathway or CDP-choline pathway, respectively, demonstrate a functional link between C1-metabolism, transmethylation processes and PC biosynthesis.Pemt 2/2 mice display only around 50% of Hcy plasma levels compared to the levels found in wild type mice [18], whereas CTa 2/2 mice show elevated (20 to 40%) plasma Hcy levels [19].
Considering the relations described between (i) the C1-metabolism and PC synthesis, (ii) observed changes in PL contents and PC signatures upon HF diet feeding and (iii) the activation of PPARa by a diacyl-phosphatidylcholine identified as a natural PPARa ligand, the following questions arise: how does a high dietary fat load impact on the hepatic C1-metabolism pathways at the levels of gene and protein expression as well as metabolites?Which role plays PPARa signaling?Can alterations in important transmethylation processes, such as DNA methylation, be observed?
To answer these questions, we analyzed gene and protein expression levels and metabolite concentrations in the liver of animals on HF and control diets.Furthermore, we examined whether PPARa is involved in the observed changes in obese mice by in vitro experiments.Since methyl-group homeostasis may also affect epigenetic processes, we measured hepatic gene expression level of DNA methyltransferases (Dnmts) and determined global and local genomic DNA methylation in livers from mice of both dietary groups.

Animal Study and Sample Collection
Eight-week-old male C57BL/6NCrl mice (Charles River Laboratories) were maintained in 12 h light/dark cycles with unlimited access to food and water.The dietary intervention was performed as previously described [5].Briefly, after 2 weeks feeding a standard laboratory chow (Ssniff GmbH, no.V1534), mice with similar mean body weight were divided randomly into a control (C) and a HF group (n = 12).The C group was given a carbohydrate2/starch-based diet, comprising 4.2% (w/w) fat (Ssniff GmbH, no.E15000-04; Table S1), while the HF group was given a beef tallow-based diet, with 34% (w/w) fat (Ssniff GmbH, no.E15741-34; Table S1) for 12 weeks.Formulations of the individual diets were made from components with the same lot number.Food intake was measured during the feeding trial and the corresponding intake of energy, protein, methionine, cystine, choline chloride, folate and vitamin B 12 were calculated for twelve weeks of feeding (Table S2).
At the end of the feeding trial, mice in a non-fasting state were anesthetized using isoflurane and blood was taken from the retroorbital sinus.After killing the mice by cervical dislocation, liver samples were collected, snap-frozen in liquid nitrogen and stored at 280uC until analysis.Animal experiments were done in accordance with the German guidelines for animal care and approved by the Department of Veterinary Affairs of the government of Oberbayern/Germany.
Cell culture.Fao rat hepatoma cells were originally cloned and characterized by Deschatrette and Weiss [20] and available at the European Collection of Cell Cultures (EACC, Salisbury, Wiltshire,UK; ECACC catalog code: 89042701).Fao rat hepatoma cells (obtained from T. Skurk, Technische Universita ¨t Mu ¨nchen, Nutritional Medicine, Freising, Germany) were cultured in Dulbecco's Modified Eagle Medium (DMEM 25 mM glucose, Invitrogen) supplemented with 10% fetal calf serum (PAA), 100 U/ml penicillin and 100 mg/ml streptomycin (PAA) in a humidified atmosphere of 95% air and 5% CO 2 at 37uC.Two days before the experimental start and prior reaching confluence, cells were preincubated with DMEM (5.56 mM glucose, Invitrogen) supplemented with 0.5% bovine serum albumin (Sigma-Aldrich), followed by stimulation with different concentrations of the PPARa agonist WY14,643 (Sigma-Aldrich) dissolved in DMSO/PBS (50:50) as vehicle for 24 h.Fao rat hepatoma cells, treated only with vehicle (without PPARa agonist) were used as non-stimulated control cells.Cells were harvested for RNA isolation using Trizol reagent (Invitrogen).

RNA Isolation and Real-time Quantitative PCR (RT-qPCR
Isolation of total liver RNA and quantification of mRNA expression by RT-qPCR were performed as previously described [5].For analysis of Fao rat hepatoma gene expression, 50 ng total RNA was used, and for normalization, the invariant gene bglucuronidase was chosen.Oligonucleotide primer sequences are listed in Tables S3 and S4.

Analysis of Amino Acids and their Derivatives
Liver tissues were ground in liquid nitrogen and 100 mg of the homogenates were dissolved in 150 ml H 2 O/MeOH (50:50), vortexed and centrifuged.40 ml of the supernatant was used to determine the concentration of amino acids and their derivatives following the iTRAQ-labeling method using the AA45/32 TM Kit (Applied Biosystems).Samples were treated according to the manufacturer's instructions and analyzed via LC-MS/MS (3200QTRAP LC/MS/MS, Applied Biosystems).Spectra were processed using the AnalystH 1.5 Software [21].Protein determination for data normalization was performed using the Bradford assay (Biorad) with BSA as standard.
Hepatic betaine, dimethylglycine (DMG) and choline concentrations were analyzed by LC-MS/MS according to Holm et al. [22] with minor modifications.5 ml liver homogenates prepared and dissolved in H 2 O/MeOH (50:50) as described above were used to determine the concentration of hepatic betaine, DMG and choline by LC-MS/MS (QTRAP5500, ABSCIEX).As internal standard d 11 -betaine (Sigma-Aldrich) was used.
The HPLC-MS/MS detection of hepatic SAM and SAH levels was performed using a Waters 2795 alliance HT, coupled to a Quattro Micro API tandem mass spectrometer (Waters Corporation) according to the previously described methodology with minor modifications [23].After homogenization of liver tissues in 1 N acetic acid and centrifugation at 12000 g for 10 min, 25 ml of the supernatants were added to 375 ml of 20 mM ammonium acetate buffer (pH 7.4) and 40 ml of internal standard (SAM and SAH), cleaned on SPE column and analyzed by HPLC-MS/MS.
Liver Hcy and cystathionine were analyzed by gas chromatography mass spectrometry (GC-MS) according to Stabler et al., with minor modifications [24].Briefly, liver tissue samples were homogenized with 0.1 M perchloric acid, centrifuged at 12000 g for 10 min, neutralized with 1 M K 3 PO 4 , and stored for final preparation as described by Stabler et al. [24].
Protein extraction and western immunoblot analysis.Western immunoblot analysis was performed as previously described with minor modifications [25].Liver tissues were homogenized in RIPA supplemented with 1% protease inhibitor cocktail and 1% phosphatase inhibitor cocktail II (Sigma).Hepatic liver protein content calculated as hepatic protein per 1 g liver tissue was similar in control (127.5610.1)and HF mice (118.366.1).Total protein extracts were separated by SDS-polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane (Whatman GmbH).For immunodetec- tion, membranes were incubated with the following primary antibodies at 1:1000 to 1:20000 dilutions in TBS-T at 4uC over night: goat anti-BHMT (Novus Biologicals), rabbit anti-CBS (Aviva Systems Biology), rabbit anti-b-actin and rabbit antihistone H3 (Cell Signaling).After several washing steps, secondary donkey anti-goat or goat anti-rabbit conjugated with IRDye 800 (Li-cor) antibodies at 1:10000 to 1:50000 dilutions in TBS-T were exposed at room temperature for 1 h.Signal detection and quantification of fluorescence intensity were performed with the Odyssey infrared imaging system (Li-cor) and by using the Odyssey Application Software 3.0 (Li-cor) for calculation of integrated intensity.Histone H3 and b-Actin were used for normalization of protein abundance.

DNA Isolation
Genomic DNA was extracted from liver of C and HF diet fed mice by 12 mAU proteinase K digestion over night and 200 mg RNaseA incubation for 2 min followed by chloroform extraction.DNA was isolated from the aqueous phase using the DNeasy Blood & Tissue Kit (Qiagen) according to the manufacturer's protocol.DNA concentration was determined on a NanoDrop ND-1000 UV-Vis spectrophotometer (NanoDrop Technologies).

Analysis of Global Genomic DNA Methylation Using Luminometric Methylation Assay (LUMA)
Global DNA methylation was quantified using LUMA as previously described [26,27] with minor variations.Briefly, 1 mg genomic DNA was cleaved with the FastDigestH restriction enzymes HpaII+EcoRI or MspI+EcoRI (Fermentas) in two separate reactions for 15 min.The digestions were performed in a 96-well format using a PSQ96 TM MA system (Biotage AB).Peak luminometric heights were calculated with the PSQ96 TM MA software.The HpaII/EcoRI and MspI/EcoRI ratios were calculated as (dGTP+dCTP)/dTTP for the respective reactions.DNA methylation was calculated from the (HpaII/MspI) ratio at the investigated sites.

Methylation-sensitive Quantitative PCR (MS-qPCR)
Local genomic DNA methylation of CpG sites was quantified using MS-qPCR as previously described [28,29] with minor modifications.For each digestion, 1 mg DNA was incubated either without enzyme (sham) or with the methylation-sensitive FastDi-gestH restriction enzymes (Fermentas) AciI or HpaII at 37uC for 5 h, followed by heat inactivation at 65uC for 20 min.For DNA methylation analysis, 5 ng of digested genomic DNA were amplified using the QuantiTectH SYBR Green PCR Kit (Qiagen) on a Mastercycler ep realplex apparatus (Eppendorf) with CpG island specific primer pairs flanking informative restriction sites (AciI, HpaII).Following conditions were used: 1 initial cycle at 95uC for 15 min followed by 40 cycles of denaturation, annealing and elongation (at 95uC for 15 s, 61uC for 30 s and 72uC for 30 s, respectively), subsequently the melting curve analysis (1.75uC/ min) was performed.For normalization, an internal control region on chromosome 18 was amplified using a primer pair flanking no informative restriction site [28,29].MS-qPCR primer sequences are shown in Table S5.C q -values were retrieved from the realplex 2.0 software (Eppendorf).Calculation of restriction site DNA methylation was performed with the modified equation 100*2 2(DCq ROI 2 DCq control) [28,29].

Bisulfite Genomic Pyrosequencing
Bisulfite conversion of genomic DNA was performed as previously described [30].Briefly, 500 ng genomic DNA and the EZ DNA Methylation-Gold Kit (Zymo Research) was used according to the manufacturer's instructions.For quantification of individual CpG-site methylation, 35 ng of bcDNA was amplified on a Rotor-gene Q (Qiagen) using the EpiTect HRM PCR Kit (Qiagen) with specific primer pairs (Table S6) and the following thermal cycling conditions: at 95uC for 5 min, followed by 50 cycles at 95uC for 10 s, at 58uC for 30 s and at 72uC for 30 s, respectively.Specificity was controlled by high resolution melting curve analysis with 0.05uC/2 s increments.The PCR products were extracted using the Wizard SV Gel and PCR clean-Up System (Promega) according to the manufacturer's instructions.For subsequent pyrosequencing the biotinylated strand was isolated and released into annealing buffer containing the specific sequencing primer (Table S6).Determination of local genomic DNA methylation was performed on a PyroMark Q24 System (Qiagen) and all pyrosequencing reagents (Qiagen) were prepared according to the manufacturer's instructions.Calculation of individual CpG-site methylation was performed by the PyroMark Q24 Software (Qiagen).

Statistical Analysis
Results are presented as mean 6 SEM.Statistical analyses were performed using Prism 4.01 software (GraphPad Software).Data were tested using Student's t-test.In case of inhomogeneous variances, data were analyzed using Student's t-test with Welch's correction.The in vitro experiment was analyzed using one-way ANOVA with tukey's test used as post hoc test.For all tests, the bilateral alpha risk was a = 0.05.Differences in liver amino acid concentrations determined with the iTRAQ-labeling method were tested using the R version 2.9.2 (R Foundation of Statistical Computing) [31].The p-values were adjusted for multiple testing using the p.adjust function within the R-library limma and the Benjamini-Hochberg method [32].

Alterations in Gene Expression of Enzymes Operating in the Hepatic Transsulfuration and Sarcosine Pathways
In a previous study, we have analyzed the development of obesity, insulin resistance and hepatic steatosis, induced by HF diet feeding, in C57BL/6N mice and detected significant alterations of hepatic PL homeostasis and specific PC signature [5].Here, we asked whether obesity and NAFLD status influence hepatic methlgroup expenditure and C1-metabolism.To detect the impact of HF diet feeding and NAFLD status on hepatic mRNA expression of key enzymes in the C1-metabolism, we performed RT-qPCR analysis using total RNA of liver tissues from these obese and control mice.The first set of selected genes analyzed (Fig. 1) comprised genes involved (i) directly in the synthesis of SAM such as methionine adenosyltransferase I alpha (Mat1a) and methionine adenosyltransferase II alpha (Mat2a), (ii) in SAH hydrolysis or transmethylation processes such as the S-adenosylhomocysteine hydrolase (Ahcy) and Pemt and (iii) in folate-dependent homocysteine remethylation like the 5-methyltetrahydrofolate-homocysteine methyltransferase (Mtr) and methylenetetrahydrofolate reductase (Mthfr).Surprisingly, no significant changes in expression levels were detected for this group of genes (Fig. 2A, B).Serine hydroxymethyltransferase 1 (Shmt1) mRNA levels were increased by 30.7612.0%(p = 0.08) in obese mice over controls but serine hydroxymethyltransferase 2 (Shmt2) remained unaffected by dietary treatment (Fig. 2B).However, in the second set of genes representing the choline oxidation and sarcosine pathway, HF diet feeding resulted in higher mRNA expression levels of genes involved in the betaine-dependent remethylation of Hcy.This included Bhmt with expression levels elevated by 107.3617.4%;(p = 0.001), while betaine-homocysteine methyltransferase 2 (Bhmt2) was not significantly affected (p = 0.16).The choline dehydrogenase (Chdh) increased by 32.065.3%(p = 0.017), and the dimethylglycine dehydrogenase precursor (Dmgdh) mRNA level was elevated by 43.3610.6%(p = 0.017) compared to control mice (Fig. 2C).The gene expression of glycine N-methyltransferase (Gnmt), important in hepatic SAM-homeostasis, was unaffected (p = 0.29) by the dietary treatment (Fig. 2C).For genes operating in the transsulfuration pathway and involved in the taurine and glutathione synthesis (Fig. 2D), reduced mRNA levels of the cystathionine b-synthase (Cbs) gene were detected with 66.362.7% of controls and marginal significance (p = 0.08), but no significant differences for cystathionase (Cth) were measured (p = 0.29).Whereas mRNA expression levels were significantly elevated for cysteine sulfinic acid decarboxylase (Csad) by 112.5638.9%,(p = 0.038) and glutathione synthetase (Gss) by 22.465.9%(p = 0.032), glutamate-cysteine ligase (catalytic subunit, Gclc) did not reveal regulation (p = 0.92).Glutamate oxaloacetate transaminase 1 (Got1), implicated in the processing of cysteine to sulfate, showed marginally decreased mRNA levels (48.767.0% of control; p = 0.06) in HF diet fed mice (Fig. 2D).

Hepatic BHMT Protein Levels Increased and CBS Levels Decreased upon HF Diet Feeding
The observed changes in gene expression of the branch-point enzymes BHMT and CBS were further analyzed on protein level by Western blot analysis using liver protein extracts from animals on HF and control diets.As shown in Figure 3, BHMT levels were increased by 61.968.3%(p,0.001)whereas CBS protein density was decreased in DIO animals to 66.666.2%(p = 0.02) of that in controls.

Metabolite Profiling Data Suggest a Repression of the Hepatic Transsulfuration Pathway
A total of 35 amino acids and their derivatives were quantified by LC-MS/MS, GC-MS/MS in hepatic tissues (Fig. 4, Table S7).Taurine and glutamine concentrations were elevated in DIO mice by 24.765.3%(p = 0.015) and 39.969.8%;(p = 0.034), respectively, as shown in Figure 4A.Concentrations for L-aamino-n-butyrate, L-citrulline, L-ornithine and hydroxyproline significantly decreased by 30% or even 50% upon HF diet feeding compared with controls.However, methionine levels remained unaffected (Fig. 4A) in obese mice.Concentrations of the methionine cycle metabolites SAM, SAH and Hcy determined by LC-MS/MS and GC-MS, respectively, did not reveal significant differences (Fig. 4B), nor did the methylation index [SAM]/[SAH] ratio change in mice fed a HF or control diet (Fig. 4C).However, the analysis of cystathionine, a major transsulfuration metabolite, by GC-MS revealed a significantly reduced concentration in obese mice of 64.067.7% (p = 0.026) compared with controls, and in the sarcosine pathway the metabolite betaine showed significantly diminished concentration in obese mice of 57.565.9%(p,0.001) compared to the controls (Fig. 4B).Furthermore, the hepatic [betaine]/[choline] ratio was significantly decreased (p,0.001) and the [dimethylglycine]/ [betaine] ratio significantly increased (p,0.001) in HF mice (Fig. 4C).

Increased PPARa Activity may Lead to the Downregulation of the Hepatic Transsulfuration Pathway
Hepatic PPARa is known to critically control gene expression of a large variety of genes involved in lipid metabolism and in particular in vand b-oxidation of fatty acids.In addition, PPARa has been suggested to regulate the transsulfuration pathway [33].To assess whether hepatic PPARa mRNA expression levels were regulated in HF mice, we performed gene expression analysis of PPARa and some of its downstream target genes by RT-qPCR analysis (Fig. 5A, B).In comparison with control, PPARa mRNA expression was elevated to 167.4617.2%(p = 0.022) and selected PPARa target genes also showed significantly higher mRNA levels, including carnitine palmitoyltransferase 1a (Cpt1a) increased to 341.3616.1% (p = 0.009), uncoupling protein 2 (Ucp2) to 193.0623.1% (p = 0.012) and acyl-coenzyme A oxidase 1 (Acox1) levels to 182.6612.6%(p,0.001)suggesting in total an increase in hepatic PPARa activity.To further examine the underlying mechanisms causing an upregulation of BHMT and a downregulation of CBS, we explored whether PPARa-activation by the agonist WY14,643 in rat hepatoma cells (Fao) can cause similar changes in gene expression in vitro as observed in the livers of obese mice.Treatment of Fao cells with increasing concentrations of WY14,643 for 24 h indeed augmented the mRNA levels of the PPARa target gene Cpt1a (25 mM, 169.468.3%;50 mM, 158.467.7%; 100 mM, 165.369.4%;p,0.001) and reduced the mRNA of Cbs (25 mM, 32.762.6%;50 mM, 37.467.5%; 100 mM, 39.1611.7%;p,0.001) but did not change Bhmt mRNA expression (p = 0.24) in comparison with untreated control cells (Fig. 5C).

Hepatic DNA Methyltransferase 3b Gene Expression, but not Hepatic Global Genomic DNA Methylation Levels, is Altered in DIO Mice
To investigate whether DIO influences epigenetic processes in the liver, DNA methylation, gene expression of DNA methyltransferases (Dnmt) as well genomic DNA methylation were determined.Animals receiving the C diet showed stronger expression for Dnmt1 and de novo Dnmt3a than for de novo Dnmt3b.Upon HF diet feeding, Dnmt1 mRNA expression was unaltered, but gene expression of Dnmt3a and Dnmt3b decreased by 25% and 54%, respectively (Fig. 6).However, global genomic DNA methylation was not affected (p = 0.87) as shown in Figure 6D.Previously, the promoter region of the rat Cbs gene was identified to be sensitive for DNA methylation [34] suggesting that the observed lower Cbs gene expression in our study could be accompanied by local genomic DNA methylation.Therefore, we measured the CpG methylation state in the CpG island of the Cbs promoter in the region P1 and P2 and of the intragenic CBS CpG island in the region I7 (Fig. 7A) with MS-qPCR.Analysis of hepatic DNA methylation in the CpG island regions P1 and P2 of the Cbs promoter in control and HF mice revealed no detectable

Discussion
DIO in mice leads to NAFLD that is characterized by hepatic accumulation of fat which could result from increased fatty acid uptake, increased hepatic de novo fatty acid synthesis, impaired oxidation of fatty acids or a reduced VLDL-mediated TG export [36].The latter is dependent on provision of sufficient PC for assembly of VLDL particles.It has been estimated that up to 30% of the hepatic PC biosynthesis originates from the PEMT pathway by a PEMT-mediated methylation of PE using SAM as a methyldonor [9,37,38].Previously, we have shown that feeding C57BL/ 6N mice a beef tallow-based HF diet resulted in hyperglycemia, hyperinsulinemia, reduced glucose tolerance and hepatic TG accumulation [5].Furthermore, we have identified changes in the hepatic PL content and PC signature suggesting a DIO-based modulation of PC [5].Recently, Rubio-Aliaga et al. also described alterations of hepatic C1-metabolism in mice upon HF diet feeding [21] demonstrating a general link between DIO-induced NAFLD and the C1-metabolism.
We here extend these observations and provide evidence that hepatic methionine homeostasis is conserved in C57BL/6N mice on a HF diet based on a PPARa-mediated downregulation of the hepatic transsulfuration pathway and an enhanced Hcy remethylation capacity associated with an elevated taurine production despite repression of hepatic transsulfuration (Fig. 8).These findings are compatible with data of Finkelstein et al. using rats fed a low methionine diet (0.25% w/w) with provision of extra dietary cystine (0.8% or 1.3% w/w) [39,40].In these rats, a hepatic methionine-sparing effect with reduced CBS activity accompanied with increased BHMT activity was observed.The lower CBS activity could only be detected under a low dietary methionine supply [39].Our findings in mice fed a HF diet with sufficient methionine (0.86%, w/w) and cystine (0.46%, w/w) suggest that, despite an adequate methionine supply, the transsulfuration pathway is downregulated whereas BHMT-dependent Hcyremethylation is increased.Furthermore, we detected lower mRNA levels of CBS and an increase of BHMT mRNA with corresponding changes on protein level.Tang et al. also identified a comparable mechanism in methionine-deprived C57BL/6 mice with a proposed post-transcriptional downregulation of hepatic CBS and increased BHMT enzyme activity associated with a transient increase in plasma Hcy level [41].Interestingly, we detected a reduced Hcy transsulfuration and an increased Hcy remethylation capacity.This finding is supported by reduced hepatic betaine concentration, increased BHMT protein expression and increased [DMG]/[betaine] ratio in HF mice, suggesting increased BHMT activity.In rats, Finkelstein et al. showed that hepatic betaine content is influenced by Hcy remethylation and is dependent on dietary protein level [42], whereas the protein intake  of control and HF mice in our study was not significantly different (Table S2).Our finding of reduced hepatic betaine concentrations in obese mice is in line with data of Kim et al. reporting significantly lower betaine levels in livers from DIO mice [43].In general, the increased hepatic BHMT protein expression, [DMG]/[betaine] ratio and reduced betaine content in our HF mice are indicative for an enhanced BHMT mediated Hcyremethylation, however we did not find any evidence for changes in the hepatic methionine cycle in obese mice.This implies that maintaining methionine levels by reduced Hcy transsulfuration and enhanced Hcy remethylation is of prime importance for hepatocytes in NAFLD despite elevated taurine levels derived from either hepatic or extrahepatic cysteine.Since methionine is an essential amino acid, a reduction in methionine levels would critically affect protein synthesis capacity for the maintenance of the hepatic proteome as well as its protein secretion function.
The development of liver steatosis upon HF feeding is known to be associated with hepatic insulin resistance [44,45,46] and enhanced rates of gluconeogenesis.HF mice analyzed in the present study exhibited similar characteristic phenotypic alterations as described previously [5].In Zucker diabetic fatty rats representing a type 2 diabetes model and in a type 1 diabetes animal model with streptozotocin an impaired insulin secretion and signaling and an increased hepatic gene expression of CBS and BHMT representing the branch-point enzymes of C1metabolism have been reported [47,48,49,50].This is to some extent in line with our data suggesting increased Hcy remethylation and improved VLDL secretion due to increased BHMT expression in obese animals [51].Furthermore, the implication of BHMT in lipid and PL metabolism of the liver and adipose tissue was recently shown by Teng et al. [52,53].However, the identified downregulation of CBS in our study contradicts a disturbed insulin signaling as described by others [47,48,50] indicating that other mechanisms may have caused the observed decrease of CBS expression.In this context, reduced hepatic Cbs mRNA level and a reduced enzyme activity were also reported for rats fed a HF diet [54,55] or for rats on HF diets treated with the PPARa agonist WY14,643 [56].Our data also provide evidence for alterations in PPARa signaling in the liver of obese mice as shown by elevated mRNA expression of Ppara in conjunction with enhanced expression of prototypical PPARa target genes such as Cpt1a, Ucp2 or Acox1 (Fig. 5A, B).Although PPARa is primarily considered to affect genes controlling lipid metabolism, it is also a regulator of hepatic amino acid metabolism [33].For example, Kersten et al. have shown that PPARa can mediate suppressive effect on urea cycle enzymes [57], and inhibitory effects of fatty acids, which are ligands of PPARa, on ureagenesis [58] and ammonia detoxification [59] have been described.In this context,  [48,49]) and PPARa (this study) on the regulation of transcription.Cross means disrupted inhibitory effect of insulin reported in hyperglycemic mice (40)(41)(42)(43).doi:10.1371/journal.pone.0057387.g008interestingly, we detected higher concentrations for glutamine and lower citrulline and ornithine concentrations in the liver of HF mice than in controls pointing to changes in ureagenesis and ammonia detoxification upon prolonged HF feeding.In addition, glutamine is known to influence fatty acid oxidation, lipolysis and glutathione biosynthesis [60,61,62].Regarding to the mechanistic role of PPARa in our study, we could demonstrate in hepatoma cells using the PPARa agonist WY14,643 that Cbs is a target gene of the PPARa signaling pathway (Fig. 5B) leading to a downregulation of the transsulfuration pathway comparably as shown by low CBS levels in HF mice.Interestingly, in vitro studies in HepG2 cells show decreased PPARa expression dependent on Hcy concentration, establishing a possible feedback signaling of the methionine cycle on PPARa signaling via Hcy [17,63].Findings from CBS-deficient mice [64,65,66] also corroborate the prominent role of the transsulfuration pathway and especially of CBS in the development of a fatty liver.
Taken together, mice fed a HF diet display an altered equilibrium between the transmethylation (methionine cycle) and processes that relate to the transsulfuration pathway such as glutathione and taurine synthesis which seem at least partially controlled by PPARa.The decrease in hepatic L-a-amino-nbutyrate levels found in our DIO mice suggest that a-ketobutyrate production from cystathionine in the transsulfuration pathway is also reduced which could affect also hepatic ophthalmic acid levels.This alternative non-thiol is produced by glutathionesynthetase when using L-a-amino-n-butyrate rather than cysteine as a substrate [67].Reduced levels of L-a-amino-n-butyrate would promote increased glutathione levels for proper redox-balance in liver.The increased taurine levels observed in DIO mice may also be indicative for changes in stress-response.The observed downregulation of Got1 gene expression known to be regulated by PPARa activity [57] together with increased gene expression of Csad and elevated hepatic taurine concentrations in obese mice points to a reduced sulfate production from cysteine compensating reduced Hcy transsulfuration in favor of increased taurine synthesis.These changes may possibly improve osmoregulatory, cytoprotective and antioxidant capacities in the steatotic liver [68,69].In the nucleus, the principal methyl-donor, SAM, provides methyl-groups for DNA methylation and histone modification which are important for epigenetic gene expression regulation, chromatin condensation and genome integrity [70,71,72].Although we observed a downregulation of hepatic de novo Dnmts, this did not seem to influence the global DNA methylation state in our DIO mice.Although, in male Sprague-Dawley rats, diabetes-mediated perturbations in C1-metabolism resulted in hepatic DNA-hypomethylation we could not confirm this in our DIO mice [73].The analysis of CpG island DNA methylation of the Cbs gene by MS-qPCR and analyzing 115 bp of a Cbs promoter CpG island (2270 until 2155) containing a putative insulin response element (PEPCK-like [TGTTTGT] motif) [35] by bisulfite conversion/pyrosequencing of liver genomic DNA from DIO mice compared with controls revealed no obvious changes in DNA methylation in the Cbs promoter or intragenic region.However, our DNA methylation analysis along the Cbs gene by MS-qPCR may have not detected DNA methylation due to the limited detection sensitivity of the MS-qPCR method at CpG sites in specific DNA sequence areas or in case of bisulfite conversion/pyrosequencing anaylsis with lower methylation frequency compared to bisulfite-sequencing shown by Uekawa et al. [34].
In conclusion, our data demonstrate that HF diet feeding in mice can induce a suppression of gene and protein expression of enzymes operating in the hepatic transsulfuration pathway in favor of an increased remethylation of Hcy to methionine.In addition, we show that the PPARa pathway is involved in the downregulation of CBS.Based on our findings, we postulate the primacy of methionine homeostasis mediated by remethylation of Hcy to methionine in liver physiology and in pathological situations such as NAFLD to ensure the maintenance of basic liver functions such as synthesis of hepatic and plasma proteins despite previously reported altered PL homeostasis and PC signature.High dietary fat intake associated with the development of hepatic steatosis is consequently linked to major changes in hepatic C1-metabolism that secondarily may also translate into changes of hepatic redoxstatus and cellular osmolyte levels.

Figure 2 .
Figure 2. Effect of HF feeding on hepatic mRNA expression levels of C1-metabolism enzymes.Data are presented as mean 6 SEM (n = 5-6).Analysis of the methionine cycle (A), the folate cycle (B), sarcosine pathway (C), and transsulfuration pathway (D).Open and grey bars represent data from animals fed control and HF diet, respectively.Asterisk indicates statistical significance (p,0.05).doi:10.1371/journal.pone.0057387.g002 or low CpG methylation of AciI and HpaII restriction sites (Cbs promoter HpaII methylation; 8.0860.54% in control animals and 8.4660.37% in animals on HF diet) whereas the AciI restriction site analysis of the Cbs intragenic region I7 showed CpG methylation of 61.6864.25% in control liver genomic DNA and 65.2463.95% in genomic DNA of HF diet liver samples.However, no significant differences of the investigated CpG sites between the groups (Cbs P1 HpaII, p = 0.19; Cbs P2 HpaII, p = 0.72; Cbs I7 AciI, p = 0.55) could be observed as shown in Figure 7B.In addition, we analyzed 115 bp of the Cbs promoter CpG island (2270 until 2155) containing a putative insulin response element (PEPCK-like [TGTTTGT] motif) [35] on the reverse strand flanked by CpGs by bisulfite conversion and subsequent pyrosequencing.The specific methylation analysis revealed only low CpG-methylation with no differences between the experimental groups except for a marginal decrease in methylation of CpG-site #6 on the forward strand (p = 0.064) as shown in Figure.7C.

Figure 7 .
Figure 7.DNA methylation analysis in Cbs gene promoter and intragenic CpG islands of hepatic genomic DNA.Data are presented as mean 6 SEM.Open and grey bars represent control and HF animals, respectively.Asterisk indicates statistical significance (p,0.05).From top to