Hypermethylation of Fads2 and Altered Hepatic Fatty Acid and Phospholipid Metabolism in Mice with Hyperhomocysteinemia*

Alterations in lipid metabolism may play a role in the vascular pathology associated with hyperhomocysteinemia (HHcy). Homocysteine is linked to lipid metabolism through the methionine cycle and the synthesis of phosphatidylcholine (PC) by phosphatidylethanolamine (PE) methyltransferase, which is responsible for the synthesis of 20–40% of liver PC. The goal of the present study was to determine if the reduced methylation capacity in HHcy is associated with alterations in liver phospholipid and fatty acid metabolism. Mice heterozygous for disruption of cystathionine β-synthase (Cbs+/-) fed a diet to induce HHcy (HH diet) had higher (p < 0.001) plasma total homocysteine (30.8 ± 4.4 μm, mean ± S.E.) than C57BL/6 mice (Cbs+/+) fed the HH diet (7.0 ± 1.1 μm) or Cbs+/+ mice fed a control diet (2.3 ± 0.3 μm). Mild and moderate HHcy was accompanied by lower adenosylmethionine/adenosylhomocysteine ratios (p < 0.05), higher PE (p < 0.05) and PE/PC ratios (p < 0.01), lower PE methyltransferase activity (p < 0.001), and higher linoleic acid (p < 0.05) and lower arachidonic acid (p < 0.05) in PE. Mice with moderate HHcy also had higher linoleic acid and α-linolenic acid (p < 0.05) and lower arachidonic acid and docosahexaenoic acid (p < 0.05) in liver PC. The first step in the desaturation and elongation of linoleic acid and linolenic acid to arachidonic acid and docosahexaenoic acid, respectively, is catalyzed by Δ(6)-desaturase (encoded by Fads2). We found hypermethylation of the Fads2 promoter (p < 0.01), lower Fads2 mRNA (p < 0.05), and lower Δ(6)-desaturase activity (p < 0.001) in liver from mice with HHcy. These findings suggest that methylation silencing of liver Fads2 expression and changes in liver fatty acids may contribute to the pathology of HHcy.

Alterations in lipid metabolism may play a role in the vascular pathology associated with HHcy. Previous studies have shown that humans and animals with HHcy have hepatic lipid accumulation, increased hepatic cholesterol biosynthesis, and decreased plasma high density lipoprotein (HDL) cholesterol and apoA-I levels with minimal changes in plasma total cholesterol levels (8 -13). Homocysteine is linked to lipid metabolism through the methionine cycle. Within the cycle, methionine is converted to S-adenosylmethionine (AdoMet), which serves as a methyl donor for numerous methyl acceptors, including phospholipids, DNA, RNA, protein, histones, and neurotransmitters (14). S-Adenosylhomocysteine (AdoHcy) is produced as a byproduct of methyl donation, and homocysteine is formed through the (reversible) liberation of adenosine from AdoHcy. In human subjects with HHcy, intracellular concentrations of AdoHcy may increase, resulting in a lower AdoMet/AdoHcy ratio, diminished methylation capacity, and global DNA hypomethylation (15,16). Similarly, mice with HHcy have reduced AdoMet/AdoHcy ratios, diminished methylation capacity, global DNA hypomethylation, and reduced methylation of the genomically imprinted H19 gene (4,6,17,18).
Phosphatidylcholine (PC) synthesis in liver occurs by two pathways, both of which are linked to homocysteine via the methionine cycle. One pathway of PC synthesis is liver-specific and involves the methylation of phosphatidylethanolamine (PE) with three AdoMet molecules to form PC by phosphatidylethanolamine N-methyltransferase (PEMT, EC 2.1.1.17) (19). The alternative pathway, the CDP-choline pathway, is responsible for the synthesis of the majority of liver PC but requires preformed choline derived from the diet or the PEMT pathway. Through this requirement for choline, the CDP-choline pathway is linked to homocysteine metabolism, because choline can be oxidized to form betaine, which can also serve as a methyl donor in the remethylation of homocysteine to methionine. The rate-limiting step in the CDP-choline pathway is the conversion of phosphocholine to CDP-choline, catalyzed by CTP:phosphocholine cytidylyltransferase (EC 2.7.7.15), following which CDP-choline is transferred to 1,2-diacylglycerol (19). Phosphatidylcholine synthesized via the PEMT pathway has higher levels of the long-chain polyunsaturated fatty acids arachidonic acid (20:4n-6) and docosahexaenoic acid (22:6n-3) at the sn-2 position, whereas PC produced via the CDP-choline pathway has higher monounsaturated and saturated fatty acids such as oleic acid (18:1n-9) and stearic acid (18:0) in the sn-1 position (20,21).
In the present study we tested the hypothesis that the reduced tissue methylation capacity in HHcy is accompanied by changes in the methylation and expression of Fads2 and alterations in liver phospholipids and long chain fatty acids. We chose to focus our study on Fads2, because it encodes ⌬(6)desaturase (EC 1.14.19.3), which catalyzes the first step in the desaturation and elongation of the dietary essential fatty acids, linoleic acid (18:2n-6) and linolenic acid (18:3n-3), to arachidonic acid (20:4n-6) and docosahexaenoic acid (22:6n-3), respectively (22). We speculate that alterations in liver phospholipid and fatty acid metabolism may contribute to the vascular pathology observed in HHcy because of the central role of liver in secreting phospholipids and fatty acids into plasma for subsequent delivery to other tissues. We used Cbs ϩ/Ϫ mice in these studies because they are susceptible to diet-induced increases in plasma tHcy, have reduced tissue methylation capacity, and show enhanced sensitivity to homocysteine-related endothelial dysfunction in aorta and mesenteric and cerebral arterioles (4,5,18).
Our findings demonstrate that the reduced methylation capacity in mice with HHcy is associated with hypermethylation of Fads2, lower Fads2 mRNA, and lower ⌬(6)-desaturase activity in liver. We also found higher PE/PC ratios in liver, primarily due to higher levels of PE with little changes in PC, and lower levels of PEMT activity in mice with HHcy. This was accompanied by higher levels of 18:2n-6 and 18:3n-3 and lower levels of 20:4n-6 and 22:6n-3 in liver PC, and higher levels of 18:2n-6 and lower levels of 20:4n-6 in liver PE. These findings suggest that the reduced hepatic methylation capacity in HHcy is associated with methylation silencing of Fads2 expression resulting in impaired synthesis of long chain polyunsaturated fatty acids and elevation of PE due to reduced synthesis of PC via the PEMT pathway.

EXPERIMENTAL PROCEDURES
Mice and Experimental Protocol-Mice heterozygous for targeted disruption of the Cbs gene (Cbs ϩ/Ϫ ) (23), on a C57BL/6 background, and their wild-type C57BL/6 littermates (Cbs ϩ/ϩ ) were used in the study. Genotyping for the wild-type Cbs allele was conducted by PCR as described previously (18). Briefly, genotyping for the disrupted Cbs allele, Cbs tm1Unc , was accomplished by PCR using the P10 primer and a primer corresponding to the neo cassette used in the targeted disruption (23). At weaning, Cbs ϩ/Ϫ and Cbs ϩ/ϩ mice were fed either a control diet (TD 05108, Harlan Teklad, Madison, WI) or a hyperhomocysteinemic (HH) diet (TD 00205, Harlan Teklad) for 3-9 weeks. The HH diet has been previously used to induce HHcy and contained double the amount of methionine (8.2 g/kg), low amounts of folic acid (0.2 mg/kg), and succinyl sulfathiazole (5.0 g/kg) to inhibit growth of intestinal bacteria, another source of folic acid (6,18). The control and HH diets contained identical amounts of choline, riboflavin, pyridoxine, cobalamin, and cysteine, but the control diet contained 4.0 g/kg methionine and 4.0 mg/kg folic acid. Both diets were devoid of cholesterol and supplied 14 and 17%, respectively, of total energy from soybean oil, which contains 55% 18:2n-6 and 8% 18:3n-3. At 6 -12 weeks of age, mice were anesthetized with 1% Avertin (2,2,2-tribromoethanol, 0.3 ml/10 g body weight, intraperitoneally), and blood was collected by cardiac puncture into EDTA (final concentration, 5 mM). Blood was centrifuged at 3000 ϫ g for 20 min at 4°C, and plasma was collected, immediately flash frozen in liquid nitrogen, and stored at Ϫ80°C until later analysis of tHcy. Samples of liver were immediately deproteinized in ice-cold 0.4 M perchloric acid, homogenized, and centrifuged. The supernatant fraction was immediately frozen and stored at Ϫ80°C for later analysis for AdoMet and AdoHcy. Additional samples of liver were flash frozen in liquid nitrogen and stored at Ϫ80°C for later extraction of genomic DNA and RNA. The protocol was approved by the University of British Columbia Animal Care Committee.
Biochemical Analyses-Plasma tHcy, defined as the total concentration of homocysteine after quantitative reductive cleavage of all disulfide bonds, was measured using a modified high-performance liquid chromatography method with fluorescence detection (24). AdoMet and AdoHcy levels in liver were determined by high-performance liquid chromatography using UV detection as described (25).
Total lipids were extracted from total liver and liver microsomal fractions (prepared as described below) by the method of Folch et al. (26). The organic phase was evaporated under nitrogen, the lipids were solubilized in chloroform:methanol:acetone:hexane (2.0:3.0:0.5:0.5, v/v), and individual classes of lipids were separated using a Waters 2690 Alliance high-performance liquid chromatograph (Waters Limited, Milford, MA). The separated lipid classes were detected and quantified by evaporative light scattering detection (Model 2000, Alltech, Mandel Scientific, Guelph, Canada) as described previously (27). Fatty acids were quantified as their respective methyl esters by gasliquid chromatography using heptadecanoic acid (17:0) as the internal standard on a Varian 3400 gas-liquid chromatograph (Varian Canada, Mississauga, Canada) equipped with a flame ionization detector (27).
For analysis of PEMT activity, microsomal fractions of liver samples were prepared and PEMT activity determined by the method of Ridgeway and Vance (28). Samples of liver were homogenized in a 225 mM sucrose and 25 mM Tris buffer at pH 7.8 containing 10 mM glutathione, 0.5 g/ml leupeptin, and 2 g/ml aprotinin. Homogenized samples were centrifuged at 2,000 ϫ g for 20 min to remove debris followed by centrifugation of the supernatant at 10,000 ϫ g for 20 min to pellet mitochondrial and nuclear fractions. The supernatant was then centrifuged at 105,000 ϫ g for 60 min to pellet microsomes. The microsomal pellets were resuspended in a buffer containing 100 mM sucrose, 50 mM potassium chloride, 40 mM potassium dihydrogen phosphate, and 30 mM EDTA at pH 7.2, and protein concentrations were determined using the protein assay kit (Bio-Rad) based on the method of Bradford (29). For PEMT activity, 25 g of protein of microsomal fractions was incubated with phosphatidylmonomethylethanolamine and S-adenosyl-[methyl-3 H]methionine at 37°C for 10 min. The reaction was terminated by the addition of chloroform:methanol (2:1, v/v). The methylated lipids were then extracted by the method of Folch (26) followed by separation by TLC using chloroform: methylacetate:n-propanol:methanol:0.25% potassium chloride (28:25:25:10:7, v/v), and the phospholipid bands were visualized by exposure to 2,7-dichlorofluroscein (Supelco Inc.). Individual bands were scraped, and the incorporation of S-adenosyl[methyl-3 H]methionine into phospholipids was quantified using a Microbeta TriLux Scintillation counter (PerkinElmer Life Sciences). ⌬(6)-Desaturase activity was also determined in microsomal fractions (1 mg of protein) following the previously described method of Garg et al. (30).
Bisulfite Pyrosequencing-The CpG-rich region of the Fads2 promoter was identified by in silico analysis using the University of California, Santa Cruz, CA, genome browser (31) and MethPrimer (32). The percent methylation of CpG sites within the Fads2 promoter ( Fig. 4) were quantified by bisulfite pyrosequencing (33). Genomic DNA was extracted from liver using the DNeasy kit and included RNase I treatment (Qiagen). Genomic DNA samples (0.5-1 g) were bisulfite-treated using the EZTM DNA Methylation Kit (Cedarlane Laboratories Ltd, Hornby, Canada) and stored at Ϫ20°C until further analysis. A 253-bp fragment corresponding to a portion of the Fads2 promoter between Ϫ477 and Ϫ224 relative to the transcriptional start site ( Fig.  4A) was amplified by PCR from bisulfite-treated DNA. The PCR reaction used HotStar Taq DNA polymerase (Qiagen) and the following primers: PMFads2F, 5Ј-GAAAGATTTT-TTTGGGTTAATGG-3Ј and PMFads2RB, 5Ј-ACTTCCCAA-TACCCCCAAAC-3Ј (IDT Inc., Coralville, IA), with the reverse primer containing a biotin at the 5Ј-end. Cycling conditions were 94°C for 15 min followed by 50 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 1 min with a final extension of 10 min at 72°C. PCR products were purified and sequenced using a PyroMark MD System (Biotage, Foxboro, MA) following the manufacturer's suggested protocol and the sequencing primer: PMFads2S, 5Ј-TGTTGATTATTGTGGAAAT-3Ј (IDT Inc.). The sequencing primer was designed to sequence a 194-bp region of the 253-bp PCR product that contains 11 CpG sites (Fig. 4A). Samples were analyzed in duplicate, and the percent methylation at each CpG site was quantified using the Pyro Q-CpG software, version 1.0.9 (Biotage).
Real-time PCR Quantification of mRNA-Total RNA was extracted from liver using the RNeasy Mini Kit (Qiagen) and included DNase I treatment to remove contaminating genomic DNA. Integrity of the RNA was assessed by confirming the presence of 18 S and 28 S rRNA on agarose gels. Synthesis of cDNA was accomplished using the High Capacity cDNA Archive Kit (Applied Biosystems). Liver RNA samples (1.0 g) were incubated with 5.0 units/l Multiscribe Reverse Transcriptase, 1ϫ random primer mix, 1ϫ dNTP mix, 1ϫ Reverse Transcription Buffer in a final reaction volume of 80 l. Samples were incubated at 25°C for 10 min followed by incubation at 37°C for 2 h and then stored at Ϫ20°C until later quantification of transcripts.
Fads2, Pcty1a, and Pemt mRNA levels were quantified by real-time PCR using the comparative Ct method (⌬⌬Ct) (34) of relative quantification and commercially available primers and TaqMan MGB probes (FAM-fluorescein dye-labeled) from Applied Biosystems specific for Fads2 (Mm00517221), Pcyt1a (Mm00447774), and Pemt (Mm00839436). ␤-Actin served as the endogenous control in the ⌬⌬Ct assay with levels quantified using mouse ACTB endogenous control primers and TaqMan MGB probes (FAM-dye-labeled, Applied Biosystems). Samples (0.1 l of cDNA reaction) were incubated with 1ϫ TaqMan Universal PCR mix and 1ϫ TaqMan primer/probe mix in a final reaction volume of 20 l. Samples were placed at 50°C for 2 min followed by incubation at 95°C for 10 min and 40 cycles of 95°C for 15 s and 60°C for 1 min in a 7500 Real Time PCR System (Applied Biosystems). Quantification of relative changes in expression were determined following the manufacturer's suggested protocol for a ⌬⌬Ct assay, and data were analyzed using the 7500 System Sequence Detection software, version 1.2.3 (Applied Biosystems). Each sample was run in duplicate, and the experiment was repeated three separate times.
Statistical Analysis-One-way analysis of variance followed by the least significant difference test for multiple comparisons was used to compare findings in Cbs ϩ/ϩ mice fed the control diet, Cbs ϩ/ϩ fed the HH diet, and Cbs ϩ/Ϫ mice fed the HH diet. These analyses were accomplished by SPSS for Windows version 12.0.1 (SPSS Inc., Chicago, IL).  (Fig. 1A). The elevation in plasma tHcy in both the Cbs ϩ/ϩ and Cbs ϩ/Ϫ mice fed the HH diet was accompanied by a significant reduction in liver methylation capacity as determined by AdoMet/AdoHcy ratio. AdoMet/AdoHcy ratios in liver were significantly lower in Cbs ϩ/ϩ mice (p Ͻ 0.05) and Cbs ϩ/Ϫ mice (p Ͻ 0.001) fed the HH diet than in Cbs ϩ/ϩ mice fed the control diet (Fig. 1D). These differences were due primarily to higher levels of AdoHcy in mice fed the HH diet, levels of which were highest in Cbs ϩ/Ϫ mice fed the HH diet and significantly different than Cbs ϩ/ϩ mice fed the HH diet (p Ͻ 0.05) and Cbs ϩ/ϩ mice fed the control diet (p ϭ 0.001) (Fig. 1C). No significant differences in levels of AdoMet in liver were observed between Cbs ϩ/Ϫ mice fed the HH diet and Cbs ϩ/ϩ mice fed the HH diet and Cbs ϩ/ϩ mice fed the control diet (Fig. 1B). We also investigated whether HHcy was accompanied by changes in levels of major lipid classes in liver. Cbs ϩ/Ϫ mice fed the HH diet had significantly higher (p Ͻ 0.01) levels of triglyceride in liver than Cbs ϩ/ϩ mice fed the control diet (Table 1). We found no significant differences in levels of total cholesterol, free cholesterol, free fatty acids, or total phospholipids between Cbs ϩ/Ϫ mice fed the HH diet and Cbs ϩ/ϩ mice fed the HH diet and Cbs ϩ/ϩ mice fed the control diet (Table 1).

Effect of Diet and Cbs Genotype on Plasma tHcy Levels, Methylation Capacity, and Lipid Classes in
HHcy and Liver Phospholipids-The reduced liver methylation capacity in mice with HHcy was accompanied by higher PE/PC ratios in liver. PE/PC ratios were significantly higher in Cbs ϩ/Ϫ mice (p Ͻ 0.001) and Cbs ϩ/ϩ mice (p Ͻ 0.01) fed the HH diet than Cbs ϩ/ϩ mice fed the control diet (Fig. 2C). These differences were primarily due to significantly higher PE levels in liver of Cbs ϩ/Ϫ mice (p Ͻ 0.001) and Cbs ϩ/ϩ mice (p ϭ 0.01) fed the HH diet than Cbs ϩ/ϩ mice fed the control diet ( Fig. 2A) with no differences in liver PC levels between the groups (Fig.  2B). Notably, the PE/PC ratios were significantly inversely asso-ciated with the AdoMet/AdoHcy ratios in liver (p ϭ 0.001, r ϭ 0.703). We also investigated whether mice with HHcy showed differences in expression of the genes encoding Pemt, responsible for the methylation of PE to PC, and, Pcyt1a, responsible for the rate-limiting step in the synthesis of PC via the CDPcholine pathway in liver (19). Liver levels of Pemt mRNA were not significantly different between Cbs ϩ/Ϫ mice and Cbs ϩ/ϩ mice fed the HH diet and Cbs ϩ/ϩ mice fed the control diet (Fig.  3A). Similarly we found no differences in levels of Pcyt1a mRNA between the diet/genotype groups (Fig. 3C). However, we did find significantly lower levels of PEMT activity in liver of Cbs ϩ/Ϫ mice (p Ͻ 0.001) and Cbs ϩ/ϩ mice (p ϭ 0.001) fed the HH diet than Cbs ϩ/ϩ mice fed the control diet (Fig. 3B). This effect was greatest in Cbs ϩ/Ϫ mice fed the HH diet with levels of PEMT significantly lower (p Ͻ 0.05) than in Cbs ϩ/ϩ mice fed the HH diet. Given these differences and the fact that liver PEMT is mostly found on the ER, we also quantified the phospholipid fractions in liver microsomes. Similar to what was observed in whole liver we found significantly higher (p Ͻ 0.001) PE/PC ratios (Fig. 2F) due primarily to significantly higher (p Ͻ 0.001) PE levels in liver microsomes (Fig. 2D). Only Cbs ϩ/Ϫ mice fed the HH diet had significantly lower (p Ͻ 0.05) levels of PC in liver microsomes (Fig. 2E). Liver Phospholipid Fatty Acids-Liver PC produced via the PEMT pathway has higher levels of long chain polyunsaturated fatty acids 20:4n-6 and 22:6n-3 at the sn-2 position, whereas PC produced via the CDP-choline pathway has higher levels of monounsaturated and saturated fatty acids such as 18:1n-9 and 18:0, at the sn-1 position (20,21). As such we investigated whether HHcy in mice produced differences in PE and PC fatty acids in total liver and liver microsomal fractions. Cbs ϩ/Ϫ mice fed the HH diet had significantly lower levels of 20:4n-6 (p Յ 0.01) and 22:6n-3 (p Ͻ 0.05), and higher levels of 18:2n-6 (p ϭ 0.01) in total   DECEMBER 21, 2007 • VOLUME 282 • NUMBER 51 liver PC than Cbs ϩ/ϩ mice fed the control and HH diets ( Table 2). Similar findings were observed in microsomal PC, but the degree of difference was even greater ( Table 2). We also found significantly higher (p Ͻ 0.05) levels of 18:3n-3 in total liver and microsomal PC of Cbs ϩ/Ϫ mice fed the HH diet than Cbs ϩ/ϩ mice fed the HH diet (Table 2) and significantly higher (p Ͻ 0.05) levels of eicosapentaenoic acid (20: 5n-3) in Cbs ϩ/Ϫ mice fed the HH diet than Cbs ϩ/ϩ mice fed the control and HH diets ( Table 2). The levels of 18:1n-9 in total liver and microsomal PC were not significantly different among the Cbs ϩ/Ϫ and Cbs ϩ/ϩ mice fed either experimental diet ( Table 2). Similar to findings for total liver and microsomal PC, levels of 20:4n-6 were significantly lower (p ϭ 0.001) in total liver and microsomal PE of Cbs ϩ/ϩ and Cbs ϩ/Ϫ mice fed the HH diet than mice fed the control diet with this effect greatest in liver microsomal PE from Cbs ϩ/Ϫ mice fed the HH diet (Table 3). This was accompanied by significantly higher levels of 18:2n-6 in total liver and microsomal PE of Cbs ϩ/Ϫ mice (p Ͻ 0.001) and Cbs ϩ/ϩ mice (p Ͻ 0.05) fed the HH diet than Cbs ϩ/ϩ mice fed the control diet ( Table 3). Levels of total liver and microsomal PE 18:2n-6 were also significantly higher (p Ͻ 0.05) in Cbs ϩ/Ϫ mice fed the HH diet than Cbs ϩ/ϩ mice fed the HH diet (Table  3). In contrast to what we found in total liver PC, we found no significant differences in levels of 18:3n-3 and 22:6n-3 in total liver PE. However, in microsomal PE, levels of 22:6n-3 were significantly lower (p Ͻ 0.001) in Cbs ϩ/ϩ and Cbs ϩ/Ϫ mice fed the HH diet than mice fed the control diet and significantly lower (p Ͻ 0.01) in Cbs ϩ/Ϫ mice than Cbs ϩ/ϩ mice fed the HH diet ( Table 3). Levels of 18:3n-3 and 20:5n-3 were significantly higher (p Ͻ 0.01) in liver microsomal PE and 20:5n-3 significantly higher (p ϭ 0.01) in total liver PE from Cbs ϩ/Ϫ mice fed the HH diet than Cbs ϩ/ϩ mice fed the control diet (Table 3).

HHcy and Hepatic Fatty Acid and Phospholipid Metabolism
However, it is important to note, that these fatty acids comprise a very small percentage of the total fatty acids in total liver and microsomal PE.
Fads2 Methylation and Expression-Because we found reduced levels of 20:4n-6 and 22:6n-3 and increased levels 18:2n-6 and 18:3n-3 in total liver and microsomal phospholipids of mice with HHcy (Tables 2 and 3), we also investigated the expression of the Fads2 gene, which encodes ⌬(6)-desaturase the enzyme that catalyzes the first step of the desaturation and elongation of 18:2n-6 and 18:3n-3 to 20:4n-6 and 22:6n-3, respectively (22). We also determined whether the promoter of Fads2 is differentially methylated in mice with HHcy, because we also found reduced methylation capacity in liver (Fig. 1). We quantified the methylation status of 11 CpGs located in the promoter region of the Fads2 gene between Ϫ417 and Ϫ224 relative to the transcriptional start site (Fig. 4A). Overall, we found that methylation of the Fads2 promoter was greater (p Ͻ 0.01) in Cbs ϩ/Ϫ mice fed the HH diet with moderate HHcy than in Cbs ϩ/ϩ mice fed the control diet (Fig. 4B). Individually, the methylation status of CpG sites 1, 3-6, 8, 10, and 11 were significantly greater in Cbs ϩ/Ϫ mice fed the HH diet than Cbs ϩ/ϩ mice fed the control diet. The methylation status of the Fads2 promoter also tended to be greater in Cbs ϩ/ϩ mice fed the HH diet with mild HHcy than in Cbs ϩ/ϩ mice fed the control diet, but the differences did not reach statistical significance. These differences in Fads2 promoter methylation were accompanied by reduced expression of Fads2. Levels of Fads2 mRNA were significantly lower in Cbs ϩ/Ϫ mice (p Ͻ 0.01) and Cbs ϩ/ϩ mice (p Ͻ 0.05) fed the HH diet (Fig. 4C). Similarly, ⌬(6)-desaturase activity was significantly lower (p Ͻ 0.001) in liver microsomes from Cbs ϩ/ϩ and Cbs ϩ/Ϫ mice fed the HH diet than mice fed the control diet (Fig. 4D). Further, a greater reduction in ⌬(6)desaturase activity was observed in Cbs ϩ/Ϫ mice fed the HH diet than in Cbs ϩ/ϩ mice fed the HH diet (Fig. 4D). To further support the concept of reduced production of 20:4n-6 and 22:6n-3 in liver from mice with HHcy we also quantified fatty acids in total liver, independent of the lipid class. We found that mice with moderate HHcy (Cbs ϩ/Ϫ mice fed the HH diet) had significantly lower (p Յ 0.01) levels of both 20:4n-6 and 22:6n-3 than Cbs ϩ/ϩ mice fed the control diet. Together these findings suggest that the reduced levels of 20:4n-6 and 22:6n-3 in liver from mice with HHcy may in part be a result of diminished capacity to desaturate their respective precursor fatty acids, 18:2n-6 and 18:3n-3, because of methylation silencing of Fads2 expression and consequent reduced ⌬(6)-desaturase activity.

DISCUSSION
This work sought to determine whether the reduced tissue methylation capacity in HHcy is accompanied by alterations in liver phospholipid and fatty acid metabolism in a mouse model of HHcy. Alterations in lipid metabolism have previously been observed in humans and animals with HHcy and include hepatic lipid accumulation, increased hepatic cholesterol biosynthesis, and reduced plasma HDL cholesterol and apoA-I levels (8 -13). We chose to study liver phospholipid and fatty acid metabolism because of the central role of liver phospholipids Values shown are means ϩ S.E.; *, p Յ 0.001 versus Cbs ϩ/ϩ mice fed the control diet; **, p Ͻ 0.05 versus Cbs ϩ/ϩ mice fed the HH diet, as determined by one-way analysis of variance.

TABLE 2 Changes in total and microsomal PC fatty acids in liver from mice with HHcy
Values are % total fatty acids given as means Ϯ S.E. and fatty acids in governing the levels and composition of circulating lipoproteins and subsequent delivery of lipids to other tissues. There are three key findings of this study. The first major finding is that the reduced methylation capacity in liver from mice with mild and moderate HHcy is associated with changes in liver phospholipid species. We found higher PE/PC ratios in total liver and microsomal fractions from mice with mild and moderate HHcy, primarily due to higher levels of PE with little changes in PC. The second major finding is that mice with mild and moderate HHcy have altered long-chain polyunsaturated fatty acid metabolism in liver. Specifically, we observed that mice with HHcy had significantly lower levels of 20:4n-6 and 22:6n-3 in liver and in liver PC than mice with normal homocysteine levels. This may be a reflection of reduced synthesis of liver PC via the PEMT pathway. PEMT is believed to preferentially utilize PE acylated with longer chain polyunsaturated fatty acids, like 20:4n-6 and 22:6n-3, as a substrate for synthesis of PC (20,21), and we did find lower levels of PEMT activity in liver from mice with HHcy. However, we did not find higher levels of 18:1n-9 and 18:0 in liver PC, as would be expected with increased synthesis of PC via the CDP-choline pathway. Most relevant, and the third major finding of this study, is that mice with HHcy have increased Fads2 promoter methylation and lower levels of liver Fads2 mRNA. Fads2 encodes ⌬(6)-desatu-rase, which catalyzes the first step in the desaturation and elongation of 18:2n-6 and 18:3n-3 to 20:4n-6 and 22:6n-3, respectively (22), and, therefore, suggests impaired synthesis of longchain polyunsaturated fatty acids. This conclusion is further supported by the findings of lower levels of 20:4n-6 and 22:6n-3 and lower ⌬(6)-desaturase activity in liver from mice with HHcy.

Diet/Cbs genotype group
We have shown previously that long-term (8 -12 months) feeding of a diet low in folic acid and high in methionine (HH diet) to Cbs ϩ/Ϫ mice produces severe HHcy (95 Ϯ 12 M) (18). In the current study we investigated the effects of short-term (3-9 weeks) feeding of the HH diet to both Cbs ϩ/ϩ and Cbs ϩ/Ϫ mice. The moderate HHcy achieved in the Cbs ϩ/Ϫ mice fed the HH diet (30.8 Ϯ 4.4 M) is similar to levels of plasma tHcy observed in patients with renal disease and who are at increased risk for cardiovascular events (35,36). As predicted, even mild HHcy, as that observed in the Cbs ϩ/ϩ mice fed the HH diet (7.0 Ϯ 1.1 M), and moderate HHcy were associated with reduced AdoMet/AdoHcy ratios in liver (Fig. 1D), indicating reduced methylation capacity.
In the current study, we hypothesized that reduced methylation capacity would be accompanied by reduced liver phospholipid methylation. We did find higher liver PE/PC ratios but this was primarily a consequence of higher levels of liver PE with little changes in liver PC (Fig. 2, A-C). Given that most of liver PEMT is found in the ER, we also quantified the phospholipids in liver microsomal fractions and found similar differences (Fig. 2, D-F). As a first step toward understanding the effect of HHcy on liver PC synthesis, we investigated the expression of the genes Pemt, responsible for the methylation of PE to PC, and Pcyt1a, responsible for the rate-limiting step in the synthesis of PC via the CDP-choline pathway (19). We found no significant differences in levels of Pcyt1a mRNA or Pemt mRNA in liver from Cbs ϩ/ϩ or Cbs ϩ/Ϫ mice fed either experimental diets (Fig. 3). However, we did find significantly lower PEMT activity in mice with HHcy. The biological significance of the reduced levels of liver PC in mice with HHcy may be related to the requirement of liver PC for liver lipoprotein synthesis and secretion (37,38). A previous study in Cbs Ϫ/Ϫ mice has shown reduced levels of apoB100 in liver but higher levels of serum very low density lipoprotein suggesting HHcy impairs the hydrolysis of very low density lipoprotein rather than enhancing liver very low density lipoprotein synthesis (12). The plasma tHcy levels were not given in this study, but previous characterizations of Cbs Ϫ/Ϫ mice have reported plasma tHcy levels of 203 M (23), which are more than five times the levels we found in Cbs ϩ/Ϫ mice fed the HH diet in the current study (Fig. 1A). On the other hand, a previous study in Cbs ϩ/Ϫ mice with diet-induced HHcy, similar to the current study, found increased hepatic very low density lipoprotein triglyceride secretion rates but attributed this to increased liver synthesis of triglyceride and cholesterol (10). Similarly, we did find elevated levels of triglyceride in liver from mice with HHcy (Table 1). More relevant however, may be the effect of HHcy-related changes in liver PC metabolism on HDL levels. Impairments in PC synthesis observed in Pemt Ϫ/Ϫ or liver-specific Pcyt1a Ϫ/Ϫ mice are associated with reductions in plasma HDL (39,40). Epidemiological studies have shown a negative association between elevations in plasma tHcy and HDL cholesterol and apolipoprotein A-I levels in human pop-ulations (8,11,41). Similarly, other mouse models of HHcy were shown to have lower plasma HDL cholesterol and apoA-I levels (8,11). Therefore, HHcy-related changes in liver PC metabolism may be responsible for changes in HDL metabolism observed by others.
The HHcy-related differences in liver phospholipid species were also accompanied by changes in liver phospholipid fatty acids. Mice with HHcy had decreased levels of 20:4n-6 and 22:6n-3 in total liver and microsomal PC. This change in PC composition may result from decreased PC synthesis via PEMT, which prefers PE with long-chain polyunsaturated fatty acids, like 20:4n-6 and 22:6n-3, as substrates (20,21). Therefore, reduced synthesis of PC via the PEMT pathway would be expected to result in reduced levels of PC acylated with fatty acids such as 20:4n-6 and 22:6n-3. We did find significantly lower levels of 20:4n-6 and 22:6n-3 in total liver and microsomal PC and significantly lower PEMT activity in mice with HHcy.
We have shown previously that reduced liver AdoMet/ AdoHcy ratio is associated with gene-specific changes in the methylation of the imprinted gene, H19 (18). In the current study we determined whether HHcy is associated with genespecific changes in the methylation of a gene with a potential role in the vascular pathology associated with HHcy. We investigated the methylation and expression of the Fads2 gene, which encodes ⌬(6)-desaturase, required for the conversion of 18:2n-6 and 18:3n-3 to 20:4n-6 and 22:6n-3, respectively. We did find increased methylation of the Fads2 promoter and decreased levels of Fads2 mRNA in liver from mice with HHcy (Fig. 4), suggesting methylation silencing of Fads2 expression. HHcy has been shown by us and others to be associated with reduced tissue methylation capacity and global DNA hypomethylation (4,6,17,18). The concept that cells could have global DNA hypomethylation but simultaneous hypermethylation of certain gene promoters is well established, especially in certain cancers, including liver cancer (42,43). The hypermethylation of the Fads2 promoter and reduced Fads2 mRNA in mice with HHcy was also accompanied by reduced levels of ⌬(6)-desaturase activity and reduced levels of 20:4n-6 and 22:6n-3 in total liver (Fig. 5). Together these findings suggest that the lower levels of 20:4n-6 and 22:6n-3 in liver from mice with HHcy may be the consequence of HHcy-induced silencing of Fads2 expression.
In summary, this study showed that the reduced methylation capacity in mice with HHcy is accompanied by methylation silencing of liver Fads2 and changes in liver phospholipid and fatty acid metabolism. Higher levels of PE were found in liver from mice with moderate and mild HHcy accompanied by significantly lower levels of PEMT activity. Significantly lower levels of 20:4n-6 and 22:6n-3, higher Fads2 promoter methylation, lower Fads2 mRNA levels and lower ⌬(6)-desaturase activity levels were also found in liver from mice with HHcy. Collectively, these findings suggest that mice with HHcy have impaired desaturation of 18:2n-6 and 18:3n-3 because of methylation silencing of Fads2 expression, resulting in decreased availability of the preferred PE substrate required by the PEMT pathway for liver PC synthesis. Given our current finding of reduced liver 20:4n-6, the next question will be to determine if this is also accompanied by decreased circulating levels of 20:4n-6 in lipoproteins and subsequent diminished delivery of 20:4n-6 to vascular tissue. The potential relevance to HHcy-related cardiovascular pathology would be significant, because 20:4n-6 is the precursor for the cytochrome P450 epoxygenase-derived eicosanoids such as epoxyeicosatrienoic acid, which have anti-inflammatory properties (44) and are candidates for endothelium-derived hyperpolarizing factor (45), previously shown to have a potential role in HHcy-related endothelial dysfunction (46).