Abstract

Elevated serum homocysteine, an intermediate of cellular one-carbon metabolism, is an independent risk factor for cardiovascular disease (CVD). Folate deficiency increases serum homocysteine and may contribute to CVD progression. Vascular smooth muscle cells (VSMCs) regulate vascular contractility, but also contribute to repair processes in response to vascular injury. Nutritional deficiencies, like folate deficiency, are thought to impact on this phenotypic plasticity, possibly by epigenetic mechanisms. We have investigated the effect of folate deficiency on VSMCs in two cell culture systems representing early and late stages of smooth muscle cells differentiation. We find that folate deficiency promotes differentiation towards a more contractile phenotype as indicated by increased expression of respective marker genes. However, microarray analysis identified markers of striated muscle as the predominant gene expression change elicited by folate deficiency. These changes are not merely a reflection of cell cycle arrest, as foetal calf serum restriction or iron deficiency do not replicate the gene expression changes observed in response to folate deficiency. Folate deficiency only has a marginal effect on global DNA methylation. DNA methylation of CpG islands associated with genes regulated by folate deficiency remains unaffected. This supports our earlier findings in a mouse model system which also did not show any changes in global DNA methylation in response to folate and vitamin B6/B12 deficiency. These data suggest that folate deficiency enhances the expression of smooth muscle marker gene expression, promotes a shift towards a skeletal muscle phenotype, and does not regulate gene expression via DNA methylation.

Introduction

Hyperhomocysteinemia is a risk factor for all facets of atherosclerotic disease including stroke, coronary heart disease and peripheral artery disease [1]. Homocysteine is generated as an intermediate of cellular one-carbon metabolism. Folate (also known as vitamin B9), vitamin B6 and vitamin B12 are all essential components of cellular one-carbon metabolism, either as substrate or as enzyme cofactors (Figure 1). Single nucleotide polymorphisms or mutations of genes involved in one-carbon metabolism are significant predictors of vascular disease. In general, a lower activity of these enzymes (e.g. methylene-tetrahydrofolate-reductase [MTHFR], methionine synthase or cystathionine-B-synthase) is associated with a higher risk of vascular disease [25].

Schematic representation of cellular one carbon metabolism.

Figure 1.
Schematic representation of cellular one carbon metabolism.

Relevant enzymes are shown: methylene tetrahydrofolate reductase (MTHFR), methionine synthase (MS), cystathionine-B-synthase (CBS), di-hydrofolate reductase (DHFR), betaine-homocysteine S-methyltransferase (BHMT). Enzymes dependent on B-vitamins are indicated [B6], [B12]. Dietary folate is a precursor for both, DNA synthesis and DNA methylation. Methotrexate is a classic example of an ‘anti-folate’ used in cancer therapy. It is a highly effective competitive inhibitor of DHFR.

Figure 1.
Schematic representation of cellular one carbon metabolism.

Relevant enzymes are shown: methylene tetrahydrofolate reductase (MTHFR), methionine synthase (MS), cystathionine-B-synthase (CBS), di-hydrofolate reductase (DHFR), betaine-homocysteine S-methyltransferase (BHMT). Enzymes dependent on B-vitamins are indicated [B6], [B12]. Dietary folate is a precursor for both, DNA synthesis and DNA methylation. Methotrexate is a classic example of an ‘anti-folate’ used in cancer therapy. It is a highly effective competitive inhibitor of DHFR.

At present, it is unclear whether homocysteine acts as a direct promoter of cardiovascular disease (CVD) or whether it is mainly an indicator of low B-vitamin status. Numerous studies in animal model systems have assessed the potential mechanisms by which homocysteine might directly influence CVD risk [6]. These experiments typically used diets deficient in folate, vitamin B6 and B12 and rich in methionine [7,8] or drinking water enriched with methionine or homocysteine [9]. Some of these studies have indeed demonstrated that indicators of vascular disease respond directly to serum homocysteine concentrations [1012]. However, in other studies, an increase in serum homocysteine was unrelated to the formation of cardiovascular lesions, whereas a reduction in serum B-vitamin levels was [7,8,13,14]. This suggests that B-vitamin deficiency may promote CVD risk by mechanisms other than causing hyperhomocysteinemia. This is further supported by findings from several large-scale human intervention trials [15,16] using supplementation regimes containing folate, B6 and B12 [16]. Typically supplementation of patients with folate and B-vitamins was successful in reducing serum homocysteine, but ineffective in preventing cardiovascular mortality [15]. This suggests that the major value of homocysteine may be as a marker of B-vitamin deficiency, rather than a mechanistic agent in vascular disease development.

One alternative hypothesis explaining the role of B-vitamin deficiency in vascular disease suggests that a lack of methyl groups available for DNA and histone methylation may alter epigenetic gene regulation in the vascular system [17]. Support for this hypothesis comes from animal experiments which show that alterations in folate supply in pregnancy lead to changes in DNA methylation in ‘metastable' gene alleles and subsequent changes in gene expression and phenotype [1820]. As these epigenetic mechanisms are important for the maintenance of differentiated phenotypes, it has been suggested that a shortage in dietary methyl donors may impact on the maintenance of the differentiation status of vascular smooth muscle cells (VSMCs) and other cell types [21]. Patients with CVD, who suffered from hyperhomocysteinemia and were given a folate supplement (15 mg oral methyltetrahydrofolate a day for 8 weeks) showed an increase in the methylation of imprinted genes suggesting that folate supply may influence the DNA methylation status of genes [22].

VSMCs display a flexible phenotype in that they are, under normal circumstances, quiescent, contractile and do not synthesise extra-cellular matrix or matrix-modifying enzymes [10,23,24]. In the context of vascular injury, VSMC becomes mobile, proliferative and secrete extracellular matrix proteins, possibly adopting a less differentiated phenotype. In VSMC isolated from New Zealand White rabbits a more proliferative phenotype is associated with a reduction in global DNA methylation [17,25]. In human atherosclerotic lesions, global DNA methylation is also decreased relative to normal tissue [25]. However, it is unclear, whether this reduction in DNA methylation is a prerequisite for increased proliferation. Alternatively, the most proliferative (and by inference, least differentiated) cells in the population of isolated smooth muscle cells may be those with a reduced DNA methylation status.

We have used two in vitro systems of smooth muscle cell differentiation to determine the influence of B-vitamin supply on differentiation status. One model, A404, represents early developmental stages of smooth muscle differentiation and one model, A7r5, represents a partially differentiated state of VSMC development. We find that folate deficiency impairs cell proliferation, increases homocysteine secretion and only leads to marginal changes in global DNA methylation. B-vitamin deficiency promotes, rather than prevents, the expression of smooth muscle cell marker genes. However, a microarray analysis of folate-deficient VSCM suggests a phenotypic shift towards a skeletal muscle phenotype. The DNA methylation status of the deficiency responsive genes is not affected. This suggests that the methylation of CpG islands in the vicinity of normal genes (in contrast with metastable alleles) is not responsive to B-vitamin status.

Materials and methods

Cells

A404 cells [26], a gift of Prof Gary Owens, University of Virginia, are derivatives of the mouse embryonic carcinoma cell line P19. The cells can be differentiated into several lineages, including a smooth muscle cell lineage, in the presence of retinoic acid. The cell line also carries a puromycin resistance gene under the control of the smooth muscle-specific myosin heavy chain 11 (MYH11) promoter. Selection of retinoic acid-treated cells in medium containing puromycin permits survival of cells which have differentiated towards the smooth muscle cell lineage but removes cells which have differentiated into other lineages (e.g. neuronal). A404 cells were grown in alpha-MEM medium (Sigma, M8042) supplemented with 7.5% of foetal calf serum (Hyclone), 2 mM glutamine (Life Technologies), and 100 U/ml penicillin/streptomycin (50 U/ml penicillin and 50 µg/ml streptomycin; Life Technologies) and 200 µg/ml hygromycin B (Roche). Cells were split 1 : 50 every 2 or 3 days and used for experiments up to passage 20. Differentiation of cells was induced by the addition of 1 µM all-trans retinoic acid (Sigma) for 3 days. In some experiments retinoic acid-treated cells were additionally treated with 0.5 µg/ml puromycin (Invivogen) for another 3 days. Rat A7r5 cells (ECACC catalogue number 86050803) were grown in DMEM (Sigma, D-6171) supplemented with 10% foetal calf serum (Sigma), 2 mM glutamine and 100 U/ml penicillin/streptomycin. All cells were maintained at 37°C and 5% CO2 in a humidified incubator.

To generate folate-deficient cells, custom made alpha-MEM and DMEM medium was used (BioConcept). The custom-made medium is equivalent to the normal medium but does not contain folic acid, vitamin B6 (pyridoxine-HCl; concentration 4 µg/ml or 23 µM in DMEM, 1 µg/ml or 5.9 µM in alpha-MEM), vitamin B12 (cobalamin; absent in DMEM, 1.36 µg/ml or 3.4 µM in alpha-MEM) and methionine (30 mg/l or 200 µM in DMEM, 15 mg/l or 100 µM in alpha-MEM), which can be added separately. While normal alpha-MEM contains both, vitamin B6 and B12, DMEM does not contain vitamin B12. However, sufficient amounts of vitamin B12 are contained in foetal calf serum. A reduced folate medium (indicated as 10F, containing a folate concentration equivalent to human serum) contained 10 ng/ml of folic acid. A folate-deficient medium (indicated as F) contained no folic acid. A medium deficient in folic acid and vitamin B6 and B12 is indicated as FB. All media were used in conjunction with conventional (i.e. non-dialysed) FCS, which contains 0.6 ng/ml of folate (as determined by radio-immunoassay). The deficient medium, therefore, allows for a limited level of cell proliferation. Cell numbers were determined by counting in a haemocytometer.

For treatment with azacytidine, cells were seeded into 6 cm dishes in conventional medium (DMEM for A7r5 cells, alpha-MEM for A404 cells). Twenty four hours later cells the medium was changed to different concentrations of azacytidine (typically 0.1, 1, 5 or 10 µM) or control medium. Seventy two hours later the medium was changed back to control medium. After a further 72 h cells were counted and genomic DNA was isolated.

Biochemistry

Homocysteine was analysed using isotope dilution GC–MS (gas chromatography and mass spectrometry) largely as described [27]. Briefly, the medium was harvested from treated cells. An amount of 100 µl of the medium was mixed with 50 µl of a 50 nM C13 labelled homocysteine standard solution and 25 µl of a 0.1 M DTT solution and incubated at room temperature for 20 min. After addition of a 2 : 1 acetonitrile/ethanol solution, the reaction was centrifuged at 13 000 rpm for 5 min. The supernatant was absorbed to a Strata-X-C SPE column (Phenomenex) activated with 1 ml of methanol. The column was washed consecutively with 0.5 ml of water and 0.5 ml of methanol and dried. Homocysteine was eluted into a tube with 700 µl of 5% NH3 in 70% methanol and dried at 90°C under nitrogen. The pellet was treated with 10 µl of 0.1 M DTT and 70 µl of 2 M NH3 for 30 min at room temperature and again dried under nitrogen at 90°C. The dried material was then resuspended in 50 µl of a 1 : 1 mix of MTBSTFA [N-Methyl-N-(tert-butyldimethylsilyl)-trifluoroacetamide] and acetonitrile and heated to 90°C for 20 min. The tertiarybutyldimethylsilyl (tBDMS) derivative of homocysteine was detected by GCMS analysis on a 30 m × 0.32 mm × 0.25 µM ZB-5MSi capillary column. Injections (1 µl) were made in the split mode with a 25 : 1 split. The M-57 fragment ions at m/z 420 and 421 for unlabelled and 1-C13 labelled homocysteine-tBDMS are monitored under electron impact selective ion monitoring conditions (EI-SIM). Concentrations of homocysteine were calculated from peak area ratios obtained from EI-SIM fitted to an LREG curve generated from standards. Homocysteine concentrations are expressed in correlation with cell numbers (as measured by µg of DNA) to evaluate the secretion of homocysteine per cell. Typically, homocysteine concentrations in the medium of cells grown in full medium are higher (∼10 µM) than in the medium of cells grown in deficient medium (∼1–5 µM). However, the number of cells in the deficient medium is substantially lower.

Folate concentrations were measured using a competitive radio-immuno-assay (RIA) assay (MP Biomedicals — SimulTRAC-S radioassay kit Vitamin Folate[125I]; catalogue number: 06B254932) following the supplier's instructions. Briefly, cultured cells were detached from plates using Trypsin-EDTA and an aliquot of the cells was counted in a haemocytometer. The remaining cells were pelleted by brief centrifugation and the cell pellet was snap frozen in liquid nitrogen. For analysis, the cells were thawed by the addition of 500 µl of standard A (blank), resuspended by vortexing and sonicated in an ultrasound water bath for 2 min. Cell culture supernatants were measured after diluting 100 µl of medium with 400 µl of standard A. One milliliter of tracer solution containing DTT was added and the samples were incubated in a boiling water bath for 15 min. The samples were then cooled to room temperature in a water bath. An amount of 200 µl of antibody beads were added to each sample and incubated at room temperature for 1 h in the dark. The beads were subsequently pelleted by centrifugation at 4°C for 10 min and the supernatant discarded. The sample tubes were read on a gamma counter (Packard Instruments). The readings were correlated with a standard curve (range between 1 and 20 ng of folate per ml) and the folate concentration was correlated with cell number.

RNA and quantitative PCR

RNA was isolated using the Trizol reagent (Sigma) following the manufacturer's protocol (1 ml of Trizol per 107 cells). Reverse transcription of RNA was done using MLV RNAse(−) reverse transcriptase (Promega) following the manufacturer's recommendations. An amount of 2 µg of total RNA was used as a template for the cDNA synthesis reaction. A one in 10 dilution of the cDNA synthesis reaction was subsequently used as a template for quantitative PCR (Applied Biosystems 7500 Fast System). At least three independent experiments were carried out for each data point and every sample was measured in triplicate.

Quantitative PCR amplifications were carried out with a final primer concentration of 0.5 µM. Oligonucleotide primers were designed using the Primer-BLAST program (NCBI). The sequences, annealing temperatures and amplicon sizes of the oligonucleotides used in this study are provided in Table 1. The quantitative PCR products were evaluated by melting point analysis and agarose gel electrophoresis. Amplifications were done at 40 cycles of 15 s at 95°C, 15 s at the indicated annealing temperature and 30 s at 72°C. Data were collected at the end of each PCR cycle. Standard curves for all genes were generated from serial dilutions of a plasmid containing the cDNA for each gene. The crossing points obtained from the sample analysis was then correlated with the standard curves to provide a concentration of the individual PCR product. Expression of the genes was then correlated with expression of the reference genes (β-actin or GAPDH) in the same sample (expressed as pg of gene per pg of reference) and expressed as fold change from control treatments.

Table 1
Oligonucleotides used for quantitative PCR
Gene ID Primer name Primer sequence Annealing Product Genbank 
rat MYH11 rsmMYH3 CTGGGGAGCTGCGTGTC 57.4°C 530 bp NM_001170600.1 
rsmMYH4 AGCGGCCTTCTCCTCATACTG 
rat CKB rbCK1 AGAAAGGGGGCAACATGAAGGAAG 58.3°C 482 bp NM_012529.3 
rbCK2 GCACTGCCCGGGTAATAAG 
rat CKM rmCK1 GCGTGGGCCTGCAGAAGATTGA 62.9°C 441 bp NM_012530.2 
rmCK2 TCCAGGGGGCGGGGCTCCAG 
rat TAGLN rSM22-1 ATGGCGTGATTCTGAGCAAGTT 57°C 415 bp NM_031549.2 
rSM22-2 GGTCGCCCATAGCCTGTC 
rat ACTA2 rsm-a-actin1 CTGGCCGAGATCTCACCGACTACC 59.6°C 490 bp NM_031004.2 
rsm-a-actin2 GAGCCGCCGATCCAGACAGAATA 
rat GAPDH rGAPDH3 GCTTTCCAGAGGGGCCATCCACA 59°C 426 bp NM_017008.4 
rGAPDH4 ACGGCAAGTTCAACGGCACAGTCA 
rat DNMT1 rDnmt1-3 GTGCCTGCCAGCTGAGTGTTGTG 60°C 483 bp NM_053354.3 
rDnmt1-4 ACCAGGGGATGAGCGTGTTGAAT 
rat DNMT3a rDnmt3a-1 GCACGGGCCGCCTCTTCTT 60°C 502 bp NM_001003957.1 
rDnmt3a-2 CACGATCGGCCCAGCAGTCTCT 
rat DNMT3b rDnmt3b-1 AGCCAGGAGACGCGAGAACAAAAG 60°C 412 bp NM_001003959.1 
rDnmt3b-2 ACCCACCAGCACCTCCAGACACTC 
rat INSIG1 rInsig1-1 TCTCCTCCGCCTGGTGGGTG 60°C 268 bp NM_022392.1 
rInsig1-2 CCCAGGCCACTTCGGGATCG 
rat IDI1 rIDI1-1 GTTGAAGTACGGCGCTCCGCA 60°C 418 bp NM_053539.1 
rIDI1-2 AGCGCTTCTGTGCTGCTCGTTT 
rat LUM rLum1 GCCCGCTGGCCTTCCAACAT 55°C 350 bp NM_031050.1 
rLum2 GGGGATTGCCATCCAAGCGCA 
rat ID1 rIDB1 TGTCGGAGCAAAGCGTTGCCA 61.5°C 331 bp NM_012797.2 
rIDB2 GCTGGAACACATGCCGCCTCGG 
rat PRND rPPD1 ACCGAAGCCCAGGTGGCTGA 60°C 373 bp NM_001102431.1 
rPPD2 CCAGCAGGCAGACCATCGCC 
rat ACTA1 rACTA1-1 GGCAATGAGCGCTTCCGTTGC 59°C 392 bp NM_019212.2 
rACTA1-2 AGACGCGGGTGCGCCTAGAAG 
rat MYH2 rMYH2-1 TGGAGGCCAGAGTGCGTGAAC 55°C 436 bp NM_001135157.1 
rMYH2-2 CACATGGGGACATGACCAAAGGC 
rat TXNIP rTIP-1 AGTGGGTACACCCCCGCCTC 55°C 300 bp NM_001008767.1 
rTIP-2 TGGCTCTGTTGCCCCATGGC 
mm DNMT1 mDnmt1-1 GGTGGATGGCCGGGTCTACTGC 60°C 489 bp NM_001199431.1 
mDnmt1-2 GGGCCCCCTTGTGAATAATCCT 
mm DNMT3a mDnmt3a-1 CACCTATGGGCTGCTGCGAAGAC 60°C 419 bp NM_007872.4 
mDnmt3a-2 GGCGGCCAGTACCCTCATAAAGT 
mm DNMT3b mDnmt3b-1 TGTCGGAGCAAAGCGTTGCCA 60°414 bp NM_001003961.4 
mDnmt3b-2 TGGCAGCGCTGAGGGAGGCACATA 
mm Oct4 mOct4-1 ACCGCCCCAATGCCGTGAAGTT 60°C 593 bp NM_013633.3 
mOct4-2 TGGGGGCAGAGGAAAGGATACAGC 
mm ACTA2 mACTA2 AGTCGCTGTCAGGAACCCTGAGACG 57°C 296 bp NM_007392.3 
mACTA2 ATCTTTTCCATGTCGTCCCAGTTG 
mm MYH11 mMYH11-1 AGGAAACACCAAGGTCAAGCA 57°C 324 bp NM_013607.2 
mMYH11-2 CCCTGACATGGTGTCCAATC 
mm SM22 mSM22a-1 TCCAGTCCACAAACGACCAAGC 57°C 328 bp NM_011526.5 
mSM22a-2 GAATTGAGCCACCTGTTCCATCTG 
Gene ID Primer name Primer sequence Annealing Product Genbank 
rat MYH11 rsmMYH3 CTGGGGAGCTGCGTGTC 57.4°C 530 bp NM_001170600.1 
rsmMYH4 AGCGGCCTTCTCCTCATACTG 
rat CKB rbCK1 AGAAAGGGGGCAACATGAAGGAAG 58.3°C 482 bp NM_012529.3 
rbCK2 GCACTGCCCGGGTAATAAG 
rat CKM rmCK1 GCGTGGGCCTGCAGAAGATTGA 62.9°C 441 bp NM_012530.2 
rmCK2 TCCAGGGGGCGGGGCTCCAG 
rat TAGLN rSM22-1 ATGGCGTGATTCTGAGCAAGTT 57°C 415 bp NM_031549.2 
rSM22-2 GGTCGCCCATAGCCTGTC 
rat ACTA2 rsm-a-actin1 CTGGCCGAGATCTCACCGACTACC 59.6°C 490 bp NM_031004.2 
rsm-a-actin2 GAGCCGCCGATCCAGACAGAATA 
rat GAPDH rGAPDH3 GCTTTCCAGAGGGGCCATCCACA 59°C 426 bp NM_017008.4 
rGAPDH4 ACGGCAAGTTCAACGGCACAGTCA 
rat DNMT1 rDnmt1-3 GTGCCTGCCAGCTGAGTGTTGTG 60°C 483 bp NM_053354.3 
rDnmt1-4 ACCAGGGGATGAGCGTGTTGAAT 
rat DNMT3a rDnmt3a-1 GCACGGGCCGCCTCTTCTT 60°C 502 bp NM_001003957.1 
rDnmt3a-2 CACGATCGGCCCAGCAGTCTCT 
rat DNMT3b rDnmt3b-1 AGCCAGGAGACGCGAGAACAAAAG 60°C 412 bp NM_001003959.1 
rDnmt3b-2 ACCCACCAGCACCTCCAGACACTC 
rat INSIG1 rInsig1-1 TCTCCTCCGCCTGGTGGGTG 60°C 268 bp NM_022392.1 
rInsig1-2 CCCAGGCCACTTCGGGATCG 
rat IDI1 rIDI1-1 GTTGAAGTACGGCGCTCCGCA 60°C 418 bp NM_053539.1 
rIDI1-2 AGCGCTTCTGTGCTGCTCGTTT 
rat LUM rLum1 GCCCGCTGGCCTTCCAACAT 55°C 350 bp NM_031050.1 
rLum2 GGGGATTGCCATCCAAGCGCA 
rat ID1 rIDB1 TGTCGGAGCAAAGCGTTGCCA 61.5°C 331 bp NM_012797.2 
rIDB2 GCTGGAACACATGCCGCCTCGG 
rat PRND rPPD1 ACCGAAGCCCAGGTGGCTGA 60°C 373 bp NM_001102431.1 
rPPD2 CCAGCAGGCAGACCATCGCC 
rat ACTA1 rACTA1-1 GGCAATGAGCGCTTCCGTTGC 59°C 392 bp NM_019212.2 
rACTA1-2 AGACGCGGGTGCGCCTAGAAG 
rat MYH2 rMYH2-1 TGGAGGCCAGAGTGCGTGAAC 55°C 436 bp NM_001135157.1 
rMYH2-2 CACATGGGGACATGACCAAAGGC 
rat TXNIP rTIP-1 AGTGGGTACACCCCCGCCTC 55°C 300 bp NM_001008767.1 
rTIP-2 TGGCTCTGTTGCCCCATGGC 
mm DNMT1 mDnmt1-1 GGTGGATGGCCGGGTCTACTGC 60°C 489 bp NM_001199431.1 
mDnmt1-2 GGGCCCCCTTGTGAATAATCCT 
mm DNMT3a mDnmt3a-1 CACCTATGGGCTGCTGCGAAGAC 60°C 419 bp NM_007872.4 
mDnmt3a-2 GGCGGCCAGTACCCTCATAAAGT 
mm DNMT3b mDnmt3b-1 TGTCGGAGCAAAGCGTTGCCA 60°414 bp NM_001003961.4 
mDnmt3b-2 TGGCAGCGCTGAGGGAGGCACATA 
mm Oct4 mOct4-1 ACCGCCCCAATGCCGTGAAGTT 60°C 593 bp NM_013633.3 
mOct4-2 TGGGGGCAGAGGAAAGGATACAGC 
mm ACTA2 mACTA2 AGTCGCTGTCAGGAACCCTGAGACG 57°C 296 bp NM_007392.3 
mACTA2 ATCTTTTCCATGTCGTCCCAGTTG 
mm MYH11 mMYH11-1 AGGAAACACCAAGGTCAAGCA 57°C 324 bp NM_013607.2 
mMYH11-2 CCCTGACATGGTGTCCAATC 
mm SM22 mSM22a-1 TCCAGTCCACAAACGACCAAGC 57°C 328 bp NM_011526.5 
mSM22a-2 GAATTGAGCCACCTGTTCCATCTG 

Primer sequences, annealing temperatures and expected product sizes are given.

Microarray

RNA for microarrays was isolated from A7r5 cells grown in medium containing 2.5, 10 or 100 ng/ml of folic acid. The isolation was carried out in cells grown for 7, 14 or 21 days in the three different media (three time points at three concentrations). An amount of 2.5 ng/ml reflects a serum concentration typically seen in folate deficiency; 10 ng/ml represents a folate sufficient status; 100 ng/ml represents serum folate concentrations seen in response to folic acid supplement intake. Five independent samples were analysed for each of the nine experimental points. RNA was isolated using the Trizol reagent (Sigma) following the manufacturer's protocol, treated with RNAse-free DNAseI (Promega) for 30 min at 37°C, extracted with phenol/chloroform and precipitated with ethanol. The quality of the RNA was assessed via the ratio of OD at 260 nm and 280 nm (only samples with an OD higher than 1.8 were used) and capillary electrophoresis analysis on Agilent Chips (the only RNA with a RIN higher than 8 was used).

Microarray analysis was done on Affymetrix rat 230-2.0 Genechips by ServiceXS. Data from Service XS, in the CEL file format (containing the raw signal intensities), were transferred to the MadMax website (https://madmax.bioinformatics.nl) where data was analysed using the statistical programming language R [28] and R-libraries offered by the Bioconductor project [29]. The data were normalised using Bioconductor and a GCRMA method, a Robust Multiarray Analysis with correction for the G : C content of the oligos, and then statistically analysed using the Limma package [30], which allowed identification of the most differentially expressed genes between different conditions using a nominal P-value <0.05 to represent statistical significance. Microarray data were deposited at the NCBI Gene Expression Omnibus under accession number GSE125502.

DNA methylation

Genomic DNA isolation was done using the Promega Wizard kit following the supplier's instructions. For all DNA methylation analyses, reference samples were generated which represent methylated and non-methylated DNA. Non-methylated control DNA was generated by whole-genome amplification (WGA) of genomic DNA using a Qiagen Repli-G kit. Methylated DNA was generated by treating isolated genomic DNA with SssI DNA methyltransferase and S-adenosyl methionine (SAM; as substrate) for 1 h at 37°C. Additional enzyme and additional SAM were added after 30 min. The activity of SssI was monitored using bacteriophage lambda DNA as control (Supplementary Figure S6). BstUI restriction digestion of lambda DNA (which is not methylated) yields a total of 25 DNA fragments. Treatment of lambda DNA with SssI and SAM fully methylates the phage DNA and prevents digestion with BstUI (recognition sequence CGCG). Different relative amounts of methylated and non-methylated DNA were mixed and assayed by LC–MS to establish a standard curve which correlates the percentage of SssI methylated DNA in the mixture and the experimentally determined % of methylated cytosine residues (Supplementary Figure S6C).

Global DNA methylation was measured using an LC–MS-based method developed by Friso et al. [31]. Briefly, genomic DNA (1 µg per reaction) was sequentially hydrolysed by digestion with nuclease P1 (2 U per reaction), venom phosphodiesterase I (0.002 U per reaction) and alkaline phosphatase (0.5 U per reaction). The resulting hydrolysate was then separated by reverse-phase HPLC. The 2′deoxycytidine and 5′methyl-2′deoxycytidine peaks were detected and quantified with reference to the internal standard 15N3 2′deoxycytidine.

Combined bisulfite restriction analysis (COBRA) was carried out as described [32]. Briefly, genomic DNA was treated with bisulfite reagent using the EZ DNA methylation kit (Zymo Research) and purified using the spin columns provided in the kit. The concentration of the converted DNA was measured using a Nanodrop device. An amount of 5 ng of bisulfite converted DNA was used as a template for a PCR amplification using gene-specific primer pairs. The primer pairs were designed using the Methyl Primer Express software (Applied Biosystems). Primer pairs for rat LINE1, rat oestrogen receptor alpha exon 1 and rat 14-3-3 sigma (SFN1) were taken from the literature [3335] (Table 2). The amplified DNA was purified over a spin column (QIAGEN Qiaquick) and 100 ng of PCR product was used for a digestion with a restriction enzyme whose recognition sequence carries a CG dinucleotide (including BstUI, Taq, SfaNI). Aliquots of the restriction reaction were separated on a 10% polyacrylamide gel and visualised by ethidium bromide staining. DNA bands were quantified using a Fuji Imaging station and the AIDA Imaging software.

Table 2
Oligonucleotides used for COBRA analyses
Gene ID Primer name Primer sequence Annealing Product Genbank 
rat LINE1 rLINE1-s TTTGGTGAGTTTGGGATA 56°C 163 bp AC229945 (Pogribny et al. [33]) 
rLINE1-as CTCAAAAATACCCACCTAAC 
rat ERα rERa-e1-s TTTTGTTGTATTAGATTTAAGGGAA 59°C 272 bp NM_012689.1 (Monje et al. [34]) 
rERa-e1-as AAAAAAAACCCCCAAACTATTAAC 
rat 14-3-3 σ rSFN1-s TTTTAGGAGTTATTTGGGTGTTGAT 60°C 229 bp XM_003750048 (Chen et al. [35]) 
rSFN1-as CCCCTAAACTAAATAAAACTACTCTCC 
rat PRND rPP2-s TTATTGGTAGTTTTTTGATGGGA 59°C 283 bp NM_001102431.1 
rPP2-as AAACCAAATAAAACCCAACAAA 
CMV CMV-BSP-s TGATTTTATGGGATTTTTTTATTT 52°C 269 bp X17403.1 
CMV-BSP-as CAACTCTACTTATATAAACCTCCCA 
Gene ID Primer name Primer sequence Annealing Product Genbank 
rat LINE1 rLINE1-s TTTGGTGAGTTTGGGATA 56°C 163 bp AC229945 (Pogribny et al. [33]) 
rLINE1-as CTCAAAAATACCCACCTAAC 
rat ERα rERa-e1-s TTTTGTTGTATTAGATTTAAGGGAA 59°C 272 bp NM_012689.1 (Monje et al. [34]) 
rERa-e1-as AAAAAAAACCCCCAAACTATTAAC 
rat 14-3-3 σ rSFN1-s TTTTAGGAGTTATTTGGGTGTTGAT 60°C 229 bp XM_003750048 (Chen et al. [35]) 
rSFN1-as CCCCTAAACTAAATAAAACTACTCTCC 
rat PRND rPP2-s TTATTGGTAGTTTTTTGATGGGA 59°C 283 bp NM_001102431.1 
rPP2-as AAACCAAATAAAACCCAACAAA 
CMV CMV-BSP-s TGATTTTATGGGATTTTTTTATTT 52°C 269 bp X17403.1 
CMV-BSP-as CAACTCTACTTATATAAACCTCCCA 

Pyrosequencing primer combinations (PCR primer pairs and sequencing primer) were designed using the Pyromark software package (Qiagen) for the following genes: skeletal muscle myosin heavy chain (MYH2), skeletal muscle α-actin (ACTA1), prion protein dublet (PRND), thioredoxin interacting protein (TXNIP) and are shown in Table 3. Analysis kits for the Insig1 (insulin-induced gene 1; PM00547638), Id1 (inhibitor of DNA binding 1; PM00543697) and Idi1 (isopentenyl-diphosphate delta isomerase; PM00504119) genes were purchased from Qiagen. Bisulfite treated DNA was used for the PCR reaction at a concentration of 10 ng/µl. Pyrosequencing was carried out using the Pyromark Gold Q24 reagents following the supplier's instructions.

Table 3
Oligonucleotides used for pyrosequencing analyses and gene sequences analysed
Gene ID Primer name Sequence Annealing Product 
rat ACTA1 ACTA1-ps-1 TTTGGTTTTTAGTATTATGAAGATTAAGGT 56°C 205 bp 
ACTA1-ps-2 ACTCCTACTTAATAATCCACATCTACTA 
ACTA1-seq TTTAGTATTATGAAGATTAAGGTG 
sequence analysed GATGAYGYGT TTGGTATGGG YGGAGATTAG GGGYGGGGGG AATTYGTAGG [+2554 to +2700 from transcriptional start] 
rat MYH2 MYH2-ps-1 TAGATTTTGAAGGTGAAAGAAGATTAGA 56°C 178 bp 
MYH2-ps-2 ACATTAAAAACCCCTAATATAAATCAC 
MYH2-seq AGATAGAGGATATGGTTATGATG 
sequence analysed ATTTATTTTT AYGAGTTYGT TGTGTTGTAT AATTTTAAAG AGYGTTAYGT AGTTTGGATG ATTTAYGT [+2379 to +2502 from transcriptional start] 
rat PRND PRND-ps-1 GGGATTTATTAAGAAGGTTGTTTTGAAG 56°C 221 bp 
PRND-ps-2 AAACCCAACAAACAAACCA 
PRND-seq GTGATTAAGGAGGTGTT 
sequence analysed GGTGATTYGT TGYGTTAAYG TTATTTAGGY GGTTAATTAG GTTGAGTTTT TTYGGGAGAA [+311 to +405 from transcriptional start] 
rat TXNIP1 TXNIP-ps-1 GGTTTAATTATGGTGATGTTTAAGAAGATT 56°C 205 bp 
TXNIP-ps-2 ATCTATTTACACTACTAAAACCCTTACAT 
TXNIP-seq ATGGTAGAGGGGAGAA 
sequence analysed GGTGGTYGGT YGGGTGATAG TGGAAGTGTG TGAAGTTATT YGAGTTAAAG TYGTTAGGAT TTTGGTTTGY G [+210 to +357] 
rat INSIG1 sequence analysed YGGAATGTTAYGTTTTTTTYGGAYGA [PM00547638] [+308 to +343 from transcriptional start] 56°C 163 bp 
rat ID1 sequence analysed YGGGTAGGYGYGTTTTAGYGTAGYGYGA [PM00543697] [+174 to +202 from transcriptional start] 56°C 208 bp 
rat IDI1 sequence analysed TTYGTAYGTTTAGTTGTTTTGYGGT [PM00504119] [−65 to −23 from transcriptional start] 56°C 112 bp 
Gene ID Primer name Sequence Annealing Product 
rat ACTA1 ACTA1-ps-1 TTTGGTTTTTAGTATTATGAAGATTAAGGT 56°C 205 bp 
ACTA1-ps-2 ACTCCTACTTAATAATCCACATCTACTA 
ACTA1-seq TTTAGTATTATGAAGATTAAGGTG 
sequence analysed GATGAYGYGT TTGGTATGGG YGGAGATTAG GGGYGGGGGG AATTYGTAGG [+2554 to +2700 from transcriptional start] 
rat MYH2 MYH2-ps-1 TAGATTTTGAAGGTGAAAGAAGATTAGA 56°C 178 bp 
MYH2-ps-2 ACATTAAAAACCCCTAATATAAATCAC 
MYH2-seq AGATAGAGGATATGGTTATGATG 
sequence analysed ATTTATTTTT AYGAGTTYGT TGTGTTGTAT AATTTTAAAG AGYGTTAYGT AGTTTGGATG ATTTAYGT [+2379 to +2502 from transcriptional start] 
rat PRND PRND-ps-1 GGGATTTATTAAGAAGGTTGTTTTGAAG 56°C 221 bp 
PRND-ps-2 AAACCCAACAAACAAACCA 
PRND-seq GTGATTAAGGAGGTGTT 
sequence analysed GGTGATTYGT TGYGTTAAYG TTATTTAGGY GGTTAATTAG GTTGAGTTTT TTYGGGAGAA [+311 to +405 from transcriptional start] 
rat TXNIP1 TXNIP-ps-1 GGTTTAATTATGGTGATGTTTAAGAAGATT 56°C 205 bp 
TXNIP-ps-2 ATCTATTTACACTACTAAAACCCTTACAT 
TXNIP-seq ATGGTAGAGGGGAGAA 
sequence analysed GGTGGTYGGT YGGGTGATAG TGGAAGTGTG TGAAGTTATT YGAGTTAAAG TYGTTAGGAT TTTGGTTTGY G [+210 to +357] 
rat INSIG1 sequence analysed YGGAATGTTAYGTTTTTTTYGGAYGA [PM00547638] [+308 to +343 from transcriptional start] 56°C 163 bp 
rat ID1 sequence analysed YGGGTAGGYGYGTTTTAGYGTAGYGYGA [PM00543697] [+174 to +202 from transcriptional start] 56°C 208 bp 
rat IDI1 sequence analysed TTYGTAYGTTTAGTTGTTTTGYGGT [PM00504119] [−65 to −23 from transcriptional start] 56°C 112 bp 

Protein analysis

Cytoplasmic protein extracts for Western blot analysis were prepared as described [36]. Briefly, cells were lysed on ice in a buffer containing 25 mM KPO4 pH7.8, 8 mM MgCl2, 1 mM EDTA, 1% Triton X-100 and 15% glycerol. The extracts were incubated on ice for 5 min and then centrifuged at 4°C for 1 min at 10 000 g. The supernatants were then aliquoted and stored at −80°C.

Western blot analyses were carried out after the semi-dry transfer of the proteins to a nitrocellulose membrane as described [37]. Smooth muscle α-actin (ACTA2, 42 kDa) and β-actin (ACTB, 42 kDa) were detected using monoclonal antibodies (Sigma A2547 and A5441, respectively) and a horse-radish-peroxidase linked anti-mouse serum (Cell Signalling Technologies #7076). All antibodies were used at a dilution of 1 : 1000. Western blots were developed using a chemiluminescent detection kit (Pierce) and the light emission recorded on a Fuji Imaging Station.

Immunohistochemistry was carried out as previously described [37].

Statistical analysis

Statistical analysis of datasets was done using one or two-way ANOVA in Microsoft Office Excel 2016 and post-hoc analysis and t-tests were carried out Graph-Pad Prism. Error bars indicate standard deviations. The number of replicates is indicated for each figure. The significance levels indicated in the figures typically refer to the comparisons between the control medium and the modified media. Principal components analysis was done using the prcomp function in R applied to scaled data, i.e. using the correlation matrix rather than the covariance matrix. Pathway analysis was carried out using the Bioconductor-TopGo (https://rdrr.io/bioc/topGO/), DAVID (https://david.ncifcrf.gov/) and ENRICHR (http://amp.pharm.mssm.edu/Enrichr/) software packages.

Results

VSMCs surround the endothelium of blood vessels and regulate vascular tone in response to exogenous and endogenous signals [23]. They also play a key role in vascular repair in response to injury. During this process, VSMCs switch from a contractile, quiescent phenotype to a proliferative, synthetic phenotype in which the cells contribute to the repair process [10].

Cellular one-carbon metabolism is fuelled by dietary folate and methionine (Figure 1). Increased serum levels of the central metabolite homocysteine have been associated with vascular disease including stroke, coronary heart disease and peripheral arterial disease [38]. This is particularly obvious in the genetic disease of cystathionine-B-synthase deficiency which leads to extremely high levels of serum Hcy (up to 100 µM compared with a level of 10 µM in healthy individuals) and death as a consequence of vascular disease in early life [39]. In addition, SNPs in enzymes involved in cellular one-carbon metabolism (e.g. MTHFR or methionine synthase) are associated with increased risk of atherosclerosis [40]. These findings link B-vitamin supply with vascular disease. Therefore, we set out to assess whether B-vitamin deficiency affects the switch of VSMCs to a pro-atherogenic phenotype.

Differentiation induces a switch in DNA methyltransferase expression in A404 cells

The differentiation capacity of VSMCs cell culture models is typically characterised by an increase of smooth muscle cell marker gene expression (e.g. MYH11, ACTA2, calponin, SM22/transgelin). To assess the effect of B-vitamin deficiency on the process of smooth muscle cell differentiation we first used the A404 model. This cell line is derived from mouse PC19 embryonal carcinoma cells and can be differentiated towards a smooth muscle cell lineage using retinoic acid. The cells also contain a puromycin marker gene under the control of the MYH11 promoter. Therefore, cells which have differentiated towards the smooth muscle lineage can be enriched by puromycin selection [26].

As a first step, we differentiated A404 cells as described and monitored the ensuing gene expression changes in smooth muscle cell marker genes ACTA2, SM22, MYH11 and the DNA methyltransferase genes Dnmt1, 3a and 3b. Retinoic acid treatment strongly increases the expression of the smooth muscle marker genes ACTA2 (smooth muscle α-actin), SM22 (transgelin 1) and MYH11 (smooth muscle myosin heavy chain) as measured by quantitative PCR (Figure 2A). This is supported by Western blot analysis and immune-histochemistry analysis of ACTA2 protein expression (Figure 2C,D). At the same time expression of the maintenance DNA methyltransferase 1 (Dnmt1) is increased significantly (9-fold) whereas expression of the de novo methyltransferase 3b is significantly decreased (Figure 2B). As expected, the relative induction of the differentiation process is further enhanced by additional treatment of the differentiated cells with puromycin (Supplementary Figure S1). The differentiation process is also accompanied by a drastic (more than 1000-fold) reduction in the expression of the stem cell marker Oct4 (Supplementary Figure S1C). Retinoic acid treatment of A404 cells for 1, 2 and 3 days reveals a gradual increase in smooth muscle marker gene expression over time (Supplementary Figure S1D). These data are consistent with published microarray data analysing the retinoic acid/puromycin driven differentiation process (Supplementary Figure S2) [41].

Analysis of smooth muscle marker and Dnmt gene expression in A404 cells.

Figure 2.
Analysis of smooth muscle marker and Dnmt gene expression in A404 cells.

A404 cells were treated with 1 µM all-trans retinoic acid (dissolved in DMSO; stock solution 1 mM) or equivalent concentration of solvent for 3 days. RNA was isolated, reverse transcribed and analysed by quantitative real-time PCR. Expression of the smooth muscle α-actin (ACTA2), SM22, myosin heavy chain (MYH11), Dnmt1, Dnmt3a and Dnmt3b genes was measured and correlated with expression of the reference gene GAPDH. (A) Fold induction of the smooth muscle cell marker genes ACTA2, SM22 and MYH11 in A404 cells treated with retinoic acid supplemented medium relative to control medium. (B) Fold induction of the DNA methyl-transferase genes Dnmt1, 3a and 3b in A404 cells treated with retinoic acid supplemented medium relative to control medium. (C) Western blot analysis of cytoplasmic extracts from A404 cells treated with retinoic acid supplemented medium ([+]RA) and control medium ([−]RA). Smooth muscle α-actin (ACTA2) and β-actin (ACTB) were detected using specific monoclonal antibodies. Statistical analysis was done by ANOVA (n = 4). * P < 0.05, ** P < 0.01, *** P < 0.001. (D) Immunohistochemistry analysis of ACTA2 protein expression in untreated ([−]RA [−]puro), retinoic acid treated ([+]RA [−]puro) and retinoic acid and puromycin treated ([+]RA [+]puro) A404 cells.

Figure 2.
Analysis of smooth muscle marker and Dnmt gene expression in A404 cells.

A404 cells were treated with 1 µM all-trans retinoic acid (dissolved in DMSO; stock solution 1 mM) or equivalent concentration of solvent for 3 days. RNA was isolated, reverse transcribed and analysed by quantitative real-time PCR. Expression of the smooth muscle α-actin (ACTA2), SM22, myosin heavy chain (MYH11), Dnmt1, Dnmt3a and Dnmt3b genes was measured and correlated with expression of the reference gene GAPDH. (A) Fold induction of the smooth muscle cell marker genes ACTA2, SM22 and MYH11 in A404 cells treated with retinoic acid supplemented medium relative to control medium. (B) Fold induction of the DNA methyl-transferase genes Dnmt1, 3a and 3b in A404 cells treated with retinoic acid supplemented medium relative to control medium. (C) Western blot analysis of cytoplasmic extracts from A404 cells treated with retinoic acid supplemented medium ([+]RA) and control medium ([−]RA). Smooth muscle α-actin (ACTA2) and β-actin (ACTB) were detected using specific monoclonal antibodies. Statistical analysis was done by ANOVA (n = 4). * P < 0.05, ** P < 0.01, *** P < 0.001. (D) Immunohistochemistry analysis of ACTA2 protein expression in untreated ([−]RA [−]puro), retinoic acid treated ([+]RA [−]puro) and retinoic acid and puromycin treated ([+]RA [+]puro) A404 cells.

Azacytidine treatment promotes smooth muscle cell differentiation in A404 cells

Differentiation of A404 cells towards a smooth muscle phenotype is associated with Dnmt1 activation. The effect of Dnmt1 activity on the differentiation process was further assessed by treating the A404 cells with different concentrations of azacytidine (an inhibitor of Dnmt1 activity). Azacytidine treatment was initiated 24 h before the 3-day retinoic acid treatment.

Azacytidine treatment reduces cell viability at concentrations above 0.1 µM. But cells remain viable even at 10 µM, although they divide much more slowly (Figure 3A). Global DNA methylation in A404 cells as measured by liquid chromatography–mass spectrometry (LC–MS) is not significantly affected by the azacytidine treatment (Figure 3B). Expression of the smooth muscle marker genes α-actin and SM22 is affected by azacytidine treatment at higher concentrations. 10 µM azacytidine increases expression levels of both marker genes in both, the untreated and the retinoic acid-treated cells (Figure 3C,D). Lower concentrations of azacytidine also increase expression of α-actin and SM22 in retinoic acid-treated A404 cells, albeit to a lesser degree (Figure 3D). Azacytidine has modest effects on the expression of Dnmt1 and Dnmt3b at higher concentrations (Figure 3C–E). Combined treatment of cells with 10 µM azacytidine and retinoic acid increase the expression of the stem cell marker Oct4 (Figure 3E). Treatment of A404 cells with aza-dCTP (rather than azacytidine) shows a similar increase in α-actin and SM22 expression in A404 cells in the absence of retinoic acid induction (data not shown). Aza-dCTP is more toxic than azacytidine and the effects of smooth muscle cell marker expression are seen at a lower concentration (0.5 µM aza-dCTP, rather than 10 µM azacytidine). These data suggest that although Dnmt1 expression in A404 cells is increased in response to retinoic acid treatment this does not increase global DNA methylation and does not appear to be a prerequisite for differentiation.

Impact of azacytidine treatment on A404 cell proliferation and global DNA methylation.

Figure 3.
Impact of azacytidine treatment on A404 cell proliferation and global DNA methylation.

A404 cells were grown in medium containing 0.1 µM, 1 µM or 10 µM of azacytidine or in normal medium for 6 days. 104 cells were seeded, and cell numbers were recorded on day 3 and 6. Genomic DNA was isolated at the end of the incubation and analysed by LC–MS. (A) Cell proliferation of cells grown in different media. The number of cells was determined by trypan blue staining at the indicated time points using a haemocytometer (n = 6). (B) Global DNA methylation. The percentage of methylated cytosine residues is shown (n = 3). (CE) Analysis of gene expression in azacytidine treated A404 cells. A404 cells were treated with azacytidine for 24 h and then induced with all-trans retinoic acid or equivalent concentration of solvent for 3 days in the continued presence of azacytidine. RNA was isolated, reverse transcribed and analysed by quantitative real time PCR. Expression of the smooth muscle α-actin (ACTA2), SM22, Dnmt1, Dnmt3b and Oct4 genes was measured and correlated with expression of the reference gene GAPDH in A404 cells treated with different concentrations of azacytidine. The expression is shown as fold change from control cells (not treated with azacytidine). (C) Fold induction of the ACTA2, SM22, Dnmt1, Dnmt3b and Oct4 genes in response to azacytidine treatment without retinoic acid. (D) Fold induction of the ACTA2, SM22 and Dnmt1 genes in response to combined retinoic acid and azacytidine treatment. (E) Expression of the Dnmt3b and Oct4 genes in response to combined retinoic acid and azacytidine treatment. Statistical analysis was done by ANOVA (n = 3). * P < 0.05, ** P < 0.01, *** P < 0.001.

Figure 3.
Impact of azacytidine treatment on A404 cell proliferation and global DNA methylation.

A404 cells were grown in medium containing 0.1 µM, 1 µM or 10 µM of azacytidine or in normal medium for 6 days. 104 cells were seeded, and cell numbers were recorded on day 3 and 6. Genomic DNA was isolated at the end of the incubation and analysed by LC–MS. (A) Cell proliferation of cells grown in different media. The number of cells was determined by trypan blue staining at the indicated time points using a haemocytometer (n = 6). (B) Global DNA methylation. The percentage of methylated cytosine residues is shown (n = 3). (CE) Analysis of gene expression in azacytidine treated A404 cells. A404 cells were treated with azacytidine for 24 h and then induced with all-trans retinoic acid or equivalent concentration of solvent for 3 days in the continued presence of azacytidine. RNA was isolated, reverse transcribed and analysed by quantitative real time PCR. Expression of the smooth muscle α-actin (ACTA2), SM22, Dnmt1, Dnmt3b and Oct4 genes was measured and correlated with expression of the reference gene GAPDH in A404 cells treated with different concentrations of azacytidine. The expression is shown as fold change from control cells (not treated with azacytidine). (C) Fold induction of the ACTA2, SM22, Dnmt1, Dnmt3b and Oct4 genes in response to azacytidine treatment without retinoic acid. (D) Fold induction of the ACTA2, SM22 and Dnmt1 genes in response to combined retinoic acid and azacytidine treatment. (E) Expression of the Dnmt3b and Oct4 genes in response to combined retinoic acid and azacytidine treatment. Statistical analysis was done by ANOVA (n = 3). * P < 0.05, ** P < 0.01, *** P < 0.001.

Folate deficiency promotes A404 smooth muscle cell differentiation

To assess the impact of folate, vitamin B6 and vitamin B12 supply on the phenotypic plasticity of smooth muscle cells, A404 cells were grown in media containing various concentrations of folate, either in the presence of vitamin B6 and B12 or in their absence. Cultured cells are typically grown in medium containing a vast excess of folate (4000 ng/ml). To study the effect of folate and B-vitamin deficiency custom made medium with a maximum folate supplementation of 1000 ng/ml was used.

Growth of A404 cells is impaired significantly in folate free medium containing vitamins B6 and B12 (Figure 4A) or devoid of B6 and B12 (Figure 4B). However, the cells still proliferate in this medium, albeit slowly, due to the presence of between 0.5 to 1 ng/ml of folate in foetal calf serum (typical concentration 0.6 ng/ml). The use of dialysed serum completely blocks cell proliferation in folate free medium (data not shown). Cellular folate stores are severely depleted after an incubation period of 6 days in folate free medium (Figure 4C). Therefore, homocysteine secretion per cell is significantly elevated in response to folate restriction (homocysteine is an intermediate of cellular one-carbon metabolism which can be regenerated to methionine in the presence of folate; Figure 1). However, overall homocysteine concentrations in the medium are higher (∼10 µM) in cells grown in the normal medium compared with cells grown in deficient medium (∼1 µM). No significant impact on global DNA methylation can be detected under any of these conditions (Figure 4E).

Impact of folate and vitamin B6/12 deficiency on A404 cells.

Figure 4.
Impact of folate and vitamin B6/12 deficiency on A404 cells.

(A,B) Proliferation of A404 cells in different cell culture media as determined by trypan blue staining at the indicated time points using a haemocytometer. Cells were grown in medium containing 100, 100, 10 or 0 ng/ml of folate either in the presence of B-vitamins (A) or in their absence (B). The folate-deficient medium (0 ng/ml) contains residual folate derived from foetal calf serum at a final folate concentration of 0.6 ng/ml. 2 × 104 A404 cells were seeded into 6 cm dishes and grown for 7 days. Cells were counted after 3, 5 and 7 days. The average of four independent experiments is shown. (C) Intracellular folate concentration in cells grown in folate deficient (F; 0 ng/ml) or folate and B-vitamin deficient (FB) media. Concentrations determined by radio-immunoassay are shown as pg folate per 104 cells. (D) Homocysteine concentrations in the cell culture medium as measured using a GC–MS method. Concentrations in nM per µg of DNA (isolated from the underlying cell layer) were calculated and the results are shown as a percentage of homocysteine in control medium. (E) Global DNA methylation. The percentage of methylated cytosine residues is shown in medium containing 0, 10, 100 or 1000 ng folic acid per ml either in the absence (FB) or presence of vitamins B6 and B12 (F). (F) Expression of the smooth muscle cell marker genes ACTA2, SM22 and MYH11 in response to F and FB media correlated to the expression of the reference gene GAPDH. Results are shown as fold change of expression relative to cells in control medium (CON). (G) Expression of the smooth muscle cell marker genes ACTA2, SM22 and MYH11 in response to retinoic acid treatment in F and FB media relative to expression of the reference gene GAPDH. Results are shown as fold change of expression relative to untreated cells (i.e. in CON medium in the absence of retinoic acid). Measurements (n = 4) were analysed by one-way ANOVA. * P < 0.05, ** P < 0.01, *** P < 0.001.

Figure 4.
Impact of folate and vitamin B6/12 deficiency on A404 cells.

(A,B) Proliferation of A404 cells in different cell culture media as determined by trypan blue staining at the indicated time points using a haemocytometer. Cells were grown in medium containing 100, 100, 10 or 0 ng/ml of folate either in the presence of B-vitamins (A) or in their absence (B). The folate-deficient medium (0 ng/ml) contains residual folate derived from foetal calf serum at a final folate concentration of 0.6 ng/ml. 2 × 104 A404 cells were seeded into 6 cm dishes and grown for 7 days. Cells were counted after 3, 5 and 7 days. The average of four independent experiments is shown. (C) Intracellular folate concentration in cells grown in folate deficient (F; 0 ng/ml) or folate and B-vitamin deficient (FB) media. Concentrations determined by radio-immunoassay are shown as pg folate per 104 cells. (D) Homocysteine concentrations in the cell culture medium as measured using a GC–MS method. Concentrations in nM per µg of DNA (isolated from the underlying cell layer) were calculated and the results are shown as a percentage of homocysteine in control medium. (E) Global DNA methylation. The percentage of methylated cytosine residues is shown in medium containing 0, 10, 100 or 1000 ng folic acid per ml either in the absence (FB) or presence of vitamins B6 and B12 (F). (F) Expression of the smooth muscle cell marker genes ACTA2, SM22 and MYH11 in response to F and FB media correlated to the expression of the reference gene GAPDH. Results are shown as fold change of expression relative to cells in control medium (CON). (G) Expression of the smooth muscle cell marker genes ACTA2, SM22 and MYH11 in response to retinoic acid treatment in F and FB media relative to expression of the reference gene GAPDH. Results are shown as fold change of expression relative to untreated cells (i.e. in CON medium in the absence of retinoic acid). Measurements (n = 4) were analysed by one-way ANOVA. * P < 0.05, ** P < 0.01, *** P < 0.001.

The expression of the smooth muscle cell marker genes ACTA2, SM22 and MYH11 is significantly increased in cells exposed to folate and B-vitamin deficiency, even in the absence of retinoic acid (Figure 4F). The differentiation process induced by retinoic acid treatment is significantly enhanced in folate/B-vitamin deficient cells (Figure 4G). Expression of Dnmt genes and their activation by retinoic acid are not significantly affected by folate/B-vitamin deficiency (data not shown). These data suggest that B-vitamin deficiency enhances, rather than inhibits, the differentiation of A404 cells in response to retinoic acid treatment. This conclusion is valid irrespective of the absolute degree of A404 smooth muscle cell differentiation.

Folate deficiency promotes vascular smooth muscle cell differentiation in A7r5 cells

To analyse whether the responses were seen in A404 cells, which represent an early stage of differentiation also applies to a more differentiated VSMC line, we analysed the bona fide rat VSMC line A7r5. A7r5 cells respond to extended cultivation at the confluence with characteristic morphological changes (including cell elongation, Figure 5A), and induction of several smooth muscle marker genes like smooth muscle α-actin (ACTA2) and smooth muscle myosin heavy chain (MYH11) (Figure 5B–D). A7r5 cells grown under conditions of folate deficiency show a reduction in cell proliferation (Figure 6A,B) even if folate levels are only reduced to the physiological concentration of 10 ng/ml (Figure 6B,C). Folate deficiency also leads to a decrease in the concentration of intracellular folate (Figure 6D), and an increase in the secretion of homocysteine per cell (Figure 6E). The increase in homocysteine secretion is diminished under conditions of additional B6 deficiency in which cell proliferation and, presumably, overall cell metabolism is attenuated (Supplementary Figure S3A). Under conditions of folate deficiency, an increase in global DNA methylation (as measured by LC–MS) is detectable (Figure 6F). In conditions of combined folate and vitamin B6 deficiency, this change in DNA global DNA methylation is not observed (Figure 6F). This change is more pronounced in the presence of vitamin B6 sufficient medium and gradually lost in decreasing medium concentrations of B6 (Supplementary Figure S3B). As in A404 cells, the expression of the smooth muscle cell marker genes MYH11 and SM22 is increased significantly in response to combined folate and B6 deficiency (Figure 6G). No significant changes can be detected in the expression of ACTA2. A trend towards an increase in gene expression is found for Dnmt3a (Figure 6H) which was confirmed in later experiments. Dnmt1 expression is unchanged and Dnmt3b expression was undetectable by qPCR in A7r5 cells (using our experimental set-up). A similar set of responses is also seen if folate deficiency is combined with methionine deficiency (Supplementary Figure S4).

Differentiation of A7r5 rat vascular smooth muscle cells.

Figure 5.
Differentiation of A7r5 rat vascular smooth muscle cells.

(A) Micrographs of A7r5 cells at exponential growth and 3 days after reaching confluence. Cells were photographed at a magnification of 40×. Scale bar: 100 µm. (B) Western blot analysis of protein extracts from growing and confluent cells using antisera directed against β-actin (ACTB) and smooth muscle α-actin (ACTA2). Extracts from HEK 293 cells (which do not express ACTA2) were used as a negative control. (C) Abundance of ACTA2 protein relative to ACTB protein in proliferating and confluent A7r5 cells as quantified by densitometry. (D) Expression of smooth muscle-specific genes SM22, smooth muscle α-actin (ACTA2), smooth muscle myosin heavy chain (MYH11), brain-specific creatine kinase (CKB) and muscle-specific creatine kinase (CKM) in A7r5 cells 3 days and 10 days after reaching confluence. The expression is shown as the percentage of expression in proliferating cells. Statistical analysis (n = 3) was done by ANOVA. * P < 0.05, ** P < 0.01, *** P < 0.001.

Figure 5.
Differentiation of A7r5 rat vascular smooth muscle cells.

(A) Micrographs of A7r5 cells at exponential growth and 3 days after reaching confluence. Cells were photographed at a magnification of 40×. Scale bar: 100 µm. (B) Western blot analysis of protein extracts from growing and confluent cells using antisera directed against β-actin (ACTB) and smooth muscle α-actin (ACTA2). Extracts from HEK 293 cells (which do not express ACTA2) were used as a negative control. (C) Abundance of ACTA2 protein relative to ACTB protein in proliferating and confluent A7r5 cells as quantified by densitometry. (D) Expression of smooth muscle-specific genes SM22, smooth muscle α-actin (ACTA2), smooth muscle myosin heavy chain (MYH11), brain-specific creatine kinase (CKB) and muscle-specific creatine kinase (CKM) in A7r5 cells 3 days and 10 days after reaching confluence. The expression is shown as the percentage of expression in proliferating cells. Statistical analysis (n = 3) was done by ANOVA. * P < 0.05, ** P < 0.01, *** P < 0.001.

Cellular responses to folate (F) and combined folate and B6 deficiency (FB) in A7r5 cells grown in the respective media for 21 days.

Figure 6.
Cellular responses to folate (F) and combined folate and B6 deficiency (FB) in A7r5 cells grown in the respective media for 21 days.

(A) Representative microscopy pictures of A7r5 cells grown in full medium (CON), folate-deficient medium (F; containing a residual concentration of 0.6 ng folate per ml derived from serum) and medium deficient in folate and vitamins B6 (pyridoxal) and B12. Cells were photographed at a magnification of 100×. Scale bar: 40 µm. (B) Impact of different medium folate concentrations (from 0 to 100 ng/ml) on cell growth over a period of 21 days (as determined by trypan blue staining at the end of the incubation period using a haemocytometer). (C) Growth of A7r5 cells in medium containing 10 ng/ml of folate (10F), no folate (F), no folate and no vitamin B6 (FB), or control medium (CON; 100 ng/ml of folate) over an 18-day period as determined by trypan blue staining at the indicated time points using a haemocytometer. (D) Cellular folate concentrations of cells grown in control or F or FB medium as measured by RIA. (E) Homocysteine concentration in medium of cells grown in control or F or FB medium (measured by LC–MS) expressed as nMol of homocysteine per µg of DNA (derived from the underlying cells). (F) Global DNA methylation as measured by LC–MS. Results are shown as a percentage of methylation in CON medium. (G) Expression of the smooth muscle cell marker genes ACTA2, SM22 and MYH11 in response to F and FB medium correlated with expression of the reference gene GAPDH. Results are shown as fold change of expression relative to control cells. (H) Expression of Dnmt1 and Dnmt3a. Measurements (n = 4) were analysed by one-way ANOVA. * P < 0.05, ** P < 0.01, *** P < 0.001.

Figure 6.
Cellular responses to folate (F) and combined folate and B6 deficiency (FB) in A7r5 cells grown in the respective media for 21 days.

(A) Representative microscopy pictures of A7r5 cells grown in full medium (CON), folate-deficient medium (F; containing a residual concentration of 0.6 ng folate per ml derived from serum) and medium deficient in folate and vitamins B6 (pyridoxal) and B12. Cells were photographed at a magnification of 100×. Scale bar: 40 µm. (B) Impact of different medium folate concentrations (from 0 to 100 ng/ml) on cell growth over a period of 21 days (as determined by trypan blue staining at the end of the incubation period using a haemocytometer). (C) Growth of A7r5 cells in medium containing 10 ng/ml of folate (10F), no folate (F), no folate and no vitamin B6 (FB), or control medium (CON; 100 ng/ml of folate) over an 18-day period as determined by trypan blue staining at the indicated time points using a haemocytometer. (D) Cellular folate concentrations of cells grown in control or F or FB medium as measured by RIA. (E) Homocysteine concentration in medium of cells grown in control or F or FB medium (measured by LC–MS) expressed as nMol of homocysteine per µg of DNA (derived from the underlying cells). (F) Global DNA methylation as measured by LC–MS. Results are shown as a percentage of methylation in CON medium. (G) Expression of the smooth muscle cell marker genes ACTA2, SM22 and MYH11 in response to F and FB medium correlated with expression of the reference gene GAPDH. Results are shown as fold change of expression relative to control cells. (H) Expression of Dnmt1 and Dnmt3a. Measurements (n = 4) were analysed by one-way ANOVA. * P < 0.05, ** P < 0.01, *** P < 0.001.

Microarray analysis detects a shift towards skeletal muscle differentiation in response to folate deficiency in A7r5 cells

To characterise the gene expression changes in response to folate deficiency more fully, A7r5 cells were grown in medium containing 100, 10 or 2.5 ng/ml of folate and the transcriptome was analysed by an Affymetrix microarray. An amount of 100 ng/ml of folate is representative of supra-physiological folate levels only detected in human serum in response to the consumption of folate supplements. An amount of 10 ng/ml folate represents an average concentration of human serum folate derived from conventional dietary intake. An amount of 2.5 ng/ml folate represents folate levels found in humans with folate deficiency [42].

RNA was harvested from the cells after 7, 14 and 21 days of cultivation and analysed using Affymetrix microarrays. A principal component analysis of the data demonstrates that the main differences in overall gene expression manifest themselves at day 14 and day 21, whereas fewer significant changes are observed at day 7 (Figure 7). At day 21 there is significant variation between the transcriptome of cells grown at 2.5 and 100 ng/ml, whereas the variation between cells grown at 2.5 and 10 ng/ml is less pronounced. This is confirmed by Venn diagrams of the gene expression changes (for data with P < 0.01 and a change bigger than 2-fold; Figure 8). The top 20 genes regulated by folate availability (increased in 2.5 ng/ml and increased at 100 ng) are shown in Table 4. A similar set of gene expression changes is found between cells grown at 2.5 and 10 ng/ml, but the amplitude of expression changes is much reduced. E.g. MYH2 expression is increased 11.5-fold in cells grown in 2.5 ng/ml compared with cells grown in 100 ng/ml. The fold change is only 6.5-fold if cell grown in 100 and 10 ng/ml are compared (Figure 9).

Principal component analysis of a transcriptome analysis of A7r5 cells grown in medium containing 2.5 ng/ml (shown in green), 10 ng/ml (shown in orange) or 100 ng/ml of folate (shown in red) for 7, 14 and 21 days.

Figure 7.
Principal component analysis of a transcriptome analysis of A7r5 cells grown in medium containing 2.5 ng/ml (shown in green), 10 ng/ml (shown in orange) or 100 ng/ml of folate (shown in red) for 7, 14 and 21 days.

The first principal component separates the data by day of analysis; the second principal component separates the data by folate concentration.

Figure 7.
Principal component analysis of a transcriptome analysis of A7r5 cells grown in medium containing 2.5 ng/ml (shown in green), 10 ng/ml (shown in orange) or 100 ng/ml of folate (shown in red) for 7, 14 and 21 days.

The first principal component separates the data by day of analysis; the second principal component separates the data by folate concentration.

Venn diagrams of the genes changed during the different incubation periods and the different folate concentrations.

Figure 8.
Venn diagrams of the genes changed during the different incubation periods and the different folate concentrations.

Genes included were significant with a P-value <0.001 and a fold change of >2. Note that very few gene changes are found at all time points between cells grown in 2.5 and 10 ng/ml.

Figure 8.
Venn diagrams of the genes changed during the different incubation periods and the different folate concentrations.

Genes included were significant with a P-value <0.001 and a fold change of >2. Note that very few gene changes are found at all time points between cells grown in 2.5 and 10 ng/ml.

Fold change of expression for six exemplary genes, which are highly responsive to modulations of folate concentrations in the medium.

Figure 9.
Fold change of expression for six exemplary genes, which are highly responsive to modulations of folate concentrations in the medium.

Thioredoxin interacting protein (TXNIP), insulin-induced gene 1 (INSIG1) and isopentenyl diphosphate isomerase (IDI1) are highly expressed in 100 ng/ml of folate but down-regulated in response to folate deficiency (A–C). PRND, skeletal muscle myosin heavy chain (MYH2) and ACTA1 are induced in response to low folate concentrations (i.e. 2.5 ng/ml) (D–F). Measurements (n = 5) were analysed by one-way ANOVA. * P < 0.05, ** P < 0.01, *** P < 0.001.

Figure 9.
Fold change of expression for six exemplary genes, which are highly responsive to modulations of folate concentrations in the medium.

Thioredoxin interacting protein (TXNIP), insulin-induced gene 1 (INSIG1) and isopentenyl diphosphate isomerase (IDI1) are highly expressed in 100 ng/ml of folate but down-regulated in response to folate deficiency (A–C). PRND, skeletal muscle myosin heavy chain (MYH2) and ACTA1 are induced in response to low folate concentrations (i.e. 2.5 ng/ml) (D–F). Measurements (n = 5) were analysed by one-way ANOVA. * P < 0.05, ** P < 0.01, *** P < 0.001.

Table 4
Top 20 gene expression changes in response to folate deficiency
Name Gene description Fold change Genbank 
Expression increased in 100 ng/ml vs 2.5 ng/ml folate 
 Txnip Thioredoxin interacting protein 8.97 U30789 
 Itgb8 Integrin subunit beta 8, transcript variant X1 8.61 XM_006240697.3 
 Lum Lumican 8.34 NM_031050 
 Insig1 Insulin induced gene 1 8.10 NM_022392 
 Idi1 Isopentenyl-diphosphate delta isomerase 6.55 NM_053539 
 Ednra Endothelin receptor type A 6.04 BF414702 
 Ptn Pleiotrophin 5.99 NM_017066 
 Itgb8 Integrin subunit beta 8, transcript variant X1 5.97 XM_006240697.3 
 Scd2 Stearoyl-Coenzyme A desaturase 2 5.84 NM_031841 
 Cdc20 Cell division cycle 20 homologue 5.44 U05341 
 Idi1 Isopentenyl-diphosphate delta isomerase 4.61 BI290053 
 LOC684841 Histone H3.2-like 4.48 XM_006222398.2 
 Slc29a1 Solute carrier family 29 (nucleoside transporters), member 1 4.46 NM_031684 
 Cenpw Centromere protein W 4.41 NM_001246319.1 
 Kif20a Kinesin family member 20A 4.38 BE111697 
 Cdca3 Cell division cycle associated 3 4.34 BF417638 
 Nrp1 neuropilin 1 4.21 AF016296 
 Arrdc4 Arrestin domain containing 4 4.18 XM_006229367.3 
 Hmgcs1 3-hydroxy-3-methylglutaryl-Coenzyme A synthase 1 3.93 NM_017268 
 Kif11 Kinesin family member 11 3.93 BE116384 
Expression increased in 2.5 ng/ml vs 100 ng/ml folate 
 Prnd Prion protein dublet 24.95 BF551984 
 Myh2 Myosin, heavy polypeptide 2, skeletal muscle, adult 11.52 BI277586 
 Id1 Inhibitor of DNA binding 1 9.79 M86708 
 Acta1 Actin, alpha 1, skeletal muscle 9.11 NM_019212 
 F3 Coagulation factor III 9.10 NM_013057 
 Tnnt1 Troponin T1, skeletal, slow 8.92 AF399874 
 Plcxd2 Phosphatidylinositol-specific phospholipase C 7.81 NM_001134481.1 
 Myh3 Myosin, heavy polypeptide 3, skeletal muscle, embryonic 7.38 NM_012604 
 Id2 Inhibitor of DNA binding 2 6.96 BC086391.1 
 Scg2 Secretogranin II 6.73 NM_022669 
 Vsnl1 Visinin-like 1 6.51 AI227991 
 Rnf181 Ring finger protein 181 6.31 AI008549 
 Tnc Tenascin C 6.31 AI176034 
 Isg12(b) Putative ISG12(b) protein 6.13 AA819034 
 Prss35 Protease, serine, 35 5.88 AA866443 
 Myl2 Myosin, light polypeptide 2, regulatory, cardiac, slow 5.38 BF419995 
 Trib3 Tribbles homologue 3 5.37 AB020967 
 Pcp4 Purkinje cell protein 4 5.10 NM_013002 
 Ppp1r1a Protein phosphatase 1, regulatory (inhibitor) subunit 1A 5.00 NM_022676 
 Plekhb2 Pleckstrin homology domain containing, family B member 2 4.99 AW254369 
Name Gene description Fold change Genbank 
Expression increased in 100 ng/ml vs 2.5 ng/ml folate 
 Txnip Thioredoxin interacting protein 8.97 U30789 
 Itgb8 Integrin subunit beta 8, transcript variant X1 8.61 XM_006240697.3 
 Lum Lumican 8.34 NM_031050 
 Insig1 Insulin induced gene 1 8.10 NM_022392 
 Idi1 Isopentenyl-diphosphate delta isomerase 6.55 NM_053539 
 Ednra Endothelin receptor type A 6.04 BF414702 
 Ptn Pleiotrophin 5.99 NM_017066 
 Itgb8 Integrin subunit beta 8, transcript variant X1 5.97 XM_006240697.3 
 Scd2 Stearoyl-Coenzyme A desaturase 2 5.84 NM_031841 
 Cdc20 Cell division cycle 20 homologue 5.44 U05341 
 Idi1 Isopentenyl-diphosphate delta isomerase 4.61 BI290053 
 LOC684841 Histone H3.2-like 4.48 XM_006222398.2 
 Slc29a1 Solute carrier family 29 (nucleoside transporters), member 1 4.46 NM_031684 
 Cenpw Centromere protein W 4.41 NM_001246319.1 
 Kif20a Kinesin family member 20A 4.38 BE111697 
 Cdca3 Cell division cycle associated 3 4.34 BF417638 
 Nrp1 neuropilin 1 4.21 AF016296 
 Arrdc4 Arrestin domain containing 4 4.18 XM_006229367.3 
 Hmgcs1 3-hydroxy-3-methylglutaryl-Coenzyme A synthase 1 3.93 NM_017268 
 Kif11 Kinesin family member 11 3.93 BE116384 
Expression increased in 2.5 ng/ml vs 100 ng/ml folate 
 Prnd Prion protein dublet 24.95 BF551984 
 Myh2 Myosin, heavy polypeptide 2, skeletal muscle, adult 11.52 BI277586 
 Id1 Inhibitor of DNA binding 1 9.79 M86708 
 Acta1 Actin, alpha 1, skeletal muscle 9.11 NM_019212 
 F3 Coagulation factor III 9.10 NM_013057 
 Tnnt1 Troponin T1, skeletal, slow 8.92 AF399874 
 Plcxd2 Phosphatidylinositol-specific phospholipase C 7.81 NM_001134481.1 
 Myh3 Myosin, heavy polypeptide 3, skeletal muscle, embryonic 7.38 NM_012604 
 Id2 Inhibitor of DNA binding 2 6.96 BC086391.1 
 Scg2 Secretogranin II 6.73 NM_022669 
 Vsnl1 Visinin-like 1 6.51 AI227991 
 Rnf181 Ring finger protein 181 6.31 AI008549 
 Tnc Tenascin C 6.31 AI176034 
 Isg12(b) Putative ISG12(b) protein 6.13 AA819034 
 Prss35 Protease, serine, 35 5.88 AA866443 
 Myl2 Myosin, light polypeptide 2, regulatory, cardiac, slow 5.38 BF419995 
 Trib3 Tribbles homologue 3 5.37 AB020967 
 Pcp4 Purkinje cell protein 4 5.10 NM_013002 
 Ppp1r1a Protein phosphatase 1, regulatory (inhibitor) subunit 1A 5.00 NM_022676 
 Plekhb2 Pleckstrin homology domain containing, family B member 2 4.99 AW254369 

Fold change of gene expression between A7r5 cells grown in folate sufficient (100 ng/ml) and folate-deficient (2.5 ng/ml) medium.

Pathway analysis was carried out using the TOP-GO, DAVID and Enrichr software suites. TOP-GO analysis ([43] evaluating the Gene Ontology-terms of the regulated genes) identifies pathways associated with cell division and cell proliferation, and cholesterol biosynthesis (GO:0006695) as processes which are significantly regulated by folate availability (Table 5). These data confirm the key role folate plays as a one-carbon donor for DNA synthesis (Figure 1).

Table 5
TOP-GO pathway analysis of transcriptome changes in response to folate deficiency
GO.ID Term Annotated Significant Expected elimKS 
GO:0007076 Mitotic chromosome condensation 17 16 5.02 3.40 × 10−7 
GO:0006695 Cholesterol biosynthetic process 65 38 19.18 1.30 × 10−6 
GO:0051301 Cell division 621 249 183.24 1.40 × 10−6 
GO:0000381 Regulation of alternative mRNA splicing 80 43 23.61 2.20 × 10−6 
GO:0070507 Regulation of microtubule cytoskeleton 288 104 84.98 6.10 × 10−6 
GO:0007059 Chromosome segregation 412 188 121.57 1.10 × 10−5 
GO:0034508 Centromere complex assembly 31 19 9.15 1.70 × 10−5 
GO:0031055 Chromatin remodelling at centromere 14 12 4.13 1.70 × 10−5 
GO:0045214 Sarcomere organization 87 45 25.67 1.90 × 10−5 
GO:0071897 DNA biosynthetic process 215 98 63.44 2.00 × 10−5 
GO.ID Term Annotated Significant Expected elimKS 
GO:0007076 Mitotic chromosome condensation 17 16 5.02 3.40 × 10−7 
GO:0006695 Cholesterol biosynthetic process 65 38 19.18 1.30 × 10−6 
GO:0051301 Cell division 621 249 183.24 1.40 × 10−6 
GO:0000381 Regulation of alternative mRNA splicing 80 43 23.61 2.20 × 10−6 
GO:0070507 Regulation of microtubule cytoskeleton 288 104 84.98 6.10 × 10−6 
GO:0007059 Chromosome segregation 412 188 121.57 1.10 × 10−5 
GO:0034508 Centromere complex assembly 31 19 9.15 1.70 × 10−5 
GO:0031055 Chromatin remodelling at centromere 14 12 4.13 1.70 × 10−5 
GO:0045214 Sarcomere organization 87 45 25.67 1.90 × 10−5 
GO:0071897 DNA biosynthetic process 215 98 63.44 2.00 × 10−5 

elimKS here gives the P-values from topGO specific Kolmogorov–Smirnov Test with the elimination of genes of significant GO-terms in parent terms.

The effect on cholesterol biosynthesis is also consistent with observations that folate supply and lipid metabolism are tightly linked [4446]. Many of the genes which are significantly down-regulated in response to folate deficiency (e.g. insulin-induced gene 1 [Insig1], isopentenyl-diphosphate delta isomerase [Idi1], Steaoryl-CoA desaturase 2 [SCD2]) are involved in aspects of lipid and cholesterol biosynthesis [47]. Cholesterol biosynthesis is also the signalling pathway most affected by folate deficiency in a WikiPathways analysis (P = 1.2 × 10−10) in the Enrichr software suite (http://amp.pharm.mssm.edu/Enrichr/). The key role of folate supply for cell proliferation is supported by the identification of mitotic cell cycle as top term in a Reactome 2016 pathway analysis (P-value: 3.2 × 10−21) and PLK1 (polo-like kinase 1) signalling (a key pathway for mitosis in embryonic cells; P-value 2.4 × 10−12).

A Go-term analysis of genes activated by folate deficiency using the DAVID software (https://david.ncifcrf.gov/) identifies muscle-specific genes (GO:0043292 and related GO-terms) with the highest enrichment score (P-value 2.8 × 10−7). This is consistent with the observation that four of the eight genes which are most strongly induced by folate deficiency are skeletal muscle-specific genes (Table 4). Striated muscle contraction is also the most significant signalling pathway identified by REACTOME analysis in the ENRICHR software suite (Supplementary Figure S5) when a gene list with all genes responsive to folate deficiency (P < 0.01; fold change >2) is analysed [48,49]. Cardiac myopathy is the disease network most closely associated with this gene list (Supplementary Figure S5B). Other genes which are activated by folate deficiency include Prnd and inhibitor of DNA binding 1 (Id1). Both of these genes are associated with pro-angiogenic function [5053]. Consistent with this observation the HIF-1a transcription factor network is the most significant NCI-Nature gene network activated in response to folate deficiency (ENRICHR) (Supplementary Figure S5A). Expression of several smooth muscle marker genes (including MYH11, ACTA2, SM22, MYL6L) and other muscle-specific marker genes (including the transcription factors serum response factor and myocardin) are increased significantly (P < 0.001), albeit by a small margin (up to ∼1.5-fold) in response to 2.5 ng/ml folate treatment (data not shown). This suggests that the higher induction rates seen in deficient medium (0.6 ng/ml, Figure 6G) are elicited by the lower concentration of folate in the medium. Genes encoding enzymes of the cellular one-carbon metabolism are only marginally changed by low folate. The biggest significant changes are found for dihydrofolate reductase (DHFR, down-regulated by 1.9-fold in 2.5 ng/ml vs 100 ng/ml of folate, P < 0.001) and methylene-tetrahydrofolate dehydrogenase (MTHFD2, up-regulated by 2-fold in 2.5 ng/ml vs 100 ng/ml of folate, P < 0.001).

Taken together the microarray results confirm that smooth muscle cell markers are increased in response to folate deficiency, but that genes associated with a skeletal muscle phenotype are increased more strongly, together with genes associated with angiogenesis.

Folate deficiency-induced gene expression changes are not caused by cell cycle arrest

The expression of the eight most responsive genes (four most highly activated in response to high folate and four most highly activated in response to low folate) was confirmed using quantitative PCR (data not shown). To assess whether the changes in gene expression were merely due to cell cycle arrest in response to folate deficiency, A7r5 cells were exposed to two other stimuli known to reduce cell proliferation. The cells were grown in medium containing 0.5% FCS or supplemented with 0.01 µM of the iron chelator deferoxamine (DFO). Both treatments significantly reduce cell proliferation (Figure 10A,B) similar to folate restriction (Figure 6C). Expression of seven genes regulated by folate deficiency (four down-regulated by low folate: TXNIP, IDI1, ID1 and INSIG1, three up-regulated by folate deficiency: MYH2, ACTA1 and PRND) was measured in response to FCS reduction and iron deficiency using quantitative PCR. Expression of IDI1 and ID1 was significantly reduced in response to FCS reduction (as under conditions of folate deficiency) and INSIG1 expression was significantly reduced in response to Fe deficiency (albeit much less than in response to F deficiency). In the case of TXNIP1, MYH2, ACTA1 and PRND only folate deficiency elicited the characteristic gene expression changes, whereas iron deficiency and low FCS did not. This suggests that most of the gene expression changes observed in the microarray are specific to folate deficiency rather than a reflection of cell cycle arrest. During this analysis, it was also found that expression of Dnmt3a was increased in response to folate deficiency (but not iron deficiency or low serum, Figure 10D) confirming the trend seen in Figure 6H.

Attenuation of A7r5 cell proliferation by low serum and iron deficiency.

Figure 10.
Attenuation of A7r5 cell proliferation by low serum and iron deficiency.

(A) Growth of A7r5 cells in medium containing 10%, 0.5% and 0.1% foetal calf serum over a period of 11 days. (B) Growth of A7r5 cells in control medium and medium containing 0.01 µM of the iron chelator DFO. (C) Expression of the genes ID1, INSIG1, TXNIP1 and IDI1 which were found to be down-regulated in response to folate deficiency. Gene expression was measured in cells grown in medium containing 0.5% serum (FCS), 0.01 µM DFO (FE) or 0.6 ng/ml folate (F). Gene expression was correlated with the expression of the reference gene β-actin. Results are shown as fold change of expression relative to control cells. (D) Expression of the genes MYH2, ACTA1 and PRND which were found to be up-regulated in response to folate deficiency. In addition, expression of the Dnmt3a gene was measured. Measurements (n = 3) were analysed by one-way ANOVA. * P < 0.05, ** P < 0.01 *** P < 0.001.

Figure 10.
Attenuation of A7r5 cell proliferation by low serum and iron deficiency.

(A) Growth of A7r5 cells in medium containing 10%, 0.5% and 0.1% foetal calf serum over a period of 11 days. (B) Growth of A7r5 cells in control medium and medium containing 0.01 µM of the iron chelator DFO. (C) Expression of the genes ID1, INSIG1, TXNIP1 and IDI1 which were found to be down-regulated in response to folate deficiency. Gene expression was measured in cells grown in medium containing 0.5% serum (FCS), 0.01 µM DFO (FE) or 0.6 ng/ml folate (F). Gene expression was correlated with the expression of the reference gene β-actin. Results are shown as fold change of expression relative to control cells. (D) Expression of the genes MYH2, ACTA1 and PRND which were found to be up-regulated in response to folate deficiency. In addition, expression of the Dnmt3a gene was measured. Measurements (n = 3) were analysed by one-way ANOVA. * P < 0.05, ** P < 0.01 *** P < 0.001.

Folate deficiency-induced gene expression changes are not associated with changes in gene-specific DNA methylation

To assess whether the observed gene expression changes are correlated with gene-specific changes in DNA methylation, A7r5 DNA derived from folate deficient and iron-deficient cells and cells grown in low serum (0.5%) were analysed using COBRA and pyrosequencing. Reference points for non-methylated DNA and methylated DNA were established by whole genome amplification and DNA methyltransferase SssI treatment, respectively, and used to establish standard curves for methylation frequency in all experiments (Supplementary Figure S6A).

Whole genome amplified A7r5 DNA only shows 0.2% of all cytosine residues being methylated when analysed by LC–MS [31] (Supplementary Figure S6D). In contrast in untreated A7r5 DNA 3.1% of all cytosine residues are methylated. Treatment of naïve A7r5 DNA with SssI and SAM increases the frequency of methylated cytosine residues to 5% (Supplementary Figure S6D). In contrast whole-genome amplified DNA treated with SssI only contains 1.2% of cytosine residues in a methylated state. This suggests that SssI treatment has limited efficiency on non-methylated genomic DNA (Supplementary Figure S6D). Genomic DNA derived from azacytidine treated A7r5 cells show progressive global demethylation as assessed by LC–MS (Supplementary Figure S7).

Gene-specific methylation of LINE1 and oestrogen receptor alpha was measured in genomic DNA derived from folate-deficient (and control) A7r5 cells using COBRA (Figure 11 and Supplementary Figures S8, S9). Methylation of LINE1 was marginally increased in response to folate deficiency (Figure 11E and Supplementary Figure S8C,D). It remained unchanged in response to cell cycle arrest inducing treatments of A7r5 cells in 0.5% FCS and DFO treatment (Figure 11F). Similarly, methylation of ERα was marginally increased in response to folate deficiency (Supplementary Figure S9D). No significant change in DNA methylation was found for the tumour suppressor gene 14-3-3 sigma (SFN, data not shown) [35]. These marginal DNA methylation changes contrast with a control experiment in which we could find a strong methylation of the CMV promoter in A404 cells compared with HEK293 cells, which correlates with the expression rates of a CMV promoter-driven nuclear GFP protein (Supplementary Figure S10). Methylation of a BstUI site in the Prnd gene, which is strongly activated in response to folate deficiency, is significantly decreased in response to folate deficiency, but not in response to reduced serum and iron deficiency (DFO treatment) (Figure 12).

COBRA analysis of DNA methylation at LINE1 (a repetitive mobile genetic element which serves as a proxy for global DNA methylation).

Figure 11.
COBRA analysis of DNA methylation at LINE1 (a repetitive mobile genetic element which serves as a proxy for global DNA methylation).

(A) Schematic representation of the LINE1 section used for DNA amplification with the primer pair LINE-1s/as (from Pogribny et al. [33]). The location of the primer binding sites, the expected PCR product and its BstUI restriction digestion products are indicated, as are the CpG nucleotides in the amplified DNA region. A BstUI site (CGCG) is maintained in methylated but not in non-methylated DNA, where it is converted to TGTG. The rate of digestion is, therefore, reflective of the methylation rate of both CG dinucleotides (i.e. increases in digestion signify increases in methylation). (B) Polyacrylamide gel analysis of the BstUI digested of LINE-1 PCR product. Different ratios of non-methylated (W) and untreated A7r5 DNA (U) were mixed prior to bisulfite conversion and PCR amplification (from 100 : 0 to 0 : 100). The PCR products were purified and digested with BstUI and separated on a 10% polyacrylamide gel alongside a molecular mass marker (NEB PCR marker). The sizes of the relevant DNA fragments are indicated. (C) Polyacrylamide gel with COBRA samples derived from A7r5 cells grown in control medium (100 ng/ml folate: CON), medium containing 10 ng/ml folate (10F), medium without added folate (0.6 ng/ml: F) and medium containing no added folate or vitamin B6 (FB). (D) Quantification of DNA methylation after image analysis of the gel shown in (B). The percentage of methylated input DNA is correlated with DNA methylation measured by BstUI digestion. (E) Quantification of DNA methylation after image analysis of the gel shown in (C). (F) Quantification of DNA methylation in A7r5 cells treated with folate-deficient medium (F), iron-deficient medium (FE) or medium with 0.5% FCS (FCS) as measured by COBRA analysis of the LINE1 element. Measurements (n = 3) were analysed by one-way ANOVA. * P < 0.05.

Figure 11.
COBRA analysis of DNA methylation at LINE1 (a repetitive mobile genetic element which serves as a proxy for global DNA methylation).

(A) Schematic representation of the LINE1 section used for DNA amplification with the primer pair LINE-1s/as (from Pogribny et al. [33]). The location of the primer binding sites, the expected PCR product and its BstUI restriction digestion products are indicated, as are the CpG nucleotides in the amplified DNA region. A BstUI site (CGCG) is maintained in methylated but not in non-methylated DNA, where it is converted to TGTG. The rate of digestion is, therefore, reflective of the methylation rate of both CG dinucleotides (i.e. increases in digestion signify increases in methylation). (B) Polyacrylamide gel analysis of the BstUI digested of LINE-1 PCR product. Different ratios of non-methylated (W) and untreated A7r5 DNA (U) were mixed prior to bisulfite conversion and PCR amplification (from 100 : 0 to 0 : 100). The PCR products were purified and digested with BstUI and separated on a 10% polyacrylamide gel alongside a molecular mass marker (NEB PCR marker). The sizes of the relevant DNA fragments are indicated. (C) Polyacrylamide gel with COBRA samples derived from A7r5 cells grown in control medium (100 ng/ml folate: CON), medium containing 10 ng/ml folate (10F), medium without added folate (0.6 ng/ml: F) and medium containing no added folate or vitamin B6 (FB). (D) Quantification of DNA methylation after image analysis of the gel shown in (B). The percentage of methylated input DNA is correlated with DNA methylation measured by BstUI digestion. (E) Quantification of DNA methylation after image analysis of the gel shown in (C). (F) Quantification of DNA methylation in A7r5 cells treated with folate-deficient medium (F), iron-deficient medium (FE) or medium with 0.5% FCS (FCS) as measured by COBRA analysis of the LINE1 element. Measurements (n = 3) were analysed by one-way ANOVA. * P < 0.05.

COBRA analysis of DNA methylation at the rat PRND gene.

Figure 12.
COBRA analysis of DNA methylation at the rat PRND gene.

(A) Schematic representation of the PRND gene section used for DNA amplification with the primer pair PP2-s and PP2-as. The location of the primer binding sites, the expected PCR product and its BstUI restriction digestion products are indicated, as are the CpG nucleotides in the amplified DNA region. (B) Polyacrylamide gel analysis of BstUI digested PRND PCR product amplified from A7r5 cells grown in control medium (100 ng/ml folate: CON), folate-deficient medium (F), iron-deficient medium (FE) or medium with 0.5% FCS (FCS). The PCR products were purified and digested with BstUI and separated on a 10% polyacrylamide gel. The sizes of the relevant DNA fragments are indicated. (C) Quantification of DNA methylation after image analysis of the gel shown in (B). The percentage of methylated input DNA is correlated with DNA methylation measured by BstUI digestion. Measurements (n = 3) were analysed by one-way ANOVA. ** P < 0.01.

Figure 12.
COBRA analysis of DNA methylation at the rat PRND gene.

(A) Schematic representation of the PRND gene section used for DNA amplification with the primer pair PP2-s and PP2-as. The location of the primer binding sites, the expected PCR product and its BstUI restriction digestion products are indicated, as are the CpG nucleotides in the amplified DNA region. (B) Polyacrylamide gel analysis of BstUI digested PRND PCR product amplified from A7r5 cells grown in control medium (100 ng/ml folate: CON), folate-deficient medium (F), iron-deficient medium (FE) or medium with 0.5% FCS (FCS). The PCR products were purified and digested with BstUI and separated on a 10% polyacrylamide gel. The sizes of the relevant DNA fragments are indicated. (C) Quantification of DNA methylation after image analysis of the gel shown in (B). The percentage of methylated input DNA is correlated with DNA methylation measured by BstUI digestion. Measurements (n = 3) were analysed by one-way ANOVA. ** P < 0.01.

To get a more comprehensive overview over gene-specific methylation in A7r5 cells in response to folate deficiency, iron deficiency and restricted FCS, seven of the genes strongly regulated by folate deficiency were assessed for gene-specific DNA methylation using pyrosequencing. CpG islands were identified using the Methyl Primer Express software (Applied Biosystems) and confirmed using the CpG finder software. All seven genes carry a CpG island in the vicinity of the gene (Supplementary Figure S11). Pyrosequencing assays for these areas were developed using the pyromark software (for Myh2, Acta1, Prnd, Txnip) or purchased as kits from QIAGEN (for Id1, Idi1 and Insig1). The overall methylation rate of the gene sequences is highly variable ranging from 5% for ID1 to 90% for PRND. This may reflect the position of the CpG islands relative to the transcriptional start site (Table 3) which is a known determinant of genome methylation [54]. The results do not show a significant change in gene-specific DNA methylation for any of the genes in response to the different treatments (Figure 13). Control reactions using mixtures of whole-genome amplified DNA and SssI treated DNA demonstrate that all sites tested are amenable to the presence and absence of DNA methylation (as shown for the Prnd gene in Figure 13G). In contrast, a significant (albeit small) decrease in DNA methylation was detected in the Prnd gene in response to azacytidine treatment (Supplementary Figure S12). Taken together with the change in DNA methylation found in the COBRA analysis, this suggests that the Prnd gene may be most susceptible to gene regulation changes by DNA methylation.

Pyrosequencing analysis of folate-responsive genes.

Figure 13.
Pyrosequencing analysis of folate-responsive genes.

Pyrosequencing analysis of CpG islands of the ACTA1 (A), MYH2 (B), INSIG1 (C), TXNIP (D), ID1 (E), IDI1 (F) and PRND genes (G,H). Genomic DNA was derived from A7r5 cells grown in low serum (0.5% FCS; indicated as FCS), iron deficiency (10 µM DFO; indicated as FE) or folate deficiency (0.6 nM folate; indicated as F). The percentage of methylated Cs at each of the polymorphism positions is indicated. (G) Quality control of the pyrosequencing analysis for the Prnd gene. Different ratios of non-methylated (WGA) and untreated A7r5 DNA were mixed prior to bisulfite conversion and PCR amplification (from 100 : 0 to 0 : 100) before being analysed by pyrosequencing at the Prnd gene. Data are the average of three independent replicates and were analysed by ANOVA. * P < 0.05.

Figure 13.
Pyrosequencing analysis of folate-responsive genes.

Pyrosequencing analysis of CpG islands of the ACTA1 (A), MYH2 (B), INSIG1 (C), TXNIP (D), ID1 (E), IDI1 (F) and PRND genes (G,H). Genomic DNA was derived from A7r5 cells grown in low serum (0.5% FCS; indicated as FCS), iron deficiency (10 µM DFO; indicated as FE) or folate deficiency (0.6 nM folate; indicated as F). The percentage of methylated Cs at each of the polymorphism positions is indicated. (G) Quality control of the pyrosequencing analysis for the Prnd gene. Different ratios of non-methylated (WGA) and untreated A7r5 DNA were mixed prior to bisulfite conversion and PCR amplification (from 100 : 0 to 0 : 100) before being analysed by pyrosequencing at the Prnd gene. Data are the average of three independent replicates and were analysed by ANOVA. * P < 0.05.

Discussion

We have analysed the effects of folate deficiency on the phenotypic plasticity of VSMCs. Our results demonstrate, firstly, that folate deficiency induces significant changes in global gene expression patterns indicating that the cells adopt a more contractile phenotype. Secondly, the overall expression pattern suggests a trans-differentiation towards a skeletal muscle phenotype. Thirdly, folate deficiency mediated alterations of cellular one-carbon metabolism do not induce DNA methylation changes in CpG islands associated with folate responsive genes.

Folate deficiency is associated with several disease phenotypes. The most important of these is neural tube defects [55,56] which has led to the introduction of fortification of flour with folic acid in some countries [57]. The caveat to this approach is the promotion of cell proliferation which can have a small but measurable cancer-promoting effect [58]. This is not surprising as many cancer drugs (e.g. methotrexate; see Figure 1) act as anti-folates interfering with folate-supported DNA synthesis. Another well-characterised disease outcome of folate deficiency is colon cancer which has been documented in animal studies and epidemiological observations in human populations [59]. Finally, folate deficiency (and one of its physiological outcomes, increased serum homocysteine levels) is associated with an increased risk of CVD [60,61]. The mechanisms by which B-vitamin deficiency impacts on CVD risk, however, are unclear.

Numerous studies have investigated the potential direct effects of homocysteine on vascular health. Several mechanistic explanations have been put forward including a promotion of inflammation, increased proliferation of VSMC in the neointima, a toxic effect on the vascular endothelium, and cell signalling alterations [62]. Some studies in model systems seemed to support a mechanistic role for homocysteine in CVD progression by promoting VSCM proliferation [11]. However other studies (including our own) failed to establish a conclusive link between serum homocysteine concentrations and vascular disease progression [7,8,13]. Specifically, our data demonstrate that plaque formation is not correlated with serum homocysteine levels. Under conditions of combined folate and B-vitamin deficiency serum homocysteine levels of ∼50 µM were detected, whereas folate deficiency alone only increased serum homocysteine to 15 µM. In contrast, significant increases in plaque formation were detected in folate deficient, but not in combined folate/B-vitamin deficient mice [8]. This is supported by another rodent model system, which specifically demonstrated that provision of high methionine seems to be the predominant driver of vascular disease and not high homocysteine levels [13]. These data argue against a direct link of serum homocysteine levels with the extent of vascular plaque formation.

Moreover, the finding that B-vitamin supplementation in cardiovascular patients reduces serum homocysteine, but appears to be of no clinical benefit has led to the suggestion that homocysteine may be a marker for B-vitamin deficiency, rather than an active agent in vascular disease progression [15,16]. These findings may be explained in two ways. Firstly, homocysteine may have no further effect on disease progression, once overt CVD is present. Alternatively, homocysteine lowering may only be clinically effective in the small group of patients which suffer from severe hyperhomocysteinemia (i.e. serum homocysteine levels of higher than 100 µM; incidentally this is the concentration which is typically used in rodent and in vitro models of hyperhomocysteinemia) [62]. Therefore, other mechanistic links between B-vitamin deficiency and CVD risk are being considered.

Hiltunen et al. observed a reduction in global DNA methylation of VSMCs in association with CVD in a mouse/rat model [25,63]. This led to the hypothesis that B-vitamin deficiency may exert its effect on CVD via an epigenetic mechanism like DNA methylation or histone methylation. A potential link between folate supply, DNA methylation, and gene expression has been demonstrated e.g. for the agouti mouse model and related ‘metastable' alleles [64]. We have, therefore, studied the effects of folate, vitamins B6 and B12, and methionine deficiency in two complementary cell culture models representing early and late differentiation stages of VSMCs.

Our results allow two major conclusions. Firstly, folate deficiency does not prevent the establishment or maintenance of a differentiated vascular smooth muscle phenotype. In fact, the expression of smooth muscle cell marker genes is increased in response to folate deficiency in both, A404 and A7r5 cells. However, the microarray analysis of A7r5 cells suggests a shift of overall gene expression towards a skeletal muscle phenotype. Trans-differentiation on A7r5 cells into skeletal muscle has been observed before as a spontaneous outcome of cultivation but was only characterised by the expression of two skeletal muscle proteins [65]. Our results suggest that an exogenous stressor like folate deficiency can promote this process. In addition, several genes involved in tissue repair are also activated in response to folate deficiency, while genes involved in cholesterol and lipid metabolism are down-regulated. This is also reflected for the overall transcriptome by pathway analyses. The close link between B-vitamin/one-carbon metabolism and lipid metabolism is well characterised and supports the findings of other researchers [4446]. Trans-differentiation of VSMC in vivo has been documented [66,67] but is controversial [68,69]. The most relevant trans-differentiation processes observed in vivo appear to be towards an osteogenic and macrophage (rather than a skeletal muscle) lineage [66,67].

We have previously studied the effects of folate and vitamin B6 and B12 deficiency in the arterial tissue of folate-deficient apoE−/− mice using proteomics [8,70]. Some of the changes seen in that analysis are also reflected in the A7r5 cells. for example, the transferrin receptor, the electron-transfer flavoprotein, lactate dehydrogenase, glutathione-S-transferase mu1, several fibrinogen polypeptides are all significantly up-regulated in conditions of folate deficiency in both analysis systems. The fact that the major changes seen in A7r5 cells are not seen in the whole-aorta is presumably due to the different sensitivities of the two analyses (microarray vs proteomics, the latter only revealing changes in highly expressed proteins) and the modest contribution of smooth muscle cells to the overall tissue composition of a complete aorta.

The second conclusion from the data presented is that changes in gene expression are not correlated with significant changes in DNA methylation of CpG islands associated with genes regulated by folate supply. This is in accordance with recent findings by McKay et al. [71] who found that gene expression and DNA methylation changes in the liver of folate-deficient mice were not correlated for most genes. In addition, Wang et al. [72] investigated the role of DNA methylation in a model of neural tube defects, and found that the expression of very few genes was changed as a consequence of DNA methylation changes. In contrast, Leclerc et al. [73] found small but significant increases in the methylation of CpG islands associated with tumour suppressor genes in colon tumours induced by folate deficiency in C57B/6 mice. However, even in that case, the DNA methylation changes detected (by pyrosequencing) were in the range of a few percent; similar to the amplitude of changes we detect in some of the assays, especially for the Prnd gene. The available data suggest that while the methylation status of some genes is sensitive to environmental changes (like folate supply) for most genes no significant methylation changes are found. To conclusively assess the specific effect of environmental changes on DNA methylation a whole-genome sequencing of bisulfite converted DNA would be necessary.

Recent technological advances have enabled the site-directed modification of DNA methylation using fusion proteins of Cas9 with Dnmt3a or Tet1 [7476] in which the methylating or demethylating activities are targeted to sites specified by guide RNAs interacting with Cas9. While these experiments clearly demonstrate a link between the methylation of target genes with their expression, it is possible that these effects only apply to a subset of genes. Our data also show profound impacts of DNA methylation on the expression of a CMV-GFP transgene stably inserted in HEK293 cells and A404 cells. Taken together these data suggest that a simple correlation between gene-specific DNA methylation changes resulting in altered gene expression, as is the case for metastable alleles, is not generally applicable. Alternatively, DNA methylation changes in one genomic location may have long-range effects on chromosome architecture resulting in changes of expression of distant genes with consequent changes in biological phenotypes.

The results in this study also raise the question as to how important folate supply is for DNA methylation. It is probable that DNA methylation of genes is mainly malleable during early stages of development (as reflected in the A404 cells). Effects of early life nutrition on CVD risk have indeed been demonstrated [77]. We only detected small changes in global or gene-specific DNA methylation in both VSMC lines in response to folate deficiency, reflecting our findings in mice [8]. Human studies have demonstrated that folate and B-vitamin supply in utero can have differing effects on the methylation of different imprinted genes [78,79]. In addition, direct analysis of gene expression correlated with these methylation changes is restricted as very few tissues which are accessible for such studies in humans (mainly lymphocytes).

DNA methylation enzymes compete for methyl groups with other cellular processes like DNA synthesis [80]. In tumour cells an increased requirement for de novo nucleotide synthesis can, therefore, decrease overall levels of DNA methylation. However, even a dramatic increase in the DNA de novo synthesis rate of 36% (e.g. in tumour cells) only reduces the DNA methylation rate by 0.5% [81] suggesting that DNA methylation is not particularly responsive to one carbon unit supply. In addition, histone methylation, another mechanism of epigenetic gene control, may also provide a target for gene regulatory effects of folate deficiency. It has been suggested that THF (derived from dietary folate) is required to protect lysine demethylase enzymes from the formaldehyde generated as a consequence of histone demethylation [82]. Folate deficiency may, therefore, impair histone demethylation thereby impacting on gene regulation. B-vitamin deficiency may also affect CVD risk indirectly via other processes, including DNA repair, micro RNA stress responses, inflammation and energy sensors, like mTOR [8386] either with or without epigenetic alterations.

Conclusions

Taken together the data shown in here demonstrate that folate deficiency promotes a contractile phenotype in VSMCs, combined with a trans-differentiation towards a skeletal muscle gene expression profile. These expression changes are not accompanied by altered DNA methylation of the regulated genes. The data also clearly show that a simple correlation of gene-specific DNA methylation changes and altered gene expression, as is the case for metastable alleles, is not generally applicable.

Abbreviations

     
  • ACTA1

    skeletal muscle α-actin

  •  
  • BHMT

    betaine-homocysteine S-methyltransferase

  •  
  • CKB

    brain-specific creatine kinase

  •  
  • CKM

    muscle-specific creatine kinase

  •  
  • COBRA

    Combined bisulfite restriction analysis

  •  
  • CVD

    cardiovascular disease

  •  
  • DFO

    deferoxamine

  •  
  • DHFR

    dihydrofolate reductase

  •  
  • EI-SIM

    electron impact selective ion monitoring conditions

  •  
  • ERα

    oestrogen receptor α

  •  
  • Hcy

    homocysteine

  •  
  • MTHFD2

    methylene-tetrahydrofolate dehydrogenase

  •  
  • MTHFR

    methylene-tetra hydro folate reductase

  •  
  • PRND

    prion protein dublet

  •  
  • RIA

    radio-immunoassay

  •  
  • SAM

    S-adenosyl methionine

  •  
  • SCD2

    Steaoryl-CoA desaturase 2

  •  
  • tBDMS

    tertiarybutyldimethylsilyl

  •  
  • TXNIP

    thioredoxin interacting protein

  •  
  • VSMCs

    Vascular smooth muscle cells

  •  
  • WGA

    whole-genome amplification

Author Contribution

A.F.K. designed and supervised the study, carried out experimental work, and wrote the manuscript, Li.Pe. and Ly.Pi. carried out experimental work, C.D.M. analysed the microarray data, S.J.D. co-designed and co-supervised the study and co-wrote the manuscript.

Funding

Microarray data shown in this manuscript have been deposited at NCBI via the Gene Expression Omnibus webpage under number: GSE125502. This work was funded by the Scottish Government Rural and Environment Science and Analytical Services Division (RESAS). The funder had no role in the design of the study, the analysis and interpretation of the data, or the publication process.

Acknowledgements

The authors would like to thank Drs Bill Rees and Perry Barrett for helpful discussions throughout the project and on the manuscript.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Ntaios
,
G.
,
Savopoulos
,
C.
,
Grekas
,
D.
and
Hatzitolios
,
A.
(
2009
)
The controversial role of B-vitamins in cardiovascular risk: an update
.
Arch. Cardiovasc. Dis.
102
,
847
854
2
Zaina
,
S.
,
Lindholm
,
M.W.
and
Lund
,
G.
(
2005
)
Nutrition and aberrant DNA methylation patterns in atherosclerosis: more than just hyperhomocysteinemia?
J. Nutr.
135
,
5
8
3
Lund
,
G.
,
Andersson
,
L.
,
Lauria
,
M.
,
Lindholm
,
M.
,
Fraga
,
M.F.
,
Villar-Garea
,
A.
et al.  (
2004
)
DNA methylation polymorphisms precede any histological sign of atherosclerosis in mice lacking apolipoprotein E
.
J. Biol. Chem.
279
,
29147
29154
4
Ghandour
,
H.
,
Chen
,
Z.
,
Selhub
,
J.
and
Rozen
,
R.
(
2004
)
Mice deficient in methylenetetrahydrofolate reductase exhibit tissue-specific distribution of folates
.
J. Nutr.
134
,
2975
2978
5
Chen
,
Z.
,
Karaplis
,
A.C.
,
Ackerman
,
S.L.
,
Pogribny
,
I.P.
,
Melnyk
,
S.
,
Lussier-Cacan
,
S.
et al.  (
2001
)
Mice deficient in methylenetetrahydrofolate reductase exhibit hyperhomocysteinemia and decreased methylation capacity, with neuropathology and aortic lipid deposition
.
Hum. Mol. Genet.
10
,
433
443
6
Thambyrajah
,
J.
and
Townend
,
J.
(
2000
)
Homocysteine and atherothrombosis—mechanisms for injury
.
Eur. Heart J.
21
,
967
974
7
Hofmann
,
M.A.
,
Lalla
,
E.
,
Lu
,
Y.
,
Gleason
,
M.R.
,
Wolf
,
B.M.
,
Tanji
,
N.
et al.  (
2001
)
Hyperhomocysteinemia enhances vascular inflammation and accelerates atherosclerosis in a murine model
.
J. Clin. Invest.
107
,
675
683
8
McNeil
,
C.J.
,
Beattie
,
J.H.
,
Gordon
,
M.-J.
,
Pirie
,
L.P.
and
Duthie
,
S.J.
(
2011
)
Differential effects of nutritional folic acid deficiency and moderate hyperhomocysteinemia on aortic plaque formation and genome-wide DNA methylation in vascular tissue from ApoE−/− mice
.
Clin. Epigenetics
2
,
361
368
9
Zhou
,
J.
,
Møller
,
J.
,
Danielsen
,
C.C.
,
Bentzon
,
J.
,
Ravn
,
H.B.
,
Austin
,
R.C.
et al.  (
2001
)
Dietary supplementation with methionine and homocysteine promotes early atherosclerosis but not plaque rupture in ApoE-deficient mice
.
Arter. Thromb. Vasc. Biol.
21
,
1470
1476
10
Gomez
,
D.
and
Owens
,
G.K.
(
2012
)
Smooth muscle cell phenotypic switching in atherosclerosis
.
Cardiovasc. Res.
95
,
156
164
11
Bloor
,
J.
,
Shukla
,
N.
,
Smith
,
F.C.T.
,
Angelini
,
G.D.
and
Jeremy
,
J.Y.
(
2010
)
Folic acid administration reduces neointimal thickening, augments neo-vasa vasorum formation and reduces oxidative stress in saphenous vein grafts from pigs used as a model of diabetes
.
Diabetologia
53
,
980
988
12
Qureshi
,
I.
,
Chen
,
H.
,
Brown
,
A.T.
,
Fitzgerald
,
R.
,
Zhang
,
X.
,
Breckenridge
,
J.
et al.  (
2005
)
Homocysteine-induced vascular dysregulation is mediated by the NMDA receptor
.
Vasc. Med.
10
,
215
223
13
Troen
,
A.M.
,
Lutgens
,
E.
,
Smith
,
D.E.
,
Rosenberg
,
I.H.
and
Selhub
,
J.
(
2003
)
The atherogenic effect of excess methionine intake
.
Proc. Natl Acad. Sci. U.S.A.
100
,
15089
15094
14
Zhou
,
J.
,
Møller
,
J.
,
Ritskes-Hoitinga
,
M.
,
Larsen
,
M.L.
,
Austin
,
R.C.
and
Falk
,
E.
(
2003
)
Effects of vitamin supplementation and hyperhomocysteinemia on atherosclerosis in apoE-deficient mice
.
Atherosclerosis
168
,
255
262
15
Ebbing
,
M.
,
Bleie
,
Ø.
,
Ueland
,
P.M.
,
Nordrehaug
,
J.E.
,
Nilsen
,
D.W.
,
Vollset
,
S.E.
, et al.  (
2008
)
Mortality and cardiovascular events in patients treated with homocysteine-lowering B vitamins after coronary angiography: a randomized controlled trial
.
JAMA
300
,
795
804
16
Martí-Carvajal
,
A.J.
,
Solà
,
I.
,
Lathyris
,
D.
and
Dayer
,
M.
(
2017
)
Homocysteine-lowering interventions for preventing cardiovascular events
.
Cochrane Database Syst. Rev.
8
,
CD006612
17
Hiltunen
,
M.O.
and
Ylä-Herttuala
,
S.
(
2003
)
DNA methylation, smooth muscle cells, and atherogenesis
.
Arterioscler. Thromb. Vasc. Biol.
23
,
1750
1753
18
Waterland
,
R.A.
(
2003
)
Do maternal methyl supplements in mice affect DNA methylation of offspring?
J. Nutr.
133
,
238
.
author reply 239
19
Waterland
,
R.A.
and
Jirtle
,
R.L.
(
2004
)
Early nutrition, epigenetic changes at transposons and imprinted genes, and enhanced susceptibility to adult chronic diseases
.
Nutrition
20
,
63
68
20
Carter
,
M.
,
Ulrich
,
S.
,
Oofuji
,
Y.
,
Williams
,
D.A.
and
Elizabeth Ross
,
M.
(
1999
)
Crooked tail (Cd) models human folate-responsive neural tube defects
.
Hum. Mol. Genet.
8
,
2199
2204
21
Gomez
,
D.
,
Swiatlowska
,
P.
and
Owens
,
G.K.
(
2015
)
Epigenetic control of smooth muscle cell identity and lineage memory
.
Arterioscler. Thromb. Vasc. Biol.
35
,
2508
2516
22
Ingrosso
,
D.
,
Cimmino
,
A.
,
Perna
,
A.F.
,
Masella
,
L.
,
De Santo
,
N.G.
,
De Bonis
,
M.L.
et al.  (
2003
)
Folate treatment and unbalanced methylation and changes of allelic expression induced by hyperhomocysteinaemia in patients with uraemia
.
Lancet
361
,
1693
1699
23
Brozovich
,
F.V.
,
Nicholson
,
C.J.
,
Degen
,
C.V.
,
Gao
,
Y.Z.
,
Aggarwal
,
M.
and
Morgan
,
K.G.
(
2016
)
Mechanisms of vascular smooth muscle contraction and the basis for pharmacologic treatment of smooth muscle disorders
.
Pharmacol. Rev.
68
,
476
532
24
Liu
,
R.
,
Leslie
,
K.L.
and
Martin
,
K.A.
(
2015
)
Epigenetic regulation of smooth muscle cell plasticity
.
Biochim. Biophys. Acta
1849
,
448
453
25
Hiltunen
,
M.O.
,
Turunen
,
M.P.
,
Häkkinen
,
T.P.
,
Rutanen
,
J.
,
Hedman
,
M.
,
Mäkinen
,
K.
et al.  (
2002
)
DNA hypomethylation and methyltransferase expression in atherosclerotic lesions
.
Vasc. Med.
7
,
5
11
26
Manabe
,
I.
and
Owens
,
G.K.
(
2001
)
Recruitment of serum response factor and hyperacetylation of histones at smooth muscle-specific regulatory regions during differentiation of a novel P19-derived in vitro smooth muscle differentiation system
.
Circ. Res.
88
,
1127
1134
27
Valerio
,
A.
,
Baldo
,
G.
and
Tessari
,
P.
(
2005
)
A rapid method to determine plasma homocysteine concentration and enrichment by gas chromatography/mass spectrometry
.
Rapid Commun. Mass Spectrom.
19
,
561
567
28
Ihaka
,
R.
and
Gentleman
,
R.
(
1996
)
R: a language for data analysis and graphics
.
J. Comput. Graph. Stat.
5
,
299
314
29
Gentleman
,
R.C.
,
Carey
,
V.J.
,
Bates
,
D.M.
,
Bolstad
,
B.
,
Dettling
,
M.
,
Dudoit
,
S.
, et al.  (
2004
)
Bioconductor: open software development for computational biology and bioinformatics
.
Genome Biol.
5
,
R80
30
Ritchie
,
M.E.
,
Phipson
,
B.
,
Wu
,
D.
,
Hu
,
Y.
,
Law
,
C.W.
,
Shi
,
W.
et al.  (
2015
)
Limma powers differential expression analyses for RNA-sequencing and microarray studies
.
Nucleic Acids Res.
43
,
e47
31
Friso
,
S.
,
Choi
,
S.-W.
,
Dolnikowski
,
G.G.
and
Selhub
,
J.
(
2002
)
A method to assess genomic DNA methylation using high-performance liquid chromatography/electrospray ionization mass spectrometry
.
Anal. Chem.
74
,
4526
4531
32
Xiong
,
Z.
and
Laird
,
P.W.
(
1997
)
COBRA: a sensitive and quantitative DNA methylation assay
.
Nucleic Acids Res.
25
,
2532
2534
33
Pogribny
,
I.P.
,
Ross
,
S.A.
,
Wise
,
C.
,
Pogribna
,
M.
,
Jones
,
E.A.
,
Tryndyak
,
V.P.
et al.  (
2006
)
Irreversible global DNA hypomethylation as a key step in hepatocarcinogenesis induced by dietary methyl deficiency
.
Mutat. Res.
593
,
80
87
34
Monje
,
L.
,
Varayoud
,
J.
,
Luque
,
E.H.
and
Ramos
,
J.G.
(
2007
)
Neonatal exposure to bisphenol A modifies the abundance of estrogen receptor alpha transcripts with alternative 5′-untranslated regions in the female rat preoptic area
.
J. Endocrinol.
194
,
201
212
35
Chen
,
T.
,
Mally
,
A.
,
Ozden
,
S.
and
Chipman
,
J.K.
(
2010
)
Low doses of the carcinogen furan alter cell cycle and apoptosis gene expression in rat liver independent of DNA methylation
.
Environ. Health Perspect.
118
,
1597
1602
36
Kolb
,
A.F.
,
Maile
,
J.
,
Heister
,
A.
and
Siddell
,
S.G.
(
1996
)
Characterization of functional domains in the human coronavirus HCV 229E receptor
.
J. Gen. Virol.
77
,
2515
2521
37
Szymanowska
,
M.
,
Hendry
,
K.A.
,
Robinson
,
C.
and
Kolb
,
A.F.
(
2009
)
EMMPRIN (basigin/CD147) expression is not correlated with MMP activity during adult mouse mammary gland development
.
J. Cell. Biochem.
106
,
52
62
38
Homocysteine Studies Collaboration
(
2002
)
Homocysteine and risk of ischemic heart disease and stroke: a meta-analysis
.
JAMA
288
,
2015
2022
39
Dayal
,
S.
,
Bottiglieri
,
T.
,
Arning
,
E.
,
Maeda
,
N.
,
Malinow
,
M.R.
,
Sigmund
,
C.D.
et al.  (
2001
)
Endothelial dysfunction and elevation of S-adenosylhomocysteine in cystathionine β-synthase-deficient mice
.
Circ. Res.
88
,
1203
1209
40
Liew
,
S.-C.
and
Gupta
,
E.D.
(
2015
)
Methylenetetrahydrofolate reductase (MTHFR) C677T polymorphism: epidemiology, metabolism and the associated diseases
.
Eur. J. Med. Genet.
58
,
1
10
41
Spin
,
J.M.
,
Quertermous
,
T.
and
Tsao
,
P.S.
(
2010
)
Chromatin remodeling pathways in smooth muscle cell differentiation, and evidence for an integral role for p300
.
PLoS ONE
5
,
e14301
42
De Bruyn
,
E.
,
Gulbis
,
B.
and
Cotton
,
F.
(
2014
)
Serum and red blood cell folate testing for folate deficiency: new features?
Eur. J. Haematol.
92
,
354
359
43
Alexa
,
A.
,
Rahnenfuhrer
,
J.
and
Lengauer
,
T.
(
2006
)
Improved scoring of functional groups from gene expression data by decorrelating GO graph structure
.
Bioinformatics
22
,
1600
1607
44
McNeil
,
C.J.
,
Hay
,
S.M.
,
Rucklidge
,
G.J.
,
Reid
,
M.
,
Duncan
,
G.
,
Maloney
,
C.A.
et al.  (
2008
)
Disruption of lipid metabolism in the liver of the pregnant rat fed folate-deficient and methyl donor-deficient diets
.
Br. J. Nutr.
99
,
262
271
45
Adaikalakoteswari
,
A.
,
Finer
,
S.
,
Voyias
,
P.D.
,
McCarthy
,
C.M.
,
Vatish
,
M.
,
Moore
,
J.
et al.  (
2015
)
Vitamin B12 insufficiency induces cholesterol biosynthesis by limiting s-adenosylmethionine and modulating the methylation of SREBF1 and LDLR genes
.
Clin. Epigenetics
7
,
14
46
Li
,
Y.
,
Zhang
,
X.
,
Sun
,
Y.
,
Feng
,
Q.
,
Li
,
G.
,
Wang
,
M.
et al.  (
2013
)
Folate deficiency during early-mid pregnancy affects the skeletal muscle transcriptome of piglets from a reciprocal cross
.
PLoS ONE
8
,
e82616
47
McFarlane
,
M.R.
,
Liang
,
G.
and
Engelking
,
L.J.
(
2014
)
Insig proteins mediate feedback inhibition of cholesterol synthesis in the intestine
.
J. Biol. Chem.
289
,
2148
2156
48
Chen
,
E.Y.
,
Tan
,
C.M.
,
Kou
,
Y.
,
Duan
,
Q.
,
Wang
,
Z.
,
Meirelles
,
G.V.
et al.  (
2013
)
Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool
.
BMC Bioinformatics
14
,
128
49
Kuleshov
,
M.V.
,
Jones
,
M.R.
,
Rouillard
,
A.D.
,
Fernandez
,
N.F.
,
Duan
,
Q.
,
Wang
,
Z.
, et al.  (
2016
)
Enrichr: a comprehensive gene set enrichment analysis web server 2016 update
.
Nucleic Acids Res.
44
,
W90
W97
50
Al-Hilal
,
T.A.
,
Chung
,
S.W.
,
Choi
,
J.U.
,
Alam
,
F.
,
Park
,
J.
,
Kim
,
S.W.
et al.  (
2016
)
Targeting prion-like protein doppel selectively suppresses tumor angiogenesis
.
J. Clin. Invest.
126
,
1251
1266
51
Perry
,
S.S.
,
Zhao
,
Y.
,
Nie
,
L.
,
Cochrane
,
S.W.
,
Huang
,
Z.
and
Sun
,
X.-H.
(
2007
)
Id1, but not Id3, directs long-term repopulating hematopoietic stem-cell maintenance
.
Blood
110
,
2351
2360
52
Wen
,
D.
,
Zou
,
Y.-F.
,
Gao
,
Y.-H.
,
Zhao
,
Q.
,
Xie
,
Y.-Y.
,
Shen
,
P.-Y.
, et al.  (
2016
)
Inhibitor of DNA binding 1 is induced during kidney ischemia-reperfusion and is critical for the induction of hypoxia-inducible factor-1α
.
Biomed Res. Int.
2016
,
4634386
53
Hung
,
Y.-H.
,
Chang
,
S.-H.
,
Huang
,
C.-T.
,
Yin
,
J.-H.
,
Hwang
,
C.-S.
,
Yang
,
L.-Y.
et al.  (
2016
)
Inhibitor of differentiation-1 and hypoxia-inducible factor-1 mediate sonic hedgehog induction by amyloid beta-peptide in rat cortical neurons
.
Mol. Neurobiol.
53
,
793
809
54
Guo
,
H.
,
Zhu
,
P.
,
Yan
,
L.
,
Li
,
R.
,
Hu
,
B.
,
Lian
,
Y.
, et al.  (
2014
)
The DNA methylation landscape of human early embryos
.
Nature
511
,
606
610
55
Czeizel
,
A.E.
(
1995
)
Nutritional supplementation and prevention of congenital abnormalities
.
Curr. Opin. Obstet. Gynecol.
7
,
88
94
PMID:
[PubMed]
56
Pitkin
,
R.M.
(
2007
)
Folate and neural tube defects
.
Am. J. Clin. Nutr.
85
,
285S
288S
57
Jacques
,
P.F.
,
Selhub
,
J.
,
Bostom
,
A.G.
,
Wilson
,
P.W.F.
and
Rosenberg
,
I.H.
(
1999
)
The effect of folic acid fortification on plasma folate and total homocysteine concentrations
.
N. Engl. J. Med.
340
,
1449
1454
58
Mason
,
J.B.
,
Dickstein
,
A.
,
Jacques
,
P.F.
,
Haggarty
,
P.
,
Selhub
,
J.
,
Dallal
,
G.
et al.  (
2007
)
A temporal association between folic acid fortification and an increase in colorectal cancer rates may be illuminating important biological principles: a hypothesis
.
Cancer Epidemiol. Biomarkers Prev.
16
,
1325
1329
59
Mathers
,
J.C.
(
2009
)
Folate intake and bowel cancer risk
.
Genes Nutr.
4
,
173
178
60
McNulty
,
H.
,
Pentieva
,
K.
,
Hoey
,
L.
and
Ward
,
M.
(
2008
)
Homocysteine, B-vitamins and CVD
.
Proc. Nutr. Soc.
67
,
232
237
61
Ward
,
M.
,
McNulty
,
H.
,
McPartlin
,
J.
,
Strain
,
J.J.
,
Weir
,
D.G.
and
Scott
,
J.M.
(
1997
)
Plasma homocysteine, a risk factor for cardiovascular disease, is lowered by physiological doses of folic acid
.
QJM
90
,
519
524
62
Antoniades
,
C.
,
Antonopoulos
,
A.S.
,
Tousoulis
,
D.
,
Marinou
,
K.
and
Stefanadis
,
C.
(
2008
)
Homocysteine and coronary atherosclerosis: from folate fortification to the recent clinical trials
.
Eur. Hear. J.
30
,
6
15
63
Hiltunen
,
M.O.
,
Tuomisto
,
T.T.
,
Niemi
,
M.
,
Bräsen
,
J.H.
,
Rissanen
,
T.T.
,
Törönen
,
P.
et al.  (
2002
)
Changes in gene expression in atherosclerotic plaques analyzed using DNA array
.
Atherosclerosis
165
,
23
32
64
Waterland
,
R.A.
and
Michels
,
K.B.
(
2007
)
Epigenetic epidemiology of the developmental origins hypothesis
.
Annu. Rev. Nutr.
27
,
363
388
65
Graves
,
D.C.
,
Yablonka–Reuveni
,
Z.
and
Yablonka-Reuveni
,
Z.
(
2000
)
Vascular smooth muscle cells spontaneously adopt a skeletal muscle phenotype: a unique Myf5/MyoD+ myogenic program
.
J. Histochem. Cytochem.
48
,
1173
1193
66
Goettsch
,
C.
,
Rauner
,
M.
,
Pacyna
,
N.
,
Hempel
,
U.
,
Bornstein
,
S.R.
and
Hofbauer
,
L.C.
(
2011
)
miR-125b regulates calcification of vascular smooth muscle cells
.
Am. J. Pathol.
179
,
1594
1600
67
Feil
,
S.
,
Fehrenbacher
,
B.
,
Lukowski
,
R.
,
Essmann
,
F.
,
Schulze-Osthoff
,
K.
,
Schaller
,
M.
et al.  (
2014
)
Transdifferentiation of vascular smooth muscle cells to macrophage-like cells during atherogenesis
.
Circ. Res.
115
,
662
667
68
Swirski
,
F.K.
and
Nahrendorf
,
M.
(
2014
)
Do vascular smooth muscle cells differentiate to macrophages in atherosclerotic lesions?
Circ. Res.
115
,
605
606
69
Bentzon
,
J.F.
and
Majesky
,
M.W.
(
2018
)
Lineage tracking of origin and fate of smooth muscle cells in atherosclerosis
.
Cardiovasc. Res.
114
,
492
500
70
Duthie
,
S.J.
,
Beattie
,
J.H.
,
Gordon
,
M.-J.
,
Pirie
,
L.P.
,
Nicol
,
F.
,
Reid
,
M.D.
et al.  (
2015
)
Nutritional B vitamin deficiency alters the expression of key proteins associated with vascular smooth muscle cell proliferation and migration in the aorta of atherosclerotic apolipoprotein E null mice
.
Genes Nutr.
10
,
446
71
McKay
,
J.A.
,
Adriaens
,
M.
,
Evelo
,
C.T.
,
Ford
,
D.
and
Mathers
,
J.C.
(
2016
)
Gene promoter DNA methylation patterns have a limited role in orchestrating transcriptional changes in the fetal liver in response to maternal folate depletion during pregnancy
.
Mol. Nutr. Food Res.
60
,
2031
2042
72
Wang
,
X.
,
Guan
,
Z.
,
Chen
,
Y.
,
Dong
,
Y.
,
Niu
,
Y.
,
Wang
,
J.
et al.  (
2015
)
Genomic DNA hypomethylation is associated with neural tube defects induced by methotrexate inhibition of folate metabolism
.
PLoS ONE
10
,
e0121869
73
Leclerc
,
D.
,
Cao
,
Y.
,
Deng
,
L.
,
Mikael
,
L.G.
,
Wu
,
Q.
and
Rozen
,
R.
(
2013
)
Differential gene expression and methylation in the retinoid/PPARA pathway and of tumor suppressors may modify intestinal tumorigenesis induced by low folate in mice
.
Mol. Nutr. Food Res.
57
,
686
697
74
Liu
,
X.S.
,
Wu
,
H.
,
Ji
,
X.
,
Stelzer
,
Y.
,
Wu
,
X.
,
Czauderna
,
S.
et al.  (
2016
)
Editing DNA methylation in the mammalian genome
.
Cell
167
,
233
247.e17
75
Vojta
,
A.
,
Dobrinić
,
P.
,
Tadić
,
V.
,
Bočkor
,
L.
,
Korać
,
P.
,
Julg
,
B.
et al.  (
2016
)
Repurposing the CRISPR-Cas9 system for targeted DNA methylation
.
Nucleic Acids Res.
44
,
5615
5628
76
Lei
,
Y.
,
Zhang
,
X.
,
Su
,
J.
,
Jeong
,
M.
,
Gundry
,
M.C.
,
Huang
,
Y.-H.
et al.  (
2017
)
Targeted DNA methylation in vivo using an engineered dCas9-MQ1 fusion protein
.
Nat. Commun.
8
,
16026
77
Guardamagna
,
O.
,
Abello
,
F.
,
Cagliero
,
P.
and
Lughetti
,
L.
(
2012
)
Impact of nutrition since early life on cardiovascular prevention
.
Ital. J. Pediatr.
38
,
73
78
Haggarty
,
P.
,
Hoad
,
G.
,
Campbell
,
D.M.
,
Horgan
,
G.W.
,
Piyathilake
,
C.
and
McNeill
,
G.
(
2013
)
Folate in pregnancy and imprinted gene and repeat element methylation in the offspring
.
Am. J. Clin. Nutr.
97
,
94
99
79
Dominguez-Salas
,
P.
,
Moore
,
S.E.
,
Baker
,
M.S.
,
Bergen
,
A.W.
,
Cox
,
S.E.
,
Dyer
,
R.A.
, et al.  (
2014
)
Maternal nutrition at conception modulates DNA methylation of human metastable epialleles
.
Nat. Commun.
5
,
3746
80
Stover
,
P.J.
(
2009
)
One-carbon metabolism–genome interactions in folate-associated pathologies
.
J. Nutr.
139
,
2402
2405
81
Cao
,
S.
,
Zhu
,
X.
,
Zhang
,
C.
,
Qian
,
H.
,
Schuttler
,
H.-B.
,
Gong
,
J.
et al.  (
2017
)
Competition between DNA methylation, nucleotide synthesis, and antioxidation in cancer versus normal tissues
.
Cancer Res.
77
,
4185
4195
82
Garcia
,
B.A.
,
Luka
,
Z.
,
Loukachevitch
,
L.V.
,
Bhanu
,
N.V.
and
Wagner
,
C.
(
2016
)
Folate deficiency affects histone methylation
.
Med. Hypotheses
88
,
63
67
83
Duthie
,
S.
(
2011
)
Folate and cancer: how DNA damage, repair and methylation impact on colon carcinogenesis
.
J. Inherit. Metab. Dis.
34
,
101
109
84
Marsit
,
C.J.
,
Eddy
,
K.
and
Kelsey
,
K.T.
(
2006
)
MicroRNA responses to cellular stress
.
Cancer Res.
66
,
10843
10848
85
Silva
,
E.
,
Rosario
,
F.J.
,
Powell
,
T.L.
and
Jansson
,
T.
(
2017
)
Mechanistic target of rapamycin is a novel molecular mechanism linking folate availability and cell function
.
J. Nutr.
147
,
1237
1242
86
Kolb
,
A.F.
and
Petrie
,
L.
(
2013
)
Folate deficiency enhances the inflammatory response of macrophages
.
Mol. Immunol.
54
,
164
172

Supplementary data