The Pharmacogenomics Journal (2009) 9, 291–305 & 2009 Nature Publishing Group All rights reserved 1470-269X/09 $32.00 www.nature.com/tpj REVIEW

Investigation of inter-individual variability of the one-carbon pathway: a bioinformatic and genetic review

DF Carr1, G Whiteley1, Genetic polymorphisms in the one-carbon folate pathway have been widely 1 1 studied in association with a number of conditions. Most of the research has A Alfirevic and M Pirmohamed , focused on the 677C4T polymorphism in the coding region of the 5, on behalf of the FolATED study 10-methylenetetrahydrofolate reductase (MTHFR) . However, there are a team total of 25 in this pathway coding for , transporters and receptors, which can be investigated using 267 tagging single 1Department of Pharmacology and Therapeutics, polymorphisms (SNPs); using SNP database (dbSNP), 38 non-synonymous University of Liverpool, Liverpool, Merseyside, UK SNPs with a minor allele frequency of 45% are present in these genes. Most of Correspondence: these variants have not been investigated in relation to disease or drug Prof. M Pirmohamed, Department of response phenotypes. In addition, their functional consequences are largely Pharmacology and Therapeutics, Sherrington unknown. Prediction of the functional effect using six publicly available Building, Ashton Street, Liverpool, Merseyside programs (PolyPhen, SIFT BLink, PMut, SNPs3D, I-Mutant2.0 and LS-SNP) was L69 3GE, UK. 4 E-mail: [email protected] limited to functionally well-characterized SNPs such as MTHFR c.677C T and c.1298A4C ranking low. Epigenetic modifications may also be important with some of these genes. In summary, to date, investigation of the one-carbon folate pathway genes has been limited. Future studies should aim for a more comprehensive assessment of this pathway, while further research is also required in determining the functional effects of these genetic variants. The Pharmacogenomics Journal (2009) 9, 291–305; doi:10.1038/tpj.2009.29; published online 7 July 2009

Keywords: one-carbon folate; genetics; bioinformatics; MTHFR; ;

Introduction

The folate pathway, commonly known as folate-mediated one-carbon metabo- lism, refers to a system consisting of a number of interdependent pathways that use a family of structurally similar and metabolically inter-convertable cofactors. These cofactors are structurally based on tetrahydrofolate (THF) and are used to chemically activate single carbons (one-carbon units) (Figure 1). They are key components in the synthesis of purine and thymidine and in the of to methionine, a process that is crucial to the regulation of normal patterns of DNA methylation (Figure 1). The one-carbon folate and methionine synthesis pathways involve a considerable number of enzymes and carrier (Figure 1). Received 27 March 2009; revised 8 May 2009; accepted 26 May 2009; published Many genetic studies have been undertaken looking for associations between online 7 July 2009 disease and polymorphisms in the multiple genes involved in one-carbon One-carbon folate pathway genetics DF Carr et al 292

Figure 1 Schematic representation of the one-carbon folate and methionine metabolism pathways.

metabolism and methionine synthesis pathways.1–4 These Folate pathophysiology and disease studies have focused, primarily, on identifying associations between disease risk and non-synonymous (ns) and pre- Elevation of homocysteine is a sensitive marker of folate viously well-characterized polymorphisms. However, given deficiency. Homocysteine, a sulfur-containing , the number of genes that are involved in the folate pathway cannot be obtained through any dietary source. It is solely (Figure 1), investigation of disease associations with other, the product of the methylation cycle, which is responsible less-well investigated genomic loci may be warranted and for its removal as well as its formation (Figure 1). Elevated may uncover new genetic associations. homocysteine levels and folate deficiency have been This review discusses the importance of the one-carbon associated with a number of disorders, including cardiovas- metabolism pathway in physiology and disease, and how cular disease, psychiatric disorders, adverse pregnancy out- genetic variation predisposes to disease and variability in comes, cancer and osteoporosis. Many of these disorders can response to medications. We have undertaken both a be explained, in part, by the interruption of folate- literature review and a bioinformatic analysis of the path- dependant one-carbon transfer and by the subsequent effect way, and will discuss the limitations of current approaches on methionine, purine and thymidylate synthesis (Figure 1). and of in silico tools that are used to predict the functionality Three specific but inter-linked underlying causes are be- of ns single nucleotide polymorphisms (nsSNPs). lieved to be important: common polymorphisms within genes of the one-carbon folate pathway; impaired of folate-metabolizing genes; and poor dietary Literature searching methods folate intake. The number of diseases associated with altered Journal articles relating to disease associations with 5,10- one-carbon metabolism has led to extensive investigations methylenetetrahydrofolate reductase (MTHFR), dihydrofo- of the genetic variants in this pathway, their metabolic late reductase (DHFR) and thymidylate synthase (TYMS) effects and the associated risk of chronic diseases.5,6 polymorphisms were identified in National Center for Bio- technology Information (NCBI) PubMed database (http:// www.ncbi.nlm.nih.gov/sites/entrez?db ¼ ), using Folate pharmacology the search terms ‘genetics’ and ‘MTHFR,’ ‘DHFR’ or ‘TYMS,’ respectively. In the case of MTHFR, articles were limited to Folate supplementation ‘Meta-analysis’ because of the very large number of Folate is found in abundance in a number of dietary sources, individual studies. In all cases, articles were limited to particularly leafy vegetables, dried beans and peas. Some English language only. foods (particularly breakfast cereals) are also fortified with

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 293

folate in the form of the synthetic analog, folic acid, which folate supplementation would be of benefit in treating can also be taken as a dietary supplement. Folic acid schizophrenic patients. There is also evidence that folic acid supplementation has been used as a preventive strategy in may improve cognitive function. In a 3-year study of the a number of clinical conditions. In 1998, a recommended elderly, a daily intake of 800 mg of oral folic acid significantly dietary allowance of 400 mg dayÀ1 of folic acid intake was improved those domains of cognitive function that deterio- established in the United States. rate with age.23

Neural tubule defects Cardiovascular disease Given its key role in thymidine synthesis and thus in DNA There has been a great deal of focus on the association replication, folate is particularly important during periods of between serum homocysteine levels and the risk of cardio- rapid cell division such as pregnancy. Adequate folate intake vascular disease (CVD). According to a meta-analysis, a during the peri-conceptional periods protects against con- decrease in serum homocysteine by 3 mmol lÀ1 should reduce genital malformations such as neural tubule defects the risk of ischemic heart disease deep vein thrombosis and (NTDs).7 The risk of NTDs is significantly decreased when stroke by 16, 25 and 24%, respectively.24 Such a reduction in a healthy diet is supplemented with multivitamins, princi- homocysteine would be achievable by folic acid supplemen- pally folic acid.8,9 Indeed, studies in the United States have tation of 0.8 mg dayÀ1.25 Other studies, however, have suggested a significant decrease in the prevalence of spina suggested that folic acid intervention, in the presence and bifida and anencephaly as the Food and Drug Administra- absence of vitamin B6, does not lower the risk of recurrent tion (FDA) mandated the addition of folic acid to all cardiovascular disease or death after acute myocardial enriched cereal grain products by January 1998.10–12 infarction.26 This finding is mirrored in studies in which patients with vascular disease showed no decrease in the risk Cancer of major cardiovascular events after supplementation with 27 Low-plasma folate levels have been associated with an folic acid and vitamin B6 and B12. increased incidence of a number of types of cancer. Folate Despite the many apparently beneficial effects of folate deficiency seems to predispose normal tissue to neoplastic supplementation, and its low incidence of toxicity,28 some transformation, and folate supplementation can suppress concern has been expressed regarding the interaction 13 the development of tumors in normal tissue. For instance, between vitamin B12 and folic acid, primarily in the elderly a 20–40% reduction in the risk of colorectal cancers and population. Folic acid is able to mask nervous system precursory adenomas has been reported in individuals with changes caused by vitamin B12 deficiency by correcting the the highest folate status as against those with the lowest.13 A associated anemia. The upper limit of recommended dietary number of meta-analyses have postulated an association intake of 1 mg, set by the Institute of Medicine, was believed between high folate and a mild protective effect against the to be unlikely to produce masking.29 However, this has risk of several cancers, including colorectal,14 pancreatic, proved controversial with some, suggesting that intake of esophageal, gastric,15 oral16 and ovarian.17 o1 mg may exert a masking effect.30 Indeed, the United Conversely, and controversially, there is some evidence to Kingdom’s Food Standards Agency (FSA) Board recom- suggest that high folate may indeed be a risk factor for some mended against mandatory folic acid fortification, partly forms of cancer. A double-blind, placebo-controlled, rando- due to the potential masking of vitamin B12 deficiency until mized trial showed that the folic acid supplementation was May 2007, when mandatory fortification was finally agreed associated with higher risk of colorectal adenoma and (http://www.food.gov.uk/healthiereating/folicfortification/). prostate cancer.18 There is also evidence to suggest that folic acid supplementation increases the risk of breast cancer Molecular and population genetics of the one-carbon in postmenopausal women.19 Whether these are true folate pathway associations or epiphenomena requires further study, but nevertheless highlights the complexity of the one-carbon The sheer number of recorded variants within the genes of folate pathway and its interaction with carcinogenesis. the one-carbon and methionine metabolism pathways highlights just how many avenues of investigation are left Psychiatric disorders unexplored in the search for genetic risk factors for disease Folate supplementation may be beneficial in the treatment and inter-individual variability in response to drugs. Most of depressive illness as an adjuvant to antidepressant attention has focused on a small number of well-character- therapy.20 However, the small numbers of studies under- ized polymorphisms in functionally well-defined genes. It is taken so far limit the conclusions that can be drawn, our intention to describe these, and additionally discuss although further studies are underway. These include a those variants that have been less well-characterized. multicentered double-blind, placebo-controlled, rando- mized trial of folic acid augmentation of pragmatic MTHFR antidepressant treatment of moderate-to-severe depression Most investigations in recent years have focused on MTHFR (FolATED).21 A possible association between folate defi- (EC, 1.5.1.20), and in particular the c.677C4T (p.A222V) ciency and has also been reported.22 How- non-synonymous polymorphism (rs1801133). Identified in ever, further investigation is required to determine whether 1995,31 the MTHFR C677T polymorphism, encoding a

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 294

valine to alanine amino-acid substitution at residue 222, has greater DHFR mRNA expression in lymphocytes compared a long established and marked effect on folate and homo- with ( þ / þ ) carriers.42 The 50-upstream 9-bp repeat poly- cysteine levels.32 Phenotypically, this variant has morphism (homozygous 6R carriers)43 and 30-UTR SNP reduced catalytic activity and thermolability, and is thus (rs34764978) (C/T and T/T carriers)44 have also been associated with elevated homocysteine levels under condi- associated with enhanced mRNA expression. tions of impaired folate status. A vast amount of research investigating MTHR C677T and A1298C (rs1801131) as One-carbon metabolism genetic variant phenotypes disease risk factors has been conducted. A summary of the Besides well-characterized polymorphisms in genes such as meta-analyses of this research is shown in Table 1. MTHFR, TYMS and DHFR, a number of genetic variants in other genes in these pathways (Figure 1) have been TYMS identified. Hyperhomocysteinemia and hypermethionemia Using 5,10-methylenetetrahydrofolate as a , TYMS have been reported as clinical consequences of functional (EC, 2.1.1.45) catalyzes the methylation of deoxyuridylate polymorphisms in genes involved in the methionine to thymidylate (dTMP). As the only de novo source of synthesis pathway. thymidylate in the cell, TYMS is critical for DNA replication and repair. This makes it an important target for chemother- Homocystinuria and hypomethionemia apeutic agents including 5- and methotrexate. A First described in 1962 in a cohort of patients with mental considerable number of genetic studies have been under- retardation,45 homocystinuria is the result of impairment of taken to determine associations between allelic variants in the methionine synthesis pathway. Major clinical manifes- the TYMS gene and both risk of disease and response to tations affect the central nervous, vascular and skeletal chemotherapeutics (Table 2). The most commonly studied systems. Sequence variations in the cystathionine-beta- polymorphism in the TYMS gene is a 28-bp tandem repeat synthase gene (CBS; EC, 4.2.1.22) leading to CBS protein region in the 50- (UTR). First described in deficiency were the first to be reported as being associated 1995,33,34 it was observed in vitro that an increase in the with homocystinuria.46,47 More recently, non-synonymous number of repeats resulted in a stepwise increase in the variant alleles have been identified in the CBS and TYMS mRNA expression. In vivo studies in human gastro- cystathionase (CTH; EC, 4.4.1.1) genes in patients with intestinal tumor samples were consistent with the in vitro hyperhomocysteinemia. Homozygotes for the c.1352G4T findings: TSER*3 carriers showed significant increases in variant allele (rs1021737) in the CTH gene encoding a both mRNA and protein expression compared with TSER*2 p.S403I amino-acid change had significantly higher plasma individuals.35 A novel G4C polymorphism in the 12th homocysteine levels than other genotypes in a cohort of 496 nucleotide of the TSER*3 has also been identified. This Caucasian individuals.48 allelic variant has been shown to decrease TYMS expres- An allelic variation in the CBS gene, encoding a p.G307S sion.36 A third characterized polymorphism in the TYMS amino-acid change, has been identified in homocystinuria gene is a 6-bp deletion in the 30-UTR, 447 bp downstream of patients of Celtic ancestry.49,50 Another protein variant, the stop codon.37 Homozygotes for this deletion show lower p.I278T (rs5742905, c.833T4C), has also been identified in mRNA expression than homozygotes for the presence of the Celtic homocystinuria patients.50,51 Heterozygotes for 6-bp deletion,38 which is also a determinant of elevated red p.I278T, however, are believed to retain some degree of blood cell folate.39 pyridoxine responsiveness to homocystinuria.50,51 The elevated homocysteine phenotype in individuals with DHFR impaired CBS or CTH is not unexpected, given the key role DHFR (EC, 1.5.1.3) reduces dihydrofolate (DHF), formed both have in the synthesis of cystathione from homocys- during dTMP synthesis, back to THF, which subsequently teine (Figure 1). rejoins the pool of folate cofactors. All folate forms in 5-Methyltetrahydrofolate-homocysteine methyltransfer- vitamin supplements and fortified food are in the form of ase (MTR; EC, 2.1.1.13) is responsible for the final step in folic acid, an unreduced folate, and requires DHFR for the methionine biosynthetic pathway. Mutations in the reduction. Thus, DHFR is vital to the bioavailability of folic MTR gene result in methylcobalamin deficiency G (cblG) acid in cellular reactions. The importance of the DHFR disorder, which is characterized by homocystinuria, and protein is perhaps illustrated by the finding that the coding elevated homocysteine and depleted methionine levels.52,53 region is highly conserved with no allelic variants having A common non-synonymous variant (p.P1173L) has been been identified.40 There are three non-coding polymorph- detected in 465% of cblG patients.54 In addition, 13 novel isms (in the 50 upstream region, intron 1 and 30-UTR) that mutations, including 5 deletions and 2 nonsense mutations have been the subject of a number of association studies were identified, which resulted in the synthesis of truncated (Table 3). The 19-bp deletion polymorphism in intron 1 in proteins that lacked portions critical for enzyme function.54 the DHFR gene has been associated with a number of 5-Methyltetrahydrofolate-homocysteine methyltransfer- diseases41 (Table 3). Individuals homozygous for the dele- ase reductase (MTRR; EC, 1.16.1.8) is responsible for the tion (À/À) have lower mean plasma total homocysteine regeneration of a functional MTR. Given that MTR is the than those homozygous for the insertion ( þ / þ ),40 with the final step in the of methionine (Figure 1), (À/À) individuals also having been shown to have 4.8-fold MTRR functionality is a key component of intracellular

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 295

Table 1 Genetic associations of commonly characterized variant alleles of the MTHFR gene

Disease Analysis cohort size Allelic variant Odds/hazard ratio (95% CI) Statistical significance Cases Controls

Colorectal cancer91–93 12 261 18 463 C677T (TT vs CC) 0.83 (0.94–1.04) (RE) P ¼ 0.001 1949 3099 C677T (T vs C) 0.83 (0.68–1.01) (FE) P40.10 10 131 13 362 C677T (T vs C) 0.93 (0.89–0.98) (FE) P ¼ 0.0003 4764 6592 A1298C (C vs A) 0.81 (0.70–0.94) (RE) P ¼ 0.005

Breast cancer94,95 6373 8436 C677T (TT vs CC) 1.05 (0.88–1.25) (RE) NS 5467 7336 C677T (T vs C) 1.02 (0.95–1.10) (RE) NS 3768 5276 A1298C (C vs A) 0.98 (0.89–1.07) (RE) NS

Bipolar disorder96 550 1098 C677T (TT vs CC) 1.82 (1.22–2.70) (FE) Unipolar depression96,97 1280 10 429 C677T (TT vs CC) 1.36 (1.11–1.67) (FE) P ¼ 0.003 Gastric cancer15,98 1584 2785 C677T (T vs C) 1.27 (1.13–1.44) (RE) 1930 2930 C677T (CC vs TT) 1.68 (1.29–2.19)

Gastric adenocarcinoma15 604 1655 C677T (CC vs TT) 1.90 (1.38–2.60) Schizophrenia96,99–101 2762 3363 C677T (TT vs CC) 1.44 (1.21–1.70) (FE) 2380 2852 C677T (T vs C) 1.13 (1.04–1.23) (FE) 1111 1454 A1298C (C vs A) 1.16 (1.03–1.31) (FE) 2265 2721 C677T (TT vs CC) 1.36 (1.07–1.72) (RE) 1119 1308 C677T (TT vs CC+CT) 1.48 (1.18–1.86) (FE)

Alzheimer’s disease102 735 837 A1298C (C vs A) 0.85 (0.73–1.00) (RE) P ¼ 0.05 Childhood ALL103 1914 2980 C677T (TT vs CC+CT) 0.88 (0.73–1.06) (FE) P ¼ 0.18 Adult ALL103,104 245 532 C677T (TT vs CC+CT) 0.45 (0.26–0.77) (FE) P ¼ 0.04 1576 1958 C677T (T vs C) 0.88 (0.76–1.02) (RE) 1377 1682 A1298C (C vs A) 0.88 (0.72–1.07) (RE) Ischemic stroke105,106 6110 8760 C677T (T vs C) 1.17 (1.09–1.26) (RE) Po0.001 C677T (TT vs CC+CT) 1.37 (1.15–1.64) (RE) Po0.001 1533 2786 C677T (T vs C) 1.46 (1.19–1.79) (RE)

First arterial stroke: children107 3235 9019 C677T (T vs C) 1.70 (1.23–2.34) (FE) Coronary heart disease108,109 26 000 31.183 C677T (TT vs CC) 1.14 (1.05–1.24) (RE) NS 11 162 12 758 C677T (TT vs CC+CT) 1.16 (1.05–1.28)

Ischemic heart disease24 12 193 11 945 C677T TT vs CC) 1.21 (1.06–1.39) (RE) Coronary artery disease110,111 735 1250 C677T (TT vs CC) 1.21 (0.87–1.68) 2419 3225 C677T (TT vs CC+CT) 1.30 (1.11–1.52)

Retinopathy in type II diabetes112 435 617 C677T (T vs C) 1.39 (1.05–1.83) (RE) Retinal vein occlusion113,114 690 2754 C677T (TT vs CC+CT) 1.332 (0.995–1.783) P ¼ 0.054 581 1080 C677T (T vs C) 1.2 (0.9–1.6)

Retinal artery occlusion113 152 435 C677T (TT vs CC+CT) 1.716 (0.977–3.014) P ¼ 0.060 Venous thrombosis115 8364 12 468 C677T (TT vs CC) 1.20 (1.08–1.32) Venous thromboembolism116 4901 7886 C677T (TT vs CC+CT) 1.2 (1.1–1.4) (RE) Hypertension in pregnancy117 3169 3044 C677T (T vs C) 1.21 (1.01–1.44) (FE) Pre-eclampsia118: all 2250 2028 C677T (TT vs CC+CT) (0.79–1.29) (RE) severe 676 1165 C677T (TT vs CC+CT) 1.38 (0.93–2.06) (RE) Neural tubule defects119: patients 219 2542 C677T (T vs C) 1.9 (1.3–2.8) mothers 166 1.7 (1.1–2.6) fathers 110 1.8 (1.1–3.1) Recurrent pregnancy loss120,121 2120 2949 C677T (TT vs CC+CT) 1.40 (1.11–1.77) 599 498 C677T (TT vs CC+CT) 1.40 (1.0–2.0)

Abbreviations: ALL, acute lymphoblastic leukemia; MTHFR, methylenetetrahydrofolate reductase; NS, no statistical significance. Data represent a summary of published meta-analyses for cohorts of mixed ethnicity. RE and FE denote the use of a random effects or fixed effects statistical model, respectively, where stated by the authors. Where no P-value is stated, significance is stated but no P-value derived.

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 296

Table 2 Genetics association of commonly characterized variant alleles of the thymidylate synthase gene

Disease Genotyped cohort size Ethnicity Allelic variant Statistical Odds/hazard significance ratio (95% CI) Cases Controls

Survival benefit in gastric cancer 51 Mixed TSER*2 P ¼ 0.002(2*/2*) treated with 5-FU122 P ¼ 0.004(2*/3*)

Survival benefit in colorectal cancer 129 Mixed TSER*3/*3 P ¼ 0.020 0.38 (0.16–0.93) treated with 5-FU123 1494del6 3R/À6bp P ¼ 0.0034 0.42 (0.22–0.82) haplotype P ¼ 0.17 0.42 (0.20–0.85) (compared with 2R/+6bp)

Risk of non-Hodgkin lymphoma124 332 713 Mixed TSER*3/*3 P ¼ 0.53 1.1 (0.77–1.6) 333 727 1494del6 P ¼ 0.02 (À/À)/ 0.57 (0.34–0.94)/ P ¼ 0.64 (+/À) 0.94 (0.71–1.2)

ALL—poor prognosis with 32 173 Mixed TSER*3/*3 P ¼ 0.005 5.2 (1.6–16.9) methotrexate treatment125,126 104 403 Mixed TSER*3/*3 P ¼ 0.006 2.1 (1.2–3.7) *3/C/6bp+ P ¼ 0.004 haplotype P ¼ 0.04 2*/6bpÀ (protective)

Methotrexate-related alopecia in 54 160 Mixed TSER*2/*2 Po0.01 5.38 (3.05–9.49) rheumatoid arthritis127

Breast cancer risk128,129 575 610 Caucasian 1494del6 — 0.97 (0.76–1.24) (À/À) 1.22 (0.81–1.85) (+/À) 473 473 Chinese 1494del6 P ¼ 0.026 —

Susceptibility to malignant lymphoma130 108 494 Japanese TSER*2 P ¼ 0.030 1.63 (1.05–2.53)

homocysteine levels. CblE type of homocystinuria, a rare abnormalities.60 A heterozygous non-synonymous allelic autosomal-recessive disorder manifesting as megaloblastic variant (p.R264H) has also been identified in individuals anemia, has been observed in patients carrying functional with autosomal-dominant inheritance of persistent hyper- mutations in the MTRR gene.55,56 methioninemia.61 S-adenosylhomocysteine (AHCY; EC, 3.3.1.1) is Hypermethioninemia responsible for catalyzing the reversible hydrolysis of Deficiencies in specific genes of the methionine synthesis S-adenosylhomocysteine to adenosine and homocysteine pathway can result in isolated hypermethioninemia. The (Figure 1). Deficiency of AHCY protein is another potential most common cause of hypermethioninemia is deficiency cause of hypermethionemia.62 A case report62 noted a of the enzyme methionine adenosyltransferase (MAT 1 Croatian boy with hypermethionemia, who possessed two alpha) (MAT1A; EC, 2.5.1.6), which catalyzes the transfer variant alleles encoding non-synonymous changes of the adenosyl moiety of ATP to methionine to form (p.W112X and p.Y143C) in 4 of the AHCY gene. In S-adenosylmethionine, a key component in DNA methyla- addition, further coding-region variants in the AHCY gene tion (Figure 1). Although many patients with MAT defi- have been shown to have no significant association with ciency are well clinically,57 other patients have been altered plasma homocysteine levels.63 This was assessed in reported to have neurological events,58 including low venous thrombosis patients selected according to their S-adenosylmethionine, calcification of basal ganglia and ability to convert methionine to homocysteine (as measured demyelination. by standardized methionine-loading test). Functional genetic polymorphisms in the gene encoding methionine adenosyltransferase 1alpha have been identified Clinical implications of other folate metabolism pathway genetic in patients with methionine adenosyltransferase deficiency polymorphisms and hypermethionemia. Identified variants of MAT1A Functional genetic variants in other genes in the methio- include a homozygous truncating mutation identified in nine synthesis and one-carbon folate metabolism pathways an individual with isolated hypermethioninemia59 and in a have been implicated in a range of conditions. Given its role girl with MAT deficiency not displaying neurological in catalyzing the conversion of betaine and homocysteine to

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 297 )

) dimethylglycine and methionine, defects in betaine-homo- ) ) ) À / À ) ) À À À / / À cysteine (BHMT; EC, 2.1.1.5) could lead to À À À / / À

À À homocystinuria. However, to date, no association between elevated homocysteine and functional genetic variants has NA NA NA been observed. An association has been described between a BHMT non-synonymous SNP (encoding a p.R239Q amino-

1.14 (0.94–1.38) 1.18 (0.93–1.51) acid substitution) (rs3733890) and the risk of placental 64 65 0.50 (0.15,1.62) (+/ 0.80 (0.15,4.62) ( 1.26 (0.96–1.66) (+/ 1.52 (1.08–2.13) ( abruption and NTDs. Gamma-glutamyl hydrolase (GGH; EC, 3.4.19.9) catalyzes the hydrolysis of folylpoly-gamma-glutamates by removing )

À g-linked glutamate and polyglutamates. Similar to a number /

À of other genes within the one-carbon folate pathway, 0.58 0.18 0.02 0.06 0.05 0.050.050.050.05 0.8 (0.4–1.5) ( 1.2 (0.6–2.2) ( NA NA

0.001 polymorphisms in the GGH gene have been associated 0.0498 2.04 (0.936–4.288) ( ¼ ¼ ¼ ¼ 4 4 4 4 4 o ¼ P P P P P P P P P 0.006 (

P with lower response to administration of the anti-folate, P genotypes) (all observed ¼

P methotrexate, for the treatment of various conditions including rheumatoid arthritis 66 and acute lymphoblastic leukemia.67 The enzyme 5-aminoimidazole-4-carboxamide ribonu- cleotide formyltransferase/IMP cyclohydrolase (ATIC; EC,

Mixed 2.1.2.3) is able to convert 10-formyl THF to THF (Figure 1). Earlier studies have described an association between individuals possessing the GG genotype of a common c.347C4G; p.T112S polymorphism (rs2372536) and a 573 1092 decreased response to methotrexate in the treatment of rheumatoid arthritis.68 A case has also been reported of AICA-ribosiduria, a disease characterized by deficiency in purine biosynthesis, where an infant possessed a K426R 533

Cases Control amino-acid substitution in ATIC on one allele and a frameshift mutation on the other.69 This individual pre- sented with dysmorphic features, severe neurological defects and congenital blindness. (AMT; EC, 2.1.2.10) forms part of the responsible for the catalysis of the conversion of 5,10-methenyl-THF to 5- formyl tetrahydrofolate (Figure 1). A number of non- synonymous allelic variants have been identified in patients Upstream 9 bp repeat 20 Caucasian 0 with typical and atypical nonketotic hyperglycinemia.70–74 Typical nonketotic hyperglycinemia presents with seizures,

T) (rs34764978)coma and 37 apnea 83 in neonates, Japanese whereas the rarer atypical

4 form presents as only developmental delay with some seizures. Three functional allelic variants (two non-synonymous (p.R135C and p.R299P) in formiminotransferase cyclodea- Upstream 9 bp repeatUpstream 9 bp repeat 109 91 234 206 -UTR SNP (C 0 0 0

19 bp Intron-1 deletion (rs70991108)19 bp Intron-1 deletion5 19 bp Intron-1 deletion5 19 bp Intron-1 deletion5 5019 219 bp Intron-1 deletion 115 Mixed 1013 244 208minase Mixed 1066 (FTCD; EC, 2.1.2.5/4.3.1.4) have been reported in individuals with glutamate formininotransferase defi- 75 40 ciency. This is an autosomal-recessive disorder and the second most common inborn error of folate metabolism.

40 Common features include physical and mental retardation and elevated serum folate. In addition to genes encoding key enzymes in the one- carbon and methionine synthesis pathways, there is also multi-vitamin users 41,43 43 44 genetic variability in genes encoding the folate transporter and receptor proteins. Among the most important are solute carrier family 19 (folate influx transporter), member 1 Genetics association of commonly characterized variant alleles of the dihydrofolate reductase gene (SLC19A1; MIM# 600424) and folate receptor 1 (FOLR1; MIM# 136430). Coordination of SLC19A1 and FOLR1 is proposed as the mechanism of folate uptake, specifically Enhanced mRNA expression Incidence of childhood leukemia/lymphoma Spina bifida: children Lower plasma homocysteine levels Risk of breast cancer: total Table 3 DiseaseSpina bifida: mothers Allelic variantAbbreviation: NA, not applicable. 5-methyl-THF, Genotyped cohort size Ethnicity in many Statistical significance Odds/hazard ratio (95% types CI) of mammalian cells.

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 298

An amino-acid substitution polymorphism encoding a northern and western Europe) population were included histidine to change at residue 27 of SLC19A1 and the number of SNPs with a minor allelic frequency (rs1051266) has been associated with decreased methotrex- (MAF) 45% was determined using the Haploview 3.2 software ate therapeutic response in individuals with rheumatoid (http://www.broad.mit.e-du/mpg/haploview/). The number of arthritis68 and childhood acute lymphoblastic leukemia.76 tagging SNPs within each loci required to represent those with This same p.H27R polymorphism has also been identified as MAF 45% was determined using the ‘Tagger’ functionality a risk factor for NTDs.77–79 An effect on absorption and with Haploview by means of a pairwise analysis of linkage cellular translocation, and blood pressure, in the elderly has disequilibrium with an R2 threshold of 0.8. also been described.80 FOLR1 polymorphisms have been Mining of data from HapMap public release #22 (Table 4) described earlier as susceptibility factors for gastric cancer in showed a total of 1252 validated SNPs within the genomic a Chinese population.81 regions of the 25 included genes, of these, 814 SNPs have a Cellular folate concentration has been shown to modulate minor allelic frequency of X5%. The analysis of 267 drug-efflux transporters, in particular ABCC1 (MIM# representative tagging SNPs is required to cover the 814 158343).82 Consistent with this is the recent finding that variants based on a pairwise analysis of linkage disequili- genetic variants in ABCC1 act as determinants of response to brium with an R2 threshold of 0.8. Further investigation of the anti-folate drug, methotrexate, in patients with psor- the NCBI dbSNP Build 129 identified 6742 SNPs within the iasis.83 25 genes. A total of 127 were classified as non-synonymous SNPs, with 38 of these validated and at a minor allelic Epigenetics frequency of X5%. The one-carbon folate metabolism system is a critical component of global methylation of CpG sites on DNA. In silico prediction of nsSNP functionality. In total, 38 non- Studies in models of folate deficiency both in vitro84 and synonymous SNPs were analyzed in silico using a panel of six in vivo85 have shown decreased DNA methylation capacity. programs (Table 5) to predict the presence of a deleterious Associations between global DNA hypomethylation in effect of the amino-acid substitution (Table 6). Each peripheral leukocytes and homozygosity for the MTHFR program uses a different approach and can differ in their C677T polymorphism have also been described.86,87 Studies prediction of deleterious SNPs. For this reason, we decided in human vascular smooth muscle cells have suggested that to analyze SNPs using a panel rather than a specific program. homocysteine could induce demethylation in the The programs selected can predict deleterious SNPs altering region of the MTHFR gene, thus upregulating mRNA stability, such as a loss of a salt bridge or burial of a polar- expression.88 It is feasible; therefore, demethylation of the charged group and, in some cases, the loss of a key residue MTHFR promoter may counter the dysfunction of the 677T involved in an of ligand binding. The input data protein variant in heterozygotes by increasing the expres- required consisted of structural or sequence information, or sion of the wild-type allele. However, this has not been in some cases both. Specific information regarding the systematically investigated. programs is summarized in Table 5. By using a range of software platforms, each requiring different input data and Bioinformatic analysis using different algorithms, a consensus of opinion on the As shown in the preceding sections, literature searching functionality of nsSNPs within the one-carbon folate methods have shown that the one-carbon and methionine pathway was obtained. Essentially, each program used was synthesis pathways are complex and consist of a number of weighted equally for its prediction of deleterious effects. enzymes. There is undoubted variability in the function of SNPs were ranked by a score out of 8 (the number of these enzymes, but to date, most of the studies have focused different prediction models used). on several key genetic variants, notably MTHFR c.677C4T. The 38 non-synonymous SNPs in the one-carbon folate Where there have been studies of the other genes, this has pathway according to the consensus output of the panel of largely focused on rare mutations and often (understand- predictive programs ranked the p.P450R amino-acid sub- ably) in single patients. In order to more fully describe the stitution of MTRR (rs16879334) as the most likely to exhibit common and rare variants in these genes in these pathways, an effect on protein functionality. To the best of our and predict their functionality, we also undertook a knowledge, however, there are no reports that this poly- bioinformatic analysis. morphism has shown functional effect on the protein or has The key genes were identified primarily from the follow- been associated with any clinical manifestations. Intrigu- ing Kyoto Encyclopaedia of Genes and Genomes (KEGG) ingly, three polymorphisms commonly associated with pathways (http://www.genome.jp/kegg/): One Carbon Pool clinical manifestations, MTHFR (p.V222A/c.677C4T and by Folate (00670), Methionine Metabolism (00680) and p.E429A/c.1298A4G) (Table 1), and CTH (p.I403S/ Folate Biosynthesis (00790). A list of 26 genes was subse- c.1352G4t),89 were ranked as low as 27, 23 and 36 out of quently compiled (Table 4). 38, respectively. Validated SNPs within these genes were identified in The fact that three well-characterized functional poly- HapMap Public Release 22, Build 35 (http://www.hapmap. morphisms were ranked so low within the panel of SNPs org/cgi-perl/gbrowse/hapmap_B35/). All database SNPs raises important issues regarding the use of in silico SNP within the CEPH (Utah residents with ancestry from functionality prediction techniques. These results would

The Pharmacogenomics Journal Table 4 Summary of data-based SNPs of one-carbon and methionine metabolism-associated genes from HapMap (release #22) and dbSNP (Build 129)

Gene Full name Gene ID MIM# KEGG HapMap public release #22) dbSNP (Build 129) (HUGO) Enzyme# Total SNPs Tagging Total Non- Validated SNPs (MAF SNPs SNPs synonymous (MAF 45%) required SNPs 45%)

AHCY S-adenosylhomocysteine hydrolase 191 180960 20cen–q13.1 3.3.1.1 31 11 1 207 7 0 ALDH1L1 aldehyde dehydrogenase 1 family, member L1 10840 600249 3q21.2 1.5.1.6 198 172 55 674 12 6 AMD1 adenosylmethionine decarboxylase 1 262 180980 6q21–q22 4.1.1.50 18 13 5 154 0 0 AMT aminomethyltransferase 275 238310 3p21.2–p21.1 2.1.2.10 5 5 2 43 1 1 ATIC 5-aminoimidazole-4-carboxamide ribonucleotide 471 601731 2q35 2.1.2.3 57 45 13 289 3 1 formyltransferase/IMP cyclohydrolase BHMT betaine-homocysteine methyltransferase 635 602888 5q13.1–q15 2.1.1.5 30 17 13 143 8 1 CBS cystathionine-beta-synthase 875 236200 21q22.3 4.2.1.22 34 25 17 230 5 0 CTH cystathionase (cystathionine gamma-) 1491 607657 1p31.1 4.4.1.1 35 24 15 172 2 1 DHFR dihydrofolate reductase 1719 126060 5q11.2–q13.2 1.5.1.3 35 27 5 277 0 0 DNMT1 DNA (cytosine-5-)-methyltransferase 1 1786 126375 19p13.2 2.1.1.37 48 31 8 294 3 2 FOLH1 folate hydrolase (prostate-specific membrane antigen) 1 2346 600934 11p11.2 3.4.17.21 65 40 10 641 5 1 FPGS folylpolyglutamate synthase 2356 136510 9q34.1 6.3.2.17 3 1 1 93 6 0 FTCD formiminotransferase cyclodeaminase 10841 229100 21q22.3 2.1.2.5 21 12 11 196 2 0

GART phosphoribosylglycinamide formyltransferase 2618 138440 21q22.11 2.1.2.2 31 17 6 168 11 2 Carr DF genetics pathway folate One-carbon GGH gamma-glutamyl hydrolase (conjugase, 8836 601509 8q12.3 3.4.19.9 29 17 7 151 4 3

folylpolygammaglutamyl hydrolase) al et MAT1A methionine adenosyltransferase I, alpha 4143 250850 10q22 2.5.1.6 42 29 12 148 1 1 MTFMT mitochondrial methionyl-tRNA formyltransferase 123263 NA 15q22.31 2.1.2.9 15 9 4 159 2 0 MTHFD1 methylenetetrahydrofolate dehydrogenase 4522 172460 14q24 1.5.1.5/ 55 32 17 385 9 2 (NADP+ dependent) 1-like 1.5.1.15/ 6.3.4.3/ 3.5.4.9 MTHFR 5,10-methylenetetrahydrofolate reductase (NADPH) 4524 236250 1p36.3 1.5.1.20 57 33 15 281 15 3 MTHFS 5,10-methenyltetrahydrofolate synthetase 10588 604197 15q25.1 6.3.3.2 88 52 19 323 1 1 (5-formyltetrahydrofolate cyclo-) MTR 5-methyltetrahydrofolate-homocysteine methyltransferase 4548 156570 1q43 2.1.1.13 206 113 5 653 6 2 MTRR 5-methyltetrahydrofolate-homocysteine methyltransferase 4552 602568 5p15.3–p15.2 1.16.1.8 60 40 15 281 10 8 reductase h hraoeoisJournal Pharmacogenomics The SHMT1 serine hydroxymethyltransferase 1 (soluble) 6470 182144 17p11.2 2.1.2.1 43 21 5 244 2 1 SLC19A1 solute carrier family 19 (folate transporter), member 1 6573 600424 21q22.3 — 25 13 2 281 8 2 TYMS thymidylate synthetase 7298 188350 18p11.32 2.1.1.45 21 15 4 255 4 0

1252 814 267 6742 127 38

Abbreviations: dbSNP, SNP database; MAF, minor allelic frequency; SNP, single nucleotide polymorphism. 299 One-carbon folate pathway genetics DF Carr et al 300

appear to question the sensitivity of the methodology in identifying functional nsSNPs associated with a character- ized phenotype. The limited number of available input protein structures for one-carbon folate pathway genes appears to bias the ranking toward nsSNPs within structu- rally well-characterized genes. The reason for the absence of structural data is often because of the difficulty in the crystallization process of, for example, membrane-bound proteins. One possible refinement of our protocol might be to produce structural data using homology modeling or threading. However, this can also be prone to considerable error. Were it possible to produce predictive structural data, these files could only be used by two programs, I-Mutant2.0 and SDM. However, whether the use of in silico predicted structural information would be beneficial to the outcomes

emidio/I-Mutant/I-Mutant.htm of this work in the case of SDM seems unlikely, as it has the http://blocks.fhcrc.org/sift/SIFT.html http://gpcr2.biocomp.unibo.it/ B http://alto.compbio.ucsf.edu/LS-SNP/Queries.html ability to introduce significant errors into the prediction algorithm. I-Mutant2.0 output may be enhanced by the use of such input data, although further investigation is required. All of the programs used within this study require different inputs and use algorithms of differing sensitivity. However, the need for more publicly available data relating to the tertiary structure appears to limit the utility of our methodology in identifying candidate nsSNPs of interest within a pathway setting.

Transgenic models Although a robust genetic association in a clinical study provides some evidence of functionality of a variant in that gene or surrounding loci, it is also important that this is shown in functional studies, particularly given the limita- tions of bioinformatic approaches outlined above. A power- ful tool for functional evaluation is the use of transgenic animal models, which can provide important information on genotype–phenotype relationships and help in evaluat- ing the clinical implications of dysfunctional folate pathway Probably damaging; possibly damaging; benignaffect protein function (low-confidencePredicted score); as tolerated pathological or neutral with a reliabilityPredicted http://genetics.bwh.harvard.edu/pph/ score as deleterious; not predicted as http://mmb2.pcb.ub.es:8080/PMut/ deleteriousPredicted as decreasing stabilitywith or a increasing reliability stability score http://www.snps3d.org confidence or neutral with high confidence, low confidence genes. To date, however, only a limited number of folate pathway gene knockout mouse models have been developed and phenotypically characterized. Of these, both MTHFR and CBS display a hyperhomocysteinemia phenotype (Table 7). A MAT1A-null mouse model has also been 89 Sequence and structure structure structure structure Sequence Predicted as disease associated with high confidence, low characterized. This knockout has impaired liver regenera- tion and spontaneous hepatocellular carcinoma. The loca- lization of these events is likely to be due to the high expression levels of MAT1A in hepatic tissue.89 Knockout mice null for the FOLR1 homolog, Folbp1,90 led to embryonic lethality, which could be reversed in hetero- zygote dams with supplementation.

Sequence profile and structural properties Sequence conservation SequenceNeural network Predicted to affectSVM—Stability protein function; predicted to SequenceSVM—Stability and SequenceSVM and and knowledge- based rules Sequence and Conclusions

135 It is clear that polymorphisms in genes in the one-carbon Summary of computational methods used in the analysis of non-synonymous SNPs 132 131 folate pathway are associated with a dynamic range of 134 136 clinical conditions. What appears to not be so dynamic is 133 the scope of genetic polymorphisms investigated in this ProgramPolyPhen Method Input Output URL Table 5 Abbreviation: SNP, single nucleotide polymorphism. SIFT Blink Pmut SNPs3D I-Mutant 2.0 LS-SNP large number of studies. A functional polymorphism, such

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 301

Table 6 Analysis and scoring parameters of non-synonymous SNPs

Minor Amino Acid Allele

Gene RS # Substitution Frequency Polyphen SIFT Blink PMut SNPs3D Profile SNPs3D Structure I-Mutant2.0 Sequence I-Mutant2.0 Structure LS-SNP 1 MTRR rs16879334 P450R 0.152 2 MAT1A rs1143693 Q119H 0.439 3 MTRR rs2287780 R415C 0.155 4 MTHFD1 rs2236225 R653Q 0.414 5 ALDH1L1 rs3796191 L254P 0.086 6 SHMT1 rs1979277 L474F 0.375 7 MTHFD2 rs1950902 R134K 0.299 8 ALDH1L1 rs1127717 D793G 0.288 9 ALDH1L1 rs9282691 E429A 0.149 10 ALDH1L1 rs2886059 V330F 0.393 11 GART rs8788 V421I 0.458 12 MTRR rs10380 H595Y 0.346 13 MTRR rs1801394 I22M 0.468 14 MTRR rs10064631 L333V 0.132 15 AMT rs34812788 A51V 0.061 16 SLC19A1 rs35786590 A558V 0.499 17 FOLH1 rs202676 H75Y 0.465 18 ALDH1L1 rs2276724 S481G 0.427 19 GART rs8971 D752G 0.264 20 MTHFR rs2274976 R594Q 0.126 21 MTHFS rs8923 T202A 0.087 22 MTR rs1805087 D919G 0.324 23 MTHFR rs1801131 E429A 0.340 24 DNMT1 rs16999593 R97H 0.135 25 ALDH1L1 rs4646750 I812V 0.143 26 DNMT1 rs8111085 I311V 0.365 27 MTHFR rs1801133 V222A 0.409 28 SLC19A1 rs1051266 H27R 0.487 29 ATIC rs2372536 T116S 0.300 30 BHMT rs3733890 R239Q 0.389 31 GGH rs1800909 C6R 0.351 32 GGH rs13270305 A31T 0.362 33 GGH rs11545078 T151I 0.153 34 MTR rs2229274 D314N 0.051 35 MTRR rs2303080 S257T 0.080 36 CTH rs1021737 I403S 0.304 37 MTRR rs1532268 S175L 0.339 38 MTRR rs162036 K350R 0.383

Program Black Grey White

Polyphen Probably Damaging Possibly Damaging Benign SIFT Blink Predicted to affect protein function Predicted to affect protein function (low confidence score) Tolerated PMut Predicted as Pathological (reliability of >5) Predicted as Pathological (reliability of <5) Predicted as Neutral SNPs3d Predicted as Deleterious - Not predicted as Deleterious I-Mutant2.0 Predicted as Decreasing Stability (reliability of >5) Predicted as Decreasing Stability (reliability of <5) Predicted not to Decrease Stability LS-SNP Predicted as Disease associated with high confidence - Predicted as Neutral

Abbreviations: nsSNP, non-synonymous SNP; SNP, single nucleotide polymorphism. In silico analysis of the deleterious effects of non-synonymous SNPs in genes of the one-carbon folate pathway (a). nsSNPs determined as deleterious by programs are marked as black; white squares indicate a non-deleterious effect and gray an intermediate effect according to the scoring parameters indicated in (b). SNPs are ranked according to the number of black scores followed by the number of gray scores. Cross-hatched squares indicate where in sufficient input data is available to produce an output.

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 302

Table 7 Transgenic animal models of impaired gene function metabolism pathway and associations with colorectal cancer. Cancer of the folate and methionine synthesis pathways Epidemiol Biomarkers Prev 2006; 15: 2408–2417. 3 Lissowska J, Gaudet MM, Brinton LA, Chanock SJ, Peplonska B, Welch R Gene Knockout mouse pathogenesis et al. Genetic polymorphisms in the one-carbon metabolism pathway 137–140 and breast cancer risk: a population-based case–control study and CBS Hyperhomocysteinemia meta-analyses. Int J Cancer 2007; 120: 2696–2703. Hyperkeratosis 4 Moore LE, Malats N, Rothman N, Real FX, Kogevinas M, Karami S et al. Polymorphisms in one-carbon metabolism and trans-sulfuration path- MAT1A89 Spontaneous hepatocellular carcinoma and way genes and susceptibility to bladder cancer. Int J Cancer 2007; 120: impaired liver regeneration 2452–2458. MTHFR85,141 Hyperhomocysteinemia (À/+ and À/À) 5 Gellekink H, den Heijer M, Heil SG, Blom HJ. Genetic determinants of Developmental retardation with cerebellar plasma total homocysteine. Semin Vasc Med 2005; 5: 98–109. 6 Molloy AM. Folate and homocysteine interrelationships including pathology (À/À) genetics of the relevant enzymes. Curr Opin Lipidol 2004; 15: 49–57. Aortic lipid deposition (À/+ and À/À) 7 Shaw GM, Schaffer D, Velie EM, Morland K, Harris JA. Periconceptional Abnormal spermatogenesis (À/À) vitamin use, dietary folate, and the occurrence of neural tube defects. Epidemiology 1995; 6: 219–226. FOLR1 (Folbp1 Embryonic lethality (À/À) 8 Milunsky A, Jick H, Jick SS, Bruell CL, MacLaughlin DS, Rothman KJ mouse homolog)90 reversed by folinic acid supplementation et al. Multivitamin/folic acid supplementation in early pregnancy of (À/+) dams reduces the prevalence of neural tube defects. JAMA 1989; 262: 2847–2852. 9 Mulinare J, Cordero JF, Erickson JD, Berry RJ. Periconceptional use of multivitamins and the occurrence of neural tube defects. JAMA 1988; 260: 3141–3145. as the MTHFR 677C4T, given its clear and well-character- 10 Honein MA, Paulozzi LJ, Mathews TJ, Erickson JD, Wong LY. Impact of ized effect on the enzymatic function of the protein and folic acid fortification of the US food supply on the occurrence of associated elevated homocysteine levels, is a good candidate neural tube defects. JAMA 2001; 285: 2981–2986. polymorphism in clinical studies where folate disruption is 11 Williams LJ, Mai CT, Edmonds LD, Shaw GM, Kirby RS, Hobbs CA et al. Prevalence of spina bifida and anencephaly during the transition to believed to have a key function. However, there are more mandatory folic acid fortification in the United States. Teratology 2002; than 25 genes in the one-carbon folate and methionine 66: 33–39. pathways. It is entirely feasible that, with greater investiga- 12 Williams LJ, Rasmussen SA, Flores A, Kirby RS, Edmonds LD. Decline in tion of some of the ‘lesser lights’ of the folate pathway the prevalence of spina bifida and anencephaly by race/ethnicity: 1995–2002. Pediatrics 2005; 116: 580–586. genes, other functionally relevant variants could be identi- 13 Kim YI. Folate and colorectal cancer: an evidence-based critical review. fied in disease association studies. Mol Nutr Food Res 2007; 51: 267–292. The use of in silico techniques represents a potentially 14 Sanjoaquin MA, Allen N, Couto E, Roddam AW, Key TJ. Folate intake useful tool in the prediction of the deleterious effects of and colorectal cancer risk: a meta-analytical approach. Int J Cancer 2005; 113: 825–828. nsSNPs. However, at present, with the limited availability of 15 Larsson SC, Giovannucci E, Wolk A. Folate intake, MTHFR polymorph- structural data, it is difficult to perform such analysis on a isms, and risk of esophageal, gastric, and pancreatic cancer: a meta- pathway-wide scale, as we have with the one-carbon folate analysis. Gastroenterology 2006; 131: 1271–1283. metabolism pathway. In future, a greater deposition of 16 Pelucchi C, Talamini R, Negri E, Levi F, Conti E, Franceschi S et al. Folate intake and risk of oral and pharyngeal cancer. Ann Oncol 2003; protein structure data on public databases will allow more 14: 1677–1681. thorough examination of all nsSNPs. This may improve 17 Navarro Silvera SA, Jain M, Howe GR, Miller AB, Rohan TE. Dietary sensitivity for detecting nsSNPs, which not only have a folate consumption and risk of ovarian cancer: a prospective cohort functional effect on the protein but also are associated with study. Eur J Cancer Prev 2006; 15: 511–515. 18 Cole BF, Baron JA, Sandler RS, Haile RW, Ahnen DJ, Bresalier RS et al. a phenotype. Folic acid for the prevention of colorectal adenomas: a randomized clinical trial. JAMA 2007; 297: 2351–2359. 19 Stolzenberg-Solomon RZ, Chang SC, Leitzmann MF, Johnson KA, Conflict of interest Johnson C, Buys SS et al. Folate intake, alcohol use, and postmenopausal breast cancer risk in the Prostate, Lung, Colorectal, The authors declare no conflict of interest. and Ovarian Cancer Screening Trial. Am J Clin Nutr 2006; 83: 895–904. 20 Taylor MJ, Carney SM, Goodwin GM, Geddes JR. Folate for depressive Acknowledgments disorders: systematic review and meta-analysis of randomized con- trolled trials. J Psychopharmacol 2004; 18: 251–256. We acknowledge the funding provided by the National Institute for 21 Roberts SH, Bedson E, Hughes DA, Lloyd KR, Moat S, Pirmohamed M Health Research under the Health Technology Assessment Pro- et al. Folate Augmentation of Treatment – Evaluation for Depression gramme Grant 04/35/08 to the FolATED trial. (FolATED): Protocol of a randomised controlled trial. BMC Psychiatry 2007; 7:65. 22 Muntjewerff JW, Blom HJ. Aberrant folate status in schizophrenic References patients: what is the evidence? Prog Neuropsychopharmacol Biol Psychiatry 2005; 29: 1133–1139. 1 Fredriksen A, Meyer K, Ueland PM, Vollset SE, Grotmol T, Schneede J. 23 Durga J, van Boxtel MP, Schouten EG, Kok FJ, Jolles J, Katan MB et al. Large-scale population-based metabolic phenotyping of thirteen Effect of 3-year folic acid supplementation on cognitive function in genetic polymorphisms related to one-carbon metabolism. Hum Mutat older adults in the FACIT trial: a randomised, double blind, controlled 2007; 28: 856–865. trial. Lancet 2007; 369: 208–216. 2 Koushik A, Kraft P, Fuchs CS, Hankinson SE, Willett WC, Giovannucci EL 24 Wald DS, Law M, Morris JK. Homocysteine and cardiovascular disease: et al. Nonsynonymous polymorphisms in genes in the one-carbon evidence on causality from a meta-analysis. BMJ 2002; 325: 1202.

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 303

25 Wald DS, Bishop L, Wald NJ, Law M, Hennessy E, Weir D et al. 45 Gerritsen T, Vaughn JG, Waisman HA. The identification of Randomized trial of folic acid supplementation and serum homo- homocysteine in the urine. Biochem Biophys Res Commun 1962; 9: cysteine levels. Arch Intern Med 2001; 161: 695–700. 493–496. 26 Bonaa KH, Njolstad I, Ueland PM, Schirmer H, Tverdal A, Steigen T 46 Boers GH, Fowler B, Smals AG, Trijbels FJ, Leermakers AI, Kleijer WJ et al. Homocysteine lowering and cardiovascular events after acute et al. Improved identification of heterozygotes for homocystinuria due myocardial infarction. N Engl J Med 2006; 354: 1578–1588. to cystathionine synthase deficiency by the combination of methio- 27 Lonn E, Yusuf S, Arnold MJ, Sheridan P, Pogue J, Micks M et al. nine loading and enzyme determination in cultured fibroblasts. Hum Homocysteine lowering with folic acid and B vitamins in vascular Genet 1985; 69: 164–169. disease. N Engl J Med 2006; 354: 1567–1577. 47 Mudd SH, Skovby F, Levy HL, Pettigrew KD, Wilcken B, Pyeritz RE et al. 28 Campbell NR. How safe are folic acid supplements? Arch Intern Med The natural history of homocystinuria due to cystathionine beta- 1996; 156: 1638–1644. synthase deficiency. Am J Hum Genet 1985; 37: 1–31. 29 Institute of Medicine. Folate. In: Dietary reference intakes of thiamin, 48 Wang J, Huff AM, Spence JD, Hegele RA. Single nucleotide roboflavin, niacin, vitamin B6, folate, vitamin B12, pantothenic acid, polymorphism in CTH associated with variation in plasma homo- biotin, and choline. National Academy Press: Washington, DC, 1998. cysteine concentration. Clin Genet 2004; 65: 483–486. pp 196–305. 49 Gallagher PM, Ward P, Tan S, Naughten E, Kraus JP, Sellar GC et al. 30 Shane B. Folate fortification: enough already? Am J Clin Nutr 2003; 77: High frequency (71%) of cystathionine beta-synthase mutation G307S 8–9. in Irish homocystinuria patients. Hum Mutat 1995; 6: 177–180. 31 Frosst P, Blom HJ, Milos R, Goyette P, Sheppard CA, Matthews RG et al. 50 Hu FL, Gu Z, Kozich V, Kraus JP, Ramesh V, Shih VE. Molecular basis of A candidate genetic risk factor for vascular disease: a common cystathionine beta-synthase deficiency in pyridoxine responsive and mutation in methylenetetrahydrofolate reductase. Nat Genet 1995; nonresponsive homocystinuria. Hum Mol Genet 1993; 2: 1857–1860. 10: 111–113. 51 Shih VE, Fringer JM, Mandell R, Kraus JP, Berry GT, Heidenreich RA 32 Ueland PM, Hustad S, Schneede J, Refsum H, Vollset SE. Biological and et al. A missense mutation (I278T) in the cystathionine beta-synthase clinical implications of the MTHFR C677T polymorphism. Trends gene prevalent in pyridoxine-responsive homocystinuria and asso- Pharmacol Sci 2001; 22: 195–201. ciated with mild clinical phenotype. Am J Hum Genet 1995; 57: 34–39. 33 Horie N, Aiba H, Oguro K, Hojo H, Takeishi K. Functional analysis and 52 Leclerc D, Campeau E, Goyette P, Adjalla CE, Christensen B, Ross M DNA polymorphism of the tandemly repeated sequences in the et al. Human : CDNA cloning and identification of 50-terminal regulatory region of the human gene for thymidylate mutations in patients of the cblG complementation group of folate/ synthase. Cell Struct Funct 1995; 20: 191–197. cobalamin disorders. Hum Mol Genet 1996; 5: 1867–1874. 34 Horie N, Takeishi K. Functional structure of the promoter region of the 53 Gulati S, Baker P, Li YN, Fowler B, Kruger W, Brody LC et al. Defects in human thymidylate synthase gene and nuclear factors that human methionine synthase in cblG patients. Hum Mol Genet 1996; 5: regulate the expression of the gene. Nucleic Acids Symp Ser 1995: 1859–1865. 34: 77–78. 54 Watkins D, Ru M, Hwang HY, Kim CD, Murray A, Philip NS et al. 35 Kawakami K, Omura K, Kanehira E, Watanabe Y. Polymorphic tandem Hyperhomocysteinemia due to methionine synthase deficiency, cblG: repeats in the thymidylate synthase gene is associated with its protein structure of the MTR gene, genotype diversity, and recognition of a expression in human gastrointestinal cancers. Anticancer Res 1999; common mutation, P1173L. Am J Hum Genet 2002; 71: 143–153. 19(4B): 3249–3252. 55 Zavadakova P, Fowler B, Zeman J, Suormala T, Pristoupilova K, 36 Mandola MV, Stoehlmacher J, Muller-Weeks S, Cesarone G, Yu MC, Kozich V et al. CblE type of homocystinuria due to methionine Lenz HJ et al. A novel single nucleotide polymorphism within the 50 synthase reductase deficiency: clinical and molecular studies and tandem repeat polymorphism of the thymidylate synthase gene prenatal diagnosis in two families. J Inherit Metab Dis 2002; 25: 461–476. abolishes USF-1 binding and alters transcriptional activity. Cancer Res 56 Leclerc D, Wilson A, Dumas R, Gafuik C, Song D, Watkins D et al. 2003; 63: 2898–2904. Cloning and mapping of a cDNA for methionine synthase reductase, a 37 Ulrich CM, Bigler J, Velicer CM, Greene EA, Farin FM, Potter JD. flavoprotein defective in patients with homocystinuria. Proc Natl Acad Searching expressed sequence tag databases: discovery and confirma- Sci USA 1998; 95: 3059–3064. tion of a common polymorphism in the thymidylate synthase gene. 57 Gaull GE, Tallan HH, Lonsdale D, Przyrembel H, Schaffner F, von Cancer Epidemiol Biomarkers Prev 2000; 9: 1381–1385. Bassewitz DB. Hypermethioninemia associated with methionine 38 Mandola MV, Stoehlmacher J, Zhang W, Groshen S, Yu MC, Iqbal S adenosyltransferase deficiency: clinical, morphologic, and biochemical et al. A 6 bp polymorphism in the thymidylate synthase gene causes observations on four patients. J Pediatr 1981; 98: 734–741. message instability and is associated with decreased intratumoral TS 58 Surtees R, Leonard J, Austin S. Association of demyelination mRNA levels. Pharmacogenetics 2004; 14: 319–327. with deficiency of cerebrospinal-fluid S-adenosylmethionine in 39 Kealey C, Brown KS, Woodside JV, Young I, Murray L, Boreham CA inborn errors of methyl-transfer pathway. Lancet 1991; 338: et al. A common insertion/deletion polymorphism of the thymidylate 1550–1554. synthase (TYMS) gene is a determinant of red blood cell folate and 59 Hazelwood S, Bernardini I, Shotelersuk V, Tangerman A, Guo J, homocysteine concentrations. Hum Genet 2005; 116: 347–353. Mudd H et al. Normal brain myelination in a patient homozygous for a 40 Gellekink H, Blom HJ, van der Linden IJ, den Heijer M. Molecular mutation that encodes a severely truncated methionine adenosyl- genetic analysis of the human dihydrofolate reductase gene: relation I/III. Am J Med Genet 1998; 75: 395–400. with plasma total homocysteine, serum and red blood cell folate levels. 60 Chamberlin ME, Ubagai T, Mudd SH, Wilson WG, Leonard JV, Chou JY. Eur J Hum Genet 2007; 15: 103–109. Demyelination of the brain is associated with methionine adenosyl- 41 Johnson WG, Stenroos ES, Spychala JR, Chatkupt S, Ming SX, Buyske S. transferase I/III deficiency. J Clin Invest 1996; 98: 1021–1027. New 19 bp deletion polymorphism in intron-1 of dihydrofolate 61 Chamberlin ME, Ubagai T, Mudd SH, Levy HL, Chou JY. Dominant reductase (DHFR): a risk factor for spina bifida acting in mothers inheritance of isolated hypermethioninemia is associated with a during pregnancy? Am J Med Genet A 2004; 124: 339–345. mutation in the human methionine adenosyltransferase 1A gene. Am 42 Xu X, Gammon MD, Wetmur JG, Rao M, Gaudet MM, Teitelbaum SL J Hum Genet 1997; 60: 540–546. et al. A functional 19- deletion polymorphism of dihydrofolate 62 Baric I, Fumic K, Glenn B, Cuk M, Schulze A, Finkelstein JD et al. reductase (DHFR) and risk of breast cancer in multivitamin users. Am S-adenosylhomocysteine hydrolase deficiency in a human: a genetic J Clin Nutr 2007; 85: 1098–1102. disorder of methionine metabolism. Proc Natl Acad Sci USA 2004; 101: 43 van der Linden IJ, Nguyen U, Heil SG, Franke B, Vloet S, Gellekink H 4234–4239. et al. Variation and expression of dihydrofolate reductase (DHFR) in 63 Gellekink H, den Heijer M, Kluijtmans LA, Blom HJ. Effect of genetic relation to spina bifida. Mol Genet Metab 2007; 91: 98–103. variation in the human S-adenosylhomocysteine hydrolase gene on 44 Goto Y, Yue L, Yokoi A, Nishimura R, Uehara T, Koizumi S et al. A novel total homocysteine concentrations and risk of recurrent venous single-nucleotide polymorphism in the 30-untranslated region of the thrombosis. Eur J Hum Genet 2004; 12: 942–948. human dihydrofolate reductase gene with enhanced expression. Clin 64 Ananth CV, Elsasser DA, Kinzler WL, Peltier MR, Getahun D, Leclerc D Cancer Res 2001; 7: 1952–1956. et al. Polymorphisms in methionine synthase reductase and

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 304

betaine-homocysteine S-methyltransferase genes: risk of placental 82 Hooijberg JH, Jansen G, Assaraf YG, Kathmann I, Pieters R, Laan AC abruption. Mol Genet Metab 2007; 91: 104–110. et al. Folate concentration dependent transport activity of the multidrug 65 Boyles AL, Billups AV, Deak KL, Siegel DG, Mehltretter L, Slifer SH et al. resistance protein 1 (ABCC1). Biochem Pharmacol 2004; 67: 1541–1548. Neural tube defects and folate pathway genes: family-based associa- 83 Warren RB, Smith RL, Campalani E, Eyre S, Smith CH, Barker JN et al. tion tests of gene–gene and gene–environment interactions. Environ Genetic variation in efflux transporters influences outcome to Health Perspect 2006; 114: 1547–1552. methotrexate therapy in patients with psoriasis. J Invest Dermatol 66 Dervieux T, Kremer J, Lein DO, Capps R, Barham R, Meyer G et al. 2008; 128: 1925–1929. Contribution of common polymorphisms in reduced folate carrier and 84 Stempak JM, Sohn KJ, Chiang EP, Shane B, Kim YI. Cell and stage of gamma-glutamylhydrolase to methotrexate polyglutamate levels in transformation-specific effects of folate deficiency on methionine cycle patients with rheumatoid arthritis. Pharmacogenetics 2004; 14: intermediates and DNA methylation in an in vitro model. Carcinogen- 733–739. esis 2005; 26: 981–990. 67 Cheng Q, Wu B, Kager L, Panetta JC, Zheng J, Pui CH et al. A substrate 85 Chen Z, Karaplis AC, Ackerman SL, Pogribny IP, Melnyk S, Lussier- specific functional polymorphism of human gamma-glutamyl hydro- Cacan S et al. Mice deficient in methylenetetrahydrofolate reductase lase alters catalytic activity and methotrexate polyglutamate accumu- exhibit hyperhomocysteinemia and decreased methylation capacity, lation in acute lymphoblastic leukaemia cells. Pharmacogenetics 2004; with neuropathology and aortic lipid deposition. Hum Mol Genet 2001; 14: 557–567. 10: 433–443. 68 Dervieux T, Furst D, Lein DO, Capps R, Smith K, Walsh M et al. 86 Stern LL, Mason JB, Selhub J, Choi SW. Genomic DNA hypomethyla- Polyglutamation of methotrexate with common polymorphisms in tion, a characteristic of most cancers, is present in peripheral reduced folate carrier, aminoimidazole carboxamide ribonucleotide leukocytes of individuals who are homozygous for the C677T transformylase, and thymidylate synthase are associated with polymorphism in the methylenetetrahydrofolate reductase gene. methotrexate effects in rheumatoid arthritis. Arthritis Rheum 2004; Cancer Epidemiol Biomarkers Prev 2000; 9: 849–853. 50: 2766–2774. 87 Axume J, Smith SS, Pogribny IP, Moriarty DJ, Caudill MA. The MTHFR 69 Marie S, Heron B, Bitoun P, Timmerman T, Van Den Berghe G, 677TT genotype and folate intake interact to lower global leukocyte Vincent MF. AICA-ribosiduria: a novel, neurologically devastating DNA methylation in young Mexican American women. Nutr Res 2007; inborn error of purine biosynthesis caused by mutation of ATIC. Am J 27: 1365–1317. Hum Genet 2004; 74: 1276–1281. 88 Wang L, Zhang J, Wang S. [Demethylation in the promoter region of 70 Kure S, Kato K, Dinopoulos A, Gail C, DeGrauw TJ, Christodoulou J MTHFR gene and its mRNA expression in cultured human vascular et al. Comprehensive mutation analysis of GLDC, AMT, and smooth muscle cells induced by homocysteine]. Wei Sheng Yan Jiu GCSH in nonketotic hyperglycinemia. Hum Mutat 2006; 27: 2007; 36: 291–294. 343–352. 89 Chen L, Zeng Y, Yang H, Lee TD, French SW, Corrales FJ et al. Impaired 71 Kure S, Mandel H, Rolland MO, Sakata Y, Shinka T, Drugan A et al. A liver regeneration in mice lacking methionine adenosyltransferase 1A. missense mutation (His42Arg) in the T-protein gene from a large FASEB J 2004; 18: 914–916. Israeli-Arab kindred with nonketotic hyperglycinemia. Hum Genet 90 Piedrahita JA, Oetama B, Bennett GD, van Waes J, Kamen BA, 1998; 102: 430–434. Richardson J et al. Mice lacking the folic acid-binding protein Folbp1 72 Kure S, Shinka T, Sakata Y, Osamu N, Takayanagi M, Tada K et al. A are defective in early embryonic development. Nat Genet 1999; 23: one-base deletion (183delC) and a missense mutation (D276H) in the 228–232. T-protein gene from a Japanese family with nonketotic hyperglycine- 91 Hubner RA, Houlston RS. MTHFR C677T and colorectal cancer mia. J Hum Genet 1998; 43: 135–137. risk: a meta-analysis of 25 populations. Int J Cancer 2007; 120: 73 Nanao K, Okamura-Ikeda K, Motokawa Y, Danks DM, Baumgartner ER, 1027–1035. Takada G et al. Identification of the mutations in the T-protein gene 92 Chen K, Jiang QT, He HQ. Relationship between metabolic enzyme causing typical and atypical nonketotic hyperglycinemia. Hum Genet polymorphism and colorectal cancer. World J Gastroenterol 2005; 11: 1994; 93: 655–658. 331–335. 74 Toone JR, Applegarth DA, Coulter-Mackie MB, James ER. Recurrent 93 Huang Y, Han S, Li Y, Mao Y, Xie Y. Different roles of MTHFR C677T mutations in P- and T-proteins of the glycine cleavage complex and a and A1298C polymorphisms in colorectal adenoma and colorectal novel T-protein mutation (N145I): a strategy for the molecular cancer: a meta-analysis. J Hum Genet 2007; 52: 73–85. investigation of patients with nonketotic hyperglycinemia (NKH). 94 Lewis SJ, Harbord RM, Harris R, Smith GD. Meta-analyses of Mol Genet Metab 2001; 72: 322–325. observational and genetic association studies of folate intakes or 75 Hilton JF, Christensen KE, Watkins D, Raby BA, Renaud Y, de la Luna S levels and breast cancer risk. J Natl Cancer Inst 2006; 98: et al. The molecular basis of glutamate formiminotransferase defi- 1607–1622. ciency. Hum Mutat 2003; 22: 67–73. 95 Zintzaras E. Methylenetetrahydrofolate reductase gene and 76 Laverdiere C, Chiasson S, Costea I, Moghrabi A, Krajinovic M. susceptibility to breast cancer: a meta-analysis. Clin Genet 2006; 69: Polymorphism G80A in the reduced folate carrier gene and its 327–336. relationship to methotrexate plasma levels and outcome of childhood 96 Gilbody S, Lewis S, Lightfoot T. Methylenetetrahydrofolate reductase acute lymphoblastic leukemia. Blood 2002; 100: 3832–3834. (MTHFR) genetic polymorphisms and psychiatric disorders: a HuGE 77 De Marco P, Calevo MG, Moroni A, Merello E, Raso A, Finnell RH et al. review. Am J Epidemiol 2007; 165: 1–13. Reduced folate carrier polymorphism (80A–4G) and neural tube 97 Lewis SJ, Lawlor DA, Davey Smith G, Araya R, Timpson N, Day IN et al. defects. Eur J Hum Genet 2003; 11: 245–252. The thermolabile variant of MTHFR is associated with depression in the 78 Morin I, Devlin AM, Leclerc D, Sabbaghian N, Halsted CH, Finnell R British Women’s Heart and Health Study and a meta-analysis. Mol et al. Evaluation of genetic variants in the reduced folate carrier and in Psychiatry 2006; 11: 352–360. glutamate carboxypeptidase II for spina bifida risk. Mol Genet Metab 98 Zintzaras E. Association of methylenetetrahydrofolate reductase 2003; 79: 197–200. (MTHFR) polymorphisms with genetic susceptibility to gastric cancer: 79 Shaw GM, Lammer EJ, Zhu H, Baker MW, Neri E, Finnell RH. a meta-analysis. J Hum Genet 2006; 51: 618–624. Maternal periconceptional vitamin use, genetic variation of infant 99 Zintzaras E. C677T and A1298C methylenetetrahydrofolate reductase reduced folate carrier (A80G), and risk of spina bifida. Am J Med Genet gene polymorphisms in schizophrenia, bipolar disorder and depres- 2002; 108: 1–6. sion: a meta-analysis of genetic association studies. Psychiatr Genet 80 Dufficy L, Naumovski N, Ng X, Blades B, Yates Z, Travers C et al. G80A 2006; 16: 105–115. reduced folate carrier SNP influences the absorption and cellular 100 Muntjewerff JW, Kahn RS, Blom HJ, den Heijer M. Homocysteine, translocation of dietary folate and its association with blood pressure in methylenetetrahydrofolate reductase and risk of schizophrenia: a an elderly population. Life Sci 2006; 79: 957–966. meta-analysis. Mol Psychiatry 2006; 11: 143–149. 81 Zhang G, Zhang QY, Miao XP, Lin DX, Lu YY. Polymorphisms and 101 Lewis SJ, Zammit S, Gunnell D, Smith GD. A meta-analysis of the mutations of the folate receptor-alpha gene and risk of gastric cancer MTHFR C677T polymorphism and schizophrenia risk. Am J Med Genet in a Chinese population. Int J Mol Med 2005; 15: 627–632. B Neuropsychiatr Genet 2005; 135: 2–4.

The Pharmacogenomics Journal One-carbon folate pathway genetics DF Carr et al 305

102 Bertram L, McQueen MB, Mullin K, Blacker D, Tanzi RE. Systematic 122 Ott K, Vogelsang H, Marton N, Becker K, Lordick F, Kobl M et al. The meta-analyses of Alzheimer disease genetic association studies: The thymidylate synthase tandem repeat promoter polymorphism: a AlzGene database. Nat Genet 2007; 39: 17–23. predictor for tumor-related survival in neoadjuvant treated locally 103 Pereira TV, Rudnicki M, Pereira AC, Pombo-de-Oliveira MS, Franco RF. advanced gastric cancer. Int J Cancer 2006; 119: 2885–2894. 5,10-Methylenetetrahydrofolate reductase polymorphisms and acute 123 Dotor E, Cuatrecases M, Martinez-Iniesta M, Navarro M, Vilardell F, lymphoblastic leukemia risk: a meta-analysis. Cancer Epidemiol Bio- Guino E et al. Tumor thymidylate synthase 1494del6 genotype as a markers Prev 2006; 15: 1956–1963. prognostic factor in colorectal cancer patients receiving fluorouracil- 104 Zintzaras E, Koufakis T, Ziakas PD, Rodopoulou P, Giannouli S, based adjuvant treatment. J Clin Oncol 2006; 24: 1603–1611. Voulgarelis M. A meta-analysis of genotypes and haplotypes of 124 Skibola CF, Forrest MS, Coppede F, Agana L, Hubbard A, methylenetetrahydrofolate reductase gene polymorphisms in acute Smith MT et al. Polymorphisms and haplotypes in folate-metabolizing lymphoblastic leukemia. Eur J Epidemiol 2006; 21: 501–510. genes and risk of non-Hodgkin lymphoma. Blood 2004; 104: 105 Cronin S, Furie KL, Kelly PJ. Dose-related association of MTHFR 677T 2155–2162. allele with risk of ischemic stroke: evidence from a cumulative meta- 125 Krajinovic M, Costea I, Chiasson S. Polymorphism of the thymidylate analysis. Stroke 2005; 36: 1581–1587. synthase gene and outcome of acute lymphoblastic leukaemia. Lancet 106 Kim RJ, Becker RC. Association between factor V Leiden, prothrombin 2002; 359: 1033–1034. G20210A, and methylenetetrahydrofolate reductase C677T mutations 126 Krajinovic M, Costea I, Primeau M, Dulucq S, Moghrabi A. Combining and events of the arterial circulatory system: a meta-analysis of several polymorphisms of thymidylate synthase gene for pharmaco- published studies. Am Heart J 2003; 146: 948–957. genetic analysis. Pharmacogenomics J 2005; 5: 374–380. 107 Haywood S, Liesner R, Pindora S, Ganesan V. Thrombophilia and first 127 Weisman MH, Furst DE, Park GS, Kremer JM, Smith KM, Wallace DJ arterial ischaemic stroke: a systematic review. Arch Dis Child 2005; 90: et al. Risk genotypes in folate-dependent enzymes and their associa- 402–405. tion with methotrexate-related side effects in rheumatoid arthritis. 108 Lewis SJ, Ebrahim S, Davey Smith G. Meta-analysis of MTHFR 677C- Arthritis Rheum 2006; 54: 607–612. 4T polymorphism and coronary heart disease: Does totality of 128 Justenhoven C, Hamann U, Pierl CB, Rabstein S, Pesch B, Harth V et al. evidence support causal role for homocysteine and preventive One-carbon metabolism and breast cancer risk: No association of potential of folate? BMJ 2005; 331: 1053. MTHFR, MTR, and TYMS polymorphisms in the GENICA study 109 Klerk M, Verhoef P, Clarke R, Blom HJ, Kok FJ, Schouten EG. MTHFR from Germany. Cancer Epidemiol Biomarkers Prev 2005; 14: 677C–4T polymorphism and risk of coronary heart disease: a meta- 3015–3018. analysis. JAMA 2002; 288: 2023–2031. 129 Zhai X, Gao J, Hu Z, Tang J, Qin J, Wang S et al. Polymorphisms in 110 Kluijtmans LA, Kastelein JJ, Lindemans J, Boers GH, Heil SG, thymidylate synthase gene and susceptibility to breast cancer in a Bruschke AV et al. Thermolabile methylenetetrahydrofolate Chinese population: a case–control analysis. BMC Cancer 2006; 6: 138. reductase in coronary artery disease. Circulation 1997; 96: 130 Hishida A, Matsuo K, Hamajima N, Ito H, Ogura M, Kagami Y et al. 2573–2577. Associations between polymorphisms in the thymidylate synthase and 111 Wu AH, Tsongalis GJ. Correlation of polymorphisms to coagulation serine hydroxymethyltransferase genes and susceptibility to malignant and biochemical risk factors for cardiovascular diseases. Am J Cardiol lymphoma. Haematologica 2003; 88: 159–166. 2001; 87: 1361–1366. 131 Ramensky V, Bork P, Sunyaev S. Human non-synonymous SNPs: server 112 Zintzaras E, Chatzoulis DZ, Karabatsas CH, Stefanidis I. The relation- and survey. Nucleic Acids Res 2002; 30: 3894–3900. ship between C677T methylenetetrahydrofolate reductase gene 132 Ng PC, Henikoff S. Predicting deleterious amino acid substitutions. polymorphism and retinopathy in type 2 diabetes: a meta-analysis. Genome Res 2001; 11: 863–874. J Hum Genet 2005; 50: 267–275. 133 Ferrer-Costa C, Gelpi JL, Zamakola L, Parraga I, de la Cruz X, 113 Cahill MT, Stinnett SS, Fekrat S. Meta-analysis of plasma homocys- Orozco M. PMUT: a web-based tool for the annotation of pathological teine, serum folate, serum vitamin B(12), and thermolabile MTHFR mutations on proteins. Bioinformatics 2005; 21: 3176–3178. genotype as risk factors for retinal vascular occlusive disease. Am 134 Yue P, Melamud E, Moult J. SNPs3D: candidate gene and SNP J Ophthalmol 2003; 136: 1136–1150. selection for association studies. BMC Bioinformatics 2006; 7: 166. 114 Janssen MC, den Heijer M, Cruysberg JR, Wollersheim H, Bredie SJ. 135 Capriotti E, Fariselli P, Casadio R. I-Mutant2.0: predicting stability Retinal vein occlusion: a form of venous thrombosis or a complication changes upon mutation from the protein sequence or structure. of atherosclerosis? A meta-analysis of thrombophilic factors. Thromb Nucleic Acids Res 2005; 33: W306–W310. Haemost 2005; 93: 1021–1026. 136 Karchin R, Diekhans M, Kelly L, Thomas DJ, Pieper U, Eswar N et al. 115 Den Heijer M, Lewington S, Clarke R. Homocysteine, MTHFR and risk LS-SNP: large-scale annotation of coding non-synonymous SNPs of venous thrombosis: a meta-analysis of published epidemiological based on multiple information sources. Bioinformatics 2005; 21: studies. J Thromb Haemost 2005; 3: 292–299. 2814–2820. 116 Ray JG, Shmorgun D, Chan WS. Common C677T polymorphism of 137 Dayal S, Bottiglieri T, Arning E, Maeda N, Malinow MR, Sigmund CD the methylenetetrahydrofolate reductase gene and the risk of venous et al. Endothelial dysfunction and elevation of S-adenosylhomocys- thromboembolism: meta-analysis of 31 studies. Pathophysiol Haemost teine in cystathionine beta-synthase-deficient mice. Circ Res 2001; 88: Thromb 2002; 32: 51–58. 1203–1209. 117 Kosmas IP, Tatsioni A, Ioannidis JP. Association of C677T polymorph- 138 Lentz SR, Erger RA, Dayal S, Maeda N, Malinow MR, Heistad DD et al. ism in the methylenetetrahydrofolate reductase gene with hyperten- Folate dependence of hyperhomocysteinemia and vascular dysfunc- sion in pregnancy and pre-eclampsia: a meta-analysis. J Hypertens tion in cystathionine beta-synthase-deficient mice. Am J Physiol Heart 2004; 22: 1655–1662. Circ Physiol 2000; 279: H970–H975. 118 Lin J, August P. Genetic thrombophilias and preeclampsia: a meta- 139 Robert K, Maurin N, Ledru A, Delabar J, Janel N, Vitvitsky V et al. analysis. Obstet Gynecol 2005; 105: 182–192. Hyperkeratosis in cystathionine beta synthase-deficient mice: an 119 van der Put NM, Eskes TK, Blom HJ. Is the common 677C–4T animal model of hyperhomocysteinemia. Anat Rec A Discov Mol Cell mutation in the methylenetetrahydrofolate reductase gene a risk Evol Biol 2004; 280: 1072–1076. factor for neural tube defects? A meta-analysis. QJM 1997; 90: 140 Vitvitsky V, Dayal S, Stabler S, Zhou Y, Wang H, Lentz SR et al. 111–115. Perturbations in homocysteine-linked redox homeostasis in a murine 120 Ren A, Wang J. Methylenetetrahydrofolate reductase C677T poly- model for hyperhomocysteinemia. Am J Physiol Regul Integr Comp morphism and the risk of unexplained recurrent pregnancy loss: a Physiol 2004; 287: R39–R46. meta-analysis. Fertil Steril 2006; 86: 1716–1722. 141 Kelly TL, Neaga OR, Schwahn BC, Rozen R, Trasler JM. Infertility in 121 Nelen WL, Blom HJ, Steegers EA, den Heijer M, Eskes TK. Hyperho- 5,10-methylenetetrahydrofolate reductase (MTHFR)-deficient male mocysteinemia and recurrent early pregnancy loss: a meta-analysis. mice is partially alleviated by lifetime dietary betaine supplementation. Fertil Steril 2000; 74: 1196–1199. Biol Reprod 2005; 72: 667–677.

The Pharmacogenomics Journal