Analysis of select folate pathway genes,PAX3, and humanT in a midwestern neural tube defect population

Analysis of Select Folate Pathway Genes,PAX3, and Human T in a Midwestern NeuralTube Defect PopulationDIMITRI TREMBATH,1 ANDREA L. SHERBONDY,1 DON C. VANDYKE,1 GARY M. SHAW,2KAREN TODOROFF,2 EDWARD J. LAMMER,3 RICHARD H. FINNELL,4 STEPHEN MARKER,5GARY LERNER,6 AND JEFFREY C. MURRAY1,7*1Department of Pediatrics, University of Iowa, Iowa City, Iowa 52242–10832March of Dimes Birth Defects Foundation, California Birth Defects Monitoring Program, Emeryville, California 946083Division of Medical Genetics, Children’s Hospital of Oakland, Oakland, California 94609–18094Department of Veterinary Anatomy and Public Health, College of Veterinary Medicine, Texas A&M University,College Station, Texas 77843–44585Minneapolis Spina Bifida Clinic, Park Nicollet Medical Center, Minneapolis, Minnesota 554046Children’s Hospital, Omaha, Nebraska 681147Department of Biological Sciences, University of Iowa, Iowa City, Iowa 52242–1324

ABSTRACT Neural tube defects (NTDs) are acommon birth defect, seen in approximately 1/1,000births in the United States. NTDs are considered acomplex trait where several genes, interacting withenvironmental factors, create the phenotype. Using aMidwestern NTD population consisting of probands,parents, and siblings from Iowa, Minnesota, and Ne-braska, we analyzed a range of candidate genes,including 5,10–methylenetetrahydrofolate reductase(MTHFR), folate receptors-a (FOLR1; hereafter abbrevi-ated ‘‘FR-a’’) and -b (FOLR2; hereafter, ‘‘FR-b’’), methio-nine synthase (hereinafter, ‘‘MS’’), T, the human homo-log of the murine Brachyury gene, and the paired-boxhomeotic gene 3 (PAX3), for association with NTDs. Wewere unable to demonstrate an association using apreviously described Ala=Val mutation in MTHFR andthe majority of our NTD populations. However, wediscovered a silent polymorphism in exon 6 of MTHFRwhich conserved a serine residue and which showedsignificant association with NTDs in our Iowa popula-tion. Analysis of exon 7 of MTHFR then demonstratedan Ala=Glu mutation which was signficantly associatedwith our Iowa NTD population; however, we could notreplicate this result either in a combined Minnesota/Nebraska or in a California NTD population. Usingpolymorphic markers for MS, FR-b, T, and PAX3, wewere unable to demonstrate linkage disequilibrium withour NTD populations. A mutation search of FR-a re-vealed one proband with a de novo silent mutation ofthe stop codon. This work provides a new panel ofgenetic variants for studies of folate metabolism andsupports, in some NTD populations, an associationbetween MTHFR and NTDs. Teratology 59:331–341,1999. r 1999 Wiley-Liss, Inc.

Neural tube defects (NTDs) are a common birthdefect, occurring in approximately 1/1,000 births in the

United States (Edmonds and James, ’90) and demon-strating racial variation, ranging from 2.2/10,000 forAmerican Indians to 5.9/10,000 for Hispanics (Lary andEdmonds, ’96). The etiology of NTDs is presumed to becomplex, with several genes, interacting with environ-mental factors, likely responsible (Seller, ’94). Manystudies, including the Medical Research Council’s Vita-min Study Research Group (’91), have demonstratedthe protective value of increased dietary folate intake(to levels of 4 mg/day; Mulinare et al., ’88; Smithells etal., ’89; Milunsky et al., ’91; Czeizel and Dudas, ’92).Folic acid supplementation decreases the recurrencerisk of NTDs by approximately 70% among women whohave already had one affected pregnancy and shows asimilar effect in women who have not had an affectedpregnancy (MRC Vitamin Study Research Group, ’91).

Although folate’s protective effect is recognized, themechanism by which some people develop low folatelevels (presumably predisposing them or their offspringto NTDs and other potential birth defects, such as cleftlip with or without cleft palate) is still a matter ofdebate. Much of the recent work has centered onenzymes within the folate pathway, such as 5,10 methy-lenetetrahydrofolate reductase (MTHFR), which con-verts 5,10–methylenetetrahydrofolate back to 5-methyl-tetrahydrofolate. Van der Put et al. (’95) and Ou et al.(’96) demonstrated the association of a C=T mutation(C677T), converting an alanine to a valine residue, inMTHFR with NTDs. This mutation is associated with

Grant sponsor: U.S. National Institutes of Health; Grant number:DE08559; Grant sponsor: U.S. Centers for Disease Control andPrevention; Grant number: CCU7132238.

*Correspondence to: Dr. Jeffrey Murray, Department of Pediatrics,University of Iowa, 200 Hawkins Drive, W229–1 GH, Iowa City, IA52242–1083. E-mail: [email protected]

Received 19 June 1998; Accepted 4 November 1998

TERATOLOGY 59:331–341 (1999)

r 1999 WILEY-LISS, INC.

decreased MTHFR activity, low plasma folate, and highplasma homocysteine and red-cell folate concentrations(van der Put et al., ’95).

Other potential candidates in the folate pathway arethe folate receptors, responsible for binding folate andbringing it into the cell. Three forms of the folatereceptor (FR) exist: a, b, and g, with a considered theprimary folate transporter of the three. Mutations inFR-a may reduce folate levels and make an individualfolate-deficient, a deficiency overcome by maternal fo-late supplementation.

The cDNA for methionine synthase (MS) has recentlybeen cloned by three different groups (Leclerc et al., ’96;Li et al., ’96; Chen et al., ’97). Methionine synthase isresponsible for converting intracellular folate and homo-cysteine to tetrahydrofolate and methionine, respec-tively. Tetrahydrofolate is a crucial ingredient in thebiosynthesis of DNA and RNA, while methionine isimportant in numerous methylation reactions. At leastone study has suggested that defects in methioninesynthase may be etiologic for NTDs (Mills et al., ’95).

In addition to the epidemiologic evidence surround-ing folate, various animal models have proposed othercandidate genes for NTDs, based either on phenotypesfrom knockout experiments or on the gene’s expressionpattern. An example of the former is the paired-boxhomeotic gene 3 (PAX3), which, when knocked out inmice, produces a severe NTD phenotype. Animals homo-zygous for mutations in PAX3 (the Splotch mouse) havea high rate of both anterior and posterior NTDs (Ep-stein et al., ’91). Hol et al. (’95) discovered a frameshiftmutation in exon 5 of PAX3 in a girl with lumbosacralmeningomyelocele and mild signs of Waardenburg syn-drome type I. Sequencing demonstrated a 5-base-pair(bp) deletion in exon 5, approximately 55 bp upstreamof the homeodomain, causing a shift in the normalreading frame and creating a premature stop codon.

An example of a candidate gene suggested by expres-sion studies in animal models is T, the human homologof the murine Brachyury gene (Edwards et al., ’96). Tgene products are vital for the formation of posteriormesoderm and for axial development in all vertebrates(Edwards et al., ’96). Using a polymorphism in intron 7of T, Morrison et al. (’96) demonstrated an associationbetween transmission of the polymorphic allele andfamilial cases of spina bifida.

To investigate these candidate genes, a traditionallinkage approach is difficult: NTDs rarely occur in largekindreds; obtaining sufficient numbers of families withmultiple affecteds for sib-pair approaches is compli-cated by low recurrence rates; and, finally, the multifac-torial nature of a complex disease precludes identifica-tion of a single causative locus. Instead, one can performmutation searches on putative candidate genes, usingtechniques such as single-strand conformational poly-morphism (SSCP) analysis or direct sequencing toidentify mutations that could be etiologic for an NTD.

Another alternative is to perform an associationstudy using polymorphic markers either close to (i.e.,

preferably within 500 kilobase-pairs (kb)) or within thesequence of a candidate gene. In association studies,one often attempts to demonstrate linkage disequilib-rium, the preferential segregation of a particular markerwith a disease phenotype on a population-wide basis(Lander and Schork, ’94, Risch and Merikangas, ’96), assuch an association may point to the existence of anearby etiologic mutation.

To determine if preferential allele transmission isstatistically significant, case-control allele frequenciesare compared, utilizing a 2 3 2 contingency table(Lander and Schork, ’94). Another approach, and oneoften used as an internal control, is to apply thetransmission disequilibrium test (TDT), developed bySpielman et al. (’93), which, using only the genotypes ofheterozygous parents with an affected child, deter-mines if there is preferential transmission of a particu-lar allele with the disease phenotype.

Using these methodologies and a Midwestern NTDpopulation consisting of cases from Iowa, Minnesota,and Nebraska, our laboratory performed associationstudies with the C677T mutation in MTHFR, a novelpolymorphism in exon 6, and a novel mutation in exon 7which converts an alanine residue to glutamic acid. Wealso performed association studies using a polymorphicmarker for FR-b, two polymorphic markers in methio-nine synthase, a polymorphism in exon 6 of PAX3, andthe intron-7 polymorphism of human T. Finally, weinvestigated the coding regions of FR-a and exons 5 and6 of PAX3 for mutations possibly etiologic in NTDs.

MATERIALS AND METHODS

DNA samples

DNA was gathered from individuals with NTDs andtheir immediate family members from three Midwest-ern U.S. clinics: 128 families from the University ofIowa Hospital School Myelodysplasia Clinic in IowaCity, Iowa; 35 families from the Park Nicollet Minne-sota Spina Bifida Clinic in Minneapolis, Minnesota;and nine families from the Children’s Hospital Develop-mental Clinic in Omaha, Nebraska. Case data for theIowa NTD population are presented in Table 1. Popula-tion control DNA came from two sources: 1) unaffectedindividual Iowa newborns or 2) Iowa families with anunaffected child (father/child/mother). Nearly 700 Cali-fornia spina bifida and control samples were availablefrom a California population. A description of thesedata can be found elsewhere (Shaw et al., 1998).

After signed informed consent was obtained, cheekswabs were performed. On some probands, blood spotswere also obtained if the patient was having otherlaboratory tests performed. DNA was extracted fromthe cheek swab/blood spot via standard alkaline lysismethods and then analyzed on an ultraviolet spectropho-tometer for concentration.

Mutation detection enhancement analysis

DNA fragments were amplified from the genomicDNA via the polymerase chain reaction (PCR). Amplifi-

332 D. TREMBATH ET AL.

cation was carried out in a total volume of 15 µlcontaining 50 ng of genomic DNA, 200 mM dNTPs, and0.30 mmol/ml of each primer in Boehringer Mannheim(Indianapolis, IN) PCR buffer with 0.5 units Taq DNApolymerase. All reactions were at 94°C for 30 sec, theappropriate annealing temperature for 30 sec, and72°C for 30 sec, for 35 cycles. Primer sequences were allpreviously published for MTHFR (Orr and Kamen, ’95;van der Put et al., ’95; Ou et al., ’96), the MS cDNA (Li etal., ’96), PAX3 (Hol et al., ’95), and T (Edwards et al.,’96; Morrison et al., ’96). Newly developed primersequences are listed in Table 2.

The amplified DNA was loaded onto 0.5 3 mutationdetection enhancement (MDE; FMC, Rockland, ME)gels containing 2.5% glycerol (for FR-a, MTHFR, MS,PAX3, and human T) or onto 6% denaturing gels (for

FR-b) and electrophoresed for 6–8 hr at 20 W and 5°C(for FR-a, MTHFR, MS, PAX3, and human T) or for 4 hrat room temperature (for FR-b). Following electrophore-sis, the gels were silver-stained and scored.

Automated sequencing of normaland mutant alleles

To analyze shifted bands that might be allelic vari-ants, the bands were cut out of the MDE/sequencinggels, and the DNA was eluted into 50 ml H2O at 37°C for1 hr, and then reamplified under the conditions de-scribed above. The reamplified DNA fragments werepurified by electrophoresis through a 2% agarose gel(for 1 hr at 100 V, 50 mA) and then sliced out of the gel.This material was purified using a Qiagen (Valencia,CA) gel purification kit and sequenced on an AppliedBiosystems 373 DNA Sequencer (Foster City, CA).

Statistical analysis

Case-control comparisons were performed using theDISLAMB program developed by Terwilliger (’95). TDTanalysis was performed using the equations publishedby Spielman et al. (’93). Association studies reportedbelow are given as comparisons between the Iowa NTDcases and population controls, which consisted of Cauca-sian children from Iowa without NTDs or other congen-tial abnormalities (e.g., cleft lip/palate). Unless other-wise noted, addition of Nebraska and Minnesota casesdid not change the significance of results. When posi-tive associations were found using the Iowa NTDcohort, Nebraska and Minnesota cases were combinedand analyzed for comparison. The California popula-tion was used only in analysis of the exon-7 mutant inMTHFR. We controlled for ethnicity in all populations,using only Caucasian probands, and in the Iowa popula-tion for the type and level of lesion, using only meningo-myelocele cases which corresponded to Van Allen clo-sure sites 1 or 5 as determined from clinical data andthe boundary criteria defined by Van Allen (’93). Sepa-rate analyses of cases excluded on the basis of type/level(e.g., being nonmeningomyelocele) were not significantunless otherwise noted. TDT was not performed in allstudies due to rarity of allelic variants (e.g., allele 2 ofPAX3) and thus small numbers of heterozygous parents.

RESULTS

Analysis of 5,10-methylenetetrahydrofolatereductase (MTHFR)

We analyzed our Iowa (IA), Nebraska (NE), andMinnesota (MN) populations for association with theC677T mutation by designing primers that flanked themutation site and by using the MDE detection system.Comparison of meningomyelocele cases (Van Allen clo-sure sites 1 and 5; Van Allen et al., ’93) and controlfrequencies was not statistically significant (see Table3). Using TDT, we were unable to demonstrate preferen-tial transmission of either allele to affected children(Table 4).

TABLE 1. Case data for the Iowa NTD population

Characteristics

Gender Male, 57 (47%) Female, 65 (53%)

EthnicityCaucasian,117 (96%)

Hispanic,3 (2%)

Black,1 (1%)

Caucasian/black,1 (1%)

Type ofdefect

Meningomyelo-cele

99 (81%) One with myelo-schisis, one withdiastematomyelia

Lipomeningomy-elocele

6 One withhydromyelia

Interspinal ipoma 1Diplomyelia 1Sacral hypoplasia 1Myelocystocele 1Myeloschisis 2Sacral agenesis 1Intramedullary

lipoma2

Meningocele 3 One withintradural lipoma

Intraspinal der-moid

1

Spina bifidaocculta

1

Diastematomyelia 1Meningoencepha-

locele1 With lumbar

meningoceleTotal 122

Motorlevel

Sensorylevel

Thoracic (-T12) 25 (20.5%) 21 (17.2%)High lumbar (L1–2) 9 (7.4%) 11 (9.0%)Mid lumbar (L3–4) 29 (23.8%) 25 (20.5%)Lumbosacral (L5–S2) 44 (36.1%) 42 (34.4%)Low sacral (S3–5) 12 (9.8%) 21 (17.2%)Asymmetric 3 (2.5%) 2 (1.6%) Lumbosacral

or lowsacral

Family historyof NTDs Yes, 46 (38%) No, 67 (55%)

Unknown,9 (7%)

Other disorders Mental retardation 31 (25%)Seizure disorders 6 (13%)Hydrocephalus 91 (75%)Neurogenic bowel and bladder 99 (81%)Neurogenic bladder alone 11 (9%)

GENETIC ANALYSIS OF NTD CANDIDATE GENES 333

However, when we compared allele frequencies be-tween IA NTD cases who were nonmeningomyeloceles(including lipomeningomyelocele, intradural lipoma,and sacral hypoplasia/agenesis) and population con-trols, we were able to demonstrate a statistically signifi-cant difference. The mutant allele (T) was more fre-quent in this subset of our NTD population than incontrols (see Table 5).

We then screened exons 1, 2, 3, 5, and 6 of MTHFR fornovel polymorphisms/mutations that could be used todetect an association. SSCP analysis revealed a silentpolymorphism, T=C, at bp 1059 in exon 6, whichconserved a serine residue (Figs. 1, 2; base-pair number-ing reflects GenBank entry UO9806 for the MTHFRcDNA). Comparison of allele frequencies between IANTD cases and population controls gave a P value of

0.001, indicating a significant association with the Callele (allele 2 in Table 6). This association was alsosignificant when allele frequencies were compared be-tween the nonmeningomyelocele subset and populationcontrols (Table 7).

Probands from the Iowa population were also haplo-typed for the C677T mutation and the T1059C polymor-phism. Haplotype frequencies were not significantlydifferent when compared to population controls (Table8), nor was there a significant difference in transmis-sion of one particular haplotype to affected children(Table 9) using the affected family-based controls(AFBAC) test.

Given the lack of association between the C677Tmutation and our IA NTD population, but linkage

TABLE 2. Novel PCR primer sequences for distinguishing candidate-gene polymorphisms*

Gene Marker location

Primer sequence (58 = 38)

Forward Reverse

MTHFR Exon 7 GGACTACTACCTCTTCTACCT ATGAACCAGGGTCCCCACTCFR-a Exon 3 AAGAATTCGCCACTGACCACAGCTCTTTCT AAAGAATTCCAGTCCCTCACCACCCCAGATCAC

Exon 4 AAAGAATTCGAGTCCTCTGTCTTCCCCCAT AAAGAATTCTCGTCCGCTAGGTCTTGTACCTGCExon 5 AAAGAATTCACCTAGCGGACGAGCTGAGCTT AAAGAATTCTGGGCCCGAACTATCTTGAGGTCTExon 6 GGGCTGGCAGACCTCAAGATAGTTC TATTCAAAGTGGCTGTCAGAGGCCC

FR-b GTGGGGAGACTTGGAGAGTT GAGGTGGTGGGCACCTGTAMS 38 UTR GAGACCACATGGGGCGTAC TATGTGCTCCTGTCCCTTCC

*Sequences of the other markers typed in this analysis have been published, for MTHFR, by Orr and Kamen (’95), by van derPut et al. [’95], and by Ou et al. (’96); for the MS cDNA, by Li et al. (’96); for PAX3, by Hol et al. (’95); and, for T, by Edwards et al.(’96) and by Morrison et al. (’96).

TABLE 3. Comparison of MTHFR C677T allelesin Iowa meningomyelocele NTD (Van Allen closure

sites 1 and 5) cases vs. population controls*

Allele 1 (T) Allele 2 (C)

Cases 15 69Controls 61 197

*x2 5 1.23, P 5 0.27, df 5 1, OR 5 0.70, 95% CI 5 (0.37, 1.32).

TABLE 4. Transmission disequilibrium test forMTHFR C677T mutation in Iowa meningomyeloceleNTD (Van Allen closure sites 1 and 5) cases’ families*

Transmitted

Untransmitted

Allele 1 Allele 2

Allele 1 0 2Allele 2 4 23

*(x2)td 5 .11, P 5 0.75, df 5 1.

TABLE 5. Comparison of MTHFR C677T allelesin Iowa nonmeningomyelocele NTD (Van Allen closure

sites 1 and 5) cases vs. population controls*



*x2 5 4.9, P 5 0.03, df 5 1, OR 5 2.42, 95% CI 5 (1.09, 5.4).Fig. 1. MDE gel of an Iowa NTD family, demonstrating the silentpolymorphism in exon 6 (T1059C) of MTHFR. Lane assignments andgenotypes: lane 1, father (2,2; C,C); lane 2, proband (1,2; T,C); lane 3,mother (1,2; T,C).


disequilibrium with the T1059C polymorphism, weundertook an analysis of the other MTHFR exons.SSCP analysis revealed a mutation in exon 7, a C=Achange at bp 1289 of the MTHFR cDNA sequencewhich, when the MTHFR cDNA was fully translated,converted an alanine residue to glutamic acid (Figs. 3,

4). The glutamic acid residue showed a strong associa-tion with our Iowa NTD population (x2 5 10.6, P 50.001; Table 10). The transmission disequilibrium test(TDT) supported this association, showing preferentialtransmission of the A allele (Glu) to affected children(x2 5 7.4, P , .01; Table 11), as did the AFBAC test (x2 55.6, P 5 0.02; Table 12). The double-mutant haplotypefor MTHFR was slightly more common in cases (19.5%)than in controls (16.5%), while the double-wild-typehaplotype was present much more frequently in con-trols (45%) vs. cases (15%). Overall, there was a signifi-cant difference when all haplotype frequencies werecompared between cases and controls (x2 5 12.9, P 50.002; Table 13).

Attempts to replicate this association in a combinedMinnesota/Nebraska cohort were unsuccessful. Com-parison of case frequencies to those of Iowa controls didnot demonstrate an association (x2 5 0.00078, P 5 .98,df 5 1, odds ratio (OR) 5 1.01, 95% confidence interval(CI) 5 (0.58, 1.76)), nor was TDT significant (x2 5 0.25,P . 0.05). We also genotyped a large California NTDpopulation, but again were unable to replicate theassociation, even when accounting for ethnicity, vita-min use, or level of lesion (data not shown).

Analysis of folate receptor-alpha (FR-a)

We performed a mutation search in the coding regionof FR-a (exons 3–6) in the IA, MN, and NE populationsusing the MDE system. The proband of one IA familydemonstrated a significant shift on MDE (Fig. 5). Theproband demonstrating the shift was a male Caucasianwith meningomyelocele, motor level S1, sensory levelS1. In this family, there was a positive family history of

Fig. 2. Sequence of (a) wild-type (T) and (b) polymorphic (C) allelesfor the MTHFR T1059C polymorphism in exon 6, illustrating the T=Cchange.

TABLE 6. Comparison of MTHFR exon-6 alleles(T1059C) in Iowa meningomyelocele NTD (Van Allenclosure sites 1 and 5) cases vs. population controls*



*x2 5 10.57, P 5 0.001, df 5 1, OR 5 3.71, 95% CI 5 (1.60,8.63).

TABLE 7. Comparison of MTHFR exon-6 alleles(T1059C) in Iowa nonmeningomyelocele NTD

(Van Allen closure sites 1 and 5) casesvs. population controls*



*x2 5 5.18, P 5 0.023, df 5 1, Fisher’s P (two-tailed) 5 0.027.

TABLE 8. Comparison of MTHFR (C677TAllele)-(T1059C Allele) Haplotypes in Iowa

Meningomyelocele (Van Allen Closure Sites 1 and 5)NTD Cases’ Families versus Population Controls*

1-1,2-1** 1-2 2-2

Cases 4 13 55Controls 16 27 93

*x2 5 2.04, P 5 0.36, df 5 2.**1-1 and 2-1 haplotypes are combined to preclude type Ierror.

TABLE 9. AFBAC analysis of MTHFR (C677TAllele)-(T1059C Allele) haplotype transmission

in Iowa meningomyelocele NTD (Van Allen closuresites 1 and 5) cases’ families*

Nontransmitted Transmitted

1-1,1-2,2-1** 13 72-2 27 33

*x2 5 2.4, P 5 0.12, df 5 1.**1-1, 1-2, and 2-1 haplotypes are combined to preclude type Ierror.

Fig. 3. MDE gel of a family, demonstrating the C1289A mutation(Ala=Glu) in MTHFR exon 7. Lane 1, father (2,2; A,A); lane 2,mother (1,1; C,C); lanes 3–5, children (1,2; A,C).


spina bifida, with a cousin similarly affected. Thisshifted band was seen only in the proband and not ineither parent or in any of 3 siblings. Paternity wasconfirmed by using a genome-wide set of CooperativeHuman Linkage Center marker (CHLC, Iowa City, IA)markers on both of the parents and on the proband todemonstrate inheritance of alleles from both parents.

Sequence analysis of the shifted band revealed aTGA=TAA change within the stop codon of exon 6 (Fig.6). This silent mutation was confirmed by sequencing

the shifted band in both the forward and reversedirections. Given the confirmation of paternity, it ap-pears that this is a de novo change within the proband’sstop codon.

We analyzed 316 control individuals via MDE vis-a-vis the proband with the TAA change. None demon-strated the TGA=TAA change within the stop codon.However, one child did demonstrate a Ser=Asn changewithin exon 6.

Analysis of folate receptor-beta (FR-b)

The sequence for FR-b was downloaded from Gen-Bank, and, via use of Primer Server, an Internet toolfrom CHLC for finding repeat sequences (http://www.chlc.org/), a pentanucleotide repeat (TGTAT)

Fig. 4. Sequence showing C1289A mutation in exon 7 of MTHFR,creating the alanine (a) to glutamic acid (b) mutation.

TABLE 10. Comparison of MTHFR exon-7 alleles(C1289A) in Iowa meningomyelocele NTD (Van Allenclosure sites 1 and 5) cases vs. population controls*

Allele 1 (C) Allele 2 (A)


*x2 5 10.6, P 5 0.001, df 5 1, OR 5 2.39, 95% CI 5 (1.4, 4.08).

TABLE 11. Transmission disequilibrium testfor MTHFR exon-7 (C1289A) mutation in Iowa

meningomyelocele NTD (Van Allen closure sites 1and 5) cases’ families*

Transmitted

Untransmitted

Allele 1 (C) Allele 2 (A)

Allele 1 (C) 1 1Allele 2 (A) 10 9

*(x2)td 5 7.4, P , .01.

TABLE 12. AFBAC analysis of MTHFR exon-7mutation (C1289A) in Iowa meningomyelocele NTD

(Van Allen closure sites 1 and 5) cases’families*

Nontransmitted Transmitted

Allele 1 (C) 13 4Allele 2 (A) 43 52

*x2 5 5.61, P 5 0.02, df 5 1.

TABLE 13. Comparison of MTHFR C677T-C1289Ahaplotypes in Iowa meningomyelocele

(Van Allen closure sites 1 and 5) NTD cases’families vs. population controls*

1,2 2,1 2,2

Cases 18 14 60Controls 19 42 51

*x2 5 12.9, P 5 0.002, df 5 2.

Fig. 5. MDE gel, demonstrating Iowa NTD proband with de novopolymorphism in exon 6 of folate receptor-a gene’s stop codon (indi-cated by arrow). Lane 1, father; lane 2, mother; lane 3, proband;lanes 4–6, unaffected siblings.


within the first intron was discovered and primers weredesigned surrounding the repeat. Thirty-seven individu-als from the IA NTD population (including 12 probands)demonstrated at least one polymorphic allele on dena-turing polyacrylamide gels (Fig. 7). Automated sequenc-ing revealed an expansion of 15 bp, or three completeTGTAT repeats, in the polymorphic fragments (Fig. 8).Using this marker, however, case-control analysis wasnot statistically significant (x2 5 0.54, P 5 0.46, df 5 1,OR 5 0.76, 95% CI 5 (0.36, 1.59)).

Analysis of methionine synthase (MS)

Three different groups have cloned the cDNA formethionine synthase (Leclerc et al., ’96; Li et al., ’96;Chen et al., ’97). Utilizing this information, we took twodifferent approaches to screening methionine synthasefor polymorphisms/mutations. Two of the groups pub-lished the untranslated-region sequence of the gene,and our first approach was to compare these sequencesutilizing Sequencher (Gene Codes Corporation, AnnArbor, MI), a sequence analysis program. There were at

least nine differences between the two published se-quences, including a poly-A tract beginning at bp 6846where one group (Chen et al., ’97) found nine As and theother group (Li et al., ’96) 10. We designed primers toeach of these and screened our IA NTD population forthis putative polymorphism.

MDE analysis revealed a polymorphism (Fig. 9): aGAA insertion at the end of the poly-A tract (Fig. 10).Comparison of allele frequencies between the IowaNTD cases and population controls was not statisticallysignificant (x2 5 0.588, P 5 0.443, df 5 1, OR 5 0.84,95% CI 5 (0.55, 1.30)).

Our second approach was to use the published se-quences of the primers utilized in cloning the cDNA toscreen our NTD populations for polymorphisms/muta-tions. This approach revealed a silent polymorphism atbp 3778 in the cDNA (C=A) which conserved anarginine residue (Figs. 11, 12). Allele frequencies werenot significantly different between our IA NTD popula-tion and population controls (x2 5 1.27, P 5 0.26, df 5 1,OR 5 0.81, 95% CI 5 (0.56, 1.17)).

Analysis of T, a human Brachyury homolog

We investigated the previously reported polymor-phism in T (a T=C change 79 bp downstream from the5’ end of intron 7) in a subset of our IA NTD population.Unlike Morrison et al. (’96), we used sporadic ratherthan familial cases of NTDs. From a small number (13)of informative families (families with at least oneheterozygous parent), we found nearly equal transmis-sion of the two alleles, and TDT was not significant(x2 5 0.07, P . 0.1).

Investigations of PAX3

We analyzed our entire IA NTD population for muta-tions in exons 5 and 6 of PAX3, which encode the

Fig. 6. Sequence of (a) wild-type and (b) polymorphic alleles for thestop-codon variants in folate receptor-a gene, illustrating theTGA=TAA change

Fig. 7. Sequencing gel of Iowa NTD family, demonstrating polymor-phism of TGTAT repeat in folate receptor-b gene (allele 2, wt; allele 1,TGTAT expansion). Lane 1, father (1,2); lane 2, proband (1,2); lane 3,unaffected sibling (1,2); lane 4. mother (2,2).

Fig. 8. Sequence of the (a) wild-type and (b, c) polymorphic alleles forTGTAT repeat in folate receptor-b gene, illustrating expansion by 3complete repeats. b: Polymorphic allele, forward sequence. c: Polymor-phic allele, reverse sequence.


homeodomain. Using MDE analysis, we were unable todemonstrate any polymorphisms in exon 5. We wereable to demonstrate a common polymorphism in exon 6(a C=A change at bp 245 of exon 6, converting athreonine residue to lysine, ACA to AAA) which we usedin a case-control analysis. Neither allele was preferen-tially associated with our Iowa NTD population (x2 50.00009, P 5 0.99, df 5 1, OR 5 1.00, 95% CI 5 (0.35,2.89)).

DISCUSSION

We report here the analysis of the genes for methy-lenetetrahydrofolate reductase (MTHFR), folate recep-tor-a (FR-a), folate receptor-b (FR-b), methionine syn-thase (MS), the human Brachyury homolog (T), andPAX3 in over 100 Midwestern families, all with at leastone child with an NTD. By performing mutationsearches and linkage disequilibrium studies, we at-tempted to identify whether any of the above geneswere significantly involved in the etiology of humanNTDs.

We were unable to demonstrate linkage disequilib-rium with the C677T mutation in MTHFR when com-paring meningomyelocele cases to population controls.

Results of our TDT analysis with cases of meningomye-locele point to nearly equal transmission of the wild-type and mutant alleles from heterozygous parents toaffected children, based on the small numbers availablefor study. However, when we confined our analysis tononmeningomyelocele cases (those with defects eitherin Van Allen closure sites 1 or 5 or at their border), wewere able to demonstrate a statistically significantassociation between the mutant allele (T) and our NTDpopulation.

This analysis, in part, confirms the work of van derPut et al. (’95) and Ou et al. (’96), who demonstrated anassociation between NTDs and the C677T mutation,with Ou et al. (’96) able to demonstrate association witha broad population of NTD cases including spina bifida,anencephaly, and encephalocele. Other groups havereported a higher frequency of the mutant allele intheir control populations and/or have also been unableto demonstrate an association with NTDs (de Franchiset al., ’93; Speer et al., ’96; Mornet et al., ’97). Theseconflicting results may demonstrate the necessity tosegregate NTD populations more specifically for associa-tion studies, on the basis of ethnicity, type of lesion,and/or Van Allen level.

The hypothesis of another etiologic mutation inMTHFR or somewhere nearby on chromosome 1 forNTDs occurring at Van Allen levels 1and 5 is supportedby our finding of an association between a silentpolymorphism in exon 6 (T1059C) and meningomyelo-cele NTDs in our IA population. The C677T mutationand T1059C polymorphism are not transmitted to-gether (i.e., are not associated). Our finding of thisassociation with the T1059C polymorphism led to thediscovery an apparently cotransmitted mutation, aC=A change at bp1289 in exon 7 which converts analanine residue to glutamic acid. The glutamic acidresidue is strongly associated with our Iowa NTDpopulation, a result supported by the TDT. However, wecould not replicate this result in either the combinedMinnesota/Nebraska or California NTD populations.

A possible explanation for the disparity between theIowa data and Minnesota/Nebraska data centers pri-marily on ethnicity. The Iowa result may be the resultof a ‘‘gradual admixture’’ model (Ewens and Spielman,’95), where the Iowa NTD population is reflecting aspecific subgroup (e.g., Czech) in the general Iowapopulation, a subgroup both with a higher incidence ofNTDs and a higher percentage of the C1289A mutation.Here, the two factors (NTDs and the C1289A mutation)are associated due to the shared ethnicity, not becauseof a causal relationship. Comparison of this subgroupwith an appropriate set of controls (Czech, to follow theabove example) would not reveal a significant associa-tion, as both cases and controls would have similarfrequencies of the two exon 7 alleles. Comparison with ageneral Iowa population (assumed, in this model, to bea mix of ethnicities and, hence, of exon 7 allele frequen-cies) may then prove falsely significant merely becausethe NTD subgroup has not had enough time to mix with

Fig. 9. MDE gel, demonstrating GAA polymorphism in 3’ UTR ofmethionine synthase gene (allele 1, GAA insertion; allele 2, wtsequence). Lane 1, father (1,2); lane 2, proband (2,2); lane 3, mother(1,2).


the population at large. The confounder to such anexplanation is TDT, which supports an associationbetween the C1289A mutant and NTDs in our Iowapopulation. According to Ewens and Spielman (’95),TDT remains a valid x2 statistic regardless of popula-tion history.

A variation on the above subgroup ethnicity argu-ment may explain the lack of association with theMinnesota/Nebraska cohort. Both sets of samples werecollected at large urban centers (Minneapolis andOmaha), which may mean the Caucasian populationsrepresent a mixture of different heritages, which de-creases the possibility of finding a particular ‘‘founder’’mutation (such as the C1289A mutant). In addition,information on type and level of lesion were morelimited for these two populations, which did not allowus to stratify the cases as we did for the Iowa NTDcohort. The lack of a separate control population ren-ders comparisons to the Iowa controls suspect, due, inpart, to the ethnicity differences described above as wellas to possible environmental differences.

More troublesome are the California data, where wecould account for ethnicity and had a geographicallymatched control group. Here, the lack of associationmay be due simply to the fact that, in the Californiapopulation, another gene within or outside of the folatepathway is the culprit in NTDs rather than MTHFR. Inaddition, the overall population admixture may bequite different in California.

During this paper’s preparation, van der Put et al.(’98) reported the discovery of the C1289A mutation and

Fig. 10. Sequences of (a, b) wild-type and (c, d) polymorphic alleles for 3’ UTR poly-A tract in methioninesynthase gene, illustrating GAA insertion. a, c: Forward sequences. b, d: Reverse sequences.

Fig. 11. MDE gel demonstrating silent polymorphism (C3778A) incDNA of methionine synthase gene. Lane 1, father (2,2; A,A); lane 2,proband (1,2; A,C); lanes 3–5, unaffected siblings, all (1,2; A,C); lane6, mother (1,2; A,C).

Fig. 12. Sequences of (a) wild-type and (b) polymorphic alleles of thecDNA of methionine synthase, illustrating the C3778A change, whichconserves an arginine residue.


their analysis of its effects in a Dutch NTD population,although they placed the mutation at bp 1298 and referto it as the 1298 (A=C) mutation. They reported thatthe mutation resulted in decreased MTHFR activity,although neither the homozygous nor heterozygousstates were associated with a higher plasma homocyste-ine or a lower plasma folate concentration. Combinedheterozygosity for the 1298 and 677 mutations resultedin significantly reduced MTHFR activity and decreasedplasma folate levels, but this combined heterozygoushaplotype was not significantly associated with NTDpatients compared to controls.

Folate receptor-a (FR-a) is the primary agent forcapturing folate and bringing it into the cell (Orr andKamen, ’95); any mutation which affects the structureof FR-a could potentially reduce folate binding andcellular levels of folate, making an individual folate-deficient and possibly increasing plasma homocysteineconcentration. One child from our Iowa NTD popula-tion demonstrated an apparent de novo silent mutationwithin the stop codon in exon 6 of FR-a. The signifi-cance of this silent mutation is best analyzed in light ofwork by Tate and Brown (’92) and Brown et al. (’93)regarding the influence of termination codon context onthe effectiveness of stop signals. According to the trans-lational termination signal database (TransTerm), thesequence TAA (the trinucleotide present in the pro-band) is a stop codon only 29.9% of the time, while TGA(the sequence present in the parents) is used as a stopcodon 48.0% of the time (Brown et al., ’93). Hence, it ispossible that translation of the folate receptor proteinwithin the proband does not effectively stop, tying upvital translational machinery. In addition to this ‘‘sink’’effect, the folate receptor protein may not be releasedfrom the translational machinery to take its place withcaveolae in the cellular membrane.

Our initial analysis of methionine synthase does notsupport an etiologic role for it in any of our NTDpopulations. Other groups have also been unable todemonstrate an association with polymorphisms in MSand different NTD populations (Morrison et al., ’97; vander Put et al., ’97).

FR-b is unlikely to bind folate at physiologic concen-trations; thus, it is not the best candidate to explain thelow folate levels/high homocysteine levels in NTD cases(Orr and Kamen, ’95). A lack of genetic association,therefore, is not contradictory with putative biochemi-cal mechanisms explaining how cells become folate-deficient.

Our findings with PAX3 are consistent with otherstudies that have failed to demonstrate an associationbetween NTDs and PAX3 (Chatkupt et al., ’95; Speer etal., ’96). Regarding human T, our finding is consistentwith a comment made by Morrison et al. (’96), whodemonstrated T association in familial cases of NTDs,but not in sporadic cases as were examined here.

Findings with MTHFR support the current adviceregarding pre- and perinatal folate supplementation.Several studies have shown that preconceptional mater-

nal folic acid intake can significantly reduce the risk ofNTDs (Mulinare et al., ’88; Smithells et al., ’89; MedicalResearch Council Vitamin Study Research Group, ’91;Milunsky et al., ’91; Czeizel and Dudas, ’92). Otherfolate-pathway candidate genes exist, including thymi-dylate synthase and cystathionine beta-synthase. Ani-mal models such as Cart1 also yield to the efficacy offolate protection against NTDs and raise the possibilitythat the postulated genetic defect may be outside thefolate pathway itself (Zhao et al., ’96). Alternatively, theprimary protective effect of preconceptional folate maybe simply to make up for a dietary deficiency (N. Klein,personal communication). Some researchers have alsoargued that the protective effect seen in a Hungarianstudy may be due to terathanasia (Hook and Czeizel,’97). Whatever its protective mechanism, folate shouldremain a standard supplement for women in their pre-and periconceptual diets.

Future work will include a further mutation search ofMTHFR and analysis of other candidates in the folatepathway. Continued haplotype analysis will also beinformative as to the effect of combinations of the abovepolymorphisms/mutations in different genes in placingindividuals at an increased risk of NTDs.

ACKNOWLEDGMENTS

We thank John Menninger for pointing out the worksof Tate and Brown (’92) regarding translation termina-tion signals, Marilyn Dolezal and other staff membersin Iowa, Nebraska, and Minnesota who gathered clini-cal information, cheek swabs, and blood spots, and thefamilies who agreed to participate in this study. Wethank Nancy Leysens and Bonnie Ludwig for perform-ing the automated sequencing, Molly Wise for technicalhelp, and Nancy Newkirk and Charles Huppmann forhelp with manuscript preparation.

LITERATURE CITEDBrown CM, Dalphin ME, Stockwell PA, Tate WP. 1993. The transla-

tional termination signal database. Nucleic Acids Res 21:3119–3123.

Chatkupt S, Hol FA, Shugart YY, Geurds MPA, Stenroos ES, Koenigs-berger MR, Hamel BCJ, Johnson WG, Mariman ECM. 1995. Ab-sence of linkage between familial neural tube defects and PAX3gene. J Med Genet 32:200–204.

Chen LH, Liu ML, Hwang HY, Chen LS, Korenberg J, Shane B. 1997.Human methionine synthase: cDNA cloning, gene localization, andexpression. J Biol Chem 272:3628–3634.

Czeizel AE, Dudas I. 1992. Prevention of the first occurrence ofneural-tube defects by periconceptional vitamin supplementation.N Engl J Med 327:1832–1835.

de Franchis R, Sebastio g, Mandato C, Andria G, Mastroiacovo P. 1995.Spina bifida, 677T=C mutation, and role of folate [letter]. Lancet346:1703.

Edmonds LD, James LM. 1990. Temporal trends in the prevalence ofcongenital malformations at birth based on the Birth DefectsMonitoring Program, United States, 1979–1987. MMWR 39:19–23.

Edwards YH, Putt W, Lekoape KM, Stott D, Fox M, Hopkinson DA,Sowden J. 1996. The human homolog T of the mouse T (Brachyury)gene: Gene structure, cDNA sequence, and assignment of chromo-some 6q27. Genome Res 6:226–233.

Epstein DJ, Vekemans M, Gros P. 1991. splotch (Sp2H), a mutation


affecting development of the mouse neural tube, shows a deletionwithin the paired homeodomain of Pax-3. Cell 67:767–774.

Ewens WJ, Spielman RS. 1995. The transmission disequilibrium test:History, subdivision, and admixture. Am J Hum Genet 57:455–464.

Hol FA, Hamel BCJ, Geurds MPA, Mullart RA, Barr FG, Macina RA,Mariman ECM. 1995. A frameshift mutation in the gene for PAX3 ina girl with spina bifida and mild signs of Waardenburg Syndrome. JMed Genet 32:52–56.

Hook EB, Czeizel A. 1997. Can terathanasia explain the protectiveeffect of folic-acid supplementation on birth defects? Lancet 350:513–515.

Lander ES, Schork NJ. 1994. Genetic dissection of complex traits.Science 265:2037–2048.

Lary JM, Edmonds LD. 1996. Prevalence of spina bifida at birth—United States, 1983–1990: A comparison of two surveillance sys-tems. MMWR 45:15–26.

Leclerc D, Campeau E, Goyette P, Adjalla CE, Christensen B, Ross M,Eydoux P, Rosenblatt DS, Rozen R, Gravel RA. 1996. Humanmethionine synthase: cDNA cloning and identification of mutationsin patients of the cblG complementation group of folate/cobalamindisorders. Hum Mol Genet 5:1867–1874.

Li YN, Gulati S, Baker PJ, Brody LC, Banderjee R, Kruger WD. 1996.Cloning, mapping, and RNA analysis of the human methioninesynthase gene. Hum Mol Genet 5:1851–1858.

Medical Research Council Vitamin Study Research Group. 1991.Prevention of neural tube defects: Results of the Medical ResearchCouncil Vitamin Study. Lancet 338:131–137.

Mills JL, McPartlin JM, Kirke PN, Lee YJ, Conley MR, Weir DG, ScottJM. 1995. Homocysteine metabolism in pregnancies complicated byneural-tube defects. Lancet 345:149–151.

Milunsky A, Jick H, Jick SS, Bruell CL, MacLaughlin DS, RothmanKJ, Willett W. 1991. Multivitamin/folic acid supplementation inearly pregnancy reduces the prevalence of neural tube defects.JAMA 262:2847–2852.

Mornet E, Muller F, Lenvoise-Furet A, Delezoide A, Col J, Simon-BouyB, Serre J. 1997. Screening of the c677T mutation on the methylene-tetrahydrofolate reductase gene in French patients with neural tubedefects. Hum Genet 100:512–514.

Morrison K, Charalambos P, Attwood J, Hol F, Lynch SA, Sampath A,Hamel B, Burn J, Sowden J, Stott D, Mariman E, Edwards YH.1996. Genetic mapping of the human homologue (T) of the mouse T(Brachyury) and a search for allele association between human Tand spina bifida. Hum Mol Genet 5:669–674.

Morrison K, Edwards YH, Lynch SA, Burn J, Hol F, Mariman E. 1997.Methionine synthase and neural tube defects. J Med Genet 34:958.

Mulinare J, Cordero JF, Erickson JD, Berry RJ. 1988. Periconceptualuse of multivitamins and the occurrence of neural tube defects.JAMA 260:3141–3145.

Orr RB, Kamen BA. 1995. Identification of a point mutation in thefolate receptor gene that confers a dominant negative phenotype.Cancer Res 55:847–855.

Ou CY, Stevenson RE, Brown VK, Schwartz CE, Allen WP, Khoury MJ,

Rozen R, Oakely GP, Adams MJ. 1996. 5,10 methylenetetrahydrofo-late reductase genetic polymorphism as a risk factor for neural tubedefects. Am J Med Genet 63:610–614.

Risch N, Merikangas K. 1996. The future of genetic studies of complexhuman diseases. Science 273:1516–1517.

Seller MJ. 1994. Vitamins, folic acid, and the cause and prevention ofneural tube defects. Ciba Found Symp 181:161–179.

Shaw GM, Rozen R, Finell RH, Wasserman CR, Lammer EJ. 1998.Maternal vitamin use, genetic variation of infant methylenetetrahy-drofolate reductase, and risk for spina bifida. Am J Epidemiol148:30–37.

Smithells RW, Sheppard S, Wild J, Schorah CJ. 1989. Prevention ofneural tube defect recurrences in Yorkshire: Final report. Lancet2:498–499.

Speer MC, Worley G, Wolpert CM, Viles K, Beaty M, Mackey JF,Aylsworth AS, Albright SG, Lucas A, Fuchs HE, George T, Pericak-Vance MA. 1996. Investigations of candidate genes in lumbosacralmyelomeningocele. Am J Hum Genet [Suppl] 59:389.

Spielman RS, McGinnis RE, Ewens WJ. 1993. Transmission test forlinkage disequilibrium: the insulin gene region and insulin-dependent diabetes mellitus (IDDM). Am J Hum Genet 52:506–616.

Tate WP, Brown CM. 1992. Translational termination: ‘‘stop’’ forprotein synthesis or ‘‘pause’’ for regulation of gene expression.Biochemistry 31:2443–2450.

Terwilliger JD. 1995. A powerful likelihood method for analysis oflinkage disequilibrium between trait loci and one or more polymor-phic marker loci. Am J Hum Genet 56:777–787.

Van Allen M. 1996. Multisite neural tube closure in humans. BirthDefects 30:203–225.

Van Allen MI, Kalousek DK, Chernoff GF, Juriloff D, Harris M,McGillivray BC, Yong S, Langlois S, MacLeod PM, Chiayat D,Friedman JM, Wilson RD, McFadden D, Pantzar J, Ritchie S, Hall J.1993. Evidence for multi-site closure of the neural tube in humans.Am J Med Genet 47:723–743.

van der Put N, Steegers-Theunissen R, Frosst P, Trijbels F, Eskes T,van den Heuvel L, Mariman E, Heyer M, Rozen R, Blom H. 1995.Mutated methylenetetrahydrofolate reductase as a risk factor forspina bifida. Lancet 346:1070–1071.

van der Put NMJ, van der Molen EF, Kluijtmans LAJ, Heil SG,Trijbels JMF, Eskes TKAB, Van Oppenraaij-Emmerzaal D, BanerjeeR, Blom HJ. 1997. Sequence analysis of the coding region of humanmethionine synthase: Relevance to hyperhomocysteinaemia in neu-ral-tube defects and vascular disease. Q J Med 90 511–517.

van der Put NMJ, Gabreels F, Stevens EMB, Smeitink JAM, TrijbelsFJM, Eskes TKAB, van den Heuvel LP, Blom HJ. 1998. A secondcommon mutation in the methylenetetrahydrofolate reductase gene:An additional risk factor for neural-tube defects? Am J Hum Genet62:1044–1051.

Zhao Q, Behringer RR, de Crombrugghe B. 1996. Prenatal folic acidtreatment suppresses acrania and meroanencephaly in mice mu-tant for the Cart1 homeobox gene. Nat Genet 13:275–283.


Documents

Analysis of select folate pathway genes,PAX3, and humanT in a midwestern neural tube defect population