RESEARCH ARTICLE

OFFICIAL JOURNAL

Insights into Severe 5,10-Methylenetetrahydrofolate

Reductase Deficiency: Molecular Genetic and Enzymatic
Characterization of 76 Patients www.hgvs.org

Patricie Burda,1 † Alexandra Schäfer,1 † Terttu Suormala,1 Till Rummel,2 Céline Bürer,1 Dorothea Heuberger,1
Michele Frapolli,1 Cecilia Giunta,1 Jitka Sokolová,3 Hana Vlášková,3 Viktor Kožich,3 Hans Georg Koch,2,4 Brian Fowler,1
D. Sean Froese,1,5∗ and Matthias R. Baumgartner1,5,6∗
1
Division of Metabolism and Children’s Research Center, University Children’s Hospital, Zurich CH-8032, Switzerland; 2 Department of Pediatrics,
University Hospital, Münster D-48149, Germany; 3 Institute of Inherited Metabolic Disorders, First Faculty of Medicine, Charles University in
Prague and General University Hospital in Prague, Prague, Czech Republic; 4 Klinikum für Kinder- und Jugendmedizin, Klinikum Braunschweig,
Braunschweig D-38118, Germany; 5 radiz – Rare Disease Initiative Zurich, Clinical Research Priority Program for Rare Diseases, University of
Zurich, Switzerland; 6 Zurich Center for Integrative Human Physiology, University of Zurich, Switzerland
Communicated by Jan P. Kraus
Received 18 November 2014; accepted revised manuscript 20 February 2015.
Published online 3 March 2015 in Wiley Online Library (www.wiley.com/humanmutation). DOI: 10.1002/humu.22779

ABSTRACT: 5,10-Methylenetetrahydrofolate reductase
(MTHFR) deficiency is the most common inherited dis-
order of folate metabolism and causes severe hyperho-
mocysteinaemia. To better understand the relationship
ing of the molecular basis of MTHFR dysfunction, and
points to the possible role of cofactor or substrate in the
treatment of patients with specific mutations.
Hum Mutat 36:611–621, 2015. 
C 2015 Wiley Periodicals, Inc.

between mutation and function, we performed molecu-
lar genetic analysis of 76 MTHFR deficient patients, fol-
lowed by extensive enzymatic characterization of fibrob-
lasts from 72 of these. A deleterious mutation was de-
tected on each of the 152 patient alleles, with one al-
lele harboring two mutations. Sixty five different mu-
tations (42 novel) were detected, including a common
splicing mutation (c.1542G>A) found in 21 alleles. Us-
ing an enzyme assay in the physiological direction, we
found residual activity (1.7%–42% of control) in 42 cell
lines, of which 28 showed reduced affinity for nicoti-
namide adenine dinucleotide phosphate (NADPH), one
reduced affinity for methylenetetrahydrofolate, five flavin
adenine dinucleotide-responsiveness, and 24 abnormal ki-
netics of S-adenosylmethionine inhibition. Missense mu-
tations causing virtually absent activity were found ex-
clusively in the N-terminal catalytic domain, whereas
missense mutations in the C-terminal regulatory domain
caused decreased NADPH binding and disturbed inhi-
bition by S-adenosylmethionine. Characterization of pa-
tients in this way provides a basis for improved diagnosis
using expanded enzymatic criteria, increases understand-
Additional Supporting Information may be found in the online version of this article.
†
These authors contributed equally to this work.
∗
Correspondence to: D. Sean Froese. E-mail: sean.froese@kispi.uzh.ch; Matthias
R. Baumgartner, Division of Metabolism and Children’s Research Center, Univer-
sity Children’s Hospital, Steinwiesstrasse 75, Zurich CH-8032, Switzerland. E-mail:
matthias.baumgartner@kispi.uzh.ch
Contract grant sponsors: Rare Disease Initiative Zurich (radiz), a clinical research
priority program for rare diseases by the University of Zurich, Switzerland; the Swiss
National Science Foundation (SNSF 31003A_138521); the research program of the Gen-
eral University Hopsital in Prague (RVO-VFN 64165); and Charles University Prague
(PRVOUK-P24/LF1/3).
KEY WORDS: methylenetetrahydrofolate; MTHFR; homo-
cystinuria; enzyme kinetics
Introduction
5,10-Methylenetetrahydrofolate reductase (MTHFR) deficiency
(MIM #607093), an autosomal-recessive disorder, is the most com-
mon congenital defect of folate metabolism [see Watkins and Rosen-
blatt, 2014]. Severe MTHFR deficiency, defined as residual activity of
less than 20% of the mean control value, is biochemically character-
ized by hyperhomocysteinaemia, homocystinuria, increased plasma
cystathionine, and low-plasma methionine. The clinical presenta-
tion is extremely variable ranging from early onset with severe neu-
rologic abnormalities to a later milder form with onset of psychiatric
and gait disturbances in the second decade or later in adulthood.
MTHFR (EC 1.5.1.20) catalyzes a two-step reaction in
which reducing equivalents are transferred first from nicoti-
namide adenine dinucleotide phosphate (NADPH) to the co-
factor flavin adenine dinucleotide (FAD) and then passed on
to 5,10-methylenetetrahydrofolate (methyleneTHF) forming 5-
methyltetrahydrofolate (methylTHF) [Guenther et al., 1999].
MethylTHF is the most common form of folate in plasma and tissues,
and serves as methyl group donor in the methylTHF-homocysteine
S-methyltransferase (methionine synthase; EC 2.1.1.13) cat-
alyzed remethylation of homocysteine to methionine [Watkins
and Rosenblatt, 2014]. Methionine is in turn activated to S-
adenosylmethionine (AdoMet), which is essential for methylation
of a large number of cellular compounds including DNA, as well as
an allosteric inhibitor of MTHFR.
The initial description of human MTHFR defined a gene of
11 exons with a 2.2-kb cDNA sequence coding for a 656 residue
(70 kDa) protein [Goyette et al., 1998]. However, longer isoforms
have since been isolated, with transcript sizes ranging from 7.5
to 9.5 kb, and one splice variant coding for a novel N-terminal


C 2015 WILEY PERIODICALS, INC.
protein sequence increasing the size by 41 amino acids to 77 kDa
[Tran et al., 2002]. Mammalian MTHFR protein is homodimeric,
and both subunits are composed of an N-terminal catalytic
domain, which includes binding sites for the substrates NADPH
and methyleneTHF as well as the cofactor FAD, linked to a
C-terminal regulatory domain [Matthews et al., 1984]. Structural
determination of an E. coli homolog, which is homotetrameric and
contains only the catalytic domain, has shown that the residues
crucial to FAD binding are scattered across this domain [Guenther
et al., 1999]. Functional studies have demonstrated allosteric
inhibition of MTHFR from AdoMet binding to the regulatory
domain [Sumner et al. 1986]; however, replacement of AdoMet
with S-adenosylhomocysteine can reverse this inhibitory effect
[Daubner and Matthews, 1982; Yamada et al., 2001].
The number of patients with MTHFR deficiency is estimated
to be over 100 [Schiff et al., 2011; Watkins and Rosenblatt, 2012].
Most harbor private mutations, with few occurring in more than
five patients. One exception is c.1141C>T (p.Arg377Cys) found
in the Amish and other populations [Goyette et al., 1996; Sibani
et al., 2003; Tonetti et al., 2003; Strauss et al., 2007]. In addition
to disease-causing mutations, 16 single-nucleotide polymorphic
(SNP) changes, and one two-nucleotide deletion/insertion, lead-
ing to an amino acid change, are known [Martin et al., 2006; Marini
et al., 2008; Pavlikova et al., 2012]. The most common and thor-
oughly investigated of these changes is c.677C>T (p.Ala222Val),
which shows an allele frequency ranging from 24% to 40% (5%–
16% c.677TT individuals) depending on the population studied
[Brattstrom et al., 1998; Hanson et al., 2001; Winkelmayer et al.,
2004; Pavlikova et al., 2012]. This SNP causes thermolability of
MTHFR [Frosst et al., 1995], moderately reduced lymphocyte en-
zyme activity [Frosst et al., 1995; van der Put et al., 1996] and
mild hyperhomocysteinaemia when folate intake is low. This vari-
ation may have clinical consequences in some patients with severe
MTHFR deficiency since expression studies have shown that the
residual activity of a deleterious mutation may be significantly low-
ered by the presence of the c.677T change in the same allele [Goyette
and Rozen, 2000; Sibani et al., 2003]. Two other SNPs, c.1298A>C
(p.Glu429Ala) and c.1793G>A (p.Arg594Gln), present in allelic fre-
quencies of approximately 32% and 5%, respectively [Hanson et al.,
2001; Rady et al., 2002; Winkelmayer et al., 2004; Pavlikova et al.,
2012], were found not to cause enzymatic thermolability or hyper-
homocysteinaemia [van der Put et al., 1998; Hanson et al., 2001].
In MTHFR deficiency, the time of onset and severity of illness
show some degree of correlation with the level of residual enzyme
activity [Watkins and Rosenblatt, 2014] and the genotype [Goyette
et al., 1995]. To better understand mechanisms underpinning this,
as well as to expand the mutational and biochemical spectrum of
MTHFR deficiency, we have identified mutations in 76 patients,
with extensive enzymatic investigation in fibroblasts from 72 of
these. Our results suggest that the type and location of mutation
is important for the manner of biochemical disruption, determines
the level of residual activity, and may give a clue as to the disease
severity/progression as well as potential therapies.
Materials and Methods
Patient Cell Lines and Mutation Identification
Cultured skin fibroblasts from 72 patients and EDTA blood sam-
ples from four additional patients (fibroblast cultures not available)
with clinical and biochemical evidence of MTHFR deficiency, ob-
tained for diagnostic studies with informed consent from the pa-
612 HUMAN MUTATION, Vol. 36, No. 6, 611–621, 2015
tients or their parents and referring clinicians, were included. Details
of cell lines 1–25 and control fibroblasts, as well as cell culture condi-
tions, have already been described [Suormala et al., 2002]. Enzyme
activities and mutations, but not extended enzymatic characteri-
zation have been previously reported for cell lines 16, 49, and 56
[Urreizti et al., 2010; termed patients 2, 4, and 5, respectively]; 33
[Bathgate et al., 2012]; 55 [Forges et al., 2010; younger sibling]; 60
[Tsuji et al., 2011]; and 71 [Crushell et al. 2012].
Genomic DNA (gDNA) was extracted from cultured patient fi-
broblast samples using the QIAamp DNA Mini Kit or from EDTA
blood using the DNEasy Blood and Tissue Kit (Qiagen, Venlo, The
Netherlands). Total RNA was extracted using the RNeasy Mini Kit
(Qiagen). To identify mutations, exons were amplified by PCR from
gDNA using flanking intronic primers (Supp. Table S1) and subse-
quent sequencing by the ABI BigDye method (Applied Biosystems,
Zug, Switzerland). In cases where no or only one mutation was
found, or to confirm splicing defects, cDNA was analyzed following
synthesis from total RNA by RT-PCR using BD Power Script Reverse
Transcriptase (BD Biosciences, San Jose, CA, USA) and the primers
listed in Supp. Table S1, with direct sequencing of RT-PCR products.
Detected mutations were confirmed in all cell lines in a second inde-
pendent gDNA sample isolated from fibroblast cultures started from
the original cryo-preserved cell stock, or from sequencing parental
gDNA when no fibroblasts were available.
The allele refractory mutation system (ARMS) [Newton et al.,
1989] was performed for cell line 7, where sequencing identi-
fied three different heterozygous mutations: c.1809 1810delinsGT,
c.1820C>G, and c.1895T>C. Briefly, a forward primer on intron
10 (MTHFR-IVS10F, all primers are listed in Supp. Table S1) and
a reverse primer-specific either for c.1895T (MTHFR-1895WT) or
for c.1895C (MTHFR-1895ASO), which also contained an addi-
tional mismatch near the 3 end to increase specificity, were used
to amplify a 295-bp fragment covering the region of exon 11 har-
boring the c.1809 1810delinsGT variant. Further, a 268-bp region
harboring the variants c.1820C>G and c.1895T>C was amplified.
For this, a forward primer-specific either for c.1809T (MTHFR-
1809WT) or for c.1809G (MTHFR-1809ASO), the latter harboring
a second mismatch near the 3 end, and a reverse primer on intron
11 (MTHFR-IVS11R) was used. All PCR products were directly
sequenced as described above.
Nucleotide numbers employed here correspond to the original
sequence described by Goyette et al. (1994, 1998) using ENSEMBL
sequence MTHFR-001 (ENST00000376592). This nomenclature
adds 12 nucleotides to the A of the ATG initiation codon. Exons
are numbered from 1 to 11 and introns from 1 to 10 according
to Goyette et al. (1998). The original numbering of nucleotides,
exons, and introns was preferred since the majority of the litera-
ture, including a large number of publications reporting effects of
the MTHFR SNPs, follow this nomenclature. All variants discov-
ered have been deposited in the clinical variant database found at
http://www.ncbi.nlm.nih.gov/clinvar/.
MTHFR Assay
All enzymatic assays were performed using the physiological for-
ward assay as described earlier [Suormala et al., 2002; Rummel
et al., 2007], with minor modifications. Briefly, specific activity
was measured in 50 mM potassium phosphate buffer, pH 6.6, un-
der saturating substrate concentrations (100 μM methyleneTHF;
Eprova AG, Scharrhausen, Switzerland; 200 μM NADPH; Sigma–
Aldrich, Buchs SG, Switzerland) with and without the addition of
FAD (75 μM; Sigma–Aldrich) to detect cofactor responsiveness.
Since the description of the original method, we have used a more
sensitive HPLC system to increase reliable detection of residual ac-
tivities from about 2%–3% of the mean control value to less than
1%. Thermolability was estimated by incubation of the assay mix-
ture with cell extract and assay buffer but without the substrates and
FAD for 5 min at 46°C, followed by the assay of activity with and
without 75 μM FAD. The Km for methyleneTHF was determined
by varying its concentration between 2.5 and 200 μM in the pres-
ence of 200 μM NADPH and 75 μM FAD. The Km for NADPH
was determined by varying its concentration between 10 and
250 μM in the presence of 100 μM methyleneTHF and 75 μM FAD.
All Km values were derived using Eadie–Hofstee plots. To estimate
the Ki for AdoMet inhibition, fibroblast extracts were preincubated
for 5 min at 37°C with variable concentrations of purified AdoMet
(2–1000 μM; Sigma–Aldrich) in the presence of assay buffer and
75 μM FAD, with reactions initiated by adding both substrates
(100 μM methyleneTHF, 200 μM NADPH). The Ki was estimated
from the linear range (slope = –1/Ki ) of the plot in which activ-
ity without AdoMet divided by activity with AdoMet was plotted
against the concentration of AdoMet.

Results

Mutation Identification
Mutations that are certain or likely to be pathogenic were detected
in each of the 152 alleles of our 76 unrelated patients, all of whom
had been referred due to elevated homocyst(e)ine and low to nor-
mal methionine suggesting MTHFR deficiency. Two mutations were
detected in 75 patients and three in one patient. These mutations,
together with the genotype of the c.677C>T polymorphic change for
each patient, are presented in Tables 1 and 2. In total, 65 different
mutations, 42 of which are novel, were identified in our patient co-
hort. Of the 23 previously published mutations that we reported in
this study, seven have been found only in our cohort [Forges et al.,
2010; Urreizti et al., 2010; Tsuji et al., 2011; Bathgate et al., 2012;
Crushell et al., 2012]. Only 19 mutations were detected in more than
one cell line, indicating that most patients carry private mutations.
The most common mutation was c.1542G>A, a splicing mutation
(for details see text below) found in 21 alleles. Analysis of SNPs in
the MTHFR gene revealed the same haplotype in patients carrying
this mutation, suggesting it arose from a single founder. All other
mutations were found in maximally eight alleles. The distribution of
mutations is as follows: 43 missense mutations (67%; 28 novel); four
nonsense mutations (6%; two novel); five deletions or duplications
(8%; four novel); 11 primarily affect splicing (16%; seven novel);
and two no-stop mutations predicting a C-terminal extension of the
protein (3%; one novel). Unusually, we found one small deletion in
the 5 untranslated region (cell line 30). Clinical data for many of
the unpublished patients will be reported in a separate publication.
Sequencing of RT-PCR products from cell lines harboring
three of the novel splice-site mutations revealed exon skip-
ping in each: c.1179-2delA resulted in skipping of exon 7
(r.1179 1359del), causing a frameshift ending in a prema-
ture stop codon (p.Trp389Trpfs∗ 1); whereas c.1644+2T>G and
c.1764+1G>T resulted in skipping of exons 9 (r.1543 1644del) and
10 (r.1645 1764del), resulting in a loss of 34 (p.Ala511 Lys544) and
40 (p.Gly545 Lys584) amino acid residues, respectively. A fourth
novel splice-site mutation, c.1542+2T>C, causes incorporation of
the first five nucleotides of intron 8 into the transcript, thereby re-
sulting in a frameshift (p.Tyr512Trpfs∗ 3). A fifth novel splice-site
mutation, c.792+1G>T, found in intron 4, was not investigated in
detail, but resembles the mutation c.792+1G>A described previously
Figure 1. Identification of an MTHFR splicing defect in patients ho-
mozygous for c.1542G>A (p.Lys510=). A: Electropherogram of RT-PCR
products. RNA extracted from three patient (Nos. 17, 69, 51) and one con-
trol fibroblast cell line was amplified using RT-PCR across a region span-
ning exons 7–9 (primers, forward: TCTACCTGAAGAGCAAGTC; reverse:
CTTCCAGAACATGAAGCTGAC). Bands were resolved using agarose gel
(1.5%) electrophoresis, with molecular mass marker and selected band
sizes (in bp) on the left. Black arrow: cDNA lacking exon 8. Expected
size with exon 8: 536 bp; without: 353 bp. Open arrows: heteroduplex ex-
pected to contain normal cDNA and cDNA with a 5-bp addition (GTGTG)
from intron 8. B: Scheme of mis-splicing due to c.1542G>A. The exons
7–9 are depicted as boxes; the sequences of exon–intron junctions are
shown by letters. The upper part indicated the wild-type allele with
normal splicing pattern. The lower part shows the abnormal splicing of
the mutant allele producing the r.1360_1542del and the r.1542_1543ins5
variant transcripts.
[Goyette et al., 1995] that showed an in-frame deletion of 57 nu-
cleotides from exon 4 (r.736 792del) due to activation of a cryptic
splice site within this exon.
We also found splicing variations from mutations remote from
exon–intron boundaries. RT-PCR analysis of the novel intronic
c.1765-18G>A mutation demonstrated generation of a cryptic 3
acceptor splice site, resulting in retention of 16 nucleotides from the
3 end of intron 10 (r.1764 1765ins16), causing a frameshift followed
by a premature termination codon (p.Asp585Glyfs∗ 14). Analysis of
c.1332G>A, a novel mutation causing a synonymous change for
p.Ser = in exon 7, revealed retention of 137 nucleotides from the
5 end of intron 7 (r.1359 1360ins137), followed by activation of a
cryptic donor splice site within intron 7. This change is predicted to
integrate 16 amino acid residues C-terminally to p.Lys449, followed
by premature termination of translation. Finally, the most common
mutation found in our cohort (c.1542G>A) codes for a synonymous
change of p.Lys510= on the last nucleotide of exon 8. Amplifica-
tion of the mutated region using RNA from cell lines homozygous
for this mutation revealed two different splicing abnormalities. In
combination with agarose electrophoresis (Fig. 1), sequencing of

HUMAN MUTATION, Vol. 36, No. 6, 611–621, 2015 613

Table 1. Mutations Detected in Patient Cell Lines with <1.5% Residual MTHFR Activity and Four Additional Patients Listed According
to the Location of the Mutations from 5 to 3 End
Nucleotide changea Predicted amino acid change Type of mutation MTHFR activityb
Allele 1 Allele 1 Allele 1
Cell line No. Allele 2 Exon/intron Allele 2 Domain Allele 2 c.677 status +FAD (%) +FAD/–FAD

20 c.188G>C Exon 1 p.Trp59Ser Catalytic Missense CC 0.6 1.00

c.188G>C Exon 1 p.Trp59Ser Catalytic Missense
66 c.249-1G>T Intron 1 Splice site Catalytic Splice site CT 0.7 0.89
c.781T>G Exon 4 p.Phe257Val Catalytic Missense
82 c.349G>A Exon 2 p.Ala113Thr Catalytic Missense TT 0.3 0.92
c.792+1G>T Intron 4 Splice site Catalytic Splice site
59 c.349G>A Exon 2 p.Ala113Thr Catalytic Missense TT 0.5 1.20
c.1025T>C Exon 5 p.M338Thr Catalytic Missense
39 c.391C>T Exon 2 p.His127Tyr Catalytic Missense CT 0.1 0.53
c.655_657del Exon 4 p.Lys215del Catalytic Small deletion
36 c.452A>C Exon 2 p.Gln147Pro Catalytic Missense CC 1.2 1.08
c.452A>C Exon 2 p.Gln147Pro Catalytic Missense
41 c.452A>C Exon 2 p.Gln147Pro Catalytic Missense CC 0.5 0.80
c.452A>C Exon 2 p.Gln147Pro Catalytic Missense
43 c.452A>C Exon 2 p.Gln147Pro Catalytic Missense CC 0.5 0.80
c.452A>C Exon 2 p.Gln147Pro Catalytic Missense
60c c.458 459delinsTT Exon 2 p.Gly149Val Catalytic Missense CC 0.8 1.14
c.458 459delinsTT Exon 2 p.Gly149Val Catalytic Missense
42 c.559C>T Exon 3 p.Arg183∗ Catalytic Nonsense CC 0.5 0.87
c.559C>T Exon 3 p.Arg183∗ Catalytic Nonsense
64 c.559C>T Exon 3 p.Arg183∗ Catalytic Nonsense CC 0.5 1.03
c.1179-2delA Intron 6 p.Trp389Trpfs∗ 1 Regulatory Splice site
74 c.776G>T Exon 4 p.Gly255Val Catalytic Missense CC 0.7 0.89
c.1179-2delA Intron 6 p.Trp389Trpfs∗ 1 Regulatory Splice site
44 c.779T>A Exon 4 p.Ile256Asn Catalytic Missense CT 0.8 1.12
c.1025T>C Exon 5 p.Met338Thr Catalytic Missense
19 c.1027T>G Exon 5 p.Trp339Gly Catalytic Missense/splicing TT 0.6 0.82
c.1027T>G Exon 5 p.Trp339Gly Catalytic Missense/splicing
23 c.1027T>G Exon 5 p.Trp339Gly Catalytic Missense/splicing TT 0.5 1.03
c.1027T>G Exon 5 p.Trp339Gly Catalytic Missense/splicing
40 c.1027T>G Exon 5 p.Trp339Gly Catalytic Missense/splicing TT 0.5 1.03
c.1027T>G Exon 5 p.Trp339Gly Catalytic Missense/splicing
48 c.1027T>G Exon 5 p.Trp339Gly Catalytic Missense/splicing TT 0.2 1.50
c.1027T>G Exon 5 p.Trp339Gly Catalytic Missense/splicing
21 c.1179-2delA Intron 6 p.Trp389Trpfs∗ 1 Regulatory Splice site CC 0.5 0.84
c.1179-2delA Intron 6 p.Trp389Trpfs∗ 1 Regulatory Splice site
65 c.1332G>A Exon 7 (p.Ser440=)/splicing Regulatory Splicing CT 1.1 1.07
c.1695G>A Exon 10 p.Trp561∗ Regulatory Nonsense
16d c.1420G>T Exon 8 p.Glu470∗ Regulatory Nonsense CC 0.6 0.95
c.1420G>T Exon 8 p.Glu470∗ Regulatory Nonsense
18 c.1420G>T Exon 8 p.Glu470∗ Regulatory Nonsense CC 0.5 0.84
c.1420G>T Exon 8 p.Glu470∗ Regulatory Nonsense
25 c.1420G>T Exon 8 p.Glu470∗ Regulatory Nonsense CC 0.6 0.95
c.1420G>T Exon 8 p.Glu470∗ Regulatory Nonsense
84e c.1420G>T Exon 8 p.Glu470∗ Regulatory Nonsense CC – –
c.1420G>T Exon 8 p.Glu470∗ Regulatory Nonsense
17 c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing CC 0.6 0.82
c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing
22 c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing CC 0.6 1.02
c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing
50 c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing CC 0.6 0.89
c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing
51 c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing CC 0.3 1.08
c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing
67 c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing CC 0.6 0.83
c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing
69 c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing CC 0.7 1.08
c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing
70 c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing CC 0.8 1.17
c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing
83e c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing CC – –
c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing
85e c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing CC – –
c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing
86e c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing CC – –
c.1542G>A Exon 8 (p.Lys510=)/splicing Regulatory Splicing
34 c.1981T>C Exon 11 p.∗ 657Argextfs∗ 50 Regulatory No-stop CC 0.9 1.06
c.1981T>C Exon 11 p.∗ 657Argextfs∗ 50 Regulatory No-stop

Novel mutations are presented in bold.

a
Numbering of the nucleotide changes and exons and introns follows the original nomenclature of Goyette et al. (1998) with +13 as the number of the A of the ATG initiation
codon.
b
Summary of fibroblast MTHFR activities are included for comparison, and include mean-specific activity assayed in the presence of 75 μM FAD (+FAD) and expressed as
percentage of the mean activity in control cells, and mean ratio of activity assayed +FAD divided by activity assayed without added FAD. In patient cells, a mean ratio higher than
1.6 indicates in vitro FAD responsiveness. Details of enzyme data are presented in Supp. Table S3.
c
Our patient 60 is the patient presented in Tsuji et al. (2011).
d
Our patient 16 is patient 2 presented in Urreizti et al. (2010).
e
Patient with only mutation analysis done (mutations confirmed by heterozygosity in the parents); no enzyme activities measured in fibroblasts.

614 HUMAN MUTATION, Vol. 36, No. 6, 611–621, 2015

 Table 2. Mutations Detected in Patient Cell Lines with >1.5% Residual MTHFR Activity Grouped According to Enzyme Characteristics and Summary of Enzymatic Data
Molecular genetic data Summary of enzymatic data
Nucleotide Predicted amino Type of mutation MTHFR activityb Ratio of values
changea acid change (patient/control)
Allele 1 Allele 1 Allele 1 +FAD/
Cell line No. Allele 2 Exon/intron Allele 2 Domain Allele 2 c.677 status +FAD (%) –FAD ratio Heat stable (%) Km c , NADPH Ki d , Ado-Met

A: Cell lines with in vitro FAD responsiveness

09 c.249-1G>T Intron 1 Splice site Catalytic Splice site CC 9.3 2.28 44.8 4.04 2.5
c.482G>A Exon 2 p.Arg157Gln Catalytic Missense
71e c.398C>A Exon 2 p.Thr129Asn Catalytic Missense CT 1.7 1.73 31.4 0.36 6.9
c.769G>T Exon 4 p.Val253Phe Catalytic Missense
32 c.482G>A Exon 2 p.Arg157Gln Catalytic Missense TT 10.4 3.05 16.8 2.94 2.8
c.482G>A Exon 2 p.Arg157Gln Catalytic Missense
55f c.535G>A Exon 3 p.Ala175Thr Catalytic Missense CT 1.7 1.96 51.3 2.29 >18
c.1178G>A Exon 6 (p.Trp389∗ )/splicing Catalytic/regulatory (Nonsense)/splicing
33g c.596C>T Exon 3 p.Ala195Val Catalytic Missense CC 17.0 6.09 36.8 0.55 2.0
c.596C>T Exon 3 p.Ala195Val Catalytic Missense
B: Cell lines with reduced affinity for NADPH
30 c.-40_-41delTC 5 UTR 5 UTR Small deletion CT 9.9 1.10 22.5 3.84 >18
c.1727C>T Exon 10 p.Pro572Leu Regulatory Missense
01 c.167G>A Exon 1 p.Arg52Gln Catalytic Missense CT 25.5 1.02 18.1 2.10 1.8
c.1274G>C Exon 7 p.Trp421Ser Regulatory Missense
61 c.214C>G Exon 1 p.Arg68Gly Catalytic Missense CT 5.1 1.14 29.8 5.90 2.9
c.1644+2T>G Intron 9 Splice site Regulatory Splice site
29 c.276_314dup Exon 2 p.Leu89 Pro101dup Catalytic Duplication CC 8.1 1.09 37.3 5.08 6.3
c.1528T>G Exon 8 p.Tyr506Asp Regulatory Missense
58 c.599G>A Exon 4 p.Gly196Asp Catalytic Missense CC 11.3 1.02 37.9 3.78 >18
c.1172G>A Exon 6 p.Gly387Asp Regulatory Missense
15 c.689_691del Exon 4 p.Ile226del Catalytic Small deletion CT 1.7 1.02 51.1 7.03 >18
c.1765-18G>A Intron 10 p.Asp585Glyfs∗ 14 Regulatory Splicing
76 c.1054C>T Exon 6 p.Pro348Ser Catalytic Missense CT 19.5 1.04 17.0 4.07 0.47
c.1618G>T Exon 9 p.Val536Phe Regulatory Missense
57 c.1072C>T Exon 6 p.His354Tyr Catalytic Missense CC 42.2 1.06 46.2 5.64 0.22
c.1072C>T Exon 6 p.His354Tyr Catalytic Missense
28 c.1100G>A Exon 6 p.Arg363His Regulatory Missense TT 19.2 1.11 16.0 5.99 0.25
c.1100G>A Exon 6 p.Arg363His Regulatory Missense
52 c.1126A>G Exon 6 p.Lys372Glu Regulatory Missense CT 6.3 1.07 11.1 5.19 0.47
c.1542+2T>C Intron 8 p.Tyr512Trpfs∗ 3 Regulatory Splice site
54 c.1141C>T Exon 6 p.Arg377Cys Regulatory Missense CC 7.4 0.98 47.8 4.85 0.57
c.1359+1G>A Intron 7 Splice site Regulatory Splice site
47 c.1142G>A Exon 6 p.Arg377His Regulatory Missense CC 30.4 1.08 47.2 5.75 0.41
c.1142G>A Exon 6 p.Arg377His Regulatory Missense
31 c.1142G>A Exon 6 p.Arg377His Regulatory Missense CC 34.8 1.05 42.4 5.74 0.32
c.1142G>A Exon 6 p.Arg377His Regulatory Missense
10 c.1274G>C Exon 7 p.Trp421Ser Regulatory Missense CT 3.8 1.13 29.0 4.04 10.7
c.1420G>T Exon 8 p.Glu470∗ Regulatory Nonsense
37 c.1644+2T>G Intron 9 Splice site Regulatory Splice site CC 1.8 1.02 52.9 3.56 >18
c.1644+2T>G Intron 9 Splice site Regulatory Splice site
56h c.1733T>G Exon 10 p.Val574Gly Regulatory Missense TT 10.3 1.07 10.7 3.80 3.9
c.1733T>G Exon 10 p.Val574Gly Regulatory Missense
75 c.1736T>G Exon 10 p.Val575Gly Regulatory Missense CC 6.5 1.02 38.8 2.89 >18
c.1736T>G Exon 10 p.Val575Gly Regulatory Missense
12 c.1764+1G>T Intron10 Splice site Regulatory Splice site TT 3.7 1.02 31.5 5.57 >18

HUMAN MUTATION, Vol. 36, No. 6, 611–621, 2015

c.1764+1G>T Intron10 Splice site Regulatory Splice site
14 c.1764+1G>T Intron10 Splice site Regulatory Splice site TT 3.0 1.10 29.7 4.36 >18
c.1764+1G>T Intron10 Splice site Regulatory Splice site
13 c.1765-18G>A Intron10 p.Asp585Glyfs∗ 14 Regulatory Splicing TT 2.5 1.06 29.5 4.44 >18
c.1765-18G>A Intron10 p.Asp585Glyfs∗ 14 Regulatory Splicing
(Continued)

615
616
Table 2. Continued
Molecular genetic data Summary of enzymatic data
Nucleotide Predicted amino Type of mutation MTHFR activityb Ratio of values
changea acid change (patient/control)
Allele 1 Allele 1 Allele 1 +FAD/
Cell line No. Allele 2 Exon/intron Allele 2 Domain Allele 2 c.677 status +FAD (%) –FAD ratio Heat stable (%) Km c , NADPH Ki d , Ado-Met

26 c.1765-18G>A Intron10 p.Asp585Glyfs∗ 14 Regulatory Splicing CC 3.5 1.10 44.9 6.29 >18
c.1765-18G>A Intron10 p.Asp585Glyfs∗ 14 Regulatory Splicing
79 c.1765-18G>A Intron10 p.Asp585Glyfs∗ 14 Regulatory Splicing CC 3.7 1.05 43.7 4.18 >18
c.1765-18G>A Intron10 p.Asp585Glyfs∗ 14 Regulatory Splicing
49h c.1780delC Exon 11 p.Leu590Cysfs∗ 72 Regulatory Small deletion CC 2.0 1.04 51.4 4.18 >18
c.1780delC Exon 11 p.Leu590Cysfs∗ 72 Regulatory Small deletion
38 c.1805T>C Exon 11 p.Leu598Pro Regulatory Missense CC 3.1 1.09 43.4 2.25 >18
c.1805T>C Exon 11 p.Leu598Pro Regulatory Missense
07 c.[1809_1810deli Exon 11 p.[Tyr599∗ ; Regulatory Nonsense CT 5.9 1.07 43.1 2.35 9.3
nsGT; 1820C>G] Exon 11 Ser603Cys] Regulatory Missense
c.1895T>C Exon 11 p.Leu628Pro Regulatory Missense
C: Cell lines with normal affinity for NADPH
04 c.148C>T Exon 1 p.Arg46Trp Catalytic Missense TT 9.7 1.08 10.2 1.13 1.6
c.167G>A Exon 1 p.Arg52Gln Catalytic Missense

HUMAN MUTATION, Vol. 36, No. 6, 611–621, 2015

35 c.148C>T Exon 1 p.Arg46Trp Catalytic Missense TT 2.3 1.06 30.8 1.28 5.9
c.1982G>C Exon 11 p.∗ 657Serextfs∗ 50 Regulatory No-stop
06 c.149G>A Exon 1 p.Arg46Gln Catalytic Missense CT 7.3 1.14 15.2 0.90 2.4
c.559C>T Exon 3 p.Arg183∗ Catalytic Nonsense
11 c.256C>T Exon 2 p.Arg82Trp Catalytic Missense TT 3.4 1.15 16.4 1.51 2.6
c.400T>C Exon 2 p.Cys130Arg Catalytic Missense
78 c.349G>A Exon 2 p.Ala113Thr Catalytic Missense CC 3.6 1.06 16.4 1.52 1.3
c.349G>A Exon 2 p.Ala113Thr Catalytic Missense
03 c.471C>G Exon 2 p.Ile153Met Catalytic Missense CC 14.8 1.01 25.1 1.11 1.7
c.1542G>A Exon 8 p.Lys510=/splicing Regulatory Splicing
53 c.560G>A Exon 3 p.Arg183Gln Catalytic Missense CC 20.6 1.02 27.1 1.36 0.83
c.560G>A Exon 3 p.Arg183Gln Catalytic Missense
68 c.560G>A Exon 3 p.Arg183Gln Catalytic Missense CC 21.8 0.99 27.4 1.23 0.59
c.560G>A Exon 3 p.Arg183Gln Catalytic Missense
77 c.685A>C Exon 4 p.Ile225Leu Catalytic Missense TT 4.2 1.06 13.9 1.15 2.4
c.685A>C Exon 4 p.Ile225Leu Catalytic Missense
62 c.772C>T Exon 4 p.Pro254Ser Catalytic Missense CC 2.0 0.98 54.3 0.57 2.1
c.772C>T Exon 4 p.Pro254Ser Catalytic Missense
73 c.1016G>A Exon 5 p.Arg335His Catalytic Missense CT 5.3 0.98 27.6 0.80 1.9
c.1179-2delA Intron 6 p.Trp389Trpfs∗ 1 Regulatory Splice site
72 c.1332G>A Exon 7 p.Ser440 = /splicing Regulatory Splicing CC 2.2 0.94 66.1 1.17 11.3
c.1644+2T>G Intron 9 Splice site Regulatory Splice site

Novel mutations are presented in bold.

a
Numbering of the nucleotide changes, as well as of exons and introns, follows the original nomenclature of Goyette et al. (1998) with +13 as the number of the A of the ATG initiation codon, except for the c.-40_-41delTC change within 5 UTR for
which the A of the ATG initiation codon is +1.
Summary of enzymatic data are given in the table for comparison and include:
b
Mean MTHFR activity in fibroblasts assayed with 75 μM FAD and expressed as percentage of the mean control value (+FAD, %), the mean +FAD/–FAD activity ratio (a ratio higher than 1.6 indicates in vitro FAD responsiveness), as well as the mean
percentage of thermo-stable activity remaining after heat treatment (46°C for 5 min) (for details see Supp. Table S3).
c
The relative Km for NADPH expressed as the ratio of mean Km in patient cells divided by mean control Km (for details see Supp. Table S4).
d
The relative Ki for AdoMet inhibition expressed as the ratio of mean Ki in patient cells divided by mean control Ki (for details see Supp. Table S4); >18 represents Ki values higher than 1 mM interpreted as the absence of inhibition.
e
Our patient 71 is the one reported in Crushell et al. (2012).
f
Our patient 55 is the younger sibling in Forges et al. (2010).
g
Our patient 33 is the proband in Bathgate et al. (2012).
h
Our patient 49 is patient 4, and our patient 56 is patient 5 in Urreizti et al. (2010).
RT-PCR products identified skipping of exon 8 (r.1360 1542del),
producing a protein product lacking the 61 amino acids encoded
by this exon, and insertion of five nucleotides from the 5 end of
intron 8 (r.1542 1543ins5), resulting in a frameshift and premature
termination of the protein. This mutation has been previously re-
ported in the homozygous state in one patient, but the molecular
and functional consequences were not studied [Richard et al., 2013].
To explore the underlying mechanisms of the novel splicing de-
fects in the MTHFR gene, we performed in silico analyses of the
changes in consensual splice sites as well as in exon splicing en-
hancers and silencers, respectively (Supp. Table S2). Evaluation of
splice sites detected either destruction of canonical sites or creation
of novel sites that were all consistent with the changes observed in
patient mRNAs. Analysis of exon splicing enhancers and silencers
indicated that some mutations altered these regulatory elements and
may have contributed to the splicing defect.
In cell line 7, three mutations were detected, c.1809 1810delinsGT
(p.Tyr599∗ ), c.1820C>G (p.Ser603Cys), and c.1895T>C
(p.Leu628Pro). To determine which mutations came from
the same allele, we applied ARMS (see Materials and Meth-
ods). Sequencing of the ARMS-PCR products revealed that the
c.1895T wild-type allele co-segregated with the mutant variants
c.1809 1810delinsGT and c.1820C>G, whereas the mutant variant
c.1895C co-segregated with the wild-type variants of c.1809 1810
and c.1820. Therefore, we conclude that the patient is compound
heterozygous for c.[1809 1810delinsGT; 1820C>G] and c.1895T>C.
MTHFR Activity and Kinetics in Patient Primary Fibroblasts
We characterized the MTHFR activity of primary skin fibroblasts
from 72 patients, using the physiological forward assay [Suormala
et al., 2002]. Detailed results are presented in Supp. Tables S3 and
S4 and are summarized in Tables 1 and 2 compared with the mean
values obtained in control cell lines. Activity was found to be very
low (<1.5% of the mean control value) in 30 cell lines (Table 1),
whereas the other 42 cell lines retained residual activity ranging
from 1.7%–42% (Table 2). The four further patients from whom
no fibroblast cultures were available are also shown in Table 1. They
were homozygous for the same mutations already detected in the
homozygous state in cell lines of the very low-activity group. In
addition to mutation type, activity was influenced by a number of
features related to cell culture conditions such as growth rate, cell
morphology, passage number, and degree of confluence on har-
vesting. Therefore, whenever possible, cultures with a low-passage
number (90% below 10, maximum 17) were used and cells were har-
vested at least 3 days post-confluence. Additionally, all assays were
performed in at least two separate experiments using fibroblasts
from independent cultures (Supp. Table S3).
Every cell line was tested for FAD responsiveness, that is, compar-
ison of activity without (–FAD) and with (+FAD) supplementation
of 75 μM FAD to the activity assay. In five cell lines, activity +FAD
was reproducibly markedly higher than without (mean ratio +FAD/–
FAD: 1.73–6.09), indicating in vitro FAD responsiveness (Table 2A).
In the other 67 cell lines, no such responsiveness was observed (mean
ratio +FAD/–FAD: 0.53–1.17), consistent with previously published
values in 75 control cell lines (0.96–1.08) [Suormala et al., 2002]. The
apparent Km of MTHFR for FAD, in the FAD-responsive cell lines,
ranged from 50 to 340 nM (data not shown). We were unable to
measure the Km for FAD in control cells since the level of riboflavin,
the precursor of FAD, in normal MEM media (0.3 μM) was suf-
ficient to saturate this enzyme with cofactor. Each FAD-responsive
cell line carried at least one missense mutation corresponding to
a residue involved in FAD binding in the E. coli MTHFR crystal
structure [Guenther et al., 1999], including p.Thr129, p.Arg157,
p.Ala175, and p.Ala195 (Table 2A). Only one cell line (No. 39) was
identified as not FAD-responsive, even though it carried a mutation
on a residue described in the E. coli enzyme to be important for
FAD binding (p.His127Tyr). This cell line additionally harbored a
deletion (p.Lys215del) and had virtually absent activity (0.1% of
mean control; Supp. Table S3), preventing accurate assessment of
FAD responsiveness.
For those 42 cell lines with sufficient residual activity (>1.5%),
we determined the apparent Km for methyleneTHF and NADPH as
well as the Ki for AdoMet (for detailed results see Supp. Table S4).
The mean Km for methyleneTHF was virtually normal for every
patient cell line (15.0–37.5 μM; control range: 17.7–42.2 μM) except
for cell line 71 (mean of three determinations: 107.9 μM) (Supp.
Table S4). This cell line also had low-residual activity (1.7%),
increased in vitro FAD responsiveness (ratio +FAD/–FAD: 1.73),
slightly increased affinity for NADPH (mean Km 0.37 times mean
control), and reduced response to AdoMet (mean Ki 6.9 times
mean control) (Table 2). Therefore, the importance of the increased
methyleneTHF Km is unclear.
We found reduced NADPH affinity in 28 cell lines, including three
with in vitro FAD responsiveness (Table 2), with Km values two to
seven times the mean control value of 27.8 μM (Supp. Table S4).
These cell lines had widely varying activity (1.7%–42% of mean
control) in routine assay conditions with 200 μM NADPH, and
many had abnormal kinetics for AdoMet inhibition. Thirteen cell
lines had virtually absent inhibition (Ki > 18 times the mean control
value of 55.6 μM; Supp. Table S4), four cell lines had reduced
inhibition (Ki four to 11 times mean control), whereas seven cell
lines had high normal or increased sensitivity to inhibition (Ki 0.22–
0.58 times mean control) (Table 2).
The Km for NADPH was normal in the remaining 13 cell lines
(range: 0.80–1.52 times control; Table 2). One of these was FAD-
responsive (No. 33), and two (Nos. 35 and 72) showed reduced
inhibition by AdoMet, whereas the remainder exhibited normal
kinetic characteristics.
SNPs and Thermolability
In addition to mutational analysis, we genotyped each of our cells
lines for the c.677C>T (p.Ala222Val) polymorphism. From our 76
mutant cell lines, genotyping revealed 58% CC, 21% CT, and 21%
TT, giving a minor allele frequency of 32%. Following heat treat-
ment of those mutant cells with detectable MTHFR activity (42 cell
lines), those with the CC genotype had 42 ± 11% activity remaining
compared with parallel cell lysates without heat treatment, those
with CT had 29 ± 14% activity remaining, and TT had 21 ± 9%.
In control cells, we previously [Suormala et al., 2002] found that
cells with the CC genotype had 62 ± 3% activity remaining, CT
had 40 ± 4% and TT 18 ± 2%. Our mutant cells show the same
trend, with CC > CT > TT; however, the absolute residual activity
was lower than expected for the CC and CT genotypes, but not TT.
This may be due to protein thermolability caused by the mutations
themselves.
Discussion
Severely Disrupting Mutations
All missense mutations detected in the 30 cell lines leading
to severely reduced (<1.5% of mean control) MTHFR activity
HUMAN MUTATION, Vol. 36, No. 6, 611–621, 2015 617
Figure 2. Schematic representation of the MTHFR gene and protein showing type location of all mutations identified in our cell cohort. The
MTHFR coding sequence is schematized as a horizontal bar in the middle, with exons, cDNA sequence, and amino acid numbers at exon boundaries
identified. The position of the three domains, catalytic, linker, and regulatory, are shown on a line above. NADPH-, methyleneTHF-, FAD-, and
AdoMet-binding domains are shown for easy visualization, but do not represent exact binding sites. NADPH is shown with a question mark over
the regulatory domain as a possible interpretation of our results.

(Table 1) were located within the N-terminal catalytic domain,
whereas all mutations in the C-terminal regulatory domain were
nonsense, splice-site/splicing, or no-stop mutations (Fig. 2). These
results clearly reinforce the concept of segregation of the N- and
C-terminal domains in the function of MTHFR, whereby substi-
tutions in the first 340 residues may result in almost complete
loss of catalytic activity, whereas in the regulatory domain only
truncating mutations have such a severe effect. In accordance with
this severely reduced MTHFR activity, all published reports on our
patients belonging to this group (Nos. 16, 19, 20, 22, 25, and 60)
describe severe clinical presentation with early onset of symptoms
and/or death before 2 years of age [Suormala et al., 2002; Tsuji
et al., 2011]. Similarly, a severe course has been reported for other
patients that are homozygous for mutations detected in this low ac-
tivity group, that is, for the nonsense mutation p.Arg183∗ [Goyette
et al., 1994], the missense mutations p.Gly149Val [Goyette et al.,
1996] and p.Trp339Gly [Kluijtmans et al., 1998; Sibani et al., 2003],
as well as the no-stop mutation p.∗ 657Serextfs∗ 50 [Tonetti et al.,
2003] that closely resembles the p.∗ 657Argextfs∗ 50 described in cell
line No. 34.
The most common mutation in our cohort is a synonymous
change, c.1542G>A (p.Lys510=), that was shown to cause defec-
tive splicing and very low-enzyme activity (Fig. 1; Table 1). A fur-
ther synonymous mutation causing defective splicing, c.1332G>A
(p.Ser440=), was detected heterozygously in two of our cell lines
(Nos. 65 and 72) with very low-enzyme activity. These findings, in
conjunction with c.486A>T (p.Gly158=), a previously described mu-
tation in Bukharian Jews associated with severe clinical presentation
and very low-enzyme activity [Ben-Shachar et al., 2012], stresses the
importance of investigating synonymous changes at the cDNA level
to detect splicing abnormalities in this gene.
Mutations Conferring Residual Activity
Our use of a sensitive assay in the physiological forward direction
allowed detailed enzymatic investigations of the 42 cell lines with
residual activity above 1.5% of the mean control value. Based on
enzyme characteristics, we divided these cell lines into three sub-
groups (Table 2): (1) five cell lines with FAD responsiveness; (2) 25
cell lines with reduced affinity for NADPH; and (3) 12 cell lines with
normal affinity for the substrates.
The 5 FAD-responsive cell lines all had at least one missense
mutation in the N-terminal catalytic domain on a residue corre-
sponding to an amino acid involved in FAD binding in the E. coli
MTHFR structure (p.Thr129Asn, p.Arg157Gln, p.Ala175Thr, and
p.Ala195Val) (Fig. 2). This suggests that the FAD-binding func-
tion of these residues is conserved in humans. These cell lines,
two with homozygous and three with heterozygous mutations
(Table 2A), demonstrated diverse enzyme characteristics, includ-
ing variation of MTHFR activity from 1.7% to 17% of the mean
control value. One other cell line (No. 39) had a missense change
on an FAD-binding residue equivalent (p.His127Tyr), in conjunc-
tion with a deletion (p.Lys215del), resulting in nearly complete

618 HUMAN MUTATION, Vol. 36, No. 6, 611–621, 2015

inactivation of the enzyme. This very low activity prevented proper
assessment of FAD-responsiveness, but supports the notion that
some FAD-binding residues are more critical than others for cat-
alytic function. Loss of FAD as a mechanism of instability is in ac-
cordance with studies of the common SNP p.Ala222Val, which have
shown that its thermolability is caused by an increased propensity to
lose FAD, resulting in dissociation of the holoenzyme into inactive
monomers [Guenther et al., 1999; Yamada et al., 2001].
Our fibroblast results suggest that patients harboring FAD-
responsive changes might be responsive to therapy including supple-
mentation with riboflavin, the precursor to FAD. Current treatment
for MTHFR deficiency includes betaine in combination with B vita-
mins, that is, foli(ni)c acid, cobalamin, pyridoxine, and in some cases
riboflavin [Watkins and Rosenblatt, 2014]. Therefore, the value of
riboflavin as a treatment is difficult to judge, since it has been little
reported and then only in combination with other agents. On the
other hand, the relation of riboflavin to homocysteine in common
diseases has been widely reported, for example, the influence of ri-
boflavin status on the relationship between MTHFR polymorphisms
and homocysteine [Hustad et al., 2000]. Also, Wilson et al. [2013]
reported a positive effect of riboflavin supplementation on blood
pressure in subjects with the c.677TT genotype. Further support
for the idea of using riboflavin in the treatment of severe MTHFR
deficiency is provided by the finding of increased stability of a partic-
ular expressed mutant enzyme to the coenzyme FAD [Sibani et al.,
2003]. Therefore, there is not yet any conclusive data in which to
evaluate the effectiveness of riboflavin on FAD-responsive patients;
however, we suggest that it might be beneficial to test its efficacy in
those patients with specific FAD-responsive mutations, due to its
stabilizing effect in the cell.
Kinetic studies revealed that reduced affinity of MTHFR for
methyleneTHF was scarce and detected only together with FAD
responsiveness. Thus, the Km for methyleneTHF was clearly ele-
vated in one cell line (No. 71) heterozygous for p.Thr129Asn, a
mutation associated with FAD-binding, and p.Val253Phe [Crushell
et al., 2012]. MethyleneTHF-binding residues have been described
in structures of E. coli MTHFR [Pejchal et al., 2006; Lee et al.,
2009]; however, none correspond to human residues p.Thr129 or
p.Val253. One folate-binding amino acid found in these E. coli struc-
tures, p.Arg279 [Pejchal et al., 2005], corresponding to p.Arg335 in
humans, was found to be mutated in one allele of a cell line stud-
ied (no. 73). However, this cell line showed no disturbances in
methyleneTHF binding. Together, these results do not support the
possibility that residues that bind methyleneTHF in E. coli perform
the same function in human MTHFR.
A major finding in this study is reduced affinity for NADPH
together with normal affinity for methyleneTHF and FAD (25 cell
lines; Table 2B). Unexpectedly, all but one cell line had at least one
allele with a mutation in the C-terminal regulatory domain, and
all 14 different missense mutations detected within the regulatory
domain were found in this cell line group (Table 2B; Fig. 2). These
data suggest that the regulatory domain at least partially mediates
NADPH binding. It is of note that in E. coli the NADH binding
site is in the catalytic domain [Pejchal et al., 2005; Lee et al., 2009].
Studies with pig liver MTHFR revealed that tryptic cleavage of the
linker between the catalytic and regulatory domains did not result in
loss of NAPDH-mediated activity [Matthews et al., 1984], implying
binding of NADPH to the catalytic domain. However, if mammalian
MTHFR exists as a head-to-tail dimer, where the regulatory region
of one subunit interacts with the catalytic region of the other, as has
been proposed [Yamada et al., 2001], cleavage of the intrasubunit
link between the catalytic and regulatory domains would not be
expected to affect activity. This would be consistent with our results,
as well as the planar rosette structure observed for porcine dimeric
MTHFR by scanning electron microscopy [Matthews et al., 1984].
In 19 of the 25 cell lines with reduced affinity for NADPH,
MTHFR activity was clearly reduced to 1.7%–11% of the mean con-
trol value. Consequently, patients from this subgroup from whom
clinical data have been reported, presented with moderate to se-
vere illness [Nos. 7, 10, 12–15; Suormala et al., 2002], or in the
first months of life with mainly neurological abnormalities and
developmental delay [Nos. 49, 56; Urreizti et al., 2010]. These rela-
tionships reinforce the correlation between residual enzyme activity
and disease severity [Suormala et al., 2002]. MTHFR activity in
the remaining six cell lines varied between 19% and 42% of mean
control, that is, levels that would be considered borderline or not
causative of MTHFR deficiency by the current definition [Watkins
and Rosenblatt, 2014]. However, the Km for NADPH in five of these
cell lines was elevated four to six times above control, and two times
in the sixth cell line (Table 2B). The clinical presentation of patient
01, the only published report of these six cell lines, was mild [Suor-
mala et al., 2002]. A recent report on a pair of siblings homozygous
for p.Arg377His has been published [Lossos et al., 2014]. In these
patients, fibroblast enzyme activity varied between 18% and 52% of
control, similar to our findings of 35% and 30% enzyme activity in
cells homozygous for the same mutation (Nos. 31, 47). The patients
described by Lossos et al. [2014] presented between 29 and 50 years
of age with progressive spastic paraparesis and polyneuropathy asso-
ciated with behavioral changes and cognitive impairment, and were
responsive to therapy. Therefore, even in individuals with high levels
of enzyme activity, kinetic abnormalities may well lead to functional
deficiency causing clinical abnormalities—reinforcing the need for
kinetic characterization beyond measurement of specific activity.
Whether high-dose substrate (precursor) therapy (with e.g. nicoti-
namide) would be of further benefit to these patients remains to be
determined.
Many of the cell lines with disturbed NADPH binding also showed
disturbed AdoMet inhibition. This may be either due to the an-
tagonistic effect previously reported for these ligands [Jencks and
Mathews, 1987], or due to direct affinity changes from mutation of
potential binding residues within the regulatory domain known to
bind AdoMet [Sumner et al., 1986]. In support of the latter expla-
nation, all seven cell lines with high-normal or increased sensitivity
to AdoMet inhibition had at least one missense mutation clustered
around the putative linker region in exon 6, whereas 10 out of the
12 cell lines with virtually no inhibition had at least one mutation
clustering around the exon 10–11 boundary (Table 2). Although the
mutational pattern is intriguing, the physiological significance of
AdoMet-binding disturbance remains unclear.
Twelve cell lines, with 16 different mutations, had residual activity
but no disturbances in affinity for the substrates or FAD. The de-
creased activity in combination with unperturbed binding kinetics
in these cell lines points to alternative mechanisms of dysfunction.
Conclusion
We have performed an extensive biochemical and molecular ge-
netic investigation into severe MTHFR deficiency. We found a good
correlation between the type and location of mutation and enzyme
level and kinetic characteristics. Our results suggest that patients that
have high-residual activity, for example, >30%, may remain unde-
tected if only specific activity is assayed, indicating the importance of
performing enzyme kinetics, especially in vitro FAD-responsiveness
and affinity values for the natural substrates to detect functional
abnormalities. However, this requires utilization of the physiological
HUMAN MUTATION, Vol. 36, No. 6, 611–621, 2015 619
forward assay. This study reinforces the notion of distinct domains
responsible for specific substrate and cofactor binding and suggests
that reduced NADPH binding may be the cause of MTHFR defi-
ciency in some patients, and that its binding is governed by the reg-
ulatory domain. Further, our findings point to the possible employ-
ment of cofactor (riboflavin) or substrate (nicotinamide) as therapy
for patients with specific mutations in this difficult to treat disorder.
The correlation of clinical data with the thorough experimental ap-
proach described here will help further clarify genotype–phenotype
relationships as well as their underlying mechanisms.
References
Bathgate D, Yu-Wai-Man P, Webb B, Taylor RW, Fowler B, Chinnery PF. 2012. Recessive
spastic paraparesis associated with complex I deficiency due to MTHFR mutations.
J Neurol Neurosurg Psychiatry 83:115.
Ben-Shachar S, Zvi T, Rolfs A, Breda Klobus A, Yaron Y, Bar-Shira A, Orr-Urtreger A.
2012. A founder mutation causing a severe methylenetetrahydrofolate reductase
(MTHFR) deficiency in Bukharian Jews. Mol Genet Metab 107:608–610.
Brattstrom L, Wilcken DE, Ohrvik J, Brudin L. 1998. Common methylenetetrahydrofo-
late reductase gene mutation leads to hyperhomocysteinemia but not to vascular
disease: the result of a meta-analysis. Circulation 98:2520–2526.
Coelho D, Kim JC, Miousse IR, Fung S, du Moulin M, Buers I, Suormala T, Burda P,
Frapolli M, Stucki M, Nurnberg P, Thiele H, et al. 2012. Mutations in ABCD4
cause a new inborn error of vitamin B12 metabolism. Nat Genet 44:1152–1155.
Crushell E, O’Leary D, Irvine AD, O’Shea A, Mayne PD, Reardon W. 2012. Methylenete-
trahydrofolate reductase (MTHFR) deficiency presenting as a rash. Am J Med
Genet A 158A:2254–2257.
Daubner SC, Matthews RG. 1982. Purification and properties of methylenetetrahydro-
folate reductase from pig liver. J Biol Chem 257:140–145.
Forges T, Chery C, Audonnet S, Feillet F, Gueant JL. 2010. Life-threatening methylenete-
trahydrofolate reductase (MTHFR) deficiency with extremely early onset: charac-
terization of two novel mutations in compound heterozygous patients. Mol Genet
Metab 100:143–148.
Frosst P, Blom HJ, Milos R, Goyette P, Sheppard CA, Matthews RG, Boers GJ, den
Heijer M, Kluijtmans LA, vanden Heuvel LP, Rozen R. 1995. A candidate genetic
risk factor for vascular disease: a common mutation in methylenetetrahydrofolate
reductase. Nat Genet 10:111–113.
Giunta C, Superti-Furga A, Spranger S, Cole WG, Steinmann B. 1999. Ehlers–Danlos
syndrome type VII: clinical features and molecular defects. J Bone Joint Surg Am
81:225–238.
Goyette P, Christensen B, Rosenblatt DS, Rozen R. 1996. Severe and mild mutations in
cis for the methylenetetrahydrofolate reductase (MTHFR) gene, and description
of five novel mutations in MTHFR. Am J Hum Genet 59:1268–1275.
Goyette P, Frosst P, Rosenblatt DS, Rozen R. 1995. Seven novel mutations in the
methylenetetrahydrofolate reductase gene and genotype/phenotype correlations
in severe methylenetetrahydrofolate reductase deficiency. Am J Hum Genet
56:1052–1059.
Goyette P, Pai A, Milos R, Frosst P, Tran P, Chen Z, Chan M, Rozen R. 1998. Gene
structure of human and mouse methylenetetrahydrofolate reductase (MTHFR).
Mamm Genome 9:652–656.
Goyette P, Rozen R. 2000. The thermolabile variant 677C→T can further reduce activity
when expressed in cis with severe mutations for human methylenetetrahydrofolate
reductase. Hum Mutat 16:132–138.
Goyette P, Sumner JS, Milos R, Duncan AM, Rosenblatt DS, Matthews RG, Rozen R.
1994. Human methylenetetrahydrofolate reductase: isolation of cDNA, mapping
and mutation identification. Nat Genet 7:195–200.
Guenther BD, Sheppard CA, Tran P, Rozen R, Matthews RG, Ludwig ML. 1999. The
structure and properties of methylenetetrahydrofolate reductase from Escherichia
coli suggest how folate ameliorates human hyperhomocysteinemia. Nat Struct Biol
6:359–365.
Hanson NQ, Aras O, Yang F, Tsai MY. 2001. C677T and A1298C polymorphisms of
the methylenetetrahydrofolate reductase gene: incidence and effect of combined
genotypes on plasma fasting and post-methionine load homocysteine in vascular
disease. Clin Chem 47:661–666.
Hustad S, Ueland PM, Vollset SE, Zhang Y, Bjorke-Monsen AL, Schneede J. 2000.
Riboflavin as a determinant of plasma total homocysteine: effect modification
by the methylenetetrahydrofolate reductase C677T polymorphism. Clin Chem
46:1065–1071.
Jencks DA, Mathews RG. 1987. Allosteric inhibition of methylenetetrahydrofolate re-
ductase by adenosylmethionine. Effects of adenosylmethionine and NADPH on
the equilibrium between active and inactive forms of the enzyme and on the
kinetics of approach to equilibrium. J Biol Chem 262:2485–2493.
620 HUMAN MUTATION, Vol. 36, No. 6, 611–621, 2015
Kluijtmans LA, Wendel U, Stevens EM, vanden Heuvel LP, Trijbels FJ, Blom HJ. 1998.
Identification of four novel mutations in severe methylenetetrahydrofolate reduc-
tase deficiency. Eur J Hum Genet 6:257–265.
Lee MN, Takawira D, Nikolova AP, Ballou DP, Furtado VC, Phung NL, Still BR,
Thorstad MK, Tanner JJ, Trimmer EE. 2009. Functional role for the conforma-
tionally mobile phenylalanine 223 in the reaction of methylenetetrahydrofolate
reductase from Escherichia coli. Biochemistry 48:7673–7685.
Litzkas P, Jha KK, Ozer HL. 1984. Efficient transfer of cloned DNA into human diploid
cells: protoplast fusion in suspension. Mol Cell Biol 4:2549–2552.
Lossos A, Teltsh O, Milman T, Meiner V, Rozen R, Leclerc D, Schwahn BC, Karp N,
Rosenblatt DS, Watkins D, Shaag A, Korman SH, et al. 2014. Severe methylenete-
trahydrofolate reductase deficiency: clinical clues to a potentially treatable cause
of adult-onset hereditary spastic paraplegia. JAMA Neurol 71:901–904.
Marini NJ, Gin J, Ziegle J, Keho KH, Ginzinger D, Gilbert DA, Rine J. 2008. The
prevalence of folate-remedial MTHFR enzyme variants in humans. Proc Natl
Acad Sci USA 105:8055–8060.
Martin YN, Salavaggione OE, Eckloff BW, Wieben ED, Schaid DJ, Weinshilboum
RM. 2006. Human methylenetetrahydrofolate reductase pharmacogenomics: gene
resequencing and functional genomics. Pharmacogenet Genomics 16:265–277.
Matthews RG, Vanoni MA, Hainfeld JF, Wall J. 1984. Methylenetetrahydrofolate re-
ductase. Evidence for spatially distinct subunit domains obtained by scanning
transmission electron microscopy and limited proteolysis. J Biol Chem 259:11647–
11650.
Markham AF. 1989. Analysis of any point mutation in DNA. The amplification refrac-
tory mutation system (ARMS). Nucleic Acids Res 17:2503–2516.
Newton CR, Graham A, Heptinstall LE, Powell SJ, Summers C, Kalsheker N, Smith JC,
Markham AF. 1989. Analysis of any point mutation in DNA. The amplification
refractory mutation system (ARMS). Nucleic Acids Res 17:2503–16.
Pavlikova M, Sokolova J, Janosikova B, Melenovska P, Krupkova L, Zvarova J, Kozich V.
2012. Rare allelic variants determine folate status in an unsupplemented European
population. J Nutr 142:1403–1409.
Pejchal R, Campbell E, Guenther BD, Lennon BW, Matthews RG, Ludwig ML. 2006.
Structural perturbations in the Ala → Val polymorphism of methylenetetrahy-
drofolate reductase: how binding of folates may protect against inactivation. Bio-
chemistry 45:4808–4818.
Pejchal R, Sargeant R, Ludwig ML. 2005. Structures of NADH and CH3-H4folate
complexes of Escherichia coli methylenetetrahydrofolate reductase reveal a spartan
strategy for a ping-pong reaction. Biochemistry 44:11447–11457.
Rady PL, Szucs S, Grady J, Hudnall SD, Kellner LH, Nitowsky H, Tyring SK, Matalon RK.
2002. Genetic polymorphisms of methylenetetrahydrofolate reductase (MTHFR)
and methionine synthase reductase (MTRR) in ethnic populations in Texas; a
report of a novel MTHFR polymorphic site, G1793A. Am J Med Genet 107:162–
168.
Richard E, Desviat LR, Ugarte M, Perez B. 2013. Oxidative stress and apoptosis in ho-
mocystinuria patients with genetic remethylation defects. J Cell Biochem 114:183–
191.
Rummel T, Suormala T, Haberle J, Koch HG, Berning C, Perrett D, Fowler B. 2007.
Intermediate hyperhomocysteinaemia and compound heterozygosity for the com-
mon variant c.677C>T and a MTHFR gene mutation. J Inherit Metab Dis 30:401.
Schiff M, Benoist JF, Tilea B, Royer N, Giraudier S, Ogier de Baulny H. 2011. Isolated
remethylation disorders: do our treatments benefit patients? J Inherit Metab Dis
34:137–145.
Shan X, Wang L, Hoffmaster R, Kruger WD. 1999. Functional characterization of
human methylenetetrahydrofolate reductase in Saccharomyces cerevisiae. J Biol
Chem 274:32613–3268.
Sibani S, Leclerc D, Weisberg IS, O’Ferrall E, Watkins D, Artigas C, Rosenblatt DS,
Rozen R. 2003. Characterization of mutations in severe methylenetetrahydrofo-
late reductase deficiency reveals an FAD-responsive mutation. Hum Mutat 21:
509–520.
Strauss KA, Morton DH, Puffenberger EG, Hendrickson C, Robinson DL, Wagner C,
Stabler SP, Allen RH, Chwatko G, Jakubowski H, Niculescu MD, Mudd SH. 2007.
Prevention of brain disease from severe 5,10-methylenetetrahydrofolate reductase
deficiency. Mol Genet Metab 91:165–175.
Sumner J, Jencks DA, Khani S, Matthews RG. 1986. Photoaffinity labeling of
methylenetetrahydrofolate reductase with 8-azido-S-adenosylmethionine. J Biol
Chem 261:7697–7700.
Suormala T, Gamse G, Fowler B. 2002. 5,10-Methylenetetrahydrofolate reductase
(MTHFR) assay in the forward direction: residual activity in MTHFR deficiency.
Clin Chem 48:835–843.
Tonetti C, Saudubray JM, Echenne B, Landrieu P, Giraudier S, Zittoun J. 2003. Re-
lations between molecular and biological abnormalities in 11 families from sib-
lings affected with methylenetetrahydrofolate reductase deficiency. Eur J Pediatr
162:466–475.
Tran P, Leclerc D, Chan M, Pai A, Hiou-Tim F, Wu Q, Goyette P, Artigas C, Milos
R, Rozen R. 2002. Multiple transcription start sites and alternative splicing in the
methylenetetrahydrofolate reductase gene result in two enzyme isoforms. Mamm
Genome 13:483–492.
Tsuji M, Takagi A, Sameshima K, Iai M, Yamashita S, Shinbo H, Furuya N, Kuro-
sawa K, Osaka H. 2011. 5,10-Methylenetetrahydrofolate reductase deficiency with
progressive polyneuropathy in an infant. Brain Dev 33:521–524.
Urreizti R, Moya-Garcia AA, Pino-Angeles A, Cozar M, Langkilde A, Fanhoe U, Esteves
C, Arribas J, Vilaseca MA, Perez-Duenas B, Pineda M, Gonzalez V, et al. 2010.
Molecular characterization of five patients with homocystinuria due to severe
methylenetetrahydrofolate reductase deficiency. Clin Genet 78:441–448.
vander Put NM, Gabreels F, Stevens EM, Smeitink JA, Trijbels FJ, Eskes TK, vanden
Heuvel LP, Blom HJ. 1998. A second common mutation in the methylenetetrahy-
drofolate reductase gene: an additional risk factor for neural-tube defects? Am J
Hum Genet 62:1044–1051.
vander Put NM, vanden Heuvel LP, Steegers-Theunissen RP, Trijbels FJ, Eskes TK,
Mariman EC, den Heyer M, Blom HJ. 1996. Decreased methylene tetrahydrofolate
reductase activity due to the 677C→T mutation in families with spina bifida
offspring. J Mol Med (Berl) 74:691–694.
Watkins D, Rosenblatt DS. 2012. Update and new concepts in vitamin responsive
disorders of folate transport and metabolism. J Inherit Metab Dis 35:665–670.
Watkins D, Rosenblatt DS. 2014. Inherited disorders of folate and cobalamin
transport and metabolism. In: Valle D, Beaudet AL, Vogelstein B, Kinzler
KW, Antonarakis SE, Ballabio A, Gibson M, Mitchell G, editors. The online
metabolic and molecular bases of inherited disease. Online Edition. New York:
McGraw-Hill.
Wilson CP, McNulty H, Ward M, Strain JJ, Trouton TG, Hoeft BA, Weber P, Roos
FF, Horigan G, McAnena L, Scott JM. 2013. Blood pressure in treated hyper-
tensive individuals with the MTHFR 677TT genotype is responsive to interven-
tion with riboflavin: findings of a targeted randomized trial. Hypertension 61:
1302–1308.
Winkelmayer WC, Sunder-Plassmann G, Huber A, Fodinger M. 2004. Patterns
of co-occurrence of three single nucleotide polymorphisms of the 5,10-
methylenetetrahydrofolate reductase gene in kidney transplant recipients. Eur
J Clin Invest 34:613–618.
Yamada K, Chen Z, Rozen R, Matthews RG. 2001. Effects of common polymorphisms
on the properties of recombinant human methylenetetrahydrofolate reductase.
Proc Natl Acad Sci USA 98:14853–14858.
HUMAN MUTATION, Vol. 36, No. 6, 611–621, 2015 621