Deletion of Mthfd1l causes embryonic lethality and
neural tube and craniofacial defects in mice
Jessica Momba, Jordan P. Lewandowskib,c, Joshua D. Bryanta, Rebecca Fitchd, Deborah R. Surmanb, Steven A. Vokesb,c,
and Dean R. Applinga,b,1
aDepartment of Chemistry and Biochemistry,bInstitute for Cellular and Molecular Biology,cSection of Molecular Cell and Developmental Biology, anddDell
Pediatric Research Institute, The University of Texas at Austin, Austin, TX 78712
Edited by David W. Russell, University of Texas Southwestern Medical Center, Dallas, TX, and approved November 30, 2012 (received for review July 3, 2012)
Maternal supplementation with folic acid is known to reduce the
incidence of neural tube defects (NTDs) by as much as 70%. Despite
the strong clinical link between folate and NTDs, the biochemical
mechanisms through which folic acid acts during neural tube
development remain undefined. The Mthfd1l gene encodes a mito-
chondrial monofunctional 10-formyl-tetrahydrofolate synthetase,
termed MTHFD1L. This gene is expressed in adults and at all stages
of mammalian embryogenesis with localized regions of higher ex-
pression along the neural tube, developing brain, craniofacial struc-
catalyzes the last step in the flow of one-carbon units from mito-
chondria to cytoplasm, producing formate from 10-formyl-THF. To
investigate the role of mitochondrial formate production during
All embryos lacking Mthfd1l exhibit aberrant neural tube closure
including craniorachischisis and exencephaly and/or a wavy neural
tube. This fully penetrant folate-pathway mouse model does not
require feeding a folate-deficient diet to cause this phenotype. Ma-
ternal supplementation with sodium formate decreases the inci-
dence of NTDs and partially rescues the growth defect in embryos
lacking Mthfd1l. These results reveal the critical role of mitochond-
rially derived formate in mammalian development, providing a
mechanistic link between folic acid and NTDs. In light of previous
studies linking a common splice variant in the human MTHFD1L
gene with increased risk for NTDs, this mouse model provides
a powerful system to help elucidate the specific metabolic mecha-
nisms that underlie folate-associated birth defects, including NTDs.
complex but poorly understood process. Not surprisingly,
neural tube defects (NTDs) have a multifactorial etiology, in-
cluding both genetic and environmental factors. The importance
of maternal folate status to NTD risk was first suggested more
than 40 y ago (1). Many human studies show that periconcep-
tional intake of supplemental folic acid can reduce the incidence
of NTDs by as much as 70% in some populations (reviewed in ref.
2). These results led to mandated fortification of all enriched
cereal grain products with folic acid in the United States begin-
ning in 1996 to ensure that women of child-bearing age would
consume adequate quantities of the vitamin. Although folic acid
fortification has decreased NTD incidence in some subpopula-
tions, fortification has not completely eliminated NTDs (3). De-
spite the strong clinical link between folate and NTDs, the
biochemical mechanisms through which folic acid acts during
neural tube development remain undefined.
Folate-dependent one-carbon (1C) metabolism is highly com-
partmentalized in eukaryotes, and mitochondria play a critical role
in cellular 1C metabolism (4). The cytoplasmic and mitochondrial
compartments are metabolically connected by transport of 1C
donors such as serine, glycine, and formate across the mitochon-
drial membranes, supporting a mostly unidirectional flow (clock-
thymidylate (dTMP), and methionine. It appears that under most
conditions, the majority of 1C units for cytoplasmic processes are
derived from mitochondrial formate (reviewed in ref. 4). This
formate is exported to the cytoplasm where it is reattached to
losure of the neural tube during development is a highly
tetrahydrofolate (THF) for use in de novo purine biosynthesis, or
further reduced for either thymidylate synthesis or remethylation
of homocysteine to methionine. The 1C unit interconverting ac-
tivities represented in Fig. 1 by reactions 1–3 (and 1m–3m in mi-
tochondria) are the central players in this intercompartmental
pathway. These crucial reactions are catalyzed by members of the
methylenetetrahydrofolate dehydrogenase (MTHFD) family in
eukaryotes.The first member ofthisfamilytobecharacterizedwas
the cytoplasmic MTHFD1 protein, a trifunctional enzyme possess-
ing 10-formyl-THF synthetase, 5,10-methenyl-THF cyclohydrolase,
and 5,10-methylene-THF dehydrogenase activities (reactions 1–3).
This enzyme incorporates formate, released from mitochondria,
into the cytoplasmic 1C THF pool as 10-formyl-THF (CHO-THF),
which is required for de novo purine biosynthesis. MTHFD1 can
THF) for dTMP synthesis (reaction 10), or for methyl group bio-
genesis via 5-methyl-THF (CH3-THF) (reaction 6).
Identification of the enzymes that catalyze reactions 1m–3m in
mammalian mitochondria has lagged behind that of the cyto-
plasmic portion of the pathway. The MTHFD2 protein is a mito-
chondrial bifunctional CH2-THF dehydrogenase/methenyl-THF
cyclohydrolase (reactions 3m and 2m) (5). However, because
MTHFD2 is expressed only in transformed mammalian cells and
embryonic or nondifferentiated tissues (6) the enzyme(s) re-
sponsible for the CH2-THF dehydrogenase/methenyl-THF cyclo-
hydrolase activities observed in adult mammalian mitochondria
(7) remained unknown. This gap was recently filled by identifica-
tion of a new mitochondrial CH2-THF dehydrogenase isozyme,
encoded by the Mthfd2l gene, expressed in embryos and in adult
tissues (8). Like MTHFD2, the MTHFD2L enzyme is bifunc-
tional, possessing both CH2-THF dehydrogenase and methenyl-
THF cyclohydrolase activities (reactions 3m and 2m).
The final step in the mammalian mitochondrial pathway to
formate (reaction 1m) is catalyzed by mitochondrial 10-formyl-
THF synthetase, encoded by the Mthfd1l gene (9). Despite
sharing 61% amino acid similarity with the cytoplasmic trifunc-
tional MTHFD1, MTHFD1L is a monofunctional enzyme, pos-
sessing only the 10-formyl-THF synthetase activity (reaction 1m)
(10). The Mthfd1l gene is expressed in most adult tissues, but at
higher levels in spleen, thymus, brain, and placenta (9, 11). The
Mthfd1l gene is also expressed at all stages of mammalian em-
bryogenesis and ubiquitously throughout the embryo but with
localized regions of higher expression along the neural tube, the
brain, craniofacial structures, limb buds, and the tail bud (12).
Moreover, metabolic tracer experiments in mouse embryonic
fibroblasts showed that more than 75% of 1C units that enter the
cytoplasmic methyl cycle are mitochondrially derived (12). Thus,
in both embryos and adults, MTHFD1L catalyzes production of
Author contributions: J.M., J.P.L., S.A.V., and D.R.A. designed research; J.M., J.P.L., J.D.B.,
R.F., and D.R.S. performed research; J.M., J.P.L., S.A.V., and D.R.A. analyzed data; and J.M.,
J.P.L., S.A.V., and D.R.A. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1To whom correspondence should be addressed. E-mail: email@example.com.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| January 8, 2013
| vol. 110
| no. 2
formate from 10-formyl-THF, the last step in the flow of 1C units
from mitochondria to cytoplasm.
To investigate the role of mitochondrial formate production
during development, we have analyzed Mthfd1l knockout mice.
We show here that loss of MTHFD1L is lethal to developing
embryos, causing fetal growth restriction and aberrant neural
tube closure with 100% penetrance in embryos that develop past
the point of neural tube closure. Although there are other folate-
related mouse models that exhibit NTDs, the Mthfd1l knockout
mouse is a fully penetrant model that does not require feeding a
folate-deficient diet to cause this phenotype. Moreover, we show
that maternal supplementation with sodium formate decreases
the incidence of NTDs and partially rescues the growth defect in
embryos lacking Mthfd1l. These results reveal the critical role of
mitochondrial formate in mammalian development, providing
a mechanistic link between folic acid and neural tube defects. In
light of previous studies linking a common splice variant in the
human MTHFD1L gene with increased risk for NTDs (13), this
mouse model provides a powerful system to help elucidate the
specific metabolic mechanisms that underlie folate-associated
birth defects, including NTDs.
Mthfd1l Is Essential in Mice. We obtained a strain of conditional
knockout ready Mthfd1l mice from the European Conditional
Mouse Mutagenesis Program (EUCOMM). In this strain, the
Mthfd1l locus is modified by the insertion of a cassette, con-
taining a splice acceptor, internal ribosome entry site, the β-ga-
lactosidase gene (LacZ) followed by a polyadenylation signal,
and the gene for neomycin phosphotransferase (Neo) between
exons 4 and 6 of Mthfd1l (Fig. 2A). This allele has three LoxP
sites: one between the polyadenylation signal and Neo and two
flanking exon 5. To generate a null allele, the mice were crossed
to a Cre deleter strain, E2a-Cre (14). Recombination at the LoxP
sites removes Neo and exon 5 to produce a disrupted allele
containing LacZ followed by a polyadenylation signal (Fig. 2A).
Transcription of the disrupted allele is expected to produce a
transcript containing exons 1–4 spliced to LacZ. The disrupted
Mthfd1l allele will herein be designated as Mthfd1lz. The geno-
type was confirmed by PCR (Fig. 2B) and RT-PCR analysis in-
dicated that the WT Mthfd1l transcript is absent in Mthfd1lz/z
embryos (Fig. 2C). Full-length MTHFD1L protein is undetect-
able in Mthfd1lz/zembryos (Fig. 2D), indicating that this is a
likely null allele. No difference in growth from weaning to 5 wk
of age was observed between Mthfd1l+/+and Mthfd1lz/+mice. To
determine the viability of homozygous null (Mthfd1lz/z) mice,
Mthfd1lz/+mice were intercrossed and the genotype distribution
was determined (Table S1). A total of 172 weanlings from 31
litters were examined. The average litter size was 5.5 pups. The
Mthfd1l genotypes were not distributed as expected for Mendelian
inheritance of the nonfunctional Mthfd1lzallele. The ratio of
Mthfd1l+/+to Mthfd1lz/+to Mthfd1lz/zwas 55:117:0, indicating
that the Mthfd1lz/zgenotype causes embryonic lethality (P =
2.0 × 10−11). If it is assumed that the Mthfd1lz/zgenotype is
lethal, Mthfd1l+/+and Mthfd1lz/+genotypes were observed in
the expected frequency (P = 0.75). Males and females were
found at the expected frequencies, and Mthfd1lz/+mice appear
healthy and breed normally.
Homozygous Deletion of Mthfd1l Results in DelayedEmbryonic Growth
and Defective Neural Tube Closure. Because we did not recover any
Mthfd1lz/zpups at birth, we sought to determine the embryonic
phenotype. Embryos were dissected from pregnant dams at E8.5–
E15.5, genotyped using yolk sac tissue, and their gross morphol-
ogy was examined. All observed Mthfd1lz/zembryos exhibited a
growth delay compared with WT and Mthfd1lz/+littermates. The
cytoplasmic and mitochondrial (m) compartments. Reactions 1, 2, and 3, 10-
formyl-THF synthetase, 5,10-methenyl-THF cyclohydrolase, and 5,10-methy-
lene-THF dehydrogenase, respectively, are catalyzed by trifunctional C1-THF
synthase (MTHFD1) in the cytoplasm. In mammalian mitochondria, reaction
1m is catalyzed by monofunctional MTHFD1L and reactions 2m and 3m by
bifunctional MTHFD2 or MTHFD2L. The other reactions are catalyzed by the
following: 4 and 4m, serine hydroxymethyltransferase (SHMT); 5, glycine
cleavage system (GCS); 6, 5,10-methylene-THF reductase; 7, methionine
synthase; 8, dimethylglycine dehydrogenase; 9, sarcosine dehydrogenase; and
10, thymidylate synthase. All reactions from choline to sarcosine are mitochon-
drial except the betaine to dimethylglycine conversion, which is cytoplasmic. Hcy,
homocysteine; AdoHcy, S-adenosylhomocysteine; AdoMet, S-adenosylmethio-
Mammalian one-carbon metabolism. Reactions 1–4 are in both the
transcripts. (A) The Mthfd1l gene, comprising 28 exons, encodes a protein
with two domains. The catalytic domain begins in exon 10. Mice in which
exon 5 was initially flanked by LoxP sites were crossed with the E2a-Cre line,
producing offspring with complete recombination at the LoxP sites (gray
arrowheads) and the resulting deletion of exon 5 and the neomycin phos-
photransferase gene (Neo). Transcription from the endogenous Mthfd1l
promoter produces a transcript containing exons 1–4 of Mthfd1l fused to an
internal ribosome entry site (IRES) and the β-galactosidase gene (LacZ) fol-
lowed by a polyadenylation site (pA). SA, splice acceptor site. (B) For geno-
typing, mouse genomic DNA was subjected to allele specific amplification
using a mixture containing F, R1 and R2 primers (Methods). A genomic
fragment of 444 bp was amplified from the WT allele and a 324-bp fragment
was amplified from the Mthfd1lzallele. (C) RT-PCR analysis of mRNA
expressed in whole E11.5 embryos using primers f in exon 4 and r in exon 14,
yielding a 1,087-bp product. M, Mthfd1L; G, glyceraldehyde-3-phosphate
dehydrogenase. (D) Immunoblot of mitochondria isolated from whole E11.5
embryos. Each lane was loaded with 7 μg of total mitochondrial protein.
Blots were probed with antibodies against MTHFD1L (M, 100 KDa) or the
mitochondrial matrix marker, Hsp60 (H, 60 KDa).
Mthfd1lz/zembryos do not produce detectable protein or full-length
| www.pnas.org/cgi/doi/10.1073/pnas.1211199110 Momb et al.
severity of the developmental delay was variable, but on average
the null embryos appeared to lag ∼0.75 d behind their littermates.
Some of the Mthfd1lz/zembryos died early during the gestational
period, but all that survived past the point of neural tube closure
(E9.5) exhibited aberrant neural tube phenotypes. Of 152
embryos dissected between E11.5–E12.5, we obtained 52
Mthfd1l+/+embryos, 74 Mthfd1lz/+embryos, and 26 Mthfd1lz/z
embryos, and we observed 28 resorptions. Of the 26 Mthfd1lz/z
embryos, 15 exhibited a clear NTD phenotype (exencephaly or
craniorachischisis) and 9 displayed a wavy neural tube phenotype
(Fig. 3 B–E). The remaining two Mthfd1lz/zembryos had com-
pletely open neural tubes but were not scored as having cra-
niorachischisis because these embryos had failed to turn,
suggesting that they may have been in the process of resorption.
The most common NTD phenotype was exencephaly with a wavy
neural tube (Fig. 3D; n = 11), or exencephaly alone (Fig. 3B, n =
3). The most severe NTD observed was craniorachischisis (Fig.
3C; n = 1). The nine Mthfd1lz/zembryos whose neural tubes had
closed all displayed a wavy neural tube with a small, aberrantly
formed head (Fig. 4B). In all, 20/24 Mthfd1lz/zembryos exhibited
a wavy neural tube, and the earliest observation of this phenotype
was at E9.5. The location of the waviness in the neural tube was
variable, but most embryos exhibited a wavy neural tube begin-
ning at approximately the same axis as the forelimb and extending
caudally past the forelimb, as depicted in Fig. 3 D and E. Because
many studies have noted an increased incidence of NTDs in
females (15), E11.5–E12.5 Mthfd1lz/zembryos were genotyped for
presence of the sex-receptor Y (SRY) locus. No bias was found
for either sex [females n = 6 (40%), males n = 9 (60%), P = 0.44].
In addition to aberrations in neural tube closure, we also
noted facial deformities in Mthfd1lz/zembryos that were most
apparent at the later stages. Compared with somite-matched WT
or heterozygous embryos, E12.5 Mthfd1lz/zembryos display im-
mature maxillary and mandibular processes (Fig. 3G). In
Mthfd1lz/zembryos, the maxillary processes of the first branchial
arch appear globular and more widely separated than in somite-
matched control embryos. In addition, the mandibular processes
are undergrown (Fig. 3 F and G; n = 7/7 embryos examined). No
surviving Mthfd1lz/zembryos were observed after E12.5, pre-
venting a later analysis of the phenotype.
Histological Analysis of Neural Tube Phenotypes. We sectioned
control (Mthfd1lz/+) and Mthfd1lz/zembryos stained for β-galac-
tosidase at E10.5 and E11.5. This allowed us to visualize re-
gionalized β-galactosidase activity, which should act as a reporter
for Mthfd1l transcription (Fig. 2A). To confirm that the LacZ
reporter recapitulated endogenous Mthfd1l gene expression, we
compared the patterns detected by β-galactosidase staining in
Mthfd1lz/+sections with endogenous Mthfd1l gene expression.
Using in situ hybridization, we observed transcript expression in
the ectoderm, underlying mesenchyme, and dorsal neural tube
(Fig. 4 C, F, and I). In the neural tube, the highest expression is
detected in the basal surface of the dorsal neuroepithelium (Fig.
4F, arrowheads). β-galactosidase activity was more restricted in
Mthfd1lz/+embryos (Fig. 4D), but is still seen within the same
region of the neural tube as the gene expression pattern. We
conclude that the LacZ reporter partially recapitulates endoge-
nous Mthfd1l gene expression with the difference most likely
being due to reduced sensitivity.
Mthfd1lz/+whole mount embryos stained for β-galactosidase
activity have the highest levels in the eyes, heart, limb, and dorsal
midline region (Fig. 4 A and J). In sectioned Mthfd1lz/+embryos,
β-galactosidase activity is predominantly detected at the basal
surface of the dorsal neuroepithelium and is rarely detected in
cells within and outside of the neural tube (Fig. 4 D, G, L, and
N). In sectioned Mthfd1lz/zembryos, β-galactosidase activity is
robustly detected both inside and outside the neural tube (Fig. 4
E, H, M, and O). Similar to Mthfd1lz/+embryos, β-galactosidase is
detected in the neuroepithelium; however, expression is less re-
stricted dorsally in the nulls. In Mthfd1lz/zembryos with closed
neural tubes, the morphology was abnormal throughout the neural
tube in all embryos sectioned (E10.5 and E11.5, n = 6). Neural
tubes had abnormally shaped lumens, including asymmetric dor-
sal-lateral bulges as well as a broader dorsal lumen that were not
seen in controls (Fig. 4 D, E, G, H, and L–O).
Dietary Supplementation with Sodium Formate. Because disruption
of Mthfd1l is expected to result in loss of mitochondrial formate
production, we sought to determine if maternal formate sup-
plementation would improve development of Mthfd1lz/zembryos.
Pregnant dams were given ad libitum access to water containing
sodium formate to achieve a calculated dose of 5,000 or 7,500 mg
sodium formate·kg−1·d−1(Methods); controls were given water
without formate. As before, we did not recover Mthfd1lz/zem-
bryos from unsupplemented dams (17 Mthfd1l+/+, 34 Mthfd1lz/+,
and 0 Mthfd1lz/zembryos, deviating significantly from the
expected Mendelian ratio; P = 0.0002). When dams were sup-
plemented with 5,000 mg·kg−1·d−1sodium formate, we obtained
4 Mthfd1l+/+, 14 Mthfd1lz/+, and 8 Mthfd1lz/zembryos from three
litters between E15.5–18.5. This genotype distribution does not
differ significantly from the expected Mendelian ratio (P = 0.50),
suggesting at least a partial rescue by formate. We next examined
the morphology of E10.5–E15.5 embryos from dams supple-
mented with 7,500 mg·kg−1·d−1sodium formate, obtaining 10
Mthfd1l+/+embryos, 31 Mthfd1lz/+embryos, and 14 Mthfd1lz/z
embryos from six litters, again conforming to the expected
Mendelian ratio (P = 0.48). Of the 14 Mthfd1lz/zembryos, 11
with WT E12.5 embryos (A), E12.5 Mthfd1lz/zembryos (B–E) exhibit a spec-
trum of neural tube defects including exencephaly (B, red arrowhead). (C)
Embryo with completely open neural folds (craniorachischisis) is indicated by
the dashed lines. Note that the embryo curves to the left so the entire open
neural tube is visible (red arrowheads). (D) Embryo with exencephaly (red
arrowhead) and a wavy neural tube. (Magnification: E, 3.2×.) (F) WT, 51-
somite embryo showing normal facial development. (G) Same Mthfd1lz/z
embryo (51 somites) imaged (D and E) displaying facial defects. The maxil-
lary processes (Ma) are globular and are broadly separated from the midline
(indicated by the frontonasal prominence, Fp). The mandibular processes
(Md) in the Mthfd1lz/zembryo appear undergrown. Embryos (D, E, and G)
are stained for β-galactosidase activity (blue). Embryos (A–D) are imaged at
the same magnification (1.25×). Embryos (F and G) are at the same magni-
Mthfd1lz/zembryos have neural tube and facial defects. Compared
Momb et al.PNAS
| January 8, 2013
| vol. 110
| no. 2
displayed normal neural tube closure and 3 had exencephaly
(Fig. 5 B and D). Thus, compared with nulls from unsupple-
mented dams, formate supplementation at 7,500 mg·kg−1·d−1
gives a significantly higher frequency of nulls with normal neural
tube closure (P = 0.028). Importantly, 6 of the 14 Mthfd1lz/z
embryos were dissected from dams at E13.5 or E15.5, whereas
no surviving Mthfd1lz/zembryos were observed after E12.5 from
unsupplemented dams. Although we could not compare sup-
plemented versus nonsupplemented embryos after E12.5 be-
cause of lethality of the unsupplemented embryos, formate
supplementation partially rescues the growth defect in Mthfd1lz/z
embryos (Fig. 5 A and B). The crown–rump length of formate
supplemented Mthfd1lz/zE11.5 embryos was significantly greater
than in unsupplemented Mthfd1lz/zembryos (5.0 ± 0.1 vs. 3.6 ± 0.3
mm, respectively; P < 0.01). Supplementation had no significant
effect on crown–rump length of WT embryos (5.9 ± 0.2 mm).
In this study, we have shown that all embryos lacking Mthfd1l
exhibit aberrant neural tube closure including craniorachischisis
and exencephaly and/or a wavy neural tube. The NTD phenotype
(exencephaly and craniorachischisis) is accompanied by abnor-
mal neural tube morphology characterized by asymmetric bulges
in the neuroepithelium and a wider lumen in wavy areas of
the neural tube. In addition to the NTD phenotype, Mthfd1lz/z
embryos show immature maxillary and mandibular process
development. Finally, we show that maternal formate supple-
mentation significantly reduces the incidence of NTDs, partially
rescues the growth defect, and allows survival past the point of
lethality seen in unsupplemented Mthfd1lz/zembryos. This
knockout mouse is a fully penetrant folate-pathway mouse model
that does not require feeding a folate-deficient diet to cause
these phenotypes. More than 10 folate-related mouse mutants
have been characterized thus far (16), but NTDs are observed in
only three: Folr1, Shmt1, and Amt. Folr1 encodes folate receptor
1, one of the major folate transport systems, and homozygous
knockout of Folr1 produces a severe folate deficiency in the
embryo that can be rescued with maternal 5-formyl-THF sup-
plementation (17). This rescue is “tunable,” and depending on
the dose of 5-formyl-THF administered to mothers during
gionalized β-galactosidase staining that ishighest intheeyes, heart, limb, anddorsal midline region(black arrowheads). (A)Dashed lineindicatesthe level ofall
sections below, at 10× and 40× magnification. (B and K) Mthfd1lz/zlittermates of embryos in (A) and (J), respectively, exhibit β-galactosidase staining that
appears superficially ubiquitous. (D, G, L, and N) Sections through Mthfd1lz/+embryos have β-galactosidase activity in the basal area of the dorsal neuro-
epithelium while Mthfd1lz/zembryos (E, H, M, and O) have a smaller neural tube, kinks in the lumen, and a broader dorsal lumen. (C, F, and I) Mthfdl1 gene
expression visualized by in situ hybridization in a whole mount embryo (C) and a section through the same embryo (F and I). Precise age of embryos: A, 37
somites; B, 34 somites; C, 33 somites; J, 44 somites; and K, 42 somites. The regions outlined by dashed boxes are magnified in the panels immediately below
them. (Magnification: A–C,1.6×; J and K, 1.25×.) Black arrowheads in D, E, F, L, and M highlight areas of β-galactosidase staining or in situ hybridization.
Aberrant neural tube morphology in Mthfd1lz/zembryos. (A and J) Mthfd1lz/+embryos at approximately E10.5 and E11.5, respectively, exhibit re-
velopment and growth in Mthfd1lz/zembryos. Pregnant dams were adminis-
tered a calculated dose of 7,500 mg·kg−1·d−1sodium formate. Mthfd1lz/z
embryos dissected at E11.5 (B) Improved growth compared with control em-
bryos from unsupplemented dams (A). E13.5 embryos from unsupplemented
(C) and supplemented (D) dams. Precise age of embryos: (A) WT, 47 somites; z/z,
36 somites; (B) WT 45 somites; and z/z, 41 somites; (D) not determined because
tails were not intact. (Scale bars, 1.0 mm.)
Maternal supplementation with sodium formate improves de-
| www.pnas.org/cgi/doi/10.1073/pnas.1211199110Momb et al.
deformities or can be rescued to birth. Homozygous knockout of
Shmt1, which encodes a cytoplasmic folate-metabolizing enzyme
(Fig. 1, reaction 4), gives rise to a low frequency of NTDs in
embryos from Shmt1−/−dams fed a folate-deficient diet (18, 19).
Amt encodes an aminomethyltransferase that is a subunit of the
mitochondrially localized glycine cleavage system (GCS), which
processes glycine to donate 1C units to THF, forming CH2-THF
(Fig. 1, reaction 5). Homozygous deletion of Amt is embryoni-
cally lethal, and Amt−/−embryos develop NTDs with 87% pen-
etrance (20). The phenotype exhibited by the Amt knockout
mouse suggests that the demand for 1C units is high during
neurulation. This is consistent with the observation that maternal
supplementation with methionine, which provides 1C units via
the methyl cycle (Fig. 1), significantly improves neurulation in
Amt null embryos, whereas folic acid has no effect (20). The
importance of the GCS to neural tube development is also
supported by the occurrence of NTDs in the nehe mouse (21).
The nehe mouse carries a hypomorphic allele of lipoic acid
synthetase, the mitochondrial enzyme that catalyzes the synthesis
of lipoic acid, an essential cofactor for GCS and several other
mitochondrial enzymes. One other folate pathway mouse model,
targeting Mthfd1 (Fig. 1, cytoplasmic reactions 1–3), also displays
disorganized neural tube closure in heterozygous Mthfd1gt/+
embryos from Mthfd1gt/+dams fed a folate-deficient diet (22).
Although the defects in this model are not classic NTDs, they are
reminiscent of the wavy neural tube phenotype we observe in
Our observation that Mthfd1lz/zembryos develop NTDs con-
firms that integrity of the mitochondrial 1C pathway is essential
for normal neural tube development. As illustrated in Fig. 1,
mitochondria possess multiple enzymes that produce CH2-THF
from various 1C donors (reactions 4m, 5, 8, and 9) (4). Embry-
onic mitochondria also possess redundant dehydrogenase/cyclo-
hydrolase enzymes that can oxidize CH2-THF to 10-CHO-THF
(MTHFD2 and MTHFD2L; reactions 2m and 3m). MTHFD2 is
expressed in transformed cells and embryonic or nondifferen-
tiated tissues, but not in adult differentiated tissues (23). Ho-
mozygous knockout of Mthfd2 is embryonic lethal, but does not
cause NTDs (24). Mthfd2 nullizygous embryos develop to about
E15.5 and display no gross developmental abnormalities, but
they are noticeably smaller and paler than WT and heterozygous
littermates. The lack of NTDs in Mthfd2 nullizygous embryos is
likely due to the existence of MTHFD2L, a second mitochondrial
dehydrogenase/cyclohydrolase. MTHFD2L is expressed in em-
bryos and adults (8) and can presumably support mitochondrial 1C
flux to the level of 10-formyl-THF in Mthfd2 nullizygous embryos.
On the other hand, only one enzyme with 10-formyl-THF syn-
thetase activity (MTHFD1L) is known to exist in mitochondria,
and this activity is required to produce formate and THF from 10-
CHO-THF (Fig. 1, reaction 1m). Thus, loss of MTHFD1L activity
is expected to completely abolish mitochondrial formate pro-
duction. This metabolic block would then starve the cytoplasm
for formate, creating a “formyl trap”, analogous to the methyl
trap seen when 5-methyl-THF accumulates in vitamin B12-de-
ficient cells (25, *). Although the Mthfd1l knockout mouse is
replete of folate, it has no way to catalyze the conversion of 10-
formyl-THF to THF plus formate, leading to a trapping of mi-
tochondrial 1C units as 10-formyl-THF, which cannot exit the
mitochondrion. This in turn would cause a deficiency in cyto-
plasmic 1C units, which are needed in stoichiometric amounts
for purine and thymidylate production as well as the methyl
cycle. The phenotypes of the Shmt1 and Amt knockout mice are
consistent with this model of mammalian 1C metabolism. The
Shmt1 gene, which encodes cytoplasmic serine hydroxymethyl-
transferase (Fig. 1, cytoplasmic reaction 4), is not essential in
embryos develop NTDs and orofacial
mice (26), indicating that mitochondrial SHMT (mitochondrial
reaction 4m) is fully capable of providing all of the 1C units
needed in both the embryo and the adult. The neural tube
phenotype in Amt nullizygous embryos lacking GCS activity
(mitochondrial reaction 5) (20) is less severe than that in
Mthfd1l knockout embryos. Presumably, the existence of alter-
native mitochondrial 1C donors (e.g., serine) allows some of the
Amt−/−embryos to develop normally. On the other hand, all
mitochondrial 1C units, whatever their source, must pass through
the MTHFD1L reaction to supply the cytoplasm with formate;
any defect in this step would be expected to cause a more severe
phenotype. The importance of mitochondrially derived formate is
supported by our results demonstrating a significant reduction in
the incidence of NTDs and partial rescue of Mthfd1lz/zembryonic
growth with maternal formate supplementation.
MTHFD1L thus controls the flux of 1C units from mito-
chondria into cytoplasmic processes such as purine and thymi-
dylate biosynthesis and the methyl cycle (Fig. 1). De novo purine
biosynthesis is essential for cell division, and S-adenosylmethio-
nine synthesis is critical for chromatin and DNA methylation,
which play essential roles during cell differentiation (27, 28) and
cell migration (29). Epigenetic modifications are particularly
dynamic, with extensive reprogramming of DNA methylation
during early embryogenesis (30). Disruption of the methyl cycle
is known to induce NTDs in cultured mouse embryos (31) and
cranial defects are observed in DNA methyltransferase 3b nul-
lizygous embryos (32). Recent studies in humans have linked
maternal folate status and occurrence of NTDs to tissue-specific
DNA hypo- and hypermethylation patterns, with decreased
DNA methylation observed in NTD brain tissue compared with
normal embryos (33). Protein methylation is also involved in
neural tube development, and hypomethylation of neural tube
proteins is accompanied by a failure of the neural tube to close in
rat embryos cultured in methionine-deficient conditions (34).
The cytoskeletal proteins β-actin and tubulin are known to be
methylated during neural tube closure (35), and proper function
of the cytoskeleton is required for cranial neural tube closure
(reviewed in ref. 36).
A common polymorphism in MTHFD1L has been shown to be
strongly associated with NTD risk in an Irish population (13),
suggesting that MTHFD1L also plays an important role in human
neural tube development. Importantly, disruption of MTHFD1L
function does not cause cellular folate deficiency (like a transport
defect); rather, it blocks a specific metabolic step: the production
and release of formate from mitochondria into the cytoplasm. This
metabolic defect causes aberrant neural tube closure including
craniorachischisis and exencephaly and/or a wavy neural tube
phenotype in 100% of Mthfd1lz/zembryos (Fig. 3). The Mthfd1l
mouse model should prove useful for identifying the mechanisms
that underlie the folate dependence of normal neural tube de-
velopment as well as for defining the metabolic pathways that are
most severely affected by cytoplasmic formate deficiency. De-
termination of these pathways will enhance our understanding of
folate-mediated 1C metabolism and may even suggest additional
interventions to protect against folate-resistant NTDs in humans.
Mice. All protocols used within this study were approved by the Institutional
Animal Care and Use Committee of The University of Texas at Austin and
conform to the National Institutes of Health Guide for the Care and Use of
Laboratory Animals. All mice were maintained on a C57BL/6 genetic back-
ground. Mice harboring a floxed conditional knockout cassette between
exons 4 and 6 of Mthfd1l were obtained from the Wellcome Trust Sanger
Institute (EUCOMM ID 37226). Mice carrying the floxed Mthfd1l allele were
mated to mice expressing Cre recombinase under control of the E2a pro-
moter (E2a-Cre) (14) to generate heterozygous Mthfd1lz/+embryos lacking
exon 5 and the neomycin resistance cassette (Fig. 2A). All mice were given ad
libitum access to water and standard mouse chow (LabDiet 5K67).
Genotyping. Genotyping was carried out by a modified PCR method (37). A
mixture of three primers was used to detect the WT and/or recombined al-
lele (Fig. 2A). To detect the WT allele, a forward primer (F) binding in the 5′
*Noronha JM, Silverman M (1962) On folic acid, vitamin B12, methionine and formimi-
noglutamic acid metabolism. Vitamin B12 and Intrinsic Factor, Second European Sympo-
sium, ed Heinrich HC (Verlag, Stuttgart), pp 728–736.
Momb et al.PNAS
| January 8, 2013
| vol. 110
| no. 2
region outside of the conditional cassette (5′-GAGTATGTGATTGCTTG- Download full-text
GACCCCCAGGTTCC-3′) and a reverse primer (R1) binding 5′ to exon 5 (5′-
TGGCTCCCGAGGTTGTCTTCTGGCTATGAT-3′) were used. Amplification using
these primers results in a 444-bp amplicon from the WT allele. Amplification
of the mutant allele uses the forward primer F and a reverse primer (R2)
complementary to a region only found in the gene-targeting cassette (5′-
CGGCGCCAGCCTGCTTTTTTGTACAAACTTG-3′). Amplification using these
primers results in a 324-bp amplicon in the presence of the mutant allele.
See Fig. 2A for a schematic of primer binding sites used to detect Mthfd1l
and Mthfd1lz. Cre recombinase was detected using 5′-GCATTACCGGTC-
GATGCAACGAGTGATGAG-3′ and 5′-GAGTGAACGAACCTGGTCGAAATCAG-
TGCG-3′ to produce a 408-bp amplicon, and SRY was detected using 5′-
TTGTCTAGAGAGCATGGAGGGCCATGTCAA-3′ and 5′-CCACTCCTCTGTGACA-
CTTTAGCCCTCCGA-3′ to detect a 273-bp amplicon.
RT-PCR. Total RNA was prepared from Mthfd1l+/+, Mthfd1lz/+, and Mthfd1lz/z
mouse embryos dissected at E11.5. First-strand cDNA was synthesized using
the SuperScript III First-Strand Synthesis System (Invitrogen) and random
hexamers. PCR was performed using a forward primer f (5-CTCACATCT-
GCTTGCCTCCA-3′) binding in exon 4 and a reverse primer r (5′-ATGTCCC-
CAGTCAGGTGAAG-3′) binding in exon 14 to amplify a 1,087-bp amplicon
from the WT transcript (see Fig. 2A for a schematic of primer binding sites).
Primers amplifying a 115-bp amplicon (forward: 5′-AGAGACGGCCGCAT-
CTTC-3′, reverse: 5′-CAAATGGCAGCCCTGGTGA-3′) from GAPDH were used
as a positive control for cDNA quality.
Mitochondrial Isolation and Immunoblotting. Mitochondria were isolated from
one embryo (Mthfd1l+/+and Mthfd1lz/+) or three embryos (Mthfd1lz/z) as
previously described (12), except embryos were homogenized by pipetting.
Proteins were separated by SDS/PAGE and immunoblotted using rabbit poly-
clonal anti-MTHFD1L (1:1,000) (11). After incubation with HRP-conjugated
goat anti-rabbit IgG(1:5,000)(Invitrogen),reactingbandswere detectedusing
ECL Plus (GE Healthcare Life Sciences). After stripping, blots were reprobed
with rabbit polyclonal anti-Hsp60 (1:1,000) (Enzo Life Sciences).
Histology. Embryos were stained for β-galactosidase activity overnight as
described previously (38). Stained embryos were embedded in paraffin,
sectioned at the level of the forelimb (4-μm thickness) and counterstained
with nuclear fast red. Mthfd1l whole mount in situ hybridization was per-
formed using a riboprobe against the 3′ UTR as previously described (12).
Embryos were then embedded in OCT medium and cryosectioned at the
level of the forelimb (12-μm thickness).
Maternal Supplementation with Sodium Formate. Mthfd1lz/+matings were set
up in a cage equipped with a water bottle containing either 0.37M or 0.55M
sodium formate. The females had access to the supplemented water at least
1 d before observation of the plug. These concentrations were calculated to
deliver either 5,000 or 7,500 mg sodium formate·kg−1·d−1, respectively,
based on an average water intake of 5 mL/d for a 25-g C57BL/6 mouse (39).
The effect of formate supplementation was analyzed by a two-sided χ2test
for NTD incidence and two-way ANOVA with Bonferroni posttest for crown–
ACKNOWLEDGMENTS. We thank Dr. Jacqueline Tabler for observational
insights. This work was supported by National Institutes of Health Grant
GM086856 (to D.R.A.) and by startup funds from the College of Natural
Sciences and the Institute for Cellular and Molecular Biology at the University
of Texas at Austin (to S.A.V.).
1. Hibbard ED, Smithells RW (1965) Folic acid metabolism and human embryopathy.
2. Ross ME (2010) Gene-environment interactions, folate metabolism and the embryonic
nervous system. Wiley Interdiscip Rev Syst Biol Med 2(4):471–480.
3. Hobbs CA, Shaw GM, Werler MM, Mosley B (2010) Folate status and birth defect risk.
Epidemiological perspective. Folate in Health and Disease, ed Bailey LB (CRC Press,
Taylor & Francis Group, Boca Raton, FL), 2nd ed, pp 133–153.
4. Tibbetts AS, Appling DR (2010) Compartmentalization of mammalian folate-medi-
ated one-carbon metabolism. Annu Rev Nutr 30:57–81.
5. Mejia NR, MacKenzie RE (1988) NAD-dependent methylenetetrahydrofolate de-
hydrogenase-methenyltetrahydrofolate cyclohydrolase in transformed cells is a mito-
chondrial enzyme. Biochem Biophys Res Commun 155(1):1–6.
6. Mejia NR, MacKenzie RE (1985) NAD-dependent methylenetetrahydrofolate de-
hydrogenase is expressed by immortal cells. J Biol Chem 260(27):14616–14620.
7. Barlowe CK, Appling DR (1988) In vitro evidence for the involvement of mitochondrial
folate metabolism in the supply of cytoplasmic one-carbon units. Biofactors 1(2):
8. Bolusani S, et al. (2011) Mammalian MTHFD2L encodes a mitochondrial methyl-
enetetrahydrofolate dehydrogenase isozyme expressed in adult tissues. J Biol Chem
9. Prasannan P, Pike S, Peng K, Shane B, Appling DR (2003) Human mitochondrial C1-
tetrahydrofolate synthase: Gene structure, tissue distribution of the mRNA, and im-
munolocalization in Chinese hamster ovary calls. J Biol Chem 278(44):43178–43187.
10. Walkup AS, Appling DR (2005) Enzymatic characterization of human mitochondrial
C1-tetrahydrofolate synthase. Arch Biochem Biophys 442(2):196–205.
11. Prasannan P, Appling DR (2009) Human mitochondrial C1-tetrahydrofolate synthase:
Submitochondrial localization of the full-length enzyme and characterization of
a short isoform. Arch Biochem Biophys 481(1):86–93.
12. Pike ST, Rajendra R, Artzt K, Appling DR (2010) Mitochondrial C1-tetrahydrofolate
synthase (MTHFD1L) supports the flow of mitochondrial one-carbon units into the
methyl cycle in embryos. J Biol Chem 285(7):4612–4620.
13. Parle-McDermott A, et al. (2009) A common variant in MTHFD1L is associated with
neural tube defects and mRNA splicing efficiency. Hum Mutat 30(12):1650–1656.
14. Lakso M, et al. (1996) Efficient in vivo manipulation of mouse genomic sequences at
the zygote stage. Proc Natl Acad Sci USA 93(12):5860–5865.
15. Harris MJ, Juriloff DM (2007) Mouse mutants with neural tube closure defects and
their role in understanding human neural tube defects. Birth Defects Res A Clin Mol
16. Harris MJ, Juriloff DM (2010) An update to the list of mouse mutants with neural tube
closure defects and advances toward a complete genetic perspective of neural tube
closure. Birth Defects Res A Clin Mol Teratol 88(8):653–669.
17. Spiegelstein O, et al. (2004) Embryonic development of folate binding protein-1
(Folbp1) knockout mice: Effects of the chemical form, dose, and timing of maternal
folate supplementation. Dev Dyn 231(1):221–231.
18. Beaudin AE, et al. (2011) Shmt1 and de novo thymidylate biosynthesis underlie folate-
responsive neural tube defects in mice. Am J Clin Nutr 93(4):789–798.
19. Beaudin AE, et al. (2012) Dietary folate, but not choline, modifies neural tube defect
risk in Shmt1 knockout mice. Am J Clin Nutr 95(1):109–114.
20. Narisawa A, et al. (2012) Mutations in genes encoding the glycine cleavage system pre-
dispose to neural tube defects in mice and humans. Hum Mol Genet 21(7):1496–1503.
21. Zhou X, Anderson KV (2010) Development of head organizer of the mouse embryo
depends on a high level of mitochondrial metabolism. Dev Biol 344(1):185–195.
22. Beaudin AE, Perry CA, Stabler SP, Allen RH, Stover PJ (2012) Maternal Mthfd1 dis-
ruption impairs fetal growth but does not cause neural tube defects in mice. Am J Clin
23. Christensen KE, Mackenzie RE (2008) Mitochondrial methylenetetrahydrofolate de-
hydrogenase, methenyltetrahydrofolate cyclohydrolase, and formyltetrahydrofolate
synthetases. Vitam Horm 79:393–410.
24. Di Pietro E, Sirois J, Tremblay ML, MacKenzie RE (2002) Mitochondrial NAD-
dependent methylenetetrahydrofolate dehydrogenase-methenyltetrahydrofolate cy-
clohydrolase is essential for embryonic development. Mol Cell Biol 22(12):4158–4166.
25. Herbert V, Zalusky R (1962) Interrelations of vitamin B12 and folic acid metabolism:
folic acid clearance studies. J Clin Invest 41:1263–1276.
26. MacFarlane AJ, et al. (2008) Cytoplasmic serine hydroxymethyltransferase regulates
the metabolic partitioning of methylenetetrahydrofolate but is not essential in mice.
J Biol Chem 283(38):25846–25853.
27. Bai S, et al. (2005) DNA methyltransferase 3b regulates nerve growth factor-induced dif-
ferentiation of PC12 cells by recruiting histone deacetylase 2. Mol Cell Biol 25(2):751–766.
28. Kobayakawa S, Miike K, Nakao M, Abe K (2007) Dynamic changes in the epigenomic
state and nuclear organization of differentiating mouse embryonic stem cells. Genes
29. Horswill MA, Narayan M, Warejcka DJ, Cirillo LA, Twining SS (2008) Epigenetic si-
lencing of maspin expression occurs early in the conversion of keratocytes to fibro-
blasts. Exp Eye Res 86(4):586–600.
30. Borgel J, et al. (2010) Targets and dynamics of promoter DNA methylation during
early mouse development. Nat Genet 42(12):1093–1100.
31. Dunlevy LP, et al. (2006) Integrity of the methylation cycle is essential for mammalian
neural tube closure. Birth Defects Res A Clin Mol Teratol 76(7):544–552.
32. Okano M, Bell DW, Haber DA, Li E (1999) DNA methyltransferases Dnmt3a and
Dnmt3b are essential for de novo methylation and mammalian development. Cell
33. Chang H, et al. (2011) Tissue-specific distribution of aberrant DNA methylation as-
sociated with maternal low-folate status in human neural tube defects. J Nutr Bio-
34. Coelho CN, Klein NW (1990) Methionine and neural tube closure in cultured rat
embryos: morphological and biochemical analyses. Teratology 42(4):437–451.
35. Moephuli SR, Klein NW, Baldwin MT, Krider HM (1997) Effects of methionine on the
cytoplasmic distribution of actin and tubulin during neural tube closure in rat em-
bryos. Proc Natl Acad Sci USA 94(2):543–548.
36. Copp AJ, Greene ND, Murdoch JN (2003) The genetic basis of mammalian neurula-
tion. Nat Rev Genet 4(10):784–793.
37. Stratman JL, Barnes WM, Simon TC (2003) Universal PCR genotyping assay that ach-
ieves single copy sensitivity with any primer pair. Transgenic Res 12(4):521–522.
38. Whiting J, et al. (1991) Multiple spatially specific enhancers are required to re-
construct the pattern of Hox-2.6 gene expression. Genes Dev 5(11):2048–2059.
39. Green EL, ed (1966) Biology of the Laboratory Mouse (Dover Publications, Inc., New
York), 2nd ed, p 706.
| www.pnas.org/cgi/doi/10.1073/pnas.1211199110Momb et al.