The syndrome called thiamine-responsive megaloblastic
anemia (TRMA) with diabetes and deafness (OMIM
249270) was first described by Rogers et al. (1). This auto-
somal recessive disorder with early childhood onset has
recently been localized to chromosome 1, band
1q23.23.3, in four large, geographically and ethnically
distinct kindreds (2). Data from several additional fam-
ilies have sharpened the TRMA interval to a 4-cM region
and provided evidence for locus homogeneity (3), sug-
gesting that a mutation in a single gene is responsible for
the triad of features that define this disease. More than
a dozen case reports have described the clinical manifes-
tations of the disorder in 31 patients (2). In all reported
cases, the anemia is corrected by pharmacologic doses of
thiamine, although macrocytosis (2) and defects in ery-
throid differentiation in vitro (4) persist after therapy.
The disorder represents a novel form of childhood
non–type I diabetes mellitus. No anti-insulin or anti-islet
antibodies have been found in patients, and in some
cases, pharmacologic doses of thiamine have reduced or
eliminated insulin requirements. Hearing loss is irre-
versible in all patients, although some investigators have
reported that thiamine therapy prevented further
progression of deafness (5). In addition to the definitive
clinical characteristics of TRMA, several patients have
had strokelike episodes and/or arrhythmias, and abnor-
malities of the retina and optic nerve (2).
Normal thiamine homeostasis involves the following
steps: (a) uptake of thiamine from the gut, (b) transport
to tissues and into cells, (c) conversion of thiamine into
the cofactor form thiamine pyrophosphate (TPP) by
thiamine pyrophosphokinase (TPK; EC 188.8.131.52), and (d)
binding of TPP to apoenzymes. Studies of human intes-
tinal thiamine absorption (6–8) suggest that two uptake
pathways exist: transport by a saturable, high-
affinity/low-capacity carrier, as well as passive intake by a
low-affinity/high-capacity system. Because thiamine con-
centrations in the intestinal lumen are low, Laforenza et
al. (8) propose that most thiamine is absorbed into the
body by way of the high-affinity transporter. A similar
pattern of thiamine uptake has also been seen in human
erythrocytes (5, 9, 10). Whether the erythroid and intes-
tinal thiamine uptake activities are identical is unknown.
After intracellular thiamine is converted to the active
cofactor TPP, TPP is incorporated into four known mam-
The Journal of Clinical Investigation|March 1999|Volume 103|Number 5
Defective high-affinity thiamine transporter leads
to cell death in thiamine-responsive megaloblastic
anemia syndrome fibroblasts
Amy R. Stagg,1Judith C. Fleming,1Meghan A. Baker,1
Massayuki Sakamoto,2Nadine Cohen,3and Ellis J. Neufeld1
1Division of Hematology/Oncology, and
2Department of Laboratory Medicine, Children’s Hospital, Dana Farber Cancer Institute,
and Harvard Medical School, Boston, Massachusetts 02115, USA
3Tamkin Molecular Genetics Facility, Technion Faculty of Medicine, Haifa 31096, Israel
Address correspondence to: Ellis J. Neufeld, Division of Hematology, Enders 720, 300 Longwood Avenue,
Boston, Massachusetts 02115, USA. Phone: (617) 355-8183; Fax: (617) 734-6791; E-mail: email@example.com
Amy R. Stagg and Judith C. Fleming contributed equally to this work.
Received for publication May 1, 1998, and accepted in revised form January 12, 1999.
We have investigated the cellular pathology of the syndrome called thiamine-responsive megaloblastic
anemia (TRMA) with diabetes and deafness. Cultured diploid fibroblasts were grown in thiamine-free
medium and dialyzed serum. Normal fibroblasts survived indefinitely without supplemental thiamine,
whereas patient cells died in 5–14 days (mean 9.5 days), and heterozygous cells survived for more than
30 days. TRMA fibroblasts were rescued from death with 10–30 nM thiamine (in the range of normal
plasma thiamine concentrations). Positive terminal deoxynucleotide transferase–mediated dUTP nick
end-labeling (TUNEL) staining suggested that cell death was due to apoptosis. We assessed cellular
uptake of [3H]thiamine at submicromolar concentrations. Normal fibroblasts exhibited saturable, high-
affinity thiamine uptake (Km400–550 nM; Vmax11 pmol/min/106cells) in addition to a low-affinity
unsaturable component. Mutant cells lacked detectable high-affinity uptake. At 30 nM thiamine, the
rate of uptake of thiamine by TRMA fibroblasts was 10-fold less than that of wild-type, and cells from
obligate heterozygotes had an intermediate phenotype. Transfection of TRMA fibroblasts with the yeast
thiamine transporter gene THI10 prevented cell death when cells were grown in the absence of
supplemental thiamine. We therefore propose that the primary abnormality in TRMA is absence of a
high-affinity thiamine transporter and that low intracellular thiamine concentrations in the mutant
cells cause biochemical abnormalities that lead to apoptotic cell death.
J. Clin. Invest. 103:723–729 (1999).
malian enzymes: the pentose phosphate shunt enzyme
transketolase and three multisubunit, multicofactor
enzyme complexes involved in oxidative decarboxylation
reactions — pyruvate dehydrogenase, α-ketoglutarate
dehydrogenase, and branched-chain keto-acid dehydro-
genase. The genes encoding the TPP-binding E1 subunits
of each of these enzymes have been localized by other
investigators and none map to chromosome 1q (2).
Several investigators have proposed a variety of puta-
tive defects of thiamine metabolism in patients with
TRMA. On the basis of low levels of α-ketoglutarate
dehydrogenase found in lymphocytes from a patient
with TRMA, Abboud et al. (11) proposed that a defective
TPP binding site on the enzyme leads to TRMA. Others
(12, 13) have cited mildly reduced levels of TPK activity
as a primary defect. Poggi et al. (5) first noted low TPP
content of TRMA erythrocytes (∼50% of controls) and
postulated that lack of a high-affinity thiamine trans-
porter may be associated with the syndrome (5, 9, 10,
14). The magnitude of the reported defect in thiamine
uptake of erythrocytes was subtle, ∼60% of normal.Addi-
tionally, these authors suggested that decreased TPK
activity contributes to TRMA, as they observed a
20%–30% decrease in TPK activity in TRMA erythrocytes
compared with controls.
No mammalian thiamine transporter has been isolated.
A thiamine transporter gene from Saccharomyces cerevisiae
has been identified (15, 16). The THI10 gene encodes a
protein with multiple membrane spanning domains,
which confers high-affinity thiamine transport in yeast.
In the present study, we demonstrate that fibroblasts
from patients with TRMA die in the absence of exoge-
nous thiamine. Using normal or mutant cells, we have
explored the thiamine uptake system and the ability of
the yeast transporter to correct the lethality in thiamine-
depleted cultures. This has enabled us to clarify previous
data and to test earlier hypotheses regarding possible
defects in TRMA. On the basis of these experiments, we
propose a model for the pathogenesis of TRMA in which
absence of a very-high-affinity thiamine transporter
causes a decrease in cell survival.
Patients. Samples were obtained from members of a large,
inbred Alaskan kindred reported previously (2). Fibroblasts
were obtained from skin biopsies of three patients, several
obligate heterozygotes (parents), and non–blood relatives
(wild-type). Informed consent was obtained from all subjects
and their parents.
Cell culture and cell lines. Diploid fibroblasts were grown from
skin samples in thiamine-replete α-MEM (1 mg/l thiamine;
GIBCO BRL, Grand Island, New York, USA), 15% FCS, 50 U/ml
penicillin, 300 µg/ml streptomycin sulfate (GIBCO BRL) in 5%
CO2 at 37°C. α-MEM minus thiamine (thi– medium) with 15%
dialyzed FCS (GIBCO BRL) was used for thiamine starvation.
Thiamine depletion was begun after cells were washed three
times in PBS and passaged into thi– medium. Experiments
shown were performed on two TRMA cell lines (first cousins),
two lines from obligate heterozygotes, and two normal controls.
Thiamine uptake assays. [3H]thiamine hydrochloride (specific
activity 15 Ci/mmol; American Radiolabeled Chemicals Inc., St.
Louis, Missouri, USA) was used for uptake assays. Thiamine
uptake was assayed in extracellular concentrations ranging
from 33 nM to 3 mM. Cells were washed with PBS, trypsinized,
and plated at a density of 1–2 ×105cells/35-mm well in 1 ml of
thi– medium. Uptake was measured in cells incubated at 37°C
in thi– medium without FCS. In unlabeled thiamine competi-
tion assays, thiamine-HCl (Sigma Chemical Co., St. Louis, Mis-
souri, USA) was added at the appropriate concentration. After
cell labeling, cells were washed three times with PBS and
harvested with 0.05% trypsin-EDTA (GIBCO BRL). The
amount of cell-associated [3H]thiamine was determined by
scintillation counting in 5 ml Ecolume scintillation fluid (ICN
Radiochemicals Inc., Costa Mesa, California, USA) using a
Beckman LS 3801instrument (Beckman Instrument Inc., Palo
Alto, California, USA). Incubation at 0°C was used as a control
for passive, cell-associated radioactivity.
Assay for α-ketoglutarate dehydrogenase. The method was adapt-
ed from a pyruvate dehydrogenase assay described by Hansen
(17). α-ketoglutarate dehydrogenase catalyzes the formation of
succinyl CoA from [1-14C]α-ketoglutarate, resulting in release
of 14CO2, which is trapped and measured. The mitochondrial
fraction (10,000-gpellet of postnuclear supernatant) was incu-
bated with 1 µCi [14C]α-ketoglutarate (Du Pont NEN Research
Products, Boston, Massachusetts, USA) diluted with nonla-
beled α-ketoglutarate to 1 mM, 1 mM coenzyme A (lithium
salt; Sigma Chemical Co.), 1.6 mM nicotinamide adenine
dinucleotide, 40 mM sodium phosphate (pH 7.4), 0.5 mM
dithiothreitol, and 5 mg/ml BSA, in the presence or absence of
1 mM TPP. The reaction was carried out in a 5-cc vial sealed
The Journal of Clinical Investigation|March 1999|Volume 103|Number 5
Reversible toxicity in TRMA–/–cells maintained in thiamine-depleted
medium. Fibroblasts from a patient with TRMA were grown for 9 days
without thiamine (a) or with 3 mM supplemental thiamine (b). Without
thiamine, all cells in thi– medium die by day 10 (c). Addition of 3 mM thi-
amine at day 9 rescues the remaining viable cells within 24 h (d). ×20 (a,
b, d); ×40 (c). Thi– medium, α-MEM minus thiamine; TRMA, thiamine-
responsive megaloblastic anemia.
Concentration of thiamine required to prevent toxicity in TRMA
0.3 nM 0.9 nM3 nM9 nM30 nM 300 nM900 nM
Cells were washed, trypsinized, and replated in varying thiamine concentrations and
then maintained for 30–40 days. Plus sign indicates cell survival, minus sign indi-
cates cell death, and plus over minus indicates morphologic changes without cell
death. Normal human plasma contains 18–32 nM free thiamine. Replete MEM con-
tains >3,000 nM. Results are from duplicate wells in three individual experiments.
TRMA, thiamine-responsive megaloblastic anemia.
with a serum cap, with a hanging basket containing a small
piece of filter paper soaked in 2.5 M NaOH, and was quenched
after 60 min with 3 M acetic acid for 1 h at 37°C. The filter
paper with adsorbed 14CO2was taken from the hanging vessel
and counted directly in fluor.The assay was linear with respect
to time and protein concentration.
Assay for conversion of thiamine to TPP. TRMA and normal
fibroblasts were plated at a density of 106cells/100-mm dish and
incubated in thi– medium. The following day, the medium was
removed and cells were incubated in thi– medium containing
2.4 µM [14C]thiamine (specific activity of 24 mCi/mmol; Amer-
sham International, Buckinghamshire, United Kingdom) at
37°C for 18 h. Cells were then washed three times with PBS on
ice, harvested with 0.05% trypsin-EDTA, centrifuged, and
washed two more times with PBS; an internal standard (unla-
beled thiamine, thiamine monophosphate, and TPP) was added;
and the protein was removed with 50 µl 5% TCA. The super-
natant was added to 1 ml of 20 mM Tris (pH 8.0). Conversion of
thiamine to TPP was analyzed by fast protein liquid chro-
matography (FPLC; Pharmacia Biotech Inc., Piscataway, New
Jersey, USA) on a Mono Q column, and fractions were collected
a rate of 1 ml/min with a set gradient from 0 to 1 M NaCl in 20
mM Tris (pH 8.0). Free thiamine ran in the void volume, and
TPP eluted with 0.35M NaCl and TMP at an intermediate posi-
tion. Internal standard peaks, followed by ultraviolet absorption
at 254 nm,showed similar recovery of all three compounds.
Apoptosis TUNEL assay. Apoptosis of TRMA fibroblasts was
evaluated by using an in situ cell detection kit (Boehringer
Mannheim Biochemicals, Indianapolis, Indiana, USA). Positive
cells were visualized using peroxidase substrate enhancer and
metal-enhanced DAB substrate (Boehringer Mannheim Bio-
chemicals). TRMA fibroblasts and normal controls were split
into thi– medium and grown on glass coverslips. Cells were
assayed on 4 consecutive days, beginning on the day when the
cell morphology was distinctly different in the mutants. A ter-
minal deoxynucleotide transferase–mediated dUTP nick end-
labeling (TUNEL) assay was performed as suggested by the
manufacturer, except that the ratio of terminal transferase
enzyme solution to label solution was 1:15. Cell were counter-
stained with 2% methyl green-pyronin.
Organic acid analysis.Mutant or control cells were plated in 35-
mm dishes and incubated overnight in thi– medium. The medi-
um was changed on day 0, and then medium was collected
from duplicate wells after 1 and 6 days. The cells from each well
were counted. Media were stored frozen before analysis. One-
milliliter samples of medium from six-well plates were treated
with hydroxylamine HCl to oximize the ketoacids, then acidi-
fied with HCl and extracted into ethyl acetate, followed by
extraction with diethyl ether. Samples were dehydrated with
Na2SO4, and the solvent was evaporated under N2. Organic
acids were derivatized with 100 µl BSTFA (Alltech Associates
Inc., Deerfield, Illinois, USA) and incubated for 10 min at 60°C.
Aliquots (1 µl) were injected for gas chromatography–mass
spectrometry analysis on a DB-1 column (dimethylpolysilox-
ane; Alltech Associates Inc.) in a Hewlett-Packard instrument
(Hewlett-Packard, Santa Clarita, California, USA). Peaks at
retention times corresponding to the expected compounds
were checked by mass spectroscopy against standard spectra
libraries. These compounds included lactate, pyruvate, 2-
ketoisovalerate, 2-keto-3-methylvalerate, 2-ketoisocaproate,
and 2-ketoglutarate. As an internal standard for each reaction,
10 µg of pentadecanoic acid was added to each medium sam-
ple before extraction and derivitization, to control for recovery.
Standards for authentic α-ketoacids were obtained from Sigma
Chemical Co. Results (peak heights) were compared by nor-
malizing to the size of the internal standard peak and correct-
ed for cell number in each dish.
Expression of yeast thiamine transporter gene in TRMA fibroblasts.
The yeast thiamine transporter gene THI10 was amplified
with Pfu DNA polymerase (Stratagene Inc., La Jolla, Califor-
nia, USA) from total yeast genomic DNA (Promega Corp.,
Madison, Wisconsin, USA) using primers flanking the open
reading frame YRL237w. The primers contained EcoRI
restriction sites to aid in subsequent cloning steps. Amplified
The Journal of Clinical Investigation| March 1999| Volume 103|Number 5
Uptake of [3H]thiamine by TRMA fibroblasts.
(a) Time course of uptake. Cells were washed
and placed in thi– medium 1 day before assay.
Uptake was measured for times shown, at
37°C, in serum-free medium supplemented
with 66 nM thiamine as described in Methods
(mutant, –/–; heterozygote, +/–; normal, +/+).
Means of duplicate wells are given. Values are
corrected for cell-associated radioactivity at
0°C. (b) Defective specific uptake in TRMA–/–
cells, assessed after 30 min incubation at 37°C.
Specific uptake was calculated by subtracting
radioactivity incorporated in the presence of 3
mM unlabeled thiamine. Means of duplicate
wells from one subject of each genotype are
represented. Error bars represent ranges for
Uptake and conversion of [14C]thiamine to TPP at steady state by nor-
mal and TRMA–/–fibroblasts
Total thiamineTPPConversion (%)
45 ± 6
58 ± 7
Cells were washed and plated in thi– medium overnight before incubation with 2.4
mM [14C]thiamine for 18 h. Cell-associated radioactivity was extracted in 5% TCA and
resolved by FPLC anion exchange chromatography as described in Methods. Deter-
minations are means ±SD of triplicate determinations. FPLC, fast protein liquid chro-
matography; thi– medium, α-MEM minus thiamine; TPP, thiamine pyrophosphate.
THI10 was cloned into the mammalian expression vector
pEF1α-neo and confirmed by restriction and sequence analy-
sis. Fibroblasts from patients with TRMA were cultured as
already described here. On the day before transfection, con-
fluent 100-mm tissue culture dishes were split 1:2 and plat-
ed in replete medium lacking antibiotics. Lipofectamine Plus
Reagent (4 mg; GIBCO BRL) was used for the transfection of
either pEF1 α-neo/THI10 or pEF1 α-neo alone into the
TRMA fibroblasts in serum-free medium. Three hours after
transfection, medium with serum was added so that the final
concentration of FCS was 15% and thiamine was 3 µM.
Transfected cells were maintained at 37°C and 5% CO2. The
medium was replaced 24 h after transfection with medium
containing 300 µg/ml G418 (GIBCO BRL) to select transfor-
mants. Medium was changed every 2–3 days. Three weeks
after transfection, G418-resistant cells were replated to be
50%–75% confluent and were maintained in medium with or
without thiamine at no more than 90% confluence.
Thiamine deprivation of TRMA fibroblasts. Fibroblasts from
patients with TRMA and from normal controls were
assessed for a thiamine-dependent phenotype by grow-
ing cells in thi– medium. After 5–14 days of thiamine
starvation, the morphology of TRMA fibroblasts began
to change (Fig. 1a). The mean time to abnormal mor-
phology was 9.5 days (±5 days, SD) in eight experiments.
Cells maintained in thiamine-replete medium remained
indistinguishable from wild-type (Fig. 1b). Within 1–3
days of morphological evidence of toxicity, cells in thi–
medium detached or died on the plates (Fig. 1c). Howev-
er, addition of thiamine (3 µM) to TRMA fibroblasts
when adverse changes first became evident rescued the
remaining cells (Fig. 1d). Rescued cells reverted to nor-
mal morphology within 24 hours and could then be
grown as before thiamine starvation. Normal fibroblasts
in thi– medium showed some decrease in growth rate
but did not die under these conditions (experiments ter-
minated at 40 days). Heterozygous cells died after 30 and
32 days on two occasions. These experiments suggest
that TRMA fibroblasts are more sensitive to thiamine
starvation and can be distinguished from normal fibrob-
lasts by their requirement for exogenous thiamine. We
postulate that normal cells survive because of the trace
amounts of vitamin (below fluorometric assay) remain-
ing in the dialyzed serum.
The concentration of thiamine necessary to sustain
TRMA fibroblasts was investigated by growing cells in
medium with thiamine concentrations ranging from 0
to 900 nM. Results are shown in Table 1. Nine nanomo-
lars of thiamine rescued mutant cells to a variable
degree, whereas 30 nM vitamin always rescued com-
pletely. This concentration is approximately that in
plasma and is 100-fold less than the amount in replete
tissue culture media (≥ 3 µM). We concluded that the
defect in TRMA cells might be observable only at
extremely low thiamine concentrations.
[3H]thiamine uptake in normal, heterozygote, and TRMA
fibroblasts. To understand the mechanism of TRMA fibrob-
last sensitivity to thiamine starvation, the cellular uptake
of thiamine was investigated at nanomolar concentrations.
[3H]thiamine of high specific activity (15 Ci/mmol,
∼15,000 cpm/pmol) was used to allow detection of uptake
at these low concentrations. Fig. 2adepicts the results of a
representative time course experiment. Fibroblasts were
incubated in 66 nM [3H]thiamine over a period of two
hours. Values were corrected for cell-associated counts in
cells incubated in an ice bath. These 0°C controls were
always <0.1 pmol/106cells. However, these results are not
corrected for nonspecific, or unsaturable, uptake. Uptake
was approximately linear for 30 minutes and approached
a plateau by approximately two hours in serum-free medi-
um. Subsequent assays were done at 30 minutes. As
demonstrated by this experiment, mutant fibroblasts show
a significant defect in thiamine uptake, and heterozygous
cells are intermediate in degree of uptake.
We next assessed uptake at varying concentrations of
thiamine in 30-minute assays. A substantial difference
between mutant and normal thiamine uptake was
observed at the lowest concentrations tested (Fig. 2b).
These results are corrected for nonspecific uptake by sub-
tracting counts obtained in the presence of 3 mM unla-
beled thiamine (50,000 to 100,000-fold excess). The
mutants exhibit ∼5% of the wild-type uptake at 33 and 66
nM thiamine under these conditions. This is consistent
with either a severe defect or absence of a high-affinity
transporter. Laforenza et al. (18) have observed a pH-sen-
sitive thiamine transport system in rat intestine that is
augmented by omeprazole. The fibroblast high-affinity
uptake system is apparently unaffected by omeprazole or
pH changes in the 6.5–8.0 range (data not shown).
Kinetic properties of high-affinity thiamine transport by fibrob-
lasts. Pilot studies suggested that thiamine uptake became
unsaturable at concentrations <1 µM. We therefore per-
formed detailed analysis of uptake by normal fibroblasts
The Journal of Clinical Investigation| March 1999|Volume 103|Number 5
Concentration dependence of high-affinity thiamine uptake in normal
fibroblasts.Cells were labeled with 66 nM [3H]thiamine diluted with unla-
beled thiamine for 30 min at 37°C. Values are corrected for nonspecific
uptake by subtracting counts associated with the presence of 500-fold
excess unlabeled compound. Results are shown for two separate experi-
ments (solid line with triangles and dotted line with circles), done on different
days with different wild-type cell lines. Inset: double reciprocal plot; the
800-nM point was eliminated as an outlier. All other data from the rec-
tangular plot were included. The least-squares regression line gives an
apparent Kmof 550 nM; apparent Vmaxfor uptake is 11 pmol/106cells/30
min. Each point is the mean of duplicate wells ± range (error bars).
in the nanomolar range as shown in Fig. 3. The proper-
ties of uptake by growing fibroblasts proved highly repro-
ducible from experiment to experiment and among cell
lines from the different normal and mutant cell lines test-
ed. We analyzed only specific, or saturable, uptake by sub-
tracting counts associated with cells in 500-fold excess of
unlabeled thiamine. Under these conditions, Lineweaver-
Burke analysis revealed a saturable process with apparent
Kmvalues from 400 to 550 nM, and a maximum trans-
port rate of 11 pmol/106cells/30 min. We were unable to
detect a saturable component in the TRMA fibroblasts in
several attempts, although a small amount of residual
transport activity could still be present. However, the
unsaturable component of thiamine uptake at thiamine
concentrations >1 µM was essentially the same in TRMA
as in wild-type cells (not shown).
Assays of other candidate enzymes for TRMA. We also
examined the function of reported candidate genes, α-
ketoglutarate dehydrogenase and TPK. We assayed α-
ketoglutarate dehydrogenase in mitochondrial pellets of
fibroblasts from two patients with TRMA or normal
controls, as described in Methods, by measuring the
coenzyme A–dependent release of 14CO2from [1-14C]α-
ketoglutarate. Results expressed as nmol CO2release/mg
protein/h (mean of duplicate determinations ± range)
were as follows: control, 250 (±42); TRMA patient no. 1,
144 (± 39); TRMA patient no. 2, 273 (± 13). The differ-
ences were not statistically significant and indicated that
this enzyme is not defective in the TRMA fibroblasts.
We next measured conversion of thiamine to TPP, as an
indirect measurement of TPK activity in vivo, in overnight
incubation of mutant or wild-type cells as described in
Methods. We found no defect in conversion of thiamine
to TPP. Indeed, the degree of conversion of radiolabeled
thiamine to TPP (Table 2) was consistently greater in the
mutants (58% ±7% vs. 45% ± 6%; mean ± SD, P = 0.043 by
two-tailed t test). We speculate that the difference is due
to smaller intracellular pools of unlabeled thiamine in
the mutant compared with the wild-type cells during
these experiments. Thiamine monophosphate was a
minor product in both wild-type and mutant cells.
Apoptosis assay. Once the phenotype of TRMA fibroblasts
in thi– medium became apparent, the mechanism of cell
death was investigated. Using the TUNEL assay for frag-
mented genomic DNA, TRMA and control fibroblasts
depleted of thiamine were assayed (Fig. 4). Cells were
assayed on day 10 of incubation in thi– medium. Both
normal and TRMA fibroblasts are TUNEL negative when
grown in thiamine-replete media (Fig. 4, a and b, respec-
tively). Normal fibroblasts remain negative for TUNEL
staining in thi– medium (Fig. 4c). However, as shown in
Fig. 4d, TRMA cells become TUNEL positive as the mor-
phology becomes abnormal. These results suggest that
TRMA fibroblasts die by programmed cell death. We were
unable to detect a DNA “laddering” pattern of apoptosis
in dishes of TRMA cells approaching cell death in thi–
medium. We postulate that only a small number of cells
undergo death at any given time. It is also possible that the
positive TUNEL staining represents a mechanism for
DNA fragmentation other than apoptosis.
Organic acid accumulation in medium of TRMA cells sub-
jected to thiamine starvation. We postulated that the thi-
amine uptake defect in TRMA cells should lead to more
rapid appearance of a thiamine-depleted biochemical
phenotype than wild-type cells. Table 3 shows the results
of organic acid analysis of medium from TRMA and con-
trol cells after one and six days of thiamine starvation.
Compared with thi– medium alone, both mutant and
wild-type medium show very little increase of the critical
organic acids on day 1. By day 6, however, all of these
compounds are significantly higher in the mutant cells.
This implies lower internal stores of TPP in the mutant
cells within a few days, despite the very high concentra-
tions of thiamine in standard medium. Although pyru-
vate is present in the medium as a supplement, it accu-
mulates severalfold in the mutant cells by day 6. Lactate
rose in both cell lines.
Rescue of thiamine-dependent lethality in TRMA fibroblasts
by S. cerevisiae THI10 gene. On the basis of the apparent
deficiency of a thiamine transporter in TRMA cells, we
tested the ability of S. cerevisiae THI10 gene (thiamine
transporter) to allow survival of mutant fibroblasts in
thiamine-depleted conditions. The THI10 gene was
cloned into expression vector pEF1 α-neo as described
in Methods and was stably transfected into TRMA cells.
As shown in Fig. 5, THI10-transfected cells, but not vec-
tor-only controls, were able to survive for more than six
weeks in thi– medium. Vector-only controls died after
13 days in thi– medium, and the experiment was termi-
nated after 44 days. These results strongly support the
hypothesis that the primary TRMA defect is a thiamine
transporter. These results do not prove that the trans-
port defect is across the plasma membrane (as opposed
to an internal membrane), but we suspect that this is the
case based on the uptake assays described.
We have demonstrated that TRMA fibroblasts are
uniquely sensitive to thiamine depletion. Cells die
after several days in thi– medium, with DNA fragmen-
The Journal of Clinical Investigation|March 1999|Volume 103| Number 5
Organic acid analysis of medium from TRMA and control fibroblasts
in the thi– medium
ConditionsMedium aloneDay 1Day 6
4.9 × 1052.1 × 1054.1 × 1057.0 × 105
Control Mutant Control
Cell number per dish
at harvest day
Organic acid content (integrator units per million cells)
Media were collected from wild-type and mutant cells on days 1 and 6 after switch-
ing to thiamine-depleted conditions and were analyzed by gas chromatogra-
phy–mass spectrometry analysis as described in Methods. Thi– medium (including
dialyzed serum) was also analyzed “medium alone.” Cells from duplicate wells were
counted on the day of harvest, and an internal standard was added to each sample
before extraction and derivatization to control for recovery.
tation as detected by TUNEL assay consistent with
apoptosis or programmed cell death. To our knowl-
edge, this is the first report of a genetic disorder of
intermediary metabolism causing apoptosis. Labeling
studies using [3H]thiamine of high specific activity
demonstrate that TRMA fibroblasts take up ∼5%–10%
of wild-type amounts of thiamine at nanomolar con-
centrations. Nonspecific (unsaturable) thiamine
uptake is not impaired in the mutants. Thus, in stan-
dard tissue culture medium (3 µM thiamine plus the
rich supply in serum), the metabolic defect is
obscured, whereas in thi– medium, characteristic
organic acids accumulate faster in TRMA cells than in
normal controls. Our results support the hypothesis
that the primary defect in TRMA is a recessive muta-
tion in the gene encoding a high-affinity thiamine
transporter. Rescue of the lethal phenotype by a
known yeast transporter provides additional support
for this hypothesis.
Rindi and colleagues (9, 10) have previously defined a
subtle defect of thiamine uptake in erythrocytes of
patients with TRMA, ∼60% of control.There are at least
two possible explanations for the differences observed
between their studies and the present work, in which we
have found greater than 10-fold reduction in [3H]thi-
amine uptake at low concentrations. First, the erythro-
cyte and fibroblast high-affinity systems might be
similar or identical, but experimental variables allow the
defect to be more readily discerned in the present
fibroblast studies. For example, we have used thiamine
of specific activity 20-fold higher than that used in pre-
vious studies (15 vs. 0.75 Ci/mmol; refs. 9, 10). This
facilitates assays at low substrate concentrations, at
which differences between mutant and normal cells are
greatest (Fig. 2). The Km of the high-affinity system
described here, ∼0.5 µΜ, is at the lower end of concen-
trations studied by Rindi et al. (9). Second, there may be
more than one specific thiamine uptake system, oper-
ating together or individually in different cell types. If
erythrocytes normally have two systems, and fibroblasts
only one, the degree of abnormality would appear more
severe in the fibroblasts. The current data do not dis-
tinguish these possibilities.
Cell death by apparent apoptosis in severely thiamine-
depleted TRMA fibroblasts raises several questions
about the pathogenesis of the disease. First, why are
patients with TRMA not severely ill without pharmaco-
logic thiamine therapy? With a single exception (19),
subjects with TRMA have not had symptoms suggestive
of severe tissue thiamine deficiency, and they have not
had organic aciduria (5, 11). Red blood cell transketolase
is also normal. We find that 10–30 nM thiamine is
enough to rescue fibroblasts (Table 1) and that thiamine
depletion does lead to organic acidosis in the cells from
patients after a few days of thiamine depletion (Table 3).
Normal plasma levels of free thiamine are on the order
of 18–33 nM (5, 13). This implies that patients with
TRMA on an adequate diet without pharmacologic
supplementation are metabolically stable, with thiamine
uptake solely by the low-affinity system.
Second, why is the deafness in TRMA progressive,
whereas the anemia, and to some degree the diabetes, are
reversible? We postulate that the thiamine requirement
of cochlea or acoustic nerve cells is substantially higher
than that of fibroblasts, which have very low energy
requirements compared with neurons, myocytes, and
other cell types. (We have not examined this phenome-
non in other TRMA cell types.) If so, even pharmacolog-
ic doses of enteral thiamine may not prevent a thiamine-
deficient phenotype in cell types with high energy usage
at all times. On the basis of our observations in fibrob-
lasts, we propose that this would lead to occasional cell
death in the most sensitive tissues. This, in turn, may
result in optic atrophy and progressive deafness. Bone
The Journal of Clinical Investigation| March 1999| Volume 103| Number 5
Apoptosis of fibroblasts subjected to thi– medium. TUNEL stain was per-
formed as described in Methods. Normal (aand b) or TRMA mutant cells
(c and d) were incubated for 10 days in the presence (a and c) or absence
(b and d) of thiamine. (a) Wild-type cells, thiamine-replete. (b) Wild-type
cells, thiamine-depleted. (c) TRMA mutant cells with thiamine. (d) TRMA
fibroblasts, thiamine-depleted. ×40. TUNEL, deoxynucleotide trans-
ferase–mediated dUTP nick end-labeling.
Complementation of TRMA fibroblasts defects with yeast thiamine trans-
porter gene THI10. Cells transfected with the THI10 gene and vector
alone were grown in the presence or absence of thiamine as described in
Methods. TRMA fibroblasts transfected with vector alone and grown in
thi– medium died after 13 days. The experiment was terminated after 44
days. (a) TRMA fibroblasts transfected with THI10 gene, thiamine-
replete. (b) TRMA fibroblasts transfected with vector alone, thiamine-
replete. (c) TRMA fibroblasts transfected with THI10 gene, thi– medium.
(d) TRMA fibroblasts transfected with vector alone, thi– medium.
marrow, in contrast, may be less sensitive to this effect
for two reasons. First, as long as the pluripotent
hematopoietic stem cells are not sensitive to cell death,
the erythroid lineage can be repopulated when patients
are thiamine replete. Persistent macrocytosis (2) and
poor erythroid colony growth in vitro (4) suggest that
even thiamine-replete erythroid progenitors are some-
what sensitive to the thiamine transport defect. Diabetes
in TRMA patients has been variably reversible with thi-
amine. It is reasonable to conclude that the beta cells of
the pancreas, or the target tissues, are of intermediate
sensitivity to cellular thiamine deficiency. Other tissues,
notably liver and muscles, which would be responsible
for organic acidemia in severe thiamine deficiency, must
be resistant to cell toxicity in the ambient thiamine con-
centration found in untreated patients with TRMA.
It is conceivable that deafness in patients with TRMA
could be prevented by parenteral, early (even prenatal)
thiamine therapy. Laforenza and colleagues (8) have
recently reported a thiamine uptake system from human
intestinal biopsies (maximum rate in duodenum but
present throughout the gut), with apparent Km∼4 µM.
Whether this system is due to the same gene product
defective in TRMA remains to be determined. The trans-
port across the intestinal epithelium into the abluminal
space was not measured. If either of these processes
depends on the TRMA gene, it is possible that even phar-
macologic doses of oral thiamine cannot provide supra-
normal levels of the vitamin in the blood and tissues. For
cells with high energy requirements, this may not be suf-
ficient in TRMA. We propose that if transintestinal
transport is also defective in TRMA, patients may do
substantially better if treated from birth (or even prena-
tally) with parenteral thiamine.
It is not yet clear how the vitamin transport defect in
TRMA is linked to programmed cell death. Matsushima
et al. (20) recently reported apoptotic cell death in the thal-
amus of thiamine-deficient rats. This suggests that the
phenomenon is not limited to TRMA. However, it does
not clarify the relationship between cellular thiamine defi-
ciency and activation of apoptosis pathways. Fibroblasts
in thi– medium release organic acids expected from
defects in the TPP requiring enzymes, i.e., branched chain
α-ketoacids and pyruvate (Table 3). This suggests that thi-
amine deprivation can lead to dysfunction in the Krebs
cycle, shutting down mitochondrial energy production.
We speculate that TRMA cells in thi– medium will be rel-
atively more impaired at these steps than normal fibrob-
lasts. Mitochondrial changes, in turn, may play a pivotal
role in activation of apoptosis. A number of possible
mechanisms have been proposed (reviewed in ref. 21).
Mitochondrial dysfunction might lead to leakage of cas-
pase activators, cytochrome C, or apoptosis-inducing fac-
tor, for example. It will be of interest in the future to exam-
ine whether TRMA cells from other kindreds exhibit the
same uptake kinetics and thiamine-dependent survival as
cells from the Alaskan patients tested here. Independent
mutations have apparently arisen on different haplotypes
in each TRMA family (2, 3), and it is possible that differ-
ences in phenotype will correlate with different genotypes.
In light of our present results, we believe that a candidate
gene for TRMA would be a thiamine transporter, respon-
sible for vitamin uptake at low extracellular thiamine con-
centration. Database homology searches have not yielded
such a gene. Isolation of the TRMA gene will facilitate fur-
ther studies of this unique biochemical disorder.
We express our gratitude to the patients and their families and
to Donna Fenske (State of Alaska Division of Public Health,
Homer, Alaska, USA). Laura Noriega assisted with thiamine-
depletion experiments. We thank Mark Fleming for advice and
thoughtful comments. This work was supported by National
Institutes of Health grants HL-49196 and HL-07574. Initial
portions of this work were supported by a grant from the
Charles H. Hood Foundation (to E.J. Neufeld).
1.Rogers, L.E., Porter, F.S., and Sidbury, J.B.J. 1969. Thiamine-responsive
megaloblastic anemia. J. Pediatr. 74:494–504.
2.Neufeld, E.J., et al. 1997. Localization of the gene for thiamine-respon-
sive megaloblastic anemia syndrome on the long arm of chromosome 1
by homozygosity mapping. Am. J. Hum. Genet. 61:1335–1341.
3.Raz, T., et al. 1998. Refined mapping of the gene for thiamine-responsive
megaloblastic anemia syndrome in a 4 cM region and evidence for genet-
ic homogeneity. Hum. Genet. 103:455–461.
4.Rotoli, B., Poggi, V., De Renzo, A., and Robledo, R. 1986. In vitroaddition
of thiamin does not restore BFU-E growth in thiamin-responsive ane-
mia syndrome. Haematologica. 71:441–443.
5.Poggi, V., et al. 1984. Thiamin-responsive megaloblastic anaemia: a dis-
order of thiamin transport? J. Inherit. Metab. Dis. 7:153–154.
6.Rindi, G., and Ferrari, G. 1977. Thiamine transport by human intestine
in vitro. Experientia. 33:211–213.
7.Hayumpa, A.M., Jr., et al.1982. Dual system of intestinal thiamine trans-
port in humans. J. Lab. Clin. Med. 99:701–708.
8.Laforenza, U., et al. 1997. Thiamine uptake in human intestinal biopsy
specimens, including observations from a patient with acute thiamine
deficiency. Am. J. Clin. Nutr. 66:320–326.
9.Rindi, G., et al. 1992. Thiamine transport by erythrocytes and ghosts in thi-
amine-responsive megaloblastic anaemia. J. Inherit. Metab. Dis.15:231–242.
10.Rindi, G., et al. 1994. Further studies on erythrocyte thiamin transport
and phosphorylation in seven patients with thiamin-responsive mega-
loblastic anaemia. J. Inherit. Metab. Dis. 17:667–677.
11.Abboud, M.R., Alexander, D., and Najjar, S.S. 1985. Diabetes mellitus,
thiamine-dependent megaloblastic anemia, and sensorineural deafness
associated with deficient alpha-ketoglutarate dehydrogenase activity. J.
12.Borgna-Pignatti, C., Marradi, P., Pinelli, L., Monetti, N., and Patrini, C.
1989. Thiamine-responsive anemia in DIDMOAD syndrome. J. Pediatr.
13.Grill, J., et al. 1991. Thiamine responsive anemia: report of a new case
associated with a thiamine pyrophosphokinase deficiency. Nouv. Rev. Fr.
14.Poggi, V., et al. 1989. Studies on thiamine metabolism in thiamine-
responsive megaloblastic anaemia. Eur. J. Pediatr. 148:307–311.
15.Enjo, F., Nosaka, K., Ogata, M., Iwashima, A., and Nishimura, H. 1997.
Isolation and characterization of a thiamin transport gene, THI10, from
Saccharomyces cerevisiae. J. Biol. Chem. 272:19165–19170.
16.Singleton, C.K. 1997. Identification and characterization of the thiamine
transporter gene of Saccharomyces cerevisiae. Gene. 199:111–121.
17.Hansen, T.L. 1982. Determination of pyruvate dehydrogenase in cultured
human fibroblasts and amniotic fluid cells. Clin. Chim. Acta. 123:45–50.
18.Laforenza, U., Orsenigo, M.N., and Rindi, G. 1998. A thiamine/H+
antiport mechanism for thiamine entry into brush border membrane
vesicles from rat small intestine. J. Membr. Biol. 161:151–161.
19.Mandel, H., Berant, M., Hazani, A., and Naveh, Y. 1984. Thiamine-
dependent beriberi in the “thiamine-responsive anemia syndrome.” N.
Engl. J. Med. 311:836–838.
20.Matsushima, K., MacManus, J.P., and Hakim, A.M. 1997. Apoptosis is
restricted to the thalamus in thiamine-deficient rats. Neuroreport.
21.Mignotte, B., and Vayssiere, J.L. 1998. Mitochondria and apoptosis. Eur.
J. Biochem. 252:1–15.
The Journal of Clinical Investigation|March 1999| Volume 103|Number 5