Báo cáo khoa học: Thiamin diphosphate in biological chemistry: new aspects of thiamin metabolism, especially triphosphate derivatives acting other than as cofactors pdf

MINIREVIEW
Thiamin diphosphate in biological chemistry: new aspects
of thiamin metabolism, especially triphosphate derivatives
acting other than as cofactors
Lucien Bettendorff and Pierre Wins
GIGA-Neurosciences, University of Lie
`
ge, Belgium
Thiamin is best known for the cofactor role of its
diphosphorylated derivative thiamin diphosphate
(ThDP; Fig. 1) in many enzymes and multienzyme
complexes [1]. The mechanism by which the thiamin
moiety of ThDP exerts its coenzyme function by
proton substitution on position 2 of the thiazolium
ring was elucidated by Ronald Breslow in 1958 [2] and
nowadays, a lot of work is devoted to understanding
the interplay between ThDP and ThDP-dependent
enzymes in catalysis [3,4].
However, thiamin phosphate derivatives other than
ThDP exist in most organisms. In microorganisms and
Keywords
adenosine thiamin triphosphate; adenylate
kinase; alarmone; Escherichia coli;
regulation; riboswitch; thiamin transport;
thiamin triphosphatase; thiamin
triphosphate; triphosphate tunnel
metalloenzymes
Correspondence
L. Bettendorff, GIGA-Neurosciences,
University of Lie
`

ge, Ba
ˆ
t. B36, Tour de
Pathologie 2, e
´
tage +1, Avenue de l’Ho
ˆ
pital,
1, B-4000 Lie
`
ge 1 (Sart Tilman), Belgium
Fax: + 32 4 366 59 53
Tel: +32 4 366 59 67
E-mail:
(Received 9 October 2008, revised 26
February 2009, accepted 12 March 2009)
doi:10.1111/j.1742-4658.2009.07019.x
Prokaryotes, yeasts and plants synthesize thiamin (vitamin B1) via complex
pathways. Animal cells capture the vitamin through specific high-affinity
transporters essential for internal thiamin homeostasis. Inside the cells, thi-
amin is phosphorylated to higher phosphate derivatives. Thiamin diphos-
phate (ThDP) is the best-known thiamin compound because of its role as
an enzymatic cofactor. However, in addition to ThDP, at least three other
thiamin phosphates occur naturally in most cells: thiamin monophosphate,
thiamin triphosphate (ThTP) and the recently discovered adenosine thiamin
triphosphate. It has been suggested that ThTP has a specific neurophysio-
logical role, but recent data favor a much more basic metabolic function.
During amino acid starvation, Escherichia coli accumulate ThTP, possibly
acting as a signal involved in the adaptation of the bacteria to changing
nutritional conditions. In animal cells, ThTP can phosphorylate some pro-

teins, but the physiological significance of this mechanism remains
unknown. Adenosine thiamin triphosphate, recently discovered in E. coli,
accumulates during carbon starvation and might act as an alarmone.
Among the proteins involved in thiamin metabolism, thiamin transporters,
thiamin pyrophosphokinase and a soluble 25-kDa thiamin triphosphatase
have been characterized at the molecular level, in contrast to thiamin
mono- and diphosphatases whose specificities remain to be proven. A solu-
ble enzyme catalyzing the synthesis of adenosine thiamin triphosphate from
ThDP and ADP or ATP has been partially characterized in E. coli , but the
mechanism of ThTP synthesis remains elusive. The data reviewed here
illustrate the complexity of thiamin biochemistry, which is not restricted to
the cofactor role of ThDP.
Abbreviations
ABC, ATB-binding cassette; AK, adenylate kinase; AThTP, adenosine thiamin triphosphate; TenA, thiaminase II; ThDP, thiamin diphosphate;
ThDPase, thiamin diphosphatase; ThMP, thiamin monophosphate; ThMPase, thiamin monophosphatase; ThTP, thiamin triphosphate;
ThTPase, thiamin triphosphatase; TPK, thiamin pyrophosphokinase; TTM, triphosphate tunnel metalloenzyme.
FEBS Journal 276 (2009) 2917–2925 ª 2009 The Authors Journal compilation ª 2009 FEBS 2917
plants, thiamin monophosphate (ThMP; Fig. 1) is
formed during the final step of thiamin biosynthesis,
but ThMP also exists in animal cells unable to carry
out de novo synthesis of thiamin. In animal tissues,
ThMP is a product of the enzymatic hydrolysis of
ThDP and has no known physiological function. Thia-
min triphosphate (ThTP; Fig. 1) was first suggested to
exist in eukaryotic cells in the 1950s, but progress in
the field was hampered by the lack of analytical tools
available to measure the tiny amounts of ThTP occur-
ring in most cells. This difficulty was overcome by the
advent of HPLC techniques in the late 1970s and early
1980s [5]. Most of these techniques rely on the precol-

umn derivatization of thiamin to highly fluorescent
thiochrome derivatives, thereby increasing sensitivity
and selectivity. Subsequent studies showed that ThTP
is present in most cells studied to date, from bacteria
to mammals [6]. Finally, the complexity of thiamin
metabolism was further highlighted by the discovery of
a hitherto unsuspected derivative: adenosine thiamin
triphosphate (AThTP; Fig. 1) [7].
It is generally assumed that the pathologies linked
to thiamin deficiency, mainly beriberi and Wernicke–
Korsakoff syndrome, are the consequence of
decreased ThDP levels, resulting in reduced activity
of key enzymes for oxidative metabolism such as
2-oxoglutarate dehydrogenase [8,9]. Although oxida-
tive metabolism is obviously very important for neu-
ronal survival, general impairment of oxidative
decarboxylation reactions does not explain the selec-
tive vulnerability of certain brain areas to thiamin
deficiency, for example the periventricular thalamic
regions and mammillary bodies in Wernicke–Korsak-
off syndrome. Likewise, it is difficult to link the
decreased activity of ThDP-dependent enzymes to, for
example, the observation that thiamin deficiency exac-
erbates plaque pathology in a mouse model of Alzhei-
mer’s disease [10]. Therefore, it is important to better
understand the role of all thiamin derivatives and to
characterize the enzymes involved in the metabolism
of thiamin phosphate derivatives. Here, we focus on
some recent developments in the field, such as
thiamin biosynthesis, transport, thiamin triphosphate

metabolism and the discovery of the new adenosine
thiamin nucleotide.
Thiamin biosynthesis and salvage
De novo thiamin biosynthesis may occur in bacteria,
some protozoa, plants and fungi [11,12]. The pathways
Fig. 1. Scheme depicting the enzymatic interconversions of thiamin derivatives in mammalian cells. Free ThDP represents the high turnover
ThDP pool, precursor of ThMP, ThTP and AThTP. This ‘rapid’ pool plays a pivotal role in the metabolism of phosphorylated thiamin deriva-
tives in eukaryotic cells. The bound ThDP represents the low turnover cofactor ThDP pool, with ThDP mostly bound to apoenzymes. 1,
Thiamin pyrophosphokinase; 2, thiamin diphosphatase; 3, thiamin monophosphatase; 4, ThTP synthase (unknown mechanism); 5,
membrane-associated and soluble thiamin triphosphatases; 6, thiamin diphosphate-adenylyl transferase; 7, adenosine thiamin triphosphate
hydrolase (postulated). Among all these enzymes, only thiamin pyrophosphokinase (1) and 25-kDa soluble thiamin triphosphatase (5) have
been characterized at the molecular level. All the other conversions occur in intact cells or in cellular extracts but the enzymes have not yet
been characterized and the genes identified (adapted and updated from Bettendorff [43]).
Thiamin metabolism and thiamin phosphates L. Bettendorff and P. Wins
2918 FEBS Journal 276 (2009) 2917–2925 ª 2009 The Authors Journal compilation ª 2009 FEBS
are complex, in particular because of the thiazole moi-
ety, a heterocycle rarely encountered in natural prod-
ucts. In all cases, the thiazole and pyrimidine moieties
are synthesized separately and then assembled to form
ThMP by thiamin-phosphate synthase (EC 2.5.1.3). The
exact biosynthetic pathways may differ among organ-
isms. In Escherichia coli and other Enterobacteriaceae,
ThMP may be phosphorylated to the cofactor ThDP
by a thiamin-phosphate kinase (ThMP + ATP «
ThDP + ADP; EC 2.7.4.16). In most bacteria and in
eukaryotes, ThMP is hydrolyzed to thiamin + P
i
by a
thiamin monophosphatase (ThMPase). Thiamin may
then be pyrophosphorylated to ThDP by thiamin

diphosphokinase (thiamin + ATP « ThDP + AMP;
EC 2.7.6.2).
Thiamin can be degraded by thiaminases [13].
Thiaminase I (EC 2.5.1.2), a pyrimidine transferase
able to use various acceptors, is found in shellfish,
the viscera of some freshwater fish, fern species (Pter-
idium aquilinum) and some microorganisms (Bacillus
thiaminolyticus). The physiological significance of this
enzyme is not known, but it is responsible for animal
(grazing ruminants or horses in pastures) as well as
human poisoning (reliance on shellfish or fish as
main food). By contrast, thiaminase II (TenA;
EC 3.5.99.2) is a hydrolase that cleaves thiamin in its
thiazole and pyrimidine moieties. It is found in some
microorganisms (Bacillus subtilis, for example) and its
significance has recently been elucidated [14]. Indeed,
TenA is involved in a salvage pathway recycling thia-
min-degradation products, such as formylamino-
pyrimidine formed in the soil, to aminopyrimidine
and hydroxypyrimidine, a building block for the bio-
synthesis of ThMP. Hydrolysis of aminopyrimidine to
hydroxypyrimidine by TenA is 100 times faster than
the hydrolysis of thiamin, and thiamin phosphate
esters (representing nearly all the intracellular thia-
min) are not hydrolyzed by TenA. This recycling
does not seem to be limited to bacteria, but could
also take place in yeast [15]. Thus, the thiaminase
activity of TenA probably has no physiological rele-
vance [14].
Thiamin biosynthesis is regulated by so-called ribos-

witches [16], consisting of ThDP-sensing noncoding
mRNA elements present in some mRNAs coding for
enzymes involved in the thiamin biosynthetic pathways
[17,18]. When plenty of ThDP is present, binding to
the riboswitch induces a conformational change in the
mRNA, sequestering the ribosome-binding site and
preventing protein synthesis. This specific feedback
mechanism is an alternative to the allosteric modula-
tion of rate-limiting enzymes of metabolic pathways by
metabolic end products.
Thiamin transport into prokaryotic and
eukaryotic cells
In bacteria, thiamin uptake occurs through ATP-bind-
ing cassette (ABC)-transporters, initiated with the
binding of thiamin or one of its phosphate derivatives
to the periplasmic protein component of the trans-
porter [19]. Interestingly, thiaminase I from B. thia-
minolyticus shares structural similarities with the
periplasmic thiamin-binding protein of the ABC
thiamin transporter from E. coli, suggesting that both
proteins share a common ancestor [19].
In eukaryotes, a plasma membrane thiamin trans-
porter was first cloned in yeast [20,21]. A second
transporter has been discovered recently in Schizosac-
charomyces pombe [22]. In animals, thiamin transport-
ers regulate thiamin homeostasis within the whole
organism with thiamin entry occurring in the small
intestine and excretion in the kidneys. Three proteins,
all belonging to the SLC19A solute carrier family,
have been implicated in thiamin transport [23].

SLC19A1, a reduced folate transporter, does not carry
thiamin, but is able to transport ThMP and ThDP
[24]. Because ThMP is present in plasma and cerebro-
spinal fluid, SLC19A1 might play a role in brain thia-
min homeostasis [24] as well as in the absorption of
ThDP from the intestine. SLC19A2 (thiamin trans-
porter 1, THTR-1) [25] and SLC19A3 (THTR-2) [26]
are specific plasma membrane thiamin ⁄ H
+
antiporters.
Both transporters are quite ubiquitously expressed in
mammalian tissues, with K
m
values in the 10
)6
–10
)5
m
range for THTR-1 [25] and in the 10
)8
–10
)7
m range
for THTR-2 [27]. In humans, mutations in SLC19A2
are responsible for thiamin-responsive megaloblastic
anemia, characterized by diabetes, deafness and
anemia [28].
Thiamin diphosphate biosynthesis
and transport into mitochondria and
peroxisomes

Thiamin diphosphate is formed in the cytosol by
an ATP : thiamin pyrophosphotransferase (thiamin
diphosphokinase or thiamin pyrophosphokinase, TPK;
EC 2.7.6.2). TPK is a homodimer of 46–56 kDa. Its
sequence was first obtained from Saccharomyces cerevi-
siae [29]. In the cytosol, a small part of the ThDP is
free and has a rapid turnover, whereas another part
binds to cytoplasmic transketolase with high affinity
[30]. However, most of the ThDP synthesized is trans-
ported into mitochondria by a carrier (SLC25A19)
that has recently been characterized in yeast [31] and
animals [32]. In humans, mutations in SLC25A19
L. Bettendorff and P. Wins Thiamin metabolism and thiamin phosphates
FEBS Journal 276 (2009) 2917–2925 ª 2009 The Authors Journal compilation ª 2009 FEBS 2919
cause Amish lethal microcephaly, a disease generally
fatal by the age of 6 months. Slc25a19
) ⁄ )
mice also
have central nervous system developmental defects,
such as an open neural tube and do not survive embry-
onic day 11. Cells cultured from Slc25a19
) ⁄ )
mice are
virtually devoid of intramitochondrial ThDP, resulting
in impairment of oxidative metabolism.
Some ThDP is also found in peroxisomes. Because
these organelles do not contain thiamin pyrophospho-
kinase activity, ThDP may be imported either by a
specific transport mechanism or it may bind first to
2-hydroxyacyl-CoA lyase and then be imported with

the enzyme as recently suggested [33].
Hydrolysis of thiamin diphosphate and
thiamin monophosphate
Hydrolysis of ThDP to ThMP may occur in most
organisms and tissues, but to date no specific thiamin
diphosphatase (ThDPase) has been characterized.
Many phosphatases are able to hydrolyze ThDP as
well as other thiamin phosphate derivatives, but gener-
ally less efficiently than nucleoside diphosphates, as is
the case of a liver microsomal nucleoside diphospha-
tase [34]. ThDPase activity is often used as a specific
marker of the Golgi apparatus. Indeed, a membrane-
associated nucleoside diphosphatase with a slight
preference for ThDP as substrate compared with
nucleoside diphosphates has been purified from rat
brain [35]. This enzyme has many properties in com-
mon with the ThDPase in the Golgi and is different
from the above-mentioned nucleoside diphosphatase
[34]. Other enzymes, especially from liver can hydro-
lyze both ThDP and nucleoside diphosphates but their
physiological function is probably to hydrolyze the
latter compounds.
ThMP can be rapidly hydrolyzed to thiamin in cul-
tured cells [24] but, except for one report in bacteria
[36], no specific ThMPase has yet been characterized.
ThMPase activity is used as a marker for spinal chord
small diameter dorsal root ganglions involved in noci-
ception [37]. ThMP is used as substrate in those stud-
ies because, in contrast to other phosphatase
substrates, it is not as easily hydrolyzed by lysosomal

acid phosphatase present in these preparations.
Thiamin triphosphate and its potential
roles
As mentioned earlier, the existence of ThTP was first
suggested in the 1950s and it was thought to have a
cofactor-independent neurophysiological role [38].
However, recent results show that ThTP is present in
most tissues and in most organisms studied to date [6],
suggesting that it might have a much more basic role
in cellular metabolism. In E. coli, ThTP is synthesized
in response to amino acid starvation and seems
required for optimal growth under these conditions
[39]. We suggested that ThTP may be a signal pro-
duced in response to changes in the nutritional envi-
ronment of the bacteria.
In multicellular organisms, the role of ThTP remains
enigmatic. It was shown to activate maxi-anion chan-
nels in inside-out patches of neuroblastoma cells [40].
These channels are thought to play a role in swelling-
induced ATP release [41]. The activation of maxi-anion
channels may be dependent on phosphorylation by
ThTP. Indeed, ThTP, like ATP, contains two
phosphoanhydride bonds with a high phosphate
energy transfer potential. Therefore, we tested whether
[
32
P]ThTP could phosphorylate proteins in vitro.
Indeed, in postsynaptic membranes from Torpedo
marmorata,[
32

P]ThTP could phosphorylate rapsyn, a
protein required for the clustering of acetylcholine
receptors at the neuromuscular junction [42]. Other, as
yet unidentified, proteins were also phosphorylated in
rodent brain. It is important to determine whether pro-
tein phosphorylation by ThTP could be part of a new
physiological signaling pathway.
Enzymatic synthesis of thiamin
triphosphate
The biosynthesis of ThTP was observed in vivo in
organisms such as bacteria [39], in cultured neuroblas-
toma cells [43] as well as in rat brain [30]. However,
the mechanism of ThTP synthesis remains an enigma.
The earliest reports proposed that an ATP : ThDP
phosphotransferase (ThDP kinase) catalyzes the reac-
tion ThDP + ATP « ThTP + ADP. Such an
enzyme system was purified [44] as a high molecular
mass multisubunit complex, but the rate of reaction
was very slow (k
cat
 1 min
)1
). Actually, it is not cer-
tain if the product of the reaction was authentic ThTP
and, with our present knowledge, it appears more
likely that the compound formed was AThTP which is
indeed formed under these conditions (see below).
In the late 1980s and early 1990s, Kawasaki and
co-workers [45,46] showed that vertebrate adenylate
kinase isoform 1 (AK1; EC 2.7.4.3), which is predomi-

nant in skeletal muscle cytoplasm, is able to synthesize
ThTP according to the reaction ThDP + ADP «
ThTP + AMP. They suggested that the in vivo syn-
thesis of ThTP occurs through this reaction, although
the rate of reaction was very low (for chicken AK1,
k
cat
 0.5 min
)1
at physiological pH) [46]. However,
Thiamin metabolism and thiamin phosphates L. Bettendorff and P. Wins
2920 FEBS Journal 276 (2009) 2917–2925 ª 2009 The Authors Journal compilation ª 2009 FEBS
after heat inactivation of AK in E. coli (E. coli has
only one AK isoform), ThTP levels are increased
rather than decreased [47] and transgenic mice lacking
AK1 have normal ThTP levels [48]. Our results suggest
that ThTP synthesis according to the above reaction is
a general property of AKs [47], but that it is not of
general physiological significance. Possible exceptions
to this rule are Electrophorus electricus electric organ
[49], as well as pig [50] and chicken [46] skeletal mus-
cles. Those tissues have a very high ThTP content, a
situation that may result from the combined effects of
a high cytosolic AK1 activity and a lack of a specific
soluble ThTPase activity (see below). This raises the
possibility that, in animals, ThTP might actually be
formed not in the cytosol, but in a different cellular
compartment. In fact, except for the very low ThTP-
synthesizing activity of AK1, we never observed
a rapid net synthesis of ThTP in soluble prepara-

tions from any biological source (B. Wirtzfeld,
A. F. Makarchikov and L. Bettendorff, unpublished
results). Therefore, it is possible that ThDP is not
formed in the cytosol but in a different subcellular
compartment.
Subcellular fractionation of rat brain showed that
the highest ThTP levels were found in the mitochon-
drial and synaptosomal fractions, the latter being also
rich in mitochondria [30]. Furthermore, when a rat
brain homogenate was incubated with ThDP, ThTP
was formed inside closed compartments [51]. The
nature of these compartments and the mechanism are
now under investigation in our laboratory.
Hydrolysis of thiamin triphosphate
In most mammalian cells, the steady-state concentra-
tions of ThTP remain low (10
)7
–10
)6
m) [6], probably
because of the presence of one or several ThTP-hydro-
lyzing enzymes with sufficient specificity and catalytic
efficiency. ThTP hydrolysis has been relatively well
studied, because the reaction can be more readily dem-
onstrated in cell extracts than ThTP synthesis. Early
studies have shown the presence of a soluble and a
membrane-associated enzyme able to hydrolyze ThTP
in rat tissues.
The latter enzyme was found to be associated with
particulate fractions (nuclear, synaptosomal and micro-

somal), was activated by Mg
2+
,Ca
2+
or Mn
2+
and
had a pH optimum around 6.5 [52]. Membrane-associ-
ated ThTPases were also described in electric organs
[49] and skeletal muscle [53], but to date all attempts
to purify these enzymes have failed. Although they
appear to be distinct from membrane-associated
ATPases, their specificity for ThTP is not established
and their catalytic efficiency could not be quantified.
Membrane-associated ThTPases from electric organs
and skeletal muscle are strongly activated by anions
[53,54], in particular by the chaotropic I
)
and NO
3
)
.
This is different from membrane-associated ThTPases
from other tissues such as the brain, that are inhibited
by these anions [53].
The cytosolic ThTPase (EC 3.6.1.28) is a soluble
protein requiring Mg
2+
as activator, Ca
2+

being
inhibitory, and having an alkaline pH optimum
( 9.0) [55]. The enzyme is expressed in most mamma-
lian tissues and was first purified from bovine brain
[56]. It is a low molecular mass protein ( 25 kDa)
with high catalytic efficiency and nearly absolute speci-
ficity for ThTP. Molecular characterization of the
human 25-kDa ThTPase [57] revealed that the
sequence does not closely resemble that of any other
protein identified in mammalian genomes. However,
bioinformatic analyses suggest that the mammalian
25-kDa ThTPases and the CyaB adenylyl cyclase from
Aeromonas hydrophila define a superfamily of domains
that can be traced back to the last universal common
ancestor [58]. This domain, called the CYTH (CyaB,
thiamin triphosphatase) domain, includes enzymes that
require divalent metal ions for activity and would play
various roles at the interface of organic polyphosphate
and nucleotide metabolism. Surprisingly, there is no
evidence that a member of this protein superfamily
exists in birds, the only known major class where it
would be absent. The pig ThTPase, though retaining
the CYTH signature, is practically devoid of ThTPase
activity probably as a result of a Glu85 fi Lys muta-
tion leading to conformational changes [59]. Indepen-
dent of this analysis by Iyer & Aravind [58], Shuman
and co-workers [60] defined a family of metal-depen-
dent phosphohydrolases which they called ‘triphos-
phate tunnel metalloenzymes’ (TTM) because their
active site is usually located within a topologically

closed hydrophilic b-barrel [60]. The proteins of this
family have common features with CYTH domains,
such as the EXEXK signature, which is a divalent cat-
ion-binding motif. The founding member of the TTM
superfamily was the yeast RNA triphosphatase Cet1
[61], but homologous sequences have been found in
archaeal and bacterial species. One of these proteins
from Clostridium thermocellum was recently found to
hydrolyze inorganic triphosphate with a much higher
catalytic efficiency than ATP [62]. There was no
adenylyl cyclase activity. It is not known whether any
of these enzymes from the TTM family would be able
to bind or to hydrolyze ThTP. It appears that the
‘CYTH–TTM’ superfamily includes enzymes with vari-
ous catalytic properties (adenylyl cyclase or inorganic
L. Bettendorff and P. Wins Thiamin metabolism and thiamin phosphates
FEBS Journal 276 (2009) 2917–2925 ª 2009 The Authors Journal compilation ª 2009 FEBS 2921
triphosphatase in some bacteria, RNA triphosphatase
in yeast, ThTPase in some animal species) but with
important common features: (a) the activity always
requires divalent metal cations, and (b) there is speci-
ficity for substrates containing a triphosphate group.
The structure of recombinant mouse 25-kDa ThT-
Pase has been determined recently [63] and the residues
responsible for binding Mg
2+
and ThTP determined
from NMR titration experiments. Although the free
enzyme has an open cleft form, the ternary complex
[ThTPase–Mg

2+
–ThTP] tends to adopt a closed tun-
nel-fold, suggesting that mammalian 25-kDa ThTPases
may be considered true members of the TTM super-
family of proteins.
Another important question is to know why ThTP
should be hydrolyzed at all? If ThTP is indeed a
signaling molecule, hydrolysis by ThTPases might
terminate its action in the same way as phosphodi-
esterases terminate the action of cAMP. This may be
true in E. coli, where the appearance of ThTP is gener-
ally transient and followed by rapid hydrolysis [7,39].
However, in mammalian cells, ThTP seems to be
continuously formed and hydrolyzed [30,43].
An important question therefore concerns the regu-
lation of soluble ThTPase activity, which would ulti-
mately control cytoplasmic ThTP concentrations.
ThTPase contains several consensus sequences for
phosphorylation by protein kinase C and casein kinase
2 [57,64]. However, it was shown recently that the
iron-regulated metastasis suppressor Ndgr-1 upregu-
lates 25-kDa ThTPase expression in several cancer cell
models [65]. ThTPase expression was inversely corre-
lated to melanoma tumor antigen p97 (MTf), an iron-
binding protein expressed at high levels in melanoma
cells [66]. The significance of these results is unclear,
but they may suggest that ThTPase expression is
linked to the degree of differentiation of cells. Indeed,
in the adult rodent brain, ThTPase is mainly found
in the highly differentiated pyramidal and Purkinje

neurons [67].
Adenosine thiamin triphosphate, a new
thiamin compound
As mentioned above, incubation of E. coli in amino
acid-deficient medium in the presence of a suitable car-
bon source led to the accumulation of ThTP. How-
ever, in the absence of a carbon source, no ThTP was
observed, but an additional peak was detected. This
raised the possibility of the existence of a new thiamin
compound. Purification and analysis by MS and
1
H-NMR showed that this compound is adenosine
thiamin triphosphate or thiaminylated ATP [7], a com-
pound not previously described. In E. coli, AThTP is
synthesized according to the reaction ThDP +
ADP (ATP) « AThTP + P
i
(PP
i
), probably by a thi-
amin diphosphate adenylyl transferase [68]. Note that
both ATP and ADP may be substrates, but not other
nucleotides. This enzyme is probably a high molecular
mass complex requiring a low molecular mass activa-
tor. When E. coli accumulate AThTP, addition of
glucose leads, within minutes, to its complete dis-
appearance. This experiment strongly suggests the
existence of at least one AThTP-hydrolyzing enzyme
that remains to be characterized. In E. coli, AThTP
might act as an alarmone, signaling carbon starvation

and ⁄ or a low energy charge. AThTP is also found at
low levels in eukaryotic organisms [7], though we have
no clue as to its role there.
Conclusion
The textbook view on the biochemical role of thiamin
is that the vitamin, after entering the cells, is pyro-
phosphorylated to ThDP, a cofactor for several
enzymes. Decreased enzyme activities caused by
decreased ThDP levels would be responsible for the
symptoms observed during thiamin deficiency. For two
decades, this view has nearly been raised to a dogma,
with practically complete ignorance of other thiamin
derivatives and the enzymes of thiamin metabolism
(Fig. 1).
In E. coli, a significant part of ThDP (up to 60%)
can be rapidly converted to either ThTP or AThTP
according to the metabolic state of the bacteria
[7,39,47]. ThTP and AThTP could act as signals
involved in the adaptation of the bacteria to stress con-
ditions. Both compounds can be rapidly formed (within
minutes) or hydrolyzed, suggesting the existence of a
complete set of enzymes able to rapidly respond to
changing environmental conditions. In eukaryotes,
ThTP and AThTP seem to be less subject to rapid
changes. Nevertheless, ThTP has a higher turnover
than the bulk of ThDP [30,43]. In rat brain, it is con-
stantly formed and hydrolyzed, but until now no spe-
cific conditions under which it accumulates or
disappears could be determined. AThTP is also formed
in eukaryotes but its potential role in these organisms

is completely unknown. It was previously suggested
that ThTP might have a specific neurochemical role
[38], but recent evidence is in favor of a much more
basic role in cellular metabolism, possibly in the fine-
tuning or integration of some metabolic processes.
In addition to clarifying the precise roles of ThTP
and AThTP, the enzymes leading to their synthesis
and hydrolysis need to be characterized. It is surprising
Thiamin metabolism and thiamin phosphates L. Bettendorff and P. Wins
2922 FEBS Journal 276 (2009) 2917–2925 ª 2009 The Authors Journal compilation ª 2009 FEBS
that 50 years after the discovery of ThTP, the mecha-
nism of its synthesis remains unclear. Membrane-asso-
ciated ThTPases have been reported to exist in
practically all organisms, including bacteria, but so far
none has been characterized from a molecular point of
view. Last, but not least, ThDPases and ThMPases,
although probably playing a role in the maintenance
of steady-state ThDP concentrations, have not been
characterized at the molecular level. The intriguing
complexity of thiamin metabolism raises numerous
questions, many of which remain unanswered. In any
event, the new developments described in this short
review strongly suggest that the simplistic view that
the cofactor ThDP is the only biologically active form
of thiamin is no longer tenable.
Acknowledgements
LB is Research Director at the F.R.S FNRS. This
work was supported by grants from the F.R.S FNRS.
References
1 Frank RA, Leeper FJ & Luisi BF (2007) Structure,

mechanism and catalytic duality of thiamin-dependent
enzymes. Cell Mol Life Sci 64, 892–905.
2 Breslow R (1958) On the mechanism of thiamin action.
IV. Evidence from studies on model systems. JAm
Chem Soc 80, 3719–3726.
3 Kale S, Ulas G, Song J, Brudvig GW, Furey W &
Jordan F (2008) Efficient coupling of catalysis and
dynamics in the E1 component of Escherichia coli
pyruvate dehydrogenase multienzyme complex. Proc
Natl Acad Sci USA 105, 1158–1163.
4 Kluger R & Tittmann K (2008) Thiamin diphosphate
catalysis: enzymic and nonenzymic covalent intermedi-
ates. Chem Rev 108 , 1797–1833.
5 Kawasaki T (1992) Vitamin B1: thiamin. In Modern
Chromatographic Analysis of Vitamins (De Leenheer
AP, Lambert WE & Nelis HJ, eds), pp. 319–354.
Marcel Dekker, New York.
6 Makarchikov AF, Lakaye B, Gulyai IE, Czerniecki J,
Coumans B, Wins P, Grisar T & Bettendorff L (2003)
Thiamin triphosphate and thiamin triphosphatase activ-
ities: from bacteria to mammals. Cell Mol Life Sci 60,
1477–1488.
7 Bettendorff L, Wirtzfeld B, Makarchikov AF, Maz-
zucchelli G, Fre
´
de
´
rich M, Gigliobianco T, Gangolf M,
De Pauw E, Angenot L & Wins P (2007) Discovery of
a natural thiamin adenine nucleotide. Nat Chem Biol 3,

211–212.
8 Shi Q, Karuppagounder SS, Xu H, Pechman D, Chen
H & Gibson GE (2007) Responses of the mitochondrial
alpha-ketoglutarate dehydrogenase complex to thiamin
deficiency may contribute to regional selective vulnera-
bility. Neurochem Int 50 , 921–931.
9 Navarro D, Zwingmann C & Butterworth RF (2008)
Region-selective alterations of glucose oxidation and
amino acid synthesis in the thiamin-deficient rat brain:
a re-evaluation using
1
H ⁄
13
C nuclear magnetic reso-
nance spectroscopy. J Neurochem 106, 603–612.
10 Karuppagounder SS, Xu H, Shi Q, Chen LH, Pedrini
S, Pechman D, Baker H, Beal MF, Gandy SE & Gib-
son GE (2008) Thiamin deficiency induces oxidative
stress and exacerbates the plaque pathology in Alzhei-
mer’s mouse model. Neurobiol Aging. doi: 10.1016/j.neu-
robiolaging.2007.12.013.
11 Webb ME, Marquet A, Mendel RR, Rebeille F &
Smith AG (2007) Elucidating biosynthetic pathways for
vitamins and cofactors. Nat Prod Rep 24, 988–1008.
12 Begley TP, Chatterjee A, Hanes JW, Hazra A & Ealick
SE (2008) Cofactor biosynthesis – still yielding fascinat-
ing new biological chemistry. Curr Opin Chem Biol 12,
118–125.
13 Murata K (1982) Actions of two types of thiaminase on
thiamin and its analogues. Ann NY Acad Sci 378, 146–

156.
14 Jenkins AH, Schyns G, Potot S, Sun G & Begley TP
(2007) A new thiamin salvage pathway. Nat Chem Biol
3, 492–497.
15 Onozuka M, Konno H, Kawasaki Y, Akaji K &
Nosaka K (2008) Involvement of thiaminase II encoded
by the THI20 gene in thiamin salvage of Saccharomy-
ces cerevisiae. FEMS Yeast Res 8, 266–275.
16 Montange RK & Batey RT (2008) Riboswitches:
emerging themes in RNA structure and function. Annu
Rev Biophys 37, 117–133.
17 Winkler W, Nahvi A & Breaker RR (2002) Thiamin
derivatives bind messenger RNAs directly to regulate
bacterial gene expression. Nature 419, 952–956.
18 Mironov AS, Gusarov I, Rafikov R, Lopez LE, Shata-
lin K, Kreneva RA, Perumov DA & Nudler E (2002)
Sensing small molecules by nascent RNA. A mechanism
to control transcription in bacteria. Cell 111, 747–756.
19 Soriano EV, Rajashankar KR, Hanes JW, Bale S,
Begley TP & Ealick SE (2008) Structural similarities
between thiamin-binding protein and thiaminase-I sug-
gest a common ancestor. Biochemistry 47, 1346–1357.
20 Enjo F, Nosaka K, Ogata M, Iwashima A & Nishimura
H (1997) Isolation and characterization of a thiamin
transport gene, THI10, from Saccharomyces cerevisiae.
J Biol Chem 272, 19165–19170.
21 Singleton CK (1997) Identification and characterization
of the thiamin transporter gene of Saccharomyces cere-
visiae. Gene 199, 111–121.
22 Vogl C, Klein CM, Batke AF, Schweingruber ME &

Stolz J (2008) Characterization of Thi9, a novel thiamin
(vitamin B1) transporter from Schizosaccharomy-
ces pombe. J Biol Chem 283, 7379–7389.
L. Bettendorff and P. Wins Thiamin metabolism and thiamin phosphates
FEBS Journal 276 (2009) 2917–2925 ª 2009 The Authors Journal compilation ª 2009 FEBS 2923
23 Ganapathy V, Smith SB & Prasad PD (2004) SLC19:
the folate ⁄ thiamin transporter family. Pflu
¨
gers Arch
447, 641–646.
24 Zhao R, Gao F & Goldman ID (2002) Reduced folate
carrier transports thiamin monophosphate: an alterna-
tive route for thiamin delivery into mammalian cells.
Am J Physiol Cell Physiol 282, C1512–C1517.
25 Dutta B, Huang W, Molero M, Kekuda R, Leibach
FH, Devoe LD, Ganapathy V & Prasad PD (1999)
Cloning of the human thiamin transporter, a member
of the folate transporter family. J Biol Chem 274,
31925–31929.
26 Eudy JD, Spiegelstein O, Barber RC, Wlodarczyk BJ,
Talbot J & Finnell RH (2000) Identification and charac-
terization of the human and mouse SLC19A3 gene: a
novel member of the reduced folate family of
micronutrient transporter genes. Mol Genet Metab 71,
581–590.
27 Said HM, Balamurugan K, Subramanian VS & Mar-
chant JS (2004) Expression and functional contribution
of hTHTR-2 in thiamin absorption in human intestine.
Am J Physiol Gastrointest Liver Physiol 286, G491–G498.
28 Fleming JC, Tartaglini E, Steinkamp MP, Schorderet

DF, Cohen N & Neufeld EJ (1999) The gene mutated
in thiamin-responsive anaemia with diabetes and deaf-
ness (TRMA) encodes a functional thiamin transporter.
Nat Genet 22, 305–308.
29 Nosaka K, Kaneko Y, Nishimura H & Iwashima A
(1993) Isolation and characterization of a thiamin
pyrophosphokinase gene, THI80, from Saccharo-
myces cerevisiae. J Biol Chem 268, 17440–17447.
30 Bettendorff L, Wins P & Lesourd M (1994) Subcellular
localization and compartmentation of thiamin deriva-
tives in rat brain. Biochim Biophys Acta 1222, 1–6.
31 Marobbio CM, Vozza A, Harding M, Bisaccia F,
Palmieri F & Walker JE (2002) Identification and
reconstitution of the yeast mitochondrial transporter for
thiamin pyrophosphate. EMBO J 21, 5653–5661.
32 Lindhurst MJ, Fiermonte G, Song S, Struys E, De
Leonardis F, Schwartzberg PL, Chen A, Castegna A,
Verhoeven N, Mathews CK et al. (2006) Knockout of
Slc25a19 causes mitochondrial thiamin pyrophosphate
depletion, embryonic lethality, CNS malformations, and
anemia. Proc Natl Acad Sci USA 103, 15927–15932.
33 Fraccascia P, Sniekers M, Casteels M & Van Veldhoven
PP (2007) Presence of thiamin pyrophosphate in
mammalian peroxisomes. BMC Biochem 8, 10.
34 Sano S, Matsuda Y & Nakagawa H (1988) Thiamin
pyrophosphatase (nucleoside diphosphatase) in the
Golgi apparatus is distinct from microsomal nucleoside
diphosphatase. J Biochem 103, 678–681.
35 Sano S, Matsuda Y & Nakagawa H (1988) Type B
nucleoside-diphosphatase of rat brain. Purification and

properties of an enzyme with high thiamin pyrophos-
phatase activity. Eur J Biochem 171, 231–236.
36 Kawasaki T, Egi Y & Sanemori H (1977) Specific thia-
min monophosphate phosphohydrolase in Micrococ-
cus denitrificans. J Bacteriol 130, 542–544.
37 Csillik B, Mihaly A, Krisztin-Peva B, Farkas I &
Knyihar-Csillik E (2008) Mitigation of nociception via
transganglionic degenerative atrophy: possible mecha-
nism of vinpocetine-induced blockade of retrograde
axoplasmic transport. Ann Anat 190, 140–145.
38 Bettendorff L (1994) Thiamin in excitable tissues:
reflections on a non-cofactor role. Metab Brain Dis 9,
183–209.
39 Lakaye B, Wirtzfeld B, Wins P, Grisar T & Bettendorff
L (2004) Thiamin triphosphate, a new signal required
for optimal growth of Escherichia coli during amino
acid starvation. J Biol Chem 279, 17142–17147.
40 Bettendorff L, Kolb HA & Schoffeniels E (1993) Thia-
min triphosphate activates an anion channel of large
unit conductance in neuroblastoma cells. J Membr Biol
136, 281–288.
41 Sabirov RZ & Okada Y (2005) ATP release via anion
channels. Purinergic Signal 1, 311–328.
42 Nghieˆ m HO, Bettendorff L & Changeux JP (2000)
Specific phosphorylation of Torpedo 43K rapsyn by
endogenous kinase(s) with thiamin triphosphate as the
phosphate donor. FASEB J 14, 543–554.
43 Bettendorff L (1994) The compartmentation of phos-
phorylated thiamin derivatives in cultured neuroblas-
toma cells. Biochim Biophys Acta 1222, 7–14.

44 Voskoboyev AI, Chernikevich IP & Luchko VS (1987)
Studies on thiamin diphosphate kinase (EC 2.7.4.15)
from brewer’s yeast: purification and some properties.
Biomed Biochim Acta 46, 3–13.
45 Shikata H, Koyama S, Egi Y, Yamada K & Kawasaki
T (1989) Cytosolic adenylate kinase catalyzes the syn-
thesis of thiamin triphosphate from thiamin diphos-
phate. Biochem Int 18, 933–941.
46 Miyoshi K, Egi Y, Shioda T & Kawasaki T (1990)
Evidence for in vivo synthesis of thiamin triphosphate
by cytosolic adenylate kinase in chicken skeletal muscle.
J Biochem (Tokyo) 108, 267–270.
47 Gigliobianco T, Lakaye B, Makarchikov AF, Wins P &
Bettendorff L (2008) Adenylate kinase-independent
thiamin triphosphate accumulation under severe energy
stress in Escherichia coli. BMC Microbiol 8, 16.
48 Makarchikov AF, Wins P, Janssen E, Wieringa B,
Grisar T & Bettendorff L (2002) Adenylate kinase 1
knockout mice have normal thiamin triphosphate levels.
Biochim Biophys Acta 1592, 117–121.
49 Bettendorff L, Michel-Cahay C, Grandfils C, De
Rycker C & Schoffeniels E (1987) Thiamin triphosphate
and membrane-associated thiamin phosphatases in the
electric organ of Electrophorus electricus . J Neurochem
49, 495–502.
50 Egi Y, Koyama S, Shikata H, Yamada K & Kawasaki
T (1986) Content of thiamin phosphate esters in
Thiamin metabolism and thiamin phosphates L. Bettendorff and P. Wins
2924 FEBS Journal 276 (2009) 2917–2925 ª 2009 The Authors Journal compilation ª 2009 FEBS
mammalian tissues – an extremely high concentration

of thiamin triphosphate in pig skeletal muscle. Biochem
Int 12, 385–390.
51 Bettendorff L, Peeters M, Wins P & Schoffeniels E
(1993) Metabolism of thiamin triphosphate in rat brain:
correlation with chloride permeability. J Neurochem 60,
423–434.
52 Barchi RL & Braun PE (1972) A membrane-associated
thiamin triphosphatase from rat brain. Properties of the
enzyme. J Biol Chem 247, 7668–7673.
53 Matsuda T, Tonomura H, Baba A & Iwata H (1991)
Membrane-associated thiamin triphosphatase in rat
skeletal muscle. Int J Biochem 23, 1111–1114.
54 Bettendorff L, Wins P & Schoffeniels E (1988) Thiamin
triphosphatase from Electrophorus electric organ is
anion-dependent and irreversibly inhibited by 4,4¢-diiso-
thiocyanostilbene-2,2¢disulfonic acid. Biochem Biophys
Res Commun 154, 942–947.
55 Hashitani Y & Cooper JR (1972) The partial purifica-
tion of thiamin triphosphatase from rat brain. J Biol
Chem 247, 2117–2119.
56 Makarchikov AF & Chernikevich IP (1992) Purification
and characterization of thiamin triphosphatase from
bovine brain. Biochim Biophys Acta 1117 , 326–332.
57 Lakaye B, Makarchikov AF, Antunes AF, Zorzi W,
Coumans B, De Pauw E, Wins P, Grisar T & Bette-
ndorff L (2002) Molecular characterization of a specific
thiamin triphosphatase widely expressed in mammalian
tissues. J Biol Chem 277, 13771–13777.
58 Iyer LM & Aravind L (2002) The catalytic domains of
thiamin triphosphatase and CyaB-like adenylyl cyclase

define a novel superfamily of domains that bind organic
phosphates. BMC Genomics 3, 33.
59 Szyniarowski P, Lakaye B, Czerniecki J, Makarchikov
AF, Wins P, Margineanu I, Coumans B, Grisar T &
Bettendorff L (2005) Pig tissues express a catalytically
inefficient 25-kDa thiamin triphosphatase: insight in the
catalytic mechanisms of this enzyme. Biochim Biophys
Acta 1725, 93–102.
60 Gong C, Smith P & Shuman S (2006) Structure–func-
tion analysis of Plasmodium RNA triphosphatase and
description of a triphosphate tunnel metalloenzyme
superfamily that includes Cet1-like RNA triphosphata-
ses and CYTH proteins. RNA 12, 1468–1474.
61 Lima CD, Wang LK & Shuman S (1999) Structure and
mechanism of yeast RNA triphosphatase: an essential
component of the mRNA capping apparatus. Cell 99,
533–543.
62 Keppetipola N, Jain R & Shuman S (2007) Novel
triphosphate phosphohydrolase activity of Clostrid-
ium thermocellum TTM, a member of the triphosphate
tunnel metalloenzyme superfamily. J Biol Chem 282,
11941–11949.
63 Song J, Bettendorff L, Tonelli M & Markley JL (2008)
Structural basis for the catalytic mechanism of
mammalian 25-kDa thiamin triphosphatase. J Biol
Chem 283, 10939–10948.
64 Lakaye B, Verlaet M, Dubail J, Czerniecki J, Bontems
S, Makarchikov AF, Wins P, Piette J, Grisar T &
Bettendorff L (2004) Expression of 25 kDa thiamin
triphosphatase in rodent tissues using quantitative PCR

and characterization of its mRNA. Int J Biochem Cell
Biol
36, 2032–2041.
65 Kovacevic Z, Fu D & Richardson DR (2008) The iron-
regulated metastasis suppressor, Ndrg-1: identification
of novel molecular targets. Biochim Biophys Acta 1783,
1981–1992.
66 Suryo Rahmanto Y, Dunn L & Richardson D (2007)
Identification of distinct changes in gene expression
after modulation of melanoma tumor antigen p97 (mel-
anotransferrin) in multiple models in vitro and in vivo.
Carcinogenesis 26, 2172–2183.
67 Czerniecki J, Chanas G, Verlaet M, Bettendorff L,
Makarchikov AF, Leprince P, Wins P, Grisar T &
Lakaye B (2004) Neuronal localization of the 25-kDa
specific thiamin triphosphatase in rodent brain. Neuro-
science 125, 833–840.
68 Makarchikov AF, Brans A & Bettendorff L (2007)
Thiamin diphosphate adenylyl transferase from E. coli:
functional characterization of the enzyme synthesizing
adenosine thiamin triphosphate. BMC Biochem 8, 17.
L. Bettendorff and P. Wins Thiamin metabolism and thiamin phosphates
FEBS Journal 276 (2009) 2917–2925 ª 2009 The Authors Journal compilation ª 2009 FEBS 2925