The resulting ylid subsequently carries out a nucleophilic attack on the keto group of pyruvate, followed by decarboxylation which leads to formation of the enam-ine intermediate which c
Trang 1Thiamin diphosphate in biological chemistry: analogues
of thiamin diphosphate in studies of enzymes
and riboswitches
Kwasi Agyei-Owusu and Finian J Leeper
Department of Chemistry, University of Cambridge, UK
Thiamin diphosphate 1 (ThDP) (Fig 1) is a coenzyme
that assists in the catalysis of carbon–carbon
bond-forming and bond-breaking reactions adjacent to a
carbonyl group A large and diverse collection of
enzymes require ThDP, which acts as an electron sink
during catalysis, stabilizing what would otherwise be
an acyl carbanion in the form of an enamine
interme-diate (Fig 2) These enzymes include pyruvate
decar-boxylase (PDC; EC 4.1.1.1) which is involved in the
formation of alcohol in anaerobic fermentation [1],
transketolase (TK; EC 2.2.1.1) which transfers a
two-carbon unit from a ketose to an aldose [2] and
pyruvate dehydrogenase (PDH; EC 1.2.4.1), a highly
complex enzyme that links the glycolytic pathway to the citric acid cycle through the formation of acetyl CoA [3]
The general catalytic cycle for ThDP-dependent enzymes, as illustrated by the cycle for PDC (Fig 2), was first elucidated by Breslow [4] and begins with deprotonation of the C-2 carbon of the thiazolium ring (now believed to be effected by the 4¢-N atom of the aminopyrimidine ring in its imino tautomer) The resulting ylid subsequently carries out a nucleophilic attack on the keto group of pyruvate, followed by decarboxylation which leads to formation of the enam-ine intermediate (which can also be drawn as an
Keywords
acetohydroxyacid synthase; enzyme crystal
structure; enzyme inhibition; pyruvate
decarboxylase; pyruvate dehydrogenase;
reaction mechanism; riboswitch; thiamin
pyrophosphate; thiamine diphosphate;
transketolase
Correspondence
F J Leeper, Department of Chemistry,
University of Cambridge, Lensfield Road,
Cambridge CB2 1EW, UK
Fax: +44 1223 336362
Tel: +44 1223 336403
E-mail: FJL1@cam.ac.uk
Website: http://www.ch.cam.ac.uk/staff/
fjl.html
(Received 9 October 2008, revised 9 March
2009, accepted 12 March 2009)
doi:10.1111/j.1742-4658.2009.07018.x
The role of thiamin diphosphate (ThDP) as a cofactor for enzymes has been known for many decades This minireview covers the progress made
in understanding the catalytic mechanism of ThDP-dependent enzymes through the use of ThDP analogues Many such analogues have been syn-thesized and have provided information on the functional groups necessary for the binding and catalytic activity of the cofactor Through these studies, the important role of hydrophobic interactions in stabilizing reaction inter-mediates in the catalytic cycle has been recognized Stable analogues of intermediates in the ThDP-catalysed reaction mechanism have also been synthesized and crystallographic studies using these analogues have allowed enzyme structures to be solved that represent snapshots of the reaction in progress As well as providing mechanistic information about ThDP-depen-dent enzymes, many analogues are potent inhibitors of these enzymes The potential of these compounds as therapeutic targets and as important her-bicidal agents is discussed More recently, the way that ThDP regulates the genes for its own biosynthesis through the action of riboswitches has been discovered This opens a new branch of thiamin research with the potential
to provide new therapeutic targets in the fight against infection
Abbreviations
3-deazaThDP, 3-deazathiamin diphosphate; PDB, Protein Database; PDC, pyruvate decarboxylase; PDH, pyruvate dehydrogenase; PDHc, pyruvate dehydrogenase complex; ThDP, thiamin diphosphate; ThTDP, thiamin 2-thiazolone diphosphate; ThTTDP, thiamin 2-thiothiazolone diphosphate; TK, transketolase; TPK, thiamin pyrophosphokinase.
Trang 2a-carbanion resonance structure) Depending on the
ThDP-dependent enzyme involved, the enamine can
attack a number of electrophilic species In the case of
PDC, protonation of the enamine occurs and finally
release of acetaldehyde regenerates the ThDP ylid
Over the course of the last 70 years, numerous
ana-logues of ThDP have been synthesized by various
research groups and this has afforded us an insight
into the catalytic mechanism of ThDP-dependent
enzymes [4] Binding and inhibition studies, as well as
X-ray crystallography involving ThDP analogues, have
provided us with information on enzyme⁄
coen-zyme⁄ substrate interactions and how reaction
interme-diates are stabilized by the enzyme More recently,
work on various enzymes as targets for antitumour
and antimalarial drugs based on ThDP has been
pur-sued with some promise [5,6] With the discovery of
ThDP-reponsive riboswitches [7], an exciting field for exploration has recently been opened up: the regula-tion of thiamin biosynthesis in some organisms by ThDP riboswitches offers a novel target for drugs based on thiamin This minireview covers the key enzymatic studies of ThDP analogues that have aided our mechanistic understanding of ThDP-dependent enzymes, as well as studies of thiamin analogues as drug molecules and in the field of riboswitches
Modifications to the aminopyrimidine moiety
Analogues of ThDP with a modified aminopyrimidine moiety (Fig 3) provided a key insight into binding of the coenzyme as well as its catalytic activity Schellen-berger et al synthesized a number of pyridyl and dea-minopyrimidine species for testing on the apoenzymes
of PDC, PDH and TK [8,9] The N1 0
-pyridyl analogue (in which the 3¢-nitrogen has been replaced by carbon) was found not only to bind to the apoenzymes tested, but also to be catalytically active It was concluded that the 3¢-nitrogen was not crucial for binding or cat-alytic activity The N3 0
-pyridyl analogue (in which the 1¢-nitrogen has been replaced by carbon), however, was found to bind relatively weakly and produce no discernible catalytic activity From this result, it could
Fig 2 Catalytic mechanism for pyruvate decarboxylase.
Fig 1 Thiamin diphosphate.
Trang 3be concluded that hydrogen bonding between a
suit-able amino acid residue and the N1 0
nitrogen was not only required for tight binding, but also was critical
for the catalytic activity of the coenzyme It was
subse-quently discovered by X-ray crystallography,
site-spe-cific mutagenesis and NMR studies of coenzyme
analogues that there is a conserved glutamate residue
in the active site of thiamin-dependent enzymes that
hydrogen bonds to the N1 0
nitrogen (Fig 4) [10] and the consequence of this hydrogen bonding is to
increase the basicity of the 4¢-amino group, its ability
to act as a proton acceptor for the C2–H of the
thiazo-lium ring being enhanced by several orders of
magni-tude [11]
By deleting the 4¢-amino group altogether or
replac-ing it with other functionalities such as a thiol,
hydroxy or dimethylamino group, its important role in
catalysis soon became apparent [9] These analogues
generally showed good binding capacity but were
cata-lytically inactive, offering further proof of the proton
acceptor role of the amino group One exception is the
4¢-monomethylated amino group which has been
shown to retain catalytic activity when bound to
trans-ketolase with b-hydroxypyruvate as substrate [12]
However, the stability of the resulting enamine inter-mediate was reduced significantly It was suggested that this is caused by impaired hydrogen bonding between the amino group and the dihydroxyethyl group attached to C-2
It has been shown that ThDP adopts a specific con-formation in the active site of the enzyme [13] This so-called ‘V-conformation’ was first suggested when 6¢-methylated ThDP analogues were tested for enzymic activity by Schellenberger [9] For PDC, it was reported that 6¢-methyl-ThDP was catalytically inac-tive, whereas 4-demethyl-6¢-methyl-ThDP was active
It was proposed that a 6¢-methyl group causes a steric clash with the 4-methyl group when the ThDP is in the V-conformation and the analogue therefore has to adopt an altered conformation Interestingly, the 4-demethylated analogue, which retains its catalytic activity, is a relatively weak binder to the apoenzyme, suggesting the requirement of hydrophobic interactions between the 4-methyl group and the apoenzyme for good binding affinity It was later reported that in transketolase the 6¢-methyl-ThDP does have 50% of the normal catalytic activity [14], whereas the 4-dem-ethyl-6¢-methyl-ThDP is inactive [15]
Crystal structures of three of these pyrimidine-modi-fied analogues, 1¢-pyridyl-, 3¢-pyridyl- and 6¢-methyl-ThDP, bound to transketolase have been obtained [16] The pyrimidine ring is found in a very similar position to that of the normal ThDP complex in all three analogues, confirming that the lack of activity of the N3 0
-pyridyl analogue is caused by the loss of hydrogen bonding and not by a change of conformation
Modifications to the thiazolium ring The positively charged thiazolium ring of ThDP has long been recognized as the catalytic centre of the coenzyme [4] Synthetic modifications to the thiazolium moiety (Fig 5) have been instrumental in providing insights into the catalytic mechanism of ThDP as well
as providing information on how reaction intermedi-ates of the coenzyme are stabilized by the various ThDP-dependent enzymes [17,18]
Modifications at C-2 Substituents at the C-2 position of ThDP can lead to potent inhibitors of ThDP-dependent enzymes Two such compounds are thiamin 2-thiazolone diphosphate (ThTDP), in which the C2–H has been replaced by C=O, and thiamin 2-thiothiazolone diphosphate (ThTTDP), in which the C2–H is replaced by C=S
Fig 3 Modifications to the aminopyrimidine ring of ThDP.
Fig 4 Amino acid residues surrounding ThDP in PDH E1 from
Escherichia coli (from PDB entry 1L8A) The carbons of ThDP are
in green and other carbons in grey Conserved glutamate residue
Glu571 hydrogen bonds to N 1 0
of ThDP Figures were drawn using
PYMOL [59].
Trang 4[18] Both are tight binding but reversible inhibitors of
the E1 component of PDH of Escherichia coli
ThTDP, in particular, was found to have a Kivalue in
the very low nm range, as measured by a number of
methods It can be seen that both ThTDP and
ThTTDP mimic the neutral enamine intermediate
com-mon to all ThDP-dependent enzymes and, as such, it
is not surprising that they bind tightly to the enzyme
An interesting result obtained from studying these two
analogues was the discovery of a new positive CD
band at 330 nm when ThTTDP is added to the E1
component of PDH [18], which is absent when either
ThDP or ThTDP is added to the subunit This
phe-nomenon is thought to be caused by the chiral
confor-mation imposed on ThTTDP when it is bound in the
active site 2-Methyl-ThDP was also found to inhibit
the enzyme by preventing formation of the ThDP
com-plex [9] Several structures of ThTDP bound to
ThDP-dependent enzymes have been solved, including human
branched-chain 2-keto acid dehydrogenase (Protein
Database [PDB] entries 2BFC and 2BFF),
benzoyl-formate decarboxylase (1YNO), oxalyl CoA
decarbox-ylase (2C31) and PDH E1 (1RP7) In the last of these
examples, the tighter binding of ThTDP than ThDP
was ascribed to an increase in the number of hydrogen
bonds between the protein and the ligand [19]
A further derivative of ThDP modified at C-2 is
tetrahydroThDP 2, which is formed by reduction of
ThDP with sodium borohydride TetrahydroThDP is
reported to inhibit yeast TK with a Kivalue of 0.4 lm
[20], yeast PDC with a Ki value of 6.5 lm [21] and
PDH complex with an IC50 of 0.046 lm [22]
Tetra-hydroThDP made by sodium borohydride reduction of
ThDP consists of two racemic diastereoisomers The
diastereoisomers have been separated and it was
found that the cis isomer was a much more potent inhibitor (Ki= 0.02–0.15 lm) than the trans isomer (Ki= 5–10 lm) [23]
Structures representing the pre-decarboxylation stage
of the reaction of PDH E1 [24–25] and pyruvate oxi-dase [26] have been obtained by incubating the native enzyme with methyl acetylphosphonate This reacts with the ThDP ylid to produce 2-phosphonolactyl-ThDP, an analogue of 2-lactyl-ThDP (Fig 2) that has -PðOMeÞO
2 in place of -CO
2 Similarly, reaction of benzaldehyde lyase with methyl benzoylphosphonate produces a phosphonate analogue of 2-mandelyl-ThDP [27] Interestingly with benzoylformate decarboxylase,
if benzoylphosphonate is used instead of its methyl ester, a phosphoryl group transfer occurs to an active-site serine residue [28] Succinylphosphonate and its monoethyl esters, inactivate a-ketoglutarate dehydro-genase [29], so presumably this also forms phosphono analogues of the predecarboxylation intermediate, though in this case no crystal structure has been reported as yet
3-Deazathiamin diphosphate and derivatives Among the most potent inhibitors of ThDP-dependent enzymes is 3-deazathiamin diphosphate (3-deazaThDP)
3, first synthesized by Hawksley et al (Fig 6) [30,31]
In this compound, a carbon atom replaces the N-3
Fig 6 3-DeazaThDP 3 and its derivatives 4 and 5, triazole ana-logues 6 and 7, and open-chain anaana-logues 8 and 9.
Fig 5 Some modifications to the thiazolium moiety that have been
studied One enantiomer of the cis isomer of tetrahydro-ThDP 2 is
shown.
Trang 5atom of ThDP resulting in a neutral thiophene ring in
place of the thiazolium ring of the coenzyme This
analogue is of the same size and steric profile as ThDP
and the only difference is the absence of a positive
charge at the 3-position, which precludes the formation
of the reactive ylid required for catalysis Enzymatic
studies with 3-deazaThDP carried out on PDC from
Zymomonas mobilis (ZmPDC) showed it to be an
essentially irreversible inhibitor, binding 25 000
times more tightly than ThDP [31] Binding to
a-keto-glutarate dehydrogenase (EC 1.2.4.2) E1 component
by 3-deazaThDP was also found to be 500 times
stron-ger than for ThDP [31] In both enzymes studied,
binding of the analogue occurred at a faster rate than
the natural coenzyme Similar results have been
obtained with pyruvate dehydrogenase and
branched-chain 2-keto acid dehydrogenase E1 components (see
below) as well as with transketolase (K M Erixon &
F J Leeper, unpublished results)
It is initially surprising to find that 3-deazaThDP
binds much tighter than the natural coenzyme The
key to rationalizing this result is to note that it is a
good mimic of the ylid structure in which the
five-membered ring is neutral overall As a consequence of
the charge being reduced from +1 to neutral, there is
an increase in hydrophobic interactions between the
enzyme’s active site residues and the analogue, which
in turn leads to tighter binding It has been shown that
the enzyme stabilizes the neutral ylid and enamine
intermediates in the catalytic cycle by providing an
active site of low dielectric constant [32] This would
assist the formation of these neutral intermediates
from their charged precursors
In studies on PDH E1, it was found that the protein
displays half-of-sites reactivity towards proteolysis of
the active-site loops when ThDP is bound but is
pro-tected from cleavage at both active sites of the a2b2
tetramer when 3-deazaThDP is bound This
observa-tion led to the ‘proton-wire’ hypothesis in which
com-munication between the active sites is mediated by the
shuttling of a proton down a solvent-filled tunnel that
connects the two active sites [33]
3-DeazaThDP has also been used to obtain crystal
structures of the ThDP-dependent enzymes with their
substrates bound In most cases, equivalent structures
with ThDP bound could not be obtained because the
reaction occurs too quickly Thus, the structure of a
complex of oxalyl CoA decarboxylase with
3-dea-zaThDP and oxalyl CoA was solved and this revealed
both the CoA binding site and the ordering of the
C-terminus of the protein to form a lid over the CoA
chain (PDB entry 2JI6) [34] In the case of
phenylpyru-vate decarboxylase, crystallizing the enzymes with
3-deazaThDP bound allowed a structure of the enzyme
to be obtained which had substrate molecules in both the active site and the regulatory site, thus unveiling the mechanism of allosteric substrate activation [35]
As well as being a very potent inhibitor, 3-deaza-ThDP has the advantage over neutral analogues such
as ThTDP in that it can be functionalized at the C-2 position in order to make very close mimics of reaction intermediates in the catalytic cycle of various ThDP-dependent enzymes In so doing, very selective and potent inhibitors of the E1 components (E1p and E1b)
of pyruvate dehydrogenase complex (PDHc) and the related branched-chain keto acid dehydrogenase com-plex (EC 1.2.4.4) have been produced (M D H Wood and F J Leeper, unpublished results) PDHc and branched-chain 2-keto acid dehydrogenase complex belong to the highly complex 2-oxoacid dehydrogenase family of enzymes Enzymes in this family have a molecular mass of between 4 and 10· 106Da PDHc comprises three different enzymic components (E1, E2 and E3) that work in tandem to catalyse the oxidative decarboxylation of pyruvate to give acetyl CoA, CO2 and NADH as products (Fig 7) This is carried out with the aid of no fewer than five coenzymes, namely ThDP, CoA, lipoic acid, NAD+ and FAD The first reaction catalysed by PDHc is the ThDP-dependent decarboxylation of pyruvate by the E1 component, an
a2b2 heterotetramer with two active sites located at the interfaces between the a- and b-subunits [33]
3-DeazaThDP proved to be a potent inhibitor of both E1p and E1b with Kivalues of 0.14 and 0.48 nm, respectively For comparison, ThDP has a KDfor both E1 components of 3 lm The time-course of the inactivation of the holoenzyme is relatively slow, because of the slow rate of unbinding of ThDP It was observed that there is a faster initial rate of inactiva-tion followed by a much slower rate This was observed for both enzymes although it is more pro-nounced for E1b This profile may be evidence for the already mentioned half-of-sites reactivity The two active sites of the E1 component are in alternate
Fig 7 Overall reaction catalysed by the pyruvate dehydrogenase complex.
Trang 6conformational states, with one being ‘open’ and the
other ‘closed’ The more rapid phase of the
inactiva-tion may be caused by binding of the inhibitor after
ThDP has unbound from the ‘open’ conformation,
whereas the slow phase may be caused by binding of
the second molecule of inhibitor in the second active
site, which has to change from a ‘closed’ to an ‘open’
conformation to allow its ThDP to unbind
Analogues 4 and 5 (Fig 6) synthesized readily from
3-deazathiamin by Friedel–Crafts acylation at the C-2
position, followed by reduction of the ketone and
pyrophosphorylation, proved to be even better
inhibi-tors for their corresponding E1 components than
deazaThDP (Fig 5) 4 closely mimics the enamine
reaction intermediate of the E1p reaction, whereas 5
mimics the enamine intermediate of E1b For the E1p
component, analogue 4 showed excellent affinity with
a Ki value of 0.1 nm Analogue 5, by contrast, showed
no detectable inhibition, thus confirming that the E1
component of PDHc shows a high degree of selectivity
for its natural substrate pyruvate over branched-chain
keto acid substrates The opposite is true for the E1b
subunit in which analogue 5 binds with a higher
affin-ity (Ki= 0.17 nm) than 4 (Ki= 0.44 nm) The fact
that 4 binds with less affinity can be attributed to its
shorter methyl side chain which would have fewer
hydrophobic interactions with the enzyme compared
with the longer branched side chain
The selectivity of both E1 components for the R and
S enantiomers of 4 and 5 was also probed
Interest-ingly neither enzyme shows any preference for either
enantiomer of 4 but E1b shows a three- to four-fold
preference for binding the R enantiomer of 5
Crystal structures have been obtained of 4 bound in
the ThDP-binding sites of phenylpyruvate
lase [35] and the branched-chain keto acid
decarboxy-lase from Lactococcus lactis [36] These structures
have helped define the conformations that the enzymes
would adopt when the reaction mechanisms have
reached the enamine intermediates Using
3-dea-zaThDP and its substituents, as well as the phosphono
analogues mentioned above, crystal structures can now
be obtained that represent ‘snapshots’ of
ThDP-depen-dent enzymes at almost every stage of the reaction
Other synthetic analogues
The synthesis of 3-deazaThDP is a 12-step procedure
which, although producing good yields, is difficult and
time-consuming Two inhibitors with marginally less
potency than 3-deazaThDP but which can be obtained
much more easily in four synthetic steps are the
triaz-ole analogues 6 (Ki= 20 pm) and 7 (Ki= 30 pm
against ZmPDC) (Fig 6) [37] A disadvantage of these triazole compounds is that they cannot be functional-ized at the 2-position to produce analogues of reaction intermediates They do, however, provide a readily accessed scaffold for synthesizing mimics of the diphosphate moiety and this is discussed in the follow-ing section
Having established that neutral analogues of the thiazolium moiety such as 3-deazaThDP bind with a much higher affinity than the natural coenzyme and knowing that the diphosphate group is the most important group for binding strength, we wondered if
‘open-chain’ analogues such as 8 and 9 would show any potency as inhibitors (Fig 6) The ‘open-chain’ analogues have the advantage of being much more readily synthesized than their thiophene-containing counterparts Initial results from affinity studies with apo-ZmPDC suggest Ki values in the nm range (K Agyei-Owusu & F J Leeper, unpublished results) These analogues are not expected to have the potency
of 3-deazaThDP because of their higher degree of con-formational freedom, but they are readily synthesized and easily functionalized into mimics of reaction inter-mediates and, with potential Kivalues in the nm range, show much promise as selective ThDP-dependent enzyme inhibitors
ThDP analogues formed during crystallization or X-ray irradiation
In several cases, the crystal structures of ThDP-depen-dent enzymes, once solved, have revealed that the thia-zolium ring of the ThDP has degraded, presumably by hydrolysis Examples include ZmPDC (PDB entry 1ZPD) and carboxyethylarginine synthase (PDB entry 2IHU), where C-2 has been lost altogether, and both yeast and Arabidopsis thaliana acetohydroxyacid syn-thases in complex with herbicides such as metsulfuron methyl (e.g PDB entries 1YHY and 1T9D), in which the ethyl pyrophosphate side chain appears to have become detached from the pyrimidine ring with at most fragments of the thiazolium ring remaining It is worth noting that hydrolysis of the thiazolium ring of ThDP generates open-chain neutral species similar to 8 and 9 In view of the fact that the enzymes bind neu-tral species more tightly than positively charged ones,
it is reasonable to assume that the enzyme causes the hydrolysis reaction to be more thermodynamically favourable
In Klebsiella pneumoniae acetolactate synthase, the structure obtained after soaking the crystals with pyru-vate appeared to show a 2-(1-hydroxyethyl)ThDP which had cyclized from the 4¢-amino group to C-2 to
Trang 7give a dihydrothiachrome derivative (PDB entry
1OZG) Pyruvate ferredoxin oxidoreductase is an
anaerobic enzyme that is inactivated by exposure to
both pyruvate and air The structure of this inactive
complex shows electron density for a noncovalently
bound pyruvate molecule and an additional atom
apparently attached to the 4¢-NH2 (PDB entry 2C3U)
It was suggested that hydroxylation of this amino
group has occurred to give the hydroxylamine [38]
Diphosphate group mimics
The diphosphate group of ThDP forms the most
important binding interactions with the enzyme by
coordinating to Mg2+[9] It has been shown that
inor-ganic diphosphate on its own will compete with the
coenzyme for its active site ThDP binds at the
inter-face of two subunits of the enzyme, with the
diphos-phate group bound to one subunit and the
aminopyrimidine moiety bound in a cavity between the
two subunits In order to synthesize potential drug
molecules with good pharmacokinetic profiles based
on ThDP, it would be essential to make mimics of the
diphosphate group that are not as highly charged but
which retain good affinity to the enzyme Because of
its high charge, the diphosphate group would make it
difficult for an analogue to penetrate the lipid
mem-branes of cells with consequent poor uptake and
bio-availability With this in mind, a number of analogues
of ThDP have been synthesized with isosteres of the
diphosphate group (Fig 8) [37]
As mentioned above, the triazole analogues of ThDP can be readily accessed and show high affinity
to the enzyme PDC Several analogues of the triazole diphosphate 7 have been prepared and provide more evidence for the critical role of diphosphate in binding For example, there is a difference in binding to ZmPDC between the diphosphate 7 and its alcohol precursor of the order of 1· 107 The general trend observed in binding studies with the diphosphate mim-ics was a decrease in affinity with decreasing anionic charge The methylene diphosphonates 10 and 11, which are trianionic, show good binding affinity rela-tive to the other diphosphate mimics studied, with Ki values only 30–40 times greater than that of 7 There is a successive decrease in affinity going from 10 and 11 (trianionic) to the phosphoramidic analogue 12 (dianionic), to carbamate 13 and malonate 14 (mono-anionic), to the iminodiacetate analogue 15, for which
no binding was observed
Pyrithiamin, riboswitches and thiamin pyrophosphokinase
Pyrithiamin (Fig 9) was first synthesized in 1941 by Tracy and Elderfield [39], the culmination of efforts by
a number of groups to produce the pyridinium isostere
of the thiazolium moiety in ThDP A 0.5 mg dose of this analogue was found to be lethal when adminis-tered to a newly weaned mouse, whereas a 0.17 mg dose retarded growth in the mouse [40] Over the years, it has been used to produce symptoms of thia-min deficiency in mice [41] and has been shown to be toxic to species of fungi, algae and bacteria [42–44] In enzymatic studies, pyrithiamin diphosphate was found
to be only a modest inhibitor of apo-TK (Ki= 110 lm) [20] and apo-PDC (Ki= 78 lm) [21], which suggested that its antithiamin effects were not the result of competition with ThDP for binding to the enzymes Moreover, a number of organisms that bio-synthesize thiamin and would be expected to upregu-late its production in the presence of pyrithiamin were found to be unable to do so [43] As far back as 1976,
Fig 8 Diphosphate mimics based on the triazole analogue of
Trang 8pyrithiamin was shown to disrupt the regulatory
mech-anisms of thiamin biosynthesis, causing a decrease in
thiamin production, although the underlying
mecha-nism for this phenomenon remained unknown
The effects of pyrithiamin on microorganisms at the
molecular level became clearer with the discovery of
ThDP riboswitches [7] Riboswitches (Fig 10) are
ele-ments of mRNA that form receptors for a specific
ligand which, when bound, controls expression of the
gene(s) encoded in the mRNA A number of different
classes of riboswitches have been discovered, each class
named after its specific ligand ThDP riboswitches are
generally sited upstream of the genes coding for the
enzymes involved in thiamin biosynthesis, transport
and the salvaging of precursors to the thiazole and
pyrimidine moieties of ThDP Depending on the
organ-ism, they can, when bound to their ligand, act either as terminators of transcription (e.g in Bacillus subtilis) or
as inhibitors of translation (by masking the ribosome binding site, e.g in E coli) or to cause mis-splicing of the RNA (in eukaryotes) resulting in premature termi-nation of translation or instability of the mRNA [45]
It has been shown that pyrithiamin is taken up into cells by thiamin transporters and pyrophosphorylated
by the enzyme TPK before binding to the ThDP ribo-switch Pyrophosphorylation of pyrithiamin by TPK has also been carried out in vitro [46] On binding of pyrithiamin diphosphate to the riboswitch, the conse-quent reduction in the levels of thiamin biosynthetic enzymes leads to the depletion of intracellular thiamin levels, preventing cell growth Pyrithiamin-resistant mutants of the bacteria B subtilis and E coli [43], the fungus Aspergillus oryzae [47] and the alga Chlamydo-monas reinhardtii [44] have all been characterized and
in each case it turns out that the mutation is in a ThDP riboswitch This causes the corresponding biosynthetic gene(s) to be constitutively expressed, no longer under the control of the riboswitch
Three separate groups have obtained crystal struc-tures of ThDP riboswitches, two using the ThiM ribo-switch from E coli [48,49] (Fig 10) and one a riboswitch from Arabidopsis thaliana [50] Two sepa-rate domains (P2⁄ P3 and P4 ⁄ P5) recognize the amino-pyrimidine and diphosphate moieties of ThDP, respectively Binding of ThDP brings these two domains together and causes the helix P1 to form, which is what ultimately controls expression of the gene(s) Interestingly, it appears that there is little binding of the thiazolium moiety of ThDP and this provides scope for the development of analogues of the coenzyme that can interact with the riboswitch In the case of the E coli riboswitch, crystal structures were obtained with ThDP analogues thiamin mono-phosphate, benfotiamine and pyrithiamin bound [49], whereas for the A thaliana riboswitch structures were solved with oxythiamin diphosphate and pyrithiamin diphosphate as ligands [50]
It is known that several types of antibiotics act by binding to ribosomal RNA [51] Riboswitches likewise have well-defined and conserved 3D structures and so pose attractive targets for the design of drug mole-cules, particularly as they do not, as far as is known, occur in animals [45]
ThDP-dependent enzymes as drug targets
Despite the essential nature of ThDP-dependent enzymes and the existence of several enzymes that
A
B
Fig 10 (A) Secondary structure of the ThiM riboswitch from
Escherichia coli Conserved nucleotides are shown in red and
conserved secondary structure in blue Helices are labelled P1 to
P5 (B) Structure of the ThDP-binding site (from PDB entry 2HOJ).
Trang 9occur in microorganisms but not in animals, inhibitors
of ThDP-dependent enzymes have not yet found use
as antibacterials However, it has been found that
three separate classes of herbicides (including
metsulfu-ron-methyl; Fig 11), which had been in use for many
years, all act by inhibiting acetohydroxyacid synthase,
the first enzyme in branched-chain amino acid
biosyn-thesis Crystal structures of these herbicides bound to
acetohydroxyacid synthase have been obtained and
show that they do not bind in the ThDP-binding site
but instead bind in the mouth of the active site,
block-ing access to the ThDP [52] Clomazone is a
commer-cially available herbicide that acts by interfering with
terpene biosynthesis Only relatively recently has it
been shown to act by inhibiting deoxyxylulose
5-phos-phate synthase, the first enzyme of the non-mevalonate
pathway to terpenes [53] However the inhibitor is not
clomazone itself but a metabolite 5-ketoclomazone
Amprolium (Fig 9) is a relative of pyrithiamin and
is widely used for the treatment or prevention of
coc-cidiosis in cats, dogs, poultry and cattle It acts by
blocking thiamin uptake into cells much more
effec-tively in the parasite than in the host animal [54] The
antibiotic metronidazole, used to treat infections by
anaerobic bacteria and protozoa, presents an
interest-ing example of a novel thiamin analogue It turns out
that metronidazole can be a substrate for thiaminase
(which catalyses the displacement of the thiazole unit
of thiamin by various nucleophiles) to form a close
analogue of thiamin (Fig 12) [55] This compound
was found to inhibit thiamin pyrophosphokinase and
it is possible that the side-effects of prolonged use
of metronidazole are caused by the resulting ThDP
deficiency
One ThDP-dependent enzyme that has gained
con-siderable attention in the last decade has been TK
This is because it has been identified as playing an
important role in cell proliferation in a number of
tumour lines TK is involved in the nonoxidative pen-tose 5-phosphate pathway that leads to formation of the ribose required for the synthesis of nucleotides and, therefore, the expression of TK and of other enzymes of the pathway is upregulated in cancer cells
It has been shown both in vitro and in vivo that target-ing TK with drugs based on ThDP leads to a decrease
in tumour growth [56] In tests on mice, oxythiamin 16 (Fig 13) was found to inhibit Ehrlich ascites tumour cell proliferation by 84% at a dose of 500 mgÆkg)1 The effect is enhanced when oxythiamin is adminis-tered alongside dehydroepiandrosterone, an inhibitor
of the oxidative pentose 5-phosphate pathway, suggest-ing that the oxidative pathway is also important
Fig 11 Structures of herbicides metsulfuron-methyl and
cloma-zone.
Fig 12 Reaction of metronidazole with thiamin catalysed by thiaminase.
Fig 13 Structures of oxythiamin, N-3¢-pyridyl-thiamin and its disul-fide prodrug 18.
Trang 10A large number of thiamin analogues have been
syn-thesized and tested for inhibition of TK in vitro, with
and without pyrophosphorylation by TPK, and in vivo,
as well as for their effect on tumour growth These
analogues included both uncharged compounds like
3-deazathiamin 3 [5] and analogues with a positively
charged thiazolium ring [57] The compound selected
for further development was the N-3¢ pyridyl analogue
17 (Fig 13) This has poor pharmacokinetic properties
leading to rapid clearance, however, and is better
administered as its prodrug 18 [58] (Thiamin itself has
poor oral absorption because of the low capacity of its
transporter in the gut and is often taken in similar
pro-drug form in order to treat or prevent symptoms of
thiamin deficiency.) The disulfide bond of prodrug 18
is reduced in vivo and cyclizes spontaneously to
produce 17 Both in vitro and in vivo tests on the
HCT-116 cell line showed encouraging EC50values for
the prodrug, which were better than those of its parent
thiazolium salt
Concluding remarks
Enzymatic studies involving analogues of ThDP have
revealed much information on the mechanistic
intrica-cies of various ThDP-dependent enzymes For
exam-ple, the role of hydrophobic interactions between the
enzyme and coenzyme in stabilizing reaction
intermedi-ates has become clearer, and the features of the
coen-zyme essential for binding and catalysis have been
elucidated New and intriguing features of some
enzymes, for example, half-of-sites reactivity in the E1
component of PDH, have also been discovered
Crys-tal structures of enzymes with analogues of reaction
intermediates bound have helped to define the roles of
the various amino acid side chains at different stages
of the reaction
The study of thiamin and its analogues has entered
a new phase with the discovery of riboswitches and
their regulatory role in the expression of genes of the
thiamin biosynthetic pathway This has, in turn, led to
a new set of therapeutic targets in the fight against
infections caused by pathogenic bacteria and fungi
Analogues targeting riboswitches in these species may
be less susceptible to resistance by the target organisms
because in most cases, more than one thiamin
ribo-switch exists to regulate thiamin biosynthesis, so
multi-ple mutations would be required to achieve resistance
Another promising area of research is the inhibition of
the enzyme TK which has been implicated in a number
of cancers Potent inhibitors of TK continue to be
developed and this will hopefully add to the arsenal of
chemotherapeutic drugs already available
Acknowledgements
We thank the EPSRC for a studentship to K.A.-O
References
1 Kluger R (1987) Thiamin diphosphate: a mechanistic update on enzymic and nonenzymic catalysis of decar-boxylation Chem Rev 87, 863–876
2 Lindqvist Y, Schneider G, Ermler U & Sundstro¨m M (1992) Three-dimensional structure of transketolase, a thiamine diphosphate dependent enzyme, at 2.5 A˚ reso-lution EMBO J 11, 2373–2379
3 Reed L (1974) Multienzyme complexes Acc Chem Res
7, 40–46
4 Breslow R (1957) Rapid deuterium exchange in thiazolium salts J Am Chem Soc 79, 1762–1763
5 Thomas AA, De Meese J, Le Huerou Y, Boyd SA, Romoff TT, Gonzales SS, Gunawardana I, Kaplan T, Sullivan F, Condroski K et al (2008) Non-charged thiamine analogs as inhibitors of enzyme transketolase Bioorg Med Chem Lett 18, 509–512
6 Eschbach M-L, Muller IB, Gilberger T-W, Walter RD
& Wrenger C (2006) The human malaria parasite
extended pyrophosphokinase with specificity for thia-mine Biol Chem 387, 1583–1591
7 Rodionov D, Vitreschak A, Mironov A & Gelfand M (2002) Comparative genomics of thiamine biosynthesis
in procaryotes New genes and regulatory mechanisms
J Biol Chem 277, 48949–48959
8 Golbik R, Neef H, Hubner G, Konig S, Seliger B, Meshalkina L, Kochetov GA & Schellenberger A (1991) Function of the aminopyrimidine part in thiamin pyrophosphate enzymes Bioorg Chem 19, 10–17
9 Schellenberger A (1967) The structure and mechanism
of the active centre of yeast pyruvate decarboxylase Angew Chem Int Ed 6, 1024–1035
10 Wikner C, Meshalkina L, Nilsson U, Nikkola M, Lindqvist Y, Sundstrom M & Schneider G (1994) Analysis of an invariant cofactor–protein interaction in thiamin diphosphate-dependent enzymes by site-directed mutagenesis – glutamic-acid-418 in transketolase is essential for catalysis J Biol Chem 269, 32144–32150
11 Kern D, Kern G, Neef H, Tittmann K, Killenberg Jabs
M, Wikner C, Schneider G & Hubner G (1997) How thiamine diphosphate is activated in enzymes Science
275, 67–70
12 Meshalkina LE, Kochetov GA, Brauer J, Hu¨bner G, Tittmann K & Golbik R (2008) New evidence for cofactor’s amino group function in thiamin catalysis by transketolase Biochem Biophys Res Commun 366, 692– 697
13 Nemeria N, Chakraborty S, Baykal A, Korotchkina LG, Patel MS & Jordan F (2007) The 1¢,4¢-iminopyrimidine