Inhibition of thiamin diphosphate dependent enzymes by 3-deazathiamin diphosphate.

Article abstract of DOI:10.1039/b403619k

3-Deazathiamin diphosphate (deazaTPP) and a second thiamin diphosphate (TPP) analogue having a benzene ring in place of the thiazolium ring have been synthesised. These compounds are both extremely potent inhibitors of pyruvate decarboxylase from Zymomonas mobilis; binding is competitive with TPP and is essentially irreversible even though no covalent linkage is formed. DeazaTPP binds approximately seven-fold faster than TPP and at least 25,000-fold more tightly (K(i) less than 14 pM). DeazaTPP is also a potent inhibitor of the E1 subunit of alpha-ketoglutarate dehydrogenase from E. coli and binds more than 70-fold faster than TPP. In this case slow reversal of the inhibition could be observed and a K(i) value of about 5 nM was calculated (ca. 500-fold tighter binding than TPP).

Full text of DOI:10.1039/b403619k

View Article Online / Journal Homepage / Table of Contents for this issue
Inhibition of thiamin diphosphate dependent enzymes by
3-deazathiamin diphosphate
Stéphane Mann, Concepcion Perez Melero, Dan Hawksley and Finian J. Leeper*
University Chemical Laboratory, Lensfield Road, Cambridge, UK CB2 1EW
Received 17th March 2004, Accepted 22nd April 2004
First published as an Advance Article on the web 25th May 2004
3-Deazathiamin diphosphate (deazaTPP) and a second thiamin diphosphate (TPP) analogue having a benzene ring
in place of the thiazolium ring have been synthesised. These compounds are both extremely potent inhibitors of
pyruvate decarboxylase from Zymomonas mobilis; binding is competitive with TPP and is essentially irreversible even
though no covalent linkage is formed. DeazaTPP binds approximately seven-fold faster than TPP and at least 25,000-
fold more tightly (Ki less than 14 pM). DeazaTPP is also a potent inhibitor of the E1 subunit of α-ketoglutarate
dehydrogenase from E. coli and binds more than 70-fold faster than TPP. In this case slow reversal of the inhibition
could be observed and a Ki value of about 5 nM was calculated (ca. 500-fold tighter binding than TPP).
oxidases and dehydrogenases oxidation of the enamine can
occur. Finally release of the product, acetaldehyde in this case,
occurs with regeneration of the ylid 2.
Introduction
Thiamin diphosphate (TPP) is a coenzyme used by a range of
different enzymes which have one thing in common: they make
Although the mechanisms of TPP-dependent reactions are
and break bonds to the carbon atom of a carbonyl group.
well known, the details of how the enzymes catalyse these
Examples include α-keto acid decarboxylases, dehydrogenases
mechanisms are not really understood. Many crystal structures
and oxidases, transketolase, deoxyxylulose synthase, benzalde-
of such enzymes have been obtained with TPP bound but in
hyde lyase and acetohydroxy acid synthase.1 The mechanisms
general crystal structures that show the substrate bound or
for all TPP-dependent enzymic reactions are similar and are
intermediates in the reaction are lacking. Exceptions are
illustrated in Scheme 1 by the mechanism for pyruvate decarb-
structures of pyruvate bound to pyruvate:ferredoxin oxido-
reductase2 (although it may be in a non-productive complex)
oxylase (PDC). In all cases the reaction is initiated by
and separate structures of erythrose 4-phosphate3 and the
active aldehyde intermediate4 bound to transketolase. In
addition structures have been solved of a radical species related
to 2-acetylTPP bound to pyruvate:ferredoxin oxidoreductase5
and of a cyclised form of 2-hydroxyethylTPP 5 bound to
acetolactate synthase6 but in neither case is it clear that it is a
true reaction intermediate that is present. The reason why it is
difficult to get such crystal structures is that the substrate and
intermediates normally undergo reactions too fast to be
detected by crystallography. One solution to this problem
would be to use an inactive analogue of the coenzyme. With
this in mind, we recently reported the synthesis of 3-deaza-
thiamin 6, which is isoelectronic with thiamin and therefore has
the same size and shape but differs only in its charge. Here we
report the conversion of 3-deazathiamin into its diphosphate
(deazaTPP) 15 and studies of the interaction of this compound
with two TPP-dependent enzymes, pyruvate decarboxylase
from Zymomonas mobilis and the E1 subunit of α-ketoglutarate
dehydrogenase from E. coli. We also report the synthesis and
studies of a second inactive analogue of TPP which has a
benzene ring in place of the thiazolium ring.
deprotonation of the TPP 1 to give the ylid 2. This then nucleo-
philically attacks the carbonyl group of the substrate, giving in
the case of PDC lactyl-TPP 3. Bond cleavage can then occur
with the thiazolium ring of the TPP moiety acting as an
electron sink; in enzymes like PDC which operate on α-ketoac-
ids it is CO2 that is lost but in some enzymes (e.g. transketolase,
benzaldehyde lyase) an aldehyde molecule can be lost instead.
The product of the bond cleavage is an enamine 4, often called
the “active aldehyde” intermediate. This then reacts with an
electrophile, which in the case of PDC is simply a proton giving
hydroxyethyl-TPP 5. In other examples a ketone or aldehyde
group of another substrate molecule is the electrophile, while in
Results
Synthetic studies
Although our previously reported synthesis of deazathiamin7
was relatively efficient, the synthesis of the thiophene ring
inevitably involved a number of steps. We thought it would be
interesting therefore to compare the inhibitory properties of
deazathiamin pyrophosphate (deazaTPP) with an analogue of
Scheme 1 Mechanism of pyruvate decarboxylase.
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 7 3 2 – 1 7 4 1
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 4
1732

View Article Online
TPP which was simpler to make. Accordingly, we synthesised
13, an analogue of thiamin having a benzene ring in place of
the thiazolium ring.
The synthesis of 13 started with m-toluoyl chloride 7
(Scheme 2). Bromination of the methyl group8 followed by
quenching the reaction mixture with methanol gave the bromo
ester 8. Nucleophilic substitution of the bromide by cyanide
followed by methanolysis of the nitrile 9 then gave the
diester 10. Reduction of both esters with LiAlH4 followed by
reoxidation of the benzylic alcohol with MnO2 gave the
aldehyde 11. Formation of the pyrimidine ring then followed
the method we reported earlier:7 base-catalysed condensation
of 11 with 3-anilinopropionitrile gave the acrylonitrile 12
as a mixture of cis and trans isomers. Finally reaction with
acetamidine and base gave the pyrimidine 13.
Scheme 3 Reagents: i, TsCl, pyr, 80–86%; ii, (Bu4N)3HP2O7, CH3CN,
72% (for 15).
conditions gave a 72% yield of deazaTPP 15 after anion
exchange chromatography, along with the elimination product
18 in 12% yield.
Studies with pyruvate decarboxylase
Inactivation of holoenzyme. Our first set of enzymic studies
employed pyruvate decarboxylase (PDC) from Zymomonas
mobilis, which had been overexpressed in E. coli and purified to
homogeneity as described previously.12 The holoenzyme in a
TPP-containing buffer was incubated with various concen-
trations of TPP analogue (15 or 17) and samples were taken at
intervals and assayed for residual activity. The inactivation of
the enzyme was slow – typically it was followed for 5 to 7 hours
and the activity was continuing to decrease at the end of this
time (Fig. 1). The enzyme activity decayed exponentially in
Scheme 2 Reagents: i, Br2 then MeOH, 65%; ii, KCN, 18-crown-6,
CH3CN, 91%; iii, MeOH, HCl, 87%; iv, LiAlH4, Et2O, 60% then MnO2,
CHCl3, 91%; v, PhNHCH2CH2CN, NaOMe, DMSO–MeOH, 48%;
vi, CH3C(᎐NH)NH2, NaOMe, EtOH, 81%.
these experiments and apparent first order rate constants (kapp
)
᎐
were obtained by fitting exponential curves to the data. It
was found that at relatively low concentrations of TPP (e.g.
1.38 µM) the rate of inhibition was essentially the same for all
concentrations of deazaTPP (1 to 15 µM) (not shown), whereas
with higher concentrations of TPP (e.g. 15 µM) the rate of
With deazathiamin 6 and the benzene analogue 13 in hand,
we now turned our attention to formation of the diphosphates.
Initially we tried the method that has been used for the
pyrophosphorylation of thiamin and related compounds such
as thiamin thiazolone.9,10 This method is to heat concentrated
phosphoric acid to drive off some water and then add
thiamin (or an analogue) and continue the heating. The result-
ing mixture contains mono-, di- and tri-phosphates of the
thiamin analogue which need to be separated from each other
and the vast excess of inorganic phosphates by anion exchange
chromatography. With the benzene analogue this procedure
did give the desired diphosphate, albeit in a low 19% isolated
yield. However, with deazathiamin decomposition apparently
occurred under these extremely acidic conditions and none of
the desired diphosphate could be isolated. A different approach
was clearly needed.
The alternative method that was now tried employed SN2
displacement of a good leaving group by a pyrophosphate ion.
This method is well known for allylic diphosphates but can also
be extended to other compounds, such as nucleoside diphos-
phates.11 Accordingly the alcohol groups of deazathiamin 6 and
the benzene analogue 13 were tosylated and the tosylates 14 and
16 (Scheme 3) were treated with tris(tetra-n-butylammonium)
hydrogen pyrophosphate in acetonitrile at room temperature.
Purification of the diphosphates by anion exchange chromato-
graphy was much easier in this case and both deazaTPP 15 and
its benzene analogue 17 could be isolated in pure form,
although still in rather low yield (33% and 24% respectively).
The major by-product in the formation of deazaTPP was found
to be the elimination product 18 (up to 70%). On varying the
temperature of the reaction it was found that the proportion of
elimination product increased with increasing temperature
whereas the proportion of the desired substitution product
increased on lowering the temperature and also on raising
the concentration of the pyrophosphate salt. The optimum
Fig. 1 Time-dependent inactivation of holo-PDC by deazaTPP (a) 1
µM, (b) 2 µM, (c) 5 µM, (d ) 10 µM; in each case TPP (15 µM) was
present. Inset: plot of kapp for the inactivation vs. [deazaTPP]/[TPP]; the
expected saturation kinetics are seen (eqn. 2), giving koff for TPP = 5.1 ×
10Ϫ3 minϪ1 and the ratio of kon for deazaTPP (ki) to kon for TPP to be
6.9.
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 7 3 2 – 1 7 4 1
1733

View Article Online
inhibition by the lower concentrations of deazaTPP was signifi-
cantly slower (Fig. 1). This is as expected if the binding of
deazaTPP is competitive with that of TPP, as shown in Scheme
4. Starting with PDC with TPP bound, TPP has to dissociate
first before deazaTPP can bind and inhibit the enzyme (hence
the very slow inactivation). Once the TPP has dissociated then
there is competition between deazaTPP and TPP for rebinding
to the apoenzyme. At the lowest concentration of TPP, deaza-
TPP binds faster than TPP at all concentrations of deazaTPP
tested; therefore the rate of inactivation is approximately
equal to the rate of dissociation of TPP and independent of
[deazaTPP]. However if the [TPP] is raised then it can compete
effectively with the lower concentrations of deazaTPP for
rebinding to the apoenzyme and thus the rate of inactivation is
slower.
Scheme 4 Kinetic scheme for the inactivation of the holo-PDC by
deazaTPP.
Fig. 2 Change in fluorescence when deazaTPP binds to apo-PDC; the
solid lines are the experimental data and the dotted lines are the
best-fit exponential decay curves; deazaTPP concentrations: (a) 1 µM,
(b) 2 µM, (c) 4 µM, (d) 8 µM, (e) 10 µM. Inset: plot of kapp for the
fluorescence change vs. [deazaTPP]; the slope of the graph gives kon for
deazaTPP (ki) and the intercept on the y-axis gives koff for deazaTPP.
The competition between deazaTPP and TPP was further
demonstrated by adding additional TPP, up to a concentration
of 50 or 100 µM, half way through one of these time-course
experiments. The result was that inactivation of the enzyme was
virtually halted but no activity that had already been lost was
recovered, suggesting that binding of deazaTPP is essentially
irreversible.
as shown in Fig. 2. The graph of kapp vs. [deazaTPP] is given in
the inset to Fig. 2. The slope gives a value of ki for deazaTPP
of 0.28 µMϪ1 minϪ1, similar to the value found by the competi-
tive inhibition experiments. The koff value for dissociation of
deazaTPP is given by the intercept on the y-axis. It can be seen
that this intercept is within experimental error of zero, again
indicating that the binding of deazaTPP is more or less
irreversible.
The kinetic scheme of Scheme 4 would give an apparent first
order rate constant for inactivation that is given by equation 1:
(1)
The total decrease of fluorescence associated with binding of
deazaTPP was roughly constant whenever the molar ratio
of deazaTPP to active sites (assuming four active sites per
tetramer) was 1 : 1 or greater and addition of TPP at the end
of the incubation caused no further decrease in fluorescence. If
the molar ratio of deazaTPP to active sites was less than 1 : 1,
however, then a smaller decrease of fluorescence was observed
and addition of TPP at the end of the incubation did cause a
further decrease in fluorescence. This shows that the stoichi-
ometry of bound deazaTPP to active sites is 1 : 1.
Rearranging equation 1, one gets equation 2:
(2)
Hence a graph of kapp vs. [deazaTPP]/[TPP] should show
Michaelis–Menten-like saturation kinetics and fitting the
appropriate curve to the experimental data will give values for
koff and ki/kon. This graph is shown in the inset to Fig. 1. From
the curve-fitting a value for koff of 5.1 × 10Ϫ3 minϪ1 and for
ki/kon of 6.9 can be obtained. This means that deazaTPP binds
nearly seven times faster than TPP does at the same concen-
tration. The rate constant for binding of TPP has previously
been reported13 to be 0.035 µMϪ1 minϪ1, which gives a value for
In order to find out whether the binding shows any
reversibility at all, apoPDC was incubated with deazaTPP
until there was no measurable activity (after addition of TPP
to reconstitute holoenzyme from any remaining apoenzyme).
Excess deazaTPP was removed by ultrafiltration and then TPP
(100 µM) was added. After 24 hours at 30 ЊC enzymic assay still
showed no measurable activity. We estimate that the activity
must have been less than 0.4% of the initial activity. This
upper limit on the recovered activity implies that the rate of
ki of 0.24 µMϪ1 minϪ1
.
Binding to apoenzyme. In view of the fact that binding of
the deazaTPP to holoPDC is so slow, because of the slow dis-
sociation of the already bound TPP, we decided to study the
binding of the analogues to apoPDC. The apoenzyme can be
prepared by adjusting the pH to 8.2 and then separating the
cofactor from the protein by either dialysis13 or gel filtration12
or, in this case, by repeated ultrafiltration. Two methods were
then used to follow binding of the TPP analogues, fluorescence
quenching and enzyme activity.
It has previously been shown that binding of TPP to
apoPDC from Z. mobilis results in a decrease in the fluor-
escence due to the tryptophan residues in the protein.13 The
fluorescence decreases over time with an apparent first-order
rate constant given by kapp = kon [L] ϩ koff, where L is the ligand
that is binding to the protein. We found that binding of the TPP
analogues, 15 and 17, to apoPDC also causes a decrease in
fluorescence. The change in fluorescence was followed over time
at 30 ЊC for a range of different concentrations of deazaTPP 15,
dissociation of deazaTPP is less than 2.8 × 10Ϫ6 minϪ1
.
Combining this with the measured rate of association (ki = ca.
0.2 µMϪ1 minϪ1), the inhibition constant Ki must be less than
14 × 10Ϫ12 M (14 pM).
In order to demonstrate that the effectively irreversible
inhibition is not due to any covalent attachment to the protein,
we attempted to remove the inhibitor from the enzyme. Initially
the inhibited enzyme was subjected to the same treatment as
was used for the removal of TPP to form apoPDC but no activ-
ity was recovered. As this method did not work, enzyme which
had been partially inactivated by treatment of the holoenzyme
with deazaTPP was instead denatured by incubation in 6.8 M
urea.14 The deazaTPP in solution was then removed by ultra-
filtration. Finally the enzyme was renatured by dilution with
buffer containing TPP but no urea. By this procedure the
enzyme activity, which had been reduced to 60% of the original
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 7 3 2 – 1 7 4 1
1734

View Article Online
value after the treatment with deazaTPP, was restored to ca.
90% of the original activity. Thus it is shown that deazaTPP can
be removed if the enzyme is denatured.
The effectively irreversible binding of deazaTPP to apoPDC
allowed us to follow the binding by measurement of enzymic
activity after addition of excess TPP. However, in order for
these measurements to be meaningful it was necessary to show
that the enzymic activity is proportional to the concentration of
free apoenzyme. PDC from Z. mobilis is a tetramer, made up of
a dimer of dimers, and there are four active sites per tetramer
located at the interface between the subunits in each dimer.15 It
was possible, therefore, that binding of deazaTPP in one active
site of the tetramer might affect the activity at one or more of
the other active sites. ApoPDC was incubated with various
substoichiometric concentrations of deazaTPP for 3 h (long
enough for effectively all the deazaTPP to bind) and then the
solution was added to the assay mixture containing TPP and
the remaining enzyme activity was measured. The resulting
graph of residual activity vs. [deazaTPP] (Fig. 3) is a straight
line with complete inactivation occurring at approximately 1 : 1
molar ratio of deazaTPP to active sites. The linearity of the
graph suggests that occupation of one site in the tetramer does
not affect the enzymic activity at any of the other sites.
Fig. 4 Binding of deazaTPP to apo-PDC followed by measuring the
loss of enzymic activity after reconstitution of the residual apoenzyme
with TPP; deazaTPP concentrations were from top curve to bottom
curve, 1.92 µM, 2.24 µM, 2.56 µM, 2.88 µM, 3.2 µM and 5 µM. Inset:
plot of kapp for the loss of activity vs. [deazaTPP]; the slope of the graph
gives kon for deazaTPP (ki).
Scheme
5
Possible two-step inactivation of the holoenzyme by
benzene analogue 17.
after addition of excess TPP will be due to the free enzyme (E)
present when the TPP is added plus a certain fraction of the
initial complex (E.17) which reverts to free enzyme rather than
going on to the irreversibly inhibited complex (E.17*). This
fraction is given by kϪ1/(kϪ1 ϩ k2).
The kinetics of Scheme 5 were simulated using a spreadsheet
program and the parameters k1, kϪ1 and k2 were adjusted to get
the best fit to the experimental data. As shown in Fig. 5, a
very good fit was obtained when k1 = 0.264 µMϪ1 minϪ1, kϪ1
=
Fig. 3 Graph of residual activity vs. molar ratio of deazaTPP to
enzyme active sites; the straight line shows that complete inactivation
occurs at a 1 : 1 molar ratio and that the enzymic activity at any active
site is not affected by occupation of other sites in the tetramer.
0.06 minϪ1 and k2 = 0.36 minϪ1. This two-step kinetic scheme is
in contrast to the one-step process suggested for deazaTPP.
Presumably the first step of binding (k1), which is partially
reversible for the benzene analogue 17, is effectively an irrevers-
ible step for deazaTPP 15 due to its better fit in the active site. If
there is a subsequent step in deazaTPP binding, corresponding
The binding of deazaTPP to apoPDC was then followed by
measuring enzymic activity in this way, complementing the
fluorescence-based method described above. In these experi-
ments apoPDC was incubated with various concentrations
of deazaTPP and samples were taken at timed intervals and
assayed. For each deazaTPP concentration the enzymic activity
decayed exponentially over time (Fig. 4) giving an apparent first
order rate constant, kapp. Plotting kapp against [deazaTPP] (inset
to Fig. 4) gave a straight line, whose slope gives the second
order rate constant of 0.16 µMϪ1 minϪ1. This value is close to
the value of 0.24 µMϪ1 minϪ1 from the competitive inhibition
experiments.
The benzene analogue 17 was also tested for binding to
apoPDC. With this compound also, it was found that binding is
effectively irreversible. The stoichiometry was also studied and
again there was a linear relationship between [inhibitor] and
residual activity with complete inactivation occurring at a 1 : 1
molar ratio of inhibitor to active sites. When the time-course
for inactivation by 17 was studied, it was found that the points
did not fit well onto a simple exponential decay curve. The
initial decay was too rapid and the later decay was too slow.
This suggested a two-step process in which the faster first step
is partially reversible and the slower second step (presumably
a conformational change) is irreversible, as in Scheme 5.
According to this scheme the residual activity that is measured
Fig. 5 Binding of the benzene analogue 17 to apo-PDC followed by
measuring the loss of enzymic activity after reconstitution of the
residual apoenzyme with TPP. The curves are the best fit obtained in
a simulation of the two-step kinetic model shown in Scheme 5 with
k1 = 0.26 µMϪ1 minϪ1, kϪ1 = 0.06 minϪ1 and k2 = 0.36 minϪ1
.
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 7 3 2 – 1 7 4 1
1735

View Article Online
to k2 for binding of 17, it is not observed in the inhibition
kinetics because the enzyme is already irreversibly inhibited
before it occurs.
TPP, the lowest concentration at which the enzyme appeared
stable.
In the inhibition experiments, the enzyme was preincubated
with TPP to ensure that it was all in the holoenzyme form and
then deazaTPP was added. Samples were removed at timed
intervals and assayed for remaining enzymic activity. At all
deazaTPP concentrations the activity decayed exponentially as
shown in Fig. 6. As with PDC, the higher concentrations of
deazaTPP all gave effectively the same rate of inhibition, which
would be the rate of unbinding of the already bound TPP.
Only when the concentration of deazaTPP was lowered to 0.25
µM or below was a significant retardation of the inhibition
observed. Fitting a Michaelis–Menten-type curve to the plot of
kapp vs. [deazaTPP] (inset to Fig. 6), as before, gave values of koff
for TPP of about 0.07 minϪ1 and of ki/kon of 74. Comparing
these values with those obtained for PDC one finds that the
TPP complex is more labile for E1o (unbinding is more than
10-fold faster) and the ratio of the rates of deazaTPP binding
to TPP binding is more than ten times greater for E1o than
PDC.
ꢀ-Ketoglutarate dehydrogenase
α-Ketoglutarate dehydrogenase is a multienzyme complex
which catalyses the NADϩ-dependent conversion of α-ketoglu-
tarate and coenzyme A into succinyl CoA. The E1 subunit
(known as E1o) is the TPP-dependent α-ketoglutarate decarb-
oxylase. In the multienzyme complex, the active aldehyde
intermediate 19 reacts with a lipoyl group attached to a lipoyl
carrier domain of the E2o subunit instead of being protonated
as the active aldehyde intermediate 4 is in Scheme 1. In this
work we used purified E1o from E. coli free of E2o (gift of
R. Frank, Department of Biochemistry). To assay the activity
of this subunit, therefore, an alternative oxidising agent must be
present in place of the lipoyl group. We employed dichloro-
phenolindophenol (DCPIP) 21 for this purpose, which has
previously been used to assay the activity of the E1 subunit of
the pyruvate dehydrogenase complex (E1p).16 The active alde-
hyde intermediate 19 formed upon decarboxylation is oxidised
by the DCPIP to give the acyl thiazolium salt 20, which rapidly
hydrolyses to regenerate the TPP ylid 2 (Scheme 6). The activity
is followed by monitoring the decrease in the absorbance due to
DCPIP at 600 nm.
Fig.
6
Time-dependent inactivation of holo-E1o by deazaTPP;
deazaTPP concentrations used were (a) 0.025 µM, (b) 0.05 µM, (c) 0.1
µM, (d) 0.25 µM, (e) 1 µM, (f ) 2.5 µM, (g) 5 µM. Inset: plot of kapp for
the inactivation vs. [deazaTPP] shows the expected saturation kinetics
and allows koff for TPP and the ratio of kon for TPP to kon for deazaTPP
(ki) to be measured.
Scheme 6 Reaction catalysed by the E1 subunit of α-ketoglutarate
dehydrogenase in the presence of DCPIP 21.
In the planned competitive inhibition experiments we did not
want to use any higher concentration of TPP than necessary as
this would only slow down the inhibition, still further. We there-
fore preincubated the enzyme with various concentrations of
TPP (5 to 100 µM) and in each case samples were taken for
assay after various times of preincubation at 30 ЊC. For TPP
concentrations of 25, 50 and 100 µM the enzymic activity
remains constant from the first sample time (10 min) onwards,
but for 5 and 10 µM [TPP] the enzymic activity was highest
after 10 min and decreased gradually thereafter. It would seem
that the binding of TPP to any apoenzyme present in the puri-
fied enzyme is complete within the first 10 min but that at a TPP
concentration of 10 µM or below there is some apoenzyme in
equilibrium with the holoenzyme and the apoenzyme is not
stable at 30 ЊC resulting in a fall of activity with time. Plotting
the initial enzyme activity measurements against concentration
and fitting a Michaelis–Menten equation (not shown) gave an
apparent KM value for TPP of about 2 to 3 µM. However this
figure is only approximate because of the instability of the
enzyme at lower [TPP] values. As a result of these experiments,
all subsequent assays were performed in the presence of 25 µM
The reversibility of the inhibition of E1o by deazaTPP
was then explored by incubating E1o with deazaTPP and
TPP (50 µM) for 30 min and then raising the concentration of
TPP to 1 mM. After a further 30 min the enzymic activity had
risen from 23% to 34% of the original activity. This system
would not have reached equilibrium in the second 30-minute
period as the unbinding of deazaTPP is slow. Nevertheless this
value along with the previously estimated values of KM and koff
for TPP and of ki/kon allow a rough estimate of all the rate
constants to be made as follows. For TPP a value for kon of
ca. 0.03 µMϪ1 minϪ1 can be calculated from the measured koff
value of 0.07 minϪ1 and the observed apparent KM value of 2 to
3 µM; for deazaTPP the kon value (i.e. ki) can be calculated as
2.22 µMϪ1 minϪ1 from the measured ki/kon value of 74. Finally
a koff value for deazaTPP can be derived by trial and error
because the expected amount of recovered activity after 30 min
in the presence of excess TPP can be calculated from any
particular estimated value of koff. A koff value for deazaTPP
of ca. 0.01 minϪ1 would give the observed 34% activity after
30 min, whereas a lower koff value would give less recovered
activity and a higher koff value would give more. A Ki value
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 7 3 2 – 1 7 4 1
1736

View Article Online
for deazaTPP of about 5 nM can then be derived from the kon
and koff values. Thus for this enzyme deazaTPP 15 binds
approximately 500 times more tightly than TPP 1.
the yeast crystal structure. The presence of a counterion to the
positively charged thiazolium salt makes it somewhat surprising
that neutral species bind so well in place of the thiazolium
salt. One can only assume that this glutamate residue is in its
protonated form when neutral species are bound. This conclu-
sion would equally apply to the neutral species in the reaction
mechanism, i.e. ylid 2 and enamine 4. This is in accord with the
conclusions drawn from their molecular mechanics calculations
by Lobell and Crout.19 These authors showed this glutamate as
being protonated when the ylid 2 attacks the pyruvate molecule
and staying protonated until the protonation of the enamine 4,
at which point the glutamate provides the required proton via
an intervening water molecule.
It should be noted that there are other TPP analogues having
a neutral ring in place of the thiazolium ring which have been
studied before, notably thiamin thiazolone and thiazolethione
diphosphates9,20–23 23 and 24 and tetrahydrothiamin diphos-
phate24 25. The results reported with these inhibitors are less
complete but, as far as they go, broadly similar to those
reported here, i.e. the inhibitors are competitive with TPP and
bind much more strongly. With the E1 subunit of pyruvate de-
hydrogenase from E. coli, inhibition by 23 does show slow
reversal upon dialysis and a Ki value in the region of 3 to 13 nM
was calculated.23 Inhibition by 25 was also found to be
reversible over a period of 2–3 hours and the Ki value for
the most inhibitory isomer (one of the enantiomers of the
cis diastereoisomer) was thought to be about 20 nM.24 For
pyruvate decarboxylase from wheat germ, inhibition by 23
shows two-step kinetics with the second step claimed to be
irreversible22 (though no experiment was described that tested
whether slow reversal of the inhibition could be observed).
Discussion
DeazaTPP 15 has been shown to be an exceptionally strong
inhibitor of both TPP-dependent enzymes studied here. For the
E1 subunit of α-ketoglutarate dehydrogenase a dissociation
constant of about 5 nM could be determined, indicating bind-
ing 500 times tighter than that of TPP. For PDC the binding of
deazaTPP is so tight as to be effectively irreversible under
normal conditions and all that we could deduce is that the dis-
sociation constant must be in the low picomolar region or less.
This is equivalent to binding that is at least 25,000 times tighter
than that of TPP, based on the reported value of 0.35 µM
for the Kd of TPP.13 For both enzymes deazaTPP also binds
more rapidly than TPP (6.9-fold and 74-fold for PDC and E1o
respectively).
The benzene analogue 17 is also an exceptionally strong
inhibitor of PDC. It binds at almost the same rate as deazaTPP
(0.23 µMϪ1 minϪ1 compared with 0.16 µMϪ1 minϪ1) and, as with
deazaTPP, no reversal of the binding could be detected.
It is remarkable that the change of charge from ϩ1 for the
thiazolium ring of TPP to zero for the equivalent thiophene
ring of deazaTPP can make such a difference to the binding
affinity. This tight binding can be ascribed to the increased
hydrophobicity of deazaTPP compared to TPP and it is clear
that the enzyme active site forms stronger interactions with the
neutral thiophene ring than the positively charged thiazolium
ring. It makes sense that the enzyme should bind compounds
with no net charge on this ring better than ones with a posi-
tively charged ring. In this way the enzyme would promote
formation of the ylid 2 and the enamine 4 from their precursors
1, 3 and 5. This stabilisation of a neutral vs. a charged species
could be achieved simply by lowering the local dielectric con-
stant in the active site and it has been estimated17 from the
emission wavelength of a fluorescent TPP analogue bound to
yeast PDC that the active site has a dielectric constant of about
14, close to that of n-hexanol.
Crystal structures of PDC15,18 show a glutamate residue
(Glu-477 in yeast and Glu-473 in Z. mobilis) which is in appar-
ent contact with the thiazolium salt (Fig. 7). As there is no
other nearby ion-pair for Glu-477, it is reasonable to assume
that the carboxylate ion and the thiazolium ring of TPP form
an intimate ion pair. The distance from N-3 of TPP to the
nearest oxygen of the glutamate residue is only 3.4 Å in
Thus the inhibition observed with deazaTPP is very much in
line with previous studies on TPP analogues 23 to 25. The
big advantage that deazaTPP has over these other inhibitors,
however, is that it should be possible to attach a variety of
groups to C-2 of deazaTPP that match the intermediates in the
various reaction mechanisms. We have already synthesised the
2-(1-hydroxyethyl) derivative of deazaTPP (the deaza analogue
of intermediate 5 in Scheme 1) and studies of the effect of this
compound on PDC are in progress and will be reported in a
future publication.
Experimental
General directions for synthesis
Melting points were determined on a Reichert melting point
apparatus and are uncorrected. IR spectra were recorded on a
Perkin-Elmer 297 FTIR spectrophotometer, using sodium
chloride plates for thin film spectra and Nujol mulls. Proton
NMR spectra were recorded on a Bruker AM/DPX 250 (250
MHz), a Bruker AM/DPX 400 (400 MHz) or a Bruker DPX
500 (500 MHz) spectrometer. All coupling constants (J) are
given in Hz. Carbon NMR spectra were recorded with proton
decoupling on either a Bruker AC/DPX 250 (62 MHz), a
Bruker AC/DPX 400 (100 MHz) or a Bruker DPX 500 (125
MHz) spectrometer. APT spectra (J-resolved spin echo) were
also run to assist in the assignments. Chemical shifts (δ) are
reported in ppm downfield from tetramethylsilane. Phosphorus
NMR spectra were recorded with proton decoupling on a
Bruker DPX 400 spectrometer at 162 MHz. Samples for phos-
Fig. 7 Structure of TPP and some of the amino acids residues
surrounding it from the crystal structure of yeast PDC18 (PDB
accession code 1PYD). Glu-477 is just above the thiazolium ring and
Ile-415 just below (in the “V” formed by the thiazolium and pyrimidine
rings).
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 7 3 2 – 1 7 4 1
1737

View Article Online
phorus NMR were prepared by dissolving the phosphate in
milliQ purified water and adjusting the pH to 7.0 with solid
carbon dioxide and then concentrating by lyophilisation prior
to dissolving in D2O. Mass spectra were run on either a Micro-
mass Q-Tof or a Micromass Concept spectrometer using elec-
tron impact (EI) or electrospray ionization (ESI) in positive ion
mode. Anion exchange was performed on a Pharmacia Acta
Prime chromatography system using a column packed with
DEAE-Sephacel (from Sigma). Analytical TLC was performed
on commercial Merck glass plates, coated to a thickness of
0.25 mm with Kieselgel 60 F254 silica. Column chromatography
was performed using Merck Kieselgel 60 (230–400 mesh)
silica using a small positive pressure of air. All solvents were
redistilled before use. Solvents and reagents for anhydrous
reactions were dried by conventional methods25 and such
reactions were performed under a small positive pressure of
argon.
δC(100 MHz; CDCl3) 40.7 (CH2), 52.0 (2 × CH3), 128.3, 128.5,
130.4 and 133.7 (4 × aryl CH), 134.2 and 134.3 (2 × aryl C) and
166.8 and 171.5 (2 × C᎐O); m/z (EI) 208.0716 (Mϩ. C H O
᎐
11 12
4
requires 208.0735).
3-(2-Hydroxyethyl)benzyl alcohol
Lithium aluminium hydride (5.9 g, 156 mmol) was added to dry
diethyl ether (125 ml), under an atmosphere of argon, and the
mixture was stirred and heated at reflux for 30 min. A solution
of diester 10 (16 g, 78 mmol) in dry diethyl ether (125 ml) was
then added dropwise such that the ether maintained a gentle
reflux. The reaction mixture was heated at reflux for a further
3 h, then cooled, poured into ice water (200 ml) and extracted
with ethyl acetate (3 × 100 ml). The combined organic layers
were washed with water and then with brine, dried (MgSO4)
and evaporated under reduced pressure. Purification by chrom-
atography on silica gel using ethyl acetate–hexane (1 : 1, Rf
0.15) as eluent gave the diol as an oil (7.1 g, 60%); δH(400 MHz;
CDCl3) 2.80 (2 H, t, J 6.5, CH2CH2OH), 3.77 (2 H, t, J 6.5,
CH2CH2OH), 4.58 (2 H, s, CH2), 7.11 and 7.16 (each 1 H, d,
J 7.5, 4-H and 6-H ), 7.18 (1 H, s, 2-H ) and 7.27 (1H, t, J 7.5,
5-H ); δC(100 MHz; CDCl3) 38.9 (ArCH2CH2), 63.3 and 64.9
(2 × CH2OH), 125.0, 127.6, 128.1 and 128.6 (4 × aryl CH) and
138.9 and 141.0 (2 × aryl C); m/z (EI) 152.0843 (Mϩ. C9H12O2
requires 152.0837).
Methyl 3-bromomethylbenzoate 8
m-Toluoyl chloride 7 (46.35 g, 0.30 mol) was heated to 180 ЊC
with vigorous stirring in a three necked round bottomed flask
fitted with a dry ice condenser. Bromine (52 g, 0.32 mol) was
added dropwise over 1 h. The reaction mixture was stirred for a
further 30 min at 180 ЊC and then cooled to room temperature
and added, with stirring, to methanol (300 ml) pre-cooled to
Ϫ78 ЊC. The mixture was left in the freezer overnight and then
the precipitate was filtered off and recrystallised from petrol-
eum ether (bp 40–60 ЊC) to give the bromo ester 8 as needles
(44.5 g, 65%), mp 46–47 ЊC (lit.8 mp 46–47 ЊC); δH(400 MHz;
CDCl3) 3.92 (3 H, s, CH3), 4.51 (2 H, s, CH2), 7.42 (1 H, t, J 7.7,
5-H ), 7.58 (1 H, dt, J 7.7 and 1.3, 4-H ), 7.96 (1H, dt, J 7.7 and
1.3, 6-H ) and 8.06 (1 H, t, J 1.3, 2-H ); δC(100 MHz; CDCl3)
32.4 (CH2), 52.2 (CH3), 128.9, 129.5 and 130.0 (aryl CH),
130.7 (aryl C), 133.4 (aryl CH), 138.2 (aryl C) and 166.4
3-(2-Hydroxyethyl)benzaldehyde 11
A solution of the above diol (5.0 g, 33 mmol) in dry chloroform
(70 ml) at room temperature was stirred vigorously with
activated manganese dioxide (20.0 g, 230 mmol) for 3 h, then
filtered through Celite and evaporated under reduced pressure
to give aldehyde 11 as an oil (4.55 g, 91%) which was used
without purification in the next step; δH(500 MHz; CDCl3)
2.94 (2 H, t, J 6.5, ArCH2CH2O), 3.91 (2 H, t, J 6.5,
CH2CH2OH), 7.47 (1 H, t, J 7.5, 5-H ), 7.51 (1 H, dt, J 7.5
and 1.6, 4-H ), 7.75 (1 H, dt, J 7.5 and 1.6, 6-H ), 7.79 (1 H, t,
J 1.6, 2-H ) and 9.98 (1 H, s, CHO); δC(125 MHz; CDCl3)
38.5 (ArCH2CH2O), 63.0 (ArCH2CH2O), 127.9, 128.9, 129.7
and 135.0 (4 × aryl CH), 136.4 and 139.6 (2 × aryl C) and
192.1 (CHO); m/z (ϩESI) 173.0581 (MϩNaϩ. C9H10O2Na
requires 173.0578).
(C᎐O).
᎐
Methyl 3-cyanomethylbenzoate 9
A mixture of bromo ester 8 (26 g, 113 mmol), potassium cyan-
ide (14.7 g, 226 mmol) and 18-crown-6 (2.5 g, 9.4 mmol) in
acetonitrile (280 ml) was stirred for 40 h at room temperature,
then filtered and evaporated under reduced pressure. A solution
of the residue in dichloromethane (100 ml) was washed with
water and then with brine, dried (MgSO4) and evaporated
under reduced pressure. Purification by silica gel chromato-
graphy (petroleum ether–ethyl acetate, 8 : 2, Rf 0.15) gave the
nitrile 9 as an oil (18 g, 91%), with spectroscopic data consistent
with those previously reported;26 δH(400 MHz; CDCl3) 3.78
(2 H, s, CH2), 3.90 (3 H, s, CH3), 7.44 (1 H, t, J 8.0, ArH ), 7.52
(1 H, m, ArH ), 7.97 (1 H, m, ArH ) and 7.98 (1 H, m, ArH );
δC(100 MHz; CDCl3) 23.4 (CH2), 52.2 (CH3), 117.3 (CN),
129.0, 129.1 and 129.2 (3 × aryl CH), 130.3 and 131.0 (2 × aryl
2-[3-(2-Hydroxyethyl)benzyl]-3-(phenylamino)acrylonitrile 12
A mixture of aldehyde 11 (4.55 g, 30 mmol) and 3-anilino-
propionitrile (4.5 g, 33 mmol) was stirred in dry dimethyl
sulfoxide (60 ml) at 40 ЊC under an atmosphere of argon. A
solution of sodium methoxide (1.66 g, 33 mmol) in methanol
(7 ml) was added dropwise and the reaction mixture was stirred
at 40 ЊC for a further 2 h, then cooled and added to vigorously
stirred ice-water. The sticky precipitate was filtered off and
recrystallised to give one isomer of the acrylonitrile 12 as
needles (4.0 g, 48%), mp 110–112 ЊC (from toluene) [Found: C,
77.65; H, 6.6; N, 9.9%; MϩNaϩ (ϩESI), 301.1304. C18H18-
N2O requires C, 77.7; H, 6.5; N, 10.1%; MϩNa, 301.1317];
νmax(Nujol)/cmϪ1 3434 (OH), 2186 (CN) and 1601, 1593 and
1497 (Ph); δH(400 MHz; CDCl3) 2.86 (2 H, t, J 6.5, ArCH2-
CH2O), 3.55 (2 H, s, CH2), 3.85 (2 H, t, J 6.5, ArCH2CH2O),
6.26 (1 H, br d, J 13.1, NH ), 6.77 (2 H, d, J 8.5, 2 × ArH ), 7.00
(1 H, t, J 7.4, ArH ), 7.17 (3 H, m, ArH ), 7.28 (3 H, m,
ArH ) and 7.35 (1 H, d, J 13.1, CHNH); δC(125 MHz; CDCl3)
32.6 and 38.8 (2 × ArCH2), 63.3 (CH2O), 82.5 (CCN), 115.0
(2 × anilino o-CH), 122.2 (CN), 122.7, 126.0, 127.7, 128.7
and 129.0 (4 × aryl CH and anilino p-CH), 129.5 (2 × anilino
m-CH), 136.8, 139.4 and 139.8 (3 × aryl C) and 140.3
(CHNH).
C), 132.2 (aryl CH) and 166.2 (C᎐O); MS (ϩEI) m/z 175.1
᎐
(Mϩ).
Methyl 3-(methoxycarbonylmethyl)benzoate 10
Methanol (150 ml) was saturated with hydrogen chloride gas at
Ϫ10 ЊC. Nitrile 9 (17 g, 97 mmol) was added dropwise with
stirring and the reaction mixture was then allowed to warm to
room temperature, stirred for 6 h, then cooled to 0 ЊC and
quenched by the addition of ice-water (150 ml), ensuring the
temperature remained below 10 ЊC. The solution was neutral-
ised by the addition of sodium hydrogen carbonate, filtered and
extracted with ethyl acetate (3 × 100 ml). The combined organic
layers were washed with brine, dried (MgSO4) and evaporated
under reduced pressure to give the diester 10 as an oil (17.5 g,
87%); νmax(Nujol)/cmϪ1 2953, 1722, 1608 and 1590; δH(400
MHz; CDCl3) 3.66 (2 H, s, CH2), 3.68 (3 H, s, CH3), 3.89 (3 H,
s, CH3) and 7.38, 7.46, 7.93 and 7.94 (each 1 H, m, ArH );
The filtrate was extracted with ethyl acetate (3 × 50 ml), and
combined organic layers were washed with brine, dried
(MgSO4) and concentrated under reduced pressure to give an
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 7 3 2 – 1 7 4 1
1738

View Article Online
oil which was a mixture of both isomers of acrylonitrile
12 (2.6 g, 31%). No attempt was made to separate the
isomers.
140.8 (pyrimidineCH) and 160.7 and 163.9 (CNCNH2); δP(162
MHz; D2O) Ϫ9.76 and Ϫ5.83 (each 1 P, d, J 21.3, OPOPO);
m/z (ϩESI) 426.0605 (MϩNaϩ. C14H19N3O7P2Na requires
426.0596).
4-Amino-5-[3-(2-hydroxyethyl)benzyl]-2-methylpyrimidine 13
3-(4-Amino-2-methylpyrimidin-5-ylmethyl)-4-methyl-5-
[2-(p-toluenesulfonyloxy)ethyl]thiophene 14
To a stirred suspension of acetamidine hydrochloride (0.68 g,
7.2 mmol) in dry ethanol (3 ml) was added dropwise a solution
of sodium methoxide (0.78 g, 14.4 mmol) in ethanol (4 ml). A
solution of 12 (1 g, 3.6 mmol) in ethanol (3 ml) was then added
dropwise and the reaction mixture was heated at reflux for 48 h
and then cooled. Addition of ice-water produced a precipitate
which was filtered off, dried and recrystallised to give the
pyrimidine 13 as colourless needles (0.71 g, 81%), mp 152–
153 ЊC (from EtOH) [Found: C, 69.1; H, 7.0; N, 17.1%; MHϩ
(ϩESI), 244.1454. C14H17N3O requires C, 69.1; H, 7.0; N,
17.3%; MH, 244.1450]; νmax(Nujol)/cmϪ1 3343 (OH), 3159
To a solution of deazathiamin 6 (280 mg, 1.1 mmol) in pyridine
(4.0 ml) at Ϫ5 ЊC was added portionwise p-toluenesulfonyl
chloride (1.0 g, 5.2 mmol). The reaction mixture was stirred at
Ϫ5 ЊC for 1 h, then quenched with cold hydrochloric acid (1 M;
10 ml) and neutralised by the addition of solid sodium
hydrogen carbonate. The mixture was extracted with ethyl
acetate (2 × 10 ml) and the combined organic layers were
washed with saturated aqueous copper sulfate, then water and
then brine, dried (MgSO4) and concentrated under reduced
pressure to give the tosylate 14 as a pale yellow solid (360 mg,
80%), mp 123–124 ЊC (from EtOH) [Found: C, 57.71; H, 5.56;
N, 9.91%; MϩNaϩ (ϩESI), 440.1076. C20H23N3O3S2 requires
C, 57.53; H, 5.55; N, 10.06%; MϩNa, 440.1079]; δH(400 MHz;
CDCl3) 1.99 (3 H, s, CH3), 2.43 (3 H, s, CH3), 2.49 (3 H, s, CH3),
3.08 (2 H, t, J 7.0, ArCH2CH2O), 3.56 (2 H, s, ArCH2), 4.13
(2 H, t, J 7.0, ArCH2CH2O), 4.82 (2 H, br s, NH2), 6.64 (1 H, s,
thiopheneH ), 7.31 (2 H, d, J 8.2, 2 × ArH ), 7.73 (2 H, d, J 8.2,
2 × ArH ) and 7.92 (1 H, s, pyrimidineH ); δC(125 MHz; CDCl3)
12.0, 21.4 and 25.2 (3 × CH3), 27.9 and 28.9 (2 × ArCH2), 69.3
(NH ), 1731 (C᎐O) and 1671 (NH ); δ (400 MHz; CDCl ) 2.46
᎐
2
2
H
3
(3 H, s, CH3), 2.80 (2 H, t, J 6.6, ArCH2CH2O), 3.72 (2 H, s,
CH2), 3.81 (2 H, t, J 6.6, ArCH2CH2O), 4.88 (2 H, br s, NH2),
6.99 (1 H, d, J 7.7, 4-H ), 7.00 (1 H, s, 2-H ), 7.10 (1 H, d, J 7.7,
6-H ), 7.23 (1 H, t, J 7.7, 5-H ) and 7.97 (1 H, s, pyrimidineH );
δC(100 MHz; CD3OD) 24.7 (CH3), 34.6 and 40.1 (2 × ArCH2),
64.1 (CH O), 115.3 (C᎐CNH ), 127.5, 128.4, 129.8 and 130.5
᎐
2
2
(4 × aryl CH), 139.0 and 141.0 (2 × aryl C), 154.7 (pyrim-
idineCH) and 163.7 and 166.3 (CNCNH2).
4-Amino-5-{3-[2-(p-toluenesulfonyloxy)ethyl]benzyl}-2-methyl-
pyrimidine 16
(CH O), 111.3 (C᎐CNH ), 119.4 (thiopheneCH), 127.6 and
᎐
2
2
129.6 (2 × o-CH and 2 × m-CH), 132.8, 132.9, 133.5, 136.7 and
144.6 (3 × thiopheneC and 2 × aryl C), 155.5 (pyrimidineCH)
and 161.3 and 166.1 (CNCNH2).
To a solution of 13 (100 mg, 0.41 mmol) in pyridine (1.5 ml)
at Ϫ5 ЊC was added portionwise p-toluenesulfonyl chloride
(313 mg, 1.64 mmol). The reaction mixture was stirred at Ϫ5 ЊC
for 2 h and then quenched with cold hydrochloric acid (1 M;
5 ml) and neutralised by the addition of solid sodium hydrogen
carbonate. The mixture was extracted with ethyl acetate (3 ×
5 ml) and the combined organic layers were washed with brine,
dried (MgSO4) and evaporated under reduced pressure to give
tosylate 16 as a yellow solid (140 mg, 86%), mp 106–107 ЊC
(from EtOH–H2O); δH(400 MHz; CDCl3) 2.42 and 2.48 (each
3 H, s, CH3), 2.90 (2 H, t, J 6.8, ArCH2CH2O), 3.71 (2 H, s,
CH2), 4.17 (2 H, t, J 6.8, ArCH2CH2O), 4.85 (2 H, br s, NH2),
6.94 (1 H, s, 2-H ), 7.00 (1 H, d, J 7.6, 4-H ), 7.01 (1 H, d, J 7.6,
6-H ), 7.20 (1 H, t, J 7.6, 5-H ), 7.28 (2 H, d, J 8.4, 2 × tosylH ),
7.77 (2 H, d, J 8.4, 2 × tosylH ) and 8.00 (1 H, s, pyrimidineH );
δC(100 MHz; CDCl3) 21.4 and 25.1 (2 × CH3), 34.3 and 35.0
3-(4-Amino-2-methylpyrimidin-5-ylmethyl)-4-methyl-5-
(2-hydroxyethyl)thiophene pyrophosphate (deazaTPP) 15
A solution of tosylate 14 (71 mg, 0.17 mmol) and tris(tetra-
butylammonium) hydrogen pyrophosphate (317 mg, 0.34
mmol) in dry acetonitrile (0.34 ml) under an inert atmosphere
was stirred at 4 ЊC for 10 h and then diluted with water (1 ml)
and purified by ion-exchange chromatography using a DEAE-
Sephacel column eluting with a gradient of 0 to 0.25 M aque-
ous ammonium bicarbonate. Lyophilisation of appropriate
fractions gave the alkene 18 (5 mg, 12%) and the pyrophosphate
ester 15 as a powder (58 mg, 72%).
For the pyrophosphate 15: δH(400 MHz; D2O) 1.94 and 2.42
(each 3 H, s, CH3), 3.04 (2 H, t, J 6.4, ArCH2CH2), 3.62 (2 H, s,
ArCH2), 4.02 (2 H, dt, J 6.8 and 6.4, CH2CH2OP), 6.89 (1 H, s,
thiophene-H) and 7.37 (1 H, pyrimidine-H); δC(125 MHz; D2O)
11.3 and 22.6 (2 × CH3), 27.4 and 28.9 (2 × ArCH2), 65.8
(2 × ArCH ), 70.1 (CH O), 112.6 (C᎐CNH ), 126.6, 127.4,
᎐
2
2
2
127.6, 128.7, 129.1 and 129.6 (8 × aryl CH), 132.8, 137.0,
137.3 and 144.5 (4 × aryl C), 155.4 (pyrimidineCH) and 161.4
and 166.1 (CNCNH2); m/z (EI) 397.1450 (Mϩ. C21H23N3O3S
requires 397.1460).
(CH O), 113.5 (C᎐CNH ), 119.4 (thiophene-CH), 133.6, 135.1
᎐
2
2
and 136.6 (3 × thiophene-C), 149.9 (pyrimidine-CH) and 162.4
and 163.8 (CNCNH2); δP(162 MHz; D2O) Ϫ5.74 and Ϫ9.91
(each 1 P, d, J 19.8, OPOPO); m/z (ϩESI) 446.0309 (MϩNaϩ.
C13H19N3O7P2SNa requires 446.0317).
4-Amino-5-[3-(2-hydroxyethyl)benzyl]-2-methylpyrimidine
diphosphate 17
To a stirred solution of tosylate 16 (60 mg, 0.15 mmol) in dry
acetonitrile (0.3 ml) under an atmosphere of argon was added
portionwise tris(tetrabutylammonium) hydrogen pyrophos-
phate (272 mg, 0.30 mmol). The reaction mixture was stirred at
room temperature for 16 h, then diluted with water (0.5 ml),
filtered through a 0.22 µm filter under positive pressure and
purified by ion-exchange chromatography on a DEAE-Seph-
acel column eluting with a gradient of 0.1 to 1 M aqueous
ammonium bicarbonate. Lyophilisation of appropriate frac-
tions gave the pyrophosphate ester 17 as a powder (15 mg,
24%); δH(400 MHz; D2O) 2.47 (3 H, s, CH3), 2.90 (2 H, t, J 6.4,
ArCH2CH2), 3.78 (2 H, s, CH2), 4.09 (2 H, dt, J 6.9 and 6.4,
ArCH2CH2O), 7.09 (1 H, d, J 7.5, 4-H ), 7.19 (1 H, s, 2-H ), 7.23
(1 H, d, J 7.5, 6-H ), 7.31 (1 H, t, J 7.5, 5-H ) and 7.56 (1 H, s,
pyrimidineH ); δC(125 MHz; D2O) 20.4 (CH3), 32.4 and 35.8
For alkene 18: δH(400 MHz, CDCl3) 2.12 (3 H, s, CH3), 2.51
(3 H, s, CH3), 3.60 (2 H, s, CH2), 4.75 (2 H, br s, NH2), 5.15
(1 H, d, J 10.9, CH᎐CH H ), 5.52 (1 H, d, J 17.2,
᎐
A
B
CH᎐CH H ), 6.65 (1 H, s, thiophene-H), 6.83 (1 H, dd, J 10.9
᎐
A
B
and 17.2, CH᎐CH ), 7.99 (1 H, br s, pyrimidine-H). δ (100
᎐
2
C
MHz, CDCl3) 12.4 (CH3), 25.5 (CH3), 29.0 (CH2), 107.0
(C᎐CNH ), 113.5 (CH᎐CH ), 119.8 (CH᎐CH ), 128.2 (thio-
᎐
᎐
᎐
2
2
2
phene-CH), 133.8, 137.4, 138.5 (3 × thiophene-C), 155.5
(pyrimidine-CH) and 162.0 and 166.0 (CNCNH2); m/z (EI)
245.0992 (Mϩ. C13H15N3S requires 245.0987).
General enzymic methods
All assays were carried out on a Cary 100 Bio UV-visible
spectrophotometer. Protein concentrations were determined
by the Bradford assay.27 Buffers were autoclaved prior to use
and kept on ice unless otherwise stated. E1o was donated by
(2 × ArCH ), 66.9 (CH O), 115.7 (C᎐CNH ), 126.9, 128.0,
᎐
2
2
2
129.2 and 129.7 (4 × aryl CH), 135.3 and 139.3 (2 × aryl C),
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 7 3 2 – 1 7 4 1
1739

View Article Online
R. Frank as an aqueous solution (60 mg mlϪ1) and stored at
4 ЊC. Solutions containing cofactors and substrates were pre-
pared freshly on the day of use.
Stoichiometry of binding to apoPDC
ApoPDC (4.1 µM of active sites) was incubated in MES–KOH
buffer (50 mM; pH 6.5; 15.5 µl) containing MgCl2 (32 mM) and
inhibitor (0.8 µM to 4.92 µM) for 3 hours at 30 ЊC. After
dilution of the solution to 200 µl with the same buffer, samples
were withdrawn (20 µl) and the enzymic activity was measured.
Assay for PDC activity
PDC activity was measured by a coupled enzyme assay using
the NADH-dependent reduction of the product, acetaldehyde,
by alcohol dehydrogenase (ADH). PDC was dissolved in
MES–KOH buffer (50 mM; pH 6.5) containing MgCl2 (5 mM),
TPP (0.1 mM), NADH (0.15 mM) and ADH (10 units per ml)
at 30 ЊC and the assay was started by addition of aqueous
sodium pyruvate (to give 5 mM). The oxidation of NADH was
monitored at 340 nm.
Measurement of the rate of inactivation of apoPDC
A solution of apoPDC (0.32 µM of active sites) in MES–KOH
buffer (50 mM; pH 6.5; 200 µl) containing MgCl2 (5 mM) and
inhibitor (1.92 µM to 5 µM for deazaTPP 15 or 2 µM to 6 µM
for benzene analogue 17) was incubated at 30 ЊC. At timed
intervals samples (20 µl) were taken and added to the assay
mixture without pyruvate (780 µl). After preincubation for
3 min at 30 ЊC, aqueous sodium pyruvate (2 µl; 2 M) was added
to initiate the assay.
Inactivation of holoPDC
A solution of holoPDC (0.029 µM), MgCl2 (5 mM) and TPP
(from 1.38 µM to 15 µM) in MES–KOH buffer (50 mM; pH
6.5) was preincubated at 30 ЊC for 60 min. After addition of
inhibitor (1 µM to 10 µM) the mixture was kept at 30 ЊC and
samples (50 µl) were taken at intervals and assayed for
PDC activity. In each case a control experiment was run
without inhibitor present to ensure that the enzyme is stable.
The residual activity given is a percentage of the value for the
control.
In order to show that the effect of deazaTPP was not due to
inhibition of alcohol dehydrogenase, the same experiment was
run using a deazaTPP concentration of 30 µM and no PDC. A
sample (50 µl) was taken and assayed as above except that acet-
aldehyde (5 mM) was used as substrate instead of pyruvate.
Consumption of NADH was very rapid, showing that the
ADH retains a high level of activity that is easily sufficient for
its role in this assay.
Assay for E1o component activity
The enzyme was preincubated at 30 ЊC for 15 min in potassium
phosphate buffer (106 mM; pH 7.0; 960 µl) containing MgCl2
(2.1 mM) and TTP (25 µM). Aqueous DCPIP (2.5 mM; 20 µl,
final concentration 50 µM) was then added and the reaction
was started by adding α-ketoglutarate (20 mM; 20 µl, final
concentration 400 µM). The decrease in absorbance at 600 nm
was monitored between 0.1 and 2 minutes. Enzyme activity is
expressed in terms of ∆A600 minϪ1
.
A range of different enzyme concentrations was tested and it
was concluded that a final concentration of 6.7 µg mlϪ1 (32 nM
of dimeric enzyme assuming it is homogeneous) was the least
that could be used in this assay while maintaining a satisfactory
level of accuracy when the enzyme was inhibited by up to 95%.
With no inhibition this concentration of enzyme gave a rate of
ca. 0.025 ∆A600 minϪ1
.
Preparation of apoPDC
Time-dependent inhibition of E1o by deazaTTP
All buffers used for preparation of the apoenzyme were pre-
viously stirred with Chelex-100 cation exchange resin (Naϩ
form) to remove bivalent metal. HoloPDC was incubated in
HEPES–KOH buffer (50 mM; pH 8.2) containing dipicolinic
acid (1 mM) for 20 min at 4 ЊC and concentrated by ultra-
filtration. This treatment was performed three times. The buffer
was then replaced by MES–KOH buffer (50 mM; pH 6.5) by
ultrafiltration and dilution (three times). The apoenzyme was
stored at 4 ЊC under N2.
A stock solution (51.84 ml) of enzyme (6.7 µg mlϪ1) in assay
buffer containing TPP (25 µM) was divided into 9 equal por-
tions and kept at 3 ЊC. Each portion in turn was pre-incubated
at 30 ЊC for 15 min and then an aliquot (960 µl) was taken
and assayed. A solution of deazaTPP (125, 50, 25, 12.5 or 5 µl of
1 mM or 125, 50, 25 or 12.5 µl of 10 µM) was added to the
remaining enzyme solution and pre-incubation was continued
at 30 ЊC. Aliquots (960 µl) were removed at timed intervals and
mixed with DCPIP and α-ketoglutarate and assayed as above.
The addition of the deazaTPP solution represents a small
dilution of the enzyme compared to the t = 0 sample taken
before addition of deazaTPP and so the measured enzyme
activities have all been adjusted to compensate for this dilution.
The results are shown in Fig. 6.
Change of fluorescence of ApoPDC
Experiments were performed in an Amico Bowman Series 2
spectrofluorimeter at 30 ЊC. Fluorescence decay was measured
at 340 nm with excitation wavelength at 300 nm. Bandwidth
was 4 nm for both excitation and emission. The solution con-
tained 0.5 µM apoPDC (2 µM of active sites), 5 mM MgCl2 and
different concentrations of deazaTPP 15 in MES–KOH buffer
(50 mM; pH 6.5).
Demonstration that binding of deazaTPP to E1o is reversible
A solution (4.8 ml) containing enzyme (13.3 µg mlϪ1), TPP (52
µM) and buffer (as above) was incubated for 15 min at 30 ЊC.
An aliquot (960 µl) was taken and assayed. A solution of
deazaTTP (1 mM; 5 µl giving 1.3 µM) was added to the remain-
ing enzyme solution and further aliquots were assayed after 5,
10 and 30 min. At this point 10 mM TPP (100 µl) was added to
the remaining 960 µl of enzyme solution. After incubation for a
further 30 min, a final 960 µl aliquot was assayed. The enzyme
activity of this final aliquot was adjusted to compensate for the
dilution caused by the addition of the TTP solution.
Denaturation–renaturation of partially inactivated PDC
HoloPDC was partially inactivated by incubation with deaza-
TPP 15 (5 µM) in the presence of TPP (16.5 µM) and the resi-
dual enzymic activity was assayed as above. A solution (0.5 ml)
containing this partially inactivated holoPDC (0.25 µM, 1 mM
monomer), TPP (100 µM), MgCl2 (5 mM), DTT (100 mM),
and urea (6.8 M) in MES–KOH buffer (50 mM; pH 6.5) was
incubated at 25 ЊC for 2 min. The inhibitor was then removed
by several rounds of ultrafiltration and dilution with the same
urea-containing buffer. Renaturation was initiated by a rapid
twenty-fold dilution of the sample (to 10 ml) with the same
buffer without urea. The volume was then reduced to 0.5 ml by
ultrafiltration and the activity of PDC was measured by the
standard enzymatic assay.
Acknowledgements
We thank the BBSRC and Syngenta for financial support,
Dr D. A. Griffin (Syngenta) for his advice and encouragement
and Prof. R. N. Perham and R. Frank (Department of
Biochemistry) for the gift of E1o.
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 7 3 2 – 1 7 4 1
1740

View Article Online
14 M. Pohl, J. Groetzinger, A. Wollmer and M. R. Kula, Eur. J.
References
Biochem., 1994, 224, 651.
15 D. Dobritzsch, S. Koenig, G. Schneider and G. Lu, J. Biol. Chem.,
1998, 273, 20196.
16 L. S. Khailova, R. Bernhardt and G. Hübner, Biokhimiya
(Moscow), 1977, 42, 113.
1 F. Jordan, Nat. Prod. Rep., 2003, 20, 184.
2 E. Chabriere, M. H. Charon, A. Volbeda, L. Pieulle,
E. C. Hatchikian and J. C. Fontecilla-Camps, Nat. Struct. Biol.,
1999, 6, 182.
3 U. Nilsson, L. Meshalkina, Y. Lindqvist and G. Schneider, J. Biol.
Chem., 1997, 272, 1864.
4 E. Fiedler, S. Thorell, T. Sandalova, R. Golbik, S. Konig and
G. Schneider, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 591.
5 E. Chabriere, C. Vernede, B. Guigliarelli, M. H. Charon,
E. C. Hatchikian and J. C. Fontecilla-Camps, Science, 2001, 294,
2559.
17 F. Jordan, H. J. Li and A. Brown, Biochemistry, 1999, 38, 6369.
18 F. Dyda, W. Furey, S. Swaminathan, M. Sax, B. Farrenkopf and
F. Jordan, Biochemistry, 1993, 32, 6165.
19 M. Lobell and D. H. G. Crout, J. Am. Chem. Soc., 1996, 118, 1867.
20 J. A. Gutowski and G. E. Lienhard, J. Biol. Chem., 1976, 251, 2863.
21 U. Nilsson, Y. Lindqvist, R. Kluger and G. Schneider, FEBS Lett.,
1993, 326, 145.
6 S. S. Pang, R. G. Duggleby, R. L. Schowen and L. W. Guddat,
J. Biol. Chem., 2004, 279, 2242.
7 D. Hawksley, D. A. Griffin and F. J. Leeper, J. Chem. Soc.,
Perkin Trans. 1, 2001, 144.
8 W. Davies and W. H. Perkin, J. Chem. Soc., 1922, 121, 2202.
9 S. Bearne, K. Karimian and R. Kluger, Biochemistry, 1988, 27, 3084.
10 H. Koike and T. Yusa, Methods Enzymol., 1970, 18, 105.
11 V. J. Davisson, D. R. Davis, V. M. Dixit and C. D. Poulter, J. Org.
Chem., 1987, 52, 1794.
12 G. Schenk, F. J. Leeper, R. England, P. F. Nixon and R. G. Duggleby,
Eur. J. Biochem., 1997, 248, 63.
22 R. Kluger, G. Gish and G. Kauffman, J. Biol. Chem., 1984, 259,
8960.
23 N. Nemeria, Y. Yan, Z. Zhang, A. M. Brown, P. Arjunan, W. Furey,
J. R. Guest and F. Jordan, J. Biol. Chem., 2001, 276, 45969.
24 P. N. Lowe, F. J. Leeper and R. N. Perham, Biochemistry, 1983, 22,
150.
25 D. D. Perrin, W. L. F. Armarego and D. R. Perrin, Purification of
Laboratory Chemicals, Pergamon, Oxford, 1980.
26 P. Chand, Y. S. Babu, S. Bantia, N. Chu, L. B. Cole, P. L. Kotian,
W. G. Laver, J. A. Montgomery, V. P. Pathak, S. L. Petty,
D. P. Shrout, D. A. Walsh and G. M. Walsh, J. Med. Chem., 1997,
40, 4030.
13 R. J. Diefenbach and R. G. Duggleby, Biochem. J., 1991, 276,
439.
27 M. M. Bradford, Anal. Biochem., 1976, 72, 248.
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 7 3 2 – 1 7 4 1
1741