Báo cáo khoa học: Catalyzing separation of carbon dioxide in thiamin diphosphate-promoted decarboxylation ppt

REVIEW ARTICLE
Catalyzing separation of carbon dioxide in thiamin
diphosphate-promoted decarboxylation
Ronald Kluger and Steven Rathgeber
Davenport Chemical Laboratories, Department of Chemistry, University of Toronto, Canada
Decarboxylases and intermediates
Thiamin diphosphate (ThDP) is a cofactor that
promotes the decarboxylation of 2-ketoacids through
formation of covalent derivatives between its C2 thia-
zolium and the carbonyl of the substrate. Combina-
tion of a protein and ThDP in a holoenzyme
provides substrate specificity and the general enzymic
advantage of reduced translational entropy that
favors addition processes [1,2]. The covalent interme-
diate undergoes cleavage of a bond to a carboxylate
group derived from the 2-ketoacid, resulting in
production of carbon dioxide. This also produces a
residual acyl anion equivalent [3–6] with a delocalized
structure that can also be represented as a neutral
enamine. The sequence is illustrated for the decarbox-
ylation of pyruvic acid by pyruvate decarboxylases in
Scheme 1.
Protonation at the basic carbon and elimination of
ThDP leads to formation of an aldehyde (giving a net
substitution of a proton for carbon dioxide). Oxidation
of the same intermediate would yield an acid, while
reaction with a carbonyl carbon gives a condensation
product. The general route is based on concepts origi-
nally developed by Breslow [7–9] based on studies of
model compounds related to ThDP. Details of reaction
patterns within that pathway reveal previously unrec-

ognized aspects of enzymic catalysis [3,10].
Synthetic analogs of the covalent intermediates have
been prepared and studied in order to arrive at a
quantitative understanding of the separate functions of
the cofactor and protein [11,12]. Spectroscopic analysis
of the conjugates of thiamin and ketoacids has enabled
specific and quantitative identification of the coenzyme
derivatives bound to proteins in enzymic reactions
[13–15].
Keywords
active site; benzoylformate decarboxylase;
carbanion; catalysis; decarboxylation;
diffusion; fragmentation; pre-association;
thiamin; thiamin diphosphate
Correspondence
R. Kluger, Davenport Chemical Laboratories,
Department of Chemistry, University of
Toronto, Toronto, Ontario M5S 3H6, Canada
Fax: +1 416 978 8775
Tel: +1 416 978 3582
E-mail:
(Received 22 July 2008, revised 2 October
2008, accepted 10 October 2008)
doi:10.1111/j.1742-4658.2008.06739.x
Thiamin diphosphate-dependent decarboxylases form addition intermedi-
ates between thiamin diphosphate (ThDP) and 2-ketoacids. Although it
appears that the intermediate should react without the intervention of cata-
lysts, evidence has clearly shown that Brønsted acid catalysis occurs
through a pre-associated system. This can promote separation of carbon
dioxide from the residual carbanion by protonation of the carbanion.

Proteins operate through pre-association and may readily promote the
separation of carbon dioxide by protonating or oxidizing the nascent carb-
anion. Alternatively, a nucleophilic side chain may trap carbon dioxide as
an unstable hemi-carbonate. Mutagenesis experiments by others have
shown that enhanced activity due to the protein in the presence of thiamin
diphosphate does not depend on the presence of any one proton donor,
consistent with pooled activity within the active site. This form of catalysis
has not been widely recognized, but should be considered an integral aspect
of enzyme-promoted decarboxylation.
Abbreviations
AHAS, acetohydroxy acid synthase; BFD, benzoylformate decarboxylase; HBnTh, 2-(1-hydroxybenzyl) thiamine; HBnThDP,
2-(1-hydroxybenzyl) thiamin diphosphate; MTh, a-mandelyl-thiamin; ThDP, thiamin diphosphate.
FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS 6089
There are significant differences between the reac-
tivity of synthetic intermediate analogs and the corre-
sponding intermediates in enzymic systems, and these
can reveal the specific role of the protein [4,5,12,16].
Quantitative differences in the decarboxylation of the
conjugates of thiamin and 2-ketoacids provide impor-
tant insights into the role of the protein as a catalyst
in the decarboxylation step of an enzyme for which
reactions are considerably faster than the comparable
unimolecular reactivity of the synthetic intermediates
[12,17,18]. Based on rate measurements in catalytic
systems, we have recently proposed that the proteins
increase the rates of decarboxylation of ThDP-derived
intermediates of 2-ketoacids through their inherent
ability to facilitate diffusion of carbon dioxide away
from the ThDP-derived intermediate, avoiding the
significant reverse reaction that is normally an inherent

part of the non-enzymic reaction [3,18,19].
The rate constant for decarboxylation of a-lactyl-
thiamin, the simplied analog of a-lactyl-thiamin
diphosphate in Scheme 1, is approximately 10
6
times
smaller than the typical k
cat
value for pyruvate decar-
boxylase [12]. However, there is no site in the likely
transition state for decarboxylation associated with the
formation of carbon dioxide that would be stabilized
by specific interaction with the protein. Therefore, we
would not expect any groups on the protein to affect
the rate, but the rate acceleration is clearly significant.
Similarly, the intermediate analog for benzoylformate
decarboxylase (BFD), the conjugate of thiamin and
benzoylformate, a-mandelyl-thiamin (MTh) (Fig. 1),
undergoes decarboxylation in neutral solution with a
rate constant that is also approximately 10
6
times
smaller than the k
cat
for BFD [11].
Catalysis by desolvation
Crosby and Lienhard produced a simplified model for
the conjugate of pyruvate and ThDP, 2-(1-carboxy-
hydroxyethyl)-3,4-dimethylthiazolium chloride) [20].
They noted that its decarboxylation rate constant is at

least 10
5
times smaller than that of the likely enzyme-
bound intermediate derived from a-lactyl-thiamin
diphosphate. They suggest that, instead of acid ⁄ base
catalysis, which would not facilitate the decarboxyl-
ation step, the enzyme could transfer the intermediate
into an environment with reduced polarity. They sup-
port this with evidence that the model intermediate’s
rate of decarboxylation is much greater in solvents
with reduced polarity.
Despite the clear change in reactivity in low-polarity
solvents, this hypothesis presents some difficulties.
BFD has a similar intermediate and also shows rate
acceleration of the conjugate of its substrate and
ThDP, but has a very hydrophilic binding site for the
substrate and coenzyme [3,21–23]. Thus, there is no
obvious way to promote the decarboxylation step in
general.
Oka fragmentation of 2-(1-hydroxyben-
zyl)-thiamin
The product of decarboxylation of the conjugate of
ThDP and benzoylformate is 2-(1-hydroxybenzyl) thia-
min diphosphate (HBnThDP). Analysis of the prod-
ucts in this reaction revealed that the C2a conjugate
base of HBnTh undergoes a reaction that destroys thi-
amin by a very rapid process that splits the pyrimidine
and thiazolium portions (Scheme 2) [16,24].
Fig. 1. The structure of a-mandelyl-thiamin (MTh), an accurate
reactivity model of the conjugate formed from ThDP and benzoyl-

formate.
Scheme 1. Covalent intermediates derived from thiamin diphosphate in the decarboxylation of pyruvate.
Catalyzing separation of carbon dioxide R. Kluger and S. Rathgeber
6090 FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS
The same products were originally observed by Oka
et al. during an attempt to catalyze condensation of
benzoin with thiamin promoted by a tertiary amine in
ethanol [25]. In that study, it is likely that HBnTh
fragmented after it formed. Our study of the products
of the spontaneous reaction of MTh revealed that
fragmentation is the specific result of a proton being
removed from C2a, a process that is subject to general
base catalysis; therefore, it must be the rate-determin-
ing step or a component of that step. Removal of the
proton from carbon rather than from the hydroxyl
group (which would lead to generation of thiamin and
benzaldehyde) is highly favored under conditions
where the pyridimine is protonated or has a positive
charge induced by alkylation [26,27].
The expected ionization of the C2a hydroxyl of
HBnTh, followed by formation of benzaldehyde and
thiamin, only occurs in more basic solutions and is
subject only to specific base catalysis by the solvent
lyate ion [26]. The occurrence of fragmentation of
HBnTh is readily detected by observing the unique
absorbance band at 328 nm that arises from the phe-
nyl thiazole ketone product [25]. The unimolecular rate
constant for fragmentation of HBnTh is approximately
10
4

s
)1
at 30 °C, which is approximately 100 times
larger than the k
cat
of BFD. Thus, the enzyme appears
to accelerate decarboxylation of MThDP and to slow
fragmentation of the anion derived from HBnThDP.
The C2a conjugate base of HBnTh from
decarboxylation – fragmentation and
its implications
Decarboxylation of MTh will produce the conjugate
base at C2a of HBnTh as the immediate product,
along with carbon dioxide. In the presence of low con-
centrations of acid components of phosphate or ace-
tate buffers, fragmentation occurs rapidly, as expected
(Scheme 3) [28]. The rate of decarboxylation is not
affected as the concentration of buffer is increased.
However, the extent of fragmentation relative to the
formation of HBnTh decreases. In the reaction cata-
lyzed by BFD, although the mechanism appears to
require formation of the analogous carbanion,
HBnThDP forms without competition from what
should be a faster fragmentation [21,22,29–31]. This
presents an interesting problem: how does the enzyme
avoid fragmentation if that process without interven-
tion of an enzyme has a lower barrier than the normal
pathway of the enzyme [32,33]?
Cryptic catalysis – decarboxylation of
MTh is enhanced by pyridine acids

Based on conventional analysis of the mechanism of
decarboxylation of a thiamin conjugate, there is no
role for a catalyst in the carbon–carbon bond-breaking
Scheme 2. Fragmentation from the C2a conjugate base of 2-(1-hydroxybenzyl)thiamin is a very fast process.
Scheme 3. Decarboxylation of MTh leads to fragmentation in the absence of an enzyme or Brønsted acid.
R. Kluger and S. Rathgeber Catalyzing separation of carbon dioxide
FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS 6091
step [5,34]. The thiazolium nitrogen is in the position
that corresponds to the carbonyl oxygen in a 2-keto-
acid. While an acid can protonate a ketone’s carbonyl
oxygen, the thiazolium nitrogen is at its maximum
electron deficiency and has no available coordination
sites. Simply, there is no place for a proton or other
cation to position itself in order to promote the reac-
tion by stabilizing a transition state that resembles the
product. This means that neither Brønsted nor Lewis
acids can play a role in promoting cleavage of the
carbon–carbon bond.
Surprisingly, it struck us as being most remarkable
when Hu and Kluger observed that pyridine buffers
promote decarboxylation of MTh [35]. Their investiga-
tions revealed that only the acid component of the
buffer is catalytically active. In addition, C-alkyl
substituted pyridine-derived acids also acted as cata-
lysts, even those with alkyl substituents adjacent to the
nitrogen center. However, no other acids or bases that
were tested were effective [18]. The second-order rate
constants for catalysis by the various pyridine deriva-
tives are essentially invariant. The lack of dependence
on pK

a
is not consistent with catalysis by Brønsted
bases from a weaker acid substrate (Fig. 2). This sug-
gests that the catalytic process is a thermodynamically
favorable proton transfer [18].
The only site that becomes available for protonation
in the decarboxylation reaction of MTh is C2a. This is
accessible only after carbon dioxide has formed. Thus,
in order for the pyridine acids to be catalytic, the for-
mation of carbon dioxide would have to be reversible
(Scheme 4). The role of the catalyst would be to add a
proton to compete for the carbanion against carbon
dioxide. This process slows the reverse reaction and in
doing so accelerates the diffusional separation of
carbon dioxide and HBnTh.
An alternative possibility is that the charge of the
protonated material provides electrostatic stabilization
in the transition state for bond cleavage, without
transferring a proton (Scheme 5). While this may be
a generally applicable type of mechanism [36], it is
unlikely to be in operation here as the proton’s
position is necessarily dynamic – rapidly associating
and dissociating.
As a measure of the significance of the electrostatic
effect, we added N-ethylpyridinium chloride and
observed that it has no effect on the rate of reaction,
and any electrostatic effect is therefore very small, if
any [18].
Pre-association mechanisms
Jencks [37] and Venkatasubban and Schowen [38]

observed catalysis in reactions where separation of
products from one another by diffusion is rate-deter-
mining. They reasoned that in such a case, the catalyst
must form an initial, stable complex with the reactant.
Applying that concept to the present case, the pyridi-
nium catalyst must be associated with MTh prior to
98
CH
3
CH
3
CH
3
CH
3
H
3
C
H
3
C
76
pK
a
54
–5
–4
–3
log K
obs

–2
–1
N
H
N
H
N
H
N
H
Fig. 2. Second-order rate constants for catalysis of the decarboxyl-
ation of MTh by Brønsted acids derived from pyridine and C-alkyl-
pyridines. The fitted line has a slope of 0, consistent with a
thermodynamically favorable proton transfer.
Scheme 4. The complex of protonated pyridine and MTh accelerates departure of carbon dioxide, trapping the carbanionic product as it
forms.
Catalyzing separation of carbon dioxide R. Kluger and S. Rathgeber
6092 FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS
the decarboxylation process. In the reported instances
[37,38], the pre-associated catalyst is held in place by
hydrogen bonding, and is in a position to transfer a
proton to the acceptor in competition with a reversible
step. However, in the decarboxylation of MTh, there
are no groups with which a proton donor could form a
hydrogen bond to promote the reaction. This suggests
that attractive forces in pre-association processes are
not necessarily limited to hydrogen bonding. Based on
modelling, using analogous materials as the basis, we
proposed that pyridinium can be associated with MTh
by face-to-face p-stacking interactions with aromatic

groups of MTh [18]. This can position the proton-
donating site near the carbanion that is being produced.
As carbon dioxide forms, the associated protonated
pyridine is in a position from which it can readily
transfer a proton to the nascent carbanionic C2a posi-
tion derived from HBnTh. The acid’s proton competes
with carbon dioxide as an electrophile. Thus, the over-
all protonation process facilitates the diffusional sepa-
ration of carbon dioxide and HBnTh. This requires
that we consider the possible existence of an additional
intermediate that leads to the rate-determining step.
The complex in which carbon dioxide remains associ-
ated with the conjugate base of HBnTh must be dis-
tinct from that in which carbon dioxide has separated.
Decarboxylation as a two-step process
Based on the idea of a pre-associated catalyst and the
observed catalysis by pyridinium, we proposed that, in
general, separation of carbon dioxide and the conju-
gate base of HBnTh is at least partially rate-determin-
ing [18]. If the barrier for addition of carbon dioxide
to the newly formed carbanion is lower than the bar-
rier to diffusional separation, then diffusion is neces-
sarily the rate-determining step. Lowering the barrier
for the diffusion step is therefore the only way to
accelerate the reaction. As diffusion is the result of a
set of physical properties, the process itself cannot be
accelerated.
Instead of affecting diffusion as a process, a reaction
can proceed faster if the process competing with diffu-
sion is slower. This is the case if the decarboxylation

step is reversible. Analysis requires consideration of
the relative magnitudes of the barriers for diffusion of
carbon dioxide and reversal of decarboxylation
(Scheme 6).
Gao et al. calculated reaction pathways for the non-
enzymic and enzymic decarboxylation of orotidine
Scheme 5. Electrostatic stabilization of the transition state for
decarboxylation of MTh.
Scheme 6. The intermediate is associated
with carbon dioxide. The lower barrier is
associated with k
)1
, and k
2
is the rate
determining step.
R. Kluger and S. Rathgeber Catalyzing separation of carbon dioxide
FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS 6093
monophosphate [39,40]. Their calculations show that
the enthalpic kinetic barrier to the reverse reaction,
addition of the carbanion to carbon dioxide, is very
small or non-existent. The barrier to the reaction is
purely entropic, and arises once carbon dioxide and
the residual anion are separated by solvent mole-
cules. Applying this idea to decarboxylation reactions
in general, diffusion of carbon dioxide can be the
rate-limiting step in a two-step process where the
barrier to reversion is lower than the barrier to dif-
fusion. If the acid catalyst suppresses the reverse
reaction, it will make the overall forward reaction

faster.
Facilitating rate-limiting CO
2
diffusion
If there is no additional enthalpic barrier to addition
of the carbanion to an associated molecule of carbon
dioxide, the rate constant will be approximately the
same as the frequency of vibration of a carbon–carbon
bond. The stretching frequency of such a bond is
typically approximately 1000 cm
)1
, which corresponds
to a rate constant of 3 · 10
13
s
)1
.
Carbon dioxide is internally polarized, with the elec-
tron deficiency at carbon creating a partial positive
charge relative to the electronegative oxygen atoms.
The center of the molecule will be attracted to the
relatively anionic C2a centre of the conjugate base
of HBnTh, introducing a barrier to separation. (The
oxygen atoms may be attracted to the cationic thia-
zolium nitrogen as well.) We estimate that the rate
constant for diffusional separation will be somewhat
smaller than for cases where there is no attractive
force, a maximum of approximately 10
8
s

)1
. The ratio
of recombination to separation based on these esti-
mates is approximately 10
5
. If an enzyme efficiently
promotes the separation process by direct protonation
of the residual anion, the acceleration is approximated
by this ratio. This ratio of 10
5
is about the same as
that of between k
cat
for BFD to the unimolecular
decarboxylation rate constant of MTh. If the enzyme
achieves this by transferring a proton to the incipient
carbanion, it will also completely suppress fragmenta-
tion of the coenzyme [3,18].
Formation of a productive complex between proton-
ated pyridine and MTh prior to decarboxylation
allows proton transfer to compete with the capture of
carbon dioxide by the nascent enamine. This provides
a catalytic route for decarboxylation simply by provid-
ing a competitor for the back reaction of carbon diox-
ide, promoting a net reaction in the forward direction.
As carbon dioxide does not have specific binding sites
in a protein, the reaction becomes effectively irrevers-
ible once it separates from its co-product after the
enamine is protonated.
Extension – enzymes always use pre-association

to promote reactions
Enzymes bind substrates into active sites that contain
multiple functional groups in close proximity. While
pre-association of two molecules in organic chemistry
is an uncommon component of catalysis, it is a
universal aspect of enzymic catalysis [41]. Therefore,
non-enzymic reactions that involve pre-association
provide information on key aspects of enzymic reac-
tions. Although intramolecular reactions model reac-
tions of bound substrates [42], the structural
relationships of functional groups and geometric
restrictions limit interpretations.
The role of the protein
We propose that the observed (spontaneous) first-order
rate constant for decarboxylation of MTh is very small
compared to the k
cat
of BFD because the enzyme con-
tains pre-associated functional groups that can serve as
the source of the proton necessary to block the return
of carbon dioxide by protonating the nascent enamine
[18]. This proposal explains why, even though the non-
enzymic decarboxylation product fragments with a rate
constant greater than the k
cat
for BFD, this is not an
issue simply because the intermediate is protonated
much more rapidly than it fragments. As fragmenta-
tion occurs from the unprotonated intermediate, the
enzyme catalyzes the reaction and at the same time

blocks the destructive pathway, without further evolu-
tion of function.
Enzymic catalysis of carboxylation –
the same issues in reverse
In enzyme reactions, the pathways for carboxylation
are not the reverse of those for decarboxylation
because aspects of energy and equilibrium make the
situation more complex. Many of the questions that
we are addressing in terms of reversible formation of
carbon dioxide during decarboxylation have been con-
sidered in depth with respect to carboxylation reac-
tions. In 1975, Sauers et al. speculated on the potential
role of carboxy phosphate in the carboxylation of bio-
tin [43]. They suggested that it is a source of localized
carbon dioxide, which is the same situation as arises
spontaneously in decarboxylation. They proposed that
the hypothetical anhydride of carbonic and phosphoric
acids, from the enzyme-catalyzed reaction of ATP and
Catalyzing separation of carbon dioxide R. Kluger and S. Rathgeber
6094 FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS
bicarbonate, would be too unstable to serve as an
intermediate. Instead, it spontaneously converts (with
a half life of less than a second) to carbon dioxide and
phosphate. The resulting unsolvated carbon dioxide is
then in a position to react with an adjacent bound
nucleophile (Scheme 7): ‘The high local concentration
of this molecule of carbon dioxide provides an effec-
tive driving force for its reaction with the bound
acceptor so long as it reacts with the acceptor more
rapidly than it dissociates into solution. A molecule

may have a high Gibbs free energy that makes it effec-
tively an ‘energy-rich’ compound as a consequence of
its fixation and decreased entropy, as well as chemical
activation ’ [43].
Furthermore, Sauers et al. address the observation
that the same enzyme facilitates the reverse process,
the decarboxylation of N
1
¢-carboxybiotin: ‘If the rate-
determining step of this reaction is the dissociation of
bound carbon dioxide, the addition of acceptor mole-
cules that decrease the steady-state concentration of
carbon dioxide at the active site would decrease the
observed rate of decarboxylation. This is consistent
with the observed inhibition of carboxybiotin break-
down by inorganic phosphate’. Thus, they propose
that carbon dioxide reacts with an alternative nucleo-
phile that gives an unstable covalent intermediate, pre-
venting it from reacting with the group that would
reverse the reaction. This strategy for accelerating
decarboxylation by facilitating the diffusion of carbon
dioxide is complementary to one in which a proton is
added to the residual organic anion. Despite the fine
logic and elegance of this proposal, the idea of revers-
ible decarboxylation as presented received little further
attention.
Catalysis by addition to carbon dioxide
An alternative means of promoting the separation of
carbon dioxide suggested in the paper by Sauers et al.
[43] involves trapping the carbon dioxide with a com-

peting nucleophile (Scheme 7, last step). A carbon
dioxide-trapping mechanism, in which a nucleophile
on the enzyme adds to carbon dioxide as it forms,
might also be a relevant step in some ThDP-dependent
decarboxylases (Scheme 8). There are indications of
this possibility in the reaction of benzoylphosphonate
as a substrate analog of benzoylformate in the reaction
catalyzed by BFD. Benzoylphosphonate appears to be
a mechanism-based inactivator: it is processed by the
enzyme leading to a product that effectively inactivates
the enzyme [44]. Crystallographic analysis reveals that
the equivalent of metaphosphate is transferred to the
hydroxyl group of an active site serine to give a
phosphate monoester. If this occurred with the normal
substrate, the metastable carbonate monoester would
result. Trapping carbon dioxide temporarily would
slow the reversal. At the same time, a proton would
have been added in place of carbon dioxide at the
basic reaction site, yielding the stable product.
AHAS – does the flavin cofactor
maintain reversibility?
Acetohydroxy acid synthase (AHAS) catalyzes decar-
boxylation of a 2-ketoacid conjugate of ThDP followed
by addition of the enamine-carbanion to a second 2-keto-
acid [45]. It is essential that the carbanion generated by
decarboxylation is not protonated in order for the C2a
center to function as a nucleophile. FAD is an essential
Scheme 7. Formation of ‘low-entropy’
carbon dioxide and formation of
carboxybiotin.

R. Kluger and S. Rathgeber Catalyzing separation of carbon dioxide
FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS 6095
cofactor for the enzyme, but no role for an oxidation
process in catalysis has been discovered, although the
flavin may be reduced slowly during the process of catal-
ysis [45,46]. As AHAS resembles pyruvate oxidase, a
functional flavoenzyme, it has been proposed that the
flavin is only an evolutionary vestige in AHAS with a
purely structural role [45,46]. Nonetheless, it is intrigu-
ing to consider how the cofactor might function in a
more active sense. It is reasonable to assume that evolu-
tion would have favored an enzyme that did not require
the complexity of the cofactor.
In terms of the general mechanism we have pre-
sented, the flavin could serve as a temporary storage
site for the electrons liberated in the decarboxylation
process, preventing addition of carbon dioxide and
reversal of the reaction, or providing a proton to
quench the intermediate in the absence of the second
substrate under dilute reaction conditions (Scheme 9).
It is well-known that redox cofactors in other enzymes
can complete an oxidation–reduction cycle as a means
of altering the reactivity of intermediates, giving the
external appearance of an inactive cofactor [47–52].
The flavin is reduced during the course of catalysis,
although a net reduction occurs only if the oxidized
substrate decomposes through an independent uncata-
lyzed pathway [45].
Implications from site-directed
mutagenesis on benzoylformate

decarboxylase
The observation of pyridine acid-catalyzed decarboxyl-
ation of MTh suggested that ThDP-dependent decar-
boxylases utilize Brønsted acids in the active site to
facilitate departure of carbon dioxide, protonating the
C2a carbanion-enamine intermediate prior to separa-
tion of carbon dioxide. This prevents reversal to form
Scheme 8. A base-catalyzed reaction with the side chain of serine
competes for carbon dioxide with the nascent carbanion of
HBnThDP.
Scheme 9. Proposal for a catalytic role for
the flavin in AHAS.
Catalyzing separation of carbon dioxide R. Kluger and S. Rathgeber
6096 FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS
the carboxylic acid. Therefore, we expect that proton
sources at the active site of the enzyme would facilitate
the reaction.
Polovnikova et al. reported the effects of site-direc-
ted mutagenesis replacements of active site groups in
BFD [53]: S26A, H70A and H281A. Their results indi-
cated that catalysis of the bound substrate is most sub-
stantially affected by H70A, for which k
cat
is reduced
by a factor of over 1000, while the reduction in
H281A was significant but smaller. However, the cata-
lytic rate constants of H70A and H281A are still
orders of magnitude higher than the rate constants for
uncatalyzed decarboxylation of MTh [11,18]. Polovnik-
ova et al. make the intriguing proposition in the

abstract of their paper that protonation of the groups
on the enzyme is distinct: ‘The residue H70 is impor-
tant for the protonation of the 2-a-mandelyl-ThDP
intermediate, thereby assisting in decarboxylation, and
for the deprotonation of the 2-a-hydroxybenzyl-ThDP
intermediate, aiding product release. H281 is involved
in protonation of the enamine’. However, the main
role proposed for H70 is as a Brønsted acid to proton-
ate the carbonyl of the substrate to promote addition
of the ylide of ThDP. On the other hand, if H70 is the
most effective proton donor in suppressing the return
of carbon dioxide, its replacement by alanine would
lead to a structural change and the loss of one site
from which a proton could be donated. However, as
proton association is a dynamic and rapid process,
catalysis is still highly effective as other groups serve
as proton sources such that the k
cat
value remains well
above the rate constant for the uncatalyzed decarbox-
ylation of MTh, suggesting that rescue is possible by
other groups [54].
Interpreting saturation mutagenesis
In a recent report on BFD, Yep et al. [54] used satura-
tion mutagenesis as a probe to determine the extent of
decrease in activity as a result of substitutions for
active site histidines H70 and H281. They report that
H281F has a k
cat
value that is 20% of that of the

native protein. This is consistent with other acid
groups being able to take on its role but with lower
efficiency. They also found that replacing H70 with
threonine or leucine decreases the activity to approxi-
mately 3% of that of the native protein, while the pre-
viously reported substitution with alanine causes a
reduction of k
cat
to 0.025% of the native level, indicat-
ing that a more serious structural change occurs. As
none of the substituted groups have a specific role to
play as a catalyst, it is likely that the structural
changes have varying effects on the arrangement of
groups in the active site that promote the departure of
carbon dioxide. Nonetheless, even the slowest mutant,
H70A, is approximately 1000 times more reactive than
MTh. Yep et al. [54] state that mutagenesis can thus
be misleading in assigning mechanistic roles, but it is
clear that the roles for the putative catalytic residues
of BFD are not discreet, and it is difficult, if not
impossible, to assign them definitive functions by any
experimental means.
Structural implications in facilitated
carbon dioxide departure – evidence
from sequence homology and
saturation mutagenesis
The relevance of specific amino acid residues can often
be deduced from their conservation in enzymes from
different species or in those sharing the same or similar
mechanisms. Given that the mechanism of ThDP-

dependent decarboxylation is similar regardless of the
substitution pattern of the substrate, a significant
amount of sequence homology should exist among
decarboxylases if specific catalytic residues are neces-
sary for catalysis. However, this is not the case [55].
Apart from the residues that bind ThDP in the active
site, the structures of pyruvate decarboxylase and BFD
are very different [23,54,56]. Thus, it is likely that the
structural features of the protein that interact with the
substrate also facilitate the decarboxylation process.
This is consistent with the observation that the H281F
mutant of BFD retains the greatest activity of the
active site variants at this position despite the different
functional groups in the side chains [54].
As steric bulk is a common structural feature of histi-
dine and phenylalanine, the role of the amino acid at
position 281 may be to influence the conformation of
ThDP-bound intermediates. Conformational fluctua-
tion of BFD intermediates in the active site is observed
in ThDP carboligation, which generates 2-hydroxy-
ketones [57]. The stereoselectivity of these reactions is a
function of the size of the acyl donor bound by thia-
min. Less differentiation was observed with small sub-
stituents on the acyl donor, presumably due to a small
difference in the kinetic barrier between conformations,
leading to attack of either face of the acyl acceptor [57].
Conclusions
ThDP-dependent enzymes catalyze remarkable reac-
tions, providing catalysis well beyond that which
would result from the cofactor alone. We have pro-

posed that decarboxylation is enhanced by the ability
of the protein to provide relatively acidic groups in
R. Kluger and S. Rathgeber Catalyzing separation of carbon dioxide
FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS 6097
proximity to the specific site of decarboxylation. The
protein accommodates the substrate specifically and
presents a resilient proton pool from sites that are con-
siderably more acidic than solvent water. While a
mutation that replaces a Brønsted acid with a hydro-
carbon lowers overall activity, the residual activity is
still much greater than that of the intermediate itself in
the absence of the protein. These results suggest that
ThDP intermediates undergo facilitated reactions in
protein environments, and that there is sufficient cata-
lytic redundancy in the sources of protons for acid
catalysis to assist departure of carbon dioxide.
Acknowledgements
We thank the Natural Sciences and Engineering
Research Council of Canada for support through a
Discovery Grant (R.K.) and a Postgraduate Scholar-
ship (S.R.).
References
1 Jencks WP & Page MI (1974) ‘Orbital steering’,
entropy, and rate accelerations. Biochem Biophys Res
Commun 57, 887–892.
2 Page MI & Jencks WP (1971) Entropic contributions to
rate accelerations in enzymic and intramolecular reac-
tions and the chelate effect. Proc Natl Acad Sci USA
68, 1678–1683.
3 Kluger R & Tittmann K (2008) Thiamin diphosphate

catalysis: enzymic and nonenzymic covalent intermedi-
ates. Chem Rev 108, 1797–1833.
4 Kluger R (1987) Thiamin diphosphate: a mechanistic
update on enzymic and nonenzymic catalysis of decar-
boxylation. Chem Rev 87, 863–876.
5 Kluger R (1997) Lessons from thiamin-watching. Pure
Appl Chem 69, 1957–1967.
6 Kluger R & Gish G (1987) Stereochemical aspects of
thiamin catalysis. In Thiamin Pyrophosphate Enzymes
(Schowen RL & Schellenberger A, eds), pp. 3–9. CRC
Press Inc., Boca Raton, FL.
7 Breslow R (1957) The mechanism of thiamine action.
II. Rapid deuterium exchange in thiazolium salts. JAm
Chem Soc 79, 1762–1763.
8 Breslow R (1957) Mechanism of thiamine action.
Participation of a thiazolium zwitterion. Chem and Ind,
893–894.
9 Breslow R (1958) Mechanism of thiamine action. IV.
Evidence from studies on model systems. J Am Chem
Soc 80, 3719–3726.
10 Schowen RL (1998) Thiamin-dependent enzymes. In
Comprehensive Biological Catalysis (Sinnott ML, ed.),
pp. 217–266. Academic Press, London.
11 Hu Q & Kluger R (2002) Reactivity of intermediates in
benzoylformate decarboxylase: avoiding the path to
destruction. J Am Chem Soc 124, 14858–14859.
12 Kluger R, Chin J & Smyth T (1981) Thiamin-catalyzed
decarboxylation of pyruvate. Synthesis and reactivity
analysis of the central, elusive intermediate, alpha-
lactylthiamin. J Am Chem Soc 103, 884–888.

13 Tittmann K, Golbik R, Hubner G & Ghisla S (1999)
Single step analysis of catalysis of pyruvate oxidase
from Lactobacillus plantarum. Kinetics, mechanism and
regulation. In Flavins and Flavoproteins 1999 (Ghisla S,
Kronek P, Macheroux P & Sund H, eds), pp. 395–400.
Agency for Scientific Publications, Berlin.
14 Tittmann K, Golbik R, Uhlemann K, Khailova L,
Schneider G, Patel M, Jordan F, Chipman DM,
Duggleby RG & Hubner G (2003) NMR analysis of
covalent intermediates in thiamin diphosphate enzymes.
Biochemistry 42, 7885–7891.
15 Wille G, Meyer D, Steinmetz A, Hinze E, Golbik R &
Tittmann K (2006) The catalytic cycle of a thiamin
diphosphate enzyme examined by cryocrystallography.
Nat Chem Biol 2, 324–328.
16 Kluger R, Lam JF, Pezacki JP & Yang C-M (1995)
Diverting thiamin from catalysis to destruction.
Mechanism of fragmentation of N(1’)-methyl-2-
(1-hydroxybenzyl)thiamin. J Am Chem Soc 117,
11383–11389.
17 Crosby J & Lienhard GE (1970) Mechanisms of
thiamine-catalyzed reactions. Kinetic analysis of the
decarboxylation of pyruvate by 3,4-dimethylthiazolium
ion in water and ethanol. J Am Chem Soc 92, 5707–5716.
18 Kluger R, Ikeda G, Hu Q, Cao P & Drewry J (2006)
Accelerating unimolecular decarboxylation by preassoci-
ated acid catalysis in thiamin-derived intermediates:
implicating Brønsted acids as carbanion traps in
enzymes. J Am Chem Soc 128, 15856–15864.
19 Kluger R & Yu D (2006) Protein-enhanced decarboxyl-

ation of the covalent intermediate in benzoylformate
decarboxylase – desolvation or acid catalysis? Bioorg
Chem 34, 337–344.
20 Crosby J, Stone R & Lienhard GE (1970) Mechanisms
of thiamine-catalyzed reactions. Decarboxylation of
2-(1-carboxy-l-hydroxyethyl)-3,4-dimethylthiazolium
chloride. J Am Chem Soc 92, 2891–2900.
21 Hasson MS, Muscate A, Kenyon GL & Petsko GE
(1996) The crystal structure of benzoylformate decarbox-
ylase. In Biochemistry and Physiology of Thiamin
Diphosphate Enzymes (Bisswanger H & Schellenberger A,
eds), pp. 174–176. A & C Intemann Verlag, Prien,
Germany.
22 Hasson MS, Muscate A, Henehan GT, Guidinger PF,
Petsko GA, Ringe D & Kenyon GL (1995) Purification
and crystallization of benzoylformate decarboxylase.
Protein Sci 4, 955–959.
Catalyzing separation of carbon dioxide R. Kluger and S. Rathgeber
6098 FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS
23 Hasson MS, Muscate A, McLeish MJ, Polovnikova LS,
Gerlt JA, Kenyon GL, Petsko GA & Ringe D (1998)
The crystal structure of benzoylformate decarboxylase
at 1.6 angstrom resolution: diversity of catalytic
residues in thiamin diphosphate-dependent enzymes.
Biochemistry 37, 9918–9930.
24 Kluger R, Lam JF & Kim C-S (1993) Decomposition
of 2-(1-hydroxybenzyl)thiamin in neutral aqueous
solutions: benzaldehyde and thiamin are not the
products. Bioorg Chem 21, 275–283.
25 Oka Y, Kishimoto S & Hirano H (1970) Vitamin B1

and related compounds. CIX. A novel cleavage of
thiamin and its homologs by the reaction with aromatic
aldehydes. Chem Pharm Bull 18, 527–533.
26 Kluger R & Moore IF (2000) Destruction of vitamin
B1 by benzaldehyde. Reactivity of intermediates in the
fragmentation of N1’-benzyl-2-(1-hydroxybenzyl)
thiamin. J Am Chem Soc 122, 6145–6150.
27 Moore IF & Kluger R (2002) Substituent effects in
carbon–nitrogen cleavage of thiamin derivatives.
Fragmentation pathways and enzymic avoidance of
cofactor destruction. J Am Chem Soc 124, 1669–1673.
28 Hu Q & Kluger R (2004) Fragmentation of the
conjugate base of 2-(1-hydroxybenzyl)thiamin: does
benzoylformate decarboxylase prevent orbital overlap
to avoid it? J Am Chem Soc 126 , 68–69.
29 Dirmaier LJ, Garcia GA, Kozarich JW & Kenyon GL
(1986) Inhibition of benzoylformate decarboxylase by
[p-(bromomethyl)benzoyl]formate. Enzyme-catalyzed
modification of thiamine pyrophosphate by halide
elimination and tautomerization. J Am Chem Soc 108,
3149–3150.
30 Dirmaier LJ, Garcia GA, Kozarich JW & Kenyon GL
(1986) The inhibition of benzoylformate decarboxylase
by p-halo-methylbenzoylformates. Fed Proc 45, 1608–
1608.
31 Hopmann RFW, Brugnoni GP & Fol B (1982) Thiamin
catalysis. 2. Kinetics and mechanism of the generation
of the yellow form of thiamin. J Am Chem Soc 104,
1341–1344.
32 Lahiri SD, Zhang G, Dunaway-Mariano D & Allen

KN (2003) The pentacovalent phosphorus intermediate
of a phosphoryl transfer reaction. Science 299, 2067–
2071.
33 Lahiri SD, Zhang G, Dunaway-Mariano D & Allen
KN (2002) Caught in the act: the structure of phos-
phorylated beta-phosphoglucomutase from Lactococcus
lactis. Biochemistry 41, 8351–8359.
34 Acevedo O & Jorgensen WL (2006) Medium effects on
the decarboxylation of a biotin model in pure and
mixed solvents from QM ⁄ MM simulations. J Org Chem
71, 4896–4902.
35 Hu Q & Kluger R (2005) Making thiamin work faster:
acid-promoted separation of carbon dioxide. JAm
Chem Soc 127, 12242–12243.
36 Warshel A, Sharma PK, Kato M, Xiang Y, Liu H &
Olsson MH (2006) Electrostatic basis for enzyme cataly-
sis. Chem Rev 106, 3210–3235.
37 Jencks WP (1976) Enforced general acid-base catalysis
of complex reactions and its limitations. Acc Chem Res
9, 425–432.
38 Venkatasubban KS & Schowen RL (1984) Protolytic
catalysis of anilide methanolysis. Spectator catalysis of
leaving-group departure. J Org Chem 49, 653–655.
39 Gao J, Byun KL & Kluger R (2004) Catalysis by
enzyme conformational change.
Top Curr Chem 238,
113–136.
40 Gao J, Ma S, Major Dan T, Nam K, Pu J & Truhlar
DG (2006) Mechanisms and free energies of enzymatic
reactions. Chem Rev 106, 3188–3209.

41 Page MI & Jencks WP (1972) Intramolecular general
base catalysis in the aminolysis of acetylimidazole and
methyl formate by diamines. J Amer Chem Soc 94,
8818–8827.
42 Kirby AJ (1980) Effective molarities for intramolecular
reactions. Adv Phys Org Chem 17, 183–278.
43 Sauers CK, Jencks WP & Groh S (1975) The alcohol–
bicarbonate–water system. Structure–reactivity studies
on the equilibria for formation of alkyl monocarbonates
and on the rates of their decomposition in aqueous
alkali. J Am Chem Soc 97, 5546–5553.
44 Bera AK, Polovnikova LS, Roestamadji J, Widlanski
TS, Kenyon GL, McLeish MJ & Hasson MS (2007)
Mechanism-based inactivation of benzoylformate
decarboxylase, a thiamin diphosphate-dependent
enzyme. J Am Chem Soc 129, 4120–4121.
45 Chipman DM, Duggleby RG & Tittmann K (2005)
Mechanisms of acetohydroxyacid synthases. Curr Opin
Chem Biol 9, 475–481.
46 Tittmann K, Golbik R, Uhlemann K, Khailova L, Patel
MS, Jordan F, Chipman DM, Duggleby RG & Schnei-
der G (2004) How thiamine works in enzymes: time-
resolved NMR snapshots of TDP-dependent enzymes in
action. In Thiamine. Catalytic Mechanisms (Jordan F &
Patel MS, eds), pp. 57–76. Marcel Dekker, New York,
NY.
47 Campbell RE, Mosimann SC, van De Rijn I, Tanner
ME & Strynadka NC (2000) The first structure of
UDP-glucose dehydrogenase reveals the catalytic
residues necessary for the two-fold oxidation. Biochem-

istry 39, 7012–7023.
48 Campbell RE, Sala RF, van de Rijn I & Tanner ME
(1997) Properties and kinetic analysis of UDP-glucose
dehydrogenase from group A streptococci. Irreversible
inhibition by UDP-chloroacetol. J Biol Chem 272,
3416–3422.
49 Yip VL, Thompson J & Withers SG (2007) Mechanism
of GlvA from Bacillus subtilis: a detailed kinetic analy-
sis of a 6-phospho-alpha-glucosidase from glycoside
hydrolase family 4. Biochemistry 46, 9840–9852.
R. Kluger and S. Rathgeber Catalyzing separation of carbon dioxide
FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS 6099
50 Yip VL & Withers SG (2006) Mechanistic analysis of
the unusual redox-elimination sequence employed by
Thermotoga maritima BglT: a 6-phospho-beta-glucosi-
dase from glycoside hydrolase family 4. Biochemistry
45, 571–580.
51 Varrot A, Yip VL, Li Y, Rajan SS, Yang X, Anderson
WF, Thompson J, Withers SG & Davies GJ (2005)
NAD
+
and metal-ion dependent hydrolysis by family 4
glycosidases: structural insight into specificity for phos-
pho-beta-d-glucosides. J Mol Biol 346, 423–435.
52 Rajan SS, Yang X, Collart F, Yip VL, Withers SG,
Varrot A, Thompson J, Davies GJ & Anderson WF
(2004) Novel catalytic mechanism of glycoside hydro-
lysis based on the structure of an NAD
+
⁄ Mn

2+
-depen-
dent phospho-alpha-glucosidase from Bacillus subtilis.
Structure 12, 1619–1629.
53 Polovnikova ES, McLeish MJ, Sergienko EA, Burgner
JT, Anderson NL, Bera AK, Jordan F, Kenyon GL &
Hasson MS (2003) Structural and kinetic analysis of
catalysis by a thiamin diphosphate-dependent enzyme,
benzoylformate decarboxylase. Biochemistry 42, 1820–
1830.
54 Yep A, Kenyon GL & McLeish MJ (2008) Saturation
mutagenesis of putative catalytic residues of benzoylfor-
mate decarboxylase provides a challenge to the accepted
mechanism. Proc Natl Acad Sci USA 105, 5733–5738.
55 Jordan F (2003) Current mechanistic understanding of
thiamin diphosphate-dependent enzymatic reactions.
Nat Prod Rep 20, 184–201.
56 Schenk G, Leeper FJ, England R, Nixon PF & Dugg-
leby RG (1997) The role of His113 and His114 in pyru-
vate decarboxylase from Zymomonas mobilis. Eur J
Biochem 248, 63–71.
57 Iding H, Dunnwald T, Greiner L, Liese A, Muller M,
Siegert P, Grotzinger J, Demir AS & Pohl M (2000)
Benzoylformate decarboxylase from Pseudomonas putida
as stable catalyst for the synthesis of chiral 2-hydroxy
ketones. Chem Eur J 6, 1483–1495.
Catalyzing separation of carbon dioxide R. Kluger and S. Rathgeber
6100 FEBS Journal 275 (2008) 6089–6100 ª 2008 The Authors Journal compilation ª 2008 FEBS