Published on Web 11/20/2002
Reactivity of Intermediates in Benzoylformate Decarboxylase: Avoiding the
Path to Destruction
Qingyan Hu and Ronald Kluger*
DaVenport Chemical Laboratory, Department of Chemistry, UniVersity of Toronto, Toronto, Canada M5S 3H6
Received August 1, 2002
Enzymes that promote the decarboxylation of 2-oxocarboxylic
acids utilize thiamin diphosphate (TDP, 1a in Scheme 1) as a
cofactor, following the general pathway that was deduced by
Breslow.1 Benzoylformate decarboxylase (BFD) is a bacterial
TDP-dependent decarboxylase that has been the subject of structural
and mechanistic studies.4 On the basis of the pattern of reactions
of R-ketoacid decarboxylases, we assume that R-mandelylTDP
Scheme 1. Thiamin and TDP Pathways for Benzoylformate
-3
-9
(MTDP, 2a) forms on the enzyme from TDP and benzoylformate,
permitting the release of carbon dioxide, to produce a carbanion-
9
enamine (3a). This is protonated to give 2-(1-hydroxybenzyl)TDP,
HBzTDP (4a), the precursor of benzaldehyde.
To study the reactivity of intermediates in the benzoylformate
decarboxylation process, we developed a synthesis of R-mande-
lylthiamin (MT, 2b). Direct analogy to producing R-lactylthiamin
from ethyl pyruvate and thiamin11 did not produce MT. This was
overcome by addition of magnesium chloride. Thus, thiamin
hydrochloride (5.0 g, 15 mmol) was suspended in 80 mL of absolute
ethanol, and 50 mL of anhydrous ethanol containing 2 equiv of
sodium ethoxide was added. Ethyl benzoylformate solution, con-
taining 8.2 mL of ethyl benzoylformate and 0.6 g of anhydrous
magnesium chloride in 50 mL of absolute ethanol, was deoxygen-
ated and immediately added under nitrogen to the basic thiamin
solution. After 30 min of stirring at 0 °C, gaseous hydrogen chloride
was added to acidify the solution, and the precipitate, consisting
of sodium chloride and thiamin, was removed by filtration. The
filtrate was concentrated, and the yellow liquid was dissolved in
CN), 5.90 (1H, d, J ) 18 Hz, H
7
a
H
b
CN), 6.82 (1H, s, H-pyrimidine),
.30 (3H, m, aromatic), 7.55 (2H, m, aromatic).
The observed first-order rate coefficient for decarboxylation of
-
4
-1
MT at pH 5, 6, and 7 (25 °C) is 3.1 ((0.3) × 10
s . For
1
-1
comparison, the enzymic kcat is 8.1 × 10 s . Thus, the decar-
boxylation step is at least 200 000 times faster on the enzyme. This
11
is similar to the acceleration in pyruvate decarboxylase. We
assume that the acceleration is achieved by desolvation of the
intermediate as it is produced from thiamin diphosphate and the
10,11
substrate.
The decarboxylation of MT tests an important mechanistic issue.
The carbanion that results from the loss of carbon dioxide from
MT should be the same as that which arises from transfer of a
proton from the C2R position of HBzT to a Brønsted base. It has
been surmised that the species generated from HBzT fragments
directly into pyrimidine and thiazole moieties1
25 mL of water and extracted with three 50 mL portions of
dichloromethane to remove excess ethyl benzoylformate. The
aqueous layer was stirred with Chelex for 1 h at pH 6 to remove
magnesium chloride. After being filtered, the aqueous solution was
lyophilized to dryness. The yellow solid was dissolved in methanol
2-14
(Scheme 2, kf):
and passed through a 3 × 30 cm cellulose column and dried to
phenyl thiazole ketone, PTK, 5-(2-hydroxyethyl-4-methyl-thiazol-
2-yl)-phenyl-methanone, and 2,5-dimethyl-pyrimidin-4-ylamine,
DPA. This requires that loss of CO2 from MT generates the
carbanion at C2R,15 whose protonation by Brønsted acids to give
HBzT will compete with the uncatalyzed fragmentation. Further-
more, because the decarboxylation step is rate determining, added
acids should not affect the observed first-order rate coefficient for
decomposition of MT. They will divert the resulting carbanion
toward HBzT and away from PTK and DPA (Scheme 2).
Consistent with these predictions, significant amounts of PTK
1
give the product. H NMR (400 MHz DCl in D
2
O relative to internal
DSS): δ 1.30 (3H, t, J ) 7.2 Hz, CH
pyrimidine), 2.49 (3H, s, CH -thiazole), 3.25 (2H, t, J ) 5.8 Hz,
CH CH OH), 3.94 (2H, t, J ) 5.8 Hz, CH CH OH), 4.43 (2H, q,
J ) 7.2 Hz, CH CH OCO), 5.36 (1H, d, J ) 18 Hz, H CN),
.86 (1H, d, J ) 18 Hz, H CN), 6.81 (1H, s, H-pyrimidine),
.32 (3H, m, aromatic), 7.54 (2H, m, aromatic). The ethyl ester
3 2 3
CH OCO), 2.41 (3H, s, CH -
3
14
2
2
2
2
3
2
a b
H
5
7
a b
H
was hydrolyzed by dissolving in 12 M HCl and left for 3 days at
room temperature. After concentration to remove excess hydrogen
i
and DPA form from MT along with HBzT (pH 6.2, 0.03 M KP ),
chloride and lyophilization, the MT chloride hydrochloride was
1
appearing with the same rate coefficient (k ). Furthermore, increas-
obtained (stored dry at -20 °C). H NMR (400 MHz, DCl in D
2
O
1
ing the concentration of buffer increases the proportion of HBzT
relative to internal DSS): δ 2.39 (3H, s, CH
3
-pyrimidine), 2.49
-thiazole), 3.23 (2H, t, J ) 5.8 Hz, CH CH OH), 3.94
2H, t, J ) 5.8 Hz, CH CH OH), 5.41 (1H, d, J ) 18 Hz, H
without affecting the observed rate constant (Figure 1).
From measurements of the reactivity of derivatives of HBzT,
(
3H, s, CH
3
2
2
1
4
(
2
2
a b
H -
we know that fragmentation of the C2R carbanion of pyrimidine-
+
5
-1
protonated HBzT (HBzT(H )) is very rapid: k
f
is ∼10 s at 40
*
To whom correspondence should be addressed. E-mail: rkluger@
3
4 -1
chem.utoronto.ca.
°C (we assume k
f
≈ 10 -10 s at 25 °C). Because we are working
1
4858 J. AM. CHEM. SOC. 2002, 124, 14858-14859
9
10.1021/ja027976r CCC: $22.00 © 2002 American Chemical Society