Synthesis with good enantiomeric excess of both enantiomers of α-ketols and acetolactates by two thiamin diphosphate-dependent decarboxylases

Article abstract of DOI:10.1016/j.bioorg.2006.09.006

In addition to the decarboxylation of 2-oxo acids, thiamin diphosphate (ThDP)-dependent decarboxylases/dehydrogenases can also carry out so-called carboligation reactions, where the central ThDP-bound enamine intermediate reacts with electrophilic substrates. For example, the enzyme yeast pyruvate decarboxylase (YPDC, from Saccharomyces cerevisiae) or the E1 subunit of the Escherichia coli pyruvate dehydrogenase complex (PDHc-E1) can produce acetoin and acetolactate, resulting from the reaction of the central thiamin diphosphate-bound enamine with acetaldehyde and pyruvate, respectively. Earlier, we had shown that some active center variants indeed prefer such a carboligase pathway to the usual one [Sergienko, Jordan, Biochemistry 40 (2001) 7369-7381; Nemeria et al., J. Biol. Chem. 280 (2005) 21,473-21,482]. Herein is reported detailed analysis of the stereoselectivity for forming the carboligase products acetoin, acetolactate, and phenylacetylcarbinol by the E477Q and D28A YPDC, and the E636A and E636Q PDHc-E1 active-center variants. Both pyruvate and β-hydroxypyruvate were used as substrates and the enantiomeric excess was analyzed by a combination of NMR, circular dichroism and chiral-column gas chromatographic methods. Remarkably, the two enzymes produced a high enantiomeric excess of the opposite enantiomer of both acetoin-derived and acetolactate-derived products, strongly suggesting that the facial selectivity for the electrophile in the carboligation is different in the two enzymes. The different stereoselectivities exhibited by the two enzymes could be utilized in the chiral synthesis of important intermediates.

Full text of DOI:10.1016/j.bioorg.2006.09.006

Bioorganic Chemistry 34 (2006) 380–393
www.elsevier.com/locate/bioorg
Synthesis with good enantiomeric excess
of both enantiomers of a-ketols and acetolactates
by two thiamin diphosphate-dependent
decarboxylases
*
Ahmet Baykal, Sumit Chakraborty, Afua Dodoo, Frank Jordan
Department of Chemistry, Rutgers the State University, Newark, NJ 07102, USA
Received 29 August 2006
Frank Jordan dedicates this paper to the memory of a valued collaborator and friend Miriam Hasson.
Abstract
In addition to the decarboxylation of 2-oxo acids, thiamin diphosphate (ThDP)-dependent decar-
boxylases/dehydrogenases can also carry out so-called carboligation reactions, where the central
ThDP-bound enamine intermediate reacts with electrophilic substrates. For example, the enzyme
yeast pyruvate decarboxylase (YPDC, from Saccharomyces cerevisiae) or the E1 subunit of the Esch-
erichia coli pyruvate dehydrogenase complex (PDHc-E1) can produce acetoin and acetolactate,
resulting from the reaction of the central thiamin diphosphate-bound enamine with acetaldehyde
and pyruvate, respectively. Earlier, we had shown that some active center variants indeed prefer such
a carboligase pathway to the usual one [Sergienko, Jordan, Biochemistry 40 (2001) 7369–7381;
Nemeria et al., J. Biol. Chem. 280 (2005) 21,473–21,482]. Herein is reported detailed analysis of
the stereoselectivity for forming the carboligase products acetoin, acetolactate, and phenylacetyl-
carbinol by the E477Q and D28A YPDC, and the E636A and E636Q PDHc-E1 active-center vari-
ants. Both pyruvate and b-hydroxypyruvate were used as substrates and the enantiomeric excess was
analyzed by a combination of NMR, circular dichroism and chiral-column gas chromatographic
methods. Remarkably, the two enzymes produced a high enantiomeric excess of the opposite enan-
tiomer of both acetoin-derived and acetolactate-derived products, strongly suggesting that the facial
selectivity for the electrophile in the carboligation is different in the two enzymes. The different
*
Corresponding author. Fax: +1 973 353 1264.
E-mail address: frjordan@newark.rutgers.edu (F. Jordan).
0045-2068/$ - see front matter Ó 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.bioorg.2006.09.006

A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
381
stereoselectivities exhibited by the two enzymes could be utilized in the chiral synthesis of important
intermediates.
Ó 2006 Elsevier Inc. All rights reserved.
Keywords: Thiamin diphosphate; Yeast pyruvate decarboxylase; Escherichia coli pyruvate dehydrogenase E1
subunit; Carboligase side reactions; b-Hydroxypyruvate; Acetoin; Acetolactate; Enzymatic chiral synthesis;
Enantiomeric excess; Circular dichroism
1. Introduction
The use of thiamin diphosphate-dependent (ThDP) enzymes in synthesis leading to car-
boligation is a well-exploited idea that has been reviewed recently [1–4]. It takes advantage
of the multiple catalytic potentials of the central enamine intermediate [5,6] as depicted
with the reaction of yeast pyruvate decarboxylase (YPDC) (Fig. 1). With some enzymes,
this carboligation, or its reverse, is the natural course of the pathway while for virtually all
2-oxoacid decarboxylases it is a side reaction. Earlier, we had reported that certain active-
center substitutions of YPDC could convert it to an effective acetoin synthase [7] (the
E477Q variant; resulting from the reaction of the enamine with acetaldehyde the product
of the reaction), or acetolactate synthase [7] (the D28A or D28N variants; resulting from
the reaction of the enamine with a second molecule of pyruvate). More recently [8], we
realized that the E636A (or E636Q) substitution at the active center of the pyruvate dehy-
drogenase complex E1 subunit from Escherichia coli (PDHc-E1) also converted it to an
effective carboligase, giving the same two products. More remarkably, the two ThDP
enzymes, both of which carry out decarboxylation of pyruvic acid as a first step, gave a
preponderance of the opposite enantiomer of both carboligase products.
Fig. 1. Mechanism of yeast pyruvate decarboxylase and of the carboligase reactions.

382
A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
It is important to note that each of these wild-type enzymes only give acetoin and at
<1% yield. In an extension of those studies, we report amplification of our qualitative
reports by determining the enantiomeric excess (ee) of each reaction. We also report stud-
ies on a frequently used pyruvate derivative b-hydroxypyruvate, which is accepted by both
enzymes and also produces the carboligase products. Finally, we also produce (R)-pheny-
lacetylcarbinol [(R)-PAC] using the YPDC variants with outstanding ee. The studies are
very much assisted by the use of circular dichroism experiments. The ability to produce
the opposite enantiomer of the same chiral compound by two closely related enzymes
makes the findings of potential synthetic significance.
2. Materials and methods
2.1. Materials
The chemicals acetoin, sodium pyruvate, (R)-(+)-a-[trimethylsilyl)oxy]-benzeneaceto-
nitrile, and methylmagnesium iodide were bought from Sigma–Aldrich and were used
without any further purification.
2.2. Methods
2.2.1. Reactions with the enzymes
Acetoin, acetolactate and PAC formation with YPDC variants were carried out in
10 mM KH2PO4 buffer (pH 6.0, the pH optimum), containing 2 mM MgCl2, 0.1 mM
ThDP [9]. Acetoin and acetolactate formation with E. coli PDHc-E1 were carried out in
10 mM KH2PO4(pH 7.0, the pH optimum for the E1 subunit) containing 2 mM MgCl2,
0.1 mM ThDP [8].
2.2.2. (R)-Phenylacetylcarbinol formation with the E477Q and D28A variants of YPDC
The E477Q and D28A variants were dissolved (0.24 and 3.35 mg/mL, respectively)
in 10 mM KH2PO4 (pH 6.0) with 2 mM MgCl2, 0.1 mM ThDP, then 4 mM benzal-
dehyde was added at 30 °C. The reaction was initiated by the addition of 4 mM
pyruvate.
2.2.3. Acetolactate formation with the E636A and E636Q PDHc-E1 variants and the D28A
YPDC variant
The D28A YPDC variant was dissolved (3.35 mg/mL) in 10 mM KH2PO4 (pH 6.0)
with 2 mM MgCl2 and 0.1 mM ThDP added at 30 °C. The E636Q and E636A PDHc-
E1 variants were dissolved (both at 1 mg/mL) in 10 mM KH2PO4 (pH 7.0) with 2 mM
MgCl2, 0.1 mM ThDP added at 25 °C. Acetolactate formation was started by addition
of 20 mM pyruvate.
2.2.4. Acetoin formation by the E636A and E636Q PDHc-E1 variants and the E477Q and
D28A YPDC variants
The E477Q and D28A YPDC variants were dissolved (0.24 and 3.35 mg/mL,
respectively) in 10 mM KH2PO4 (pH 6.0) with 2 mM MgCl2, 0.1 mM ThDP, and
100 mM acetaldehyde was added at 30 °C. The E636Q and E636A PDHc-E1 variants

A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
383
were dissolved (both at 1 mg/mL) in 10 mM KH2PO4 at pH 7.0 with 2 mM MgCl2,
0.1 mM ThDP added at 25 °C. Acetoin formation was started by addition of 20 mM
pyruvate.
2.2.5. 1,3,4-Trihydroxy-2-butanone formation by parental PDHc-E1 and its E636A and
E636Q variants and the E477Q YPDC variant
PDHc-E1 and its E636Q and E636A variants were dissolved (1 mg/mL) in 10 mM
KH2PO4(pH 7.0) with 2 mM MgCl2 and 0.1 mM ThDP added at 25 °C. The E477Q
YPDC variant was dissolved (0.24 mg/mL) in 10 mM KH2PO4 (pH 6.0) with 2 mM
MgCl2, 0.1 mM ThDP added at 25 °C. To initiate formation of 1,3,4-trihydroxy-2-buta-
none, the reaction was started with simultaneous addition of 40 mM b-hydroxypyruvate
and 100 mM glycolaldehyde. Parental PDHc-E1 did not need glycolaldehyde to produce
1,3,4-trihydroxy-2-butanone.
2.2.6. 2,4-Dihydroxy-2-hydroxymethyl-3-oxo-butanoic acid formation by the PDHc-E1
E636A and E636Q variants
The E636Q and E636A variants were dissolved (1 mg/mL) in 10 mM KH2PO4 (pH 7.0)
with 2 mM MgCl2 and 0.1 mM ThDP added at 25 °C. The reaction was initiated by addi-
tion of 40 mm b-hydroxypyruvate.
2.2.7. Chemical synthesis
To a solution of 2 mL of 3.0 M methylmagnesium iodide (6 mmol) in 7 mL ether was
added dropwise 1 mL of (R)-(+)-a-[(trimethylsilyl)oxy]-benzeneacetonitrile (4.61 mmol)
in 8 mL of ether. After the addition was completed the reaction mixture was left at room
temperature for 2 h. During the first five minutes of the reaction the mixture turned cloudy
yellow and after 10 min precipitation was observed. After 2 h of reaction time the mixture
was placed on ice for 15 min. Next, the mixture was poured on 50 g of ice containing 2 mL
of concentrated sulfuric acid and was left overnight to hydrolyze the trimethylsilyl group.
The ether layer was first rinsed with 20 mL of 10% HCl then with brine. The organic layer
was dried over magnesium sulfate and the volatiles were removed in vacuo. NMR chem-
ical shifts and GC–MS fragmentation patterns confirmed the product. NMR (DMSO-d6/
TMS): d (ppm) 2.05 (s, 3H), 4.25 (s, 1H), 5.05 (s, 1H), 7.24 (m, 5H). GC–MS, [M+] 150,
107, 79, 51 [10].
2.2.8. Overexpression and purification of the YPDC, PDHc-E1, and their active center
variants
Overexpression and purification of the PDHc-E1 and its E636A and E636Q variants
was carried out following the protocol described previously for E. coli PDHc-E1 [8]. Over-
expression of the E477Q and D28A active site variants of YPDC was carried out following
the protocol described previously [11]. The E477Q variant has a His6-tag attached to its C-
terminal residue hence it was purified on a Talon column [11]. The D28A variant was puri-
fied following the protocol reported previously [11].
2.2.9. Circular dichroism (CD)
Experiments were carried out on an AVIV 202 and an Applied Photophysics Chirascan
spectrometer. Spectra were recorded in the 250–350 nm range. For each sample, the base-
line was first recorded with the buffer and then appropriate amounts of MgCl2 and ThDP

384
A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
was added and the spectrum was re-recorded to make certain that there is no signal in the
wavelength range studied. Product formation was monitored at the ellipticity maximum of
the particular compound. After termination of the reaction, the enzyme was removed with
a Centricon microcentriprep unit (1.5 mL and the supernatant was scanned with CD. For
acetoin and (R)-PAC samples a 0.1 cm, and for acetolactate a 1.0 cm path length cell was
used.
2.2.10. Chiral GC analysis
Analysis was carried out on a HP 5950 gas chromatograph equipped with a Chiraldex
B-DM chiral column (Astec, Advanced Separation Technologies, Inc.) and a flame ioni-
zation detector at a flow rate of 0.7 mL/min. Acetoin was analyzed at 50 °C, while pheny-
lacetylcarbinol was analyzed at 100 °C. A 20 lL volume of sample was injected with a
100:1 split ratio. Acetoin and phenylacetylcarbinol were extracted from the reaction mix-
ture with chloroform and injected onto the Chiraldex B-DM chiral column. Since this col-
umn interacts more favorably with the (S)-enantiomer thus delaying its retention time, we
could readily assign the enantiomers. We could not analyze acetolactate samples with the
chiral GC method since it would be decarboxylated under GC conditions. Therefore,
acetolactate samples were scanned with CD and the enantiomeric excess was calculated
with the reported molar ellipticity [12].
2.2.11. Determination of steady-state kinetic parameters for b-hydroxypyruvate with
YPDC
Activity was assayed with the NADH/alcohol dehydrogenase (ADH) coupled assay
[13], converting the glycolaldehyde to ethylene glycol and monitoring the depletion of
NADH at 340 nm. One unit of YPDC activity is defined as the amount of protein
required to convert 1.0 lmol of b-hydroxypyruvate to glycolaldehyde per minute at
25 °C. Kinetic studies were carried out on a Varian Cary 300 Bio UV–visible spectro-
photometer. The kinetic parameters were calculated by curve fitting using the Sigma-plot
program (Systat). The total sample solution was 1 mL. YPDC (5 lL of 6.88 mg/mL,
containing 34.4 lg enzyme) was added to assay buffer containing 20 mM KH2PO4
(pH 6.0), 1 mM ThDP, 2 mM MgCl2, 20 U of yeast ADH (360 U/mg), and 0.15 mg
NADH (both from Sigma, St. Louis, MO) was added to the mixture. The enzyme
was assayed with 0.5–100 mM of b-hydroxypyruvate and activity was measured by mon-
itoring the NADH depletion at 25 °C using the coupled enzyme assay.
2.2.12. Determination of steady-state kinetic parameters for b-hydroxypyruvate with PDHc-
E1
The PDHc-E1 activity was measured by reconstituting PDHc activity with saturating
amount of E2-E3 subcomplex (the mass ratio of E1 to E2–E3 sub-complex was 1:5) using
a Varian Cary 300 Bio UV–visible spectrophotometer, monitoring the b-hydroxypyruvate-
dependent reduction of NAD+ at 340 nm [14]. One unit of activity is defined as the
amount of NADH produced (micromoles per minute per milligram of E1). The kinetic
parameters were calculated by curve fitting using the Sigma-plot program (Systat). The
reaction medium contained in 1 mL: 0.1 M Tris–HCl (pH 8.0), 1 mM MgCl2, 2 mM sodi-
um pyruvate, 2.5 mM NAD+, 0.2 mM ThDP, and 2.8 mM DTT at 30 °C [8]. The reaction
was initiated by adding 0.1 mM CoA. Activity was measured with 2, 5, 10, 20, 30, 40, and
100 mM of b-hydroxypyruvate at 25 °C.

A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
385
3. Results
3.1. Phenylacetylcarbinol formation with YPDC variants
Phenylacetylcarbinol is the major product formed when pyruvate and benzaldehyde are
added to a phosphate buffer (pH 6.0) containing the E477Q or D28A variants of YPDC,
along with the required cofactors ThDP and Mg2+. The CD spectrum of this reaction
develops a distinct band with a maximum at 283 nm and the band persists even after pro-
tein removal with Micro Centriprep and heat treatment (Fig. 2). The negative CD band
corresponds to the (R)-enantiomer. Phenylacetylcarbinol is a stable compound and can
be characterized by GC–MS showing the expected parent peak, fragmentation pattern,
1
and the expected H NMR spectrum.
After protein removal, phenylacetylcarbinol was extracted into chloroform and was
analyzed with the chiral column for estimation of the enantiomeric excess (ee), using a chi-
ral B-DM column from Chiraldex. An approximately 20 lL sample was injected with a
1:100 split ratio at a flow rate of 0.7 mL/min and at 100 °C. The retention time for the
(R) and (S) enantiomers under these GC conditions is 23.5 and 26.1 min, respectively.
To confirm that the column preferentially binds the (S) enantiomer, thus delaying its
retention time as suggested by the manufacturer’s specifications, authentic (R)-phenylace-
tylcarbinol was synthesized and analyzed under these GC conditions. Indeed, we con-
firmed that the first peak to emerge corresponds to the (R)-enantiomer, and the second
one to the (S)-enantiomer (Fig. 3).
The results of phenylacetylcarbinol production with the E477Q and D28A YPDC
indicated that both enzyme variants form only of (R)-PAC according to the CD spectrum
(figure not shown). The chiral GC chromatograms confirm formation solely of the
(R)-enantiomer and no detectable (S)-enantiomer (Fig. 3). The molar ellipticity for
(R)-PAC was calculated to be 30736 deg cm2 dmolꢀ1 at 283 nm.
Fig. 2. CD spectra of (R)-phenylacetylcarbinol formation with the E477Q YPDC variant. Conditions are
described under Section 2.

386
A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
Fig. 3. Chiral GC analysis of PAC produced by the E477Q and D28A YPDC variants vs. authentic (R)-PAC as
reference.
Unlike the E477Q variant, the D28A variant could also catalyze the condensation of
two pyruvate molecules yielding acetolactate (Fig. 4), but formation of acetolactate is
pyruvate concentration dependent and with 4 mM pyruvate the rate of acetolactate forma-
tion is very slow. The reaction to form (R)-PAC was examined with equimolar concentra-
tions of pyruvate and benzaldehyde (4 mM). Under these conditions no acetolactate
1
formation was observed, as also confirmed with H NMR (data not shown).
Fig. 4. Time development of the CD spectra of acetolactate formed with the D28A YPDC variant. Time delay
between each scan was approximately 4 min. Conditions are described under Section 2.

A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
387
3.2. Acetoin formation with YPDC and PDHc-E1 variants
Acetoin (3-hydroxy-2-butanone) is the major product formed when pyruvate and acet-
aldehyde are used as substrates with the YPDC and PDHc-E1 variants. CD spectra of the
reaction mixtures with these variants show a major CD band with a maximum at 278 nm.
The negative and the positive band correspond to the (R)- and (S)-enantiomers, respective-
1
ly. The structure of the product formed was confirmed with GC–MS and H NMR.
The yield and the stereochemistry of acetoin formed with each variant are different. The
YPDC variants and the PDHc-E1 variants catalyze the formation of opposite
enantiomers. It can be clearly seen in the CD spectrum that the YPDC variants preferen-
tially catalyze formation of (R)-acetoin and the PDHc-E1 variants catalyze formation of
(S)-acetoin (Fig. 5).
Assignment of the absolute configurations was again achieved with chiral gas chroma-
tography (Fig. 6). The product formed was extracted into chloroform and injected into the
GC at 0.7 mL/min and at 50°. The E477Q variant produced 98% yield of acetoin (4.1 mg)
with 94% ee, while the D28A variant produced 68% yield (2.9 mg) with 88% ee (Table 1).
Apparently, both YPDC variants preferentially catalyze formation of the (R) enantio-
mer. Both variants can achieve high enantiomeric excess showing that the facial selectivity
is very good in both cases. The reason for the lower yield of acetoin produced with the
D28A variant is that with 20 mM pyruvate and 100 mM acetaldehyde there is acetolactate
being produced as well. The molar ratio of acetolactate to acetoin formed was 30:70
1
according to H NMR.
The PDHc-E1 variants will produce acetolactate instead of acetoin in the absence of
exogenouos acetaldehyde, but with saturating concentration of acetaldehyde they will pro-
duce acetoin as the major product and only negligible amounts of acetolactate. Unlike the
YPDC variants, the PDHc-E1 variants catalyze the formation of (S)-acetoin. Percent yield
for acetoin production was calculated based on the starting pyruvate concentration and
Fig. 5. CD spectra of acetoin produced by the E636A and E636Q PDHc-E1 variants and the E477Q and D28A
YPDC variants. (a) PDHc-E1 E636Q, (b) PDHc-E1 E636A, (c) YPDC D28A, and (d) YPDC E477Q.
(kmax = 278 nm). Conditions are described under Section 2.

388
A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
Fig. 6. Chiral GC analysis of the acetoin produced by E477Q YPDC and E636Q PDHc-E1 vs. racemic acetoin
reference.
Table 1
Enantiomeric excess of acetoin, phenylacetylcarbinol, and acetolactate formed by the various enzymes used and
percent yields for the products formed under the specified conditions
Enzyme variant
Enantiomeric excess and yield of producta
PAC
Acetoin
Acetolactate
YPDC-E477Q
YPDC-D28A
PDHc-E1 E636A
PDHc-E1 E636Q
>99% (R)b(62)
>99% (R)b(60)
Not produced
Not produced
94% (R)b(98)
88% (R)b(68)
24% (S)b(92)
70% (S)b(93)
Not produced
40% (S)c(92)
60% (R)c(93)
>99% (R)c(95)
a
Yield in parenthesis in %.
ee was calculated by Chiral-GC.
ee was calculated from CD data.
b
c
the amount of product formed. The percent yield of acetoin was 92% (3.8 mg) with the
E636A, and 93% (3.9 mg) with the E636Q variant.
Interestingly, the enantiomeric excess achieved with PDHc-E1 variants varied signifi-
cantly with the substituent: the E636Q variant formed (S)-acetoin with 70% ee but the
E636A variant gave (S)-acetoin with only 24% ee (Table 1). This shows the significant
effect of the amino acid side chain on the stereochemical outcome of the reaction. A clear
comparison of the enantiomers formed according to the chiral-GC analysis of the YPDC
and PDHc-E1 variants vs. racemic acetoin mixture is presented in Fig. 5.
3.3. Acetolactate formation with YPDC and PDHc-E1 variants
Acetolactate is the major product formed when pyruvate is used as the sole substrate
with the D28A YPDC variant and with the E636A or E636Q PDHc-E1 variants.

A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
389
Acetolactate is the result of the condensation of a second pyruvate molecule onto the
enamine. The D28A YPDC variant produces (S)-acetolactate while PDHc-E1 variants
produce (R)-acetolactate (Fig. 7). The negative and the positive CD bands correspond
to the (R)- and (S)-enantiomers, respectively.
The assignment of absolute configuration of the acetolactate enantiomers was based on
literature values. The enzyme acetohydroxyacid synthase catalyzes the formation
(S)-acetolactate corresponding to a positive CD band according to Chipman and
coworkers [12]. The E477Q YPDC variant doesn’t catalyze acetolactate formation. The
percent yield of acetolactate formed with the D28A YPDC and the E636A and E636Q
PDHc-E1 variants is 92% (5.8 mg), 93% (5.9 mg), and 95% (6.0 mg), respectively.
Acetolactate is a thermally unstable compound, which is converted to racemic acetoin
on decarboxylation irrespective of the enantiomer of acetolactate undergoing decarboxyl-
ation, hence it was not possible to analyze it under GC conditions. We therefore used the
reported molar ellipticity for acetolactate [12]. The samples were scanned with CD after
protein removal and based on the starting concentration of pyruvate and the molar ellip-
ticity, the following ee were calculated: D28A YPDC, 40%; E636A PDHc-E1, 60%; and
E636Q PDHc-E1, >99%.
3.4. b-Hydroxypyruvate as substrate for YPDC and PDHc-E1
First, it was demonstrated that b-hydroxypyruvate is indeed a substrate for the two
enzymes. With YPDC, normal decarboxylation results in glycolaldehyde formation. Data
for the b-hydroxypyruvate-dependent glycolaldehyde production by YPDC were fitted to
nH
nH
nH
the Hill equation with substrate inhibition [9]; m ¼ V max½Sꢁ =ðS þ ½Sꢁ ð1 þ ½Sꢁ=KiÞÞ,
0:5
which yielded, Vmax = 0.379 0.007 U/mg, nH= 1.71 0.09, S0.5 = 2.01 0.063 mM,
Ki app = 1.156 0.387 M; kcat = 1.46 sꢀ1, kcat/S0.5 = 0.726 sꢀ1 mMꢀ1. With PDHc-E1,
Fig. 7. CD spectra of acetolactate produced by the E636A and E636Q PDHc-E1 variants and the D28A YPDC
variants. (a) PDHc-E1 E636Q, (b) PDHc-E1 E636A, and (c) YPDC D28A. (kmax = 300 nm). Conditions are
described under Section 2.

390
A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
we had to reconstitute the entire complex (by adding E2–E3 sub-complex) to demonstrate
activity and the data for the b-hydroxypyruvate-dependent 1,3,4-trihydroxy-2-butanone
production by PDHc-E1 were fitted to the Michaelis–Menten equation;
nH
nH
m ¼ V max½Sꢁ =ðKnmH þ ½Sꢁ Þ which yielded, Vmax = 3.342 0.305 U/mg, Km = 7.768
1.706 mM, kcat = 11.1 sꢀ1, kcat/Km = 1.43 sꢀ1 mMꢀ1
.
3.5. 1,3,4-Trihydroxy-2-butanone formation with YPDC and PDHc-E1 variants
When b-hydroxypyruvate was used as substrate along with glycolaldehyde as co-sub-
strate, 1,3,4-trihydroxy-2-butanone was produced with both the YPDC and PDHc-E1
variants. Except with parental PDHc-E1, both YPDC variants and PDHc-E1 variants
need saturating concentrations of glycolaldehyde as the second substrate. Again, the
YPDC variants produce the R-enantiomer in excess and the PDHc-E1 variants produce
the S-enantiomer in excess (Fig. 8); we have not determined the ee of the reactions.
3.6. 2,4-Dihydroxymethyl-3-oxo-butanoic acid formation with PDHc-E1 and YPDC
variants
2,4-Dihydroxymethyl-3-oxo-butanoic is the sole product formed from the reaction of
the E636A, and E636Q PDHc-E1 variants with b-hydroxypyruvate as the sole substrate.
The CD spectrum shows two bands corresponding to the same species with a maximum at
301 nm (Fig. 9; apparently with b-hydroxypyruvate, there are two CD bands in this
carboligase product). The product formed is the (S)-enantiomer. The D28A YPDC vari-
ant produces the (R)-enantiomer (data not shown).
The kmax values of the CD bands for the compounds studied are: acetoin, 278 nm;
acetolactate, 300 nm; phenylacetylcarbinol, 283 nm; 1,3,4-trihydroxy-2-butanone,
279 nm; and 2,4-dihydroxymethyl-3-oxo-butanoic acid, 301 nm.
Fig. 8. CD spectra of 1,3,4-trihydroxy-2-butanone produced by parental PDHc-E1 and its E636A and E636Q
variants and the E477Q YPDC variant. (a) PDHc-E1 E636Q + b-hydroxypyruvate and glycolaldehyde, (b)
PDHc-E1 E636A + b-hydroxypyruvate and glycolaldehyde, (c) PDHc-E1 + b-hydroxypyruvate, and (d) YPDC
E477Q + b-hydroxypyruvate and glycolaldehyde. (kmax = 278 nm). Conditions are described under Section 2.

A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
391
Fig. 9. CD spectra of 2,4-dihydroxy-2-hydroxymethyl-3-oxo-butanoic acid produced by the PDHc-E1 E636A
and E636Q variants. (a) PDHc-E1 E636A + b-hydroxypyruvate and (b) PDHc-E1 E636Q + b-hydroxypyruvate.
(kmax = 301 nm). Conditions are described under Section 2.
4. Discussion
This report quantifies the yield and stereochemical outcome of the carboligase reactions
of two ThDP-dependent enzymes which share a common pathway through decarboxyl-
ation and convert pyruvate to different enantiomers of acetoin and acetolactate. The active
center variants of YPDC and PDHc-E1 were shown to produce high yields and high ee,
depending on the conditions used. Further, we showed that the YPDC active center vari-
ants produce essentially a single enantiomer of phenylacetylcarbinol in good yield. We
note that Crout and coworkers had published extensively on the use of wild-type YPDC
and pyruvate decarboxylase from Zymomonas mobilis, to produce a-ketols such as
(R)-PAC and derivatives as well as aliphatic acyloins [15–17]. We also report that
b-hydroxypyruvate a commonly used substrate analog can also undergo carboligase reac-
tions catalyzed by YPDC variants and PDHc-E1 from E. coli leading to the substituted
derivatives of acetoin and acetolactate, again providing excess of the same enantiomer
as produced with pyruvate by the same enzyme. This is important to show since
b-hydroxypyruvate is often used with the enzyme transketolase as it produces the 1,2-
dihydroxyethylidine ThDP or enamine on decarboxylation that is sufficiently stable for
high-resolution X-ray studies [18]. The kmax values of the CD bands for the acetoin and
acetolactate derivatives produced from b-hydroxypyruvate are very similar to those from
acetoin and acetolactate itself. This raises a significant complication in the interpretation
of UV–vis and CD data on ThDP enzymes. As we had reported, the enamine-derived from
aliphatic 2-oxo acids has a kmax near 295 nm [5,6], the 10,40-imino tautomer of ThDP when
enzyme-bound has a kmax between 300 and 310 nm with a positive CD band [19–22], and,
as is here shown, acetolactate (or derivative derived from b-hydroxypyruvate), also gives
rise to a positive or negative CD band with kmax of 300–302 nm (sign depending on the
particular enantiomer) and a corresponding UV signal.
A comparison of the stereochemical outcome of the carboligase reaction is instructive.
To explain why the two enzymes prefer to form different enantiomers of the same

392
A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
carboligase product, we need to consider the facial selectivity of the approach of the carbon-
yl compound to the enamine intermediate (Fig. 10). Apparently, the two enzymes preferen-
tially recognize the opposite faces (re or si) of the incoming carbonyl compound (aldehyde
or 2-oxo acid). One can then ask whether electrostatic or size effects control the recognition.
Superimposition of the carbonyl groups of the two products, suggests that the relative sizes
of the two other (non-carbonyl) groups, rather than the charge at the carboxylate, influence
the preferred approach leading to the predominant enantiomer. This conclusion is reached
since PDHc-E1 preferentially synthesized (S)-acetoin and (R)-acetolactate, while YPDC
formed (R)-acetoin and (S)-acetolactate. That YPDC and pyruvate decarboxylase from
Zymomonas mobilis produce opposite enantiomers of acetoin had been noted earlier, and
Lobell and Crout carried out molecular modeling using the 3-dimensional X-ray structure
of YPDC to provide an explanation for its selectivity in acetoin formation [23].
Our results suggest that engineering of the bacterial pyruvate dehydrogenase complex
E1 subunit could be exploited to produce the opposite enantiomer of many compounds
than the one readily produced by pyruvate decarboxylases from several sources and by
acetohydroxyacid synthetases.
Fig. 10. Proposed mechanism for the formation of the R- and S-enantiomers by the two enzymes.

A. Baykal et al. / Bioorganic Chemistry 34 (2006) 380–393
393
Acknowledgments
The authors are grateful for financial support from: NIH-GM-050380 and NIH-GM-
062330, as well as the NIH-MBRS GM-060826 at Rutgers-Newark (Ann Cali, PI).
References
[1] P.W. Ward, A. Singh, Curr. Opin. Biotechnol. 11 (2000) 520–526.
[2] M. Pohl, G.A. Sprenger, M. Muller, Curr. Opin. Biotechnol. 15 (2004) 335–342.
¨
[3] M. Mu¨ller, G.A. Sprenger, Thiamine: catalytic mechanisms in normal and disease states, in: F. Jordan, M.S.
Patel (Eds.), Marcel Dekker, Inc. New York, (2004) 77–112.
[4] B. Lingen, M. Pohl, A.S. Demir, A. Liese, M. Muller, Thiamine: catalytic mechanisms in normal and disease
¨
states, in: F. Jordan, M.S. Patel (Eds.), Marcel Dekker, Inc. New York, (2004) 113–130.
[5] F. Jordan, Nat. Prod. Rep. 20 (2003) 184–201.
[6] F. Jordan, N. Nemeria, Bioorg. Chem. 33 (2005) 190–215.
[7] E.A. Sergienko, F. Jordan, Biochemistry 40 (2001) 7369–7381.
[8] N. Nemeria, K. Tittmann, E. Joseph, L. Zhou, M.B. Vazquez-Coll, P. Arjunan, G. Hubner, W. Furey, F.
¨
Jordan, J. Biol. Chem. 280 (2005) 21,473–21,482.
[9] E.A. Sergienko, F. Jordan, Biochemistry 40 (2001) 7382–7403.
[10] J. Brussee, E.C. Ross, A. Van Der Gen, Tetrahedron Lett. 29 (1988) 4485–4488.
[11] M. Liu, E.A. Sergienko, F. Guo, J. Wang, K. Tittmann, G. Hu¨bner, W. Furey, F. Jordan, Biochemistry 40
(2001) 7355–7368.
[12] M. Vinogradov, A. Kaplun, M. Vyazmensky, S. Engel, R. Golbik, K. Tittmann, K. Uhlemann, L.
Meshalkina, Z. Barak, G. Hubner, D.M. Chipman, Anal. Biochem. 342 (2005) 126–133.
¨
[13] H. Holzer, G. Schultz, C. Villar-Palasi, Jutgen-Sell, J. Biochem. Z. 327 (1956) 331–344.
[14] N. Nemeria, A. Volkov, A. Brown, J. Yi, L. Zipper, J.R. Guest, F. Jordan, Biochemistry 37 (1998) 911–922.
[15] V. Kren, D.H.G. Crout, H. Dalton, D.W. Hutchinson, W. Konig, M.M. Turner, G. Dean, N. Thomson, J.
Chem. Soc. Chem. Commun. (1993) 341–343.
[16] S. Bornemann, D.H.G. Crout, H. Dalton, V. Kren, M. Lobell, G. Dean, N. Thomson, M.M. Turner, J.
Chem. Soc. Perkin Trans. 1 (1996) 425–430.
[17] S. Bornemann, D.H.G. Crout, H. Dalton, D.W. Hutchinson, G. Dean, N. Thomson, M.M. Turner, J. Chem.
Soc. Perkin Trans. 1 (1993) 309–311.
[18] E. Fiedler, S. Thorell, T. Sandalova, R. Golbik, S. Ko¨nig, G. Schneider, Proc. Natl. Acad. Sci. USA 99
(2002) 591–595.
[19] F. Jordan, Z. Zhang, E. Sergienko, Bioorg. Chem. 30 (2002) 188–198.
[20] F. Jordan, N.S. Nemeria, S. Zhang, Y. Yan, P. Arjunan, W. Furey, J. Am. Chem. Soc. 127 (2003)
12,732–12,738.
[21] N. Nemeria, A. Baykal, E. Joseph, S. Zhang, Y. Yan, W. Furey, F. Jordan, Biochemistry 43 (2004)
6565–6575.
[22] A. Baykal, L. Kakalis, F. Jordan, Biochemistry 45 (2006) 7522–7528.
[23] M. Lobell, D.H.G. Crout, J. Am. Chem. Soc. 118 (1996) 1867–1873.