Branched-chain keto acid decarboxylase from Lactococcus lactis (KdcA), a valuable thiamine diphosphate-dependent enzyme for asymmetric C - C bond formation

Article abstract of DOI:10.1002/adsc.200700057

The thiamine diphosphate-dependent, branched-chain 2-keto acid decarboxylase from Lactococcus lactis sup. cremoris Bl157 (KdcA) is a new valuable enzyme for the synthesis of chiral 2-hydroxy ketones. The gene was cloned and the enzyme was expressed as an

Full text of DOI:10.1002/adsc.200700057

FULL PAPERS
DOI: 10.1002/adsc.200700057
Branched-Chain Keto Acid Decarboxylase from Lactococcus
lactis (KdcA), a Valuable Thiamine Diphosphate-Dependent
À
Enzyme for Asymmetric C C Bond Formation
Dçrte Gocke,a Cong Luan Nguyen,a Martina Pohl,a,c,* Thomas Stillger,b
Lydia Walter,b and Michael Müllerb
a
Institute of Molecular Enzyme Technology, Heinrich-Heine University of Düsseldorf, Research Center Jülich, 52426
Jülich, Germany
Fax : (+49)-2461-61-2490; e-mail: ma.pohl@fz-juelich.de
Institute of Pharmaceutical Sciences, Albert-Ludwigs-University of Freiburg, Albertstr. 25, 79104 Freiburg, Germany
b
Fax : (+49)-761-203-6351
Both groups contributed equally to the publication
c
Received: January 24, 2007; Revised: April 3, 2007
Supporting information for this article is available on the WWW under http://asc.wiley-vch.de/home/.
Abstract: The thiamine diphosphate-dependent, looses activity at higher temperatures. In this work
branched-chain 2-keto acid decarboxylase from Lac- the carboligase activity of KdcA is demonstrated for
tococcus lactis sup. cremoris B1157 (KdcA) is a new the first time. The enzyme shows an exceptionally
valuable enzyme for the synthesis of chiral 2-hydroxy broad substrate range and, most strikingly, it catalyz-
ketones. The gene was cloned and the enzyme was es the carboligation of different aromatic aldehydes
expressed as an N-terminal hexahistidine fusion pro- as well as CH-acidic aldehydes such as phenylacetal-
tein in Escherichia coli. It has a broad substrate dehyde and indole-3-acetaldehyde with aliphatic al-
range for the decarboxylation reaction including dehydes such as acetaldehyde, propanal, and cyclo-
linear and branched-chain aliphatic and aromatic propanecarbaldehyde, yielding chiral 2-hydroxy ke-
keto acids as well as phenyl pyruvate and indole-3- tones in high enantiomeric excess. Noteworthy, the
pyruvate. The dimeric structure of recombinant donor-acceptor selectivity is strongly influenced by
KdcA is in contrast to the tetrameric structure of the nature of the respective substrate combination.
other 2-keto acid decarboxylases. The enzyme is
stable between pH 5 and 7 with a pH optimum of Keywords: benzaldehyde lyase; benzoylformate de-
pH 6–7 for the decarboxylation reaction. While carboxylase; bioorganic chemistry; biotransforma-
À
KdcA is sufficiently stable up to 408C it rapidly tions; C C coupling; pyruvate decarboxylase
Introduction
in the process of cheese ripening due to their decar-
boxylation activity of 2-keto acids which are formed
Thiamine diphosphate-dependent enzymes such as through transamination of the corresponding
pyruvate decarboxylase (PDC, E.C. 4.1.1.1), benzoyl- branched-chain amino acids.[9] Recently the substrate
formate decarboxylase (BFD, E.C. 4.1.1.7) and ben- binding site of KdcA has been probed by site-directed
zaldehyde lyase (BAL, E.C. 4.1.2.38) have been inten- mutagenesis studies based on a homology model
sively studied with respect to their carboligation activ- using the structure of pyruvate decarboxylase from
ity, offering an easy access to chiral 2-hydroxy ketones Zymomonas mobilis as a template.[12]
from aldehydes.[1–8] In order to enlarge the range of
In this study we investigated the carboligase prop-
accessible 2-hydroxy ketones, we studied the carboli- erties of KdcA for the first time. Compared to other
gase properties of a recently described branched- decarboxylases, KdcA has a broader substrate range
chain keto acid decarboxylase (E.C. 4.1.1.72). Two and accepts besides acetaldehyde several aliphatic al-
highly homologous enzymes have been found in dif- dehydes such as propanal, butanal, isobutyraldehyde
ferent Lactococcus lactis strains: Lactococcus lactis and cyclopropanecarbaldehyde as donor and/or ac-
sup. cremoris B1157[9,10] (KdcA) and Lactococcus ceptor in the enzyme-catalyzed acyloin condensation.
lactis IFPL730[11] (Kivd). These enzymes are involved Moreover, KdcA catalyzes the carboligation of eno-
Adv. Synth. Catal. 2007, 349, 1425 – 1435
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1425

Dçrte Gocke et al.
FULL PAPERS
lizable CH-acidic aldehydes such as indole-3-acetalde- in contrast to other tetrameric 2-keto acid decarboxy-
hyde and phenylacetaldehyde. The substrate range of lases, such as PDC[13,14] and BFD.[15] However, other
the decarboxylase and the carboligase reaction, the ThDP-dependent enzymes are known to be active as
pH and temperature-dependent stability and activity,
a
dimer, such as acetohydroxy acid synthase
as well as the stereoselectivity of KdcA are reported. (AHAS)[16] and transketolase (TK).[17] Meanwhile the
three-dimensional structure of KdcA has been deter-
mined proving the dimeric structure of the enzyme
Results and Discussion
(unpublished results).
Cloning, Overexpression and Purification
Decarboxylase Activity
The coding gene (kdca, gene bank CAG34226) was
cloned and overexpressed in E. coli as an N-terminal The investigation of the substrate range of decarboxy-
hexahistidine fusion protein. Addition of the His-tag lation is of interest to deduce information about the
led to an N-terminal elongation of the protein by 23 acyl donor spectrum for carboligase activity of KdcA.
amino acids: MGSSHHHHHHSSGLVPRGSHMAS.
If a respective 2-keto acid is a substrate for the
A 15-L fed-batch fermentation of the recombinant enzyme, the binding of the corresponding aldehyde to
E. coli strain resulted in 720 kU (decarboxylase activi- the C-2 atom of ThDP located in the active center is
ty) KdcA in 1.2 kg cells. The enzyme was purified by most likely. In the case of carboligation the ThDP-
immobilized metal chelate chromatography yielding bound aldehyde is ligated to a second acceptor alde-
1.7 g KdcA from 1.2 kg cells.
hyde molecule. Therefore, the aldehydes which are
products of the decarboxylation reaction can, in prin-
ciple, be used at least as acyl donors for the carboliga-
tion reaction.
Determination ofthe Native Molecular Weight
KdcA has an exceptionally broad substrate range
The native molecular weight of the recombinant for the decarboxylation reaction as determined by
KdcA was determined by size-exclusion chromatogra- Smit et al.[10] and Yep et al.[12] (Table 1). In contrast to
phy to afford a molecular weight of about 146 kDa other decarboxylases like PDC or BFD, KdcA accepts
(for log Mr/Kav plot see Supporting Information). As different aliphatic branched-chain as well as enoliza-
the calculated size of the monomeric subunit is ble 2-keto acids such as phenyl pyruvate and indole 3-
63.337 kDa, the observed data correlate best with a pyruvate. Although the natural branched-chain ali-
dimeric structure of KdcA in the native state. This is phatic substrate 1 as well as its analogue 2 (Table 1)
Table 1. Substrate range and kinetic data of the decarboxylation reaction.[a] Data are compared with the literature: Smit
et al. used crude extracts of KdcA without any tag, Yep et al. used a purified variant with a C-terminal hexahistidine tag and
own data were obtained with a purified variant containing an N-terminal hexahistidine tag.
2-Keto acid
Relative activity [%]:
Kinetic data:
own data compared to [Smit et al.][10]
(Yep et al.)[12]
own data (Yep et al.)[12]
Vmax [U/mg]
KM [mM]
1
2
3
4
5
100 [100] (100)
181.64Æ0.21 (47.3)
5.02Æ0.21 (2.8)
26.3 [31] (100)
34.3Æ0.2 (48.2)
0.264Æ0.011 (3.7)
38.1 [27.8]
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
19
0
1426
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1425 – 1435

Branched-Chain Keto Acid Decarboxylase from Lactococcus lactis
Table 1. (Continued)
FULL PAPERS
2-Keto acid
Relative activity [%]:
Kinetic data:
own data compared to [Smit et al.][10]
(Yep et al.)[12]
own data (Yep et al.)[12]
Vmax [U/mg]
KM [mM]
6
(18)
n.d. (8.6)
3.67Æ0.31
n.d.
n.d. (1.3)
7
1.2 [1.3]
9.3 [7.3]
15 [19] (20.6)
13 [25.3] (26.5)
0
29.77Æ6.44
n.d.
8
9
n.d. (9.8)
n.d. (12.8)
n.d.
n.d. (1.3)
n.d. (0.6)
n.d.
10
11
12
13
14
15
16
8.4 (15.2)
n.d. (7.2)
15.69Æ0.21 (26.6)
n.d. (2.9)
n.d.
n.d. (7.5)
0.127Æ0.007 (0.21)
n.d. (0.63)
n.d.
8.6 [7] (56.3)
(6)
1.6
1.1
n.d.
n.d.
17
0.85 [3.5]
1.55Æ0.6
0.234Æ0.024
[a]
Conditions: Activity towards different 2-ketoacids was measured using the coupled continuous assay. All substrates were
applied in a concentration of 30 mM, except 17, which was applied with 1 mM. Relative activities for 2, 13, and 17 have
been calculated based on the maximal velocities.
show the highest reaction velocities, the KM values observed up to 50 mM. With phenyl pyruvate the
are significantly higher than those of linear aliphatic v/[S] plot is hyperbolic up to 12 mM (Figure 1a) fol-
2-keto acids, phenyl pyruvate, and indole-3-pyruvate.
lowed by a progressive decay of activity up to 30 mM
Kinetic constants were determined for the physio- and a subsequent rapid complete loss of activity (not
logical substrate 3-methyl-2-oxobutanoic acid (1), 4- shown). In accordance with the observation of Yep
methyl-2-oxopentanoic acid (2) as well as phenyl pyr- et al.[12] KdcA exhibits an extremely low KM value for
uvate (13) and indole-3-pyruvate (17) (Table 1). With phenyl pyruvate (0.13 mM), which is about 40-times
substrates 1 and 2 hyperbolic v/[S] plots have been lower compared to the natural substrate 1. However,
Adv. Synth. Catal. 2007, 349, 1425 – 1435
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1427

Dçrte Gocke et al.
FULL PAPERS
Optimal Cofactor Concentration
As in all ThDP-dependent enzymes the cofactors in
KdcA are bound non-covalently to the active site.
The main contribution to the binding arise from the
coordinative interaction of the diphosphate moiety
via Mg2+ to the protein as well as from hydrophobic
and ionic interactions between protein site chains and
the thiazole and pyrimidine rings of ThDP.
For stability of the holoenzyme most ThDP-depen-
dent enzymes require the addition of cofactors to the
buffer. In case of KdcA the addition of 2.5 mM
MgSO4 and 0.1 mM ThDP to the buffer is sufficient
to keep the enzyme stable and active.
pH-Dependent Activity and Stability
KdcA shows a pH optimum of pH 6–7 for the decar-
boxylation of 1 (Figure 2). This activity optimum
overlaps very well with the stability optimum of
KdcA in potassium phosphate buffer, where no loss
of activity can be detected at pH 5–7 within 60 h.
Rapid inactivation occurs at pH 4 (<2 h) and slower
inactivation at pH 8 (half-life: 40 h).
Figure 1. v/[S]-plot of the KdcA-catalyzed decarboxylation
of (top) phenyl pyruvate (13) and (bottom) indole-3-pyru-
vate (17). Buffer: 50 mM potassium phosphate, 2.5 mM
MgSO4, 0.1 mM ThDP, pH 6.8, 308C. Data have been ob-
tained in triplicate using the coupled decarboxylase assay.
Figure 2. Determination of the pH optimum of the KdcA-
catalyzed decarboxylation of 3-methyl-2-oxo-butanoic acid
(1). Data were obtained using the coupled decarboxylase
assay.
the maximal velocity in the presence of phenyl pyru-
vate is only about 9% of the velocity obtained with 1.
Our data correlate well with those obtained by Smit
et al.[10] who investigated KdcA without any hexahisti-
dine tag (Table 1). However, we could not reproduce
the high relative activity of 56.3% towards phenyl Temperature Optimum for Activity and Stability
pyruvate compared to the natural substrate 1 which
was found by Yep et al. with a KdcA-variant carrying Under initial rate conditions (90 sec), the temperature
a C-terminal hexahistidine tag (Table 1).
optimum was observed at 508C. Further increases in
In the case of indole-3-pyruvate kinetic investiga- temperature resulted in a fast decay of activity with a
tion has been hampered by the low solubility (about midpoint of thermal inactivation (Tm) at about 628C
1.3 mM) and the strong absorbance of the substrate in (Figure 3).
aqueous buffer. However, as the KM value is also very
From these data, the activation energy of the decar-
low (0.23 mM), saturation is nearly achieved at 1 mM boxylase reaction was calculated from a lnVmax/[1/T]
G
(Figure 1 b). Compared to 3-methyl-2-oxobutanoic plot in the range of 25–408C as 8.5 kJmolÀ1, which is
acid (1) the KdcA shows 0.85% relative activity with relatively low compared to other 2-keto acid decar-
indole-3-pyruvate.
boxylases. For benzoyl formate decarboxylase from
1428
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1425 – 1435

Branched-Chain Keto Acid Decarboxylase from Lactococcus lactis
FULL PAPERS
Figure 4. Stability of KdcA in 50 mM potassium phosphate
*
buffer, pH 6.8, 2.5 mM MgSO4, 0.1 mM ThDP ( ), in the
Figure 3. Determination of the temperature optimum and
the midpoint of thermal inactivation of the KdcA-catalyzed
decarboxylation of 3-methyl-2-oxobutanoic (1). Data were
obtained using the direct decarboxylase assay.
^
same buffer plus 20% (v/v) DMSO ( ) and in the same
buffer plus 15% (v/v) PEG-400 ( ). Activities were related
to the starting activity in buffer without additives. The up to
20% higher activity in the case of ( ) and ( ) after short
time incubation is due to normal fluctuations in decarboxy-
lase activity observed after solution of lyophilized enzyme.
&
*
^
Pseudomonas putida and for pyruvate decarboxylase
from Zymomonas mobilis activation energies of
38 kJmolÀ1[18] and 43 kJmolÀ1, respectively, (unpub- to give (S)-5-hydroxy-2,7-dimethyloctan-4-one (27)
lished) have been determined.
with an enantiomeric excess (ee) of 30–47%. BAL
In order to optimize the conditions for the applica- catalyzes this reaction with a specific activity of
tion of KdcA in enzymatic syntheses, the temperature 0.05 U/mg resulting in the formation of (R)-27 (ee
stability of the enzyme in 50 mM potassium phos- 80%).[21,22] Thus, although the activity of KdcA is dis-
phate buffer, pH 6.8 in the presence of 2.5 mM tinctly lower for this transformation it opens the way
MgSO4 and 0.1 mM ThDP was determined. While the to both enantiomers of aliphatic acyloins.
enzyme is sufficiently stable up to 408C (half-life:
Surprisingly, KdcA is not able to catalyze the self-
80 h) it rapidly looses activity at higher temperatures carboligation of isobutyraldehyde (19) in detectable
(508C: half-life 9 h, 558C: half-life 4 h).
amounts, although 19 is the reaction product of the
physiological decarboxylation of 3-methyl-2-oxobuta-
noic acid (1) and should therefore fit into the active
center. We assumed an inhibition or inactivation of
KdcA caused by isobutyraldehyde (19). In order to
Organic Solvents
The biotransformation of aromatic aldehydes is often test this hypothesis the enzyme was incubated with in-
hampered by their low solubility in aqueous systems. creasing concentrations of 19 (5, 20, 40 mM) for 26 h
Since the addition of either 20% (v/v) DMSO or and the decrease of activity of KdcA was followed by
15% (v/v) PEG-400 has been successfully applied for measuring the residual decarboxylase activity using
BAL- and BFD-catalyzed carboligase reactions,[1,3,19,20] the direct decarboxylase assay. The data presented in
these organic solvents were tested with KdcA. The Figure 5 show a concentration-dependent inactivation
enzyme is completely stable in the presence of 20% of KdcA, which explains the negative results of the
(v/v) DMSO (half-life: 150 h), whereas it is rapidly in- carboligation studies with this aldehyde.
activated in the presence of 15% (v/v) PEG-400
(half-life: 6 h) (Figure 4).
To investigate the reversibility of the inactivation,
isobutyraldehyde (19) was removed from the 40 mM
sample after complete inactivation of KdcA using ul-
trafiltration. The restoration of activity was followed
by measuring the decarboxylase activity of the
enzyme. After 2 h 7% and after 16 h 30% of the orig-
inal activity were recovered (data not shown), demon-
strating that the inactivation of the enzyme is at least
Carboligase Activity
Carboligation ofAliphatic Aldehydes
Since KdcA is involved in the production of partially reversible. Furthermore, kinetic studies using
branched-chain aliphatic aldehydes, isovaleraldehyde the direct decarboxylase assay in the presence of
(18) and isobutyraldehyde (19) were tested at first for 40 mM isobutyraldehyde resulted neither in a decay
carboligation (Table 2, entries 1 and 2). As summar- of Vmax nor did isobutyraldehyde affect the KM value
ized in Table 2, the enzyme catalyzes the self-carboli- for 3-methyl-2-oxobutanoic acid (1) (data not shown).
gation of 18 with a low specific activity of 0.004 U/mg We therefore conclude that the observed inactivation
Adv. Synth. Catal. 2007, 349, 1425 – 1435
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1429

Dçrte Gocke et al.
FULL PAPERS
Table 2. Results of KdcA-catalyzed carboligation reactions.
No. Substrate A
Substrate B
Product(s)
Enantiomeric exess
1
2
3
18
19
20
-
27
30-47% (S)
-
-
-
-
28
29
46% (R)
4
5
21
22
-
-
>98% (R)
-
-
-
6
23
-
-
30a and
30b
30a: (R) 93%; 30b: (R) 92%;
30a/30b: 60:40
7
21
21
21
21
21
20
8
24
25
18
26
31b
32b
33b
34b
>98% (R)[a]
96.5% (R)
88% (R)
9
10
11
98% (R)[a]
12
13
22
20
35a
96.5% (R)
36a and
36b
13
17
20
20
ee n.d.; 36a/36b: 80:20
ee n.d.
14
37a
[a]
Absolute configuration was determined by comparison of HPLC data and with regard to mechanistic aspects of KdcA
catalysis; n.d.=not determined.
1430
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1425 – 1435

Branched-Chain Keto Acid Decarboxylase from Lactococcus lactis
FULL PAPERS
demonstrated in Table 2 (entry 7) the mixed carboli-
gation of acetaldehyde and benzaldehyde resulted in
the formation of nearly equal amounts of (R)-2-hy-
droxypropiophenone (ee 93%) (2-HPP, 30a) and (R)-
phenylacetyl carbinol (ee 92%) (PAC, 30b) besides
traces of (R)-benzoin (29) as was deduced from
GC/MS, HPLC and NMR. The combination of ben-
zaldehyde with larger aliphatic aldehydes gave re-
markable results: the selectivity of KdcA shifted com-
pletely to the PAC derivatives 31b–34b, if propanal
(24), butanal (25), isovaleraldehyde (18) or cyclopro-
panecarbaldehyde (26) were applied (Table 2,
entry 8–11). This is in contrast to similar BAL-cata-
lyzed reactions, which resulted, for example, in mix-
tures of the HPP (32a) and PAC (32b) derivatives
using benzaldehyde and propanal as substrates (un-
published results).
&
Figure 5. Inactivation of KdcA by isobutyraldehyde (19) ( :
~
*
^
0 mM, : 5 mM, : 20 mM : 40 mM). After the indicated
time intervals 50 mL samples were withdrawn and residual
activity was measured by the direct decarboxylase assay.
Subsequently, 3,5-dichloro- (22) and 3,5-dimethoxy-
benzaldehyde (23) were tested in KdcA-catalyzed car-
boligations together with acetaldehyde. Whereas no
reaction occurred with 23 the dichloro derivative 22 is
in the presence of isobutyraldehyde is a slow process clearly accepted as a donor aldehyde, yielding selec-
and not detectable under initial rate conditions. tively the 2-HPP derivative 35a with high enantiomer-
Besides the branched-chain aldehydes other alde- ic excess (ee 96.5%) (Table 2, entry 12).
hydes such as acetaldehyde, propanal and cyclopropa-
The 2-hydroxy ketones 29–35 are formed with high
necarbaldehyde were tested as substrates for the car- enantioselectivity (ee >88%). The absolute (R)-con-
boligase reaction. In all cases carboligase products figuration was assigned unambiguously to benzoin
were observed by means of GC/MS. The specific ac- (29), 2-hydroxypropiophenone (30a), phenylacetyl
tivity for acetoin formation (0.052 U/mg) is 13-fold carbinol (30b), 32b, 33b, and 35a by using circular di-
higher than the catalytic activity towards the forma- chroism or by comparison with authentic samples.
tion of the branched-chain 2-hydroxyketone 27 The absolute configuration of the hydroxy ketones
(Table 2, entry 3). The (R)-enantiomer of acetoin (28) 31b and 34b is assumed to be (R), too, according to a
is formed predominantly (ee 46%) as proven by com- comparison with products obtained from BAL and
parison with the product obtained from the PDC PDC catalysis.
(Saccharomyces cerevisiae)-catalyzed reaction.
Carboligation ofPhenylacetaldehyde and
Indole-3-acetaldehyde with Acetaldehyde
Carboligation ofAromatic Aldehydes
We could show that KdcA catalyzes the self-ligation CH-acidic aldehydes such as phenylacetaldehyde and
of benzaldehyde (21) to (R)-benzoin (29) with high indole-3-acetaldehyde are prone to enolization and
enantiomeric excess (>98%), whereas 3,5-dichloro- thus are difficult substrates for enzymatic[23] and also
benzaldehyde (22) or 3,5-dimethoxybenzaldehyde non-enzymatic transformations. Moreover, indole-3-
(23) were not ligated to symmetric benzoin deriva- acetaldehyde is very instable and decomposes rapidly,
tives (Table 2, entries 4–6).
which renders its direct application in biotransforma-
tions almost impossible.[24]
Phenylacetaldehyde is stable and commercially
available; however, in aqueous buffer solution an
aldol reaction occurs spontaneously. Application of
this aldehyde in KdcA-catalyzed carboligations with
Mixed Carboligation ofAromatic and Aliphatic
Aldehydes
As shown above KdcA accepts various aliphatic alde- acetaldehyde gives the 2-hydroxy ketones 36a and b
hydes as well as benzaldehyde as acyl donor and ac- besides significant amounts of the aldol product.
ceptor. However, it was unknown whether benzalde-
To overcome these problems, we made use of
hyde derivatives like 22 or 23 could function as selec- KdcAꢁs ability to decarboxylate the corresponding 2-
tive donors or acceptors for this enzyme. Therefore, keto acid of both aldehydes, which offers an easy way
we tested the catalytic activity of KdcA with various to generate these aldehydes in situ. Referring to the
mixtures of aromatic and aliphatic aldehydes. As reaction mechanism, the decarboxylation of a 2-keto
Adv. Synth. Catal. 2007, 349, 1425 – 1435
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1431

Dçrte Gocke et al.
FULL PAPERS
After digestion of the amplified gene with EcoRI the
gene was first ligated into the vector pBluescript in order to
allow restriction by NheI and EcoRI. Subsequently the kdca
gene (1.6 kb) was ligated into the vector pET28a (Novagen)
carrying already the information for an N-terminal His-tag
yielding the final construct pET28a-KdcA-His-N. Transfor-
acid results in a reaction intermediate with the corre-
sponding aldehyde bound to the cofactor ThDP. This
so-called “activated aldehyde” can react with a fur-
ther acceptor aldehyde. If the KdcA-catalyzed decar-
boxylation of phenyl pyruvate (13) and indole-3-pyru-
vate (17) is performed in the presence of acetalde-
hyde the 2-hydroxy ketones 36a, b and 37, respective-
ly, are formed predominantly in good yield (Table 2,
entries 13 and 14).
mation of the expression host E. coli BL21(DE3) was per-
N
formed by electroporation. Identity of the gene with the
published sequence[10] was confirmed by DNA sequencing
(Sequiserve, Germany).
Expression and Purification
Conclusions
A fed-batch cultivation according to the method of Korz
et al.[25] was performed in a 40-L Techfors reactor (Infors
AG, CH) at 308C, pH 7.0. Dissolved oxygen saturation was
regulated between 30–40% saturation by increasing the stir-
rer speed and the air flow rate and subsequently by a pres-
sure increase during the high oxygen demanding overexpres-
sion. After a preliminary growth phase protein expression
was induced by the addition of 1.5 mM IPTG at an OD600
~60. From 15 L culture 1.2 kg E. coli cells with a specific ac-
tivity of 21 U/mg in the crude cell extract were gained. Cells
were harvested by a separator (Westfalia Separator AG)
and stored at À208C.
KdcA is a valuable new biocatalyst for asymmetric
C C bond formation. The enzyme can be produced
À
easily on a large scale in E. coli and purified using im-
mobilized metal affinity chromatography. The
branched-chain keto acid decarboxylase is stable up
to 408C at pH 6.8, and also in the presence of water-
miscible organic solvents such as 20% (v/v) DMSO.
The enzyme is able to catalyze the carboligation of a
broad range of aldehydes to chiral 2-hydroxy ketones.
Most interestingly the selectivity of the products
formed in the mixed acyloin synthesis of benzaldehyde
and derivatives with different aliphatic acceptor alde-
hydes strongly depends on the substrate combination.
The product composition can be shifted from a selec-
tively formed PAC derivative of the aromatic aldehyde
in the presence of large aliphatic aldehydes, over a
nearly equal ratio of the HPP and PAC derivatives in
the presence of acetaldehyde to a selectively formed
HPP derivative in the presence of 3,5-dichlorobenzal-
dehyde (Table 2). A further important achievement is
the selective asymmetric synthesis of 2-hydroxy ke-
tones containing phenylacetaldehyde and indole-3-
acetaldehyde. Good results have been obtained by in
situ production of these enolizable (CH-acidic) alde-
hydes by enzymatic decarboxylation of the correspond-
ing 2-keto acid in the presence of acetaldehyde, with
both reactions being catalyzed by KdcA (Table 2).
Purification of KdcA to homogeneity (>95%) was per-
formed by immobilized nickel chelate affinity chromatogra-
phy followed by size exclusion chromatography using a pu-
rification protocol previously developed for benzoyl formate
decarboxylase[18] with the following alterations: The pH of
all buffers used was set to 6.8. Buffers contained 2.5 mM
MgSO4 and 0.1 mM ThDP. Proteins unspecifically bound
were eluted with 50 mM imidazole. KdcA was eluted with
250 mM imidazole. After purification the enzyme was either
freeze-dried or diluted with 50% (v/v) glycerol and stored
at À208C.
Decarboxylase Activity
One unit of decarboxylase activity is defined as the amount
of KdcA which catalyzes the decarboxylation of 1 mmol 3-
methyl-2-oxobutanoic acid (1) or another 2-keto acid per
minute under standard conditions (pH 6.8, 308C).
To investigate the decarboxylase activity of KdcA two
continuous decarboxylase assays were used. In the coupled
decarboxylase assay horse liver alcohol dehydrogenase (HL-
ADH) is used to reduce the aldehydes obtained from the
KdcA-catalyzed decarboxylation. Thereby NADH is oxi-
dized to NAD+. The decay of NADH is followed spectro-
photometrically for 90 sec at 340 nm.
Experimental Section
Cloning ofKdcA
The KdcA gene was amplified by PCR from the genomic
DNA of Lactococcus lactis subspecies cremoris B1157 which
was kindly provided by NIZO food research B. V. (Ede, NL).
The strain was grown in M17’ medium and isolation of ge-
nomic DNA was performed using the DNA-Tissue Kit
(Qiagen). The following primers have been used for cloning:
Assay composition: 700 mL buffer A (50 mM potassium
phosphate buffer, pH 6.8, 2.5 mM MgSO4, 0.1 mM ThDP),
100 mL 3-methyl-2-oxobutanoic acid (or another 2-keto
acid) 300 mM in buffer A (final concentration 30 mM;
except indole-3-pyruvate: 1 mM), 100 mL NADH 2.5 mM in
buffer A (final concentration 0.25 mM), 50 mL HL-ADH
1432
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1425 – 1435

Branched-Chain Keto Acid Decarboxylase from Lactococcus lactis
FULL PAPERS
(Sigma–Aldrich) 5 Units/mL in buffer A (final concentra-
tion 0.25 U/mL). The assay was started by addition of 50 mL
KdcA.
Further, a direct decarboxylase assay following the direct
decay of 3-methyl-2-oxobutanoic acid (1) was developed in
order to measure KdcA activity under NADH degradating
conditions.
Assay composition: 950 mL 3-methyl-2-oxobutanoic acid
(1) (60 mM in buffer A). The reaction was started by addi-
tion of 50 mL KdcA solution. To avoid a background by ab-
sorption of isobutyraldehyde, the decay of 3-methyl-2-oxo-
butanoic acid (lmax =320 nm) (1) was followed at 340 nm
(e=0.017 LmmolÀ1 cmÀ1).
250 mm; TGC
G
time program (oven): T0min =608C, T3min =608C, T14min
2808C (heating rate 208C·minÀ1), T19min =2808C). The enan-
tiomeric excess was determined by chiral GC, employing a
Shimadzu GC 2010, fitted with an FS Lipodex D column
(50 m”0.25 mm) and an FID detector, or by chiral HPLC
employing an HP 1100 HPLC system (Agilent) fitted with a
diode-array detector. NMR spectra were recorded on a
Bruker DPX-400. Chemical shifts are reported in ppm rela-
tive to CHCl3 (1H NMR: d=7.27) and CDCl3 (13C NMR: d
=77.0) as internal standards. CD spectra were recorded on
a JASCO J-810 spectropolarimeter using acetonitrile as sol-
vent.
Kinetic constants were calculated by non-linear regression
using Origin 7G SR4 (OriginLab Coop., Northampton,
USA). Background activities, measured in the absence of Carboligase Activity, Representative Examples for
KdcA, were subtracted.
the KdcA-Catalyzed Synthesis of2-Hydroxy Ketones:
(R)-1-Hydroxy-1-phenylpentan-2-one (32b): Benzaldehyde
(29 mg, 0.27 mmol) was dissolved in a mixture of dimethyl
sulfoxide (3 mL) and potassium phosphate buffer [12 mL,
50 mM, pH 6.8, containing MgSO4 (2.5 mM) and ThDP
(0.1 mM)]. To this solution 19 mg (0.27 mmol) butanal were
added. After addition of KdcA (3.3 mg protein) the reaction
was stirred slowly at 308C for 104 h. The reaction mixture
was extracted with diethyl ether (25 mL) and the organic
layer washed with brine and dried over Na2SO4. Evapora-
tion of the solvent and purification of the crude product,
which contained besides 32b small amounts of substrates
and 5-hydroxyoctan-4-one, by flash column chromatography
afforded (R)-1-hydroxy-1-phenylpentan-2-one (32b) as low-
viscous oil; yield: 15.6 mg (32%, 96.5% ee). HPLC: (Chiral
OM, n-hexane/2-propanol, 98:2, 0.5 mLminÀ1, 408C) Rt
(S)=30.1 min, Rt (R)=36.5 min; [a]2D3: À130.4 (c 0.1 g/
100 mL, CDCl3); CD (acetonitrile): l (De) [nm] (mol.
CD)=199 (+18.02), 211 (+9.5), 218 (+9.7), 231 (+0.9), 238
Protein Determination
Protein determination was performed according to Brad-
ford[26] using BSA as a standard.
Determination ofMolecular Mass
Size-exclusion chromatography was performed using a Su-
perdex G200 prep grade column [total volume 122 mL
(Ø 1.6 cm)] (Amersham) and 50 mM potassium phosphate
buffer, pH 6.5, including 2.5 mM MgSO4, 0.1 mM ThDP and
150 mM KCl. The coefficient of available volume [Kav=
(VeÀVo)/(VtÀVo), Ve: elution volume of the respective
G
protein, Vt: total volume, elution volume of blue dextran)
for KdcA and the standard proteins have been determined
twice with a standard deviation of 0.2%.
Calibration was performed using ribonuclease A (13.7 kD,
Kav=0.66), chymotrypsinogen A (25 kD, Kav=0.6), oval-
bumin (43 kD; Kav=0.48), BSA (67 kD, Kav=0.4), aldo-
lase (158 kD, Kav=0.29), catalase (232 kD, Kav=0.27), fer-
ritin (440 kD, Kav=0.15), and thyroglobulin (669 kD, Kav=
0.07). The Kav coefficient of KdcA was determined as 0.33.
Data were plotted as log Mr over Kav resulting in a linear
correlation with a R2 =0.9919. Flow: 1 mL/min. Sample:
2 mL KdcA (1 mg/mL) in 50 mM potassium phosphate
buffer, pH 6.5, including 2.5 mM MgSO4, 0.1 mM ThDP
1
(À0.1), 283 (À7.3); H NMR (400 MHz, CDCl3): d=0.81 (t,
3H, J=7.5 Hz, CH3), 1.46–1.65 (m, 2H, CH2), 2.25–2.41 (m,
2H, CH2), 4.37 (bs, OH), 5.08 (s, 1H, CHOH), 7.3–7.4 (m,
5H, ArH) – contains 2 mol% acyloin; 13C NMR (100 MHz,
CDCl3): d=13.5 (CH3), 17.1, 39.7 (CH2), 79.7 (CHOH),
127.4, 128.6, 128.9 (CHar), 138.08 (Cq), 209.4 (C=O); GC-
MS: Rt =9.02 min; MS (70 eV, EI): m/z (%)=178 (1%)
[M+], 107 (100%), 79 (42%).
(R)-1-Hydroxy-1-phenylbutan-2-one (31b): HPLC (Chiral
OM, n-hexane/2-propanol, 95:5, 0.5 mLminÀ1, 408C): ee
>98.5%, Rt =(S) 21.6 min, Rt (R)=25.0 min; 1H NMR
(400 MHz, CDCl3): d=1.01 (t, 3JH,H =7.4 Hz, 3H, CH3),
2.29–2.46 (m, 2H, CH2), 4.3–4.4 (bs, 1H, OH), 5.1 (s, 1H,
CHOH), 7.3–7.41 (m, 5H, Ar-H) – contains 3 mol% benzal-
dehyde and acyloin; 13C NMR (100 MHz, CDCl3): d=7.6
(CH3), 31.1 (CH2), 79.4 (CHOH), 127.3 (2Ar), 128.6, 128.9
(CHar), 138.3 (Cq), 210.1 (C=O); GC-MS Rt =8.44 min; MS
(70 eV, EI): m/z (%)=164 (0.1%) [M+], 107 (100%), 79
(81%).
(R)-1-Hydroxy-1-phenyl-4-methylpentan-2-one (33b): Iso-
lated yield: 25%; ee=88%; HPLC, Chiralcel OD-H, n-
hexane/2-propanol, 98:2, 0.5 mLminÀ1, 258C, Rt (S)=
22.9 min, Rt (R)=29.5 min; [a]23: À256.2 (c 0.2 g/100 mL,
CDCl3); CD (acetonitrile): l (DDe) [nm] (mol. CD)=199
(+20.9), 218 (+11.05), 231 (+0.9), 237 (+0.04), 283 (À8.5);
1H NMR (400 MHz, CDCl3): d=0.75 (d, 3H, J=6.5 Hz,
CH3), 0.89 (d, 3H, J=6.5 Hz, CH3), 2.04–2.31 (m, 3H, CH,
Determination ofpH and Temperature Optima
The pH optimum was measured using the coupled decar-
boxylase assay. For determination of the temperature opti-
mum the direct decarboxylase assay was used.
Stability Investigations
For investigation of the stability towards pH, temperature,
and organic solvents KdcA was incubated under the reaction
conditions given in the Figure legends and residual activity
was assayed with the coupled decarboxylase assay.
Analytical Performance
The conversion was followed by GC/MS, employing an HP
6890 series GC system fitted with an HP 5973 mass selective
detector (Hewlett Packard; column HP-5MS, 30 m ”
Adv. Synth. Catal. 2007, 349, 1425 – 1435
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1433

Dçrte Gocke et al.
FULL PAPERS
CH2), 4.39 (d, J=4.1 Hz, OH), 5.05 (d, 1H, J=4.1, CHOH),
7.29–7.41 (m, 5H, ArH) – contains 20 mol% acyloin and
substrate; 13C NMR (100 MHz, CDCl3): d=22.2 (CH3), 22.4
(CH3), 24.6 (CH), 46.6 (CH2), 80.0 (CHOH), 127.4, 128.6,
128.9 (CHar), 137.9 (Cq), 209.0 (C=O); GC-MS: Rt =
9.30 min; MS (70 eV, EI): m/z (%)=192 (1%) [M+], 136
(1%) , 107 (100%), 79 (58%);
Acknowledgements
Skilful technical assistance of Ilona Frindi-Wosch (enzyme
purification and size exclusion chromatography) and Elke
Breitling (synthesis of indole-3-acetaldehyde) is gratefully ac-
knowledged. The authors thank Degussa AG for financial
support.
(R)*-2-Cyclopropyl-1-hydroxy-1-phenylethanone
(34b):
Isolated yield 14%; ee=98%; HPLC (Chiracel OD-H, n-
hexane/2-propanol, 95:5, 0.5 mLminÀ1, 408C): Rt (S)=
16.6 min, Rt (R)=21.15 min; 1H NMR (400 MHz, CDCl3):
d=0.78–0.86 (m, 1H), 0.94–1.2 (m, 3H), 1.83–1.9 (m, 1H),
4.39 (bs, 1H, OH), 5.27 (s, 1H, CHOH), 7.3–7.45 (m, 5H,
Ar-H) – contains 20 mol% acyloin; 13C NMR (100 MHz,
CDCl3): d =12.1, 12.7 (CH2), 17.6 (CH), 80.1 (CHOH),
127.7, 128.6, 128.9 (CHar), 138.1 (Cq), 209.6 (C=O); GC-MS:
Rt =9.45 min; MS (70 eV, EI): m/z (%)=176 (3%) [M+],
107 (100%), 79 (71%).
References
[1] T. Stillger, M. Pohl, C. Wandrey, A. Liese, Org. Proc.
Res. Develop. 2006, 10, 1172–1177.
[2] M. Knoll, M. Müller, J. Pleiss, M. Pohl, ChemBioChem
2006, 7, 1928–1934.
[3] A. S. Demir, O. Sesenoglu, P. Dünkelmann, M. Müller,
Org. Lett. 2003, 5, 2047–2050.
(R)-1-(3,5-Dichlorophenyl)-2-hydroxypropan-1-one (35a):
HPLC (Chiralpak AD, n-hexane/2-propanol, 98:2,
0.75 mLminÀ1, 208C): ee=96.5%, Rt (S)=20.4 min, Rt (R)=
25.4 min; CD (acetonitrile): l (De) [nm] (mol. CD)=212
(+4.9), 211 (+9.5), 245 (À2.0), 289 (+1.65); 1H NMR
(400 MHz, CDCl3): d=1.46 (d, 3H, J=7.1, CH3), 3.56 (bs,
1H, OH), 5.07 (q, 1H, J=7.1, CHOH), 7.61 (t, 1H, J=2.07,
ArH), 7.78 (d, 2H, J=2.07, ArH) – contains 8 mol% 3,5-di-
chlorobenzaldehyde; 13C NMR (100 MHz, CDCl3): d=21.9
(CH3), 69.6 (CHOH), 126.9, 133.5 (CHar), 135.9, 136.0 (Cq),
200.2 (C=O); GC-MS: Rt =10.04 min; MS (70 eV, EI): m/z
(%)=218 (7%) [M+], 175 (100%), 147 (43%), 111 (71%).
3-Hydroxy-1-phenylbutan-2-one (36a) and 1-phenyl-2-hy-
droxybutan-3-one (36b): 36a: yield: 49%; ee=not deter-
[4] P. Dominguez de Maria, T. Stillger, M. Pohl, S. Wallert,
K. Drauz, H. Grçger, H. Trauthwein, A. Liese, J. Mol.
Catal. B: Enzym. 2006, 38, 43–47.
[5] P. Siegert, M. J. McLeish, M. Baumann, H. Iding,
M. M. Kneen, G. L. Kenyon, M. Pohl, Protein Eng.
Des. Sel. 2005, 18, 345–357.
[6] M. Pohl, B. Lingen, M. Müller, Chem. Eur. J. 2002, 8,
5288–5295.
[7] A. S. Demir, O. Sesenoglu, E. Eren, B. Hosrik, M.
Pohl, E. Janzen, D. Kolter, R. Feldmann, P. Dünkel-
mann, M. Müller, Adv. Synth. Catal. 2002, 344, 96–103.
[8] P. Dünkelmann, D. Kolter-Jung, A. Nitsche, A. S.
Demir, P. Siegert, B. Lingen, M. Baumann, M. Pohl, M.
Müller, J. Am. Chem. Soc. 2002, 124, 12084–12085.
[9] G. Smit, B. A. Smit, W. J. Engels, FEMS Microbiol.
Rev. 2005, 29, 591–610.
[10] B. A. Smit, J. Vlieg, W. J. M. Engels, L. Meijer, J. T. M.
Wouters, G. Smit, Appl. Environ. Microbiol. 2005, 71,
303–311.
[11] M. de La Plaza, P. Fernandez de Palencia, C. Pelaez, T.
Requena, FEMS Microbiol. Lett. 2004, 238, 367–374.
[12] A. Yep, G. L. Kenyon, M. J. McLeish, Bioorg. Chem.
2006, 34, 325–336.
1
mined; H NMR (400 MHz, CDCl3, 300 K): d=1.4 (d, 3H,
J=7.1 Hz, CH3), 3.55 (d, 1H, J=4.6 Hz, OH), 3.77 (d, 1H,
J=15.8 Hz, CH2), 3.83 (d, 1H, J=15.8 Hz, CH2), 4.3–4.36
(m, 1H, CHOH), 7.18–7.34 (m, 5H, ArH) – contains 19
mol% 36b; 13C NMR (100 MHz, CDCl3, 300 K): d=19.7
(CH3), 44.5 (CH2), 72.2 (CHOH), 127.2, 128.7, 129.4, 133.1,
210.04 (C=O); GC-MS: Rt =8.6 min; MS (70 eV, EI): m/z
(%)=164 (1%) [M+], 146 (30%), 121 (41%), 103 (38%),
91 (100%).
36b: yield: 12%; ee=not determined; 1H NMR
(400 MHz, CDCl3, 300 K): d =2.19 (s, 3H, CH3), 2.87 (dd,
1H, J=14.2 Hz, J=7.5 Hz, CH2), 3.12 (dd, 1H, J=14.2 Hz,
J=4.6 Hz, CH2), 3.52 (d, 1H, J=5.3 Hz, OH), 4.38–4.42 (m,
1H, CHOH), 7.18–7.34 (m, 5H, ArH) – contains 36a;
13C NMR (100 MHz, CDCl3, 300 K): d=25.8 (CH3) 39.8
(CH2) 77.6 (CHOH), 126.8, 128.5, 129.2, 209.4 (C=O); GC-
MS: Rt =8.6 min; MS (70 eV, EI): m/z (%)=164 (1%)
[M+], 146 (30%), 121 (41%), 103 (38%), 91 (100%).
3-Hydroxy-1-(3-indolyl)-butan-2-one (37a): Yield: 23%;
ee=not determined; 1H NMR (400 MHz, CDCl3, 300 K):
d=1.46 (d, 3H, J=7.1 Hz, CH3), 3.43 (d, 1H, J=4.9 Hz,
OH), 3.96 (s, 2H, CH2), 4.38–4.45 (m, 1H, CHOH), 7.14–
7.18 (m, 1H, ArH+1H, NCH), 7.21–7.25 (m, 1H, ArH),
7.39 (d, 1H, ArH), 7.53 (d, 1H, ArH), 8.06–8.19 (bs, 1H,
NH); GC-MS: Rt =12.54 min; MS (70 eV, EI): m/z (%)=
203 (9.5%) [M+], 130 (100%).
[13] P. Arjunan, T. Umland, F. Dyda, S. Swaminathan, W.
Furey, M. Sax, B. Farrenkopf, Y. Gao, D. Zhang, F.
Jordan, J. Mol. Biol. 1996, 256, 590–600.
[14] D. Dobritzsch, S. Kçnig, G. Schneider, G. Lu, J. Biol.
Chem. 1998, 273, 20196–20204.
[15] M. S. Hasson, A. Muscate, M. J. McLeish, L. S. Polovni-
kova, J. A. Gerlt, G. L. Kenyon, G. A. Petsko, D.
Ringe, Biochemistry 1998, 37, 9918–9930.
[16] A. K. Chang, R. G. Duggleby, Biochem. J. 1997, 327,
161–169.
[17] Y. Lindqvist, G. Schneider, U. Ermler, M. Sundstrom,
Embo J. 1992, 11, 2373–2379.
[18] H. Iding, T. Dünnwald, L. Greiner, A. Liese, M.
Müller, P. Siegert, J. Grçtzinger, A. S. Demir, M. Pohl,
Chem. Eur. J. 2000, 6, 1483–1495.
[19] F. Hildebrand, S. Kühl, M. Pohl, D. Vasic-Racki, M.
Müller, C. Wandrey, S. Lütz, Biotechnol. Bioeng. 2007,
96, 835–843.
[20] P. Dominguez de Maria, T. Stillger, M. Pohl, S. Wallert,
K. Drauz, H. Grçger, H. Trauthwein, A. Liese, J. Mol.
Catal. B: Enzym. 2005, 38, 43–47.
1434
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1425 – 1435

Branched-Chain Keto Acid Decarboxylase from Lactococcus lactis
FULL PAPERS
[21] E. Janzen, M. Müller, D. Kolter-Jung, M. M. Kneen,
M. J. McLeish, M. Pohl, Bioorg. Chem. 2006, 34, 345–
361.
[22] P. Dominguez de Maria, M. Pohl, D. Gocke, H. Grçger,
H. Trauthwein, L. Walter, M. Müller, Eur. J. Org.
Chem. 2007, in press; DOI 10/1002/ejoc.200600876.
[23] A. Schütz, R. Golbik, K. Tittmann, D. I. Svergun,
M. H. Koch, G. Hübner, S. Kçnig, Eur. J. Biochem.
2003, 270, 2322–2331.
[24] J. Koga, T. Adachi, H. Hidaka, J. Biol. Chem. 1992,
267, 15823–15828.
[25] D. J. Korz, U. Rinas, K. Hellmuth, E. A. Sanders, W.-D.
Deckwer, J. Biotechnol. 1995, 39, 59–65.
[26] M. M. Bradford, Anal. Biochem. 1976, 72, 248–254.
Adv. Synth. Catal. 2007, 349, 1425 – 1435
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1435