Genome mining for innovative biocatalysts: New Dihydroxyacetone aldolases for the chemist's toolbox

Article abstract of DOI:10.1002/cctc.201500014

Stereoselective carboligating enzymes were discovered by a genome mining approach to extend the biocatalysis toolbox. Seven hundred enzymes were selected by sequence comparison from diverse prokaryotic species as representatives of the aldolase (FSA) family diversity. The aldol reaction tested involved dihydroxyacetone (DHA) and glyceraldehyde-3-phosphate. The hexose-6-phosphate formation was monitored by mass spectrometry. Eighteen enzymes annotated either as transaldolases or aldolases were found to exhibit a DHA aldolase activity. Remarkably, six of them proven as aldolases, and not transaldolases, shared very limited similarities with those currently described. Multiple sequence alignment performed on all enzymes revealed a Tyr in the new DHA aldolases as found in FSA_coli instead of a Phe usually found in transaldolases. Four of these DHA aldolases were biochemically characterised in comparison with FSA_coli. In particular, an aldolase from Listeria monocytogenes exhibited interesting catalytic properties. Exploiting nature′s catalyst mines: A universal high-throughput screening strategy based on mining genomes and selection of enzyme representatives of the aldolase family is applied. Out of ten hits proven as dihydroxyacetone aldolases, one from Listeria monocytogenes exhibited highly interesting catalytic properties, comparable to those of FSA_coli.

Full text of DOI:10.1002/cctc.201500014

DOI: 10.1002/cctc.201500014
Full Papers
Genome Mining for Innovative Biocatalysts: New
Dihydroxyacetone Aldolases for the Chemist’s Toolbox
Christine GuØrard-HØlaine,[a, b] VØronique de Berardinis,[c, d, e] Marielle Besnard-Gonnet,[c, d, e]
Ekaterina Darii,[c, d, e] Marine Debacker,[a, b] Adrien Debard,[c, d, e] Carlos Fernandes,[a, b]
Virgil HØlaine,[a, b] Aline Mariage,[c, d, e] Virginie Pellouin,[c, d, e] Alain Perret,[c, d, e] Jean-
Louis Petit,[c, d, e] Martine Sancelme,[a, b] Marielle Lemaire,*[a, b] and Marcel Salanoubat*[c, d, e]
Stereoselective carboligating enzymes were discovered by
a genome mining approach to extend the biocatalysis toolbox.
Seven hundred enzymes were selected by sequence compari-
son from diverse prokaryotic species as representatives of the
aldolase (FSA) family diversity. The aldol reaction tested in-
volved dihydroxyacetone (DHA) and glyceraldehyde-3-phos-
phate. The hexose-6-phosphate formation was monitored by
mass spectrometry. Eighteen enzymes annotated either as
transaldolases or aldolases were found to exhibit a DHA aldol-
ase activity. Remarkably, six of them proven as aldolases, and
not transaldolases, shared very limited similarities with those
currently described. Multiple sequence alignment performed
on all enzymes revealed a Tyr in the new DHA aldolases as
found in FSAcoli instead of a Phe usually found in trans-
aldolases. Four of these DHA aldolases were biochemically
characterised in comparison with FSAcoli. In particular, an aldol-
ase from Listeria monocytogenes exhibited interesting catalytic
properties.
Introduction
The need for innovative biocatalysts with improved or unnatu-
ral properties is still crucial thanks to the great potential of en-
zymes with synthetically useful properties. The bottleneck of
the biocatalytic processes is often limited by the availability of
suitable enzymes.[1] To meet this requirement, three main strat-
egies are accessible: i) evolution of known enzymes,[2] ii) de no-
vo design of new biocatalysts by computational design,[3] iii)
search for new enzymes in the biodiversity.[2a,3a,4] Owing to the
tremendous microbial biodiversity, it is highly probable that
many valuable activities have still to be found among wild-
type enzymes that have evolved over billions of years.[5] Aldol-
ases belong to a well-established class of enzymes catalysing
one of the most important reactions in organic synthesis, ste-
reoselective carbon–carbon ligation. They proved to be attrac-
tive and powerful for the synthesis of natural products and
complex bioactive small molecules.[6] Despite their broad toler-
ance to the aldehyde acceptor (electrophile), their use for syn-
thesis is limited by the narrow ketone donor (nucleophile) spe-
cificity. Thus there is a real need to find new aldolases able to
display broader donor specificities, in particular, for nonphos-
phorylated substrates to increase the chemist’s toolbox of ver-
satile biocatalysts.[6c,g] The discovery of fructose-6-phosphate
aldolases (FSAAcoli and FSABcoli) in Escherichia coli has opened
up exciting new perspectives.[7] Indeed, FSAscoli possesses the
unique capacity to react efficiently with at least four different
non-phosphorylated donor substrates (dihydroxyacetone,
DHA; hydroxyacetone, HA; hydroxybutanone and glycolalde-
hyde) and various acceptors.[7a,8] Given the extraordinary po-
tential of FSAs, using wild-type or variants, many syntheses
were described to prepare especially antidiabetic (glycosidase
inhibitors), anti-infective compounds or phosphorylated sugars
as metabolites.[6c,9] Since their discovery in 2001, no new natu-
ral efficient DHA aldolases have been identified[6g] from other
microorganisms unlike most other aldolases. However, E. coli
transaldolase B (TalBcoli) and DHAP-dependent rhamnulose
aldolase display a slight aldol promiscuous activity towards
DHA that has been improved by mutagenesis.[10]
[a] Dr. C. GuØrard-HØlaine,+ M. Debacker, Dr. C. Fernandes, Dr. V. HØlaine,
M. Sancelme, Prof. M. Lemaire+
Clermont UniversitØ, UniversitØ Blaise Pascal
ICCF, BP 10448, 63000 Clermont-Ferrand (France)
E-mail: marielle.lemaire@univ-bpclermont.fr
[b] Dr. C. GuØrard-HØlaine,+ M. Debacker, Dr. C. Fernandes, Dr. V. HØlaine,
M. Sancelme, Prof. M. Lemaire+
CNRS, UMR 6296,BP 8002663177 Aubi›re (France)
[c] Dr. V. de Berardinis,+ M. Besnard-Gonnet, Dr. E. Darii, A. Debard, A. Mariage,
V. Pellouin, Dr. A. Perret, J.-L. Petit, Dr. M. Salanoubat+
CEA, DSV, IG, Genoscope
2 rue Gaston CrØmieux, 91057 Evry (France)
E-mail: salanou@genoscope.cns.fr
[d] Dr. V. de Berardinis,+ M. Besnard-Gonnet, Dr. E. Darii, A. Debard, A. Mariage,
V. Pellouin, Dr. A. Perret, J.-L. Petit, Dr. M. Salanoubat+
CNRS-UMR 8030
2 rue Gaston CrØmieux, 91057 Evry (France)
[e] Dr. V. de Berardinis,+ M. Besnard-Gonnet, Dr. E. Darii, A. Debard, A. Mariage,
V. Pellouin, Dr. A. Perret, J.-L. Petit, Dr. M. Salanoubat+
UniversitØ d’Evry Val d’Essonne
boulevard FranÅois Mitterrand, 91057 Evry (France)
[+] These authors contributed equally to this work.
Here we used a universal large-scale LC–MS screening
method to prospect among different enzyme families for DHA
aldolases using d,l-glyceraldehyde 3-phosphate (d,l-G3P) as
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201500014.
ChemCatChem 2015, 7, 1871 – 1879
1871
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
On each microtiterplate, E. coli cell lysate was used as a nega-
tive control with FSAAcoli as a positive control. Hits were con-
sidered as positive when the signal was at least three-fold
higher than the negative control. This methodology led to the
selection of 19 positive hits (Supporting Information, Table 5).
To confirm these results, the corresponding genes were cloned
with a His-tag and 17 enzymes were successfully purified by
nickel affinity chromatography (Supporting Information,
Table 5). Among the two unsuccessful purifications,
A0A0E4G3U3 (ALDlisteria), one of the strongest hits, was purified
in its native form (untagged) by fast protein liquid chromatog-
raphy (FPLC). The formation of a hexulose-6-phosphate was
confirmed by LC–MS analysis for these 18 purified enzymes.
Among them, only two were annotated as aldolases (FSA),
whereas the other 16 were annotated as transaldolases in Uni-
protKB database (Supporting InformationTable 5). To analyse
these two activities further, known spectrophotometric assays
were applied as described in Figure 1 (2)–(4). For aldolase ac-
tivity, fructose-6-phosphate (F6P) formation was monitored by
following NADPH appearance (Figure 1 (2)). Retroaldolisation
of d-F6P was also assayed through d-G3P formation by moni-
toring NADH disappearance (Figure 1, entry 3).[7a] FSAAcoli was
used as a positive control. Transaldolase activity was assayed
by using F6P and glycolaldehyde as an acceptor. d-G3P forma-
tion was monitored as for retroaldolase activity assay (Figure 1
(4)). TalBcoli was used as a positive control.
Scheme 1. Aldolisation catalysed by DHA aldolases.
the acceptor substrate (Scheme 1). In this study, to explore the
DHA aldolase families, we applied a sequence-driven approach
that successfully found new nitrilases from the natural diversi-
ty.[11] To our knowledge no such large genome mining ap-
proach for new aldolases discovery have been described in lit-
erature.
Results and Discussion
Enzyme selection and preliminary characterisation
We reasoned that natural promiscuous catalytic properties are
widespread in several enzyme families. Thus sequence compar-
isons against UniprotKB proteins with already known aldolases
(EC 4.1.2 with the exception PLP dependent glycine aldolases),
including FSA from E. coli and transaldolases (EC 2.2.1), were
conducted and allowed selection of 1148 candidate proteins
from 313 different prokaryote species representative of the al-
dolase/transaldolase diversity. Among the candidate proteins,
731 genes (64%) from approximately 230 species were suc-
cessfully cloned in an expression vector using automated plat-
form facilities. Out of these, overexpression of 571 proteins
(79%) were visible on sodium dodecyl sulfate polyacrylamide
gel electrophoresis (SDS PAGE). The sequences of nine new
DHA aldolases and five new transaldolases have been submit-
ted to European Nucleotide Archive and UniprotKB Id are re-
ported in the Supporting Information Table 2b.
DHA aldolase activity was clearly detected for the two anno-
tated as FSA and for six enzymes annotated as transaldolases.
The transaldolase activity was confirmed for 10 out of the 16
putative transaldolases. The six enzymes annotated as “transal-
dolases” active for both aldolisation and retroaldolisation reac-
tions did not display any transaldolase activity (Table 1). This is
corroborated by amino acid sequence comparisons that show
that these enzymes revealed low sequence identity (<30%
identity) with various transaldolases but display a higher se-
quence similarity with FSAAcoli (<41% identity, Supporting In-
formation, Table 6). Five among these six aldolases are from
various Streptococcus species and shared sequence identities
between 60% and 79%. Interestingly, ALDlisteria from Listeria
monocytogenes that seemed to be the most efficient enzyme
during the preliminary studies shares no sequence identity
with FSAAcoli. The family of transaldolases and related aldolases
are divided into five subfamilies. Subfamily 1 contains the
“classical” transaldolases, whereas subfamily 5 contains FSAs
from E. coli.[12] FSAs are specified by the presence of the three
signatures from the protein sequence analysis and classifica-
tion Interpro database: IPR001585 corresponding to transaldol-
ases and related aldolases families, IPR04731 and finally
IPR023001 (see Supporting Information Figure 1). Only B7NAG9
corresponded to this context, the seven other DHA aldolases
simply shared the IPR001585 (See Supporting Information,
Table 7).
Proteins of interest were identified by LC–MS by monitoring
hexulose-6-phosphate formation produced by the aldol addi-
tion of dihydroxyacetone (DHA) as the donor and d,l-G3P as
the acceptor (Figure 1 (1)).
Figure 1. High-throughput screening (HTS) methodology and initial catalytic
activity characterisation. G3P=glyceraldehyde-3-phosphate, F6P=fructose-
6-phosphate, DHA=dihydroxyacetone, DHAP=dihydroxyacetone phos-
phate, GOP=glycerolphosphate, G6P=glucose-6-phosphate, 6-PGL=6-
phosphonoglucono-1,5-lactone, X=xylulose, GLA=glycolaldehyde,
TPI=triose phosphate isomerase, GPDH=glycerol-3-phosphate dehydro-
genase, GPI=glucose-6-phosphate isomerase, G6PD=glucose-6-phosphate
dehydrogenase
A multiple sequence alignment including these enzymes
and FSAAcoli and TalBcoli as references demonstrate that essen-
tial catalytic and substrate binding residues[10a,13] are highly
conserved in these eight aldolases despite a low level of se-
quence conservation with FSAAcoli for seven of them (Figure 2).
ChemCatChem 2015, 7, 1871 – 1879
www.chemcatchem.org
1872
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 1. Experimental validation of the 18 new aldolases/transaldolases performed on purified enzymes. Enzymes from E. coli strain K12 are used as refer-
ences.
UniprotKB id Annotation
Genome
Aldolase activity Retroaldolisation activity Transferase activity
P78055[a]
B7NAG9
FSA A[b]
FSA 1[b]
E. coli (strain K12)
E. coli O17:K52:H18
Streptococcus pneumoniae
Listeria monocytogenes serotype 4b str. H7858
+
+
+
+
+
+
+
+
À
À
À
À
A0A0D6J3Z8 FSA[b]
A0A0E4G3U3 transaldolase
(ALDlisteria
A0A0E4C393 transaldolase
ALDSmut transaldolase
A0A0D6H018 transaldolase
(ALDSpyog
A0A0E4C363 transaldolase
)
Streptococcus suis
Streptococcus mutans serotype c
Streptococcus pyogenes serotype M1
+
+
+
+
+
+
À
À
À
)
Streptococcus gordonii
+
+
À
(ALDSgord
)
Q8E738
transaldolase family protein Streptococcus agalactiae
+
À
À
À
À
À
À
À
À
À
À
À
+
À
À
À
À
À
À
À
À
À
À
À
À
+
+
+
+
+
+
+
+
+
+
+
P0A870[a]
Q7M7U6
Q4JVK0
transaldolase B
transaldolase
transaldolase
Escherichia coli (strain K12)
Wolinella succinogenes
Corynebacterium jeikeium
Shewanella baltica
A0A0E3VZZ2 transaldolase
A0A0E4C370 transaldolase
Q3IG49
Q8NQ64
A9E0W9
A1A1M8
A0A0E4C394 transaldolase
Q1H0R4 transaldolase
Pseudomonas mendocina
Pseudoalteromonas haloplanktis
Corynebacterium glutamicum
Oceanibulbus indolifex
Bifidobacterium adolescentis
Neisseria meningitidis
transaldolase
transaldolase
probable transaldolase
transaldolase
Methylobacillus flagellatus
[a] Known enzymes served as references; [b] FSA=fructose-6-phosphate aldolase.
(C2DGL7,
A0A0E4G3V8,
A0A0E4C372, Q8Y990) with a ty-
rosine were originated from phy-
logenetically diverse organisms
(Aeromonas, Enterococcus, Clostri-
dium etc.) extending the diversi-
ty of organism far beyond the
Listeria and the Streptococcus
genera (Supporting Information,
Table 8).
Biochemical characterization
of these 11 enzymes confirmed
that the presence of the tyrosine
is related to DHA aldolase activi-
ty and that of the phenylalanine
to transaldolase activity (Sup-
porting Information, Table 9). For
C2DGL7 and A0A0E4G3V8, no
expression could be detected on
Figure 2. Specific residues identified in the active site. Numbers correspond to FSAAcoli sequence. Residues in-
volved in binding water molecules and/or deprotonation are shown in blue. Residues conferring apolar properties
on the donor side are shown in purple. Residues involved in phosphate binding are shown in green. The catalytic
lysine is highlighted in red. A 3D structure of FSAAcoli with these highlighted conserved residues was proposed
(Supporting Information, Figure 3).
Interestingly, the eight DHA aldolases have a tyrosine as in
FSAAcoli (Tyr131) instead of a phenylalanine usually found in
transaldolases.[14] In FSAAcoli, this tyrosine contributes to the al-
dolase activity by binding a water molecule together with
Gln59 and Thr109 in the active site.[13] To reinforce the correla-
tion between the presence of the tyrosine residue and the al-
dolase activity, we purified the 11 annotated transaldolases
from the collection for which LC–MS screening results did not
allow their initial selection. Multiple sequence alignment re-
vealed that four of them have a tyrosine and one from Micro-
scilla marina (A1ZYC6) a leucine instead of a phenylalanine
(Supporting Information, Figure 2). These four enzymes
SDS PAGE. Notably, no significant activity could be detected
for A1ZYC6 harbouring a leucine in the same position despite
a good overexpression.
As illustrated by the phylogenetic tree (Figure 3), new trans-
aldolases characterised in this study were found in all the dif-
ferent subfamilies of already described transaldolases except in
subfamily 2 which is composed of transaldolases of higher
plants.[14] We suggest that the new DHA aldolases could be
either grouped in subfamily 5 or comprise a new subfamily
since they did not share any of the two FSA Interpro signature
(Figure 3, in squared dash points).
ChemCatChem 2015, 7, 1871 – 1879
www.chemcatchem.org
1873
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
using hydroxyacetone (HA) and
d-G3P, which are the best sub-
[7b,9d]
strates of FSAAcoli
.
By analo-
gy with FSAAcoli, which is highly
stereoselective, the expected
product should be exclusively
1-deoxy-d-F6P (d-d-F6P). The
aldols were isolated as their
barium salts and studied by
13C NMR spectroscopy. In all
cases, only one compound was
detected and identified as d-d-
F6P. Thus, the new aldolases
displayed a high stereoselectiv-
ity leading to the (3S, 4R)
adduct. With this information in
hand, deeper biochemical inves-
tigations were conducted.
The specific activities for the
retroaldol reaction were deter-
mined on purified enzymes
Figure 3. Phylogenetic tree including reference enzymes of each transaldolase subfamily (in black)[14] with bio-
chemical data or 3D structures with the enzymes from our study shown in colour (blue: transaldolase activity;
green: aldolase activity; red: no activity). Column 1 indicates the sequence variation corresponding to Tyr131 in
FSA; column 2 indicates the respective family, if applicable, or the activity determined in this study (FSA-like or
transaldolase-like); column 3 and 4 indicate the annotation in databases according to sequence similarities and
the source organism. nd: not done.
(Supporting
Information,
Table 12). ALDlisteria displayed the
highest specific activity, 5-fold
higher than the one found for
Genes encoding transaldolase-like proteins have been found
associated with catabolic operons in some organisms, but their
functions remain unknown.[15] An analysis of the genomic con-
text of the new DHA aldolases (Supporting Information,
Table 10) revealed that a large majority (nine enzymes) are in-
cluded in a gene cluster with predicted genes involved in
sugar transport systems such as FSABcoli. The bacterial sugar
phosphotransferase system (PTS) catalyses the transfer of the
phosphoryl group from phosphoenolpyruvate to its sugar sub-
strates concomitant with the translocation of these sugars
across the bacterial membrane. The DHA aldolases are general-
ly annotated as Tal-like proteins in these clusters and many of
them are flanked by a pyruvate formate lyase and a glycerol
dehydrogenase (Supporting Information, Table 10).
FSAAcoli. Besides this result, ALDSgord and ALDSmut were found 3-
fold more active than FSAAcoli
.
FSAAcoli was found to be particularly thermostable even if
not originating from a thermophilic microorganism. To contin-
ue the comparison, thermostability properties of these four en-
zymes were studied. The activity of all enzymes but one
(ALDSmut) remained unchanged after a 40 min incubation at
708C, which is similar to that of FSAAcoli (Supporting Informa-
tion, Figure 3). Thus this property could be exploited to simpli-
fy the purification process of these three enzymes. Moreover,
despite its presence in the E. coli genome, FSAAcoli, is not ex-
pressed under usual growth conditions and will not contami-
nate the heat-stable overexpressed enzymes.[7b]
The pH profiles of these enzymes were revealed to be rela-
tive similar in between and with FSAscoli, showing an optimum
at 8.5. Remarkably, ALDSgord , and ALDSpyog retained 97 and 90%
of their respective activity when lowering the pH to 7. This
result could be exploited to conduct aldol reactions in the
presence of acceptors sensitive to alkaline pH. Five buffers
(phosphate; N-(2-hydroxyethyl)piperazine-N’-propanesulfonic
acid, HEPES; tetraethylammonium, TEA; tris(hydroxymethyl)-
aminomethane, Tris and glycylgycine, glygly) were then tested,
and no significant difference was revealed except for ALDSmut
for which 50% of activity was lost in the presence of HEPES.
Steady-state kinetic parameters were determined for the
four aldolases by using FSAAcoli as a reference. The results are
presented in Table 2. The retroaldol reaction was tested on
three substrates. Among them, d-d-F6P was revealed as the
best substrate for all the aldolases, in particular for FSAAcoli. AL-
Characterization of the most promising new DHA aldolases
A biochemical characterisation of four aldolases was then con-
ducted to evaluate their biocatalytic capacity (A0A0D6H018-
from S. pyogenes (ALDSpyog), A0A0E4C363 from S. gordonii
(ALDSgord), Q8DVJ4 from S. mutans (ALDSmut)and ALDlisteria). Kinet-
ic parameters were determined and compared to those of
FSAAcoli. Specificity spectra towards the acceptor substrates
were studied by varying their hydrophobicity, polarity and
chain length. These assays were conducted on purified en-
zymes (see Experimental Section).
First we determined the stereoselectivity of these enzymes.
The spectrophotometric assays based on the retroaldolisation
of F6P (Figure 1) indicated that the (3S, 4R) configuration could
be processed but did not exclude the processing of the three
others. To analyse this further, small-scale reactions were per-
formed in the same conditions as for the four enzymes by
D
listeria was the more efficient on d-F6P and d-arabinose-5-phos-
phate (d-A5P), followed by ALDSgord. Interestingly, the result
that d-A5P was found as a substrate suggests that glycolalde-
ChemCatChem 2015, 7, 1871 – 1879
www.chemcatchem.org
1874
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
highest conversion rates. Penta-
nal was partially converted after
24 h incubation by all enzymes
except FSAAcoli for which a total
reaction was observed. Con-
cerning the aldose substrates,
the increase of carbon-chain
length reduced the conversion
rates. Thus, glyceraldehyde was
found to be the best acceptor
followed by erythrose, which
was a moderate substrate for
Table 2. Kinetic parameters determination for various substrates at pH 8.0 in 50 mm glycylglycine buffer (kcat
calculated for one active site)
[b]
Substrate
Kinetic
FSAAcoli
ALDSgord
ALDSmut
ALDSpyog
ALDlisteria
parameters
Km [mm]
19Æ1
28Æ1
39Æ6
18Æ2
8.8Æ0.4
d-F6P[a]
kcat [sÀ1
]
0.60Æ0.01
0.32Æ0.002
4.1Æ0.2
0.14Æ0.01
3.1Æ0.5
0.08Æ0.02
1.2Æ0.1
0.06Æ0.01
2.3Æ0.1
k
cat/Km [sÀ1 mmÀ1
]
]
]
]
]
0.26Æ0.02
Km [mm]
kcat [sÀ1
0.28Æ0.02
25Æ1
0.97Æ0.06
32Æ1
0.67Æ0.07
12Æ1
0.39Æ0.02
10Æ1
0.56Æ0.04
17Æ1
d-d-F6P[a]
d-A5P[a]
d,l-G3P[a]
DHA[a]
]
kcat/Km [sÀ1 mmÀ1
89Æ6
33Æ2
18Æ2
26Æ2
30Æ2
Km [mm]
0.40Æ0.05
0.43Æ0.05
1.1Æ0.2
0.29Æ0.01
0.55Æ0.02
1.9Æ0.1
0.41Æ0.03
0.40Æ0.03
1.0Æ0.1
0.12Æ0.01
0.25Æ0.01
2.1Æ0.1
0.05Æ0.001
0.46Æ0.01
9.2Æ0.2
kcat [sÀ1
]
ALDSmut and ALDSpyog
. Alde-
kcat/Km [sÀ1 mmÀ1
hydes with more than four car-
bons were no acceptor. Howev-
er, the C5 phosphorylated ac-
ceptor (ribose-5-phosphate) was
accepted by all enzymes with
the same conversion rate. The
presence of this phosphate
group enhanced the catalytic
efficiency.
Km [mm]
0.5Æ0.1
6Æ2
0.23Æ0.03
19Æ2
0.54Æ0.09
11Æ2
0.49Æ0.07
13Æ2
0.12Æ0.01
13Æ1
kcat [sÀ1
]
kcat/Km [sÀ1 mmÀ1
12Æ5
80Æ10
20Æ5
27Æ5
108Æ10
Km [mm]
16Æ1
50Æ5
27Æ2
24Æ2
64Æ3
kcat [sÀ1
]
6.9Æ0.1
0.43Æ0.01
18Æ2
6.6Æ0.4
0.24Æ0.02
8.2Æ0.5
0.34Æ0.03
22Æ1
kcat/Km [sÀ1 mmÀ1
0.36Æ0.05
0.34Æ0.02
[a] F6P=fructose-6-phosphate, G3P=glyceraldehyde-3-phosphate, d-d-F6P=1-deoxy-d-fructose-6-phosphate,
A5P=arabinose-5-phosphate, DHA=dihydroxyacetone. [b] Native protein purified by FPLC.
As previously illustrated,[9f]
A5P gave no direct aldolisation
with HA. The TLC analysis indi-
hyde is an acceptor of the aldol reaction (retroaldolisation re-
action lead to glycolaldehyde). This activity is exceptional
among aldolase families and until present was only demon-
strated for FSAscoli.
cated after a few minutes that A5P was cleaved into glycolal-
dehyde and G3P. These two compounds quickly reacted with
HA to give 1-deoxy-d-xylulose and d-d-F6P, respectively. The
cleavage could not be avoided even by adding 10 equivalents
of HA in order to saturate the active site with the donor.
In addition, all the enzymes exhibited similar specificity for
DHA. As the sole acceptor assayed, d,l-G3P was revealed as an
excellent acceptor substrate, particularly for ALDlisteria and
ALDSgord . ALDlisteria showed the highest catalytic efficiency for
the aldol reaction with G3P (9-fold higher than FSAAcoli, the
lowest one).
To demonstrate the biocatalytic potential of these aldolases,
small-scale synthesis of three compounds was performed. The
best new aldolase ALDlisteria was chosen for the proof of con-
cept. Thus propanal, d-glyceraldehyde and 4-nitrobutanal were
used as acceptor substrates with DHA as the donor (see Sup-
porting Information, Scheme S1). These acceptors were select-
These results highlight that the broad tolerance towards var-
ious substrates described for FSAAcoli is also a property of
these new aldolases. ALDlisteria and ALDSgord are the most prom-
ising enzymes for biocatalysis applications.
ed because they lead to aldols previously prepared with
[9b,16]
FSAAcoli
.
To note, 4-nitrobutanal can lead to a nitrocyclitol
To evaluate each aldolase’s synthetic potential, various ac-
ceptor substrates were reacted with HA, which is the best
donor for FSAcoli. The reactions were monitored by thin layer
chromatography (TLC) over a period of 24 h.
Table 3. TLC analysis for various enzymes.[a]
[c]
Acceptor
FSAAcoli
ALDSgord
ALDSmut
ALDSpyog
ALDlisteria
Reaction conditions were standardised[16] for all the enzymes
to compare the results (see Experimental Section). In the case
of phosphorylated acceptors, two TLCs were performed either
in a very polar eluent adapted to phosphorylated compounds
or in a classical one to follow HA disappearance. Depending
on the enzyme assayed, this methodology enabled classifica-
tion of the substrates into four categories: high, medium, low,
and no substrate. The aldehydes were chosen according to
their hydrophobicity (benzaldehyde, pentanal) or hydrophilicity
(phosphorylated or polyhydroxylated compounds) and carbon-
chain length. Results are summarised in Table 3. The five en-
zymes accept alkyl aldehydes as substrates except benzalde-
hyde. The smallest one, formaldehyde, was fully converted in
24 h with all tested enzymes. FSAAcoli and ALDlisteria gave the
formaldehyde
benzaldehyde
pentanal
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
&
d-Gly[b]
d-erythrose
2-d-d-ribose[b]
d-glucose
d-R5P[b]
d-A5P[b]
d-G6P[b]
[a] Conditions: 50 mm acceptor, 50 mm donor (HA); enzyme (0.07 mg),
H2O pH 8; &: good, total conversion after 3 h; &: medium, total conver-
sion after 12 h, &: low, incomplete conversion after 12 h, &: no conver-
sion after 24 h. [b] d-Gly=d-glyceraldehyde; 2-d-d-ribose=2-deoxy-d-
ribose; d-R5P=d-ribose-5-phosphate, d-G6P=d-glucose-6-phosphate.
[c] Native protein purified by FPLC.
ChemCatChem 2015, 7, 1871 – 1879
www.chemcatchem.org
1875
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
as described in our previous work through an intramolecular
Henry reaction[9b] and propanal was described as a poor sub-
strate by Fessner et al.[16] The three acceptors were substrates
and produced the corresponding expected aldol in 85%, 25%
and 70% yields, respectively. The analytical data were consis-
tent to those described in literature. The most interesting
result relies on the successful conversion of propanal requiring
eight time less enzyme, leading to a two times higher yield in
This approach has extended the chemist’s enzymatic tool-
box with new efficient stereoselective carboligating biocata-
lysts. Further investigations are underway on the search for
other unusual biocatalysed aldol reactions.
Experimental Section
General
a reaction three times faster than the same reaction performed
[16]
d-fructose-6-phosphate disodium salt hydrate, glycolaldehyde
dimer, 1,3-dihydroxyacetone dimer, d,l-glucose 3-phosphate dieth-
yl acetal barium salt, d-(+)-glyceraldehyde, d-ribose-5-phosphate
disodium salt hydrate, d-glucose-6-phosphate dipotassium salt hy-
drate, b-nicotinamide adenine dinucleotide reduced disodium salt
hydrate, b-nicotinamide adenine dinucleotide phosphate sodium
salt, glycylglycine, HEPES, triethanolamine, sodium phosphate
monobasic dihydrate and Trizma hydrochloride, a-glycerophos-
phate dehydrogenase/triosephosphate isomerase from rabbit
muscle, glucose-6-phosphate dehydrogenase and phosphoglucose
isomerase were purchased from Sigma–Aldrich. Hydroxyacetone
was purchased from Fluka and purified by silica gel chromatogra-
phy. Formaldehyde, benzaldehyde and d-glucose were purchased
from Avocado. Valeraldehyde was purchased from Acros. d-eryth-
rose was purchased from Alfa Aesar. 2-deoxy- d-ribose was pur-
chased from Lancaster Synthesis. d-arabinose-5-phosphate and
d-deoxyfructose-6-phosphate were synthesised in the laboratory as
previously described.[9d] Synthetic gene was done by Genewiz. Oli-
gonucleotides were from Sigma-Genosys.
with wild-type FSAAcoli
.
ALDlisteria in its N-terminal His-tag version leads to the forma-
tion of highly aggregated proteins and cannot be purified
properly. Considering its high catalytic properties, a study was
performed to tackle this problem. Alternatives were tested
such as the use of an E. coli BL21 strain optimised for express-
ing rare codons, a synthetic gene with optimised codons for
E. coli, purification with and without an N-terminal His-tag and,
finally, a His-tag introduced at the C-terminus. This last con-
struction allowed us to purify ALDlisteria efficiently by nickel af-
finity chromatography (Supporting Information, Figure 5 and
Table 11). To ensure that the His-tag introduced in terminal po-
sition did not affect the catalytic properties, the Km on F6P was
determined. The presence of the His-tag did not change the
biochemical properties as Km was found at 10.3 mm (for native
enzyme Km =8.8 mm, see Table 2).
Conclusions
Selection of enzymes
A universal high-throughput screening strategy based on
mining genomes and selection of enzymes representative of
the biodiversity was applied to search for new aldolase activi-
ties. High-performance liquid chromatography/mass spectrom-
etry revealed as a powerful and appropriate technique for the
aldol detection, thus bypassing fastidious probe synthesis used
in UV/Vis or fluorescence spectrometry. A total of 26 new bio-
catalysts, including transaldolases, were found and their activi-
ties were validated on purified enzymes. Among the 10 new
DHA aldolases, eight were wrongly annotated as “transaldol-
ases” in the UniprotKB database.
The collection of new aldolases was developed in a three-step pro-
cess. First, a set of proteins was created by collecting information
on known representative of the classes of aldolase and transaldol-
ase. The list of EC numbers was selected on the basis of substrate
structure: DHAP/DHA, pyruvate aldolases and transaldolases (as
they display promiscuous aldolase activity) were chosen (Support-
ing Information, Table 1). Second, this set was compared to Uni-
protKB database and the metagenome from Genoscope using low
stringency parameters (>30% of identity, on 80% of the length)
resulting in the selection of 11 717 candidate enzymes. Third, a clus-
tering based on 80% identity was applied to create putative iso-
functional groups. Representatives of each cluster were selected
for cloning and screening processes. The choice of the primers was
based on the Primer3 program.[17] Specific extensions are added to
the primers for cloning into our plasmid pET22b(+) (Novagen)
modified for ligation independent cloning.[18]
These enzymes such as FSAscoli possess a tyrosine as a third
coordinator of the catalytic water molecule, whereas the trans-
aldolases have a phenylalanine.[10a] The sequences of these
new DHA aldolases do not harbour Interpro motifs specific to
FSAs, and this feature could not allow FSA activity prediction.
Nevertheless, their catalytic properties were found to be very
close to those of the known FSAAcoli and phylogenetic studies
suggest proximity with transaldolase subfamily 5.
Cloning and high-throughput screening
All steps, from primers purchase to cell lysate preparation, were
performed in 96-microwell plates. The cloning was done in the
modified plasmid pET22b(+), the protein production in E. coli
BL21Star (DE3) plysE strains (Invitrogen) and BL21-CodonPlus
(DE3)-RIPL (Agilent technologies), cell lysate preparations and pro-
tein concentrations were performed as previously described.[11] Pri-
mers used for His-tagged variants are described in Supporting In-
formation, Table 2. When strains of the selected proteins were not
available, PCR amplifications were performed on another DSM
strain DNA and the clones were sequenced (Supporting Informa-
tion Table 2 and Table 2b).
Four new DHA aldolases were more deeply characterised.
They were stereoselective for the (3S, 4R) configuration, and
exhibited similar acceptor substrate specificity. Their tolerance
towards different donors, their various pH profiles, and better
catalytic properties than those of FSAAcoli make them promis-
ing biocatalysts for further application in organic synthesis.
Indeed, the most active ALDlisteria was confirmed to be a good
catalyst for use in organic synthesis with dihydroxyacetone as
a donor.
ChemCatChem 2015, 7, 1871 – 1879
www.chemcatchem.org
1876
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Biochemical assays were performed in 96-well microtiterplates.
Enzyme assays were performed in a final volume of 100 mL contain-
ing 1 mm dithyothreitol, 3 mL cell lysate (0.05 to 0.1 mgmLÀ1 of
total proteins), 10 mm of each substrate (50 mm glycylglycine
buffer, pH 7.5). Assays were performed overnight at RT and then re-
actions were stopped by adding 0.01 volume equivalents of tri-
fluoroacetic acid. After centrifugation, a 1/20 dilution was done
before LC–MS injection by transferring 10 mL volumes of each well
of acidified reaction media in a 96-well daughter microtiterplate
(Figure 1). The controls were prepared as described above replac-
ing enzyme cell lysate by E. coli BL21 empty vector cell lysate.
Spectrophotometric enzymatic assay
One unit (U) of DHA aldolase is defined as the amount of enzyme
that cleaves one micromole of d-F6P to afford d-G3P and DHA per
minute at pH 8 (glycylglycine 50 mm buffer) and 258C. For transal-
dolase activity, glycolaldehyde is added as an acceptor substrate
with F6P as the donor to afford d-G3P and xylulose.
Enzymatic assays were performed with purified enzymes. In each
enzymatic assay, the formation of one product was measured
using different coupled enzymatic systems for which NADH or
NADPH consumption have been monitored at 340 nm (Figure 1).
Aldolase activity in the direction of F6P formation from d,l-G3P
and DHA was measured by monitoring F6P production using the
coupled enzymatic system glucose-6-phosphate isomerase and
glucose-6 phosphate dehydrogenase.[7a] Aldolase activity in the ret-
roaldolisation direction from F6P and transferase activity from F6P
and glycolaldehyde were analysed by monitoring G3P appearance
using the coupled enzymatic system triose phosphate isomerase
and glycerol-3-phosphate dehydrogenase.
Purification of aldolases
Selected enzymes were overexpressed and purified by loading the
clear crude cell extract onto a Ni–NTA column (QIAGEN), according
to the manufacturer’s instructions. The wild-type ALDlisteria was pro-
duced as an enzyme without a His-tag. Purification was conducted
from a 400 mL culture as previously reported[19] except that the
Ni–NTA purification step was replaced by a 25 mL Q-Sepharose
Fast Flow (GE Lifesciences) ion-exchange chromatography. The
enzyme purity was analysed by SDS PAGE. Protein concentrations
were determined by the Bradford method with bovine serum albu-
min as the standard (Bio-Rad) and the samples were stored at
À808C. ALDlisteria was later produced with a C-terminal His-tag (see
Supporting Information, Sections 7 and 8.)
Steady-state kinetic parameters for F6P, dF6P, A5P, d,l-G3P
and DHA
Each kinetic assay (1 mL) was performed in glycylglycine buffer
(50 mm pH 8.0). In the case of F6P, A5P, deoxyfructose-6-phosphate
(dF6P), the substrate cleavage rate was measured by monitoring
d-G3P appearance with glycerol-3-phosphate dehydrogenase
triose phosphate isomerase auxiliary enzymes (10 U) and NADH
(0.36 mm). In the case of DHA and d,l-G3P, the substrate formation
rate was measured by monitoring F6P appearance with glucose-6
phosphate dehydrogenase (8 U)/glucose-6-phosphate isomerase
(13.5 U) auxiliary enzymes and NADP+ (1 mm)[7a] (for more informa-
tion see Supporting Information).
Partially purified samples of non His-tagged ALDlisteria were ob-
tained by heating the cell free extract at 708C for 45 min before
centrifugation. Then the protein was dialysed against 25 mm gly-
cylglycine buffer, pH 8.0.
Enzymatic assay by LC–MS analysis
Stereochemistry
LC–MS–MS analyses were performed on a Hybrid triple quadru-
pole-linear ion trap mass spectrometer (QTRAP 5500 from ABSciex,
Toronto, Canada) equipped with an ESI source and coupled to
a Dionex UltiMate 3000 RS LC system (Thermo Scientific Dionex
Corporation, Sunnyvale, CA, USA). The QTRAP mass spectrometer
was operated in the ESI negative ion mode with the following pa-
rameters: ion source (IS) 4500 V, curtain gas (CUR) 20 a.u., tempera-
ture (TEM) 5008C, gas 1 (GS1) 45 a.u., gas 2 (GS2) 60 a.u., CID
medium. MS–MS experiments were performed by using MRM scan
type.
Reaction: DHAP (0.5 mmol) was added to triose phosphate isomer-
ase (150 U) and HA (3 eq, 1.5 mmol) in water (pH 8.0, 20 mL). Once
the solution homogenised, the reaction was initiated by aldolase
addition (20 U). The reaction proceeded at RT under gentle agita-
tion (100–200 rpm). Every 20 min, 3 more equivalents of HA were
added (total 9 equiv. over 80 min).
Purification protocol
The reaction was stopped by decreasing the pH to 3.0 resulting in
partial enzyme precipitation. The pH was then adjusted to 6.0 and
BaCl2·2H2O (2 equiv.) was added. The solution was centrifuged
15 min at 4000 rpm at 48C and the pellet was discarded. After par-
tial concentration under vacuum, 5 volumes of ethanol were
added. The solution was left to stand overnight at 48C and then
centrifuged. The sugar barium salt was obtained as white powder
after one ethanol rinsing step followed by two others with ace-
tone. The stereochemistry was determined by NMR. All spectra
samples were identical to the literature one (1-deoxy-d-fructose-6-
phosphate)
For the first part of experiments, the ESIMS–MS method for the de-
tection of an hexose-6-phosphate and G3P including two most in-
tense transitions for each compound (monitoring of phosphate
moieties ([H2PO4]À and [PO3]À) was employed (Supporting Informa-
tion, Table 3). For final experiments, the MRM method including
a set of transitions related to the monitoring of phosphate moiet-
ies and sugar cleavage product ions of hexose phosphates was
used (Supporting Information, Table 4).
The reverse-phase chromatography method was used on an ACQ-
UITY UPLC BEN C18 column (1.7 mm, 2.1”150 mm) from Waters
(Milford, USA). The flow rate was 300 mLminÀ1, the sample injection
volume was 10 mL. The column was thermostated at 508C. The
mobile phase A was 10 mm ammonium carbonate in water, and
the mobile phase B was acetonitrile. Linear gradient from 0 to
100% of B was applied over a period of 7 min. The data processing
was performed by using Analyst software (ABSciex).
Acceptor screening
Aldolase (0.07 mg) in glycylglycine buffer solution (50 mm pH 8.0)
was added to a solution (200 mL total reaction volume) containing
HA as donor (50 mm) and an acceptor compound (50 mm) in
ChemCatChem 2015, 7, 1871 – 1879
www.chemcatchem.org
1877
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
water (pH 8.0). The reaction proceeded at RT under gentle agita-
tion (100–200 rpm). The reaction was monitored at regular inter-
vals by TLC (chloroform/methanol 9:1 or 8:2 and ammonium/etha-
nol 5:6) and worked up after 24–48 h depending on the substrate
consumption. The relative conversion rates were determined on
the basis of densitometric TLC monitoring.
Acknowledgements
This research was supported by the French Agence Nationale
pour la Recherche (ANR Genozyme, ANR-08-GENM-026). We
thank Christine Pelle for ALDlisteria purification and Andrew Tolo-
nen for English corrections.
Sequence comparison and phylogenetic analysis
Keywords: biocatalysis · CÀC coupling · enzymes · gene
expression · mass spectrometry
All protein-versus-protein alignments were calculated with gapped
BLASTp and the BLOSUM62 scoring matrix. Phylogenetic analysis
was performed using MAFFT[20] for multiple sequence alignment,
QUICKTREE[21] for tree building and TreeDyn[22] for tree rendering.
All the enzyme UniprotKB identifiers are reported in the tree. The
aldolases used as a reference were taken from Samland et al.[14]
and others (J9MJK9 from Fusarium oxysporum).[23]
[1] a) R. Wohlgemuth, Curr. Opin. Biotechnol. 2010, 21, 713–724; b) S.
Wenda, S. Illner, A. Mell, U. Kragl, Green Chem. 2011, 13, 3007–3047;
c) K. Drauz, H. Grçger, O. May, Enzyme Catalysis in Organic Synthesis,
3rd ed., Wiley-VCH, 2012.
[2] a) T. Davids, M. Schmidt, D. Bottcher, U. Bornscheuer, Curr. Opin. Chem.
Biol. 2013, 17, 215–220; b) U. Bornscheuer, G. Huisman, R. Kazlauskas,
S. Lutz, J. Moore, K. Robins, Nature 2012, 485, 185–194.
[3] a) H. Kries, R. Blomberg, D. Hilvert, Curr. Opin. Chem. Biol. 2013, 17,
221–228; b) L. Giger, S. Caner, R. Obexer, P. Kast, D. Baker, N. Ban, D. Hil-
vert, Nat. Chem. Biol. 2013, 9, 494–498; c) E. Althoff, L. Wang, L. Jiang,
L. Giger, J. Lassila, Z. Wang, M. Smith, S. Hari, P. Kast, D. Herschlag, D.
Hilvert, D. Baker, Protein Sci. 2012, 21, 717–726.
General procedure for preparative synthesis
A sample (1.5 mL, 25 mg of protein) of ALDlisteria aldolase (synthetic
gene, partially purified by heat treatment for 45 min at 708C in
phosphate buffer pH 7.5, 100 mm) was added to a solution
(18.5 mL) containing 130 mg of DHA (70 mm) and a final acceptor
concentration of 50 mm (propanal, d-glyceraldehyde or nitrobuta-
nal) at pH 7.5. After 8 h for propanal and nitrobutanal or 24 h for
d-glyceraldehyde, the reaction was lyophilised. The solid obtained
was suspended in EtOH and then centrifugated. After evaporation
of the supernatant the crude oil was purified by column chroma-
tography on silica gel to provide the expected pure compound. In
the case of propanal the experimental procedure and the work-up
were slightly different and are explained in detail in the corre-
sponding paragraph below.
[4] a) G. Behrens, A. Hummel, S. Padhi, S. Schatzle, U. Bornscheuer, Adv.
Synth. Catal. 2011, 353, 2191–2215; b) M. Guazzaroni, A. Beloqui, P. Go-
lyshin, M. Ferrer, World J. Microbiol. Biotechnol. 2009, 25, 945–954.
[5] a) M. Ferrer, F. Martinez-Abarca, P. Golyshin, Curr. Opin. Biotechnol. 2005,
16, 588–593; b) P. Golyshin, Curr. Opin. Biotechnol. 2005, 16, 585–587.
[6] a) P. ClapØs, W. D. Fessner, Sci. Synth. Stereosel. Synth. 2011, 2, 677–734;
b) P. ClapØs, W. D. Fessner, G. A. Sprenger, A. K. Samland, Curr. Opin.
Chem. Biol. 2010, 14, 154–167; c) P. ClapØs, X. Garrabou, Adv. Synth.
Catal. 2011, 353, 2263–2283; d) M. Brovetto, D. Gamenara, P. Mendez,
G. Seoane, Chem. Rev. 2011, 111, 4346–4403; e) M. Müller, Adv. Synth.
Catal. 2012, 354, 3161–3174; f) K. Fesko, M. Gruber-Khadjawi, Chem-
CatChem 2013, 5, 1248–1272; g) P. Falcicchio, S. Wolterink-Van Loo,
M. C. R. Franssen, J. van der Oost, Extremophiles 2014, 18, 1–13.
[7] a) M. Schürmann, G. A. Sprenger, J. Biol. Chem. 2001, 276, 11055–11061;
b) I. Sµnchez-Moreno, L. Nauton, V. ThØry, A. Pinet, J.-L. Petit, V. de Berar-
dinis, A. K. Samland, C. GuØrard-HØlaine, M. Lemaire, J. Mol. Catal. B
2012, 84, 9–14.
5,6-Dideoxy-d-threo-hexulose: Yield (85%, 182 mg), propanal
(60 mg) was added followed by two other additions of a 60 and
150 mg portion after 2 and 4 h, respectively, until total DHA con-
sumption. The volatile status of this acceptor was exploited and
the column chromatography was not necessary in that case to
[8] a) X. Garrabou, J. A. Castillo, C. GuØrard-HØlaine, T. Parella, J. Joglar, M.
Lemaire, P. ClapØs, Angew. Chem. Int. Ed. 2009, 48, 5521–5525; Angew.
Chem. 2009, 121, 5629–5633; b) M. Sugiyama, Z. Hong, P.-H. Liang,
S. M. Dean, L. J. Whalen, W. A. Greenberg, C.-H. Wong, J. Am. Chem. Soc.
2007, 129, 14811–14817.
obtain
a pure compound. Simple concentration under high
vacuum of the above mentioned EtOH solution afforded the de-
sired aldol. The spectral data are identical to those described in
1
Ref. [16]. H NMR (400 MHz, D2O): d=4.60 (d, J=19.4 Hz, 1H, 1-HA),
[9] a) J. A. Castillo, C. GuØrard-HØlaine, M. Gutierrez, X. Garrabou, M. San-
celme, M. Schürmann, T. Inoue, V. HØlaine, F. Charmantray, T. Gefflaut, L.
Hecquet, J. Joglar, P. ClapØs, G. A. Sprenger, M. Lemaire, Adv. Synth.
Catal. 2010, 352, 1039–1046; b) F. Camps Bres, C. GuØrard-HØlaine, C.
Fernandes, J. Castillo, M. Lemaire, Tetrahedron: Asymmetry 2013, 24,
1075–1081; c) W.-D. Fessner, D. Heyl, M. Rale, Catal. Sci. Technol. 2012,
2, 1596–1601; d) I. Sµnchez-Moreno, V. HØlaine, N. Poupard, F. Charman-
tray, B. LØgeret, L. Hecquet, E. Garcia-Junceda, R. Wohlgemuth, C. GuØr-
ard-HØlaine, M. Lemaire, Adv. Synth. Catal. 2012, 354, 1725–1730; e) M.
Mifsud, A. Szekrenyi, J. Joglar, P. ClapØs, J. Mol. Catal. B 2012, 84, 102–
107; f) C. GuØrard-HØlaine, M. Debacker, P. ClapØs, A. Szekrenyi, V. HØ-
laine, M. Lemaire, Green Chem. 2014, 16, 1109–1113.
[10] a) S. Schneider, T. Sandalova, G. Schneider, G. A. Sprenger, A. K. Samland,
J. Biol. Chem. 2008, 283, 30064–30072; b) S. Schneider, M. Gutierrez, T.
Sandalova, G. Schneider, P. ClapØs, G. A. Sprenger, A. K. Samland, Chem-
BioChem 2010, 11, 681–690; c) X. Garrabou, J. Joglar, T. Parella, R. Cre-
huet, J. Bujons, P. ClapØs, Adv. Synth. Catal. 2011, 353, 89–99.
[11] C. Vergne-Vaxelaire, F. Bordier, A. Fossey, M. Besnard-Gonnet, A. Debard,
A. Mariage, V. Pellouin, A. Perret, J. Petit, M. Stam, M. Salanoubat, J.
Weissenbach, V. de Berardinis, A. Zaparucha, Adv. Synth. Catal. 2013,
355, 1763–1779.
4.48 (d, J=19.3 Hz, 1H, 1-HB), 4.34 (d, J=2.3 Hz, 1H, 3-H), 3.89
(ddd, J=2.3, 6.4, 7.7 Hz, 1H, 4-H), 1.67–1.48 (m, 2H, 5-H), 0.92 ppm
(t, J=7.5 Hz, 3H, 6-H). 13C NMR (100 MHz, D2O): d=213.2 (C2), 77.1
(C3), 73.5 (C4), 65.9 (C1), 25.4 (C5), 9.4 ppm (C6).
d-Fructose: Yield (25%, 45 mg, low yield owing to purification pro-
cess). Column chromatography on silica gel: 90:10 CH2Cl2/MeOH
then 80:20 CH2Cl2/MeOH when DHA was removed. The 1H NMR
spectra is provided in Supporting Information.
(1S,2S,3R,6R)-1-(Hydroxymethyl)-6-nitrocyclohexane-1,2,3-triol:
For a better control of the intramolecular Henry reaction, the enzy-
matic reaction was performed at pH 8.5. yield (70%, 145 mg).
Column chromatography on silica gel: 90:10 CH2Cl2/MeOH. The
1
spectral data are identical to those described in Ref. [9b]. H NMR
(400 MHz, MeOD): d=4.81 (dd, J=3.9, 12.9 Hz, 1H, 6-H), 3.83 (d,
J=11.0 Hz, 1H, 7-H), 3.75 (ddd, J=4.6, 9.4, 11.6 Hz, 1H, 3-H), 3.39
(d, J=9.4 Hz, 1H, 2-H), 3.35 (d, J=11.0 Hz, 1H, 7-H), 2.49–2.36 (m,
1H, 5-Heq), 2.10–1.86 (m, 2H, 5-Heq, 4-Heq), 1.38–1.26 ppm (m, 1H,
4-Hax). 13C NMR (100 MHz, MeOD): d=86.3 (C6), 77.0 (C1), 75.0 (C2),
70.9 (C3), 61.8 (C7), 29.5 (C4), 24.6 ppm (C5).
[12] A. K. Samland, G. A. Sprenger, Int. J. Biochem. Cell Biol. 2009, 41, 1482–
1494.
ChemCatChem 2015, 7, 1871 – 1879
www.chemcatchem.org
1878
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[13] S. Thorell, M. Schürmann, G. A. Sprenger, G. Schneider, J. Mol. Biol.
2002, 319, 161–171.
[14] A. K. Samland, S. Baier, M. Schürmann, T. Inoue, S. Huf, G. Schneider,
G. A. Sprenger, T. Sandalova, FEBS J. 2012, 279, 766–778.
[19] A. Kreimeyer, A. Perret, C. Lechaplais, D. Vallenet, C. Medigue, M. Sala-
noubat, J. Weissenbach, J. Biol. Chem. 2007, 282, 7191–7197.
[20] K. Katoh, K. Misawa, K. Kuma, T. Miyata, Nucleic Acids Res. 2002, 30,
3059–3066.
[15] a) C. Alcantara, L. A. Sarmiento-Rubiano, V. Monedero, J. Deutscher, G.
Perez-Martinez, M. J. Yebra, Appl. Environ. Microbiol. 2008, 74, 5731–
5740; b) M. Tangney, J. K. Brehm, N. P. Minton, W. J. Mitchell, Appl. Envi-
ron. Microbiol. 1998, 64, 1612–1619; c) M. J. Yebra, A. Veyrat, M. A.
Santos, G. Perez-Martinez, J. Bacteriol. 2000, 182, 155–163; d) J. Reizer,
A. Reizer, M. H. Saier, Jr., Microbiology 1995, 141, 961–971.
[16] M. Rale, S. Schneider, G. Sprenger, A. Samland, W. Fessner, Chem. Eur. J.
2011, 17, 2623–2632.
[21] K. Howe, A. Bateman, R. Durbin, Bioinformatics 2002, 18, 1546–1547.
[22] F. Chevenet, C. Brun, A. L. Banuls, B. Jacq, R. Christen, BMC Bioinf. 2006,
7, 439.
[23] E. Kourtoglou, D. Mamma, E. Topakas, P. Christakopoulos, Process Bio-
chem. 2008, 43, 1094–1101.
Received: January 7, 2015
[17] S. Rozen, H. Skaletsky, Methods Mol. Biol. 2000, 132, 365–386.
[18] C. Aslanidis, P. J. de Jong, Nucleic Acids Res. 1990, 18, 6069–6074.
Published online on June 11, 2015
ChemCatChem 2015, 7, 1871 – 1879
www.chemcatchem.org
1879
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim