Broadening deoxysugar glycodiversity: Natural and engineered transaldolases unlock a complementary substrate space

Article abstract of DOI:10.1002/chem.201002942

The majority of prokaryotic drugs are produced in glycosylated form, with the deoxygenation level in the sugar moiety having a profound influence on the drug's bioprofile. Chemical deoxygenation is challenging due to the need for tedious protective group manipulations. For a direct biocatalytic de novo generation of deoxysugars by carboligation, with regiocontrol over deoxygenation sites determined by the choice of enzyme and aldol components, we have investigated the substrate scope of the F178Y mutant of transaldolase B, TalB^F178Y, and fructose 6-phosphate aldolase, FSA, from E. coli against a panel of variously deoxygenated aldehydes and ketones as aldol acceptors and donors, respectively. Independent of substrate structure, both enzymes catalyze a stereospecific carboligation resulting in the D-threo configuration. In combination, these enzymes have allowed the preparation of a total of 22 out of 24 deoxygenated ketose-type products, many of which are inaccessible by available enzymes, from a [3 -8] substrate matrix. Although aliphatic and hydroxylated aliphatic aldehydes were good substrates, D-lactaldehyde was found to be an inhibitor possibly as a consequence of inactive substrate binding to the catalytic Lys residue. A 1-hydroxy-2-alkanone moiety was identified as a common requirement for the donor substrate, whereas propanone and butanone were inactive. For reactions involving dihydroxypropanone, TalB^F178Y proved to be the superior catalyst, whereas for reactions involving 1-hydroxybutanone, FSA is the only choice; for conversions using hydroxypropanone, both TalB^F178Y and FSA are suitable. Structure-guided mutagenesis of Ser176 to Ala in the distant binding pocket of TalB^F178Y, in analogy with the FSA active site, further improved the acceptance of hydroxypropanone. Together, these catalysts are valuable new entries to an expanding toolbox of biocatalytic carboligation and complement each other well in their addressable constitutional space for the stereospecific preparation of deoxysugars.

Full text of DOI:10.1002/chem.201002942

FULL PAPER
DOI: 10.1002/chem.201002942
Broadening Deoxysugar Glycodiversity: Natural and Engineered
Transaldolases Unlock a Complementary Substrate Space
Madhura Rale,[a] Sarah Schneider,[b] Georg A. Sprenger,[b] Anne K. Samland,[b] and
Wolf-Dieter Fessner*[a]
Abstract: The majority of prokaryotic
drugs are produced in glycosylated
form, with the deoxygenation level in
the sugar moiety having a profound in-
fluence on the drugꢀs bioprofile. Chem-
ical deoxygenation is challenging due
to the need for tedious protective
group manipulations. For a direct bio-
catalytic de novo generation of deoxy-
sugars by carboligation, with regiocon-
trol over deoxygenation sites deter-
mined by the choice of enzyme and
aldol components, we have investigated
the substrate scope of the F178Y
structure, both enzymes catalyze a ste-
reospecific carboligation resulting in
the d-threo configuration. In combina-
tion, these enzymes have allowed the
preparation of a total of 22 out of 24
deoxygenated ketose-type products,
many of which are inaccessible by
available enzymes, from a [3ꢁ8] sub-
strate matrix. Although aliphatic and
hydroxylated aliphatic aldehydes were
good substrates, d-lactaldehyde was
found to be an inhibitor possibly as a
consequence of inactive substrate bind-
ing to the catalytic Lys residue. A 1-hy-
droxy-2-alkanone moiety was identified
as a common requirement for the
donor substrate, whereas propanone
and butanone were inactive. For reac-
tions involving dihydroxypropanone,
TalBF178Y proved to be the superior cat-
alyst, whereas for reactions involving 1-
hydroxybutanone, FSA is the only
choice; for conversions using hydroxy-
propanone, both TalBF178Y and FSA are
suitable. Structure-guided mutagenesis
of Ser176 to Ala in the distant binding
pocket of TalBF178Y, in analogy with the
FSA active site, further improved the
acceptance of hydroxypropanone. To-
gether, these catalysts are valuable new
entries to an expanding toolbox of bio-
catalytic carboligation and complement
each other well in their addressable
constitutional space for the stereospe-
cific preparation of deoxysugars.
mutant of transaldolase B, TalBF178Y
,
and fructose 6-phosphate aldolase,
FSA, from E. coli against a panel of
variously deoxygenated aldehydes and
ketones as aldol acceptors and donors,
respectively. Independent of substrate
Keywords: biocatalysis · carbohy-
drates · enzymes · protein engineer-
ing · stereoselectivity
Introduction
conjugation greatly further enhances the accessible chemical
diversity by employing more than 100 different sugars.[2,3]
Most of these sugars are deoxygenated and unknown in eu-
karyotes and the role that they play in many physiological
processes is attributed in part to the enhanced hydrophobici-
ty that they display with respect to the oxygenated ana-
logues. A variety of non-glycosylated therapeutics could be
improved by glycoconjugation, including mitomycin,[4] mor-
phine,[5] ifosfamide mustard,[6] podophyllotoxin,[7] colchi-
cine,[8] rapamycin,[9] or taxol.[10] Natural glycodiversity has
recently inspired the development of both in vitro and in
vivo strategies for manipulating the sugar biosynthetic ma-
chinery with a view to generating novel “glycol-random-
ized” natural product analogues[2,11] in the search for im-
proved pharmaceutical or agricultural drug lead com-
pounds.[12] The biosynthesis of most carbohydrate moieties
found in natural products proceeds via nucleotide-activated
hexose intermediates by different multistep combinations in-
volving oxidation, reduction, deoxygenation, epimerization,
isomerization, group transfer, and rearrangement reactions
before the glycoconjugation step.[2] In particular, the enzy-
matic creation of deoxygenation sites is a most challenging
step.[13] Chemical methods to prepare deoxysugars by total
synthesis[14] or by regiospecific deoxygenation of natural
The majority of prokaryotic small-molecule natural products
interesting for pharmaceutical applications, such as those in
clinical use for the treatment of bacterial or fungal infec-
tions, cancer, and other human diseases, are produced in gly-
cosylated form. Structural changes in the sugar attachment
can have a profound influence on the drugsꢀ bioactivity,
pharmacokinetic properties, and target specificity at the
tissue, cellular, or molecular level.[1] Although natural prod-
ucts by themselves constitute a library of core structures en-
compassing a large and privileged structural diversity, glyco-
[a] Dr. M. Rale, Prof. Dr. W.-D. Fessner
Institut fꢂr Organische Chemie und Biochemie
Technische Universitꢃt Darmstadt
Petersenstrasse 22, 64287 Darmstadt (Germany)
Fax : (+49)6151-16-6636
E-mail: fessner@tu-darmstadt.de
[b] Dr. S. Schneider, Prof. Dr. G. A. Sprenger, Dr. A. K. Samland
Institut fꢂr Mikrobiologie, Universitꢃt Stuttgart
Allmandring 31, 70550 Stuttgart (Germany)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201002942.
Chem. Eur. J. 2011, 17, 2623 – 2632
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2623

monosaccharides[15] usually require tedious multistep conver-
sions due to the need for protective group manipulations.
A more direct biocatalytic generation of deoxysugars is
possible by de novo construction using carboligation en-
zymes with immediate control over sites of deoxygenation
by the choice of suitable enzyme and aldol components.[16]
In particular, dihydroxyacetone phosphate-dependent aldo-
lases have been instrumental in the generation of a plethora
of natural and non-natural sugars[17] and isomerases have
been used to convert ketoses into aldoses, including various
deoxysugars.[18] Despite these numerous examples, limita-
tions to this approach are the often stringent selectivity of
aldolases for their nucleophilic substrate, which limits the
structural diversity accessible to each individual catalyst,
and the typical selectivity for phosphorylated substrates,
which introduces an unsought economic burden. Thus, we
set out a program to investigate novel enzymes from the
transaldolase family as potential catalysts for asymmetric
synthesis with a particular focus on deoxysugar preparation.
Transaldolase is a ubiquitous enzyme in all domains of
life, which, via a Schiff base intermediate, catalyzes the re-
versible transfer of dihydroxyacetone (DHA) among ketose
donors and aldose acceptors in the pentose phosphate path-
way. As a first result of our joint program focusing on the
directed evolution of transaldolase B from E. coli (TalB)[19]
towards novel specificities, very recently some of us report-
ed that replacement of a single amino acid residue, phenyla-
lanine 178, by tyrosine in the active site confers true aldo-
lase, instead of transaldolase, activity to the mutant enzyme
(TalBF178Y; Scheme 1).[20]
the construction of their active sites; all 11 functionally
equivalent amino acid residues in the active site show a sim-
ilar spatial orientation (Figure 1) even if they are not con-
Figure 1. Comparison of the X-ray structures of the active site of FSA
(dark gray; pdb 1L6w)[23] and TalBF178Y (light gray; pdb 3cwn).[20] The
figure was prepared by using PyMOL.[25]
served in sequence.[23,24] However, in the course of this study
we found that engineered TalBF178Y and wild-type FSA have
a rather different, but highly complementary capacity, for
completely stereoselective applications in deoxysugar syn-
thesis. We demonstrate their rather general utility as tools
for asymmetric synthesis by the preparation of a large set of
stereochemically homogeneous deoxysugars from simple
precursors and discuss their individual strengths and limita-
tions with respect to differences in protein structure.
Herein we report on the donor and acceptor substrate
space of this engineered TalBF178Y mutein in comparison
with the related wild-type fructose 6-phosphate aldolase
(FSA) from E. coli, another member of the transaldolase
class.[21] The latter was recently studied for application in the
synthesis of iminocyclitols using azido or Cbz-protected
amino aldehydes.[22] Despite the poor similarity between
their primary protein sequence (18.1% identity, 35.1% simi-
larity), TalBF178Y and FSA show a highly similar 3D structure
at the level of the monomer fold ((b/a)8 barrel) as well as in
Results and Discussion
To evaluate the utility of TalBF178Y as a potential catalyst for
the preparation of deoxysugars, variably deoxygenated
donors (Figure 2) and acceptors (Figure 3) were selected for
small-scale preparative reactions under standard conditions
to gain an insight into the tolerance of structural modifica-
tions and the consequences for the stereoselectivity of car-
Scheme 1. Exemplary metabolic reaction of wild-type transaldolase (TalBwt) and the evolved catalytic activity of TalBF178Y for reversible d-threo-aldol for-
mation shared by wild-type fructose 6-phosphate aldolase (FSA).
2624
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 2623 – 2632

Broadening Deoxysugar Glycodiversity
FULL PAPER
DHA as a donor, TalBF178Y indeed produced a single prod-
uct, as indicated by a new spot on TLC that was subsequent-
ly identified as d-fructose 1a by its typical sets of NMR sig-
nals (Scheme 2). Spectral correlation also ascertained a ste-
Figure 2. Aldol donors considered in this study.
Figure 3. Systematic structural variation of aldol acceptors showing the
stereoconfiguration and degree of functionalization.
boligations. Sufficiently high substrate concentrations
(100 mm acceptor aldehyde, 150 mm donor) were chosen to
compensate for potentially low Km values and reactions
were performed under standard conditions in 50 mm Gly-
Gly buffer at pH 8.5 and 258C. The outcomes were then an-
alyzed in comparison with the corresponding reactions with
FSA as a reference enzyme. The progress of the reaction
was monitored by TLC/HPLC and by in situ NMR (1H, 13C)
experiments in D2O for a period of 24–48 h. The latter anal-
ysis served to provide an unambiguous proof of product
constitution as well as relative configuration. The NMR data
also clearly reflect the relative kinetic differentiation among
substrate classes in the direction of synthesis because the re-
verse aldol cleavage reaction becomes disfavored owing to
the cyclization of primary products to form more stable fur-
anose or pyranose isomers, which thereby withdraws prod-
ucts from the aldol equilibrium (except for substrate C).[16,17]
Finally, reaction mixtures were lyophilized and products pu-
rified by silica gel chromatography for individual characteri-
zation.
Scheme 2. Structural set of deoxysugars accessible by TalBF178Y/FSA cat-
alysis using substrate combinations from Figures 3 and 4. a: R=CH2OH;
b: R=CH3; c: R=C2H5.
To test the suitability of different donor substrates, non-
phosphorylated d-glyceraldehyde ((R)-A) was the most
plausible acceptor because it bears an absolute configuration
identical to the natural substrate d-glyceraldehyde 3-phos-
phate (GA3P). Although the original screening of the TalB
mutant libraries had been performed for their ability to syn-
thesize fructose 6-phosphate (Fru6P) from DHA and
GA3P,[20] the latter was dismissed as a potential reference
component for judging synthetic applications because of its
inherent instability towards decomposition.[26] By using
reoselective 3S,4R aldol addition as NMR analysis showed
no indication of the formation of other diastereomers (>
95% de; see the Supporting Information). The stereoselec-
tivity of the addition was further confirmed by an enzyme-
coupled photometric assay for the formation of d-fruc-
tose.[27] The relative kinetic parameters were determined by
using DHA as donor and dl-GA or the enantiopure d-GA
as acceptor (Table 1). For the donor substrate DHA,
TalBF178Y and FSA showed similar kinetic properties and cat-
Chem. Eur. J. 2011, 17, 2623 – 2632
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2625

W.-D. Fessner et al.
Table 1. Comparison of the steady-state kinetics for FSA and TalBF178Y using donor substrate DHA and ac-
ceptor substrates glyceraldehyde 3-phosphate and glyceraldehyde to form d-fructose 6-phosphate and d-fruc-
tose, respectively.[a]
leads to rapid inactivation.
Identical reactions, but run with
FSA instead for comparison, re-
vealed a distinctly different re-
activity pattern.[21b] The FSA-
catalyzed reaction with DHA
showed only a very low conver-
sion to 1a, which corroborates
earlier findings that DHA is a
poor substrate for this enzyme
with (R)-A as an acceptor[21b,22a]
but stands in clear contrast to
the reported equal efficiency of
Kinetic parameters[b]
DHA+dl-GA3PÐFru6P
DHA+dl-GAÐd-Fru[c]
FSA
TalBF178Y
TalBF178Y
G
TalBF178Y
ACTHNGUTER(NNUG d-GA)
Vmax [Umgꢀ1
Km [mm]
]
31
1.9
20
2.4
0.87
>120
0.54
0.86
59
0.53
9.0
[d]
k
catA
]
13
12
kcat/Km [mꢀ1 sꢀ1
]
6.6ꢁ103
5.0ꢁ103
<4.3
[a] Data are average values of two independent measurements. [b] The enzyme-coupled assays employed only
detect the specific formation of d-fructose 6-phosphate and d-fructose, respectively. [c] Ref. [27]. [d] kcat was
calculated as turnover number per active site, that is, monomeric subunit.
alytic efficiencies.[20] Although the TalB mutant showed
good reactivity with both sources of glyceraldehyde (dl-
and d-GA), no activity was observed by using FSA under
all three donors DHA, HA, and HB with GA3P as the ac-
ceptor.[22b] On the other hand, the FSA-catalyzed reaction of
(R)-A with HA to yield 1b was more rapid than that using
TalBF178Y, and notably, with HB it gave a modest conversion
to 1c as a single product whereas TalBF178Y gave none. Nei-
ther catalyst is able to accept propanone as a nucleophile.
By using wild-type transaldolase, as expected no product
formation could be detected from any of these non-natural
substrate combinations.
the
conditions
of
the
spectrophotometric
assay
(<0.05 Umgꢀ1). This is in agreement with previous reports
that d-GA is a very weak substrate for FSA, especially in
combination with DHA as donor.[21b,28a] For TalBF178Y, sub-
strate inhibition was only observed at rather high d-GA
concentrations (>100 mm). In comparison, by using the
phosphorylated acceptor dl-GA3P, both FSA and TalBF178Y
showed similar Vmax and Km values of 31 and 20 Umgꢀ1 and
1.9 and 2.4 mm, respectively (Table 1). This resulted in simi-
lar catalytic efficiencies (kcat/Km) of 6600 and 5000mꢀ1 sꢀ1, re-
spectively. At concentrations >4 mm, strong substrate inhib-
ition was observed with FSA, but not with TalBF178Y within
the range of dl-GA3P concentrations studied. Although the
data demonstrates that the phosphorylated substrate is
clearly preferred by TalBF178Y because the catalytic efficiency
for dl-GA3P is three orders of magnitude larger than for
dl-GA, it is of practical significance that TalBF178Y shows
considerable activity at a Vmax of around 0.9 Umgꢀ1, even
with the non-phosphorylated substrate.
To elucidate a possible stereochemical substrate discrimi-
nation, the relative reaction rates were determined with (S)-
A (l-glyceraldehyde) under otherwise identical conditions.
TalBF178Y-catalyzed addition of DHA led to a rapid conver-
sion with l-sorbose 2a identified as the only product by
TLC and NMR analysis. Initially, the conversion of (S)-A
seemed to proceed more rapidly than that of the enantiomer
(R)-A according to TLC analysis. However, a competitive
experiment using rac-A (dl-glyceraldehyde) under kinetical-
ly controlled conditions with in situ 1H NMR monitoring
identified a 5:2 ratio of 1a/2a, which clearly demonstrates a
preference for (R)-A but indicates that TLC cannot be used
as the only method of product analysis because of potential
errors arising from unequal staining intensity. An unexpect-
ed but true preference for (S)-A was exhibited with HA as
the donor although the conversion to the corresponding 1-
deoxy-l-sorbose 2b was less rapid than for DHA. Comple-
mentary FSA-catalyzed reactions with acceptor (S)-A
showed the same overall trend of donor qualities as with
(R)-A. Again, DHA proved a weak donor for FSA leading
to slow conversion to 2a, whereas HA and HB proved to be
much superior donors with completely regio- and stereose-
lective conversions to 2b and the higher homologue 2c, re-
spectively. FSA also clearly displayed a preference for the
2S-configured acceptor.
Despite a recent report stating hydroxyacetone (HA) to
be unacceptable to TalBF178Y on account of a negative HPLC
analysis,[20] we envisaged chances to bring about conversion
under appropriate reaction conditions as the molecular size
is largely comparable to that of DHA as the perceived “nat-
ural” substrate. Remarkably, with (R)-A as an acceptor, the
mutant displayed moderate tolerance for the non-natural
deoxygenated HA substrate reaching >50% conversion
(TLC analysis) within the time window of the analysis. Ac-
cording to the NMR analysis, the reaction yielded the 1-
deoxy analogue 1b exclusively, which points to a unique
binding orientation of HA and regiospecific deprotonation
towards a hydroxyenamine as the activated nucleophile. No
reaction took place in the presence of non-hydroxylated
propanone, however, which indicates that the hydroxymeth-
yl unit is an indispensable structural precondition for the
aldol donor component. By using 1-hydroxybutanone (HB)
as a potential chain-extended nucleophile, the protein pre-
cipitated at donor concentrations as low as 50 mm with no
detectable product formation. Thus, by using (R)-A as the
acceptor, TalBF178Y shows a clear donor preference for the
“natural” substrate DHA followed by HA, whereas HB
By using the 2-deoxy analogue of glyceraldehyde B, elimi-
nating configurational predilection, TalBF178Y demonstrated
excellent conversion of DHA to afford 5-deoxy-d-fructose
3a as a single product in an isolated yield of 83%. As the
acceptor lacks a chiral reference center, the absolute config-
uration was verified by optical rotation ([a]2D5 =ꢀ60.7 (c=
1.04, H2O); lit.:[18a] [a]D25 =ꢀ64.6 (c=4.3, H2O)). Changing
the donor to HA also showed good conversion within 24 h
(ca. 66% by NMR spectroscopy) to furnish 1,5-dideoxyhex-
ulose 3b as the sole product. For comparison, a remarkably
2626
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 2623 – 2632

Broadening Deoxysugar Glycodiversity
FULL PAPER
good reaction was also observed in the FSA-catalyzed addi-
tion of DHA to B, whereas reactions using HA and HB
donors again proceeded smoothly with near quantitative
conversion to provide deoxysugars 3b and 3c, respectively.
In the presence of DHA, propanal (C; “2,3-dideoxyglycer-
aldehyde”) was converted by TalBF178Y to a moderate extent
furnishing the corresponding open-chain “hexulose” 4a.
Even in combination with HA this substrate was well toler-
ated, producing the dihydroxyketone 4b. By using C, FSA
again catalyzed the addition of DHA fairly well and con-
verted HA and HB rapidly to provide 4b and 4c, respec-
tively.
Scheme 3. Reactions of acceptor F, also acting as a competing nucleo-
phile. a: R=CH2OH; b: R=CH3; c: R=C2H5.
Enantiomers of lactaldehyde (“3-deoxyglyceraldehyde”)
(R)-D and (S)-D were expected to provide more insight into
the capacity of the enzyme catalysts for kinetic enantiomer
differentiation. By using (R)-D, TalBF178Y catalyzed a rather
slow conversion with DHA to yield 6-deoxyhexulose 5a as a
single product. Furthermore, no reaction was observed on
changing the donor to HA. This behavior was rather per-
plexing considering the structural relationship with acceptor
(R)-A, which has an identical configuration at C-2, and with
acceptor C, which has an identical degree of deoxygenation
at C-3, both of which were better substrates than (R)-D. In-
terestingly, FSA showed no activity at all in the reactions of
(R)-D with each of the three donors. Indeed, (R)-D was
identified as a strong inhibitor of FSA activity when the
latter was assayed in the cleavage of Fru6P (see the Sup-
porting Information).
In contrast, the outcome with the enantiomeric (S)-D was
quite different as TalBF178Y catalyzed a rapid conversion
with DHA to the expected hexulose 6a and showed a mod-
erate rate with HA to afford 6b. FSA acted as anticipated,
yielding a less productive reaction with DHA and speedy
conversions to products 6b and 6c with HA and HB, respec-
tively. Thus, in comparison with the reactions with fully hy-
droxylated A, with 3-deoxygenated acceptors D not only
were the donor preferences of TalBF178Y and FSA more pro-
nounced, but also the preference for the S enantiomer.
The TalBF178Y-catalyzed reaction with DHA and com-
pound E, the symmetrical analogue of D lacking an element
of chirality, proceeded well to furnish a single product 7a,
whereas with HA the sugar 7b was generated at only a
modest rate. Similarly, FSA catalysis was moderately pro-
ductive with HA and HB, providing deoxy- and dideoxysu-
gars 7b and 7c, respectively, whereas with DHA only a fair
reaction occurred. Thus, with both enzymes acceptor E be-
haved similarly to the S enantiomer of D, although it was
slightly attenuated. Apparently, steric hindrance directly ad-
jacent to the reactive carbonyl group of the acceptor compo-
nent seems to be tolerated rather well by both catalysts.
It was speculated that the structurally most simple accept-
or, 2-hydroxyaldehyde F, offers a borderline case with the
potential of acting both as a good electrophile as well as a
potential donor.[28a] By using TalBF178Y in the presence of
DHA, F was converted into d-xylulose 8a as the major
product. However, TLC analysis as well as distinct signals in
the NMR experiments indicated the presence of a second
component that consequently was established to be d-
threose stemming from the homoaldolization of
F
(Scheme 3). This novel side-reaction has recently been also
observed by other authors for FSA[28] and its occurrence in
the TalBF178Y-catalyzed reaction reflects the similar catalytic
environment of these two catalysts. However, the self-aldoli-
zation was only a minor activity (<10% relative to cross-
aldol product 8a) in comparison with that observed with
FSA for which d-threose formation was by far the predomi-
nant pathway. Considering the lower reactivity of TalBF178Y
with HA compared with DHA, a more pronounced compe-
tition was to be expected for the donor HA towards F
(cross- and self-aldolization) and indeed observed, although
modest conversion into the cross-aldol product 8b was also
attained. In line with the observations of others,[28] FSA-cat-
alyzed reactions of acceptor F with DHA showed the pres-
ence of large amounts of d-threose and practically no con-
version to the cross-aldol product 8a. In contrast, from the
FSA-catalyzed reactions with HA and HB, the major prod-
ucts were the cross-aldol products 8b and 8c, respectively,
which were accompanied by only modest amounts of d-
threose.
From the entire set of reactions discussed above, each
component of the [3ꢁ8] product library (Scheme 2, Table 2)
could be isolated generally in good-to-high yields after a suf-
ficiently extended reaction time. This outcome is facilitated
by the fact that the enzymatic carboligation equilibrium gen-
erally favors product formation due to the ensuing cycliza-
tion of primary structures to more stable cyclic furanose or
pyranose forms (Scheme 2), which withdraw the products
from the retro-aldol equilibrium.[16] For components involv-
ing DHA, TalBF178Y proved to be clearly superior, whereas
for components involving HB, FSA is the only choice. For
conversions based on the component HA, both TalBF178Y
and FSA are suitable with a subtle preference for each de-
pending on particular combinations (Figure 4). Thus, the
two catalysts complement each other well to address a large
constitutional space of regiospecifically deoxygenated sugars
that in part (specifically, all compounds in series b and c)
were not accessible with available enzymes before.[16,17] The
only exceptions (5b,c) observed are caused by the low reac-
tivity of (R)-D (d-lactaldehyde) with both biocatalysts. De-
Chem. Eur. J. 2011, 17, 2623 – 2632
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2627

W.-D. Fessner et al.
Hydroxybutanone (HB)
Table 2. Classification of substrate quality.
Electrophile
Dihydroxyacetone (DHA)
Hydroxyacetone (HA)
TalBwt
TalBF178Y
FSA
Product
TalBF178Y
FSA
Product
TalBF178Y
FSA
Product
d-glyceraldehyde ((R)-A)
l-glyceraldehyde ((S)-A)
3-hydroxypropanal (B)
propanal (C)
d-lactaldehyde ((R)-D)
l-lactaldehyde ((S)-D)
2-methyl-2-hydroxypropanal (E)
glycolaldehyde (F) [b]
–
–
–
–
–
–
–
–
+++
+++
+++
++
+
+
1a
2a
3a
4a
5a
6a
7a
8a
++
+++
++
++
–
+++
+++
+++
+++
–
+++
++
+++
1b
2b
3b
4b
5b
6b
7b
8b
–
–
–
–
–
–
–
–
+++
+++
+++
++
1c
2c
3c
4c
5c
6c
7c
8c
++
++
–
++
–
+++
+++
++
++
+
++
+
+
++
++
+
+
[a] Substrates are indicated as good (+++; conversion >75%), moderate (++; 25–75%), fair (+; <25%), or poor (–; no product detectable by TLC).
Data are based on TLC analysis after 1 h of reaction time (100 mm acceptor, 150 mm donor, 20 mg FSA or 10 mg TalBF178Y; glycyl-glycine buffer 50 mm
at pH 8.5). [b] All reactions involving F are accompanied by d-threose formation from self-aldolization.
which hydroxylation at the 2- or 3-position facilitates the
correct binding of the aldehyde electrophile in the active
site.[16,31]
It became evident from X-ray studies of DHAP aldolases
that the absolute stereospecificity of the configuration at the
nucleophilic carbon usually is fully controlled by the enzyme
as stereospecific deprotonation is assisted by hydrogen
bonding of the hydroxy group to give a cis or trans enedio-
late (or hydroxyenamine equivalent) with spatial protection
of one prochiral hemisphere of the “carbanion” equivalent
from the protein backbone. Diastereoselectivity results from
two-fold hydrogen bonding of the aldehyde carbonyl by
active-site residues to activate and correctly position the
electrophilic center for stereoselective attack by the enolate
Figure 4. Three-dimensional illustration of the relative rates of product
(or enamine) nucleophile with retention of its configura-
formation (y axis) upon catalysis by TalBF178Y (dark bars) or FSA (light
tion.[31] The 5:2 kinetic preference for R-configured alde-
bars) for the [3ꢁ8] substrate array of structurally related aldol donors (x
hyde A is in line with the configuration of the natural d-
GA3P substrate. However, the inverse preference for (S)-D
seems to be peculiar. The latter may result from an over-
riding inhibition by the R antipode, as observed in the in-
ability to form adducts 5b,c, both for TalBF178Y or FSA cata-
lysts. Contrary to the situation with A, the nonpolar end of
the acceptor may lead to a nonproductive binding mode,
possibly by covalent binding to the catalytic lysine 132 of
TalB through carbinolamine (or imine) formation, which
would competitively interfere with the regular mode of cat-
alysis. Support for the latter option may be seen in the ap-
parent formation of an aldehyde-derived carbinolamine
ligand during crystallization for X-ray structure determina-
tion of FSA.[23]
axis) and acceptors (z axis).
spite the deliberate restriction of this study to small alde-
hydes, it is apparent that for both catalysts the scope of ac-
ceptor substrates is rather wide, in common with other aldo-
lases, and even tolerates steric bulk directly adjacent to the
reactive carbonyl group. It ought to be mentioned that the
preparation of 3a is the first efficient one-step preparation
of 5-deoxy-d-fructose, which is discussed as a potential arti-
ficial sweetener,[18a,29] from inexpensive starting materials.
Both catalysts deliver aldol adducts having a common d-
threo configuration, identical to that of fructose 1,6-bisphos-
phate aldolase (FruA).[30] According to the NMR analysis of
the reaction monitored in situ and of the crude product mix-
tures, all conversions proceeded with excellent stereoselec-
tivity. Products 1a–c, 2a–c, and 6a–c from enantiomerically
pure 2-hydroxyaldehydes (R)-A, (S)-A, and (S)-D allow an
immediate determination of the diastereoselectivity due to
the internal reference of the specific absolute configuration
introduced by the aldehyde chiral center; the formation of a
single stereoisomer indicates complete selectivity in the
asymmetric carboligation step. Similarly, from the 2-hydroxy-
aldehydes E and F, single products 7a–c and 8b,c were ob-
tained and the 3-hydroxyaldehyde B yielded single diaste-
reomers 3a–c. This correlates with numerous observations
using FruA and other DHAP-dependent aldolases, for
Most remarkable, however, are the TalBF178Y- and FSA-
catalyzed conversions involving C in the generation of single
diastereomers 4a–c for three reasons: 1) Such substrate
combinations are challenging for aldolases, particularly for
transaldolase descendants designed by nature for the con-
version of highly polar phosphorylated and polyhydroxylat-
ed substrates because more generic structures with rather
low oxygenation levels possess a decreased aldol acceptor
electrophilicity and lack opportunities for hydrogen bonding
other than to aldehyde carbonyl; 2) these are the first exam-
ples of non-carbohydrate products with this low-level substi-
tution pattern produced by an aldolase;[16,17,32] 3) aliphatic
a,b-dihydroxy ketones of the type 4b,c with a strictly de-
2628
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 2623 – 2632

Broadening Deoxysugar Glycodiversity
FULL PAPER
Figure 5. Comparison of the X-ray structures of the active site of wild-type TalB (B, pdb 1onr)[24a] and FSA (C, pdb 1L6w).[23] For both enzymes the bind-
ing pocket corresponding to the C1 carbon of the Schiff base intermediate is shown. Except for the Schiff base forming lysine, the residues and active
site surfaces are stained according to their polarity profile (polar or charged residues in red, nonpolar residues in yellow). For comparison, the reduced
Schiff base intermediate with Lys132 is shown for TalBwt (A, pdb 1ucw).[24b] The figure was prepared by using PyMOL.[25]
fined absolute and relative syn configuration are currently
not accessible with similar efficiency by enantioselective
chemical catalysis.[33] For known FruA enzymes, for compar-
ison, diastereoselectivity with propanal amounts to only
96% de with the FruA from rabbit muscle or Staphylococ-
cus carnosus.[30]
With increasing deoxygenation increments in the substrate
components, selectivity for donor type and acceptor enantio-
meric configuration becomes more pronounced when com-
pared with the fully hydroxylated combination, which seems
to point to a more compact transition state and thus higher
energetic discrimination of the respective reaction pathways
of deoxygenated compounds. This effect is potentially
caused by the reduced hydrogen-bonding interactions avail-
able to substrates with a lower number of hydroxy functions,
both intramolecularly as well as to protein residues
(Figure 5).
Indeed, a comparison of the residues that surround the
C1 carbon of the glyceryl moiety in the Schiff base inter-
mediate within a distance of 5 ꢅ in the two enzymes, FSA
and wild-type TalB,[23,24] reveals that some of the polar resi-
dues in the active site of TalB are replaced by nonpolar resi-
dues in FSA (Table 3). This includes the residues Asn154
and Ser176 in TalB, which form a hydrogen bond to the hy-
droxy group at C1. These residues are replaced by the ali-
phatic side-chains Leu107 and Ala129 in FSA. The conse-
quences become even more apparent with a glance at the
structures (Figure 5). The surface of the binding pocket ac-
commodating the C1 moiety in TalB is mainly stained in
red, which corresponds to polar and charged residues form-
ing this cleft. In contrast, the binding site in FSA is predomi-
nantly stained in yellow and indicates a hydrophobic surface
formed from nonpolar residues. These differences in polarity
and the absence of hydrogen-bond donors might explain
why FSA 1) prefers HA with a nonpolar methyl group at C1
over DHA as donor substrate and 2) how this preference
may contribute to the regiospecificity of the aldol reaction
with HA (and HB) to produce 1-deoxysugars exclusively.
On the other hand, DHA can form more interactions with
the active site of TalB than HA. Therefore, the binding of
DHA is preferred over HA in the case of TalB (or its
mutant).
As a control experiment to verify this interpretation of
nucleophilic substrate recognition sites, the polar residue
Ser176 of TalBF178Y, involved in the binding of the terminal
CH2OH function of the donor, was replaced by a nonpolar
Ala residue in a step towards mimicking the more hydro-
phobic binding environment of FSA (Table 3). The relative
donor preference was evaluated in a competition experi-
ment by using equal concentrations of both DHA and HA
nucleophiles together with 3-hydroxypropionaldehyde (B),
selected for its lack of configurational complications and for
the pyranoid cyclization mode that stabilizes the derived
products. Indeed, the double mutant TalBF178Y,S176A showed
practically identical conversion of HA to 1,5-dideoxy-d-
threo-hexulose 3b in comparison with the DHA-derived 5-
deoxy-d-threo-hexulose 3a, according to an NMR analysis
(ratio 3a/3b=1:1). Under identical kinetically controlled
conditions, the single mutant TalBF178Y almost exclusively
produced 3a, whereas FSA catalysis exclusively gave 3b
(see the Supporting Information). This finding complements
well an inverse mutagenesis experiment with FSA in which
FSAA129S shows improved activity with the more polar DHA
donor.[28b]
Table 3. Residues located within 5 ꢅ distance of the glyceryl C1 carbon
in the reduced Schiff base intermediate of TalB (pdb 1ucw)[24] and the
corresponding residues in FSA (pdb 1L6w).[23] Conserved residues are
shown in italics.
TalB Thr33 Ser94 Asn154 Thr156 Ser176 Met223 Ala225 Thr243
FSA Thr26 Phe57 Leu107 Thr109 Ala129 Leu163 Ala165 Thr185
Chem. Eur. J. 2011, 17, 2623 – 2632
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2629

W.-D. Fessner et al.
In comparison with the productive binding of HA by both
of these TalB muteins, their behavior towards the higher ho-
mologous HB remains puzzling. Although TalBF178Y led to
rapid precipitation upon exposure to HB concentrations
even as low as 10 mm, the double mutant TalBF178Y,S176A
seemed to be more stable but still precipitated at concentra-
tions >50 mm HB. Under no reaction conditions was prod-
uct formation (3c) detectable. The limitations for the use of
HB as a donor substrate seem to be rather particular to the
structure of the compound. The bulk of the hydrophobic
moiety alone cannot be the imposing factor because propa-
nal (C) is a fairly good substrate that is tolerated at much
higher concentrations. As a first hypothesis, the extra meth-
ylene group in HB may not be adaptable in the active site
without causing major reorientation of active-site residues
thereby causing a destabilization of the overall fold and/or
subunit aggregation state. Replacement of Ser176 by Ala
seems to ameliorate the effect but without allowing com-
plete adaptation. Possibly, the larger size of Met223 relative
to Leu163 in FSA is another crucial factor to be considered
for further mutagenesis experiments.
The fact that propanone (acetone) is inappropriate for
both TalBF178Y and FSA as a nucleophile even at high sub-
strate concentration indicates that the hydroxyacetyl portion
in the donor is an absolute structural requirement in either
or both Schiff base formation and nucleophile generation.
Homoaldol formation from glycolaldehyde, observed for
both catalysts, albeit at different relative rates, is proof that
this aldehyde can bind and act similarly as a donor, in struc-
tural analogy to the hydroxymethyl ketone donors. The very
poor donor efficiency of TalBF178Y towards glycolaldehyde
relative to the distinct potency of FSA, which results in an
incompletely filled void in the donor pocket upon binding of
an aldehyde instead of a ketone group, may be interpreted
as a higher adaptive flexibility of the FSA active site.
Indeed, the efficiency of glycolaldehyde binding seems to
correlate with the more pronounced inhibition (nonproduc-
tive Schiff base formation) observed with d-lactaldehyde, as
discussed above.
Conclusion
We have studied two enzymes, TalBF178Y and FSA, as new
entries to an expanding toolbox of biocatalytic carboligation
and demonstrated that these catalysts are useful and reliable
for preparative applications similar to the “classic” fructose
1,6-bisphosphate aldolase (FruA) but without the need for a
costly phosphorylated nucleophile. The catalysts show an in-
teresting complementary tolerance for the aldol donor com-
ponent with TalBF178Y preferring the fully hydroxylated
DHA nucleophile and FSA the more deoxygenated nucleo-
phile, and even the elongated aliphatic HB that is unaccept-
able toTalBF178Y. Although FSA shows a somewhat broader
substrate tolerance, utilizing HB as efficiently as HA, its re-
actions involving DHA are sluggish and less productive; this
catalyst prefers a higher degree of deoxygenation in its
donor as well as acceptor substrates. TalBF178Y is able to use
glycolaldehyde as an aldol donor but, in comparison with
FSA, only as a minor side-reaction to the use of this com-
pound as an aldol electrophile. Factors governing the dis-
criminate substrate selectivity have been elucidated by site-
specific mutagenesis of the polar Ser176 residue in TalBF178Y
to approach an FSA-like active-site composition.
In combination, these enzymes have allowed the prepara-
tion of a total of 22 out of 24 structures, many of which are
unique for enzymatic carboligation, from a [3ꢁ8] substrate
matrix of deoxygenated ketose-type products with a specific
d-threo or syn-3S,4R configuration (4S,5R in case of the HB
nucleophile) by using variously modified aldehydes. The
only two omissions in the matrix concern the special case of
d-lactaldehyde, which seems to pose a specific problem of
inactive substrate binding. Further studies to adapt the sub-
strate tolerance of the enzymes for alternative donors and
acceptors, and towards further preparative applications are
in progress. The results will be communicated in due course.
Experimental Section
On the other hand, d-GA and other non-phosphorylated
aldol acceptors containing varying degrees and location of
deoxygenation sites have been demonstrated to be good
substrates, despite the fact that the metabolic function of
the enzymes targets fully oxygenated and phosphorylated
sugar substrates. Thus, an efficient one-step preparation of
5-deoxyfructose, which is of interest as a potential artificial
sweetener,[29] was achieved starting from inexpensive start-
ing materials. Although this study was restricted to small
C2–C3 aldehydes for the sake of limiting the substrate library
to a manageable size, we expect that larger structures will
behave similarly, as is well documented for DHAP aldolases
and related enzymes.[16,17] Significantly, the two catalysts
even seem to tolerate a-branching in the acceptor compo-
nent rather well, causing steric hindrance directly adjacent
to the reactive carbonyl group. This and other aspects need
to be further investigated.
Enzyme sources: Fructose 6-phosphate aldolase (FSA; 3.2 Umgꢀ1 pro-
tein) and transaldolase B mutant F178Y (TalBF178Y; 0.32 Umgꢀ1 protein)
were prepared as lyophilized powders according to published proced-
AHCTUNGTRENNUNG
ures.[20,21a] For both biocatalysts, activity was assayed as the cleavage rate
of Fru6P with the formation of GA3P being monitored by coupled enzy-
matic consumption of NADH;[21a] the protein was quantified by the Brad-
ford assay.[34] One unit of FSA or TalBF178Y is defined as the amount of
protein that will cleave 1 mmol of Fru6P to afford d-GA3P and DHA per
minute at 258C and pH 8.5 (glycyl-glycine buffer, 50 mm).
Determination of kinetic constants: The Km value for dl-GA3P was de-
termined within a concentration range of 0.14–11.2 mm (FSA) and 0.28–
11.2 mm (TalBF178Y) at 308C in 50 mm glycyl-glycine buffer (pH 8.5) con-
taining 1 mm dithiothreitol (DTT). The formation of Fru6P was detected
as described previously[20,21a] using the coupling enzymes phosphoglucose
isomerase and glucose 6-phosphate dehydrogenase (activities were moni-
tored at 340 nm for 10 min). The concentration of DHA was kept con-
stant at 300 mm (FSA) or at 150 mm (TalBF178Y). The concentration of
DHA was saturating as the Km value is 62 mm for FSA and 30 mm for
TalBF178Y [20]
The specific activity was plotted against the concentration of
.
dl-GA3P. As a result of the strong substrate inhibition observed in some
reactions the kinetic constants were calculated from Hanes plots[35] by
2630
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 2623 – 2632

Broadening Deoxysugar Glycodiversity
FULL PAPER
using SigmaPlot 9.0. The formation of d-fructose from DHA (150 mm)
and dl-GA or d-GA was followed by analyzing the formation of
NADPH (0.5 mm NADP+) in 50 mm glycyl-glycine buffer (pH 8.5) con-
taining 1 mm DTT by using the coupling enzymes fructokinase (0.5 U)
from Zymomonas mobilis,[36] phosphoglucose isomerase (0.5 U), and glu-
cose 6-phosphate dehydrogenase (0.5 U). ATP (1.2 mm) and MgCl2
(10 mm) were added.[27] The Km value for dl-GA (or d-GA) was deter-
mined by using aldehyde concentrations ranging from 2 to 140 mm for
TalBF178Y. The concentration of DHA was kept constant at 150 mm.
[1] a) A. C. Weymouth-Wilson, Nat. Prod. Rep. 1997, 14, 99–110; b) T.
Vogt, P. Jones, Trends Plant Sci. 2000, 5, 380–386; c) K. M. Koeller,
C.-H. Wong, Nat. Biotechnol. 2000, 18, 835–841; d) J. S. Thorson,
T. J. Hosted, J. Jiang, J. B. Biggins, J. Ahlert, Curr. Org. Chem. 2001,
5, 139–167; e) V. Kren, L. Martꢆnkovꢇ, Curr. Med. Chem. 2001, 8,
1303–1328; f) P. Jones, T. Vogt, Planta 2001, 213, 164–174; g) K.
Keegstra, N. Raikhel, Curr. Opin. Plant Biol. 2001, 4, 219–224;
h) J. S. Thorson, T. Vogt in Carbohydrate-Based Drug Discovery
(Ed.: C.-H. Wong), Wiley-VCH, Weinheim, 2002, pp. 685–712; i) C.-
H. Wong, M. C. Bryan, P. T. Nyffeler, H. Liu, E. Chapman, Pure
Appl. Chem. 2003, 75, 179–186; j) M. S. Butler, J. Nat. Prod. 2004,
67, 2141–2153; k) H. Cheng, X. H. Cao, M. Xian, L. Y. Fang, T. B.
Cai, J. J. Ji, J. B. Tunac, D. X. Sun, P. G. Wang, J. Med. Chem. 2005,
Site-directed mutagenesis: Site-directed mutagenesis for the replacement
of Ser176 with Ala was carried out by using a modified version of the
QuikChange protocol. In this case the plasmid pJF119talBF178Y[20] was
used as template and the PfuUltra DNA Polymerase (Stratagene) for
amplification. The following primers were used: Forward 5’-
ggcgtgttcctgatcgcgccgtacgttggc-3’, reverse 5’-gccaacgtacggcgcgatcaggaa-
cacgcc-3ꢀ (mutated codon is underlined). The correct sequence was con-
firmed by custom sequencing (GATC, Konstanz, Germany).
ˇ
ˇ
48, 645–652; l) V. Kren, T. Rezanka, FEMS Microbiol. Rev. 2008,
32, 858–889.
[2] C. J. Thibodeaux, C. E. Melancon, H.-w. Liu, Angew. Chem. Int. Ed.
2008, 47, 9814–9859.
General procedures for substrate evaluation
[3] M. M. K. Boysen, Chem. Eur. J. 2007, 13, 8648–8659.
[4] A. Ghiorghis, A. Talebian, R. Clarke, Cancer Chemother. Pharma-
col. 1992, 29, 290–296.
[5] a) R. T. Brown, N. E. Carter, F. Scheinmann, N. J. Turner, Tetrahe-
dron Lett. 1995, 36, 1117–1120; b) G. N. Jenkins, A. V. Stachulski, F.
Scheinmann, N. J. Turner, Tetrahedron: Asymmetry 2000, 11, 413–
416.
[6] a) J. Pohl, B. Bertram, P. Hilgard, M. R. Nowrousian, J. Stꢂben, M.
Wießler, Cancer Chemother. Pharmacol. 1995, 35, 364–370; b) M.
Veyhl, K. Wagner, C. Volk, V. Gorboulev, K. Baumgarten, W. M.
Weber, M. Schaper, B. Bertram, M. Wiessler, H. Koepsell, Proc.
Natl. Acad. Sci. USA 1998, 95, 2914–2919.
In situ NMR experiments: All component solutions were prepared in
D2O. Solutions of donor and aldehyde acceptor components at final con-
centrations of 150 and 100 mm, respectively, in a total volume of 1 mL
were added to a solution of the aldolase (20 mg FSA or 10 mg TalBF178Y
as lyophilized powder) in glycyl-glycine buffer (50 mm, pH 8.5). From this
mixture, a sample of 700 mL was withdrawn and carefully transferred to
an NMR tube. The progress of the reaction progress was monitored by
recording H NMR spectra (500 MHz) after 0, 1, 12, and 24 h of reaction
time (48 h for slow reactions). The NMR experiments were analyzed for
proof of stereochemical preference as well as effective conversion.
To assess the donor preferences of FSA and mutants TalBF178Y and
TalBF178Y,S176A, competition experiments (in D2O) were performed as fol-
lows. Equal stoichiometric quantities of DHA, HA, and acceptor alde-
hyde (each at 100 mm final concentration) in a total volume of 1 mL
were added to the aldolase solution (700 mL; 20 mg FSA; 10 mg
TalBF178Y; 10 mg TalBF178Y,S176A) in glycyl-glycine buffer (50 mm, pH 8.5).
The reactions were followed by recording 1H NMR spectra (500 MHz)
after 0, 1, 6, 12, and 24 h.
1
[7] T. F. Imbert, Biochimie 1998, 80, 207–222.
[8] A. Ahmed, N. R. Peters, M. K. Fitzgerald, J. A. Watson, F. M. Hoff-
mann, J. S. Thorson, J. Am. Chem. Soc. 2006, 128, 14224–14225.
[9] M. Abel, R. Szweda, D. Trepanier, R. W. Yatscoff, R. T. Foster, Iso-
technika International Inc., Edmonton, Alberta, Canada, U. S. Pat.
No. US 7,160,867, 2007.
[10] D. Liu, S. Sinchaikeul, P. V. G. Reddy, M. Chang, S. Chen, Bioorg.
Med. Chem. Lett. 2007, 17, 617–620.
Preparative synthesis: Lyophilized fructose 6-phosphate aldolase powder
(200 mg; 100 mg of TalBF178Y) was added to a solution (10 mL total reac-
tion volume) containing the respective donor (DHA, HA, HB; 150 mm)
and acceptor (A–F; 100 mm) components in glycyl-glycine buffer (50 mm,
pH 8.5) and the resultant mixture was stirred at room temperature. The
addition of DTT (1 mm) recommended for kinetic assays[20, 21a] was omit-
ted in the preparative reactions as the enzymes were found to be stable
in its absence. The reaction was monitored at regular intervals by TLC
(chloroform/methanol; 5:1 or 2:1) and worked up after 24–48 h depend-
ing on the rate of consumption of the aldehyde. The relative rates of con-
version were determined on the basis of densitometric TLC monitoring,
by signal integration of in situ NMR experiments, as well as by gravimet-
ric quantification of the isolated product. Conversion is denoted as +++
(>75%), ++ (25–75%), + (<25%), or – (no detectable product for-
mation). For product characterization, the individual reaction mixtures
were worked up by lyophilization of the crude reaction mixture followed
by silica gel column chromatography of the residue using chloroform/
methanol as the eluent (20:1–5:1) to provide the pure product. The prod-
ucts were characterized by recording their mass and NMR spectra. The
1H and 13C NMR spectra were recorded and whenever required further
information was obtained by 13C DEPT, 1H–1H COSY, and HSQC ex-
periments.
[11] a) H. G. Floss, J. Ind. Microbiol. Biotechnol. 2001, 27, 183–194;
b) C. T. Walsh, ChemBioChem 2002, 3, 124–134; c) J. M. Langen-
han, B. R. Griffith, J. S. Thorson, J. Nat. Prod. 2005, 68, 1696–1711;
d) C. J. Thibodeaux, C. E. Melancon, H.-w. Liu, Nature 2007, 446,
1008–1016; e) H. B. Bode, R. Mꢂller Angew. Chem. 2007, 119,
2195–2198; Angew. Chem. Int. Ed. 2007, 46, 2147–2150; Angew.
Chem. Int. Ed. 2007, 46, 2147–2150; f) A. Luzhetskyy, C. Mendez,
J. A. Salas, A. Bechthold, Curr. Top. Med. Chem. 2008, 8, 680–709;
g) C. Mendez, A. Luzhetskyy, A. Bechthold, J. A. Salas, Curr. Top.
Med. Chem. 2008, 8, 710–724; h) J. Haerle, A. Bechtold, Methods
Enzymol. 2009, 458, 309–333.
[12] D. J. Newman, G. M. Cragg, K. M. Snader, J. Nat. Prod. 2003, 66,
1022–1037.
[13] a) A. Trefzer, J. A. Salas, A. Bechthold, Nat. Prod. Rep. 1999, 16,
283–299; b) X. He, G. Agnihotri, H.-w. Liu, Chem. Rev. 2000, 100,
4615–4661; c) X. M. He, H.-w. Liu, Annu. Rev. Biochem. 2002, 71,
701–754; d) X. He, H.-w. Liu, Curr. Opin. Chem. Biol. 2002, 6, 590–
597.
[14] A. Kirschning, M. Jesberger, K. U. Schçning, Synthesis 2001, 0507–
0540.
[15] R. M. de Lederkremer, C. Marino, Adv. Carbohydr. Chem. Biochem.
2007, 61, 143–216.
[16] a) W.-D. Fessner, C. Walter, Top. Curr. Chem. 1997, 184, 97–194;
b) T. D. Machajewski, C. H. Wong, Angew. Chem. 2000, 112, 1406–
1430; Angew. Chem. Int. Ed. 2000, 39, 1352–1375.
[17] a) W.-D. Fessner in Modern Aldol Reactions (Ed.: R. Mahrwald),
Wiley-VCH, Weinheim, 2004, pp. 201–272; b) A. K. Samland, G. A.
Sprenger, Appl. Microbiol. Biotechnol. 2006, 71, 253–264; c) W.-D.
Fessner in Asymmetric Organic Synthesis with Enzymes (Eds.: V.
Gotor, I. Alfonso, E. Garcꢆa-Urdiales), Wiley-VCH, Weinheim,
Acknowledgements
This work was supported by the Deutsche Forschungsgemeinschaft
within the framework of SPP1170 (grants Fe244/7-2 to W.D.F. and Sp503/
3-2 to G.A.S.) and by ESF project COST CM0701. We would like to
thank T. Sandalova for her help in preparing the PyMOL graphics.
Chem. Eur. J. 2011, 17, 2623 – 2632
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2631

W.-D. Fessner et al.
2008, pp. 275–318; d) P. Clapꢈs, W.-D. Fessner, G. A. Sprenger,
A. K. Samland, Curr. Opin. Chem. Biol. 2010, 14, 154–167.
[18] a) J. R. Durrwachter, D. G. Drueckhammer, K. Nozaki, H. M.
Sweers, C.-H. Wong, J. Am. Chem. Soc. 1986, 108, 7812–7818;
b) W.-D. Fessner, C. Gosse, G. Jaeschke, O. Eyrisch, Eur. J. Org.
Chem. 2000, 125–132.
[19] a) G. A. Sprenger, U. Schçrken, G. Sprenger, H. Sahm, J. Bacteriol.
1995, 177, 5930–5936; b) A. K. Samland, G. A. Sprenger, Int. J. Bio-
chem. Cell Biol. 2009, 41, 1482–1494.
[20] S. Schneider, T. Sandalova, G. Schneider, G. A. Sprenger, A. K.
Samland, J. Biol. Chem. 2008, 283, 30064–30072.
[21] a) M. Schꢂrmann, G. A. Sprenger, J. Biol. Chem. 2001, 276, 11055–
11061; b) M. Schꢂrmann, M. Schꢂrmann, G. A. Sprenger, J. Mol.
Catal. B 2002, 19–20, 247–252.
[22] a) J. A. Castillo, J. Calveras, J. Casas, M. Mitjans, M. P. Vinardell, T.
Parella, T. Inoue, G. A. Sprenger, J. Joglar, P. Clapes, Org. Lett.
2006, 8, 6067–6070; 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; c) A. L. Concia, C. Lozano, J. A. Cas-
tillo, T. Parella, J. Joglar, P. Clapes, Chem. Eur. J. 2009, 15, 3808–
3816.
[23] S. Thorell, M. Schꢂrmann, G. A. Sprenger, G. Schneider, J. Mol.
Biol. 2002, 319, 161–171.
[24] a) J. Jia, W. Huang, U. Schçrken, H. Sahm, G. A. Sprenger, Y.
Lindqvist, G. Schneider, Structure 1996, 4, 715–724; b) J. Jia, Y.
Lindqvist, U. Schçrken, G. A. Sprenger, G. Schneider, Protein Sci.
1997, 6, 119–124.
[30] a) M. D. Bednarski, E. S. Simon, N. Bischofberger, W.-D. Fessner,
M.-J. Kim, W. Lees, T. Saito, H. Waldmann, G. M. Whitesides, J.
Am. Chem. Soc. 1989, 111, 627–635; b) M. T. Zannetti, C. Walter,
M. Knorst, W.-D. Fessner, Chem. Eur. J. 1999, 5, 1882–1890.
[31] Class I aldolases: a) A. Dalby, Z. Dauter, J. A. Littlechild, Protein
Sci. 1999, 8, 291–297; b) C. L. M. J. Verlinde, P. M. Quigley, J. Mol.
Model. 1999, 5, 37–45; c) K. H. Choi, J. Shi, C. E. Hopkins, D. R.
Tolan, K. N. Allen, Biochemistry 2001, 40, 13868–13875; d) M. St-
Jean, J. Lafrance-Vanasse, B. Liotard, J. Sygusch, J. Biol. Chem.
2005, 280, 27262–27270; e) C. LowKam, B. Liotard, J. Sygusch, J.
Biol. Chem. 2010, 285, 21143–21152; Class II aldolases: f) W.-D.
Fessner, A. Schneider, H. Held, G. Sinerius, C. Walter, M. Hixon,
J. V. Schloss, Angew. Chem. 1996, 108, 2366–2369; Angew. Chem.
Int. Ed. Engl. 1996, 35, 2219–2221; g) D. R. Hall, G. A. Leonard,
C. D. Reed, C. I Watt, A. Berry, W. N. Hunter, J. Mol. Biol. 1999,
287, 383–394; h) S. M. Zgiby, G. J. Thomson, S. Qamar, A. Berry,
Eur. J. Biochem. 2000, 267, 1858–1868; i) A. C. Joerger, C. Goße,
W.-D. Fessner, G. E. Schulz, Biochemistry 2000, 39, 6033–6041;
j) D. R. Hall, C. S. Bond, G. A. Leonard, C. I. Watt, A. Berry, W. N.
Hunter, J. Biol. Chem. 2002, 277, 22018–22024; k) M. Krçmer, I.
Merkel, G. E. Schulz, Biochemistry 2003, 42, 10560–10568.
[32] Aldolase antibodies with similar substrate profiles have been report-
ed: T. Hoffmann, G. Zhong, B. List, D. Shabat, J. Anderson, S. Gra-
matikova, R. A. Lerner, C. F. Barbas, J. Am. Chem. Soc. 1998, 120,
2768–2779.
[33] a) P. J. Walsh, K. B. Sharpless, Synlett 1993, 605–606; b) W. Notz, B.
List, J. Am. Chem. Soc. 2000, 122, 7386–7387; c) K. Sakthivel, W.
Notz, T. Bui, C. F. Barbas, J. Am. Chem. Soc. 2001, 123, 5260–5267;
d) M. Markert, M. Mulzer, B. Schetter, R. Mahrwald, J. Am. Chem.
Soc. 2007, 129, 7258–7259; e) Y. Zhao, A. W. Mitra, A. H. Hoveyda,
M. L. Snapper, Angew. Chem. 2007, 119, 8623–8626; Angew. Chem.
Int. Ed. 2007, 46, 8471–8474; f) B. Schetter, C. Stosiek, B. Ziemer,
R. Mahrwald, Appl. Organomet. Chem. 2007, 21, 139–145; g) K.
Aelvoet, A. S. Batsanov, A. J. Blatch, C. Grosjean, L. G. F. Patrick,
C. A. Smethurst, A. Whiting, Angew. Chem. 2008, 120, 780–782;
Angew. Chem. Int. Ed. 2008, 47, 768–770; h) J. Paradowska, M. Ro-
gozinska, J. Mlynarski, Tetrahedron Lett. 2009, 50, 1639–1641.
[34] M. M. Bradford, Anal. Biochem. 1976, 72, 248–254.
[25] W. L. DeLano, The PyMOL Molecular Graphics System, Version
1.3, Schrçdinger LLC, 2002. (http://www.pymol.org).
[26] J. P. Richard, J. Am. Chem. Soc. 1984, 106, 4926–4936.
[27] S. Schneider, M. Gutiꢈrrez, T. Sandalova, G. Schneider, P. Clapꢈs,
G. A. Sprenger, A. K. Samland, ChemBioChem 2010, 11, 681–690.
[28] a) X. Garrabou, J. A. Castillo, C. Guerard-Helaine, T. Parella, J.
Joglar, M. Lemaire, P. Clapes, Angew. Chem. 2009, 121, 5629–5633;
Angew. Chem. Int. Ed. 2009, 48, 5521–5525; b) J. A. Castillo, C.
Guerard-Helaine, M. Gutierrez, X. Garrabou, M. Sancelme, M.
Schꢂrmann, T. Inoue, V. Helaine, F. Charmantray, T. Gefflaut, L.
Hecquet, J. Joglar, P. Clapes, G. A. Sprenger, M. Lemaire, Adv.
Synth. Catal. 2010, 352, 1039–1046.
[35] C. S. Hanes, Biochem. J. 1932, 26, 1406–1421.
[36] P. Weisser, R. Krꢃmer, H. Sahm, G. A. Sprenger, J. Bacteriol. 1995,
177, 3351–3354.
[29] a) O. R. Martin, S.-L. Korppi-Tommola, W. A. Szarek, Can. J. Chem.
1982, 60, 1857–1862; b) R. J. Woods, W. A. Szarek, V. H. Smith Jr.,
J. Am. Chem. Soc. 1990, 112, 4732–4741.
Received: October 12, 2010
Published online: February 2, 2011
2632
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 2623 – 2632