A tailor-made chimeric thiamine diphosphate dependent enzyme for the direct asymmetric synthesis of (S)-benzoins

Article abstract of DOI:10.1002/anie.201405069

Thiamine diphosphate dependent enzymes are well known for catalyzing the asymmetric synthesis of chiral α-hydroxy ketones from simple prochiral substrates. The steric and chemical properties of the enzyme active site define the product spectrum. Enzymes catalyzing the carboligation of aromatic aldehydes to (S)-benzoins have not so far been identified. We were able to close this gap by constructing a chimeric enzyme, which catalyzes the synthesis of various (S)-benzoins with excellent enantiomeric excess (>99) and very good conversion. Hybrid vigor: Combining the active-site characteristics of two thiamine diphosphate (ThDP) dependent enzymes solved the long-standing problem of enzymatic asymmetric (S)-benzoin synthesis starting from commercially available benzaldehydes. The resulting rationally designed hybrid enzyme provides access to various (S)-benzoins with excellent enantiomeric excess and good conversion.

Full text of DOI:10.1002/anie.201405069

.
Angewandte
Communications
DOI: 10.1002/anie.201405069
Biocatalysis
A Tailor-Made Chimeric Thiamine Diphosphate Dependent Enzyme
for the Direct Asymmetric Synthesis of (S)-Benzoins**
Robert Westphal, Constantin Vogel, Carlo Schmitz, Jꢀrgen Pleiss, Michael Mꢀller,
Martina Pohl,* and Dçrte Rother
In memory of Ayhan S. Demir
Abstract: Thiamine diphosphate dependent enzymes are well
known for catalyzing the asymmetric synthesis of chiral a-
hydroxy ketones from simple prochiral substrates. The steric
and chemical properties of the enzyme active site define the
product spectrum. Enzymes catalyzing the carboligation of
aromatic aldehydes to (S)-benzoins have not so far been
identified. We were able to close this gap by constructing
a chimeric enzyme, which catalyzes the synthesis of various
(S)-benzoins with excellent enantiomeric excess (> 99%) and
very good conversion.
able benzaldehydes. This direct synthesis is still a unique
feature of nonenzymatic synthetic organic chemistry.[5]
Whereas the R-selective enzymatic synthesis of benzoins
has already been achieved,[6] S-selective synthesis is limited by
the specific active-site architecture of ThDP-dependent
enzymes (Figure 1). Introduction of the S-pocket concept
T
hiamine diphosphate (ThDP) dependent enzymes are
proven catalysts for the stereo- and regioselective synthesis
À
of chiral a-hydroxy ketones through asymmetric C C bond
formation starting from aldehydes, a-keto acids, and
ketones.[1] Whereas enzymes are available for the enantio-
complementary synthesis of a broad range of symmetrical and
mixed aliphatic, araliphatic, and aromatic products,[2] so far no
biocatalyst is available to selectively access (S)-benzoins
starting from benzaldehydes. Instead, other enzymatic routes
became available, which involve kinetic resolution as well as
deracemization of rac-benzoins,[3] or asymmetric reduction of
1,2-diarylethane-1,2-diones (benzils).[4] Although these
approaches are mostly characterized by high stereoselectivity,
they all share the same drawback: the racemic or prochiral
starting material must be chemically synthesized beforehand.
Such additional steps reduce the eco-efficiency and sustain-
ability of the process compared to the direct enzymatic
synthesis of (S)-benzoins starting from commercially avail-
Figure 1. A schematic presentation of the scope and limitations of
carboligations catalyzed by ApPDC-E469G and PfBAL. (S)-Benzoin (1a)
formation is prevented either by a small donor-binding site (A), or an
S-pocket that is inaccessible for the antiparallel orientation of the
acceptor benzaldehyde relative to the ThDP-bound donor benzalde-
hyde (as a hydroxybenzyl group; B). Dashed rectangle: donor-binding
site; dotted circle: S-pocket region; rectangle: acetaldehyde; hexagon:
benzaldehyde; gray: donor; cyan: antiparallel-oriented acceptor, which
leads to the respective (S)-product; yellow: parallel-oriented acceptor,
which leads to the respective (R)-product.
[*] Dr. R. Westphal,[+] C. Schmitz, Prof. Dr. M. Pohl, Dr. D. Rother
IBG-1: Biotechnology, Forschungszentrum Jꢀlich GmbH
Wilhelm-Johnen-Strasse, 52425 Jꢀlich (Germany)
E-mail: ma.pohl@fz-juelich.de
Dipl.-Biol. C. Vogel,[+] Prof. Dr. J. Pleiss
Institute of Technical Biochemistry, University of Stuttgart
Allmandring 31, 70569 Stuttgart (Germany)
enabled the design of an S-selective variant of pyruvate
decarboxylase from Acetobacter pasteurianus (ApPDC) for
the formation of (S)-phenylacetylcarbinol (PAC).[7] Replace-
ment of a glutamate residue (E469) by glycine allowed
benzaldehyde as an acceptor to bind predominantly in
antiparallel orientation relative to the ThDP-bound donor,
which is a prerequisite for S-selectivity. However, ApPDC-
E469G only provides a small donor-binding site that prefer-
entially stabilizes small aliphatic donor substrates, thereby
preventing the formation of (S)-benzoin [(S)-1a] (Fig-
ure 1A). Benzaldehyde lyase from Pseudomonas fluorescens
(PfBAL) is a powerful but strictly R-selective catalyst for the
Prof. Dr. M. Mꢀller
Institute of Pharmaceutical Sciences
Albert-Ludwigs-University Freiburg
Albertstrasse 25, 79104 Freiburg (Germany)
[+] These authors contributed equally to this work.
[**] This work was financed by the German Research Foundation (DFG)
in the framework of the Research Group “Diversity of asymmetric
thiamine catalysis” (FOR 1296). We thank Doris Hahn, Ursula
Mackfeld, and Heike Offermann for skillful technical support.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201405069.
9376
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 9376 –9379

Angewandte
Chemie
synthesis of 1a.[6b] In contrast to ApPDC, PfBAL provides
a large donor-binding site, which is ideal for stabilizing
benzaldehyde in that position. However, owing to a missing S-
pocket, PfBAL only allows the parallel arrangement of the
donor and acceptor benzaldehydes prior to carboligation.
Furthermore, S-pocket engineering is limited in PfBAL by
the position of the protein backbone of the respective a-helix
in the S-pocket region (Figure 1B).
To solve the long-standing problem of enzymatic (S)-
benzoin synthesis, a rational hybridization approach was
followed, in which the active-site characteristics of the variant
ApPDC-E469G[7] and PfBAL were combined. Two
approaches were conceivable: 1) the introduction of a large
S-pocket into PfBAL, or 2) the extension of the donor-
binding site of ApPDC-E469G. Herein, we show that the
second approach indeed resulted in the first tailor-made S-
selective ThDP-dependent enzyme variant for the formation
of various benzoins.
A combination of modeling studies and a comprehensive
sequence analysis of the amino acid distribution in 186
homologous PDC sequences and 43 BAL sequences led to the
identification of a threonine residue (T384) as the pivotal
factor that influences the size of the donor-binding site in
ApPDC-E469G. T384 limits the space for benzaldehyde in the
donor-binding site and is conserved in 92% of ApPDC
homologous sequences (Figure 2A), whereas glycine was
found at the equivalent position in all PfBAL homologous
sequences. T384 in ApPDC-E469G was replaced with glycine
through site-directed mutagenesis in order to mimic the
PfBAL donor-binding site. As a result, an enlarged donor-
binding site with putatively sufficient space for benzaldehyde
was obtained (Figure 2B). Moreover, short molecular dynam-
ics simulations revealed that a neighboring tryptophan
residue (W388) underwent minor conformational changes,
probably because of interactions with the ThDP-bound
benzaldehyde donor (hydroxybenzyl), thereby further open-
ing up the donor-binding site.
Biochemical characterization of the new variant proved
that the single mutation of T384 to glycine indeed altered the
chemoselectivity of ApPDC-E469G, which subsequently
preferred benzaldehyde as the donor. Whereas no significant
formation of 1a in the homocoupling of benzaldehyde was
observed with ApPDC-E469G (Table 1, entry 1), double
Table 1: Enzymatic synthesis of (S)-1a as catalyzed by ApPDC variants.[a]
Entry
ApPDC variant
ee [%][b]
Conv. [%][c]
1
2
3
4
5
6
E469G
E469G/T384G
E469G/T384G/I468G
E469G/T384G/I468A
E469G/T384G/I468V
E469G/T384G/I468V/W543F
n.d.[d]
59
66
87
76
<1
52
23
95
40
36
95
[a] Reaction conditions: 50 mm triethanolamine buffer, pH 8.0, 2 mm
MgSO4, and 0.1 mm ThDP; 1 mgmLÀ1 enzyme; 18 mm benzaldehyde;
208C, 6 h; [b] determined by chiral-phase HPLC; [c] Determined by
chiral-phase HPLC based on the consumption of benzaldehyde; [d] Not
determined.
variant ApPDC-E469G/T384G catalyzed the benzoin forma-
tion with 52% conversion under the tested conditions
(entry 2). The preference for benzaldehyde as the donor
was also demonstrated in the mixed carboligation of acetal-
dehyde and benzaldehyde by variant ApPDC-E469G/T384G,
which resulted in the formation of a mixture of 1a and 2-
hydroxypropiophenone (2-HPP), whereas PAC, the main
product obtained with wild-type ApPDC or variant ApPDC-
E469G,[7] was only detected in trace amounts. Furthermore,
variant ApPDC-E469G/T384G catalyzed both the synthesis
of (S)-1a from benzaldehyde with moderate stereoselectivity
(59% ee, entry 2) and mixed carboligation towards 2-HPP
with good S-selectivity (91% ee).
To improve the moderate S-selectivity of ApPDC-E469G/
T384G, two strategies are possible: stabilization of the
antiparallel (“S-pathway”) or destabilization of the parallel
acceptor orientation (“R-pathway”) prior to carboligation.[8]
To suppress the R-pathway in ApPDC-E469G/T384G, the
active site was examined for residues that potentially stabilize
the acceptor benzaldehyde in the parallel orientation. By
performing molecular modeling, I468 and W543 were iden-
tified (Figure 3A) as residues that could stabilize parallel-
oriented benzaldehyde through nonpolar interactions or p-
stacking.
I468 was replaced with valine, glycine, or alanine to
increase the distance of the respective side chain to parallel-
oriented benzaldehyde and thus disrupt the stabilization of
the R-pathway (Figure 3B). In all cases, the new variants
showed improved S-selectivity for the formation of (S)-1a
(Table 1, entries 3–5). Out of the different variants, the triple
variant E469G/T384G/I468A performed best with respect to
ee (87%, S) and conversion (95%) under standard reaction
conditions. The high conversion is particularly surprising
Figure 2. Stereoview of the donor-binding sites of ApPDC-E469G (A)
and ApPDC-E469G/T384G (B) with benzaldehyde bound as a hydrox-
ybenzyl group (gray) to C2 of ThDP (orange). A) ApPDC-E469G is not
able to properly bind benzaldehyde in the small donor-binding site,
which is mainly restricted by T384 (red). B) The donor-binding site
could be opened for benzaldehyde by replacing T384 with glycine.
Moreover, the equilibrated structure revealed conformational changes
to W388 (green) that additionally opened the donor-binding site.
Angew. Chem. Int. Ed. 2014, 53, 9376 –9379
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
9377

.
Angewandte
Communications
cluster in the ApPDC variant. In addition to improved S-
selectivity in the homocoupling reaction of benzaldehyde,
ApPDC-E469G/T384G/I468A/W543F
also
revealed
enhanced stereoselectivity for the synthesis of (S)-2-HPP
(> 99% ee) starting from benzaldehyde as the donor and
acetaldehyde as the acceptor substrate.
To evaluate the synthetic potential of the newly designed
variants, a substrate screening with substituted benzaldehyde
derivatives was performed with ApPDC-E469G/T384G/
I468A and ApPDC-E469G/T384G/I468A/W543F (Table 2)
under optimized reaction conditions (reduced temperature of
158C) for the synthesis of (S)-1a. meta-Substituted benzalde-
hyde derivatives turned out to be the best substrates in the
homocoupling reaction in terms of S-selectivity and conver-
sion. Except for 3-fluorobenzaldehyde (1e), all of the meta-
substituted benzaldehydes (Table 2, entries 5–9) were trans-
formed with higher S-selectivity than benzaldehyde (Table 2,
entry 1). The reactions with variant E469G/T384G/I468A/
W543F again showed higher S-selectivity and lower conver-
sion than those with variant E469G/T384G/I468A. Remark-
ably, variant E469G/T384G/I468A/W543F catalyzed the syn-
thesis of enantiopure 1 f–i (> 99% ee (S), Table 2, entries 6–
9). These synthesis reactions could be successfully scaled-up
from analytical scale (300 mL) to preparative scale (20 mL)
with final product concentrations of up to 4 gLÀ1 (yields of
isolated product 61–85%, Table 2).
Figure 3. Stereoview representation of the possible stabilization of the
parallel-oriented acceptor benzaldehyde (yellow) in the active sites of
ApPDC-E469G/T384G (A) and ApPDC-E469G/T384G/I468A/W543F
(B). A) In E469G/T384G, parallel-oriented benzaldehyde is putatively
stabilized by I468 (blue) and W543 (purple), neither of which directly
influence its antiparallel orientation (cyan). B) In variant E469G/
T384G/I468A/W543F, the stabilization of parallel-oriented benzalde-
hyde by A468 (blue) and F543 (purple) is no longer possible.
because multiple active-site muta-
tions frequently result in a drastic
decrease in enzymatic activity, as
has been demonstrated for other
ThDP-dependent enzymes.[8]
Table 2: Synthesis of benzoins (1a–l) as catalyzed by ApPDC variants.[a]
E469G/T384G/
I468 A
E469G/T384G/
I468 A/W543F
Entry
Ar
Product
ee [%][b]
Conv. [%][c]
(yield [%])[d]
ee [%][b]
Conv. [%][c]
To further improve the stereo-
selectivity of variant E469G/
T384G/I468A, a fourth mutation at
the position of tryptophan 543 was
introduced. W543 is part of the C-
terminal a-helix, which covers the
entrance to the active site of
ApPDC (Figure 3A), and of an
aromatic cluster (for details, see
Chapter 2 of the Supporting Infor-
mation) that might be relevant for
structural stabilization. This posi-
tion was subjected to site-saturation
mutagenesis by using NDT codon
degeneracy. Only few variants
showed significant carboligation
activity during screening, but all of
(yield [%])[d]
1
2
3
4
5
6
7
8
C6H5
2-FC6H4
1a
1b
1c
1d
1e
1 f
1g
1h
1i
89 (S)
58 (R)
75 (R)
n.d.[e]
87 (S)
91 (S)
95 (S)
96 (S)
98 (S)
64 (S)
>99 (R)
n.d.
92 (85)
36
<5
n.c.
80 (82)
97
85 (84)
30 (70)
93
53
10
98 (S)
21 (R)
n.d.
26 (66)
<5
n.c.
2-ClC6H4
2-MeOC6H4
3-FC6H4
3-ClC6H4
3-BrOC6H4
3-IC6H4
3-MeOC6H4
4-FC6H4
4-ClC6H4
4-MeOC6H4
n.d.
n.c.
93 (S)
>99 (S)
>99 (S)
>99 (S)
>99 (S)
85 (S)
n.d.
36
48 (72)
30
11
58 (61)
<5
9
10
11
12
1j
1k
1l
n.c.
<5
n.d.
n.c.
[a] Reaction conditions: see Table 1, reaction temperature: 158C; [b,c] See Table 1; [d] Yields of isolated
product after preparative synthesis and conversion of >90% after 24 h and 48 h, (reaction conditions:
see Chapter 4 of the Supporting Information); [e] Not determined; [f] No conversion.
the active variants were S-selective for the formation of 1a.
Among these variants, E469G/T384G/I468A/W543F revealed
improved S-selectivity for the synthesis of 1a, with 95% ee
and a conversion of 36% (Table 1, entry 6). In comparison to
W543, the increased distance between the phenyl ring of the
parallel-oriented acceptor benzaldehyde and F543 (6.8 ꢀ
compared to 3.7 ꢀ) is assumed to prevent stabilizing inter-
actions (Figure 3B). Furthermore, the physicochemical prop-
erties at position 543 were preserved by the substitution of
tryptophan by phenylalanine, which might be beneficial to
maintaining the structural stability conferred by the aromatic
In contrast to meta-substituted benzaldehydes, ortho- and
para-substituted benzaldehydes, with the exception of 4-
fluorobenzaldehyde (1j), were converted with low conversion
to yield an excess of the R-enantiomer.
These results confirm our findings from recent studies on
2-succinyl-5-enolpyruvyl-6-hydroxy-3-cyclohexene-1-carbox-
ylate synthase (MenD), which revealed that meta-substituted
benzaldehydes are converted with higher S-selectivity and
conversion than ortho- and para-substituted benzaldehydes
and unsubstituted benzaldehyde.[8,9] However, the reasons for
the preference for meta-substituted benzaldehydes in S-
9378
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 9376 –9379

Angewandte
Chemie
selective carboligation reactions and the switch in stereose-
lectivity in the case of ortho- and para-substituted benzalde-
hydes are still not understood. In this case, a combination of
steric and electronic interactions between the substrate and
the S-pocket might influence the stereoselectivity. The
elucidation of this phenomenon is now the subject of further
investigation.
Keywords: asymmetric catalysis · biocatalysis · C–C coupling ·
enantioselectivity · protein engineering
.
[1] a) P. Hoyos, J. V. Sinisterra, F. Molinari, A. R. Alcantara, P.
Domiguez de Maria, Acc. Chem. Res. 2010, 43, 288 – 299; b) M.
Brovetto, D. Gamenara, P. Saenz Mꢁndez, G. A. Seoane, Chem.
Rev. 2011, 111, 4346 – 4403.
In summary, the rational design of a hybrid substrate-
binding site, which combines the active-site characteristics of
the S-selective ApPDC-E469G variant and PfBAL, enabled
the creation of a new biocatalyst for the synthesis of different
(S)-benzoins with excellent ee values and good conversion
when starting from commercially available benzaldehydes.
These results highlight the robustness of ThDP-dependent
enzymes with respect to active-site mutations, as well as their
tremendous catalytic potential in asymmetric carboligation
reactions. Moreover, this hybridization approach might open
the field for the combinatorial assembly of selectivity-
determining modules from different ThDP-dependent
enzymes and offer a new perspective on thiamine catalysis
with respect to the design of variants with novel catalytic
activities.
[2] a) M. Mꢂller, D. Gocke, M. Pohl, FEBS J. 2009, 276, 2894 – 2904;
b) M. Mꢂller, G. A. Sprenger, M. Pohl, Curr. Opin. Chem. Biol.
2013, 17, 261 – 270; c) H. C. Hailes, D. Rother, M. Mꢂller, R.
Westphal, J. M. Ward, J. Pleiss, C. Vogel, M. Pohl, FEBS J. 2013,
280, 6374 – 6394.
[3] a) A. S. Demir, M. Pohl, E. Janzen, M. Mꢂller, J. Chem. Soc.
Perkin Trans. 1 2001, 633 – 635; b) M. C. Fragnelli, P. Hoyos, D.
Romano, R. Gandolfi, A. R. Alcantara, F. Molinari, Tetrahedron
2012, 68, 523 – 528.
[4] A. S. Demir, A. Peruze, D. Umut, E. Umut, J. Mol. Catal. B 2008,
55, 164 – 168.
[5] D. Enders, U. Kallfass, Angew. Chem. 2002, 114, 1822 – 1824;
Angew. Chem. Int. Ed. 2002, 41, 1743 – 1745.
[6] a) A. S. Demir, T. Dꢂnnwald, H. Iding, M. Pohl, M. Mꢂller,
Tetrahedron: Asymmetry 1999, 10, 4769 – 4774; b) A. S. Demir, ꢃ.
Sesenoglu, E. Eren, B. Hosrik, M. Pohl, E. Janzen, D. Kolter, R.
Feldmann, P. Dꢂnkelmann, M. Mꢂller, Adv. Synth. Catal. 2002,
344, 96 – 103; c) P. Dꢂnkelmann, D. Kolter-Jung, A. Nitsche, A. S.
Demir, P. Siegert, B. Lingen, M. Baumann, M. Pohl, M. Mꢂller, J.
Am. Chem. Soc. 2002, 124, 12084 – 12085.
[7] D. Rother (neꢁ Gocke), G. Kolter, T. Gerhards, C. L. Berthold, E.
Gauchenova, M. Knoll, J. Pleiss, M. Mꢂller, M. Pohl, Chem-
CatChem 2011, 3, 1587 – 1596.
[8] R. Westphal, D. Hahn, U. Mackfeld, S. Waltzer, M. Beigi, M.
Widmann, C. Vogel, J. Pleiss, M. Mꢂller, D. Rother, M. Pohl,
ChemCatChem 2013, 5, 3587 – 3594.
[9] R. Westphal, S. Jansen, C. Vogel, J. Pleiss, M. Mꢂller, D. Rother,
M. Pohl, ChemCatChem 2014, 6, 1082 – 1088.
Experimental Section
All chemicals were purchased from Sigma Aldrich. Benzaldehydes
were freshly distilled before use. The generation, expression, and
purification of ApPDC variants are described in the Supporting
Information. Descriptions of sequence and structural analyses, as well
as reaction details and product analytics, can also be found in the
Supporting Information.
Received: May 7, 2014
Published online: July 15, 2014
Angew. Chem. Int. Ed. 2014, 53, 9376 –9379
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 9379