Carving the active site of almond R-HNL for increased enantioselectivity

Article abstract of DOI:10.1002/anie.200500435

Structure-guided redesign of the active site of almond (R)-PaHNL5 for increased enantioselectivity resulted in four improved muteins. In particular, mutation V360I gave enhanced conversion rates and enantioselectivities higher than 96% ee for two structur

Full text of DOI:10.1002/anie.200500435

Angewandte
Chemie
Communications
The active site of almond hydroxynitrile lyase was redesigned to give
highly enantioselective enzyme variants for the synthesis of (R)-2-
hydroxy-4-phenylbutyronitrile. K. Gruber, A. Glieder et al. describe their
rationalization and realization of these biocatalysts in their commu-
nication on the following pages.
4700 ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200500435
Angew. Chem. Int. Ed. 2005, 44, 4700 – 4704

Angewandte
Chemie
Protein Engineering
trile (1b) and (R)-2-hydroxy-4-phenyl-3-butene nitrile (2b)
emerged as the most interesting representatives in terms of
commercial use. Enantiopure a-hydroxycarboxylic acids,
which are important intermediates for the synthesis of a
class of angiotensin-converting enzyme inhibitors (ACEi)
known as “prils”,[5] can be derived from these cyanohydrins by
acid hydrolysis. Compound 2a was described as a notoriously
recalcitrant substrate, but palladium-catalyzed hydrogenation
of 2b also yields (R)-2-hydroxy-4-phenylbutyric acid.[6] In
addition 2b can be used as a versatile intermediate for
asymmetric epoxidation, dihydroxylation, and halogen addi-
tion.[7] Only R enantiomer 2b is a buildingblock of
pharmacologically active “prils”, and commercial production
calls for high enantiomeric excesses (> 95% ee), high yields
(> 95%), an environmentally benign process, and economic
reaction times with a low enzyme/substrate ratio. Enantiomer
separation is feasible by preferential crystallization of dia-
stereomers, although with limited yields (68%).[8] Other
synthetic routes, for example, enantioselective reduction of
keto or diketo esters, suffer from complicated procedures or
Carving the Active Site of Almond R-HNL for
Increased Enantioselectivity**
Roland Weis, Richard Gaisberger, Wolfgang Skranc,
Karl Gruber,* and Anton Glieder*
Hydroxynitrile lyases (HNLs) catalyze quantitative, stereo-
selective carbon–carbon bond formation in the addition of
HCN to aldehydes or ketones yieldingenantiopure cyanohy-
drins, which are key intermediates for numerous synthetic
routes.[1] Both R- and S-selective HNLs are widely present in
nature, and several genes have been cloned and expressed.[1]
Recombinant almond (Prunus amygdalus) (R)-HNL isoen-
zyme 5 (PaHNL5) is secreted to the culture supernatant of
Pichia pastoris, which can be directly used for biocatalytic
conversions in water or biphasic systems without prior
enzyme purification or immobilization. Further prerequisites
for an industrial application of this enzyme are its stability at
acidic pH (suppression of the unselective non-enzymatic
background reaction) and its enantioselectivity. Although the
high stability of recombinant PaHNL5 already enabled
[5,9]
require expensive startingcompounds.
Consideringimplementation of these chiral building
blocks on a large scale, we investigated a biocatalytic route
usingvery low amounts (i.e. 17 mgenzyme per mmol 1a) of
recombinant PaHNL5 (Scheme 1). This route starts with
cheap substrates, minimizes the number of unit operations,
and offers optional crystallization after cyanohydrin hydrol-
ysis to recover pure product. Conversion of 1a to 1b using
recombinant PaHNL5 was almost complete (93%) after 4 h
with 89.4% ee (Table 1, entry 1). However, to avoid enantio-
selective crystallization as an additional step in the synthetic
route, we needed a more selective mutein.
Directed evolution of enzymes has proved to be an
efficient tool to influence the enantioselectivity of many
(mostly bacterial) enzymes.[10] However, although expression
of PaHNL isoenzymes by E. coli is feasible, highly active
PaHNL5 could only be expressed in Pichia pastoris,[1,11] which
is not a reliable host system for laboratory evolution.
Furthermore a codon dilemma[12] impedes access to all
possible amino acid exchanges by point mutations, especially
when only a few thousand mutants can be screened. Because
of our recent success in designing a more active mutein for
sterically demandingmandelonitrile derivatives, [1a] we started
a structure-guided approach to increase the enantioselectivity
of this lyase. To our knowledge this is the first report on a
successful structure-guided design of lyases to improve
enantioselectivity, although rational approaches to Manihot
esculenta HNL variants with reduced substrate transport
limitations resulted in improved enantioselectivity for a
variety of substrates as well.[13]
stereoselective enzymatic syntheses in water or biphasic
[1,2]
systems even with slow-reactingsubstrates,
the enantiose-
lectivity of PaHNL5 is in some cases still too low for
biocatalysis on a large scale.
Amongthe broad substrate range accepted by PaHNL5[3]
those with the aldehyde functionality separated from an
aromatic moiety by an aliphatic linker represent an especially
interesting group. The aromatic ring, which plays an impor-
tant role in the correct recognition and binding of the natural
substrate mandelonitrile (benzaldehyde cyanohydrin),[4] still
has to fit into the closer region around the active site. On the
other hand its distance from the functional group and the
conformational flexibility of the alkyl linker reduce its
influence on the stereoselectivity of the reaction. Among
such aromatic substrates, 3-phenylpropionaldehyde (1a) and
3-phenylpropenal (2a, trans-cinnamaldehyde), and their cor-
respondingcyanohydrins ( R)-2-hydroxy-4-phenylbutyroni-
[*] R. Weis, R. Gaisberger, Prof. Dr. K. Gruber, Prof. Dr. A. Glieder
Research Centre Applied Biocatalysis GmbH
Petersgasse 14, 8010 Graz (Austria)
Fax: (+43)316-873-4072
E-mail: karl.gruber@a-b.at
glieder@glieder.com
Dr. W. Skranc
DSM Fine Chemicals Austria Nfg. GmbH & Co KG
R & D Center Linz
We modeled the complexes of PaHNL5 with (R)-1b and
(S)-1b. For both substrates equivalent bindingmodes were
observed with respect to the position of the phenyl group as
well as the mechanistically important polar interactions of the
hydroxy and cyano groups with His498 and His460
(Figure 1).[4] Differences were observed only for the inter-
actions of PaHNL5 with the alkyl linker chain of 1b, which
was oriented towards Ala111 in the S enantiomer and
towards Val360 in the R enantiomer (Figure 1, see also
St.-Peter-Strasse 25, 4021 Linz (Austria)
[**] This research was supported byDSM, the FFG, the province of
Styria, the SFG, and the city of Graz. We thank H. Mandl from the
Research Centre Applied Biocatalysis, I. Wirth and O. Maurer from
DSM for excellent technical support, and K. Faber, C. Kratky, and
W. Kroutil for valuable comments on the manuscript, and P. Naggl
for the color artwork.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2005, 44, 4700 –4704
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4701

Communications
while preservinghydrophobicity. Ala111 and Val360
interact with the alkyl chain of the substrate from
opposite sides and therefore are promisingsites for
mutations. The same is true for Leu331 and Leu343,
which mainly interact with the phenyl group and form
part of the access tunnel leadingto the active site. In
total, we studied 24 mutants in silico, and 12 of
those—with the largest predicted changes in enantio-
selectivity—were prepared and analyzed experimen-
tally. In a first test of the hydrocyanation of 1a with
only low amounts of each mutein, V360I yielded the
most selective enzyme (Table 1, entry 13). As
expected, A111G was less stereoselective. In a more
detailed examination, 1 mgeach of the best four
muteins (entries 10–13) were analyzed for conversion
of 2 g(15 mmol) of 1a (i.e. 67 mgenzyme per mmol
substrate). L331A, V360M, and V360I (Table 2,
entries 3–5) resulted in almost complete substrate
conversion within 4 h with significantly improved
enantioselectivity. At the end of the conversion 1b
produced with mutein A111GV360I also showed
slightly higher enantiopurity (91.8% ee), although
the conversion was a little slower.
To explore turnover rates, the amount of the
enzyme was adjusted such that with each individual
mutein the conversion of 2 g(15 m m) of 1a was in a
roughly linear range during the first 20 min. Specific
activities under these conditions (Table 2) served for
comparison of the individual muteins for cyanohydrin
synthesis, although due to the nonlinear kinetics of
PaHNL5-catalyzed reactions in biphasic systems,
Scheme 1. a) Enantioselective synthesis of (R)-2-hydroxy-4-phenylbutyronitrile (1b) and
(R)-2-hydroxy-4-phenylbutene nitrile (2b) from 3-phenylpropionaldehyde (1a) and 3-phen-
ylpropenal (2a), respectively, for the production of (R)-2-hydroxy-4-phenylbutyric acid, an
intermediate for “prils”. b) Examples of “prils”.
such numbers cannot be translated to kcat values for higher
enzyme concentrations. The introduction of isoleucine at
position 360 raised the ee to more than 96% and accelerated
the transformation of 1a almost sixfold. Thus the conversion
Table 1: Comparison of the constructed muteins to the unmodi-
fied PaHNL5 for the enantioselective hydrocyanation of 1a.[a]
EmnetryEnzy
Conv. [%]
ee [%]
1
2
3
4
5
6
7
8
PaHNL5
A111VL331A
A111G
A111VL343A
A111V
L331F
A111GV360M
L343A
L343F
V360M
A111GV360I
L331A
93
30
40
40
16
80
75
83
95
93
92
94
98
89.4
ꢀ0
15.2
27.6
60.2
63.1
64.7
68.8
74.4
86.7
87.6
88
9
10
11
12
13
V360I
95.3
[a] Reaction conditions: 30 mmol of 1a and 0.5 mg of enzyme
(17 mgmmolÀ1 substrate), 4 h, 108C, pH 3.4.
video material in the SupportingInforma-
tion). We took advantage of this pseudo
mirror symmetry and redesigned the sub-
strate bindingsite of PaHNL5 to tailor the
accessible volume for one or the other
enantiomer, similarly to other successful
examples of laboratory-evolved enzymes.[14]
Amino acids in the bindingsite were
exchanged with residues of different sizes
Figure 1. Stereoimage of the superposition of modeled complexes of (R)-1b (blue) and
(S)-1b (pink) bound to the active site of PaHNL5. Amino acid side chains are shown in
light gray, the FAD cofactor in yellow. N and O atoms are shown in blue and red, respec-
tively. Residues selected for mutagenesis are labeled in red, and their van der Waals vol-
umes are represented bysmall spheres. The image was prepared with the software PyMol
(http://www.pymol.org).
4702
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 4700 –4704

Angewandte
Chemie
was complete within a couple of hours despite a low enzyme/
substrate ratio. In contrast, in a recent report at least 30 times
more enzyme[6] was employed. Mutations L331A and, sur-
required amount of biocatalyst 10- to 30-fold to produce
enantiopure (R)-2-hydroxy-4-phenylbutyric acid and deriva-
tives thereof. From trans-cinnamaldehyde (2a) very low
amounts of catalyst with mutation V360I were needed to
produce cyanohydrin 2b with an optical purity of ꢀ 98% ee
within 3 h. The conversion time for this recalcitrant substrate
was reduced from 168 h to 3 h. Interestingly our designed and
more selective muteins were also more active. This is in
contrast to reports on the design of more selective hydrolases,
where just the reactivity with the wrongenantiomer was
reduced.[17] PaHNL5-V360I is currently the most efficient
catalyst for syntheses of enantiomerically pure building
blocks for pharmaceutically active “prils”, therefore suggest-
ingitself for industrial application on a large scale.
prisingly, A111GV360I also resulted in
a moderately
increased catalytic rate. Compared to other muteins, however,
the turnover rate of A111GV360I decreased duringthe
course of the reaction.
All muteins with increased enantioselectivity for the
synthesis of 1b were also tested in the conversion of
15 mmol of the more rigid 2a to 2b (Table 3). Two muteins
were also more selective with 2a, reaching ꢀ 98% ee with
mutation V360I. In addition all mutations except V360M
caused a faster conversion of 2a than with the original enzyme
PaHNL5. The higher rigidity of 2a compared to 1a and the
larger volume of the methionine residue are likely reasons for
the weaker activity of the V360M variant. With only 27 mgof
PaHNL5-V360I per mmol 2a, conversion to 2b was almost
complete after 3 h with an excellent enantiomeric excess of
ꢀ 98%. In contrast, Gerrits and co-workers reported a
reaction time of 168 h for 97% yield and 98% ee.[15] Recently
also the Sheldon lab reported highly enantioselective and fast
reactions by employing cross-linked enzyme aggregates of
native almond HNL.[16]
Received: February 4, 2005
Published online: July 6, 2005
Keywords: ACE inhibitors · asymmetric synthesis · biocatalysis ·
.
hydroxynitrile lyase · protein engineering
Structure-guided design was employed to generate
recombinant PaHNL5 variants suitable for large-scale stereo-
selective synthesis of aromatic cyanohydrins in water-based
systems. From 24 desigend enzyme variants, 12 of which were
prepared, four well-expressed, sit-specific mutants of the
Pahnl5 gene with significantly increased stereoselectivity
were obtained. All muteins were stable at low pH. As a
highlight, the variant PaHNL5-V360I exhibited extraordinary
catalytic properties. The enantiomeric excess for the con-
version of 1a was improved to > 96% despite a very low
enzyme/substrate ratio. This particular mutein reduced the
[1] a) A. Glieder, R. Weis, W. Skranc, P. Poechlauer, I. Dreveny, S.
Majer, M. Wubbolts, H. Schwab, K. Gruber, Angew. Chem. 2003,
115, 4963 – 4966; Angew. Chem. Int. Ed. 2003, 42, 4815 – 4818;
b) M. H. Fechter, H. Griengl, Food Technol. Biotechnol. 2004,
42, 287 – 294.
[2] a) M. Bauer, R. Geyer, M. Boy, H. Griengl, W. Steiner, J. Mol.
Catal. B 1998, 5, 343 – 347; b) U. Kragl, U. Niedermeyer, M. R.
Kula, C. Wandrey, Ann. N. Y. Acad. Sci. 1990, 613, 167 – 175.
[3] a) F. Effenberger, B. Horsch, F. Weingart, T. Ziegler, S. Kuhner,
Tetrahedron Lett. 1991, 32, 2605 – 2608; b) F. Effenberger,
Angew. Chem. 1994, 106, 1609 – 1619; Angew. Chem. Int. Ed.
Engl. 1994, 33, 1555 – 1564.
Table 2: Conversion and enantioselectivityof selected muteins and the original PaHNL5 for the hydrocyanation of 1a.[a]
Reaction time
1 h
[b] [%]
2 h
4 h
Conv.[b] [%]
EntmryEenzy
Conv.
ee [%]
Conv.[b] [%]
ee [%]
ee [%]
TOF[c] [sÀ1
]
1
2
3
4
5
PaHNL5
A111GV360I
L331A
V360M
V360I
72
70
78
78
86
89.0
90
90.7
93.6
96.0
87
83
90
93
96
90.1
90.6
93.3
94.0
96.6
96
94
95
97
98
90.2
91.8
92.9
94.6
96.7
2497Æ208
3379Æ208
2865Æ312
6812Æ837
14397Æ416
[a] Reaction conditions: 15 mmol of 1a and 1 mg of enzyme (67 mg enzymemmolÀ1 substrate), 108C, pH 3.4. [b] Conversions. [c] TOF: turnover
frequency.
Table 3: Conversion and enantioselectivityof selected muteins and the parent PaHNL5 for the hydrocyanation of 2a.[a]
Reaction time
1 h
2 h
3 h
[b]
EnmtreyEnzy
Conv.
ee
Conv.[b]
ee
Conv.[b]
ee
TOF[c] [sÀ1
]
1
2
3
4
5
PaHNL5
V360M
L331A
A111GV360I
V360I
13
9
19
21
70
95.2
90.3
94.5
95.8
98.0
22
14
29
36
90
96.2
92.7
95.4
96.8
97.9
30
18
39
47
97
96.4
92.6
95.7
96.8
97.6
110Æ19
n.d.[d]
n.d.[d]
149Æ16
458Æ32
[a] Reaction conditions: 15 mmol of 2a and 0.4 mg of enzyme (27 mg enzymemmolÀ1 substrate), 108C, pH 3.4). [b] Conversion and ee are shown in
percent. [c] TOF: turnover frequency. [d] n.d.: not determined since the ee was not improved.
Angew. Chem. Int. Ed. 2005, 44, 4700 –4704
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4703

Communications
[4] a) I. Dreveny, K. Gruber, A. Glieder, A. Thompson, C. Kratky,
Structure 2001, 9, 803 – 815; b) I. Dreveny, C. Kratky, K. Gruber,
Protein Sci. 2002, 11, 292 – 300.
[5] R. K. Tikare, WO 0242244A, 2004 [Chem. Abstr. 2002, 136,
401540].
[6] A. Matsuyama, I. Takase, Y. Ueda, Y. Kobayashi, US 5429935,
1995 [Chem. Abstr. 1995, 123, 141884].
[7] a) A. Schummer, H. Yu, H. Simon, Tetrahedron 1991, 47, 9019 –
9034; b) H. Yu, H. Simon, Tetrahedron 1991, 47, 9035 – 9052.
[8] M. Kottenhahn, K. Drauz (Degussa AG), US 5616727, 1997
[Chem. Abstr. 1997, 126, 305791].
[9] a) H. U. Blaser, E. Schmidt, Asymmetric Catalysis on Industrial
Scale: Challenges, Approaches and Solutions, VCH, Weinheim,
2004, pp. 91 – 103; b) M. Studer, S. Burkhardt, A. F. Indolese, H.-
U. Blaser, Chem. Commun. 2000, 1327 – 1328.
[10] a) O. May, P. T. Nguyen, F. H. Arnold, Nat. Biotechnol. 2000, 18,
317 – 320; b) K. E. Jaeger, T. Eggert, Curr. Opin. Biotechnol.
2004, 15, 305 – 313; c) B. Lingen, J. Groetzinger, D. Kolter, M.-R.
Kula, M. Pohl, Protein Eng. 2002, 15, 585 – 593.
[11] a) H. Ayguen, S. Wojczewski, S. Rosmus (Biospring GmbH), DE
10251547, 2004 [Chem. Abstr. 2002, 140, 401356]; b) R. Weis, P.
Poechlauer, R. Bona, R. Luiten, M. Wubbolts, H. Schwab, A.
Glieder, J. Mol. Catal. B 2004, 29, 211 – 218.
[12] B. Steipe, Curr. Top. Microbiol. Immunol. 1999, 243, 55 – 86.
[13] H. Buhler, F. Effenberger, S. Forster, J. Roos, H. Wajant,
ChemBioChem 2003, 4, 211 – 216.
[14] M. Chen-Goodspeed, M. A. Sogorb, F. Wu, F. M. Raushel,
Biochemistry 2001, 40, 1332 – 1339; M. T. Reetz, F. Daligault, B.
Brunner, H. Hinrichs, A. Deege, Angew. Chem. 2004, 116, 4170 –
4173; Angew. Chem. Int. Ed. 2004, 43, 4078 – 4081.
[15] J. P. Gerrits, W. F. Willeman, A. J. J. Straathof, J. J. Heijnen, J.
Brussee, A. van der Gen, J. Mol. Catal. B 2001, 15, 111 – 121.
[16] L. M. van Langen, R. P. Selassa, F. van Rantwijk, R. A. Sheldon,
Org. Lett. 2005, 7, 327 – 329.
[17] a) U. T. Bornscheuer, Curr. Opin. Biotechnol. 2002, 13, 543 – 547;
b) M. T. Reetz, Proc. Natl. Acad. Sci. USA 2004, 101, 5716 –
5722; c) D. Rotticci, J. C. Rotticci-Mulder, S. Denman, T. Norin,
K. Hult, ChemBioChem 2001, 2, 766 – 770.
4704
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 4700 –4704