X-ray structure and designed evolution of an artificial transfer hydrogenase

Article abstract of DOI:10.1002/anie.200704865

A structure is worth a thousand words: Guided by the X-ray structure of an S-selective artificial transfer hydrogenase, designed evolution was used to optimize the selectivity of hybrid catalysts. Fine-tuning of the second coordination sphere of the ruthenium center (see picture, orange sphere) by introduction of two point mutations allowed the identification of selective artificial transfer hydrogenases for the reduction of dialkyl ketones. (Chemical Equation Presented)

Full text of DOI:10.1002/anie.200704865

Communications
DOI: 10.1002/anie.200704865
Artificial Metalloenzymes
X-Ray Structure and Designed Evolution of an Artificial Transfer
Hydrogenase**
Marc Creus, Anca Pordea, Thibaud Rossel, Alessia Sardo, Christophe Letondor, Anita Ivanova,
Isolde LeTrong, Ronald E. Stenkamp,*and Thomas R. Ward*
While the optimization of enzyme performance is a well-
established and active field of research, the creation of
activity out of nothing within an existing protein scaffold
remains a daunting task.[1–3] To overcome this challenge, a
catalytically active organometallic moiety anchored within a
protein scaffold represents an attractive starting point for
both chemical and genetic optimization of the resulting
artificial metalloenzyme.[4–9] Guided by the structure of an
artificial transfer hydrogenase based on biotin–streptavidin
technology, we have implemented a designed evolution
protocol to identify both R- and S-selective variants for
reduction of acetophenone derivatives (up to 96% ee) as well
as dialkyl ketone substrates (up to 90% ee).
exploited to afford artificial metalloenzymes for ester hy-
drolysis,[17] dihydroxylation,[18] sulfoxidation,[19,20] epoxida-
tion,[21,22] and Diels–Alder reactions.[23–25]
The mechanism of the transfer hydrogenation of aromatic
ketones using piano-stool complexes incorporating amino-
sulfonamide ligands is well-established. The enantiodiscrimi-
À
nation event relies on C H···p interactions between the
substrate and the h6-bound arene.[26] Within this context, the
reduction of dialkyl ketones remains challenging.[27] Nature
relies mostly on NAD(P)H-containing enzymes for such
reactions, with the entire second coordination sphere of the
active site in alcohol dehydrogenases being tailored for such
tasks.[28,29]
An early report by Wilson and Whitesides[10] inspired the
use the biotin–streptavidin technology to produce artificial
hydrogenases for the enantioselective reduction of nitrogen-
protected dehydroamino acids[11–14] as well as for the reduc-
tion of acetophenone derivatives by transfer hydrogena-
tion.[15,16] Previous studies on artificial transfer hydrogenases
have allowed the identification of promising systems for the
reduction of acetophenone derivatives: [h6-(benzene)RuCl-
In recent years, there has been an increasing effort to
combine rational design features into Darwinian evolutionary
protocols.[30–34] Designed evolution combines rational deci-
sions on sites of mutations with rounds of screening to perfect
those elements of enzyme function that cannot be predicted.
With this aim, [h6-(benzene)RuCl(Biot-p-L)]ꢀS112K Sav was
crystallized, and its X-ray structure is displayed in Figure 1
(PBD reference code 2QCB). Structural details are collected
in the Supporting Information. In the refined model (1.58-Š
resolution, R = 0.168, Rfree = 0.187), several relevant features
are identified:
1) The piano-stool moiety and the S112K side chain are only
partially occupied (20% and 50%, respectively, Fig-
ure 1a). This finding may be traced back to a) conforma-
tional flexibility of the piano-stool moiety within the host
protein, b) ruthenium decomplexation during cocrystalli-
zation in the presence of 1.0m sodium citrate, or c) a short
contact with a neighboring biotinylated complex (Ru···Ru
separation 4.44 Š) within the streptavidin tetramer that
hinders the occupancy of the adjacent biotin binding site
in this ordered conformation.
2) Short contacts between the Ru complex and amino acids
in several loop regions can be identified (Figure 1b).
3) Incorporation of the bulky biotinylated complex [h6-
(benzene)RuCl(Biot-p-L)] within S112K Sav does not
lead to a major structural reorganization of the host
protein. Compared to wild-type (WT) Sav (PDB reference
code 1STP), a root mean square (RMS) of 0.276 Š is
computed for all 121 Ca atoms present (Figure 1c).
4) Despite the use of a “racemic” piano-stool complex for
crystallization, the configuration at ruthenium is S in the
crystal structure (Figure 1b). Most interestingly, the (S)-
Ru configuration in a homogeneous system leads to (S)-
phenylethanol reduction products,[26] corresponding to the
preferred enantiomer produced with [h6-(benzene)RuCl-
(Biot-p-L)]ꢀS112K Sav.[15]
(Biot-p-L)]ꢀS112K Sav
(Sav = streptavidin,
= N’-(4-Biotinamidophenylsulfonyl)ethylenediamine;
Biotin-p-L
see
Figure 2) and [h6-(p-cymene)RuCl(Biot-p-L)]ꢀS112A Sav
afford (S)- and (R)-phenylethanol reduction products, respec-
tively.[15,16] Alternative anchoring strategies have been
[*] Dr. I. LeTrong, Prof. R. E. Stenkamp[++]
Departments of Biological Structure and Biochemistry and the
Biomolecular Structure Center, University of Washington
Box 357420 Seattle, WA 98195-7420 (USA)
Fax:(+1)206-543-1524
E-mail: stenkamp@u.washington.edu
Dr. M. Creus,[+] A. Pordea,[+] T. Rossel, A. Sardo, Dr. C. Letondor,
Dr. A. Ivanova, Prof. T. R. Ward[++]
Institute of Chemistry, University of Neuchâtel
Av. Bellevaux 51, CP 158, 2009 Neuchâtel (Switzerland)
Fax: (+41)32-718-2511
E-mail: thomas.ward@unine.ch
[+] These authors contributed equally to this work.
[++] Corresponding authors: Prof. Stenkamp for the X-ray structure, Prof.
Ward for all other matters.
[**] This work was funded by the Swiss National Science Foundation
(Grants FN 200021-105192 and 200020-113348), the Roche Foun-
dation as well as the FP6 Marie Curie Research Training network
(MRTN-CT-2003-505020) and the Canton of Neuchâtel. We thank
Umicore Precious Metals Chemistry for a loan of ruthenium. We
thank C. R. Cantor for the streptavidin gene.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
1400
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 1400 –1404

Angewandte
Chemie
Each of the immobilized isoforms was combined with
either [h6-(benzene)RuCl(Biot-p-L)] or [h6-(cymene)RuCl-
(Biot-p-L)] and tested for the reduction of ketones 1 and 3
(Scheme 1). Although the purification and immobilization
Figure 1. X-ray crystal structure of [h6-(benzene)RuCl(Biot-p-L)]ꢀS112K
Sav. a) Close-up view (only monomer B (blue) occupied by the
biotinylated catalyst (ball-and-stick representation); monomers A
(green), C (orange), and D (yellow)). b) Highlight of amino acid side-
chain residues displaying short contacts with Ru. The absolute config-
uration at ruthenium is S. c) Superimposition of the structure of [h6-
(benzene)RuCl(Biot-p-L)]ꢀS112K Sav with the structure of biotinꢀcore
streptavidin (PDB reference code 1STP, only monomers A and B
displayed for clarity; biotin: white stick, core streptavidin: white tube).
6
À
d) Ru Ca distances extracted from the X-ray structure of [h -
Scheme 1. Substrates, reduction products, and operating conditions
used for the designed evolution of artificial transfer hydrogenases. h6-
arene=benzene, p-cymene; Sav mutant: K121X, L124X, S112A K121X,
S112K K121X, S112A L124X, S112K K124X. The catalytic runs were
performed at 558C for 64 h using the mixed buffer NaO2CH (0.48m),
B(OH)3 (0.41m), and 3-(N-morpholino)propanesulfonic acid (MOPS,
0.16m) at pHinitial 6.25. Ru/substrate/formate ratio 1:100:4000
(benzene)RuCl(Biot-p-L)]ꢀS112K Sav; monomers: A black, B blue,
C green, and D red.
5) Substituting the capping h6-benzene ligand with the
bulkier h6-p-cymene may force the latter biotinylated
complex to adopt a different position or configuration at
the metal center to avoid steric clash between the h6-arene
and the amino acid residues 112KA and 121KA.
with biotin–sepharose led to an erosion of both the activity
and the selectivity (Table 1), this protocol significantly
hastened the screening process to identify trends. The results
obtained with the immobilized artificial metalloenzymes are
summarized in the form of a fingerprint (Figure 2). The most
promising results were reproduced using the purified non-
immobilized catalyst in the presence of representative
substrates (Scheme 1 and Table 1, see the Supporting
Information for a comprehensive list of all catalytic runs).
On the basis of this screening, several interesting features
can be identified:
On the basis of this structure, we selected positions K121
and L124 for saturation mutagenesis, using the genetic
backgrounds of WT Sav, S112A Sav, or S112K Sav to afford
a total of 117 Sav isoforms (See the Supporting Information).
Position K121 was of particular interest, as both K121A and
K121B residues (A and B refer to the respective monomers;
see Figure 1) may interact both with the h6-arene (i.e. K121A)
and with the incoming substrate (i.e. K121B). We targeted
residue L124, as we speculated that saturation mutagenesis at
this position may subtly alter the position of the biotinylated
catalyst (H3CLeu···OSORRLigand 3.53 Š).
To accelerate the optimization process, a straightforward
extraction–immobilization protocol with biotin–sepharose
was implemented to capture functional Sav from crude
cellular extracts.[14] Indeed, we hypothesized that, owing to
the high affinity of biotin for streptavidin, one biotin-binding
site could be used for immobilization, leaving up to three
binding sites to accommodate the biotinylated catalyst.
1) Irrespective of the substrate, the nature of the h6-arene
plays an important role in determining the preferred
enantiomer (Figure 2).[15,16] Strikingly, substitution of h6-
benzene for h6-p-cymene in the presence of S112A K121N
Sav affords near mirror images for the reduction of both
substrates 1 and 3 (Table 1 entries 4–7).
2) Both substrates 1 and 3 behave similarly in terms of
selectivity (Table 1 and Figure 3b). While the enantiose-
lectivity for the reduction of aryl ketone derivatives can be
Angew. Chem. Int. Ed. 2008, 47, 1400 –1404
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1401

Communications
Table 1: Summary of selected results for the catalytic experiments with
chain at position S112 may allow increased influence of
beneficial K121X or L124X mutations.
either biotin–sepharose immobilized or purified homogeneous artificial
metalloenzymes [h6-(arene)RuH(Biot-p-L)]ꢀSav.
4) Overall, saturation mutagenesis at position K121 is more
effective for the optimization of enantioselectivity than
saturation mutagenesis at position L124 (Figure 3d). We
speculate that this effect may be due to the influence of the
K121X side chains both on the piano-stool complex itself
(K121XA) and on the trajectory of the incoming prochiral
substrate (K121XB; Figure 1b). This latter interaction may
be particularly important for the challenging dialkyl
ketone substrates 3 and 11, as it provides a second-
coordination-sphere interaction with the substrate, remi-
niscent of natural enzymes.
5) Methyl alkyl and methyl aryl ketones afford good levels of
conversion and selectivity (up to 96% ee (R) for 6, 92% ee
(S) for 8, and 90% ee (R) for 12; Table 1, entries 13–15). In
contrast, ketones bearing a longer alkyl group give
comparatively modest conversions (Table 1, entries 16–
18), suggesting that such artificial metalloenzymes may be
further evolved to display high substrate specificity. Such
substrate specificity is similar to yeast alcohol dehydro-
genase which in general only accepts aldehydes and
methyl ketones as substrates.[29]
Entry Arene
Sav isoform
Ketone
ee (conv [%])
extract[a,b] pure[a,c]
1
2
3
4
5
6
7
8
9
p-cymene WT Sav
1
1
1
1
1
3
3
1
3
1
3
3
5
7
11
9
9
13
65 (81)
87 (93)
benzene
WT Sav
À39 (31) À57 (43)
p-cymene L124V
83 (90)
91 (96)
benzene
S112A K121N
À55 (99) À75 (98)
50 (63) 70 (89)
À62 (95) À72 (quant.)
58 (81) 70 (78)
À59 (64) À65 (94)
82 (84) 88 (99)
p-cymene S112A K121N
benzene
p-cymene S112 AK121N
benzene S112K L124H
p-cymene S112A K121T
K121R
11 p-cymene S112A K121S
12 benzene S112A K121W
13 p-cymene L124V
S112A K121N
10 benzene
À64 (96) À68 (95)
59 (80)
77 (98)
84 (99)
96 (97)
À92 (quant.)
90 (quant.)
87 (20)
80 (86)
–
–
–
–
–
–
14 benzene
15 p-cymene S112A K121T
16 p-cymene L124V
17 benzene
18 p-cymene S112A K121T
S112A K121N
S112A K121N
À92 (54)
46 (50)
[a] Positive and negative ee values correspond to the R and S enantiom-
ers, respectively. [b] Immobilized protein from crude cellular extract.
[c] Purified non-immobilized protein. –=not performed.
In summary, the struc-
tural characterization of
[h6-(benzene)RuCl(Biot-p-
L)]ꢀS112K Sav
has
allowed us to implement a
designed evolution proto-
col for the optimization of
artificial transfer hydroge-
nases.
immobilization step using
biotin–sepharose, per-
A straightforward
formed on crude cellular
extracts, allowed the iden-
tification
of
[h6-(p-
cymene)RuH(Biot-p-
L)]ꢀS112A K121T Sav for
the enantioselective reduc-
tion of dialkyl ketones (up
to 90% ee for 12; Table 1,
entry 15). This artificial
metalloenzyme is attained
by an evolutionary path
involving a combination of
two poorly adapted muta-
Figure 2. Fingerprint display of the results for the chemogenetic optimization of the reduction of ketones 1
and 3 in the presence of biotin–sepharose-immobilized artificial metalloenzymes [h6-(arene)RuH(Biot-p-
L)]ꢀSav mutant. Catalytic runs which could not be performed (insufficient soluble protein expression, see the
Supporting Information) are represented by white triangles.
tions that would not nor-
mally be selected individu-
ally (Figure 3e). Interac-
optimized to levels of greater than 90% ee with Sav
isoforms bearing a single point mutation,[15,16] a designed
evolution protocol was required to identify Sav double
mutants that afford dialkyl reduction products in up to
90% ee.
tion between the host protein and the substrate clearly
contributes to enantioselectivity. This notion is consistent
with the evolution of artificial transfer hydrogenases for
À
dialkyl ketones, which cannot rely on a C H···p interaction as
an enantiodiscriminating element. Thus, screening for enan-
tioselectivity likely optimizes interactions between substrate
and the second coordination sphere provided by the protein,
exerting a strong evolutionary pressure toward substrate
3) Artificial metalloenzymes incorporating the S112A muta-
tion give better enantioselectivities than the systems
bearing the S112K mutation (Figure 3c). A small side
1402
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 1400 –1404

Angewandte
Chemie
Figure 3. Enantioselectivity trends of artificial transfer hydrogenases. Graphical summary for the immobilized protein arranged according to:
6
~
~
^
^
a) The nature of the capping h -arene: benzene ( ) and p-cymene ( ). b) The nature of the products 2 ( ) or 4 ( ). c) The nature of the S112X
~
~
^
*
^
residue: S112A ( ), S112K ( ), S112 ( ). d) The position of the site of mutation: K121X ( ) or L124X ( ). e) Reconstructed evolutionary path of
the best R- and S-selective hybrid catalysts for the reduction of 4-phenyl-2-butanone 3. Only the best single mutations and those present in the
best double mutants are listed.
[9] M. Breuer, K. Ditrich, T. Habicher, B. Hauer, M. Kesseler, R.
Stürmer, T. Zelinski, Angew. Chem. 2004, 116, 806; Angew.
Chem. Int. Ed. 2004, 43, 788.
[10] M. E. Wilson, G. M. Whitesides, J. Am. Chem. Soc. 1978, 100,
306.
specialization. Such specialized artificial enzymes, obtainable
by designed evolution with modest screening efforts, hold
great potential for applications in both white and red
biotechnology.
[11] C.-C. Lin, C.-W. Lin, A. S. C. Chan, Tetrahedron: Asymmetry
1999, 10, 1887.
[12] G. Klein, N. Humbert, J. Gradinaru, A. Ivanova, F. Gilardoni,
U. E. Rusbandi, T. R. Ward, Angew. Chem. 2005, 117, 7942;
Angew. Chem. Int. Ed. 2005, 44, 7764.
[13] M. T. Reetz, J. J.-P. Peyeralans, A. Maichele, Y. Fu, M. Maywald,
Chem. Commun. 2006, 4318.
[14] U. E. Rusbandi, C. Lo, M. Skander, A. Ivanova, M. Creus, N.
Humbert, T. R. Ward, Adv. Synth. Catal. 2007, 349, 1923.
[15] C. Letondor, N. Humbert, T. R. Ward, Proc. Natl. Acad. Sci.
USA 2005, 102, 4683.
Received: October 19, 2007
Published online: January 4, 2008
Keywords: asymmetric catalysis · chemogenetic optimization ·
.
directed evolution · hydrogenation · metalloenzymes
[1] M. A. Dwyer, L. L. Looger, H. W. Hellinga, Science 2004, 304,
1967.
[2] H.-S. Park, S.-H. Nam, J. K. Lee, C. N. Yoon, B. Mannervik, S. J.
Benkovic, H.-S. Kim, Science 2006, 311, 535.
[3] B. Seelig, J. W. Szostak, Nature 2007, 448, 828.
[4] D. Qi, C.-M. Tann, D. Haring, M. D. Distefano, Chem. Rev. 2001,
101, 3081.
[16] C. Letondor, A. Pordea, N. Humbert, A. Ivanova, S. Mazurek,
M. Novic, T. R. Ward, J. Am. Chem. Soc. 2006, 128, 8320.
[17] R. R. Davies, M. D. Distefano, J. Am. Chem. Soc. 1997, 119,
11643.
[18] T. Kokubo, T. Sugimoto, T. Uchida, S. Tanimoto, M. Okano, J.
Chem. Soc. Chem. Commun. 1983, 769.
[19] F. van de Velde, L. Kꢀnemann, F. van Rantwijk, R. A. Sheldon,
Chem. Commun. 1998, 1891.
[5] C. M. Thomas, T. R. Ward, Chem. Soc. Rev. 2005, 34, 337.
[6] C. Letondor, T. R. Ward, ChemBioChem 2006, 7, 1845.
[7] Y. Lu, Angew. Chem. 2006, 118, 5714; Angew. Chem. Int. Ed.
2006, 45, 5588.
[20] A. Mahammed, Z. Gross, J. Am. Chem. Soc. 2005, 127, 2883.
[21] K. Okrasa, R. J. Kazlauskas, Chem. Eur. J. 2006, 12, 1587.
[8] T. Ueno, T. Koshiyama, S. Abe, N. Yokoi, M. Ohashi, H.
Nakajima, Y. Watanabe, J. Organomet. Chem. 2007, 692, 142.
Angew. Chem. Int. Ed. 2008, 47, 1400 –1404
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1403

Communications
[22] A. Fernandez-Gacio, A. Codina, J. Fastrez, O. Riant, P.
[29] K. Faber, Biotransformations in Organic Chemistry, 5th ed.,
Springer, Berlin, 2004.
[30] A. Schmid, J. S. Dordick, B. Hauer, A. Kiener, M. Wubbolts, B.
Witholt, Nature 2001, 409, 258.
Soumillon, ChemBioChem 2006, 7, 1013.
[23] G. Roelfes, B. L. Feringa, Angew. Chem. 2005, 117, 3294; Angew.
Chem. Int. Ed. 2005, 44, 3230.
[24] M. T. Reetz, N. Jiao, Angew. Chem. 2006, 118, 2476; Angew.
Chem. Int. Ed. 2006, 45, 2416.
[31] H. E. Schoemaker, D. Mink, M. G. Wubbolts, Science 2003, 299,
1694.
[25] G. Roelfes, A. J. Boersma, B. L. Feringa, Chem. Commun. 2006,
635.
[32] C. A. Voigt, S. L. Mayo, F. H. Arnold, Z.-G. Wang, Proc. Natl.
Acad. Sci. USA 2001, 98, 3778.
[26] M. Yamakawa, I. Yamada, R. Noyori, Angew. Chem. 2001, 113,
2900; Angew. Chem. Int. Ed. 2001, 40, 2818.
[27] A. Schlatter, M. K. Kundu, W.-D. Woggon, Angew. Chem. 2004,
116, 6899; Angew. Chem. Int. Ed. 2004, 43, 6731.
[28] L. A. LeBrun, D.-H. Park, S. Rmaswamy, B. W. Plapp, Biochem-
istry 2004, 43, 3014.
[33] M. T. Reetz, M. Bocola, J. D. Carballeira, D. Zha, A. Vogel,
Angew. Chem. 2005, 117, 4264; Angew. Chem. Int. Ed. 2005, 44,
4192.
[34] M. T. Reetz, Nat. Protoc. 2007, 2, 891.
1404 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 1400 –1404