Artificial metalloenzymes for enantioselective catalysis: The phenomenon of protein accelerated catalysis

Article abstract of DOI:10.1016/j.jorganchem.2004.09.032

We report on the phenomenon of protein-accelerated catalysis in the field of artificial metalloenzymes based on the non-covalent incorporation of biotinylated rhodium-diphosphine complexes in (strept)avidin as host proteins. By incrementally varying the [Rh(COD)(Biot-1)]⁺ vs. (strept)avidin ratio, we show that the enantiomeric excess of the produced acetamidoalanine decreases slowly. This suggests that the catalyst inside (strept)avidin is more active than the catalyst outside the host protein. Both avidin and streptavidin display protein-accelerated catalysis as the protein embedded catalyst display 12.0- and 3.0-fold acceleration over the background reaction with a catalyst devoid of protein. Thus, these artificial metalloenzymes display an increase both in activity and in selectivity for the reduction of acetamidoacrylic acid.

Full text of DOI:10.1016/j.jorganchem.2004.09.032

Journal of Organometallic Chemistry 689 (2004) 4868–4871
www.elsevier.com/locate/jorganchem
Note
Artificial metalloenzymes for enantioselective catalysis:
the phenomenon of protein accelerated catalysis
J e´ r oˆ me Collot, Nicolas Humbert, Myriem Skander, G e´ rard Klein, Thomas R. Ward *
Institute of Chemistry, University of Neuchtel, Av. Bellevaux 51, CP 2, CH-2007 Neuch aˆ tel, Switzerland
Received 18 August 2004; accepted 14 September 2004
Available online 14 October 2004
Abstract
We report on the phenomenon of protein-accelerated catalysis in the field of artificial metalloenzymes based on the non-covalent
incorporation of biotinylated rhodium–diphosphine complexes in (strept)avidin as host proteins. By incrementally varying the
+
Rh(COD)(Biot-1)] vs. (strept)avidin ratio, we show that the enantiomeric excess of the produced acetamidoalanine decreases
[
slowly. This suggests that the catalyst inside (strept)avidin is more active than the catalyst outside the host protein. Both avidin
and streptavidin display protein-accelerated catalysis as the protein embedded catalyst display 12.0- and 3.0-fold acceleration over
the background reaction with a catalyst devoid of protein. Thus, these artificial metalloenzymes display an increase both in activity
and in selectivity for the reduction of acetamidoacrylic acid.
Ó 2004 Elsevier B.V. All rights reserved.
Keywords: Biotin–avidin; Streptavidin; Enantioselective catalysis; Second coordination sphere; Hydrogenation; Artificial metalloenzyme; Bioinor-
ganic chemistry; Protein-accelerated catalysis
1
. Introduction
enzymatic catalysis is complementary to transition met-
al-based systems, although the scope of reactions is lim-
ited to enzymatic processes. From a combinatorial
perspective however, the potential of directed evolution
techniques in optimizing an enzymeÕs selectivity is
unrivalled [7].
We have recently demonstrated that incorporation of
an achiral rhodium–diphosphine moiety into a protein
environment can yield highly enantioselective hydrogen-
ation catalysts [8].
Inspired by the seminal work of Kaiser [9,10], several
groups have developed methods to covalently modify
proteins by incorporating transition-metal catalysts to
yield hybrid catalytic systems with promising properties
[7,10].
The approach we follow relies on non-covalent (i.e.,
supramolecular) interactions between the metal catalyst
and the protein. As no chemical coupling step is re-
quired upon addition of the catalyst precursor to the
protein, the integrity of organometallic species is
Enantioselective catalysis is one of the most efficient
ways to synthesize high-added value enantiomerically
pure organic compounds [1]. In recognition for their
pioneering accomplishments in transition metal medi-
ated enantioselective catalysis, the 2001 Nobel Prize in
chemistry was awarded to Knowles [2], Noyori [3] and
Sharpless [4].
As the subtle details which govern enantioselection
cannot be reliably predicted or computed, catalysis relies
more and more on a combinatorial approach. In the
past couple of years, several groups have made impor-
tant contributions in this field [5].
Biocatalysis offers an attractive alternative for the
synthesis of enantiopure products [6]. In many aspects,
*
Corresponding author. Tel.: +41 32 718 2516; fax: +41 32 718
2511.
E-mail address: thomas.ward@unine.ch (T.R. Ward).
0
022-328X/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jorganchem.2004.09.032

J. Collot et al. / Journal of Organometallic Chemistry 689 (2004) 4868–4871
4869
warranted [11]. As suggested by Whitesides, the biotin–
avidin system [12] offers an attractive scaffold to perform
such experiments [13,14].
The principle of the biotin–avidin technology (often
referred to as molecular velcro) relies on the extraordi-
nary affinity of biotin for either avidin or streptavidin
1
4
ꢁ1
(
K ꢀ 10
M ) [12]. Most importantly, it is generally
a
accepted that derivatization of the valeric acid side chain
of biotin does not affect significantly the strength of the
biotin–avidin interaction as most stabilizing contacts are
located on the bicyclic framework of (+)-biotin [12]. In
the past twenty years, the biotin–avidin technology has
found numerous applications in various fields of bio-
technology including ELISA, immunolabeling, affinity
targeting, drug delivery etc. [12]. In the context of this
work, it should be emphasized that both avidin and
streptavidin are tetrameric proteins constituted by four
identical subunits which can bind up to four biotin mol-
ecules with no binding cooperativity.
Scheme 2.
Performing the reduction of acetamidoacrylic acid
in avidin as host protein, the rhodium coenzyme
[
quantitative yield with 39% ee.
As the catalyst devoid of its protein environment
yields essentially racemic product under identical condi-
tions, we reasonned that it is imperative to ensure that
catalysis proceeds exclusively within the protein
the problem of Coulomb repulsion between the host
protein and the cationic catalyst precursor. The reaction
performed under identical conditions afforded (R)-ace-
tamidoalanine in nearly quantitative yield with 94% ee
[8]. Upon lowering the pH to 4.0, the reaction proceeded
quantitively (Scheme 2).
+
Rh(COD)(Biot-1)] affords (S)-acetamidoalanine in
(
Scheme 1). As avidin possesses an isoelectric point
pI = 10.4, we speculated that at pH 7.0, the cationic cat-
alyst may display a significantly reduced affinity for the
positively charged avidin. The modest ee may thus be a
reflection of two competing catalytic cycles:
2. Results and discussion
Both in the fields of homogeneous [15] and heteroge-
neous [16] catalysis, the concept of ligand acceleration
has proven very valuable. In ligand-accelerated cataly-
sis, the addition of a ligand increases the reaction rate
of a catalytic transformation which proceeds even in
the absence of added ligand.
(
i) inside the protein which affords an highly enantio-
enriched product;
ii) outside the protein which affords racemic product.
(
In the area of artificial metalloenzymes, the same con-
+
cept may apply as the ‘‘coenzyme’’ [Rh(COD)(Biot-1)]
Streptavidin, which displays only 30% sequence
homology with avidin possesses a much lower isolectric
point (pI = 6.4) [12]. Using streptavidin may thus solve
was shown to be active (and unselective) in the absence
of host protein (Scheme 1). As the interaction between
the biotinylated moiety and the host protein is non-
covalent, it is inherently reversible. The amount bound
and free catalyst (abbreviated lbound and lfree, respec-
tively) is determined by the stability constant of the
+
supramolecular edifice [Rh(COD)(Biot-1)] –(strept)avi-
din and the individual concentrations (We use LehnÕs
notation for host–guest interactions [17]; (strept)avidin
denotes both avidin and streptavidin) [18]. Because of
the very low solubility of acetamidoacrylate, we could
not perform a thorough kinetic analysis to determine
the individual rate constants kcat–prot and kcat for the
reaction within the protein and outside the protein,
respectively. Assuming that both competing catalytic
cycles (within and outside (strept)avidin) proceed
according to the same mechanism, we reasonned that
we could determine the relative rates kcat–prot:kcat by
Scheme 1.

4870
J. Collot et al. / Journal of Organometallic Chemistry 689 (2004) 4868–4871
+
incrementally varying the [Rh(COD)(Biot-1)] to (str-
ept)avidin ratio. By plotting the enantiomeric excess of
the product as a function of this ratio, it is should be
possible to estimate the extent of protein-acceleration
As this procedure produces a single equation with two
unknowns kcat–prot and kcat, we can only extract the ratio
rather than the individual rate constants.
If no protein acceleration is operative, k
= kcat.
cat–prot
+
(
Fig. 1).
Hydrogenation experiments were performed using
both avidin and streptavidin loaded with varying
At eight equivalents [Rh(COD)(Biot-1)] vs. protein,
lbound = lfree = 4 as (strept)avidin possesses four bind-
ing sites. Since a = 0.97 (i.e., 94% ee (R) for [Rh(COD)
+
+
amounts of [Rh(COD)(Biot-1)] . The [Rh(COD)(Biot-
1
(Biot-1)] –streptavidin) and b = 0.50 (racemic material
+
produced by [Rh(COD)(Biot-1)] ), the % (R) predicted
+
)] -to-protein ratio was varied incrementally from 1:1
to 10:1 and the enantiomeric excess is plotted against
this ratio, Fig. 1. The reactions were performed as de-
scribed in [8]. Two noteworthy features emerge:
at eight equivalents and with no protein acceleration is
% (R) = 73.5 (i.e., 47% ee (R)). The broken lines in
Fig. 1 correspond to the ee calculated with no protein
acceleration, assuming that both reactions proceed
according to the same mechanism.
(
(
i) protein acceleration and
ii) cooperativity between the four catalytic sites.
By performing a least square minimization on the cal-
culated and measured % ee for a given kcat–prot:kcat ratio,
we estimate the relative rates. For streptavidin and avi-
din, we compute a kcat–prot:kcat ratio of 2.95 and 11.96,
respectively.
(i) Both avidin and streptavidin display protein accel-
erated catalysis. The extent of protein acceleration in
streptavidin (where the (R)-product is produced prefer-
entially) can be estimated by
+
(ii) Increasing the [Rh(COD)(Biot-1)] -to-protein ra-
tio from 1:1 to 4:1 (i.e., all four biotin binding sites occu-
pied) leads to a slight increase in enantioselectivity (from
k
cat–prot ꢂ l
ꢂ a þ kcat ꢂ l ꢂ b ;
free
free
bound
%
ðRÞcalculated
¼
ð1Þ
k
cat–prot ꢂ lbound þ kcat ꢂ l
8
9% to 94% ee (R) for streptavidin and from 34% to 39%
where kcat–prot and kcat are the rate constants for the
reaction within the protein and outside the protein,
respectively, lbound and lfree are the number of pro-
(S) for avidin). This suggests that there is a slight coop-
erativity on the enantioselectivity between the four bio-
tin binding sites.
+
tein-bound complexes ([Rh(COD)(Biot-1)] –(strept)avi-
+
din) and protein-free [Rh(COD)(Biot-1)] moieties; a
+
and b are the % (R) produced by [Rh(COD)(Biot-1)] –
3. Conclusion
+
strept)avidin and by [Rh(COD)(Biot-1)] , respectively.
(
+
By incrementally varying the [Rh(COD)(Biot-1)] -to-
strept)avidin ratio, we have unraveled the phenomenon
(
of protein-accelerated catalysis, similar to ligand accel-
erated catalysis. We have shown that the hydrogenation
of acetamidoacrylic acid within artificial metalloenzyme
proceeds at a faster rate than the corresponding hydro-
genation outside of the protein.
We believe that this phenomenon is caused by the
affinity of the hydrophobic substrate for the hydropho-
bic catalytic pocket within the host protein. Although
the ‘‘active site’’ was by no means optimized to accomo-
date an enantioselective hydrogenation event, it is pleas-
ing to witness an increase both in activity and in
selectivity for this reaction, prototypical of the homoge-
neous catalysis kingdom. Current efforts in the group
are directed towards the determination of individual rate
constants using a more soluble substrate.
Acknowledgements
+
Fig. 1. Enantioselectivity as a function of the [Rh(COD)(Biot-1)] -to-
This work was funded by the Swiss National Science
Foundation, CERC3 and the Canton of Neuch aˆ tel. We
thank Belovo Egg Science and Technology for a gener-
ous gift of egg white avidin. We thank C.R. Cantor for
the streptavidin gene, J.-M. Neuhaus, P. Sch u¨ rmann
(strept)avidin ratio. The triangles correspond to the ee (R enantiomer)
obtained with streptavidin as host protein; the squares correspond to
the ee (S enantiomer) obtained with avidin as host protein. The broken
lines represent the theoretical ee with no protein acceleration,
emphasizing the effect of the protein on the rate of the reduction.

J. Collot et al. / Journal of Organometallic Chemistry 689 (2004) 4868–4871
[9] E.T. Kaiser, D.S. Lawrence, Science 226 (1984) 505.
4871
(
Univ. Neuch aˆ tel) and P. Arosio (Univ. Milano) for
their help in setting up the streptavidin production.
[
[
[
10] D. Qi, C.-M. Tann, D. Haring, M.D. Distefano, Chem. Rev. 101
(
2001) 3081.
11] M. Ohashi, T. Koshiyama, T. Ueno, M. Yanase, H. Fujii, Y.
Watanabe, Angew. Chem. Int. Ed. 42 (2003) 1005.
References
12] M. Wilchek, E.A. Bayer (Eds.), Methods in Enzymology: Avidin–
Biotin Technology, Academic Press, San Diego, 1990.
[
1] E.N. Jacobsen, A. Pfaltz, H. Yamamoto (Eds.), Comprehensive
Asymmetric Catalysis, Springer, Berlin, 1999.
[13] M.E. Wilson, G.M. Whitesides, J. Am. Chem. Soc. 100 (1978)
306.
[
[
[
[
[
2] R. Noyori, Angew. Chem. Int. Ed. 41 (2002) 2008.
3] K.B. Sharpless, Angew. Chem. Int. Ed. 41 (2002) 2024.
4] W.S. Knowles, Angew. Chem. Int. Ed. 41 (2002) 1998.
5] C. Gennari, U. Piarulli, Chem. Rev. 103 (2003) 3071.
6] K. Faber, Biotransformations in Organic Chemistry, fifth ed.,
Springer, Berlin, 2004.
[14] C.-C. Lin, C.-W. Lin, A.S.C. Chan, Tetrahedron Asym. 10 (1999)
1887.
[15] D.J. Berrisford, C. Bolm, K.B. Sharpless, Angew. Chem. Int. Ed.
Engl. 34 (1995) 1059.
[16] M. Garland, H.-U. Blaser, J. Am. Chem. Soc. 112 (1990) 7048.
[17] J.-M. Lehn, Supramolecular Chemistry: Concepts and Perspec-
tives, VCH, Weinheim, 1995.
[
[
7] M.T. Reetz, Proc. Natl. Acad. Sci. USA 101 (2004) 5716.
8] J. Collot, J. Gradinaru, M. Skander, N. Humbert, A. Zocchi,
T.R. Ward, J. Am. Chem. Soc. 125 (2003) 9030;
[18] Although the biotin–(strept)avidin is among the strongest
non-covalent interactions known, we have shown that the
introduction of a bulky and charged rhodium moiety dramat-
ically diminishes the affinity association constant (unpublished
results).
M. Skander, N. Humbert, J. Collot, J. Gradinaru, G. Klein, A.
Loosli, J. Sauser, A. Zocchi, F. Gilardoni, T.R. Ward, J. Am.
Chem. Soc. 126 (2004) in press.