Artificial metalloenzymes through cysteine-selective conjugation of phosphines to photoactive yellow protein

Article abstract of DOI:10.1002/cbic.201000159

(Chemical Equation Presented) Pinning phosphines on proteins: A method for the cysteine-selective bioconjugation of phosphines has been developed. The photoactive yellow protein has been site-selectively functionalized with phosphine ligands and phosphine transition metal complexes to afford artificial metalloenzymes that are active in palladium-catalysed allylic nucleophilic substitution reactions.

Full text of DOI:10.1002/cbic.201000159

DOI: 10.1002/cbic.201000159
Artificial Metalloenzymes through Cysteine-Selective Conjugation of
Phosphines to Photoactive Yellow Protein
Wouter Laan,[a] Bianca K. MuÇoz,[a] Renꢀ den Heeten,[b] and Paul C. J. Kamer*[a]
The embedding of organometallic catalysts into the chiral envi-
ronment of proteins and DNA to develop enantioselective
hybrid transition metal catalysts is attracting increasing atten-
tion.[1] Phosphine ligands are ubiquitous in transition metal
chemistry and can afford extremely reactive and versatile ho-
mogeneous catalysts. Consequently, various efforts have been
made to create hybrid catalysts with this attractive class of
ligands. The covalent embedding of phosphine ligands into
DNA[2] and the application of phosphine-functionalized DNA in
asymmetric catalysis has recently been described.[3] So far, pro-
tein-based artificial metalloenzymes containing phosphine li-
gands have mainly been developed by using noncovalent an-
choring strategies. The introduction of biotinylated phosphines
to (strept)avidin has been a particularly successful approach
for the development of enantioselective artificial metalloen-
zymes.[4] Furthermore, the potential of using antibodies for the
development of hybrid catalysts through supramolecular an-
choring of phosphine catalysts has recently been demonstrat-
ed.[5]
site-selective covalent conjugation of phosphine ligands and
phosphine–transition-metal complexes to a protein, and we
report the application of some of the hybrids in catalysis.
Photoactive yellow protein (PYP) is a small (15 kDa) water-
soluble photoreceptor protein from the bacterium Halorhodo-
spira halophila (Figure 1).[8] The protein has a strong absorb-
ance peak at 446 nm due to its chromophore, p-hydroxycin-
namic acid, which is located in a small hydrophobic binding
pocket and covalently linked by a thioester linkage to the
unique cysteine, Cys69, of the protein.[9] The (recombinant)
PYP apo-protein can be reconstituted in vitro with activated
forms of the p-hydroxycinnamic acid chromophore or other
chromophore derivatives: the use of the thiophenyl ester, the
anhydride and the imidazolide of p-hydroxycinnamic acid can
all lead to highly efficient and selective formation of the de-
sired thioester linkage with the protein.[10]
We decided to explore whether this reconstitution approach
could be adopted for the site-selective coupling of phosphine
ligands to the cysteine of PYP by using phosphino-carboxylic
acids. The only by-products of the activation of a carboxylic
acid with N,N-carbonyldiimidazole (CDI) to form the reactive
imidazolide and subsequent coupling reaction are imidazole
and CO2. Because they are easily removed from the protein
after coupling, we chose to use CDI-activated phosphine
ligands for the protein functionalization. The imidazolides of
phosphino-carboxylic acids 1–7 were synthesised by treating
them with an excess of N,N-carbonyldiimidazole (CDI) in DMF
(Scheme 1). For all ligands, a shift of the signal in the 31P NMR
spectrum occurs upon imidazolide formation, thus allowing
the extent of activation to be monitored by NMR.
In contrast, no examples of artificial metalloenzymes based
on robust covalent phosphine conjugation to a protein have
been reported to date, and so the protein structure space
combined with this class of ligand remains limited. Neverthe-
less, Reetz has modified the active-site serine of a number of
lipases with a diphosphine coupled to a phosphonate inhibitor,
but unfortunately the resulting hybrids turned out to be hy-
drolytically unstable, which hampered application in catalysis.[6]
The unique reactivity of the nucleophilic thiol side chain of
cysteine makes it a very attractive target for site-selective bio-
conjugation to proteins, which has previously been used for
the covalent anchoring of synthetic catalysts. However, be-
cause of the nucleophilic character of phosphines, the most
common methods for cysteine-selective bioconjugation, such
as, disulfide bridge formation or alkylation by using haloaceta-
mides and maleimides are incompatible with phosphines.
Thus, alternative strategies need to be explored for this class
of ligand. De Vries et al. turned to less-nucleophilic phosphites
that could be covalently attached to papain. This resulted in
an active, but nonselective hydrogenation catalyst.[7] Following
a different approach, we have developed for the first time the
Treatment of the PYP with imidazolides of 1–7 afforded in
all cases the desired conjugate in high yield and with excellent
chemoselectivity. The predominant LC-MS signal found for
each conjugate corresponds to PYP containing free phosphine
(Table 1 and Figures S1–S3 in the Supporting Information).
While the bidentate ligands 4 and 7 were the least reactive,
the use of a larger excess still afforded excellent conversion.
The lower reactivity might be due to the increased steric hin-
drance encountered by these bulkier bidentate ligands or a
faster rate of hydrolysis of the imidazolide. Due to their nucleo-
philicity, free phosphines react similarly to free thiols with re-
agents used for the colorimetric detection of thiol groups. This
prohibited the use of such assays to determine the specificity
and efficiency of the coupling reactions. Instead, we relied on
mass spectrometry to determine the extent of modification. Al-
though in most cases a signal corresponding to the unmodi-
fied protein was still observed, the ionization efficiency of each
hybrid was found to be about 80–100-fold less than that of
the parent-protein, therefore revealing that all ligands coupled
with more than 90% efficiency (see the Supporting Informa-
[a] Dr. W. Laan, Dr. B. K. MuÇoz, Prof. Dr. P. C. J. Kamer
School of Chemistry, University of St Andrews
North Haugh, KY16 9ST, St. Andrews (UK)
Fax: (+44)1334463808
E-mail: pcjk@st-andrews.ac.uk
[b] Dr. R. den Heeten
Van ’t Hoff Institute for Molecular Sciences, University of Amsterdam
Nieuwe Achtergracht 166, 1018 WV Amsterdam (NL)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201000159.
1236
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1236 – 1239

Table 1. Conjugation of phosphines 1–7 to PYP WT and PYP R52G.
Protein
Ligand
Conversion [%][a]
MWcalcd
MWobs
WT
WT
WT
WT
WT
WT
WT
R52G
R52G
R52G
R52G
1
2
3
4
5
6
7
1
2
3
4
>90
>90
>90
>90
>90
>90
>90
>90
>90
>90
>90
16150.1
16150.1
16150.1
16386.0
16116.9
16144.9
16468.4
16050.7
16050.7
16050.7
16286.6
16150.0
16149.9
16151.1
16385.6
16117.1
16143.4
16466.9
16050.3
16048.8
16052.9
16287.8
[a] Estimated from relative LC-MS signal intensities.
In holo-PYP, Arg52 shields the binding pocket from the sol-
vent.[11] Substituting a glycine for this residue is expected to
lead to increased accessibility and could afford artificial metal-
loenzymes with different activity and selectivity. Similar conju-
gation results were indeed obtained when the mutant protein
PYP R52G was used, thus indicating that the conjugation
methodology is not restricted to the wild-type (WT) protein.
Further analysis of selected conjugates by 31P {1H} NMR spec-
troscopy confirmed the introduction of the phosphine ligands
as free phosphines; in each case, a relatively broad peak with a
chemical shift typical for the respective free phosphine was
found. For example, PYP WT-1 displayed a signal at d=
À6.3 ppm (Figure 2, top), while PYP R52G-4 showed a signal at
d=À20 ppm (Figure S4), these are typical values for free tri-
phenylphosphine derivatives. To the best of our knowledge,
this is the first demonstration of covalent, site-selective conju-
gation of free phosphines to a protein host providing a long-
term stable linkage.
Scheme 1. Synthesis of phosphine-PYP conjugates by imidazolide formation
of phosphine-carboxylic acids. a) DMF, overnight, RT; b) 50 mm Tris-HCl/H2O,
pH 8.0, overnight, RT.
Next, we explored the potential of the established conjuga-
tion method for the synthesis of catalytic palladium-containing
hybrids. Proteins contain numerous potential donor sites for
palladium, as was clearly shown by the coordination of palladi-
Figure 1. Crystal structure of PYP (PDB ID: 2PHY).[9a] The cysteine employed
for bioconjugation and the arginine flanking the binding pocket are high-
lighted.
tion). The amount of double coupling observed was negligible
for all ligands.
To confirm that the conjugates were exclusively formed by
thioester formation, conjugates were incubated either with
dithiothreitol (DTT) or at pH 14, the conditions that result in
cleavage of the thioester linkage in native PYP.[9b,c] Both treat-
ments result in complete disappearance of the LC-MS signal
corresponding to the hybrids (Figures S6–S9), thus confirming
the chemoselective conjugation of the ligands by a thioester
bond to the unique cysteine of PYP. Even though the use of
imidazolides is a classical way to modify proteins through
lysine acylation, the lysines of PYP appear to be sufficiently
unreactive at pH 8 to prevent any competitive amine modifica-
tion. Furthermore, analysis of conjugates stored at 48C at neu-
tral pH for several weeks revealed the linkage to be stable
over prolonged periods under these conditions.
Figure 2. 31P{1H} NMR analysis of phosphine-modified PYP. Top: PYP WT
modified with 1; bottom: PYP R52G modified with 1-Pd(allyl)Cl.
ChemBioChem 2010, 11, 1236 – 1239
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1237

um–allyl moieties at multiple sites of ferritin.[12] To ensure dis-
crete coordination of the transition metal to the phosphine in
the presence of the alternative donor atoms of proteins,[12] we
conjugated preformed palladium complexes of some of the
ligands. Treatment of crude imidazolides of ligands 1, 5 and 6
with [Pd(allyl)Cl]2 (Scheme 2) afforded yellow complexes that
ladium remains bound to the phosphine ligand. Analysis of the
metalloproteins by circular dichroism (CD) spectroscopy re-
vealed that the introduction of the synthetic catalysts has no
significant effect on the structure of the protein (Figure S13).
Allylic substitution is a synthetically important catalytic re-
action for which significant asymmetric induction has been
achieved with various transition-metal-modified macromole-
cules.[13–16] The PYP-based hybrid metalloproteins were evaluat-
ed in palladium-catalysed allylic amination, by employing the
model substrate 1,3-diphenylprop-2-enyl acetate and benzyla-
mine as the nucleophile (Scheme 3 and Table 2). When the re-
actions were performed in aqueous buffered solution, irrepro-
Scheme 2. Formation of palladium–allyl complexes of the imidazolides of
ligands 1 (top), 5 (n=3) and 6 (n=5, bottom). a) DMF, RT, 30 min.
Scheme 3. Asymmetric allylic amination: addition of benzylamine to 1,3-di-
phenylprop-2-enyl acetate.
exhibited broad signals around d=23 ppm in 31P NMR spectra,
which are typical values for this type of phosphine–palladium
complex. By using the same procedure for the conjugation of
the free phosphines, these palladium complexes could also be
selectively and efficiently coupled to the protein, as revealed
by LC-MS (Figures 3, S10 and S11). In each case, a species with
Table 2. Allylic substitution of 1,3-diphenyl-2-propenyl acetate with ben-
zylamine catalysed by phosphine-palladium modified PYP.[a]
Protein Ligand Conversion [%] Protein Ligand Conversion [%]
WT
WT
1
5
70
53
WT
R52G
6
1
88
90
[a] Reaction conditions: 40 nmol PYP-L-Pd, benzylamine/ propenyl ace-
tate/Pd=110:50:1, 50 mm bicine, pH 8.5, 50% DMF, 0.5 cm3, 258C, 24 h.
ducible conversion was observed, this was presumably due to
solubility problems with the substrate and/or inaccessibility of
the metal centre. When the reactions were performed in a mix-
ture of DMF and buffered aqueous solution (1:1), all metallo-
proteins gave good and reproducible conversions to the ami-
nation product, and hydrolysis of the substrate was negligible.
Unfortunately, no enantiomeric excess was obtained. This
could result from (partial) denaturation of the protein due to
the high concentration of organic solvent (the structures of
PYP and PYP R52G unfold above 5% DMF, as determined by
CD, data not shown) causing the metal site to be too far re-
moved from the chiral environment of the protein. No conver-
sion was observed when the reaction was performed with un-
modified protein in the presence of palladium precursor, thus
showing that the observed activity indeed stems from phos-
phine coordinated palladium. Unfortunately, the synthesis of
palladium complexes of ligands 2–4 proved to be difficult to
reproduce. In some cases, red complexes were obtained that
did not couple to the proteins, this is presumably due to the
formation of palladium black.
Figure 3. ESI-MS analysis of PYP WT modified with 6-Pd(allyl)Cl.
a mass corresponding to the protein containing the ligand and
palladium–allyl was observed. The apparent loss of the chlo-
ride either occurs during the LC-MS analysis procedure, or this
ligand is displaced by an amino acid upon formation of the
conjugates. To verify that the procedure affords conjugates in
which the palladium is still coordinated to the phosphine
ligand, R52G modified with 1-Pd(allyl)Cl was characterized by
31P NMR spectroscopy. The conjugate exhibits a broad signal at
d=23 ppm, (Figure 2, bottom) almost identical to the shift
found for the unconjugated complex, thus confirming that pal-
In summary, we have for the first time established a method
for the cysteine-selective bioconjugation of phosphines and
1238
www.chembiochem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1236 – 1239

J. Am. Chem. Soc. 2003, 125, 9030–9031; c) M. T. Reetz, J. J. Peyralans, A.
Maichele, Y. Fu, M. Maywald, Chem. Commun. 2006, 4318–4320; d) C. C.
Lin, C. W. Lin, A. S. C. Chan, Tetrahedron Asymmetry 1999, 10, 1887–
1893.
have used this method to synthesise artificial metalloenzymes.
We are currently exploring the use of more water-soluble sub-
strates for the allylic amination reactions that would obviate
the need for organic cosolvents and the concomitant protein
denaturation. Furthermore, the established conjugation
method can be applied to the coupling of rhodium–phosphine
complexes (unpublished results), thus providing a platform for
the synthesis of phosphine-based artificial transition metallo-
enzymes for diverse catalytic transformations.
[5] H. Yamaguchi, T. Hirano, H. Kiminami, D. Taura, A. Harada, Org. Biomol.
Chem. 2006, 4, 3571–3573.
[6] M. T. Reetz, M. Rentzsch, A. Pletsch, M. Maywald, Chimia 2002, 56, 721–
723.
[7] L. Panella, J. Broos, J. F. Jin, M. W. Fraaije, D. B. Janssen, M. Jeronimus-
Stratingh, B. L. Feringa, A. J. Minnaard, J. G. de Vries, Chem. Commun.
2005, 5656–5658.
[8] T. E. Meyer, E. Yakali, M. A. Cusanovich, G. Tollin, Biochemistry 1987, 26,
418–423.
[9] a) G. E. Borgstahl, D. R. Williams, E. D. Getzoff, Biochemistry 1995, 34,
6278–6287; b) J. J. Van Beeumen, B. V. Devreese, S. M. Vanbun, W. D.
Hoff, K. J. Hellingwerf, T. E. Meyer, D. E. Mcree, M. A. Cusanovich, Protein
Sci. 1993, 2, 1114–1125; c) W. D. Hoff, P. Dux, K. Hard, B. Devreese, I. M.
Nugteren-Roodzant, W. Crielaard, R. Boelens, R. Kaptein, J. Van Beeu-
men, K. J. Hellingwerf, Biochemistry 1994, 33, 13959–13962.
[10] a) Y. Imamoto, T. Ito, M. Kataoka, F. Tokunaga, FEBS Lett. 1995, 374, 157–
160; b) U. K. Genick, S. Devanathan, T. E. Meyer, I. L. Canestrelli, E. Wil-
liams, M. A. Cusanovich, G. Tollin, E. D. Getzoff, Biochemistry 1997, 36,
8–14.
Acknowledgements
We thank Prof. Dr. D. J. Cole-Hamilton for fruitful discussion. The
plasmids and E. coli strain for expression of PYP and PYP R52G
were kindly provided by Dr. Michael van der Horst and Prof. Dr.
K. J. Hellingwerf (University of Amsterdam). This research was fi-
nanced by the NRSCC and the European Union: NEST Adventure
STREP Project artizymes contract No. FP6-2003-NEST-B3 15471;
Marie Curie Excellence Grants, artizyme catalysis, contract no.
(MEXT-2004-014320), http://europa.eu.int/mariecurie-actions.
[11] U. K. Genick, G. E. O. Borgstahl, K. Ng, Z. Ren, C. Pradervand, P. M. Burke,
ˇ
V. Srajer, T. Teng, W. Schildkamp, D. E. McRee, K. Moffat, E. D. Getzoff, Sci-
ence 1997, 275, 1471–1475.
[12] S. Abe, J. Niemeyer, M. Abe, Y. Takezawa, T. Ueno, T. Hikage, G. Erker, Y.
Watanabe, J. Am. Chem. Soc. 2008, 130, 10512–10514.
[13] P. Fournier, R. Fiammengo, A. Jꢂschke, Angew. Chem. 2009, 121, 4490–
4493; Angew. Chem. Int. Ed. 2009, 48, 4426–4429.
Keywords: bioconjugation
metalloenzymes · transition metals
· catalysis · hybrid catalysts ·
[14] J. Pierron, C. Malan, M. Creus, J. Gradinaru, I. Hafner, A. Ivanova, A.
Sardo, T. R. Ward, Angew. Chem. 2008, 120, 713–717; Angew. Chem. Int.
Ed. 2008, 47, 701–705.
[15] C. A. Christensen, M. Meldal, Chem. Eur. J. 2005, 11, 4121–4131.
[16] A. Agarkov, S. J. Greenfield, T. Ohishi, S. E. Collibee, S. R. Gilbertson, J.
Org. Chem. 2004, 69, 8077–8085.
[1] a) J. Steinreiber, T. R. Ward, Coord. Chem. Rev. 2008, 252, 751–766; b) T.
Ueno, S. Abe, N. Yokoi, Y. Watanabe, Coord. Chem. Rev. 2007, 251, 2717–
2731.
[2] M. Caprioara, R. Fiammengo, M. Engeser, A. Jaschke, Chem. Eur. J. 2007,
13, 2089–2095.
[3] L. Ropartz, N. J. Meeuwenoord, G. A. van der Marel, P. W. N. M. van Leeu-
wen, A. M. Slawin, P. C. J. Kamer, Chem. Commun. 2007, 1556–1558.
[4] a) M. E. Wilson, G. M. Whitesides, J. Am. Chem. Soc. 1978, 100, 306–307;
b) J. Collot, J. Gradinaru, N. Humbert, M. Skander, A. Zocchi, T. R. Ward,
Received: March 10, 2010
Published online on April 29, 2010
ChemBioChem 2010, 11, 1236 – 1239
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1239