Direct screening for phosphatase activity by turnover-based capture of protein catalysts.

Full text of DOI:10.1002/1521-3773(20020301)41:5<775::AID-ANIE775>3.0.CO;2-F

COMMUNICATIONS
F
Direct Screening for Phosphatase Activity by
Turnover-Based Capture of Protein Catalysts**
S
H
F
S
surface
hydrolysis
Pi
surface
CHF2
S
S
–O
– O
O
– O
P
Jason R. Betley, Sandro Cesaro-Tadic,
Abdelaziz Mekhalfia, James H. Rickard,
Hazel Denham, Lynda J. Partridge,
O
N
O
N
H
O
H
3
Andreas Pl¸ckthun,* and G. Michael Blackburn*
protein
protein
S
Nu:
H
In studies towards the generation and identification of
catalytic antibodies with phosphate monoesterase activity, we
have developed a direct selection approach for covalent
modification of catalysts founded on mechanism-based phos-
phatase inhibitors. It builds on the work of Halazy et al.[1] in
the use of o- and p-fluoromethylphenols to generate quinone
methides as suicide substrates for glucosidases. Withers and
co-workers conceived the use of 4-difluoromethylphenyl
phosphate 1 as a suicide substrate for a human prostatic acid
F
H
Nu
F
S
surface
S
– O
O
S
N
O
surface
N
H
O
4
H
Scheme 1. Mechanism of covalent binding of the suicide substrate 3 by
formation of quinone methide 4. Following phosphate hydrolysis, the
protein catalyst becomes covalently attached to a carrier surface.
the surface of a solid support and also subsequent release of
the trapped protein species by disulfide reduction with
dithiothreitol. Covalent attachment to the protein results
from the formation of a reactive quinone methide[5] species 4
by elimination of fluoride after protein-catalyzed cleavage of
the phosphate monoester bond. The catalytic protein (anti-
body or enzyme) then becomes covalently trapped as a
consequence of the addition of a nucleophile to the quinone
methide at or near the active site.
Substrate 3 was synthesized in six steps from 2-hydroxy-5-
nitrobenzaldehyde (Scheme 2).[6] Phosphorylation of phenol 5
led to the quantitative formation of phosphate triester 6.
Fluorination of aldehyde 6,[7] followed by quenching with ice
N
S
NaO
NaO
O
S
P
O
CHF2
CHF2
1
NaO
NaO
NH
O
P
n
O
O
N
O
H
3 n = 1–3
OH
HO
HO
CHF2
H
O
N
S
S
O
OH
O
N
N
O
H
2
phosphatase and a phosphotyrosine phosphatase,[2] while
Myers and Widlansky described the use of 4-monofluoro-
methylphenyl phosphate as a substrate.[3] This strategy was
later adapted by Janda et al.[4] to pan a phage Fab library of
9 Â 103 members against the bovine serum albumin (BSA)
conjugate of a reactive substrate 2 to select catalytic anti-
bodies with galactosidase activity.
O
P
O
P
EtO
EtO
EtO
EtO
OH
O
O
CHO
a
b
CHF2
CHO
NO2
5
6
7
NO2
NO2
c
Our design of reagent 3 was developed in line with these
examples (Scheme 1). The reactive 2-difluoromethylphenyl
phosphate is linked by a variable spacer arm to a pyridyl-2'-
disulfide. This enables the suicide substrate to be anchored to
O
CHF2
EtO
EtO
H
N
O
P
O
O
P
RO
CHF2
OR
O
d
N
O
S
S
H
N
[*] Prof. A. Pl¸ckthun, S. Cesaro-Tadic
Biochemisches Institut der Universit‰t Z¸rich
Winterthurerstrasse 190, 8057 Z¸rich (Switzerland)
Fax : (‡41)1-635-5712
8
10 R = Et
R = H
NH2
e
3
Scheme 2. Synthesis of 3: a) (EtO)2POCl, Et3N, CH2Cl2; b) Et2NSF3
(2.1 equiv), 08C, 25 min; c) H2, Pd/C (10%); succinic anhydride, iPr2Et,
CH2Cl2; d) 2-PySSCH2CH2NH2 (9), EDC, CH2Cl2; e) Me3SiBr, CH2Cl2,
then MeOH.
E-mail: plueckthun@biocfebs.unizh.ch
Prof. G. M. Blackburn, J. R. Betley, A. Mekhalfia, J. H. Rickard
Krebs Institute, Department of Chemistry
University of Sheffield
Sheffield, S3 7HF (UK)
water, and purification by chromatography on silica gel gave 7.
The nitro group of 7 was catalytically hydrogenated without
affecting the difluoromethyl group, and the resulting aniline 8
was treated with succinic anhydride. The succinyl hemiamide
product was purified by means of silica-gel chromatography
with a trace amount of acetic acid in the eluent. The thiol
exchange moiety 2-PySSCH2CH2NH2 (9), prepared from bis-
2-aminoethyldisulfide dihydrochloride in two steps,[8] was
Fax : (‡44)1142-738-673
E-mail: g.m.blackburn@sheffield.ac.uk
H. Denham, Dr. L. J. Partridge
Department of Molecular Biology and Biotechnology
University of Sheffield
Western Bank, Sheffield, S10 2TN (UK)
[**] This work was supported by BBSRC studentships (to J.B. and H.D.),
by an EPSRC studentship (to J.H.R.), and by an EC TMR studentship
(to S.C.-T.).
Angew. Chem. Int. Ed. 2002, 41, No. 5
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
1433-7851/02/4105-0775 $ 17.50+.50/0
775

COMMUNICATIONS
coupled to 8 with EDC (N-ethyl-N'-dimethylaminopropylcar-
bodiimide) to give disulfide ester 10. Finally, selective
cleavage of the phosphate ethyl esters of 10 with bromotri-
methylsilane[9] gave the suicide substrate 3 (n ˆ 1).[10] Two
higher homologues of 3 were synthesized by using glutaric
anhydride and adipic anhydride, respectively.[11]
BSA was modified with Traut×s reagent[12] to convert some
of the available e-amino groups of surface lysine residues into
thiols. Disulfide exchange with the suicide substrate 3 was
undertaken immediately at 48C. The dephosphorylation of
the BSA 3 conjugate with alkaline phosphatase was deter-
mined by measuring the phosphate release with resorufin
(PiPer Phosphate Assay Kit, Molecular Probes). We found
that all coupled substrate molecules were accessible to the
enzyme and that the turnover compared well with that of the
uncoupled substrate 3. The suicide reagent 3 BSA conjugate
was immobilized onto the surface of standard Maxisorp
ELISA plates by using various coating concentrations (ELI-
SA ˆ enzyme-linked immuno-sorbent assay). Alkaline phos-
phatase was added for various time periods followed by
extensive washing in Tris-buffer pH 9.0 and 0.1m glycine
HCl pH 2.2. Any enzyme that was covalently trapped through
the suicide reaction with 3 was detected by using a murine
anti-alkaline phosphatase antibody and an anti-mouse-HRP
secondary antibody (Sigma) with TMB substrate (3,3',5,5'-
tetramethylbenzidine). Maximum linkage of alkaline phos-
Figure 1. Plot showing turnover-based covalent interaction with enzyme at
three concentrations of suicide substrate 3 (n ˆ 2). The sensorgram shows
change of biosensor response units (RU) with time. Substrate 3 was
immobilized on the thiol-activated chip surface by disulfide exchange
(step a) with intensities of 5, 11, and 20 RU corresponding (according to its
molecular mass) to about 0.1, 0.2, and 0.4 mm surface concentrations on
flow cells 2(green line), 3 (blue line) and 4 (red line), respectively. Flow
cell 1 (black line) was an underivatized reference. Excess thiol was
deactivated in all flow cells (step b) and alkaline phosphatase (0.1 mm)
was injected for 4 h (step c). The enzyme was bound at intensities of
249 RU and 336 RU in flow cells 3 and 4, respectively. Noncovalently
attached protein was finally washed off with guanidinium chloride
(6m)(step d). The remaining signal intensities of 215 RU and 295 RU
correspond to covalently trapped enzyme, which make up 86% and 88% of
the entire bound phosphatase in flow cells 3 and 4, respectively. Enzyme
binding at the lowest inhibitor concentration was very weak (11 RU). No
binding of alkaline phosphatase to the reference surface in flow cell 1 was
observed. The enzyme inhibitor complex was finally released by disulfide
bond cleavage (step e).
[13]
phatase to the wells was obtained after incubation for 3 d
(OD450 ˆ 1.5 with 10 mg mLÀ1 BSA 3 conjugate and 10 units
of phosphatase per well).
In developing this selection tool, a major concern not fully
resolved by earlier work is whether noncatalytic proteins can
also be trapped. The results of Janda et al.[4] for an antibody
with galactosidase activity show multiple turnovers for the
covalently captured protein catalyst. This implies that pro-
teins may be trapped at amino acid residues not at the active
site, conceivably not even in the same protein molecule.
Furthermore, Raushel and co-workers have shown that the
loss of catalytic activity resulted from the precipitation of a
phosphate triesterase, caused by multiple modifications by a
quinone methide species.[14] To explore this possibility with 3,
alkaline phosphatase (10 units/well) was incubated with
BSA 3 for 3 d, and after appropriate washing steps, the p-
nitrophenyl phosphate (3 mm in 10 mm diethanolamine
buffer, pH 9.0, 50 mL/well) was added. The presence of
trapped alkaline phosphatase was demonstrated by ELISA
as described above.[15] No formation of p-nitrophenol was
observed, even after prolonged incubation.[16] It follows that
the enzyme has indeed been trapped either within the active
site or sufficiently close to it, so that the bulky BSA conjugate
occludes the active site. These results show that the suicide
trapping of alkaline phosphatase by substrate 3 leads to
the simultaneous immobilization and inactivation of the
enzyme.
step c). More enzyme was bound at higher concentrations of
the inhibitor on the surface. After injecting the alkaline
phosphatase, a continuous flow of buffer was initiated. Only a
small decrease in signal intensity was observed, which showed
that little dissociation of phosphatase from the surface-bound
complex had occurred. Most importantly, binding of the major
part of the enzyme (>85%) was not impaired even after
extensive treatment with 6m guanidinium chloride. Clearly,
the enzyme was covalently attached to the immobilized
inhibitor.
The slow but progressive binding of alkaline phosphatase
upon prolonged incubation for several hours in ELISA
experiments or after prolonged flow of enzyme over the
surface with immobilized inhibitor indicates a turnover-based
interaction and is inconsistent with the rapid formation of a
Michaelis complex of enzyme and substrate. Such binding
would not be expected to give kinetically resolvable binding
or dissociation traces in a BIAcore experiment, and at the
protein concentrations used (small relative to KM) its plateau
height would be extremely small.[18] The proportion of
irreversibly bound enzyme was found to be dependent on
the duration of the injection of alkaline phosphatase, which is
consistent with the time dependence observed on microtitre
plates. Unrelated proteins (e.g. BSA) did not activate the
mechanistic trap nor did they bind at all to the suicide
inhibitor. These results clearly show that inhibitor 3 traps the
The ability of substrate 3 to serve as a turnover-based
inhibitor for phosphatases was further investigated by means
of BIAcore analysis.[17] Inhibitor 3 was covalently immobilized
on a BIAcore chip in a defined orientation by disulfide
exchange. Upon injection, binding of alkaline phosphatase to
the inhibitor resulted in an increase in the signal (Figure 1,
776
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
1433-7851/02/4105-0776 $ 17.50+.50/0
Angew. Chem. Int. Ed. 2002, 41, No. 5

COMMUNICATIONS
[10] All new compounds were fully characterized by means of 1H, 19F, and
32P NMR spectroscopy and by HR-MS.
[11] J. H. Rickard, PhD thesis, University of Sheffield (UK), 2000.
[12] R. L. Jue, J. M. Lambert, L. R. Pierce, R. R. Traut, Biochemistry 1978,
17, 5399 5406; R. Singh, L. Kats, W. A. Bl‰ttler, J. M. Lambert, Anal.
Biochem. 1996, 236, 114 125.
[13] H. Denham, PhD thesis, University of Sheffield (UK), 1998.
[14] L.-C. Lo, C.-H. L. Lo, K. D. Janda, D. B. Kassel, F. M. Raushel,
Bioorg. Med. Chem. Lett. 1996, 6, 2117 2200.
[15] When the alkaline phosphatase was preincubated with 10 mm EDTA
(which inhibits 95% of phosphatase activity), binding to the trapping
agent was halved.
[16] Alkaline phosphatase incubated for 3 d at 378C in an ELISA plate
coated with untreated BSA still cleaved p-nitrophenyl phosphate.
[17] K. M. M¸ller, K. M. Arndt, A. Pl¸ckthun, Anal. Biochem. 1998, 261,
149 158.
[18] The Michaelis constant KM for 3 was found to be 250 mm, measured by
phosphate release from uncoupled inhibitor 3 by alkaline phosphatase
in solution. An affinity of this magnitude will lead, at best, to a tiny
plateau since, because of the fast off-rate, equilibration with the
surface is instantaneous.
[19] ™Catalytic Antibodies∫: G. M. Blackburn, A. Datta, H. Denham, P.
Wentworth, Adv. Phys. Org. Chem. 1998, 31, 249 391.
[20] B. Stec, M. J. Hehir, C. Brennan, M. Nolte, E. R. Kantrowitz, J. Mol.
Biol. 1998, 277, 647 662; ™Chemistry and enzymology of phospha-
tases∫: T. S. Widlanski, W. Taylor, Compr. Nat. Prod. Chem. 1999, 5,
139 162.
phosphatase efficiently and selectively, thereby displaying the
fundamental requirements for turnover-based enzyme inacti-
vation.
We wondered whether a catalyst with low turnover rates,
which would be expected to initially arise from selection
experiments with a ™first-generation∫ non-enzyme protein
library (e.g. catalytic antibodies),[19] can also be identified by
means of this screening procedure. As a model for such a case,
we investigated a mutant of alkaline phosphatase, S102A,
which lacks the primary nucleophile at the active site. Even
though the turnover rate for the S102A enzyme is about four
orders of magnitude lower than for the wild-type, it is still
substantially faster than the uncatalyzed reaction (kcat/kuncat

105).[20] Briefly, the mutant enzyme was injected in the
BIAcore instrument in the same way as the wild-type enzyme
to bind to the suicide substrate, and noncovalent bound
protein was washed off with guanidinium chloride. More than
30% of the total bound enzyme remained after this step,
demonstrating significant covalent coupling. The covalently
coupled protein could be removed by reductive cleavage of
the disulfide-containing linker that connects the suicide
substrate to the surface. Thus, poorly active protein can still
be identified by means of this turnover-based screening
procedure. This strongly suggests that the rate of covalent
inactivation depends on the successful reaction of a second
nucleophile with the quinone methide rather than on the
primary hydrolysis. However, under the same conditions, less
enzyme is trapped than in the case of wildtype enzyme, thus
suggesting that the suicide inhibitor may be potentially
selective for catalytic efficiency. BSA does not show any
covalent binding when used as a control in these reactions,
excluding the possibility that the covalent binding might
simply be a result of random reactivity of surface nucleophiles
on the protein.
The Inhibiting Influence of Aromatic Solvents
on the Activity of Asymmetric
Hydrogenations**
Detlef Heller,* Hans-Joachim Drexler,
Anke Spannenberg, Barbara Heller, Jingsong You, and
Wolfgang Baumann*
Dedicated to Franz Hein (1892 1976)
The synthesis and successful deployment of 3 is now being
applied to the selection of catalysts from large protein
libraries. Moreover, BIAcore analysis is demonstrated herein
to represent a useful approach for the direct screening of
library members and enables real-time analysis of the
sequence of steps necessary for catalyst selection from a
library of proteins. The relative advantages of o-trifluoro, o-
difluoro, and o-monofluoromethylphenyl phosphate suicide
substrates[11] are currently under investigation.
Complexes of ruthenium, iridium, and especially rhodium
have been used in the homogeneously catalyzed asymmetric
hydrogenation of prochiral olefins, ketones, and imines.[1]
Hydrogenations are usually carried out in simple alcohols,
but aromatic solvents, water, or alcohol/aromatic solvent
mixtures can also be used. It has been reported that aromatic
solvents such as benzene can inhibit asymmetric hydrogena-
[*] Priv.-Doz. Dr. D. Heller, Dr. W. Baumann, Dr. H.-J. Drexler,
Dr. A. Spannenberg, Dr. B. Heller
Received: September 7, 2001 [Z17867]
Institut f¸r Organische Katalyseforschung, Universit‰t Rostock e.V.
Buchbinderstrasse 5/6, 18055 Rostock (Germany)
Fax : (‡49)381-46693-83
[1] S. Halazy, V. Berges, A. Ehrhard, C. Danzin, Bioorg. Chem. 1990, 18,
330 344.
[2] Q. Wang, U. Dechert, F. Jirik, S. G. Withers, Biochem. Biophys. Res.
Commun. 1994, 200, 577 583.
E-mail: detlef.heller@ifok.uni-rostock.de
wolfgang.baumann@ifok.uni-rostock.de
Dr. J. You
[3] J. K. Myers, T. S. Widlanski, Science 1993, 262, 1451 1453.
[4] K. D. Janda, L.-C. Lo, C.-H. L. Lo, M.-M. Sim, R. Wang, C.-H. Wong,
R. A. Lerner, Science 1997, 275, 945 948; K. D. Janda, US 5571681,
1996.
[5] M. Wakselman, New J. Chem. 1983, 7, 439 447.
[6] J. R. Betley, PhD thesis, University of Sheffield (UK), 1997.
[7] W. J. Middleton, J. Org. Chem. 1975, 40, 574 578.
[8] S. Pinitglang, PhD thesis, University of London (UK), 1996.
[9] G. M. Blackburn, Chem. Ind. 1981, 834; G. M. Blackburn, D. E. Kent,
Chem. Commun. 1981, 134 138; C. E. McKenna, J. Schmidhauser,
Chem. Commun. 1979, 739 740.
Department of Chemistry, Sichuan University
610064 Chengdu (China)
[**] We would like to thank the Deutsche Forschungsgemeinschaft as well
as the Fonds der Chemischen Industrie for their generous support of
this work. We are also indebted to Prof. Dr. U. Rosenthal and Dr. D.
Selent for helpful discussions. Franz Hein prepared bis(h6-arene)-
chromium(i)-complex cations already in 1919; their true structure as
hexahapto complexes was only realized more than 35 years later.
Supporting information for this article is available on the WWW under
http://www.angewandte.com or from the author.
Angew. Chem. Int. Ed. 2002, 41, No. 5
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
1433-7851/02/4105-0777 $ 17.50+.50/0
777