Synthesis and enzymatic evaluation of a P1 arginine aminocoumarin substrate library for trypsin-like serine proteases

Article abstract of DOI:10.1016/S0960-894X(00)00460-1

A method for the solid-phase synthesis of P₁ arginine containing peptides via attachment of the arginine side-chain guanidine group is described. This procedure is applied to the preparation of a tetrapeptide, P₁ arginine aminocoumarin PS-SCL. This library was validated by using it to determine the P₄-P₂ specificity for thrombin and comparing the results to the known thrombin subsite specificity. This is the first reported example of a PS-SCL library containing a P₁ arginine. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0960-894X(00)00460-1

Bioorganic & Medicinal Chemistry Letters 10 (2000) 2291±2294
Synthesis and Enzymatic Evaluation of a P1 Arginine
Aminocoumarin Substrate Library for Trypsin-Like
Serine Proteases
Philip D. Edwards,* Russell C. Mauger, Kevin M. Cottrell, Frank X. Morris,
Kara K. Pine, Mark A. Sylvester, Clay W. Scott and Stephen T. Furlong
Departments of Chemistry and Neuroscience, AstraZeneca Pharmaceuticals, PO Box 15437, Wilmington, DE 19850-5437, USA
Received 9 May 2000; accepted 2 August 2000
AbstractÐA method for the solid-phase synthesis of P1 arginine containing peptides via attachment of the arginine side-chain
guanidine group is described. This procedure is applied to the preparation of a tetrapeptide, P1 arginine aminocoumarin PS-SCL.
This library was validated by using it to determine the P4±P2 speci®city for thrombin and comparing the results to the known
thrombin subsite speci®city. This is the ®rst reported example of a PS-SCL library containing a P1 arginine. # 2000 Elsevier Science
Ltd. All rights reserved.
Historically, the amino acid sequence ¯anking the clea-
vage site of a protein substrate has been used to develop
synthetic substrates and peptidic inhibitors of proteoly-
tic enzymes. However, the natural cleavage sequence
may not be the optimal sequence for binding to the
enzyme's active site. This was dramatically demon-
strated for interleukin-converting enzyme (ICE) by
Rano et al.1 who used a positional scanning synthetic
combinatorial library (PS-SCL) to prepare and analyze
all possible tetrapeptide aminocoumarin substrates
containing an aspartic acid residue at the P1 position.
When this library was screened against ICE, it was dis-
covered that the optimal P4±P1 substrate sequence was
WEHD not YVAD as previously believed. The success-
ful construction of this PS-SCL was facilitated by the
ability to attach the invariant P1 aspartate residue to a
solid-phase resin via the carboxylate of the P1 aspartate
side chain.
attached to the resin via its guanidine side chain. At the
time we conceived this project, there were no reported
methods for conducting SPPS using an arginine residue
linked to a polymer via its side chain. In general, the
versatility of the PS-SCL approach to substrate libraries
is limited by the inability to attach most amino acids to
a solid support via their side chain. Recently, Backes et
al.2 reported a strategy to overcome this generic pro-
blem in which the carboxyl group of the P1 residue is
attached to the resin via a safety catch linker. Displace-
ment of the peptide with 7-amino-4-methylaminocou-
marin followed by side-chain deprotection a€ords the
peptide aminocoumarin substrate. This method avoids
the necessity of attaching the P1 residue to the resin via
its side chain, and should, in principle, enable produc-
tion of PS-SCL substrate libraries with any P1 residue.
However, in the report by Backes et al.,2 only a P1 lysine
library for thrombin and plasmin was described, not a
P1 arginine library. Since arginine, not lysine, is the
preferred P1 residue for these trypsin-like enzymes, the
question is raised as to whether synthetic diculties
prevented the preparation of a P1 arginine substrate
library. Due to the therapeutic importance of trypsin-
like serine proteases, the need for P1 arginine PS-SCL
substrate libraries is great. In this paper, we describe the
®rst reported process for constructing a P1 arginine
aminocoumarin (AMC) tetrapeptide substrate PS-SCL
via solid-phase techniques, and validate the ®delity of
the substrate library using the trypsin-like serine pro-
tease thrombin.
Since trypsin-like serine proteases prefer a basic arginine
residue at the P1 position, we sought to prepare a PS-
SCL in which the invariant P1 arginine residue was
*Corresponding author. Tel.: +1-302-886-8371; fax: +1-302-886-5382;
e-mail: philip.edwards@astrazeneca.com
Abbreviations: DMAP, 4 - (dimethylamino)pyridine; DIPEA, diiso-
propylethylamine; DIPCI, diisopropyl carbodiimide; HATU, N-
[(dimethylamino)-1H-1,2,3-triazol[4,5-b]pyridin-1-ylmethylene]-N-
methylmethan aminium hexa¯uorophosphate N-oxide; HOBt, 1-
hydroxybenxtriazole hydrate; PS-SCL, positional scanning synthetic
combinatorial library; SPPS, solid-phase peptide synthesis.
0960-894X/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(00)00460-1

2292
P. D. Edwards et al. / Bioorg. Med. Chem. Lett. 10 (2000) 2291±2294
While there have been several reports of the solid-phase
synthesis of amidines and guanidines,3 little has been pub-
lished on the solid-phase synthesis of peptides containing
arginine linked to the resin through the side-chain gua-
nidine group. As we began considering a strategy for
attaching an arginine to a solid-phase resin via its side
chain guanidine group, two reports appeared in the lit-
erature describing such a procedure.4,5 Zhong et al.4
employed an aromatic sulfonyl linker which is con-
ceptually similar to the procedure we ultimately employed.
However, attachment of the arginine guanidine to the
resin-bound sulfonyl group required 4 N KOH at 75 ꢀC
for 2 days, conditions incompatible with the presence of
an aminocoumarin amide required in our synthesis.
Most relevant to our needs was a poster presentation by
Urban and Dattilo describing the solid-phase prepara-
tion of a P1 arginine tripeptide p-nitroanilide substrate
using a novel CMtr linker. The CMtr is a modi®ed Mtr
containing a carboxyl group for attachment to the
resin.5 We used a modi®ed version of this procedure for
preparing the intermediate Boc-Arg (CMtr)-Amc 7
(Scheme 1). The chemistry is straightforward with a few
notable exceptions. The sulfonyl chloride 3 is relatively
unstable, and the reaction time for its formation should
be limited to 1 h. Furthermore, 3 should be used in the
subsequent sulfonamide forming reaction immediately
upon solvent removal. We have also found that the best
overall yields are obtained when sulfonamide formation
and ester hydrolysis are carried out in two independent,
sequential reactions.
A signi®cant amount of e€ort was required to identify
appropriate conditions for attaching 7 to a solid sup-
port (Scheme 2). We initially investigated two di€erent
PEG polystyrene resins. Since the amino group attach-
ment point in these resins is distant from the cross-
linked resin backbone, we felt that these resins would
a€ord a more ecient amide coupling with the rela-
tively bulky acid 7. However, the physical properties of
the coupled products were such that severe mechanical
losses were encountered. These losses continued to plague
us throughout the solid-phase synthetic sequence. To
overcome this problem and still maintain an extended
attachment point to the resin, we attached a b-alanine
linker to an AgroPoreTM-NH2 resin.6 When coupled
with 7, the resin bound intermediate 9 had excellent
handling properties and no signi®cant mechanical losses
were observed. The couplings of both the b-alanine to
the resin and of 7 to 8 were monitored by following the
growth of the amide carbonyl absorption peak of a bead
using solid-phase IR since a negative Kaiser test could
not be obtained. Maximum loading was considered the
point where there was no further increase in the car-
bonyl absorption, and was achieved using a two-step
sequence involving basic coupling conditions (HATU
and DIPEA) followed by a more acidic system (DIPCI
Scheme 2. (i) Boc-bAla-OH, HATU, DIPEA, DMF, 18 h, repeat for
3 h; (ii) HOBT, DIPCI, DMAP (cat), DMF, 18 h, repeat for 3 h; (iii)
Ac2O, DIPEA, DMF, 2 h; (iv) 40% TFA/CH2Cl2, 0.5 h; (v) HATU,
DIPEA, DMAP (cat), DMF, 18 h; (vi) HOBt, DIPCI, DMAP (cat),
DMF, 18 h; (vii) 40% TFA/CH2Cl2, 1 h; (viii) isokinetic mixture or
known amino acid, HATU, DIPEA, DMF, 2 h, repeat procedure
once; (ix) 20% piperidine, DMF, 15 min, repeat procedure once; (x)
thioanisole/TMSBr/triisopropyl silane/phenol/TFA (ref 8).
Scheme 1. (i) Ethyl bromoacetate, K2CO3, DMF, rt, 18 h, 75%; (ii)
HSO3Cl, CH2Cl2, 4 ꢀC, 1 h, 85%; (iii) Boc-Arg(Z)2-OH, Boc2O, Py,
THF, rt, 18 h, 76%; (iv) Pd(OH)2, EtOH, 1 N HCl, H2/48 psi, 4 h,
93%; (v) 1 N NaOH (2 equiv), acetone, rt, 0.5 h, 32%; (vi) 1 N NaOH
(1 equiv), THF/EtOH (1:1), rt, 3 h, 82%.

P. D. Edwards et al. / Bioorg. Med. Chem. Lett. 10 (2000) 2291±2294
2293
and HOBt). Unreacted amino groups on the resin were
then capped with acetic anhydride.
The strategy behind the tetrapeptide PS-SCL is the gen-
eration of three sublibraries, one library for each sequence
position evaluated (P2, P3, and P4). Each sublibrary is
composed of one pot for each amino acid evaluated. In
the current study, 20 amino acids were investigated
resulting in 20 pots per sublibrary. In each pot, the residue
at one sequence position (e.g., P2) is known. Using a P2
sublibrary as an example, this is achieved by dividing 9
into 20 pots and coupling a di€erent amino acid in each
pot. In the next coupling sequence, an isokinetic mix-
ture of the 20 amino acids is used to generate all possi-
ble 20 tripeptides in each pot in equal proportions. For
the ®nal coupling, the same isokinetic mixture is used to
generate in each of the 20 pots all possible 400 tetra-
peptide substrates with a di€erent, known P2 residue.
Following cleavage from the resin, each pot in the sub-
library is assayed against the enzyme and the optical
density (OD) measured. Since each pot contains all
possible tetrapeptides with a known P2 residue, those
pots displaying the highest OD contain substrates that
are most eciently cleaved and therefore the preferred
P2 residues. Similar analysis of the P3 and P4 sub-
libraries identi®es the preferred P3 and P4 residues.
Figure 1. MALDI-TOF mass spectra (re¯ectron mode, alpha-cyano
hydroxycinnamic acid matrix) of P3 sublibrary pots (a) Ac-O-Val-O-
Arg-Amc and (b) Ac-O-Asp-O-Arg-Amc.
Solid-phase synthesis of each sublibrary was accom-
plished using Fmoc protected single amino acids or an
isokinetic mixture of 20 amino acids, HATU coupling,
and piperidine deprotection. The loading was monitored
using the Kaiser test, with a negative test being obtained
before proceeding with the deprotection. The HATU
coupling conditions and proportions of amino acids in
the isokinetic mixture were those developed by Herman
et al.7 Following cleavage from the resin, the crude
material was isolated by solvent removal and subsequent
ether trituration. No further puri®cation was employed
so as to avoid selective removal of any peptides. Initially,
cleavage of the peptides from the resin using a standard
TFA:H2O:phenol:TIS (87.5:5:5:2.5) cocktail and a PEG
polystyrene resin a€orded only a 5% yield of peptide
mixture. Changing to the AgroPoreTM-NH2 resin and
employing a stronger, TMSBr-based cleavage cocktail8
increased the yield to 50±70% (75±105 mg/pot).
Ac-O-Asp-O-Arg-Amc pots are 705 and 721, respec-
tively. The centers of distribution in Figure 1 are in
good agreement with these calculated values.
While the chemical characterization of the library sug-
gested that equimolar mixtures of peptides were present in
the pots analyzed, the most important and practical
means for validating the ®delity of the library was to
assay each sublibrary against an enzyme with known
substrate speci®city. Thrombin was chosen because it is
a trypsin-like serine protease of current medicinal inter-
est. In addition, thrombin has a varied subsite speci®city
that we felt would be ideal for validating the library: an
extremely strong preference for proline at P2; a rela-
tively broad speci®city at P3; and a preference for
hydrophobic residues at P4.9 The libraries were assayed
by incubating each sample with enzyme and measuring
the ¯uorescence.
We conducted a limited chemical characterization of the
library. The P3 sublibrary Ac-O-Val-O-Arg-Amc and
Ac-O-Asp-O-Arg-Amc pots were analyzed by means of
MALDI-TOF mass spectrometry. The mass spectral
data was assessed for the presence of three features: (1)
that the smallest (-G-V-G-) and largest (W-V-W-)
molecular weight peptides were present; (2) that the
center of the distribution for each of the two pots was
shifted from each other by 16, the di€erence in mole-
cular weight between valine and aspartic acid; and (3)
that the molecular weight distribution was centered
around the average molecular weight. Figure 1 is a
comparison of the mass spectra for the -Val- and -Asp-
pots. As anticipated, the high and low molecular weight
compounds were present and the center of distribution
between each library was shifted by 16. The average
molecular weights for the Ac-O-Val-O-Arg-Amc and
The results against each sublibrary are depicted in Fig-
ure 2. In the P2 library, only ®ve amino acids demon-
strated any signi®cant hydrolysis by thrombin, all small
hydrophobic residues. Proline (P) was clearly preferred.
Consistent with the literature, a broad variety of resi-
dues were tolerated at P3, including basic, polar, and
both large and small hydrophobic residues. Proline,
which is a unique cyclic amino acid that imparts a turn
into the peptide backbone, was completely inactive at
the P3 position. Signi®cantly, valine (V), leucine (L) and
isoleucine (I) were preferred over phenylalanine (F) as
recently reported for peptidic chloromethyl ketone inhi-
bitors.10 Particularly interesting is the comparison of
aspartic acid (D) and its amide asparagine (N) with

2294
P. D. Edwards et al. / Bioorg. Med. Chem. Lett. 10 (2000) 2291±2294
Backes et al.2 using their P1 lysine substrate library. The
subtle di€erences observed between the two libraries,
such as detectable hydrolysis of nonproline P2 sub-
strates in the current P1 arginine library, may be due
either to di€erent detection threshold levels in the
enzyme assays, or due to di€erent subsite speci®cities
between P1 arginine and P1 lysine substrates.
In summary, a procedure has been developed for e-
ciently attaching an arginine to a solid support via its
guanidine side chain, and cleaving it from the resin fol-
lowing subsequent elaboration. This methodology was
used to construct a P1 arginine aminocoumarin sub-
strate PS-SCL, and validate the library against throm-
bin, a trypsin-like serine protease. This extremely
important class of enzymes prefers an arginine at P1,
and the methodology described herein is the ®rst pro-
cess reported to allow construction of PS-SCL substrate
libraries containing a P1 arginine.
References and Notes
1. Rano, T. A.; Timkey, T.; Peterson, E. P.; Rotonda, J.;
Nicholson, D. W.; Becker, J. W.; Chapman, K. T.; Thron-
berry, N. A. Chem. Biol. 1997, 4, 149.
2. Backes, B. J.; Harris, J. L.; Leonetti, F.; Craik, C. S.; Ell-
man, J. A. Nat. Biotechnol. 2000, 18, 187.
3. (a) Josey, J. A.; Tarlton, C. A.; Payne, C. E. Tetrahedron
Lett. 1998, 39, 5899. (b) Roussell, P.; Bradley, M.; Matthews,
I.; Kane, P. Tetrahedron Lett. 1997, 38, 4861. (c) Bonnat, M.;
Bradley, M.; Kilburn, J. Tetrahedron Lett. 1996, 37, 5409.
4. Zhong, H. M.; Greco, M. N.; Maryano€, B. E. J. Org.
Chem. 1997, 62, 9326.
5. Urban, J.; Dattilo, J. W. Abstracts of Papers, 15th Amer-
ican Peptide Symposium, Nashville, TN, 14±19 June, 1997.
6. AgroPoreTM-NH2 resin was purchased from Argonaut
Technologies.
7. Herman, L. W.; Tarr, G.; Kates, S. A. Mol. Divers. 1996, 2,
147.
Figure 2. Substrate speci®cities for human thrombin. Each bar is an n
of 3. The x axis represents substituted amino acids (B=d-alanine;
J=norleucine). The y axis is AMC production expressed as a percen-
tage of the maximum rate observed in each experiment.
8. (a) Hughes, J. L.; Leopold, E. J. Tetrahedron Lett. 1993, 48,
7713. (b) Guo, L.; Funakoshi, S.; Fujii, N.; Yajima, H. Chem.
Pharm. Bull. 1988, 36, 4989.
9. Bode, W.; Turk, D.; Karshikov, A. Protein Sci. 1992, 1,
426.
glutamic acid (E) and its amide glutamine (Q). Both D
and N were inactive, while E and Q, which have an addi-
tional methylene in their side chain, were good substrates,
suggesting a critical side chain length requirement.
Finally, hydrophobic residues are preferred at P4. These
results agree remarkably well with those obtained by
10. Vindigni, A.; Dang, Q. D.; DiCera, E. Nat. Biotechnol.
1997, 15, 891.