Toward protein-cleaving catalytic drugs: Artificial protease selective for myoglobin

Article abstract of DOI:10.1016/S0968-0896(03)00216-5

A protein-cleaving catalyst highly selective for a disease-related protein can be used as a catalytic drug. As the first protein-cleaving catalyst selective for a protein substrate, a catalyst for myoglobin (Mb) was designed by attaching the Cu(II) or Co(

Full text of DOI:10.1016/S0968-0896(03)00216-5

Bioorganic & Medicinal Chemistry 11 (2003) 2901–2910
Toward Protein-Cleaving Catalytic Drugs: Artificial Protease
Selective for Myoglobin
Joong Won Jeon, Sang Jun Son, Chang Eun Yoo, In Seok Hong and Junghun Suh*
School of Chemistry and Center for Molecular Catalysis, Seoul National University, Seoul 151-747, South Korea
Received 17 March 2003; accepted 24 March 2003
Abstract—A protein-cleaving catalyst highly selective for a disease-related protein can be used as a catalytic drug. As the first
protein-cleaving catalyst selective for a protein substrate, a catalyst for myoglobin (Mb) was designed by attaching the Cu(II) or
Co(III) complex of cyclen to a binding site searched by a combinatorial method using peptide nucleic acid monomers as building
units. Various linkers were inserted between the catalytic Co(III) center and the binding site of the Mb-cleaving catalyst. Kinetic
data revealed catalytic turnover of the Mb cleavage by the Cu(II) or Co(III) complex. MALDI-TOF MS revealed cleavage of the
polypeptide backbone of Mb at selected positions. N-Terminal sequencing of the cleavage products identified the cleavage site and
provided evidence for the hydrolytic nature of the Mb cleavage. Various chelating ligands were tested as the ligand for the Co(III)
center of the Mb-cleaving catalyst. Among the nine chelating ligands examined, only cyclen and its triaza-monooxo analogue
manifested catalytic activity.
# 2003 Elsevier Science Ltd. All rights reserved.
Introduction
substrate selectivity and catalytic rate would be
obtained by connecting a binding site that recognizes
the target protein to a catalytic group. On complexation
of the protein-cleaving catalyst to the target protein, the
effective molarity18 of the catalytic grouptoward a
peptide linkage of the target protein can increase to a
sufficiently high level to allow facile cleavage of the
peptide bond. To suppress cleavage of non-target pro-
teins, the catalytic center should have very low protein-
cleaving activity when unattached to the binding site.
In search of new drugs, efforts are made to design
molecules that specifically block the activity of disease-
related proteins such as enzymes, receptors, and ion
channels.1ꢀ4 Even if a drug molecule is bound to its
target protein very strongly, at least an equivalent
amount of the drug is needed to inactivate the protein.
If a disease-related protein is deactivated upon cleavage
of its polypeptide backbone by a synthetic catalyst, the
catalyst can be used as a drug. Because a catalytic
amount of the drug can destroy the protein, the dosage
can be reduced substantially by using the catalyst.
Designing protein-cleaving catalysts highly selective for
target proteins, therefore, would become a powerful
tool in drug discovery. To establish the idea of protein-
cleaving catalytic drugs, it is essential to synthesize the
first artificial protease selective for a target protein.
As the catalytic moiety of a target-selective artificial
protease, either a purely organic catalytic center or a
metal complex may be used. Each of the organic artifi-
cial proteases reported to date uses an artificial active
site comprising two or more organic catalytic groups.
On the other hand, a single metal complex can often act
as an effective catalytic site for artificial metallopro-
teases. In the present study, we chose to use metal
complexes as the catalytic center of the first target-
selective artificial protease.
Several synthetic catalysts with proteolytic activity
lacking substrate selectivity have been reported.5ꢀ17
Both organic functional groups and metal complexes
have been exploited as the catalytic groups in those
artificial proteases. A protein-cleaving catalyst with high
As the binding site of the target-selective artificial pro-
tease, small organic compounds such as inhibitors or
antagonists already reported to have high affinity
toward the target protein may be used. Alternatively, a
new binding site may be searched by the combinatorial
approach. The catalyst obtained by the latter method
*Corresponding author. Tel.: +82-2-886-2184; fax: +82-2-874-3704;
e-mail: jhsuh@snu.ac.kr
0968-0896/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0968-0896(03)00216-5

2902
J. W. Jeon et al. / Bioorg. Med. Chem. 11 (2003) 2901–2910
does not necessarily bind to the active site of the pro-
tein. In the present study, we adopted the combinatorial
method to obtain the binding site of the target-selective
artificial protease using a metal complex as the catalytic
site. In this article, selection of the lead catalyst selective
for myoglobin (Mb) and effects of variation in the shape
of the linker connecting the binding site and the cataly-
tic site as well as the structure of the chelating ligand of
the catalytic site are described.19
Results
In the present study directed toward the first protein-
cleaving catalyst selective for a target protein, we chose
metal complexes of cyclen (Cyc) as the catalytic center
since a previous study11 indicated that the Cu(II) com-
plex of Cyc has very low proteolytic activity unless
activated by additional catalytic elements.
In the present study, we constructed a combinatorial
library of Cyc derivatives containing analogues of pep-
tide nucleic acid (PNA)21,22 to search the binding site of
a protein-cleaving catalyst. Since organic functional
groups of some proteins are exploited in recognition of
nucleobases of nucleic acids,23 nucleobases of PNAs
may be conversely used in recognition of proteins. The
library can be presented as CycAc(Q)nLysNH2 where Q
is PNA monomer A*, G, T*, or C. PNA contains
nucleobases that can be used for base-pairing with
nucleobases of DNA. We used modified nucleobases A*
and T* instead of A and T. A* and T* recognize T and
A, respectively. A* and T*, however, do not recognize
each other.24 Base-pairing among PNA mixtures pre-
sent in the library, therefore, can be suppressed by using
A* and T*.
The library of CycAc(Q)nLysNH2 (total concentration:
ca. 70 mM) was mixed with an aqueous solution of
Cu(II)Cl2 (350 mM) to generate the library of Cu(II)Cy-
cAc(Q)nLysNH2 where Cu(II) is bound to the Cyc moi-
ety. Cu(II) forms a strong complex with Cyc (log
Kf=16.8 at pH 7).25 With the Cu(II)Cyc library con-
taining 7- or 8-mer PNAs, no evidence was obtained for
cleavage of proteins (10 mM) such as bovine serum
albumin, g-globulin, elongation factor P, gelatin A,
gelatin B, and horse heart Mb at 37 ꢁC and pH 7 as
checked by sodium dodecyl sulfate polyacrylamide gel
electrophoresis (SDS-PAGE). The Cu(II)Cyc library
containing 9-mer PNAs, however, clearly showed activ-
ity for cleavage of Mb. Four groups of library with the
known PNA monomers positioned next to Cu(II)Cyc
were subsequently synthesized and tested for their
activity and A* was identified as the best monomer for
that position. By repeating the search for the rest of the
nine positions occupied by PNA monomers, 1 (Fig. 1)
was chosen as the ligand of the best catalyst. The bind-
ing site for Mb was, therefore, obtained by using the
functional groups in 1 such as the pyrimidine and
purine bases and amide groups. The structures of 1 and
its analogues (2–17) synthesized in the present study are
illustrated in Figure 1.
The PNA derivatives were synthesized by automated
synthetic procedures with the fmoc-derivatives of A*,
T*, G, and C (A*X, T*X, GX, and CX, respectively) as
well as Na-fmoc-Ne-boc-l-Lys (LysX) and the boc-deri-
vative of CycAcOH (1X). The library of Cyc-containing
PNA oligomers with the general structure of
CycAc(Q)nLysNH2 was constructed by the Split-Pool
method. In the construction of the library, it was
assumed that the carboxyl groups of A*X, T*X, GX,
and CX have identical reactivity in coupling with the
amino groups linked to the polymer support.
The position of the metal complex of Cyc relative to the
PNA 9-mer was varied. Cyc was attached either to the
C- (2) or the N-terminus of the PNA oligomer. When
Cyc was attached to the N-terminus, the catalytic site
contained either one (3–8) or two (9) Cyc units. To
suppress the conformational freedom of the Cyc-acetyl
moiety of 1, methyl groupwas attached to the acetyl

J. W. Jeon et al. / Bioorg. Med. Chem. 11 (2003) 2901–2910
2903
portion of the Cyc-acetyl moiety leading to 3. Amino
acids such as Gly, l-Asp, L-Pro, l-Lys, and b-Ala were
inserted between the Cyc-acetyl moiety and the PNA
portion of 1 to obtain 4–8. For synthesis of 2–9 by the
automated procedures, 2Y, 3X, and 9X synthesized in
this study as summarized in Schemes 1–3 and GlyX,
AspX, ProX, and BalX (N-fmoc derivatives of Gly,
l-Asp b-t-butyl ester, l-Pro, and b-Ala, respectively)
purchased from commercial sources were used as addi-
tional building blocks.
In view of the wide range of potential application of the
target-specific protein-cleaving catalysts to drug design,
whether the Cyc moiety of 1 can be replaced by other
chelating ligands is important. In the present study,
therefore, 10–17 were tested to evaluate effectiveness of
various alternative chelating ligands. For synthesis of
10–17 by the automated procedures, 10X–17X were
synthesized as summarized in Schemes 4–11 and used as
additional building blocks.
The stock solution of Cu(II) complex of 1 was prepared
by adding an aqueous solution of CuCl2 to 1 (1.2 equiv)
at pH 6.0. Due to the kinetic inertness26 of Co(III)
complexes, direct insertion of Co(III) ion to the chelat-
ing ligands is not easy. Instead, the Co(III) complexes of
1–14 were generated in situ by incorporating Co(II) ion
to the respective Cyc derivatives and then oxidizing the
complexed Co(II) ion in methanol according to the lit-
erature27 procedure. For 15–17, Co(III) complexes were
not obtained by the method employed in this study.
Binding of Co(III) ion to aza ligands was confirmed by
appearance of lmax around 510 nm in Vis spectrum and
the matrix-assisted laser desorption/ionization time-of-
flight mass spectrum (MALDI-TOF MS).
Scheme 2. Synthesis of 3X.
Figure 1. Structures of 1–17.
Scheme 3. Synthesis of 9X.
Scheme 4. Synthesis of 10X.
Scheme 5. Synthesis of 11X.
Scheme 1. Synthesis of 2Y.

2904
J. W. Jeon et al. / Bioorg. Med. Chem. 11 (2003) 2901–2910
Reaction rates for cleavage of protein substrates were
followed by SDS-PAGE according to the methods
described elsewhere.5ꢀ7,11 An example of the plot of
[Mb] (total concentration of uncleaved Mb; estimated
from the electrophoretic bands of Mb) against time is
illustrated in Figure 2. The time-dependent decrease in
[Mb] was fitted to pseudo-first-order kinetic equations
to obtain pseudo-first-order rate constant (ko).
straight line to obtain ko, Michaelis–Menten scheme
predicts that the kinetic behavior does not conform to
first-order kinetics when Co is smaller than [Mb]o (the
initially added concentration of Mb) even when Co
> >Km. Moreover, rate data based on electrophoretic
measurement are not very accurate. These may be rela-
ted to the scattered data points of Figure 3 at Co
<[Mb]o.
Although the structure of 1 was obtained by using the
Cu(II) complex, detailed kinetic analysis was performed
with the Co(III) complex due to the higher catalytic
activity of the Co(III) complex revealed by the kinetic
data of Figure 2. The dependence of ko on Co (the
initially added concentration of the catalyst) measured
with Co(III)1 at pH 7.5 is illustrated in Figure 3.
Although the plot of ln [Mb] against time was fitted to a
Although the kinetic data are somewhat scattered, the
two straight lines drawn in Figure 3 intersect at
Co=[Mb]o. This intersection agrees with strong binding
of Mb to Co(III)1: in terms of Michaelis–Menten para-
meters, Km < <Co and, thus, Km < <5 mM. Further-
more, ko measured with Co greater than [Mb]o
corresponds to kcat
.
For the Co(III) complexes of 2–14, activity for the Mb
cleavage was observed only with 3, 6, 7, 9, and 10. The
dependence of ko on Co for the Co(III) complexes of 3,
6, 7, 9, and 10 was the same as that observed with that
of 1. For 3, 6, 7, 9, and 10, therefore, Km < <Co and Km
< <10 mM when analyzed in terms of Michaelis–Men-
ten scheme. Furthermore, ko measured with Co greater
than [Mb]o corresponds to kcat
.
The kcat values thus measured at various pHs for the
cleavage of Mb by Co(III)1 and Co(III)10 are illu-
strated in Figures 4 and 5. The pKa values reflected in
the bell-shaped pH profiles can be assigned to the
Scheme 6. Synthesis of 12X.
Scheme 10. Synthesis of 16X.
Scheme 11. Synthesis of 17X.
Scheme 7. Synthesis of 13X.
Scheme 8. Synthesis of 14X.
Scheme 9. Synthesis of 15X.
Figure 2. Decrease in [Mb] during incubation of Mb with Cu(II)1 (*;
curve a, [Mb]o=7.9 mM, [Cu(II)1]o=2.0 mM) or Co(III)1 (*; curve b,
[Mb]o=4.7 mM, [Co(III)1]o=0.47 mM) at pH 7.5 and 37 ꢁC. The
curves were obtained as indicated in the text: ko=5.7 ꢂ 10ꢀ3 hꢀ1 for
curve a and 9.4 ꢂ 10ꢀ3 hꢀ1 for curve b.

J. W. Jeon et al. / Bioorg. Med. Chem. 11 (2003) 2901–2910
2905
ionization of functional groups of the complex formed
between Mb and Co(III)1 or Co(III)10. It is noteworthy
that the catalyst is most active at the physiological pH.
The kcat values measured at the optimum pH for the
Co(III) complexes of 1, 3, 6, 7, 9, and 10 are summarized
in Table 1.
Since only two protein fragments were formed by
Co(III)3, the product solution obtained by incubation
of Mb (9.8 mM) with Co(III)3 (8.5 mM) for 7 days at pH
8.0 and 37 ꢁC was subject to N-terminal sequencing by
Edman degradation. Results of the N-terminal sequen-
cing indicated that Leu89-Ala90 was the cleavage site in
agreement with the cleavage site proposed on the basis
of MALDI-TOF MS results. Moreover, the successful
N-terminal sequencing indicates that the amino group
was generated after peptide cleavage in support of the
hydrolytic nature of the Mb cleavage.
MALDI-TOF MS (Fig. 6) of a reaction mixture
obtained by incubation of Mb with Co(III)1 disclosed
that Mb was dissected into two pairs of proteins.
Attempts to identify the N-terminal sequences of the
cleavage products were unsuccessful since individual
protein fragments were not separated cleanly by elec-
trophoresis. For the cleavage of Mb by the Co(III)
complexes of 3, 6, 7, 9, and 10, however, MALDI-TOF
MS taken for the solutions obtained by incubation of
Mb with the Co(III) complexes revealed single cleavage
site, as exemplified by the spectrum illustrated in Figure 7.
The molecular weights of protein fragments disclosed by
MALDI-TOF MS are summarized in Table 2 together
with the possible cleavage sites for the protein fragments
deduced from the primary structure of Mb.
When Fe(III), Hf(IV), Pt(IV), Zr(IV), Pd(II), or Ce(IV)
ion was added to 1 and then incubated with Mb, Mb
cleavage was not observed. When other proteins such as
Figure 5. pH dependence of kcat for cleavage of Mb by Co(III)10 at
37 ꢁC. The bell-shaped curve was obtained by treating the Co(III)10–
Mb complex as a diprotonic acid (pKa1=7.4, pKa2=8.3) and by
assuming that the monoprotonated species is reactive.
Table 1. Values of kcat measured at optimum pH and 37 ꢁC for the
Co(III) complexes of 1, 3, 6, 7, 9, and 10
Catalyst
kcat (10ꢀ3 hꢀ1 a
)
Optimum pH
Co(III)1
Co(III)3b
Co(III)6
Co(III)7
Co(III)9
Co(III)10
22
2.8
2.4
2.4
4.4
8.9
7.5
7.5
7.5
7.5
7.5
8.0
Figure 3. The plot of ko against Co for cleavage of Mb
([Mb]o=4.7 mM) by Co(III)1 at pH 7.5 and 37 ꢁC (0.05 M buffer;
addition of 0.5 M NaCl did not affect the rate data appreciably).
Straight lines a (Co <So) and b (Co ꢃSo) stand for vo=So (vo: initial
velocity) and ko, resepctively, as predicted by Michaelis–Menten
scheme under the condition of Co > >Km.
aRelative standard deviation: 10–20%.
bFor the two enantiomers of 3X, only one may lead to an active arti-
ficial protease. Thus, the observed kcat value may underestimate the
actual catalytic capability.
Figure 6. MALDI-TOF MS taken after incubation of Mb (12 mM)
with Co(III)1 (3.5 mM) at pH 6.0 and 37 ꢁC for 85 h. Mass values (m/z)
are 16953 (A), 9892 (B), 8909 (C), 16953/2 (D), 8045 (E), and 7074 (F);
A and D are peaks for Mb (M.W. 16953) and two pairs (B/F and C/E)
are for proteins formed by cleavage of Mb.
Figure 4. pH dependence of kcat for cleavage of Mb by Co(III)1 at
37 ꢁC. The bell-shaped curve was obtained by analyzing the data by
treating the Co(III)1-Mb complex as a diprotonic acid (pKa1=5.5,
pKa2=8.7) and by assuming that the monoprotonated species is reactive.

2906
J. W. Jeon et al. / Bioorg. Med. Chem. 11 (2003) 2901–2910
To test the proteolytic activity of the library, the Cyc
moieties of the library were converted to the Cu(II)
complexes and the mixture of Cu(II)CycAc(Q)nLysNH2
was incubated with various proteins. Although the
nucleobases of A* and T* do not recognize each other,
those of G and C can form complexes with each other.
A significant portion of the members of the library can
form duplexes with other members containing com-
plementary sequences, especially when the G/C content
of the PNA oligomer is high. Those duplexes might
have low affinity for the target protein and, thus, have
limited proteolytic activity. Although a large number of
PNA oligomers were generated by the combinatorial
method, only a small portion may exist as unfolded
unimolecular identities in the buffer solution. In 1, which
contains a PNA oligomer chosen as the most active
sequence, six nucleobases out of nine are A* or T*.
Figure 7. MALDI-TOF MS taken after incubation of Mb (9.8 mM)
with Co(III)7 (9.8 mM) at pH 7.5 and 37 ꢁC for 96 h. Mass values (m/z)
are 16953 (A), 16953/2 (B), 7068 (C), and 9888 (D); A and B are peaks
for Mb and C and D are for proteins formed by cleavage of Mb.
Table 2. Molecular weights of protein fragments disclosed by
MALDI-TOF MS and cleavage sites proposed to account for the
protein fragments
Catalyst
m/z of fragments
Cleavage site proposed
on the basis of m/z valuesa
Compounds 1–18 synthesized with the automated syn-
thesizer were contaminated with small amounts of
impurities as exemplified by the MALDI-TOF MS of
Co(III)1 illustrated in Figure 8. In most of the Co(III)
complexes, the Co(III) ion do not considerably dis-
sociate from the Cyc portion under the conditions of
measurement of MALDI-TOF MS. Purification of the
PNA derivatives by HPLC did not remove the MALDI-
TOF MS peaks of the impurities significantly. Except
for 2, Cyc moiety was attached in the last synthetic step
and, therefore, the contaminants do not have the Cyc
portion which is essential for the catalytic activity. For
2, the contaminants do not contain some of the nucleo-
bases which are important for complex formation with
Mb. It was assumed, therefore, that the presence of
small amounts of the contaminants do not affect the
catalytic outcome significantly.
Co(III)1
Co(III)3
Co(III)6
Co(III)7
Co(III)9
Co(III)10
7074, 9891, 8045, 8909
7066, 9897
7066, 9898
Leu89-Ala90, Leu72-Gly73
Leu89-Ala90
Leu89-Ala90
7068, 9888
7068, 9884
7075, 9896
Leu89-Ala90
Leu89-Ala90
Leu89-Ala90
aMr of protein fragments to be obtained: 7059 and 9893 for cleavage at
Leu89-Ala90, 8056 and 8896 for cleavage at Leu72-Gly73
.
albumin, g-globulin, elongation factor P, gelatin A, and
gelatin B were incubated with Cu(II)1 or Co(III)1, pro-
tein cleavage was not observed. When Mb was treated
with CuCl2, Cu(II)Cyc, or Co(III)Cyc, Mb was not
degraded. An analogue of 1 was prepared where the
PNA residue next to the CycAc unit is C instead of A*
as indicated by 18. No catalytic activity was observed
for Co(III)18 and Cu(II)18 in the cleavage of Mb. The
degree of cleavage of amide bonds of Cu(II)1 or
Co(III)1 was not significant when followed for several
days under the conditions of kinetic measurements of
the present study by measuring MALDI-TOF MS and
the catalytic activity for Mb cleavage.28 The rate for
cleavage of Mb by Cu(II)1 or Co(III)1 was unaffected
by the removal of O2 from the reaction mixtures.
Upto 2.5 or 6 molecules of Mb were cleaved by each
molecule of Cu(II)1 or Co(III)1, respectively, for the
data of Figure 2, indicating the catalytic nature of the
action of Cu(II)1 and Co(III)1. The number of Mb
molecules cleaved per catalyst molecule was small due
to the small value of kcat as a consequence of the low
effective molarity of the catalytic center in the Mb–cat-
alyst complex. Catalytic turnover by Co(III)1, in sp ite
of the exchange-inertness of Co(III) ion, indicates that
the protein-cleavage products dissociate from the
Co(III) ion effectively. In this regard, carboxylates
bound to Co(III) complexes can be freed hydrolytically
through C–O bond cleavage.29
Discussion
Libraries of PNA oligomers with the general structure
of CycAc(Q)nLysNH2 were prepared with an automated
synthesizer by the Split-Pool method. In each step of
attaching a new PNA monomer to the amino group
provided by the polymer support, equal amounts of
A*X, T*X, GX, and CX were added assuming equal
reactivity of the four monomers in the coupling reac-
tion. The library of CycAc(Q)nLysNH2 contains a large
number of PNA oligomers. The theoretical number of
the PNA oligomers contained in the library is 4n
(262,144 when n=9), but the actual number might be
considerably smaller than this number because the
assumption of the equal reactivity may not be valid.
Figure 8. MALDI-TOF MS of Co(III)1; m/z 2907.39 (M+H)+
(C111H153N64O25S2Co calcd 2906.91).

J. W. Jeon et al. / Bioorg. Med. Chem. 11 (2003) 2901–2910
2907
The rate for cleavage of Mb by Cu(II)1 or Co(III)1 was
unaffected by the removal of O2 from the reaction mix-
tures. This result and previous observation8,13,15 for
hydrolytic cleavage of peptide bonds by Cu(II) com-
plexes of tetraaza ligands and Co(III) complexes suggest
the hydrolytic nature of cleavage of Mb by the catalysts
examined in this study. The hydrolytic nature of the
Mb cleavage was further supported by the successful
N-terminal sequencing of the cleavage products
obtained with Co(III)3 by Edman degradation, which
demonstrates generation of a primary amino group by
the Mb cleavage.
Indeed, the analogues cleaved Mb at one position as
summarized in Table 2 and N-terminal sequencing
experiment performed with the cleavage products
obtained with Co(III)3 clearly identified the cleavage
site.
The PNA 9-mer contained in 1 is presumably bound on
the surface of Mb, although no information is available
at present for the structure of the complex formed
between the PNA 9-mer and Mb. It would be possible
to design protein-cleaving catalysts by attaching the
catalytic center either to the N- or the C-terminus of the
PNA portion of 1 if both of the termini are located on
the surface of Mb in the PNA–Mb complex. In
Co(III)2, the Co(III)Cyc is attached to the C-terminus.
With Co(III)2, however, Mb was not cleaved. It is pos-
sible that Co(III)Cyc is not located in a productive
position in the Co(III)2-Mb complex. It is also possible
that the l-Lys moiety of Co(III)1 is essential for recog-
nition of Mb and that Co(III)2 is inactive due to the
absence of the l-Lys moiety.
The lack of catalytic activity of Co(III)18 in Mb clea-
vage in contrast to Co(III)1 indicates that the sequence
of the PNA oligomer contained in 1 is important for the
catalytic activity, in view of the subtle difference in the
structure of 1 and 18. The Cu(II) complex of 1 as well as
the Co(III) complexes of 1, 3, 6, 7, 9, and 10 can be
regarded as artificial proteases selective for Mb on the
basis of the catalytic turnover illustrated in Figure 2,
hydrolytic cleavage of the peptide bonds, and recognition
of Mb.
In Co(III)3, the methyl groupwas attached to the Cyc-
acetyl moiety in an attempt to suppress rotational free-
dom of the acetyl moiety connecting Cyc to the binding
site. The catalytic activity, however, was considerably
reduced. The additional methyl groupdid not raise the
effective molarity of the catalytic center in the Mb–cat-
alyst complex. In Co(III)4–Co(III)8, an amino acid
moiety is inserted between the Cyc-acetyl moiety and
the binding site. When Gly or b-Ala was inserted
[Co(III)4 or Co(III)8], the catalytic activity was almost
completely lost. Elongation of the linker by three or
four atoms could raise the rotational freedom of the
linker, reducing the effective molarity of Co(III)Cyc in
the catalyst–Mb complex. When l-Aspwas inserted
[Co(III)5], the catalytic activity was lost again. The
extra carboxyl groupintroduced by Aspdid not pro-
duce an active conformation for the catalyst. On the
other hand, catalytic activity was manifested when
l-Pro or l-Lys [Co(III)6 or Co(III)7] was inserted. The
unique conformation of l-Pro apparently provided the
catalyst with productive conformation. The cationic
center introduced to the linker by the e-amino groupof
l-Lys in Co(III)7 provided a productive conformation
presumably by electrostatic interaction with the Mb
surface. It is noteworthy that Mb is cleaved at the same
position even when the linker was elongated by three
atoms in Co(III)6 and Co(III)7 compared with Co(III)1
and Co(III)3. In Co(III)9, an additional Co(III)Cyc
moiety is introduced to the e-amino groupof the l-Lys
moiety of Co(III)7. The catalytic activity, however, was
not considerably improved upon introduction of the
additional Co(III)Cyc moiety. Since identical protein
fragments were obtained for Co(III)7 and Co(III)9, it
appears that the Co(III)Cyc attached to the a-amino
group, rather than the e-amino group, of the l-Lys
portion in Co(III)9 is the catalytic group.
Among the metal ions tested, Co(III) manifested the
highest protein-cleaving activity upon complexation to
1. It is noteworthy that Co(III) complexes may be more
suitable for medical uses compared with Cu(II) com-
plexes since metal transfer to metal-abstracting materi-
als in living body should be substantially slower for the
Co(III) complexes due to the exchange-inertness of
Co(III). The kcat measured with Co(III)1 at pH 7.5 and
37 ꢁC corresponds to the half-life of 30 h. This may be
compared with the half-life30 of 200 years for sponta-
neous hydrolysis of unactivated amides measured under
identical conditions. The kcat may be further compared
with the kcat of 0.18 hꢀ1 (at the optimum pH of 9 and
25 ꢁC; corresponding to a half-life of 3.8 h) measured
with a catalytic antibody31 with peptidase activity eli-
cited by a joint hybridoma and combinatorial antibody
library approach in the hydrolysis of an amide sub-
strate. Both Co(III)1 and the catalytic antibody man-
ifest high affinity for their substrates. The kcat of the
Mb-selective artificial protease can be improved further
by raising the effective molarity of the catalytic group
toward the scissile peptide linkage in the complex
formed between Mb and the artificial protease.
Mb is cleaved at two different positions by the action of
Co(III)1 producing two pairs of protein fragments. It is
not clear at present whether the two cleavage sites
involve different Co(III)1–Mb complexes or originate
from an identical one. More information on binding
mode of the PNA portion of the catalysts would be
obtained by crystallographic studies. Since it was
unsuccessful to identify the N-terminal residues of the
cleavage products obtained with Co(III)1, analogues of
Co(III)1 containing both the PNA moiety and the
Co(III)Cyc unit were synthesized. A change in the rela-
tive position of the PNA portion and the Co(III)Cyc
would induce different catalytic behavior, and, thus, Mb
might be cleaved at one site, allowing identification of
the newly formed N-terminal amino acid residue.
Metal ions can play several catalytic roles in hydrolysis
of peptide bonds.32,33 Metal ion itself can act as a
Lewis acid to enhance the electrophilicity of the car-
bonyl carbon of the scissile bond upon coordination to

2908
J. W. Jeon et al. / Bioorg. Med. Chem. 11 (2003) 2901–2910
the carbonyl oxygen. Metal-bound hydroxide ion can
act as a nucleophile that attacks at the carbonyl carbon
of the peptide bond. In addition, the metal-bound
water molecule can act as a general acid to protonate
the leaving nitrogen in the rate-determining expulsion
of the amine from the tetrahedral intermediate. Some-
times, the metal ion can play several catalytic roles
simultaneously.
the metal center is important in the design of the pro-
tein-cleaving catalytic drugs.
In the complex formed between the Co(III) complexes
and Mb, the distance between the Co(III) center and the
carbonyl groupof the target amide bond would be
affected by the nature of the chelating ligand. Thus, the
Co(III) ion chelated by 1 may occupy the most produc-
tive position among various chelating ligands examined
here. A small difference in the distance between the cat-
alytic center and the reaction site can cause a dramatic
change in catalytic rates.18 If the marked difference seen
with the Co(III) complexes of 1 and 10–14 is due to this
effect, relative activities of 1 and 10–14 would be differ-
ent for individual catalytic drugs targeted to various
proteins.
The Co(III) complexes of Cyc and related tetraaza
ligands catalyze the hydrolysis of phosphodiesters
including DNA.34ꢀ40 The mechanism of 19 has been
suggested for the phosphodiester hydrolysis.34,36 Pro-
tein-cleavage by Co(III)Cyc of the Mb-cleaving cata-
lysts involves hydrolysis of peptide bonds. A mechanism
(20) similar to 19 can be proposed for the peptide
hydrolysis by Co(III)1 or its analogues. An analogous
mechanism involving formation of a four-membered
ring has been proposed for metal-catalyzed amide
hydrolysis.34,41 The rate-determining stepfor the
mechanism of 20 would be expulsion of the amine moi-
ety from the tetrahedral intermediate (21), which
requires the protonation of the leaving amine. The bell-
shaped pH profiles of kcat illustrated in Figures 4 and 5
agree with the mechanism of 20/21, since protonation of
the hydroxo (20) or oxo (21) ligand at low pHs and the
deprotonation of the ammonium ion (21) at high pHs
would reduce the reactivity.
The intrinsic catalytic power of the Co(III) center
should be controlled by the structure of the ligand. The
Lewis acidity of Co(III) would depend on the nature of
the ligating atoms as well as relative positions of the
ligating atoms. In addition, the steric strain involved in
the rate-determining transition state should be affected
by the structure of the chelating ligands. For the phos-
phodiester hydrolysis by Co(III) complexes of tetraaza
ligands (19), for example, the catalytic rate was sensi-
tive to the structure of the chelating ligands. This was
attributed to the stabilization of the four-membered
ring in the transition state,34,35 since the steric strain in
the ring should be affected by the structure of the che-
lating ligands of the Co(III) center. Similarly, the steric
strain of the four-membered ring of mechanism 20
should depend on the structure of the chelating ligands.
If the marked difference seen with the Co(III) com-
plexes of 1 and 10–14 is due to the intrinsic reactivity
of the Co(III) center originating from electronic and/or
steric effects of the ligands, Cyc is better than 10–17 as
the chelating ligand for protein-cleaving catalytic
drugs.
The effectiveness of protein-cleaving catalytic drugs
exploiting metal complexes as the catalytic centers
would depend on the structure of the chelating ligands
of the catalytic centers. In this regard, various chelating
ligands were tested in the present study. Compounds 1
and 10–14 have characteristic structural features: 1
contains a macrocyclic tetraaza ligand, 10 contains a
macrocyclic triaza-monooxo ligand, 11 contains a
macrocyclic triaza ligand, 12 contains an acyclic tetra-
aza ligand with one aromatic nitrogen atom, 13 con-
tains an acyclic triaza ligand with one aromatic
nitrogen atom, and 14 contains an aliphatic acyclic
tetraaza ligand. How the catalytic outcome is affected
by the chelating ligands with the various structural
features was examined.
The catalytic center and the binding site are the two
essential components of target-selective artificial pro-
teases. A large number of small molecules are known to
have high affinity for various disease-related proteins.
Those molecules can be employed as the binding sites of
the artificial proteases. Small molecules with high affi-
nity for a target enzyme or receptor may be searched by
the combinatorial method. Even if they have high affi-
nity, they cannot be used as drugs unless the target
proteins are considerably deactivated upon complexa-
tion with the small molecules. Those molecules, how-
ever, can be used as effective binding sites for the
artificial proteases selective for target proteins.
MALDI-TOF MS of the product solutions revealed
that an identical peptide bond of Mb is cleaved by
Co(III)1 and Co(III)10. Co(III)10 complexed to Mb is
2–3 times less reactive than Co(III)1 bound to Mb as
revealed by the kcat values summarized in Table 1. On
the other hand, Co(III) complexes of 11–14 did not
show any catalytic activity to cleave Mb, although the
PNA portions of the complexes should have been
bound to Mb. Marked differences are observed for the
catalytic activities of Co(III) complexes of 1 and 10–14,
demonstrating that selection of the chelating ligand for
The present study demonstrates that target-selective
artificial proteases can be designed by choosing appro-
priate binding sites, linkers, and catalytic units. The
activity of a target-selective artificial protease would be
improved by the increase in its affinity to the target
protein and by proper orientation of the catalytic group
in the catalyst–protein complex. If the targets are dis-
ease-related proteins, the artificial proteases can serve as
protein-cleaving catalytic drugs.

J. W. Jeon et al. / Bioorg. Med. Chem. 11 (2003) 2901–2910
2909
Experimental
groupof the newly introduced building block was
removed to generate an amino groupwhich was used
for further coupling with the carboxyl group of another
building block. The PNA oligomer derivatives were
detached from the solid support by treatment with TFA
and cresol, and were precipitated by addition of cold
ether. The carboxy groupattached to the resin in the
first coupling step was released as a carboxamide. Then,
the protecting groups attached to the nucleobases and
chelating ligands were removed with TFA and tri-
fluoromethanesulfonic acid. In the synthesis of the
library of CycAc(Q)nLysNH2 by the Split-Pool method,
equal amounts of A*X, T*X, GX, and CX were used in
the reaction mixture employed in coupling with the
amino ends attached to the polymer support. The same
general procedure was used for the synthesis of 1–18.
Synthesis of catalysts
PNA derivatives were synthesized by automated proce-
dures using an Expedite Model 8909 Nucleic Acid
Synthesis System with various building blocks. Building
blocks GX, CX, GlyX, AspX, ProX, BalX, and LysX
were purchased from commercial sources whereas A*X
and T*X as well as 1X were synthesized as reported20
previously and 2Y, 3X, and 9X–17X were synthesized as
1
summarized in Schemes 1–11. 2Y: H NMR (300 MHz,
CDCl3) d 7.72 (m, 2H), 7.57 (m, 2H), 7.40–7.16 (m,
19H), 5.05–4.85 (br s, 6H), 4.37 (m, 1H), 4.20–4.18 (m,
2H), 4.06–3.95 (br s, 4H), 3.70 (br s, 2H), 3.40–3.10 (br
m, 18H), 2.83–2.78 (br s, 4H); HRMS exact mass
1013.1403 (M+H)+, calcd for C55H62N7O12 1013.1370.
3X: 1H NMR (300 MHz, CDCl3): d 3.53–3.30 (br, 14H),
2.96 (br s, 4H), 1.43 (m, 27H); MS (MALDI-TOF) m/z
531.75 (M+H)+ (C25H47N4O8 calcd 531.67). 9X: 1H
NMR (300 MHz, CD3OD) 4.14 (m, 1H), 3.17–3.46 (br
m, 28H), 3.14 (t, 2H), 2.79–2.70 (br m, 8H), 1.73 (m,
1H), 1.55 (m, 1H), 1.36 (m, 54H), 1.33–1.20 (m, 4H);
MS
(MALDI-TOF)
m/z
2851.49
(M+H)+
(C111H154N64O25S2 calcd 2848.97) for 1, 2879.63
(M+H)+ (C111H154N66O25S2 calcd 2876.99) for 2,
2864.43 (M+H)+ (C112H156N64O25S2 calcd 2863.00)
for 3, 2906.46 (M+H)+ (C113H157N65O26S2 calcd
2906.21) for 4, 2964.86 (M+H)+ (C115H159N65O28S2
MS
(MALDI-TOF)
m/z
1172.48
(M+H)+
calcd
2964.06)
for
5,
2945.94
(M+H)+
(C56H103N10O16 calcd 1172.49). 10X: 1H NMR
(300 MHz, CDCl3): d 7.34–7.25 (m, 10H), 5.11 (m, 4H),
4.23–4.08 (br m, 4H), 3.84–3.50 (br m, 10H), 3.25–3.15
(br m, 4H); MS (MALDI-TOF) m/z 500.50 (M+H)+
(C116H161N65O26S2 calcd 2946.08) for 6, 2977.85
(M+H)+ (C117H166N66O26S2 calcd 2977.14) for 7,
2920.36 (M+H)+ (C114H159N65O26S2 calcd 2920.05)
for 8, 3191.87 (M+H)+ (C127H186N70O27S2 calcd
3189.43) for 9, 2851.49 (M+H)+ (C111H153N63O26S2
1
(C26H34N3O7 calcd 500.58). 11X: H NMR (300 MHz,
CDCl3): d 7.34 (m, 10H), 5.15 (d, 4H), 3.38 (m, 10H),
2.74 (br s, 4H); MS (MALDI-TOF) m/z 456.06
calcd
2849.85)
for
10,
2807.51
(M+H)+
(C109H149N63O25S2 calcd 2805.90) for 11, 2858.87
(M+H)+ (C112H150N64O25S2 calcd 2856.95) for 12,
2816.37 (M+H)+ (C110H145N63O25S2 calcd 2813.88)
for 13, 2824.04 (M+H)+ (C109H152N64O25S2 calcd
2822.93) for 14, 2781.94 (M+H)+ (C107H147N63O25S2
1
(M+H)+ (C24H30N3O6 calcd 456.52). 12X: H NMR
(300 MHz, CDCl3): d 7.63 (t, 1H), 7.14 (br d, 2H), 5.85
(br s, 1H), 4.54 (s, 2H), 4.39 (s, 2H), 3.81 (br s, 2H) 3.45
(br s, 4H), 1.45 (m, 27H); MS (MALDI-TOF) m/z
1
539.75 (M+H)+ (C26H43N4O8 calcd 539.65). 13X: H
calcd
2779.87)
for
15,
2739.11
(M+H)+
NMR (300 MHz, CDCl3): d 7.63 (t, 1H), 7.28 (d, 1H),
7.18 (d, 1H), 5.45 (br s, 1H), 4.58 (s, 2H), 4,40 (d, 2H),
4.16 (s, 2H), 1.50 (m, 18H); MS (MALDI-TOF) m/z
(C105H142N62O25S2 calcd 2736.80) for 16, 2839.36
(M+H)+ (C109H145N61O29S2 calcd 2837.86) for 17,
2824.08 (M+H)+ (C111H155N61O26S2 calcd 2823.96)
for 18.
1
396.52 (M+H)+ (C19H30N3O6 calcd 396.47). 14X: H
NMR (300 MHz, CDCl3): d 3.98 (s, 2H), 3.32–3.16 (m,
6H), 2.65 (br s, 6H), 1.45 (m, 27H); MS (MALDI-TOF)
m/z 505.74 (M+H)+ (C23H45N4O8 calcd 505.64). 15X:
1H NMR (300 MHz, CDCl3): d 5.18 (br s, 2H), 3.23–
3.18 (m, 6H), 2.76–2.70 (m, 4H), 1.46 (s, 18H); MS
(MALDI-TOF) m/z 362.53 (M+H)+ (C16H32N3O6
calcd 362.44). 16X: 1H NMR (300 MHz, CDCl3): d
5.35–5.25 (br d,1H), 3.90 (d, 2H), 3.39 (m, 2H), 3.20 (m,
2H), 1.47–1.42 (m, 18H); MS (MADLI-TOF) m/z
Quantification of PNA derivatives was carried out with
UV spectra by approximating that the absorbance at
260 nm of a PNA derivative was equal to the sum of
those of constituent nucleobases. Insertion of Co(III)
ion to the chelating ligands of 1–18 was carried out
according the literature method by insertion of Co(II)
ion followed by air-oxidation of the Co(II) ion to
Co(III).27
1
319.42 (M+H)+ (C14H27N2O6 calcd 319.38). 17X: H
NMR (300 MHz, CDCl3): d 3.45 (s, 4H), 2.78 (t, 2H),
2.50 (t, 2H), 1.79 (dd, 2H), 1.46 (s, 18H); MS (MALDI-
TOF) m/z 332.46 (M+H)+ (C16H30N1O6 calcd 332.42).
Measurements
In kinetic measurements, the temperature was con-
trolled with an immersion circulator manufactured by
Fisher Scientific. pH measurements were carried out
with a Dongwoo Medical DP-880 pH/Ion meter.
The degree of cleavage of proteins was measured by
SDS-PAGE with a Mighty Small II SE 250 model. Den-
sities of the electrophoretic bands were analyzed with a
AlphaImagerTM 2200 model and a AlphaEaseTM model.
MALDI-TOF MS analysis was performed with a Voy-
The protocol of the commercial automated system can
be summarized as follows: The solid support was a Sieber
amide resin built on crosslinked polystyrene onto which
polyethyleneglycol was grafted. The amino group gener-
ated on treatment of the resin with trifluoroacetic acid
(TFA) was coupled with the carboxyl group of a building
block with the aid of O-(7-azabenzo-triazol-1-yl)-1,1,3,3-
tetramethyluronium hexafluorophosphate. The resulting
resin was treated with acetic anhydride to block unreac-
ted amino group. By treatment with piperidine, fmoc
TM
ger-DE STR Biospectrometry Workstation Model.
UV-Vis spectra were taken with a Beckman DU 68
spectrophotometer. N-Terminal sequencing was carried

2910
J. W. Jeon et al. / Bioorg. Med. Chem. 11 (2003) 2901–2910
out by Korea Basic Science Research Institute with a
Procise 491 protein sequencer. Distilled and deionized
water was used for preparation of buffer solutions.
Buffers (0.05 M) used in this study were acetate (pH 4.5–
5.0), 2-(N-morpholino)ethanesulfonate (pH 5.5–6.5), N-
2-hydroxyethylpiperazine-N0-ethanesulfonate (pH 7–8),
N-tris(hydroxymethyl)methyl-3-aminopropanesulfonate
(pH 8.5) and boric acid (pH 9). All buffer solutions were
filtered with 0.45 mm Millipore microfilter and auto-
claved before use in the kinetic measurements. The
stock solution of Mb was prepared by dissolving horse
heart Mb (purchased from Sigma, used without further
purification) in water and was kept at 4 ꢁC. Mb is oxi-
dized to metMb in the presence of oxygen. The Mb used
in the present study was also in the met form as checked
by its Vis spectrum.42
17. Suh, J. Synlett 2001, 9, 1343.
18. Kirby, A. J. Adv. Phys. Org. Chem. 1980, 17, 183.
19. Preliminary results were published in ref 20.
20. Jeon, J. W.; Son, S. J.; Yoo, C. E.; Hong, I. S.; Song, J. B.;
Suh, J. Org. Lett. 2002, 4, 4155.
21. Uhlmann, E.; Peyman, A.; Breipohl, G.; Will, D. W.
Angew. Chem., Int. Ed. 1998, 37, 2796.
22. Nielson, P. E. Curr. Opin. Struct. Biol. 1999, 9, 353.
23. Stryer, L. Biochemistry, 4th ed.; Freeman: New York,
1995; p792.
24. Lohse, J.; Dahl, O.; Nielson, P. E. Proc. Natl. Acad. Sci.
U.S.A. 1999, 96, 11804.
25. Izatt, R. M.; Pawlak, K.; Bradshaw, J. S. Chem. Rev.
1991, 91, 1721.
26. Greenwood, N. N.; Earnshaw, A. Chemistry of the Ele-
ments; Reed: Oxford, 1997; p1123.
27. Castillo-Blum, S. E.; Sosa-Torres, M. E. Polyhedron 1995,
14, 223.
28. In Cu(II)1 or Co(III)1, 20 amide groups are present.
Intramolecular catalysis by the Cu(II) or Co(III) center in the
hydrolysis of the first amide grouplinked to the CycAc moiety
involves the transition state containing a five-membered ring.
Lack of appreciable intramolecular catalytic activity of the
Cu(II) or Co(III) center for the hydrolysis of the first amide
groupcan be ascribed to involvement of considerable steric
strain in the five-membered ring structure. For the other amide
groups, the intramolecular catalysis would not be effective in
view of the ring size of the transition state.
Acknowledgements
This work was supported by Korea Research Founda-
tion Grant (KRF 2001-015-DS0029).
References and Notes
29. Well, M. A.; Bruice, T. C. J. Am. Chem. Soc. 1977, 99,
5341.
30. Smith, R. S.; Hansen, D. E. J. Am. Chem. Soc. 1998, 120,
8910.
31. Gao, C.; Lavey, B. J.; Lo, C.-H. L.; Datta, A.; Went-
worth, P., Jr.; Janda, K. D. J. Am. Chem. Soc. 1998, 120,
2211.
1. Silverman, R. B. The Organic Chemistry of Drug Design
and Drug Action; Academic: San Diego, 1992; p52, 146.
2. Ganellin, C. R., Roberts, S. M., Eds. Medicinal Chemistry,
2nd ed. Academic: London, 1993.
3. Veerapandian, P., Ed. Structure-Based Drug Design. Mar-
cel Dekker: New York, 1997.
4. Leff, P., Ed. Receptor-Based Drug Design. Marcel Dekker:
New York, 1998.
5. Suh, J.; Hah, S. S. J. Am. Chem. Soc. 1998, 120, 10088.
6. Suh, J.; Oh, S. J. Org. Chem. 2000, 65, 7534.
7. Oh, S.; Chang, W.; Suh, J. Bioorg. Med. Chem. Lett. 2001,
11, 1469.
8. Sutton, P. A.; Buckingham, D. A. Acc. Chem. Res. 1987,
20, 357.
9. Hegg, E. L.; Burstyn, J. N. J. Am. Chem. Soc. 1995, 117,
7015.
32. Suh, J. Acc. Chem. Res. 1992, 25, 273.
33. Suh, J. In Perspectives on Bioinorganic Chemistry; Hay, R.
W., Dilworth, J. R., Nolan, K. B., Eds.; JAI: London, 1996;
Vol. 3, p115.
34. Chin, J.; You, X. J. Am. Chem. Soc. 1988, 110, 223.
35. Chin, J. Acc. Chem. Res. 1991, 24, 145.
36. Dixon, N. E.; Geue, R. J.; Lambert, J. N.; Moghaddas, S.;
Pearce, D. A.; Sargeson, A. M. Chem. Commun. 1996, 1287.
37. Hettich, R.; Schneider, J.-J. J. Am. Chem. Soc. 1997, 119,
5638.
10. Suh, J.; Oh, S. Bioorg. Med. Chem. Lett. 1996, 6, 1067.
11. Jang, B.-B.; Lee, K. P.; Min, D. H.; Suh, J. J. Am. Chem.
Soc. 1998, 120, 12008.
12. Parac, T. N.; Ullmann, G. M.; Kostic, N. M. J. Am.
Chem. Soc. 1999, 121, 8663.
13. Moon, S.-J.; Jeon, J. W.; Kim, H.; Suh, M. P.; Suh, J. J.
Am. Chem. Soc. 2000, 122, 7742.
14. Takarada, T.; Yashiro, M.; Komiyama, M. Chem. Eur. J.
2000, 6, 3906.
38. Hettich, R.; Schneider, J.-J. J. Chem. Soc., Perkin Trans. 2
1997, 2069.
39. Jeung, C.-S.; Kim, C. H.; Min, K.; Suh, S. W.; Suh, J.
Bioorg. Med. Chem. Lett. 2001, 11, 2401.
40. Jeung, C.-S.; Song, J. B.; Kim, C.-H.; Suh, J. Bioorg. Med.
Chem. Lett. 2001, 11, 3061.
41. Chin, J.; Jubian, V.; Mrejen, K. J. Chem. Soc., Chem.
Commun. 1990, 1326.
42. Bodwell, C. E.; McClain P. E. In The Science of Meat and
Meat Products, 2nd ed.; Price, J. F., Schweigert, B. S., Eds.;
W. H. Freeman: San Francisco, 1971; p97.
15. Suh, J.; Moon, S.-J. Inorg. Chem. 2001, 40, 4890.
16. Suh, J. Adv. Supramol. Chem. 2000, 6, 245.