Toward a rational design of peptide inhibitors of ribonucleotide reductase: Structure-function and modeling studies

Article abstract of DOI:10.1021/jm000335r

Mammalian ribonucleotide reductase, a chemotherapeutic target, has two subunits, mR1 and mR2, and is inhibited by AcF¹TLDADF⁷, denoted P7. P7 corresponds to the C-terminus of mR2 and competes with mR2 for binding to mR1. We report re

Full text of DOI:10.1021/jm000335r

36
J . Med. Chem. 2001, 44, 36-46
Tow a r d a Ra tion a l Design of P ep tid e In h ibitor s of Ribon u cleotid e Red u cta se:
Str u ctu r e-F u n ction a n d Mod elin g Stu d ies
Bari A. Pender,‡ Xu Wu,‡ Paul H. Axelsen,§ and Barry S. Cooperman*,‡
Departments of Chemistry and Pharmacology, University of Pennsylvania, Chemistry Building,
Philadelphia, Pennsylvania 19104
Received August 3, 2000
Mammalian ribonucleotide reductase, a chemotherapeutic target, has two subunits, mR1 and
mR2, and is inhibited by AcF1TLDADF7, denoted P7. P7 corresponds to the C-terminus of mR2
and competes with mR2 for binding to mR1. We report results of a structure-function analysis
of P7, obtained using a new assay measuring peptide ligand binding to mR1, that demonstrate
stringent specificity for Phe at F7, high specificity for Phe at F1, and little specificity for the
N-acyl group. They support a structural model in which the dominant interactions of P7 occur
at two mR1 sites, the F1 and F7 subsites. The model is constructed from the structure of
Escherichia coli R1 (eR1) complexed with the C-terminal peptide from eR2, aligned sequences
of mR1 and eR1, and the trNOE-derived structure of mR1-bound P7. Comparison of this model
with similar models constructed for mR1 complexed with other inhibitory ligands indicates
that increased F1 subsite interaction can offset lower F7 subsite interaction and suggests
strategies for the design of new, higher affinity inhibitors.
I. In tr od u ction
demonstrated previously that the N-acetylated peptide
AcF1TLDADF7, denoted P7, which corresponds to the
seven C-terminal residues of mouse R2 (mR2), binds to
mouse R1 (mR1), blocking subunit association and
inhibiting enzyme activity.19 Our choice of P7 was based
on the strong homology among eucaryotes and some
viruses for the seven residues at the C-terminus of
eucaryotic R2, a homology that has been reinforced by
new additions to the sequence database (Table 1).
Subsequent structure-function studies20 resulted in the
following conclusions: (1) P7 is the minimal core peptide
length required; deletions result in loss of inhibitory
activity, while extensions at the N-terminus have little
effect. (2) A free carboxylate is required at the C-
terminus. (3) Positions 1 and 7 have a strong preference
for Phe, as seen by a >40-fold loss of activity on Leu
substitution. (4) The charged/polar residues at positions
2 and 6 are not crucial to peptide binding, although
these residues are evolutionarily conserved in mam-
malian R2 C-termini.20 As demonstrated by trNOE
NMR,21,22 the mR1-bound structures of both P7 and
AcYTLDADF have reverse turn conformations. On the
basis of these structures, we have designed and syn-
thesized two new sets of compounds: (1) cyclic lactam
peptides, one of which (AcFcyc[ELDK]DF, denoted
cycP7) is a 2.5-fold more potent inhibitor of mRR than
P7,23,24 and (2) a tetrahydropyran-based â-turn mimetic
which was less successful.25
Ribonucleotide reductase (RR) catalyzes the reduction
of ribonucleotides to 2′-deoxyribonucleotides, the rate-
limiting step of DNA biosynthesis. Class I RRs, which
include RR from eucaryotes, Escherichia coli, and
certain bacteriophages and viruses, are composed of two
subunits, R1 and R2. R1 contains the active site as well
as binding sites for allosteric ligands, while R2 contains
a µ-oxo bridged diferric center, which is required for
generating, and presumably stabilizing, a catalytically
essential tyrosine radical. Enzyme activity is initiated
via electron transfer between the active site in R1 and
the conserved tyrosine radical in R2, and it is completely
dependent on the association of the two subunits.1,2
RR has been shown to be linked with malignant
transformation and tumor cell proliferation3 and is a
well-recognized target for cancer chemotherapeutic and
antiviral agents.3,4 The RR inhibitors hydroxyurea and
2′-deoxy-2′,2′-difluorocytidine (gemcitabine) are used in
clinical practice to treat certain types of leukemia3 and
pancreatic cancer,5 respectively. Clinical trials are
underway to broaden the use of these compounds,
sometimes in combination with others, for both cancer
and HIV chemotherapy.6-13 Because both compounds
have toxic side effects,6,14-16 and it is clear that resis-
tance to RR inhibitors may emerge during therapy,17
the development of safer, more efficacious ribonucleotide
reductase inhibitors remains a high priority for RR-
targeted chemotherapies.
Our efforts continue to focus on generating more
potent inhibitors of mRR. Below we describe the syn-
thesis of three positional libraries derived from P7, of
general structures AcFTLDADX, AcXTLDADF, and XC-
(O)FTLDADF. These libraries are tested using a new
assay that measures the ability of a given potential
ligand to compete with Sepharose-FTLDADF binding
to mR1, giving results equivalent to effects on mRR
activity. These and other structure-function results are
rationalized in terms of a structural model of the
Inhibitors that disrupt enzyme quaternary structure
remains an attractive goal, because of the intrinsically
higher specificity offered by this approach.18 We have
* To whom to address correspondence. Tel: (215) 898-6330. Fax:
(215) 898-2037. E-mail: cooprman@pobox.upenn.edu.
‡ Department of Chemistry.
§ Department of Pharmacology.
10.1021/jm000335r CCC: $20.00 © 2001 American Chemical Society
Published on Web 12/08/2000

Design of Peptide Inhibitors of Ribonucleotide Reductase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 1 37
Ta ble 1. Comparison of Seven C-Terminal Residues of R2 and Residues 339, 340, 343, 725, 728, and 729 (mR1 numbering) at the R1
Peptide Binding Sitea
a
Sequences were downloaded from the protein sequence database (EMBL, Switzerland). The binding site alignment is based on the
observed peptide binding site in the E. coli crystal structure26 and multiple sequence alignments made using CLUSTALW at the Institute
for Biomedical Computing, Washington University of St. Louis, St. Louis, MO.

38 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 1
Pender et al.
mR1.P7 complex that is based on the crystal structure
of the R1 from E. coli (eR1) complexed with the C-
terminal peptide from E. coli R2 (eR2),26 the aligned
sequences of mR1 and eR1, and the trNOE-derived
structure of P7 bound to mR1. A similar approach is
used to model the structure of cycP7 bound to mR1. The
combination of the structure-function and modeling
results provides specific suggestions for the development
of new, high affinity inhibitors of mRR.
peptide, 3, a result extending the earlier observation
that Leu substitution lowers inhibitory activity >40-
fold.20 The Gln-substituted peptide 4 also binds poorly,
so that any possible increases in affinity due to H-
bonding do not outweigh the loss in affinity on substitu-
tion with a nonaromatic residue. Moreover, the precise
electronic character of the aromatic ring is apparently
critical, judging by the decreases in binding activity
when the isosteric27 fluoro-phenyl or pyridyl groups
(peptides 5-8) substitute for phenyl. Thus, a small drop
is seen for the o-F substitution, while decreases on
F5-Phe or pyridyl substitution are much larger.
Second, the size and placement of the aromatic
residue are crucial, as shown by the poor binding
activities of the peptides substituted either with the
other coded aromatic amino acids, Tyr, Trp, and His
(9-11), or with Phg (12). Elsewhere28 it has been
reported that homoPhe substitution at position lowers
affinity only 2-fold, so that the F7 subsite appears less
sensitive to overfilling than to underfilling (12). Inter-
estingly, substitution with the D-Phe isomer (13) results
in only a relatively small decrease in binding affinity
(7-fold). This substitution may distort the peptide in a
manner similar to the cyclic peptide cycP7, though with
a less favorable effect on binding, as discussed below.
Third, we note a large loss in binding activity on
replacing the carboxylate with carboxamide (14), im-
plicating the need for a negative charge at the C-
terminus. Consistent with this conclusion is the earlier
finding of a 50-fold drop in inhibitory potency for the
peptide AcFTLDADFAA (15),20 which extends P7 at the
C-terminus.
Fourth, the F7 subsite is very constrained sterically,
since ring substitution at the o-, m-, and p-positions of
a C-terminal Phe all lead to lower binding (16-25). The
amino, hydroxy-, and methoxy-substituted derivatives
were designed as possible hydrogen bond donors and
acceptors. Clearly, any possible positive effect of such
substitution is more than outweighed by negative steric
and/or other factors.
Fifth, the F7 subsite has a hydrophobic character, as
demonstrated most clearly by the correlation between
hydrophobicity and affinity for the six peptides with
p-substitution in the C-terminal Phe. Thus, the two
substituents with positive Π values (Table 3a), p-CH3
and p-Cl (22, 23), confer much smaller losses in affinity
(3-4-fold reductions), the p-OCH3 (24) derivative which
has a Π value close to zero confers an intermediate loss
(16-fold worse), whereas the four substituents with
negative Π values, p-OH (9), p-NH2 (25), and p-NO2 (19,
which is 47% of the o-OMe:p-NO2 mixture, see Experi-
mental section) exhibit dramatic losses in binding
affinity (g20-fold reductions). In addition, the affinities
of peptides with coded aromatic amino acids in the
C-terminal position also correlate with their relative
hydrophobicities: Phe (1) > Trp (10) > Tyr (9), His (11)
> Gln (4) (Table 3b).
II. Resu lts a n d Discu ssion
A New m R1-P ep tid e Bin d in g Assa y. In earlier
work we evaluated the efficacy of peptide inhibitors in
an in vitro RR activity assay, which yields a Ki value
(in the form of an IC50) for a given inhibitor.19,20
However, this assay, which is based on a column
separation of product from substrate, is cumbersome for
screening a large number of compounds, and requires
materials that are both costly and radioactive.
Here we introduce a new, simpler assay that mea-
sures the direct competition between an inhibitory
ligand and P7 for the peptide-binding site on the mR1
protein, eliminating the need for both mR2 protein and
radioactive substrate. The assay is based on an applica-
tion of the FTLDADF-Sepharose affinity column intro-
duced earlier for mR1 purification.19 It measures the
amount of mR1 eluted when a mixture of mR1 and the
candidate inhibitor are drawn through the affinity
column by centrifugation. The higher the affinity of an
inhibitory ligand for mR1, the more mR1 is eluted.
Comparison of the results obtained with P7 allows a
quantitative estimate of the Kd of the candidate inhibi-
tory ligand (see Experimental Section). The close agree-
ment in the results obtained for several potential
inhibitory ligands in both the activity and binding
assays (Table 2) validates the new assay. Importantly,
the new assay directly tests that the candidate molecule
is inhibiting enzyme activity by binding to the R2
C-terminal peptide site, rather than by another mech-
anism. This will be especially important for the testing
of smaller, less peptide-like inhibitors that should result
from the more detailed understanding of the peptide
binding site reported in this work (see below).
Str u ctu r e-F u n ction Stu d ies. The strong conser-
vation of F1 and F7 in the C-terminal sequences of
eucaryotic R2 (Table 1) suggests that these residues may
have strong interactions with mR1, at loci denoted the
F1 and F7 subsites, respectively, and raises the question
of whether gains in affinity toward mR1 could be
achieved by substitution at these positions in P7.
Further, acetylation of the terminal N is known to be
important for P7 binding, but is unclear whether
acylation with another group would not confer higher
affinity. These points are addressed with the three
single position libraries described below, the results for
which are summarized in Table 2. These results also
led to a preliminary investigation of the binding of
simple N-protected derivatives of Phe and Leu.
Summing up, all substitutions for Phe at the C-
terminus of P7 lead to decreased binding activity. The
very stringent binding requirements that can be in-
ferred from these results make it unlikely that peptide
binding affinity can be increased through substitution
at the C-terminus, at least for linear peptides.
(a ) Libr a r y I. AcF TLDADX. Results for library I,
including peptides 2-25, lead to the following conclu-
sions regarding the F7 subsite, and the properties a
residue must have for optimal binding to the F7 subsite.
First, the substituted residue must be aromatic. This
is shown by the poor binding of the Cha-substituted
(b) Libr a r ies II, AcXTLDADF , a n d III, XC(O)-

Design of Peptide Inhibitors of Ribonucleotide Reductase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 1 39
Ta ble 2. Relative Binding Affinities of Peptide Inhibitors of Mouse RR
a
d
f
New binding assay. b Activity assay. c Reference 20. Assayed as a 53:47 18:19 mixture. e Reference 21. S. Nair, private communication.
g
Reference 19.

40 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 1
Ta ble 3. Measures of Hydrophobicity
Pender et al.
(a) Comparison of the Hydrophobic Character
of Substituents on an Aromatic Ring (Π)50
substituent
Π valuea
pCl
+0.71
+0.56
-0.02
-0.28
-0.67
-1.23
pCH3
pOCH3
pNO2
pOH
pNH2
(b) Comparison of the Hydrophobicities
of the Coded Amino Acids51
amino acid
hydrophobicityb (kcal/mol)
Leu
Phe
Trp
Tyr
His
Gln
-3.98
-2.04
-1.39
+1.08
+5.60
+6.48
a
b
Reference 50. Reference 51.
F TLDADF . Library II probes the importance of the
N-terminal Phe. The results obtained (Table 2) parallel
those obtained with library I in indicating the impor-
tance of an aromatic residue at this position, as each of
the substitutions containing an aliphatic side chain,
including Cha (26) and Leu (27), had quite reduced
binding activity (>30-fold). By contrast, each of the
substitutions containing an aromatic side chain had
measurable binding activity, albeit lower than P7 itself,
and the effects of identical substitutions [Tyr (30), Phg
(31)] were less deleterious at position 1 than at position
7. We interpret these results as suggesting a lower
binding stringency at the F1 subsite than at the F7
subsite, raising the possibility that affinity might yet
be increased through suitable substitution at this site,
e.g., substituted naphthyl (32) derivatives.
Library III probes the importance of the nature of the
acyl group capping the terminal nitrogen. Evidence that
the amide formed by N-terminal acylation contributes
to peptide affinity, rather than simply removing the
negative effect of a positively charged ammonium group,
is provided by the result that the desamino derivative
(33) has very poor inhibitory activity. However, the
positive effect seems to be confined to the amide group
itself, and is largely independent of the nature of
carbamide substitution. Minor enhancement in binding
activity is obtained with large aromatic acyl groups (35-
38), but the effects are less than 2-fold vis-a`-vis P7, and
even a formyl capping group (42) affords a peptide
having an affinity only 2-fold less than P7 itself. These
results suggest that the group attached to the carbonyl
has little interaction with mR1, consistent with our
earlier finding that the acetylated forms of the mR2
C-terminal octamer and nonamer have the same inhibi-
tory activity as P7.20
F igu r e 1. E. coli R1 peptide binding site. (a) Stereoview of
the peptide binding site in the eR1 crystal structure with
bound eR2 peptide (pdb reference code 1RlR). Colored in red
is the eR2 peptide. Colored in blue are the residues within 7
Å of the peptide, including the RI and R13 helicies. (b)
CLUSTALW alignment of mR1 residues with eR1 residues 7
Å from the eR2 peptide.
weaker than P7) was virtually identical. By contrast,
Boc-Phe showed very much poorer inhibition of mRR
activity (g100-fold weaker than P7), indicating that
FmocPhe is interacting primarily via the Fmoc group
rather than via Phe. This suggestion is supported by
the finding that FmocLeu inhibited mRR equally as well
as FmocPhe. The important implications of relatively
strong FmocPhe and FmocLeu binding are considered
further below.
Mod els of P 7 a n d cycP 7 Bou n d to m R1. Three-
dimensional models of P7 and cycP7 bound to mR1 were
constructed, based on (a) the known crystal structure
of the complex between eR1 and 10 C-terminal residues
of bound eR2 C-terminal peptide26 (the structure of mR1
is unavailable); (b) the aligned primary structures of eR1
and mR1 (Figure 1)sa direct comparison of the eR1 and
mR1 sequences showed them to be 29% identical and
53% homologous,29 constituting compelling evidence for
similar three-dimensional structures;30 and (c) struc-
tures of the mR1-bound forms of P721,22 and cycP723
determined by trNOE. The success of these models in
rationalizing the structure-function results summa-
rized in Table 2 and related earlier results, as described
below, provides a basis for confidence that they will be
useful in guiding the future design of P7 analogues with
heightened binding activity toward mR1. Below we use
(c) N-P r otected P h e a n d Leu . The clear indication
of strong interactions at the F7 and F1 subsites led us
to investigate N-terminally protected Phes as ligands
for mR1, as assayed via inhibition of mRR activity,
because of the greater dynamic range of this assay.
Interestingly, FmocPhe inhibited mRR with an appar-
ent Ki only 17-fold higher than P7. Moreover, such
inhibition arises from competitive binding to mR1, since
the activity of FmocPhe in the binding assay (13-fold

Design of Peptide Inhibitors of Ribonucleotide Reductase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 1 41
mR1 numbering throughout in discussing the roles of
specific residues.
In the E. coli enzyme, the 10 eR2 C-terminal residues
bind into a shallow cleft between the antiparallel
R-helices 13 (residues 335-345) and I (residues 718-
732) in eR1 (Figure 1). Three major lines of evidence
suggest especially strong conservation of this interface
region. First, biochemical and genetic results show that
residues in mR1 and a viral R1 involved in R2
C-terminal peptide binding align with helix RI in eR1.
Thus, an azidophenyl derivative of FTLDADF was found
to photoincorporate into the 724-735 peptide of mR1,31
and HSV-R1 variants at positions 1090 and 1091,
corresponding to mR1 residues 716 and 717, have been
isolated from strains of HSV-1 that show weak resis-
tance to an R2 C-terminal peptidomimetic inhibitor.32
In addition, the RI sequence partially overlaps a highly
conserved R1 sequence (residues 731-738) that includes
the functionally important Tyr737 and Tyr738 residues.33
Second, in the eR1 crystal structure, eR1 residues
Tyr339 in helix R13 and Tyr728 in helix RI form a strongly
interacting binding pocket for the C-terminal Leu of the
bound eR2 peptide. Inspection of a multiple sequence
alignment reveals a conserved covariance of these three
residues (Table 1). Specifically, for virtually all R2
sequences ending in Leu (procaryotes and some viruses)
residues 339 and 728 are always Phe or Tyr, whereas
for virtually all R2 sequences ending in Phe (eucaryotes,
including mR2, and other viruses) residues 339 and 728
are always Met and Trp, respectively. Since the sum of
the side chain volumes of two Phe residues and one Leu
residue is almost exactly equal to the sum of the
volumes of one Phe, one Trp, and one Met residue,34 R1
residues 339 and 728 and the C-terminal residue of R2
appear to constitute a conserved hydrophobic cluster of
constrained volume at the R1:R2 interface.
Third, mR1-bound P7 has a distinct reverse turn
conformation. Residues TLDA comprise this turn, and
their 12 backbone heavy atoms can be overlayed with
the corresponding atoms of residues LSNF of the bound
E. coli R2 C-terminal peptide (DIDDLSNFQL) bound
to eR1 with an RMSD of 0.35 ( 0.02 Å.21,22
F igu r e 2. Homology models of mR1 bound with (a) P7; (b)
cycP7. Images were generated using GRASP.52 The binding
surface is color coded by distance, with red areas of the model
having essentially no contact with the peptide (3 Å from
nearest point), blue areas having intermediate contacts (2-3
Å), and white areas having close contacts (1-2 Å). Residues
F1 and F7 are colored in black. The extent of the F1 and F7
subsites, as well as selected residues important for peptide
binding (see text for discussion), are indicated in yellow.
Especially noteworthy in contrasting panels a and b are the
enlarged interaction surface at the F1 subsite in panel b, which
now includes Tyr726 and K719 [these residues are in the red
regions in panel a], and the interactions at the F7 subsite with
residues Met339, Lys340, and Trp728 in panel a, which are absent
in panel b.
Along with these similarities, there is also an impor-
tant difference between R2 C-terminal binding to eR1
vs mR1. P7 binds to mR1 with a binding affinity of 10
µM and appears to contain virtually the entire binding
determinant for mR2 interaction with mR1.20,35 In
contrast, the eR2 C-terminal octamer binds only poorly
to eR1 (Ki of 370 µM), and it requires the 20-mer
C-terminal peptide to achieve a dissociation constant
of 20 µM.36 This difference suggests that there are
contacts between P7 and mR1 that are either not made
between eR1 and the eR2 C-terminal octamer or are
much weaker. On the basis of the strong sensitivity of
P7 binding to mR1 to substitution at both the F1 and
F7 positions, we believe it likely that the relatively high
affinity seen for P7 binding to mR1 reflects significant
mR1 interaction with both the C-terminal and N-
terminal halves of P7. This contrasts with the eR2
C-terminal octamer (D*D1LSNFQL7, numbered so as to
be comparable to P7) interaction with eR1, which, as is
clear from examination of the eR1-R2 C-terminal struc-
ture (pdb 1rlr; see also ref 26), occurs principally via
contacts to F5 and L7. Noteworthy in this connection is
the clear homology between the four C-terminal resi-
dues of P7 (-D4ADF7) and the four C-terminal residues
of eR2 (-N4FQL7) vs the lack of homology between the
N-terminal three residues of P7 (F1TL3-) and the 5th-
7th residues from the C-terminus of eR2 (D1LS3-).
(a ) Bou n d P 7. The model of bound P7 (Figure 2a),
constructed as described (see Experimental Section),
contains extensive areas of interaction between both the
N- and C-terminal Phes of the peptide and mR1,

42 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 1
Pender et al.
consistent with the structure-function results sum-
marized in Table 2. Two aromatic residues within the
F7 subsite, Trp728 and Phe725, form edge-to-face interac-
tions with F7. Trp728 is one of the covariant cluster of
residues discussed above, and Phe725 is strongly con-
served when both the N- and C-termini of the R2
C-terminal peptide are Phe (Table 1). F7 also packs with
the side chains of Met339, another of the covariant
tant differences. Most importantly, cycP7 shows an
increased interaction at the F1 subsite at the expense
of a much reduced interaction at the F7 subsite. This
difference results from the C-terminal Phe folding
inward, packing with L3, in a manner that precludes
its being able to extend into the F7 subsite. In so doing
it makes a strong interaction with Tyr726 which now also
interacts with L3, effectively extending the F1 subsite.
Other differences at the F1 subsite vs bound P7 include
stronger interactions of F1 with Phe725 and Glu343 and
of the acetyl group with several residues, especially the
Gln396 side chain, and a weaker L3 interaction with
Glu343. A second significant difference concerns D4,
which salt bridges with Lys719, whereas in bound P7 it
interacts with the P7 backbone only.
The inward folding of the C-terminal Phe residue of
the cyclic peptide may help explain the surprisingly
small decrease in relative binding affinity (9-fold) of
peptide (15), in which the F7 residue is substituted with
D-Phe. Naively, one would expect this substitution to
completely disrupt binding at the F7 subsite, with an
accompanying large loss in binding affinity. We suspect
that AcFTLDAD(D-Phe) may adopt a conformation
similar to that of cycP7, in which the D-Phe side chain
folds back into the peptide itself, reinforcing interaction
with the F1 subsite, though not to the same extent as
the L-Phe side chain in cycP7.
Con clu sion s Goin g F or w a r d . Why does cycP7 bind
2.5 times more tightly to mR1 than does P7,24 despite
having lost the energetically favorable interaction with
the F7 subsite? It is reasonable to assume that the
answer lies in the differences noted above. Our working
hypothesis is that expansion of the interaction at the
F1 subsite in cycP7 compensates for loss of F7 interaction
with the F7 subsite, since the salt bridges seen in our
models are solvent exposed and are not expected to
make major contributions to binding stability. The
importance of the F1 subsite is also suggested by our
finding that FmocPhe and FmocLeu show reasonably
tight binding to mR1 (Table 2). Modeling studies on
FmocPhe and FmocLeu in aqueous solution show that
in both molecules the fluorene packs very close to the
phenyl or the isobutyl side chain. Docking these con-
formations into the model of mR1 (not shown) demon-
strates that the fluorene group binds at the F1 subsite,
occupying the binding region of F1 and L3 in P7. The F7
subsite is simply too small to accommodate either
molecule. It follows from these considerations that very
high affinity ligands for mR1 could be developed that,
while maintaining an expanded contact with the F1
subsite, are also able to make strong contacts with the
F7 subsite. The design, synthesis, and testing of such
molecules is underway.
residues, and with Glu343
.
F1 also shows strong interactions at the F1 subsite,
including favorable edge-to-face interactions with Tyr717
as well as hydrophobic packing with Glu387 and Phe725
.
L3, which is constrained to interact with F1 by trNOE
results, interacts with Asn345 as its primary protein
contact, as well as with Glu343. The importance of these
interactions is shown by the 20-fold drop in affinity on
substituting L3 with Ala.7 In contrast to F7, the F1 and
L3 interactions with mR1 occur along the surface of the
protein and do not localize to a pocket. This is consistent
with the observed lower stringency for binding at the
F1 subsite than at the F7 subsite. The existence and
properties of the F1 subsite offer attractive rationales
both for the higher affinity of P7 for mR1 vs the eR2
C-terminal octamer for eR1, since no interactions com-
parable to the F1 subsite are seen between eR1 and the
corresponding D1LS3 residues in the eR2 C-terminus,
and for the promising binding of the simple molecules
FmocPhe and FmocLeu (Table 2), each of which could
bind to this subsite via the Fmoc group.
The interactions made by the three carboxylates in
the C-terminal half of P7 are also of interest. The
C-terminal carboxylate forms a salt bridge with Lys340
,
which is highly conserved in eucaryotic and some viral
R1s (Table 1), whereas D6 salt bridges with the even
more highly conserved Lys729. This scheme differs from
that seen in the E. coli structure, in which Lys729 forms
the salt bridge with the C-terminus of the E. coli
peptide, which, unlike P7, does not have a free carboxy-
late in the penultimate position. The C-terminus-Lys340
salt bridge appears critical for the mR1.P7 interaction,
given the low binding activity of the C-terminal amide
peptide 14. By contrast, the D6-Lys729 salt bridge adds
little to the energetics of P7 interaction with mR1, since
Ala substitution at position 6 results in little loss of
affinity.20 The third carboxylate, D4, shows no interac-
tions with mR1 but does hydrogen bond through its side
chain to the amide of L3. This interaction is observed
in the trNOE structure of P7 and may be crucial to
holding the peptide in a reverse turn conformation,
given the 30-fold loss in affinity on substituting D4 with
Ala.20
Finally, the model shows relatively few contacts
between mR1 and either residues T2 and A5 or the acetyl
group. This is in accord with the general insensitivity
of binding affinity to substitution at the 2,520 and N-acyl
positions (Table 2).
III. Ch em istr y
Libraries II and III were synthesized by solid phase
peptide synthesis, using standard reagents and proce-
dures as outlined in the Experimental Section. Library
I was generated by solution phase coupling of C-
terminally protected Phe derivatives with the hexamer,
fully protected except for the C-terminus. Phe deriva-
tives that could be purchased with a free N-terminus
and protection at the C-terminus were coupled directly.
However, most Phe derivatives were only commercially
(b) Bou n d cycP 7. The model of bound cycP7 (Figure
2b) shows interesting similarities and differences vis-
a`-vis bound P7. Both molecules share common interac-
tions at the F1 subsite, and for both the D6 side chain
makes strong hydrogen bonds with Lys729. In addition,
residues E2 and K5 show no interaction with mR1,
paralleling the lack of interaction with mR1 of P7
residues T2 and A5. There are, however, several impor-

Design of Peptide Inhibitors of Ribonucleotide Reductase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 1 43
and Boc-pentafluoro-L-Phe were purchased from ChemImpex
International (Wood Dale, IL). Boc-Phe was purchased from
Sigma. Fmoc-2-hydroxy-L-Phe, Fmoc-2-methoxy-L-Phe, Fmoc-
3-hydroxy-L-Phe, Fmoc-3-methoxy-L-Phe, Fmoc-4-methyl-L-
Phe, and Fmoc-L-cyclohexylalanine were a gift from RSP amino
acids (Worcester, MA). Fmoc-p-Nitro-L-Phe, Fmoc-p-Amino-
(Boc)-L-Phe, Fmoc-L-Tyr(tBu), H-L-Phe-OtBu‚HCl, H-L-Trp-
OtBu‚HCl, H-L-Phg-OtBu‚HCl, H-L-Phe-NH2, and all other
Fmoc amino acids were purchased from Bachem Bioscience
(King of Prussia, PA). FmocPhe-Wang resin, HBTU, and
HOBT were purchased from Nova Biochem (La J olla, CA). The
solvents for peptide synthesis were purchased from Fisher
Scientific (Pittsburgh, PA). MicroSpin columns, activated CH
Sepharose 4B, and 8-3H-GDP were purchased from Amersham
Pharmacia Biotech (Piscataway, NJ ). Phenyl Boronate Agarose
gel (PBA-60) was purchased from Millipore (Bedford, MA). All
other chemicals were from Aldrich or Sigma (St. Louis, MO).
The plasmid pET-M2 (pET3a-R2) encoding the mouse R2
subunit was a gift from Dr. Lars Thelander39 and was grown
in BL21(DE3) cells. mR2 was purified by a published method,40
with slight modification as outlined in the thesis of Christian
Hamann.41 The gene encoding the large (R1) subunit of mouse
ribonucleotide reductase was a gift from Dr. Ingrid Caras
(Genentech).42 mR1 was expressed and purified as described
previously.43
available as the N-terminally protected residues. Such
derivatives were reacted with tert-butyl trichloroace-
timidate to protect the C-terminus, followed by N-
terminus deprotection. Reactions for library I were
typically on the scale of 5 µmol. All peptides were
purified to homogeneity by HPLC and gave molecular
ions in the mass spectrum consistent with their molec-
ular weights, with one exception (see below). The parent
peptide (P7, AcFTLDADF) synthesized using this cou-
pling procedure had a binding affinity for mR1, as
measured by the binding assay, that was identical to
that of P7 synthesized earlier by standard automated
solid phase synthesis,20 providing strong evidence for
the lack of racemization during library I generation.
The mass spectrum of AcFTLDADPhe(o-OH) 16 gave
a molecular ion 18 mass units less than expected. We
attribute this to condensation of the o-hydroxy group
with the C-terminus, forming a lactone which is a
3-subsituted-3,4-dihydrocoumarin. Such lactone forma-
tion is favored under the acidic conditions employed for
deprotection. However, known pH effects on both the
rate and equilibrium position of 3,4-dihydrocoumarin
ring opening37,38 strongly suggest that the ring-opened
form dominates under our binding assay conditions, at
pH 8.5.
New R1-P ep tid e Bin d in g Assa y. The preparation of
FTLDADF-Sepharose was described previously.19 Briefly, CH
Sepharose 4B (10 g) is suspended in 50 mL of 1 mM HCl,
subsequently washed on a Buchner funnel with 1.95 L of 1
mM HCl, and dried under suction. The filter cake is added to
50 mL of 0.1 M NaHCO3 containing 15 mg of FTLDADF. The
IV. Con clu sion s
reaction mixture is agitated on
a shaker table at room
We report the use of a new, more efficient and less
expensive assay to directly measure the binding of a
peptide ligand to the R2 C-terminal peptide binding site
of mR1, demonstrate that such binding parallels peptide
ligand inhibitory effects on mRR activity, and determine
binding affinities to mR1 of peptides formed by substi-
tution at the N-terminal, F1, and F7 positions of the R2
C-terminal peptide N-Ac-FTLDADF (denoted P7), as
well as of N-protected Phe and Leu. The results of these
studies, and of earlier studies of P7 substitutions, are
consistent with a structural model of the mR1.P7
complex in which the dominant interactions of P7 occur
at two distinct mR1 loci, the F1 and F7 subsites. The
model is created using QUANTA 97 and is based on (a)
the crystal structure of the R1 from E. coli (eR1)
complexed with the C-terminal peptide from E. coli R2
(eR2), (b) the aligned sequences of mR1 and eR1, (c) the
trNOE-derived structure of P7 bound to mR1. Compari-
son of this model with similar models constructed for
the mR1 complex with both a high affinity cyclic peptide
analogue of P7 and with FmocPhe indicates that in-
creased interaction at the F1 subsite can offset loss of
F7 subsite interaction, and it suggests strategies for the
design of new, higher affinity inhibitors of mRR.
temperature, and A260 is monitored to follow release of the
imidazole group from the resin. Once the A260 levels off (∼4
h), the reaction is quenched by adding 50 mL of 2 M glycine,
and agitated for 1 h. FTLDADF-Sepharose is collected on a
fritted funnel (coarse), washed with 500 mL of 50 mM Tris-
Cl, pH 7.6, and suspended and stored in 60 µM NaN3.
For the R1-peptide binding assay wet column material
(∼ 75 µL) is aliquotted into MicroSpin columns, followed by
equilibration with 2 × 75 µL of buffer A (50 mM Tris, pH 8.5,
10 mM DTT, 100 µM dTTP). Added DTT keeps R1 in reduced
form, preventing aggregation due to intermolecular disulfide
formation. Added dTTP (100 µM) ensures all R1 is maintained
in the dimeric state. The assay works best when the elution
of mR1 is minimal in the absence of added inhibitor and
maximal in the presence of a potent inhibitor, as tested with
100 µM P7. In maintaining this balance, it is usually desirable
to mix FTLDADF-Sepharose with underivatized Sepharose 4B-
Cl. A typical mixing ratio is 60:40: i.e., a column would have
45 µL of FTLDADF-Sepharose prepared as above and 30 µL
of Sepharose 4B-Cl. In addition, pH 8.5 is superior to pH 7.5.
Complete elution of R1 corresponded to 70 ( 10% of that
applied.
After each addition of buffer (or later of wash or sample),
columns are spun for 1 min at 735g in a Heraeus pico Biofuge
(VWR). Premixed samples of mR1 (4.4 µM) and inhibitor
(typically 100 µM) in buffer A (50 µL total) are loaded onto
individual columns and allowed to sit for 30 min. The columns
are then centrifuged, and the eluate is collected in an eppen-
dorf tube followed by a 50 µL wash with buffer A, collected in
the same tube. The column material is regenerated by washing
with 2 × 150 µL of 6 M guanidinium chloride, followed by 2 ×
150 µL wash and subsequent storage in 60 µM NaN3. Eluate
from each sample is analyzed for R1 content by Bradford
assay.44 All samples are run in duplicate. Table 4 shows sample
data from our column assay, demonstrating an increase in
protein elution upon introduction of higher concentrations of
inhibitor.
V. Exp er im en ta l Section
Ma ter ia ls. AcFT(tBu)LD(OtBu)AD(OtBu) was purchased
from SynPep corporation (Dublin, CA). FTLDADF either was
purchased from Research Genetics (Huntsville, AL), or was
prepared by Dr. Alison Fisher, as described.20 AcFTLDADF
was also prepared by Dr. Alison Fisher, as described.20 An
ASI431A peptide synthesizer (Applied Biosystem Inc.) was
used to synthesize peptide precursors, Fmoc-FT(tBu)LD-
(OtBu)AD(OtBu)F-Wang resin for library II and Fmoc-T(tBu)-
LD(OtBu)AD(OtBu)F-Wang resin for library III, following the
operation procedure in the manual. Boc-2-fluoro-L-Phe, Boc-
2-chloro-L-Phe, Boc-2-pyridyl-L-alanine, and Boc-3-pyridyl-L-
alanine were purchased from American Peptide Company
(Sunnyvale, CA). H-D-Phe-OtBu‚HCl, H-His(Trt)-OtBu‚HCl,
Ca lcu la tin g Kd s Usin g th e R1-P ep tid e Bin d in g Assa y.
The amount of eluted mR1 in the presence of a ligand
competing with Sepharose-FTLDADF for mR1 binding pro-
vides a measure of the affinity of mR1 for the free ligand. To
convert this information into a quantitative estimate of ligand

44 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 1
Pender et al.
Ta ble 4. Sample Data from the New R1-Peptide Binding
Assay
The precision with which relative Ks can be determined from
eq 10 is limited by the term {[mR1]e,a - [mR1]e,0} as [mR1]e,a
approaches [mR1]e,0 for very weak inhibitors. Thus, values
above 30 have large average deviations and are considered
unreliable.
R1 eluted (µg)
[inhibitor] (µM)
AcFTLDADF
Fmoc Phe
In Vit r o R ib on u cleot id e R ed u ct a se Act ivit y Assa y.
Ribonucleotide reductase activity was assayed at 37 °C by
monitoring the conversion of [3H]-GDP to [3H]-dGDP es-
sentially as previously described.19,20 In a volume of 70 µL, each
assay contained 60 mM HEPES, pH 7.6, 10 mM MgCl2, 8.75
mM NaF, 25 mM DTT, 300 µM dTTP, 50 µM FeCl3, 100 µM
[8-3H]-GDP, 2 µg of R1, 4 µg of R2, and varying amounts of
peptide inhibitors. Reactions were initiated with the radiola-
beled substrate and quenched by immersion in a boiling water
bath for 5 min. Tritiated samples were frozen and lyophilized
to dryness to reduce backgrounds. Lyophilized samples were
reconstituted in 1 mL of 50 mM Tris-HCl, pH 8.5, with 100
mM Mg(OAc)2 (buffer B) and centrifuged at 6000g for 10 min
in a microfuge to remove denatured protein. Reconstituted
samples were loaded onto phenylboronate-agarose columns (2
mL bed volume), preequilibrated with 10 mL of buffer B.
Deoxyribonucleoside diphosphates elute in 8 mL of buffer B.
Unreacted ribonucleoside diphosphate substrates were recov-
ered and columns were regenerated by treatment with 10 mL
of 50 mM sodium citrate, pH 5.9. Recovery typically ranged
between 75% and 100% of label. Radioactivity in aliquots of
both buffer B and citrate fractions was measured by liquid
scintillation counting.
Peptide stocks were made by adding base (either 0.1 M
NaOH or saturated NaHCO3) to a suspension of lyophilized
peptide in water until full solution was obtained (final pH 8.0-
8.5). The concentrations of the peptide stocks were determined
from analytical HPLC injections by comparing the area under
the absorption peak (215 nm) with a standard sample of
AcFTLDADF. The validity of this approach was confirmed for
several peptides using extinction coefficients for either phen-
ylalanine (258 nm) or for the side chains of the phenylalanine
analogues.
10
30
100
300
1000
1.3
2.1
3.8
7.7
10.9
0.9
1.3
2.7
8.7
binding relative to that of the standard inhibitor, P7, each
assay is standardized by measuring mR1 elution both with
no inhibitor present and in the presence of P7. The following
equations are pertinent, where Ka is the dissociation constant
of ligand La, KP7 is the dissociation constant of P7, [Lc]eff is
the effective concentration of column-bound FTLDADF, Kc is
the dissociation constant for the binding of the column-bound
FTLDADF to mR1, [mR1]t is the total mR1 concentration in
the column prior to centrifugation, and [mR1]e is the measured
concentration of eluted mR1 protein following centrifugation.
Kc ) [mR1][Lc]eff/[mR1Lc]
Ka ) [mR1][La]/[mR1La]
(1)
(2)
(3)
(4)
[mR1]t ) [mR1] + [mR1Lc] + [mR1La]
[mR1]e ) [mR1] + [mR1La]
Combining eqs 1, 2, and 3 yields
[mR1]t ) [mR1](R + â)
where â ) 1 + ([La]/Ka)
and R ) [Lc]eff/Kc
(5)
C-Ter m in a l P r otection of N-Ter m in a lly P r otected
P h en yla la n in e An a logu es. Reactions were performed typi-
cally on a scale of 100 µmol, following a published procedure.45
Typically, 200 µL of 1 M tert-butyl-2,2,2-trichloroacetimidate
in cyclohexane was added to 1 mL of 0.1 M phenylalanine
analogue in DCM, and the reaction was either maintained at
room temperature for ∼2 days or refluxed for 3-5 h. Reactions
were stopped by evaporating solvent under reduced pressure
when no further consumption of starting material was noted
by TLC (silica, 0.5% MeOH in CHCl3), using ninhydrin
staining. Fully protected phenylalanine analogues were puri-
fied by silica gel flash chromatography. Each demonstrated
an M + H or M + Na peak in electrospray mass spectrometry
(Micromass platform mass spectrometer, Manchester, U.K.)
consistent with its molecular mass.
Combining eqs 2 and 4 yields
[mR1] ) [mR1]e/â
(6)
(7)
Combining eqs 5 and 6 and rearranging yields
R ) â {([mR1]t/[mR1]e) - 1}
Determination of [mR1]e in the presence and absence of an
inhibitor La obviates the need to evaluate R explicitly, since
R ) {1 + [La]/Ka}{[mR1]t/[mR1]e,a - 1} )
{[mR1]t/[mR1]e,0 - 1} (8)
N-Ter m in al Depr otection of P h en ylalan in e An alogu es.
Reactions were performed typically on a scale of 10-20 µmol.
FmocPhe derivatives in DCM (2 volumes, 0.1 M) were depro-
tected by addition of one volume of diethylamine. Boc-Phe
derivatives in ethyl acetate (3 volumes, 0.27 M) were depro-
tected as described,46 with slight modification, by reaction with
a 5-fold excess of 4 M HCl in dry dioxane (1 volume). In both
cases, reactions were monitored by TLC as above and were
complete in 1-2 h at room temperature. No purification of
these materials was performed prior to the coupling reaction
described below.
Syn th esis of Libr a r y 1: AcF TLDADX. Reactions were
performed, under argon, with either one phenylalanine ana-
logue or with two phenylalanine analogues simultaneously.
Typically, 4.5 mg (5 µmol, 1 equiv) of AcFT(tBu)LD(OtBu)-
AD(OtBu), 1.4 mg of HOBt (10 µmol, 2 equiv), and 0.7 µL of
triethylamine (TEA) (5 µmol, 1 equiv) were dissolved in a
minimum amount of DCM at 4 °C. EDC (1.5 mg, 7.5 µmol, 1.5
equiv) and additional equivalent(s) of TEA, if needed to
neutralize hydrochloride salts, were added to the C-terminally
protected amino acid (2 equiv total). Reaction was initiated
where [mR1]e,a is the eluted protein in the presence of an
inhibitor La, and [mR1]e,0 is the eluted protein with no inhibitor
present. Solving for Ka:
Ka ) [La][mR1]e,0{[mR1]t - [mR1]e,a}/[mR1]t{[mR1]e,a
-
[mR1]e,0} (9)
In the work presented in this paper, free [La] is well ap-
proximated by total [La].
Although Kd values calculated directly from eq 9 are subject
to considerable variation [e.g., the value found for P7 was 9 (
5 µM (12 determinations)], the dissociation constant for any
inhibitor La relative to the P7 standard, calculated from eq
10, is quite reproducible. Such relative values are reported in
Table 2.
Ka/KP7 ) [La]{[mR1]t - [mR1]e,a}{[mR1]e,P7 - [mR1]e,0}/
[LP7]{[mR1]e,a - [mR1]e,0}{[mR1]t - [mR1]e,P7} (10)
where [mR1]e,P7 is the protein eluted in the presence of P7.

Design of Peptide Inhibitors of Ribonucleotide Reductase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 1 45
by addition of the peptide solution to the solution of EDC and
amino acid(s). Occasionally, a small volume of DMF was added
to increase amino acid solubility. The reaction was started on
ice and allowed to warm to room temperature overnight,
checked for completion by TLC on silica (9 mL:1 mL:1 drop
DCM/MeOH/AcOH), and evaporated to an oil under reduced
pressure. The crude oil was dissolved in 30 mL of ethyl acetate,
washed two times with 10 mL of 1 M HCl, once with 10 mL of
saturated NaHCO3, and once with 15 mL of brine. The organic
layer was dried over MgSO4 and evaporated to dryness under
reduced pressure. The protected 7-mer was deprotected in 95%
TFA using anisole (50-fold molar excess) as a scavenger. For
AcFTLDADW, deprotection was carried out in 50% TFA, 3.1%
ethanedithiol, and 2% water in DCM. The peptide was allowed
to sit for 1-2 h at room temperature, and the TFA was
evaporated under reduced pressure. The crude peptide then
was dissolved in 1:1 acetonitrile/water with 0.1% TFA and
lyophilized overnight to yield a dry powder.
Syn th esis of Libr a r y II. AcXTLDADF . Fmoc-T(tBu)LD-
(OtBU)AD(OtBu)F-Wang resin (22 mg, 0.01 mmol of peptide)
was treated with 20% piperidine in NMP (10 mL/1 g resin)
for 30 min at room temperature to remove the Fmoc-group.
The resin was removed by filtration, and washed with NMP
and DCM (3 mL each time, 3 times each solvent), before being
reacted for 2 h at room temperature with a solution of Fmoc-
amino acid (0.10 mmol) in 5 mL NMP containing HBTU (15.2
mg, 0.04 mmol), HOBT (6.1 mg, 0.04 mmol), and DIEA (10
µL, 0.05 mmol). Completion of the reaction was verified by
Kaiser test.47 N-Terminal acetylation with acetic anhydride
was carried out as described below. Peptides were deprotected
and cleaved from the resin by treatment with 95% TFA (10
mL/1 g resin) for 90 min at room temperature20 and precipi-
tated and washed exhaustively with anhydrous diethyl ether.
tions of FmocPhe and Fmoc Leu were created using the GB/
SA continuum model in MacroModel (Schro¨dinger Inc., J ersey
City, NJ ). Structural models of mR1 complexes were created
using QUANTA 97 (Molecular Simulations, Inc., Burlington,
MA) and the crystal structure of E. coli R1 (RCSB Protein Data
Bank code 1RLR).26,49 This structure includes a bound decamer-
ic peptide (DIDDLSNFQL) corresponding to the C-terminal
residues of the E. coli R2 protein. We used this decameric
peptide, residues in eR1 within 7 Å of the decameric peptide,
and the sequence alignment of mR1 and eR1 from CLUSTALW
(Figure 1) to build a model of mR1. Residue side chains in eR1
were manually changed to the corresponding side chains in
mR1 at positions in which their sequences differed, and the
decameric peptide was reduced in length to the acetylated
heptameric peptide. Distance constraints from the trNOE
experiments for P7 and cycP7, numbering 74 and 69, respec-
tively, were imposed on the peptides, and positional constraints
were imposed on all backbone atoms of the mR1 model.
Exhaustive energy minimization was then performed using
the adapted basis Newton-Raphson algorithm, a radial
dielectric function, and the Charmm force field implemented
within QUANTA 97. The model for bound P7 (Figure 2a)
exhibits 18 lower bound violations vis-a`-vis the trNOE con-
straints, which averaged 0.2 Å each (0.5 Å being the largest
deviation), and two upper bound violations of 0.1 and 0.9 Å.
The corresponding values for the model of bound cycP7 (Figure
2b) are 12 lower bound violations which averaged 0.2 Å each
(0.8 Å being the largest deviation), and 4 upper bound
violations totaling only 0.05 Å. An especially important
constraint for the models of both bound P7 and bound cycP7,
was that between the side chains of F1 and L3. Two other
constraints important for the P7 model were those between
the side chains of A5 and F7, as well as the T2-A5 constraint,
which helps define the turn.
Syn th esis of Libr a r y III. XF TLDADF . Fmoc-FT(tBu)LD-
(OtBu)AD(OtBu)F-Wang resin was deblocked with 20% pip-
eridine in NMP as described above. Three methods of N-ter-
minal acylation were employed following deblocking, all using
0.01 mmol of peptide bound to resin. Method 1: Resin was
reacted for 1 h at room temperature with 2 mmol of anhydride
(acetic, propionic, butyric, or benzoic) in 4 mL of NMP
containing DIEA (0.09 mL, 0.5 mmol) and HOBT (9.2 mg, 0.06
mmol). Method 2: Resin was reacted for 2 h at room temper-
ature with 0.10 mmol of carboxylic acid (phenylacetic, p-
methoxylphenylacetic, or p-nitrophenylacetic) in 5 mL of NMP
containing HBTU (15.2 mg, 0.04 mmol), HOBT (6.1 mg, 0.04
mmol), DIEA (10 µL, 0.05 mmol). Method 3: HCO-FTLDADF
was synthesized using the mixed anhydride procedure.48 At 0
°C, 0.5 mL of acetic anhydride was added dropwise to 1 mL of
formic acid, and the mixture was stirred for 10 min. The
reaction mixture was then stirred at room temperature for 1
h, diluted by addition of 3 mL of DCM, and added into the
reaction vessel containing peptide. The reaction vessel was
shaken for 2 h at room temperature. Completion of all capping
reactions was verified by Kaiser test. Peptides were depro-
tected and cleaved from the resin by treatment with 95% TFA
(10 mL/1 g resin) for 90 min at room temperature, and
precipitated and washed exhaustively with anhydrous diethyl
ether.
Abbr evia tion s. Abbreviations for common amino acids are
in accordance with the recommendations of IUPAC. Additional
abbreviations: Boc, tert-butoxycarbonyl; cycP7, AcFcyc[ELDK]-
DF; DCM, dichloromethane; DIEA, N,N-diisopropylethy-
lamine; DMF, N,N-dimethyl fomamide; EDC, 1-ethyl-3,3-
dimethylaminopropylcarbodiimide; DTT, dithiothreitol; eRR,
E. coli RR; Fmoc, fluorenylmethoxycarbonyl; HBTU, 2-(1H-
benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophos-
phate; HEPES, N-2-hydroxyethylpiperazine-N′-2-ethanesulfon-
ic acid; HOBT, N-hydroxybenzotriazole; mRR, mouse RR;
NMP, N-methylpyrrolidone; P7, AcFTLDADF; RR, ribonucle-
otide reductase; TEA, triethylamine; TFA, trifluoroacetic acid;
Tris-Cl, tris(hydroxymethyl)aminomethane hydrochloride; tr-
NOE, transfer nuclear Overhauser effect.
Ack n ow led gm en t. This work was supported by
NIH Grant CA58567 to B.S.C. B.A.P. was also sup-
ported by NIH Training Grant T32 GM 07229. We
thank Dr. Alison Fisher for providing P7, Dr. Shri Nair
for providing the inhibition results with peptide 33, Dr.
J ian Liu for modeling the conformations of FmocPhe
and FmocLeu in solution, and Professor Dale Mierke
for helpful discussions.
P u r ifica tion a n d Ch a r a cter iza tion of P ep tid es. All
peptides were purified by HPLC (Synchropak C18 column,
Micra Scientific Inc., or Microsorb-MV C-18, (Keystone Scien-
tific, Bellefonte, PA) with a Perkin-Elmer Series 4 Chromato-
graph HPLC) using a linear gradient of acetonitrile in 0.1%
TFA. Collected fractions were lyophilized and reinjected to
assess purity and determine peptide stock concentration. Each
peptide had either or both M + H and M + Na peaks in its
mass spectrum consistent with its molecular mass, and each
was >95% pure, except in two cases. AcFTLDAD(oOH-Phe)
14 was only 85% pure. AcFTLDAD(oOMe-Phe 18)/AcFTLDAD-
(pNO2-Phe) 19 were prepared in the same reaction and were
co-purified and assayed for mR1 binding as a mixture having
a 53:47 composition, as estimated by mass spectrometric
analysis. Both samples had very low affinity for mR1.
Refer en ces
(1) J ordan, A.; Reichard, P. Ribonucleotide reductases. Annu. Rev.
Biochem. 1998, 67, 71-98.
(2) Stubbe, J .; van der Donk, W. A. Ribonucleotide reductases:
radical enzymes with suicidal tendencies. Chem. Biol. 1995, 2,
793-801.
(3) Szekeres, T.; Fritzer-Szekeres, M.; Elford, H. L. The enzyme
ribonucleotide reductase: target for antitumor and anti-HIV
therapy. Crit. Rev. Clin. Lab. Sci. 1997, 34, 503-28.
(4) Robins, M. J . Mechanism-based inhibition of ribonucleotide
reductases: new mechanistic consideration and promising bio-
logical applications. Nucleosides Nucleotides 1999, 18, 779-93.
(5) Hui, Y. F.; Reitz, J . Gemcitabine: a cytidine analogue active
against solid tumors. Am. J . Health Syst. Pharm. 1997, 54, 162-
70.
(6) Stevens, M. R. Hydroxyurea: an overview. J . Biol. Regul.
Homeost. Agents 1999, 13 (3), 172-5.
Molecu la r Mod elin g. Models for the solution conforma-

46 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 1
Pender et al.
(7) Romanelli, F.; Pomeroy, C.; Smith, K. M. Hydroxyurea to inhibit
human immunodeficiency virus-1 replication. Pharmacotherapy
1999, 19 (2), 196-204.
(8) Hainsworth, J . D.; Burris, H. A., III; Litchy, S.; Erland, J . B.;
Hon, J . K.; Brierre, J . E.; Greco, F. A. Gemcitabine and
vinorelbine in the second-line treatment of nonsmall cell lung
carcinoma patients: a minnie pearl cancer research network
phase II trial. Cancer 2000, 88 (6), 1353-8.
(9) Berlin, J . D.; Adak, S.; Vaughn, D. J .; Flinker, D.; Blaszkowsky,
L.; Harris, J . E.; Benson, A. B., III A phase II study of
gemcitabine and 5-fluorouracil in metastatic pancreatic cancer:
an Eastern Cooperative Oncology Group Study (E3296). Oncol-
ogy 2000, 58 (3), 215-8.
(10) Lorusso, V.; Manzione, L.; De Vita, F.; Antimi, M.; Selvaggi, F.
P.; De Lena, M. Gemcitabine plus cisplatin for advanced
transitional cell carcinoma of the urinary tract: a phase II
multicenter trial. J . Urol. 2000, 164 (1), 53-6.
(11) Matano, E.; Tagliaferri, P.; Libroia, A.; Damiano, V.; Fabbrocini,
A.; De Lorenzo, S.; Bianco, A. R. Gemcitabine combined with
continuous infusion 5-fluorouracil in advanced and symptomatic
pancreatic cancer: a clinical benefit-oriented phase II study. Br.
J . Cancer 2000, 82 (11) 1772-5.
(12) Nagourney, R. A.; Link, J . S.; Blitzer, J . B.; Forsthoff, C.; Evans,
S. S. Gemcitabine plus cisplatin repeating doublet therapy in
previously treated, relapsed breast cancer patients. J . Clin.
Oncol. 2000, 18 (11), 2245-9.
(13) Adjei, A. A.; Erlichman, C.; Sloan, J . A.; Reid, J . M.; Pitot, H.
C.; Goldberg, R. M.; Peethambaram, P.; Atherton, P.; Hanson,
L. J .; Alberts, S. R.; J ett, J . Phase I and pharmacologic study of
sequences of gemcitabine and the multitargeted antifolate agent
in patients with advanced solid tumors. J . Clin. Oncol. 2000,
18 (8), 1748-57.
(14) Linskens, R. K.; Golding, R. P.; van Groeningen, C. J .; Giaccone,
G. Severe acute lung injury induced by gemcitabine. Neth. J .
Med. 2000, 56 (6), 232-5.
(15) Santini, D.; Tonini, G.; Abbate, A.; Di Cosimo, S.; Gravante, G.;
Vincenzi, B.; Campisi, C.; Patti, G.; Di Sciascio, G. Gemcitabine-
induced atrial fibrillation: a hitherto unreported manifestation
of drug toxicity. Ann. Oncol. 2000, 11 (4), 479-81.
(16) Banach, M. J .; Williams, G. A. Purtscher retinopathy and
necrotizing vasculitis with gemcitabine therapy. Arch. Ophthal-
mol. 2000, 118 (5), 726-7.
(17) Mansson, E.; Spasokoukotskaja, T.; Sallstrom, J .; Eriksson, S.;
Albertioni, F. Molecular and biochemical mechanisms of flu-
darabine and cladribine resistance in a human promyelocytic
cell line. Cancer Res. 1999, 59, 5956-63.
(18) Duan, J .; Liuzzi, M.; Paris, W.; Lambert, M.; Lawetz, C.; Moss,
N.; J aramillo, J .; Gauthier, J .; Deziel, R.; Cordingley, M. G.
Antiviral activity of a selective ribonucleotide reductase inhibitor
against acyclovir-resistant herpes simplex virus type 1 in vivo.
Antimicrob Agents Chemother. 1998, 42, 1629-35.
(29) Aggarwal, R. Alignment of the mouse, HSV, and malarial R1
subunit sequences of ribonucleotide reductase to the E. coli R1
crystal structure using sequence- and structure-based alignment
techniques. University of Pennsylvania, Philadelphia, 1997.
(30) Sander, C.; Schneider, R. Database of homology-derived struc-
tures and the structural meaning of sequence alignment.
Proteins 1991, 9, 56-68.
(31) Davis, R.; Thelander, M.; Mann, G. J .; Behraven, G.; Soucy, F.;
Beaulieu, P.; Lavalle´e, P.; Graslund, A.; Thelander, L. Purifica-
tion, characterization, and localization of subunit interaction
area of recombinant mouse ribonucleotide reductase. J . Biol.
Chem. 1994, 269, 23171-6.
(32) Bonneau, A. M.; Kibler, P.; White, P.; Bosquet, C.; Danserau,
N.; Cordingley, M. G. Resistance of Herpes Simplex Virus Type
1 to peptidomimetic ribonucleotide reductase inhibitors: Selec-
tion and characterization of mutant isolates. J . Virol. 1996, 70,
787-93.
(33) Ekberg, M.; Sahlin, M.; Eriksson, M.; Sjo¨berg, B. M. Two
conserved tyrosine residues in protein R1 participate in an
intermolecular electron transfer in ribonucleotide reductase. J .
Biol. Chem. 1996, 34, 20655-9.
(34) Grantham, R. Amino acid difference formula to help explain
protein evolution. Science 1974, 185, 862-4.
(35) Hamann, C. S.; Lentainge, S.; Li, L.-S.; Salem, J . S.; Yang, F.-
D.; Cooperman, B. S. Chimeric small subunit inhibitors of
mammalian ribonucleotide reductase: a dual function for the
R2 C-terminus? Protein Eng. 1998, 11 (3), 219-24.
(36) Climent, I.; Sjo¨berg, B.-M.; Huang, C. Y. Carboxyl-terminal
peptides as probes for Escherichia coli ribonucleotide reductase
subunit interaction: kinetic analysis of inhibition studies.
Biochemistry 1991, 30, 5164-71.
(37) Mattoo, B. N. Spectrophotometric study of the hydrolysis of
coumarin and dissociation of cis-coumarinic acid. Trans. Faraday
Soc. 1957, 53, 760-6.
(38) Bowden, K.; Hanson, M. J .; Taylor, G. R. Reactions of carbonyl
compounds in basic solutions. Part 1. The alkaline ring fission
of coumarins. J . Chem. Soc. (B) 1968, 2, 174-7.
(39) Mann, G. J .; Graslund, A.; Ochiai, E.; Ingemarson, R.; Thelander,
L. Purification and characterization of recombinant mouse and
herpes simplex virus ribonucleotide reductase R2 subunit.
Biochemistry 1991, 30, 1939-47.
(40) Studier, F. W.; Rosenberg, A. H.; Dunn, J . J .; Dubendorff, J . W.
Use of T7 RNA Polymerase to direct expression of cloned genes.
Methods Enzymol. 1990, 185, 60-89.
(41) Hamann, C. S. Purification, characterization and activity of
chimeric Escherichia coli, Mouse and Plasmodium falciparum
small subunits of type 1 ribonucleotide reductase. University of
Pennsylvania, Philadelphia, 1994, p 181.
(42) Caras, I. W.; Levinson, B. B.; Fabry, M.; Williams, S. R.; Martin,
D. W. J . Cloned mouse ribonucleotide reductase subunit M1
cDNA reveals amino acid sequence homology with Escherichia
coli and herpesvirus ribonucleotide reductases. J . Biol. Chem.
1985, 260, 7015-22.
(43) Salem, J . S.; Scott, C. P.; Li, L.-S.; Cooperman, B. S.; Rubin, H.
High level expression of the large subunit of mouse ribonucle-
otide reductase in a baculovirus system. FEBS Lett. 1993, 323,
93-5.
(19) Yang, F.-D.; Spanevello, R. A.; Celiker, I.; Hirschman, R.; Rubin,
H.; Cooperman, B. S. The carboxyl terminus heptapeptide of the
R2 subunit of mammalian ribonucleotide reductase inhibits
enzyme activity and can be used to purify the R1 subunit. FEBS
Lett. 1990, 272, 61-4.
(44) Bradford, M. M. A rapid and sensitive method for the quanti-
tation of microgram quantities of protein utilizing the principle
of protein-dye binding. Anal. Biochem. 1976, 72, 249-54.
(45) Thierry, J .; Yue, C.; Potier, P. 2-phenyl isopropyl and tert-Butyl
trichloroacetimidates: useful reagents for ester preparation of
N-Protected amino acids under neutral conditions. Tetrahedron
Lett. 1998, 39, 1557-60.
(46) Gibson, F. S.; Bergmeier, S. C.; Rapoport, H. Selective removal
of an N-BOC protecting group in the presence of a tert-butyl ester
and other acid-sensitive groups. J . Org. Chem. 1994, 59, 3216-
18.
(47) Kaiser, E.; Colescott, R. L.; Bossinger, C. D.; Cook, P. I. Color
test for detection of free terminal amino groups in the solid-phase
synthesis of peptides. Anal. Biochem. 1970, 34, 595-8.
(48) Zoete, V.; Bailly, F.; Catteau, J .-P.; Bernier, J .-L. Design,
synthesis and antioxidant properties of ovothiol-derived 4-mer-
captoimidazoles. J . Chem. Soc., Perkin Trans. 1997, 1, 2983-8.
(49) Hansch, C.; Leo, A. In Substituent constants for correlation
analysis in chemistry and biology; J ohn Wiley & Sons: New
York, 1979; p 48-52.
(50) Creighton, T. E. In Proteins: Structures and Molecular Proper-
ties, 2nd ed.; W. H. Freeman and Co.: New York, 1993; p 154.
(51) Berman, H. M.; Westbrook, J .; Feng, Z.; Gilliland, G.; Bhat, T.
N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E. The Protein Data
Bank. Nucleic Acids Res. 2000, 28, 235-42.
(20) Fisher, A.; Yang, F.-D.; Rubin, H.; Cooperman, B. S. R2 C-
terminal peptide inhibition of mammalian and yeast ribonucle-
otide reductase. J . Med. Chem. 1993, 36, 3859-62.
(21) Fisher, A.; Laub, P. B.; Cooperman, B. S. NMR structure of an
inhibitory R2 C-terminal peptide bound to mouse ribonucleotide
reductase R1 subunit. Nature Struct. Biol. 1995, 2, 951-5.
(22) Laub, P. B.; Fisher, A.; Cooperman, B. S. Small C-terminal
Inhibitory Peptide of Mouse Ribonucleotide Reductase As Bound
To The Large Subunit. Nmr, 26 Structures, Protein Databank
Id: 1AFT, MMDB Id: 5993 1997.
(23) Pellegrini, M.; Liehr, S.; Fisher, A. L.; Laub, P. B.; Cooperman,
B. S.; Mierke, D. F. Structure-Based Optimization of Peptide
Inhibitors of Mammalian Ribonucleotide Reductase. Biochem-
istry 2000, 39, 12210-5.
(24) Liehr, S.; Barbosa, J .; Smith, A. B., III; Cooperman, B. S.
Synthesis and biological activity of cyclic peptide inhibitors of
ribonucleotide reductase. Org. Lett. 1999, 1, 1201-4.
(25) Smith, A. B., III; Sasho, S.; Barwis, B. A.; Sprengeler, P.;
Barbosa, J .; Hirschmann, R.; Cooperman, B. S. Design and
synthesis of a tetrahydropyran-based inhibitor of mammalian
ribonucleotide reductase. Bioorg. Med. Chem. Lett. 1998, 8,
3133-6.
(26) Uhlin, U.; Eklund, H. Structure of ribonucleotide reductase
protein R1. Nature 1994, 370, 533-9.
(27) Filler, R., Kobayashi, Y., Yagupolskii, L. M., Eds. In Organof-
luorine Compounds in Medicinal Chemistry and Biomedical
Applications; Elsevier: New York, 1993; p 2.
(52) Nicholls, A.; Bharadwaj, R.; Honig, B. GRASP: graphical
representation and analysis of surface properties. Biophys. J .
1993, 64, 166-70.
(28) Guindon, Y.; Lavallee, P.; Sumanas, R.; Consentino, G. P.
Ribonucleotide reductase inhibitors. E.P. 0 383 190 A2.
J M000335R