Article

pubs.acs.org/jmc

Ligand Bioactive Conformation Plays a Critical Role in the Design of

Drugs That Target the Hepatitis C Virus NS3 Protease
Steven R. LaPlante,*,†,§ Herbert Nar,‡ Christopher T. Lemke,† Araz Jakalian,† Norman Aubry,†
and Stephen H. Kawai*,†,⊥
†
Department of Chemistry, Boehringer-Ingelheim (Canada) Ltd., Research and Development, Laval, Québec H7S 2G5, Canada
‡
Department of Medicinal Chemistry, Boehringer Ingelheim Pharma KG, 88397 Biberach an der Riss, Germany
*
S Supporting Information

ABSTRACT: A ligand-focused strategy employed NMR, X-

ray, modeling, and medicinal chemistry to expose the critical
role that bioactive conformation played in the design of a
variety of drugs that target the HCV protease. The bioactive
conformation (bound states) were determined for key
inhibitors identified along our drug discovery pathway from
the hit to clinical compounds. All adopt similar bioactive
conformations for the common core derived from the hit
peptide DDIVPC. A carefully designed SAR analysis, based on
the advanced inhibitor 1 in which the P1 to P3 side chains and
the N-terminal Boc were sequentially truncated, revealed a
correlation between affinity and the relative predominance of the bioactive conformation in the free state. Interestingly,
synergistic conformation effects on potency were also noted. Comparisons with clinical and recently marketed drugs from the
pharmaceutical industry showed that all have the same core and similar bioactive conformations. This suggested that the variety
of appendages discovered for these compounds also properly satisfy the bioactive conformation requirements and allowed for a
large variety of HCV protease drug candidates to be designed.

■ INTRODUCTION
The binding of a ligand to a macromolecule may be viewed as
great interest to the pharmaceutical industry, clearly illustrates
this. Inspection of the scientific and patent literature for
involving the collision of two flexible objects whose shapes inhibitors of this enzyme reveals only peptidic species.
must be of sufficiently complementary to allow for the Furthermore, they are nearly all derived from the N-terminal
formation of an encounter complex. Subsequent mutual product sequence DDIVPC.
adaptations may then occur which stabilize the final ligand− One can view DDIVPC as incorporating a flexible ‘scaffold’
protein complex.1 The shape and conformational properties of consisting of a polypeptide backbone in which a particular
the ligand, in both the free and bound states, must be residue is cyclized (i.e., the P2 proline). Whether carried out
considered for a clear understanding of such biomolecular consciously or not, optimization in the varied directions
recognition processes.2 This said, deciphering the details of reported to date has involved concurrent improvement of the
such factors governing binding events are often viewed as individual subcontacts and maintenance of suitable conforma-
impractical in a drug discovery setting. Nonetheless, we believe tional properties. In this work, the conformation and dynamic
that a lucid comprehension of the conformational and dynamic properties of the key peptidyl inhibitor 1 of HCV protease were
aspects of binding events, on the small molecule side, can be systematically monitored by a number of NMR methods and
exploited to advance drug design and optimization efforts. through the use of a thorough truncation SAR (molecular
The binding of peptidic species to proteins such as protease dissection). The goal of this exercise was to acquire a clear
enzymes represents an especially problematic case. A protease understanding of the roles of various pharmacophores in
generally binds its substrates or other peptide ligands through a maintaining the bioactive conformation. This represents an
number of important points of contact throughout a large essential step to the design or further optimization of such
active site. The peptides themselves exhibit rather complex therapeutic agents as it can greatly aid in separating
conformational behavior that is highly dependent on the nature conformational factors and subcontact aspects of the global
of the side chains. Such targets are well-known to be very frugal binding event.
with respect to offering interesting non-peptidic compounds,
even after the screening of very large compound collections. Special Issue: HCV Therapies
Moreover, even modest modifications to the peptide backbone
are generally not well-tolerated. The NS3 protease of the Received: August 30, 2013
hepatitis C virus (HCV), an enzyme which has been a target of

© XXXX American Chemical Society A dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX
Journal of Medicinal Chemistry Article

Figure 1. Shown are the bound structures of DDIVPC (X-ray structure with inhibitor side chains labeled in yellow), BILN 127 (NMR and X-ray
based model), 1 (X-ray structure), Faldaprevir (X-ray structure), and BILN 2061 (X-ray based model), all superimposed on the enzyme derived from
the X-ray structure of the Faldaprevir complex. The ligand−enzyme complexes are displayed with atoms colored by atom -type except for the
common scaffold, which is colored red. The complexes were derived by X-ray crystallography for DDIVPC (PDB code 4JMY) and 1 (where the urea
N-terminus of 1′, PDB code 4K8B, was changed to a Boc), and the complex of BILN 127 was docked and consistent with NMR transferred NOESY
data and X-ray structures of related compounds. The complex involving Faldaprevir was derived by X-ray crystallography (PDB code 3P8O). The
complex involving BILN 2061 was based on the modification of an X-ray structure3f,6 of a related macrocyclic compound and energy minimization.
See also Materials and Methods and Supporting Information.

B dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

■
Article

RESULTS AND DISCUSSION (Faldaprevir) (“linear series”)8 and BILN 2061 (Ciluprevir)
(“macrocycle series”).7 Both molecules successfully demon-
Summary of SAR. Our drug discovery efforts with regard
strated a reduction in viral load in humans.9 As shown in Figure
to HCV protease drew on our history of studying peptidyl
1, these two molecules also maintain the same conformation of
inhibitors of another viral enzyme. We noted during our human
the peptide skeleton when protease-bound. The conformational
cytomegalovirus (HCMV) campaign that, although very poor
features of the linear series is discussed in detail below, whereas
ligands, N-terminal cleavage products were complexed in a well-
the conformational optimization and design of the macrocycle
defined manner by this serine protease.4a This prompted us to
series has been addressed elsewhere.2c,6,7a,b,9a
investigate the N-terminal cleavage products of HCV protease. Bound Conformation: Structure of the Inhibitor 1′:
Using NMR,3a we discovered that the hexapeptide Asp-Asp-Ile- HCV Protease Complex. The conformation of inhibitor 1′
Val-Pro-Cys (DDIVPC; Figure 1) also binds its target in a well- bound to the active site of the protease was unequivocally
defined way and were eventually found by us and others to be established by the X-ray structure of the complex (Figure 3). In
surprisingly potent inhibitors of the enzyme.3a,b,4b,5,7b Enzy- this structure, π-stacking between the P2 quinolines of two
matic assays subsequently found that the compound inhibited complexed inhibitor molecules due to the face-to-face packing
HCV protease with an IC50 of 71 μM (Ki = 19 μM).3a,b A of the two protomers in the asymmetric unit was observed (see
recent publication speculates as to the nature of this exceptional Supporting Information). It was therefore important to verify
product inhibition of a serine protease.3g whether the P2 groups, when the complex is in solution, have
As a lead molecule, DDIVPC was particularly attractive in the same conformation as in the crystal structure. NMR
view of the C-terminal carboxylate that imparts good solubility, transferred nuclear Overhauser effect (transferred NOESY)
as well as the dramatically improved specificity (notably with experiments were employed to evaluate the protease-bound
respect to elastase) since an activated ‘warhead’, commonly conformation of compound 1 in solution. The key interhy-
required for serine protease inhibition, is not present.2c The X- drogen distances from P2-Hγ to Q-H3 and from Q-H6 to the
ray structure of this key hexapeptide, complexed to HCV Boc were measured to be 1.8 and 2.9 Å, respectively. (The
protease, is reported here (Figure 1; top right). The binding of system used to designate atoms and amino acid positions10 in
the ligand occurs through the four C-terminal residues (IVPC) compounds 1 and 1′ are shown in Figure 2). In the crystal
with the N-terminal aspartates solvent exposed, corroborating
earlier NMR binding studies.3a,d The critical P1−P4 core binds
in an extended conformation, a general feature of complexes of
serine proteases and their peptidyl ligands.
A summary of our extensive optimization effort is shown in
Figure 1. A number of modifications,3b,c notably addition of the
aryloxy group to the P2 residue, provided BILN 127 and a 10-
fold improvement in binding.3a NMR and modeling suggested
that the introduced aryl moiety lies over part of the catalytic
triad of the enzyme (vide infra).3a It was also apparent that the
bound conformation is similar to that of DDIVPC (both
displayed in red). A battery of NMR methods, including 13C T1
and transferred 13C T1 experiments,2c,d,3a found that important
differences existed between the free and bound states of the this
ligand (also see Supporting Information). These investigations
formed the basis of a number of rational, structure-based
optimization strategies including macrocyclization.6,7b
N-Truncation coupled with further side chain and aryloxy-
optimizations3c,d afforded the Boc-tripeptide 1 (BILN 1508) Figure 2. Atom numbering used in this study. Inhibitor 1, X = O;
with a Ki of 0.020 μM.3 Shortly after, the X-ray structure of 1′, a inhibitor 1′, X = NH.
very close urea analogue of carbamate 1 in complex with the
protease, was solved. This complex is described in detail below structure and after hydrogen atoms were added, the
and shows that the peptide scaffold is also bound in a very corresponding key distances were similar and were determined
similar fashion as seen for DDIVPC and BILN 127. The large to be 2.0 and 3.2 Å. It was later found that similar distances
tricyclic P2-substituent indeed lies over part of the catalytic were observed for compound 1′ using transferred NOESY
triad of the protease. Compound 1 became a molecule of experiments. Taken together, the NMR studies confirm that the
central importance to our understanding of the SAR as crucial bound crystal and solution structures correlate very well and
structural information concerning its binding to the target that inhibitors 1 and 1′ are complexed, not surprisingly, in
could be gleaned from the crystallographic structure of the identical manners.
protease complex with inhibitor 1′. Furthermore, the NMR As shown in Figure 3, compound 1′ is bound to the active
spectra (1H and 13C) were very well-suited for a wide range of site of HCV protease through an anti-parallel β-sheet between
studies, notably ROESY and 13C NT1 analyses. It must be the inhibitor backbone (scaffold) and the E2-strand of the
clarified that SAR had revealed that the carbamate-capped protease (Figure 3C and E). There are apparent hydrogen
(Boc) tripeptides have better cell-based activity than their urea bonds between the P1-NH and the carbonyl of Arg155, as well
counterparts, which accounts for why the present studies (as as a pair of H-bonds between the P3 residue (carbonyl and
well as optimization efforts) involved the former series. NH) and Ala157 of the enzyme. The terminal urea-NH is also
Subsequent efforts to improve potency and biopharmaceut- within H-bonding distance of the latter. These interactions are
ical properties led to two series exemplified by BI 201335 normally observed for peptide ligands complexed to serine
C dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX
Journal of Medicinal Chemistry Article

Figure 3. (A) Full view of the X-ray structure of HCV protease (green, with the catalytic triad colored magenta), the NS4A peptide cofactor
(orange), and 1′ colored yellow and as atom-type (PDB code 4K8B). (B) Same as panel A but for apo HCV protease (PDB code 1BT7). (C) View
showing the Cγ-exo conformation of the P2 proline ring and the orientation of the quinoline substituent, which lies over the protease catalytic acid
side chains of Arg155 and Asp181. (D) View showing the shallow pocket of HCV protease with bound 1′ (green). (E) View of inhibitor 1′ (green)
bound to the active site of HCV protease (white carbon atoms) showing canonical hydrogen bonds between the inhibitor backbone and the E2
strand (P3 to Ala157 and P1(NH) to Arg155). The side chain of Phe154 lies at the bottom of the S1 pocket. (F) View showing the interactions
between the terminal carboxylate and the oxyanion hole, the latter composed of the backbone amides of residues 137−139. The hydrogen-bonding
network involving the protease catalytic triad His57, Asp81, and Ser139 is also shown.

protease active sites and are globally referred to as the involved in binding of the inhibitor, from which it is oriented
‘canonical’ binding mode. away.
The terminal carboxylate forms a network of hydrogen bonds The P1 vinylcyclopropyl group protrudes into the shallow S1
to the active site (Figure 3F). One of the oxygen atoms is pocket. The N-terminal portion of the peptide is conforma-
within H-bonding distance of the backbone NH groups of tionally extended, and with little S3 pocket to speak of, the P3
Gly137, Ser138, and Ser139, which constitutes the oxyanion tert-butyl side chain lies on the surface of the protease and is
hole. The second oxygen is within H-bonding distance of the ε- solvent-exposed. The P2-quinoline6 moiety is oriented
proton of the catalytic His57. This residue is presumed to exist perpendicular to the gross plane of the proline ring. The latter
in the protonated state and is within H-bonding distance to is puckered such that the substituted γ-carbon is below the
Asp81.3f The catalytic Ser side chain does not appear to be plane of the other pyrrolidine atoms (referred to as a Cγ-exo
D dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX
Journal of Medicinal Chemistry Article

conformation; see Figure 3C). Also shown in this view, the of the molecules studied herein have two well-defined sets of
large P2 residue lies over the Asp-His catalytic pair,3 as well as NMR resonances, arising from very slowly (on the NMR time
the side chain of Arg155 of the E2 strand involved in canonical scale) interconverting peptide conformers of the tertiary amide
binding. These interactions appear largely hydrophobic, linkage between the P2 and P3 residues. However, the cis-
although some electrostatic interaction between the Arg155 conformer exists as a minor population compared to that of the
guanidinium and the aryl group methoxy substituent likely relevant trans-conformer which is present in 87%.
contributes to binding. Close analysis of the free-state data in Table 1 reveals that the
Free versus Bound Conformations of 1/1′. Having predominant conformation of 1 greatly resembles that of the
determined the bound conformation of inhibitor 1′ by X-ray bound state. There are short P1-NH/P2-αH distances that are
crystallography (and establishing that it is also that of indicative of an extended, trans-peptide orientation for the P1-
compound 1), comparisons could then be made with the P2 residues. The longer distance between the P3-NH and P3-
predominant free-state conformation of molecule 1 using NMR αH is consistent with the J-coupling data, indicating an
rotating-frame nuclear Overhauser effect (ROESY) experi- extended orientation. For the Ψ-torsion angle, a key correlation
ments11 (Table 1). It must be emphasized that although was that observed between the P3-tert-butyl group (P3-Hγ) and
the P1-HβC (3.3 Å), which also defines a primarily extended
Table 1. Apparent Interproton Distances (Å) Derived the conformation. A number of crosspeaks between the Q-H3 and
from ROESY of Free Inhibitor 1 and HCV Protease-Bound Q-H5 hydrogens and those of the P2 proline ring indicate that
Inhibitor 1′ from X-ray Crystallographic Dataa the predominant disposition of the quinoline group is such that
the B-ring is oriented away from the peptide backbone. Key
crosspeak free compound 1 bound compound 1′
correlations also include an intense crosspeak between Q-H3
P1-NH−P1-HβC 2.8 3.0 and P2Hγ as well as that between Q-H5 and P2Hα. The
P1-NH−P2-HβA 4.6 4.4 predominance of a Cγ-exo proline pucker (as observed in the
P1-NH−P2-Hα 2.3 2.2 bound state) is indicated by the strong P2HβA/P2HδA
P1-Hγ−P1-HβB 2.6 4.0 crosspeak (and corroborated by the lack of a peak between
P1-Hγ−P1-HβC 2.8 2.9 the corresponding B-ring hydrogens). Therefore, the ROESY
P1-HδA−P1-HβC 2.6 1.9 data for inhibitor 1 points to the f ree molecule existing
P1-HδA−P1-HδB 2.0 1.9 predominantly in a conformation closely resembling that when
P1-HδA−P1-HβB 3.5 4.6 bound to the protease as indicated by the X-ray crystallographic
P2-Hα−P2-HβB 2.3 2.4 structure of HCV protease−inhibitor 1′ complex described
P2-Hγ−P2-HβB 3.4 2.7
herein and the accompanying NMR work (Table 1).
P2-Hγ−P2-HβA 2.5 2.4
Systematic Truncation of Inhibitor 1. Knowing that the
P2-Hγ−P2-HδB 2.6 2.7
predominant free-state conformation of inhibitor 1 is very
P2-HδA−P2-HβA 3.2 2.5
similar to that when bound by the enzyme, we focused on
P3-NH−P3-Hα 3.5 3.0
establishing which structural features of the ligand contribute to
P3-NH−P2-HδB 4.9 4.5
stabilizing this bioactive conformation. A systematic ‘molecular
P3-Hγ−P1-HβC 3.3 3.9
dissection’ approach was undertaken in which sequential
Q-H6−Q-OMe 3.4 3.9
removal of portions of molecule 1 was performed and the
Q-H8−Q-OMe 2.1 2.2
resulting effects on binding affinity monitored. While a
Q-H3−P2-HβB 2.7 3.1
particular group in the parent structure can obviously
Q-H3−P2-HδB 3.8 4.5
contribute to the total binding energy through either the
Q-H3−P2-HδA 2.8 4.3
stabilization of the bioactive conformation or the formation of
Q-H3−P2-Hγ 2.0 2.0
favorable contacts with the active site, or both, we focused on
Q-H5−P2-Hα 3.8 4.0
the former, monitoring a number of parameters reflecting the
Q-H5−P2-Hβb > 4.9
conformational/dynamic behavior throughout the series of
Q-H3−Boc 5.5 7.0
compounds. Sequential truncations targeted the P3 side chain
P3-NH−Boc 4.1 4.5
and N-terminus (Table 2), the P2 aryloxy group (Table 3), and
P2-HδA−Boc 4.5 6.5
Q-H5−Boc 3.6 2.4
the P1 side chain (Table 4). The cropping steps were planned
a
out such that no charged or highly polar groups would appear
Distances for free compound 1 were derived from 200 ms ROESY in the new compounds so that we could assume that all
data (in DMSO-d6 solvent). Crosspeak volumes were scaled to the Q-
H5−Q-H6 correlation assigned a distance of 2.4 Å. Scaling also
truncated molecules would bind in the same fashion as parent
included multiplying the volumes of crosspeaks that involved methyl inhibitor 1. Nonetheless, a selection of truncated ligands were
and tert-butyl groups by factors of 0.66 and 0.5, respectively. “>” docked into the enzyme (see Supporting Information,
indicates no crosspeak observed. Distances for bound 1′ were derived Appendix 13).
from the X-ray crystal structure. Hydrogens were added to the crystal Free Conformations of P3 and N-Terminal Truncated
structure. Inhibitors. The free-state conformational properties of
molecules 1−3 were first investigated by an array of NMR
ROESY crosspeak volumes were converted to distances methods. All of the present NMR studies were carried out
between hydrogen atoms, these values should not be taken as under identical conditions in DMSO-d6, a solvent in which
literal values (interatomic distances), but rather as a convenient most compounds are soluble and for which conformational data
means of identifying predominant solution conformations. compare well with those obtained in water, allowing for a
Given a sufficient number of key ROESY crosspeaks, the convenient comparison of solution data. Perhaps the most
predominant conformation was identified since crosspeak areas expeditious measure12 of the flexibility of the Boc-P3 moiety,
reflect interproton distances.11 It should also be noted that all specifically the rotational freedom about the Φ angle, are the
E dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX
Journal of Medicinal Chemistry Article

Table 2. Binding Parameters and Conformational Indices of HCV Protease Inhibitors 1−5a

a
The conserved bonds/atoms of the scaffold derived from the lead peptide DDIVPC are colored red, and all other bonds/atoms are colored black.
Note that the inherent solubility imparted by the carboxylates made testing for activity feasible at high concentrations. Ki and ΔGbinding values were
calculated from IC50 data as described in Materials and Methods.
1
H NMR JNH‑αH coupling constants, which are provided in hydrogens remained quite constant within the series. Differ-
Table 2. The value of 8.3 Hz observed for compound 1 ences, however, were observed for the Boc-P3 region of the
indicates a predominantly extended conformation in solution, molecules. Nearly all of the Boc-tert-butyl crosspeaks observed
consistent with the X-ray derived bound conformation. A clear for inhibitor 1 were not observed in the case of the P3-Gly
trend in the J values was observed, the coupling constant analogue 3. An indication of the greater mobility of the Boc-P3
decreasing as bulk was removed from the side chain, indicating portion of inhibitors 2 and 3 are the changes in P3-NH and P3-
that other conformations are present in the P3-Ala analogue 2. αH ROESY distances, being shorter than those measured for 1.
The value of 5.6 Hz for molecule 3 is consistent with a freely This is due, in part, to freer Φ-bond rotation (corroborated by
rotating (on the NMR time scale) αC-NH bond. the J-coupling data) that allows the protons to come into
ROESY NMR data were also acquired for comparison of the proximity, whereas they were held apart in the predominantly
free-state properties of compounds 1−3 and are provided in extended conformation assumed by compound 1.
ROESY analysis was also carried out for the P3 tert-
Table S14 in the Supporting Information. As mentioned earlier,
butylacetyl inhibitor 4. While the derived distances reflect
the interproton distances must not be taken as literal values.
generally similar conformational properties for the P1 and P2
Changes in the intensities of ROESY crosspeaks for equivalent
residues (data not shown), two key points of comparison are
groups of related molecules may reflect changes in the notable with respect to the parent compound 1. The crosspeak
predominant conformation but can also arise from differences corresponding to the P3-tert-butyl (Hγ’s) and the P1-HβC was
in the dynamic ensemble average about a predominant not observed. Furthermore, a correlation (4.3 Å) between the
conformation. It must therefore be kept in mind that signals tert-butyl and the quinoline-H6 is present. The equivalent
arising from conformers with a relatively low population may be ROESY peak is absent in the spectrum of 1. Thus, the terminal
emphasized in cases where two hydrogens are in very close acyl group in 4 moves freely in contrast to 1, underscoring the
proximity, owing to the exponential relationship between role of the Boc-amino moiety in maintaining the latter in the
crosspeak intensity and interproton distance.11 bioactive extended conformation. This is clearly corroborated
Globally, the distances measured for inhibitors 1−3 were by NMR 13C spin-lattice relaxation times (13C T1) (vide infra
similar between protons within the P1 and P2 residues (Table and Tables S15−S17 in the Supporting Information).
S14 in the Supporting Information). However, there appear to The above analysis exemplifies how ROESY data may be
be differences in the predominant orientation of the P1 unit used not only to gain information concerning the preferred
with respect to the P2 residue (P1-NH/P2-HβA). The distances conformation of a molecule in solution but also to shed light on
between the quinoline protons and the P2 proline ring its dynamic properties relative to similar analogues. Another
F dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX
Journal of Medicinal Chemistry Article

Table 3. Binding Parameters and Conformational Indices of HCV Protease Inhibitors 1 and 6−10a

a
The conserved bonds/atoms of the scaffold derived from DDIVPC are colored red, and all other bonds/atoms are colored black. Ki and ΔGbinding
values were calculated from IC50 data as described in Materials and Methods.

method of obtaining insight into the flexibilities or mobilities of carbons in the latter. A clear indication of increased mobility in
molecules is through 13C T1 times by NMR.2a,c,3a,7b The the Boc-P3 portion with reduction of the size of the side chain
chemical shifts of the trans-conformers of inhibitors 1−5 and was provided by the Boc-tert-butyl T1 values, which sequentially
relevant NT1 values are provided in Table S16 (Supporting increase from 0.51 to 0.55 to 0.58 s in going from 1 to 3. The
Information). Care was taken to only use T1 values in cases higher mobility of the tert-butylacetyl group in 4 as compared
where the trans- and cis-signals were adequately resolved. to 1 was clearly evidenced by comparison of the NT1 values for
Resonance overlap was especially problematic in the case of both the P3-Cα and P3-Cγ(Me) signals, which corroborated
molecule 2. Comparison of the data obtained for inhibitors 1 the ROESY data.
and 3 showed a slight increase in the NT1 values, indicative of a Free Conformations of P2 and P1 Truncated
slightly greater degree of mobility for the P2 proline ring Inhibitors. The conformational properties of free inhibitors
G dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX
Journal of Medicinal Chemistry Article

Table 4. Binding Parameters and Conformational Indices of HCV Protease Inhibitors 1, 11 and 12.a

a
The conserved bonds/atoms of the scaffold derived from DDIVPC are colored red, and all other bonds/atoms are colored black. Ki and ΔGbinding
values were calculated from IC50 data as described in the Materials and Methods.

6, 8, and 10, for which structural elements of the P2-aryloxy NT1 data (Table S17, Supporting Information) also revealed
group have been sequentially removed, were also studied. A that, even for the P1-glycyl inhibitor 12, the mobility of the rest
striking difference between this series and that described above of the molecule is quite similar to that of the parent compound
is the invariance of the P3 JNH‑αH values (Table 3). In all cases, 1. While analysis of the present series does not allow for any
they remain within a range consistent with an extended Φ- conclusions as to changes in the mobility of the P1-residue
angle. In addition, the trans-conformers of the P2-P3 peptide itself, the great decreases in binding are no doubt due to both
bond predominated in all cases, which greatly facilitated the the losses of S1−P1 contacts as well as increases in the mobility
analysis and allowed for far more complete 13C T1 data sets to of the terminal carboxylate group.
be obtained (see Table S17, Supporting Information). Relationship between Inhibitor Conformation and
Truncation of the P2-aryloxy group clearly resulted in an Binding. Taken together, the J-coupling, ROESY, and 13C NT1
increase in the mobility of all regions of the molecules. For the data for the free inhibitors provides a mutually reinforcing
series of compounds 1, 6, 8, and 10, the T1 values arose fairly picture of the conformational and dynamic consequences of
evenly from 0.51 to 0.62 s for the Boc methyl carbons. The removing steric bulk from the P3 side chains in the series of
effect of the P2-aryl substituent on the motion of the proline inhibitors 1−3. Sequential reduction of the size of the side
ring is evident. While modest increases in NT1 values chain from a tert-butyl to methyl to hydrogen results in
accompanied truncation to the pyridine 8, indicating an stepwise losses of 1.6 and 1.3 kcal/mol. The large substituent in
increase in the flexibility of the pyrrolidine, the unsubstituted compound 1 constrains the Boc-P3 moiety to an extended
prolyl inhibitor 10 exhibited much greater mobility, notably at fashion such that the NH and CO groups are correctly
the γ-position. oriented to form the key hydrogen bonds to Ala157 of HCV
In terms of ROESY analysis, only key crosspeaks were protease, accounting for the reduced potencies of 2 and 3. This
analyzed in a semiquantitative manner. For molecules 6 and 8, effect, however, is contingent on the presence of the bulky Boc-
the key correlations defining the Cγ-exo proline ring pucker group as evidenced by the greater degree of mobility of
(P2-HβA−P2-HδA) and the proximity of the P3 and P1 side inhibitor 4, whose binding energy is 2.5 kcal/mol less than that
chains (P3-Hγ−P1-HβC) are present, being slightly less intense of parent compound 1. We believe that these conformational
compared to the equivalent signals for compound 1. Analysis of factors can be invoked to explain, in large part, the observed
the prolyl peptide 10 could not, however, be carried out due to changes in binding reflected in the Ki and ΔGbinding values for
extensive signal overlap, notably for the P1 and P2 residues. the present inhibitors.
Examination of the P2Hα signal (1D spectrum) revealed a very We have described a formal method for presenting SAR
different coupling pattern compared to all of the other aimed at assessing the contributions of combinations of groups
molecules, indicating that the lack of a γ-substituent results in to biomolecular recognition events14 and have applied it to the
a much different conformational behavior. P3 and N-terminal truncations described above. This “chemical
The influence of the P1 residue on the global conformational double-mutant” analysis clearly showed that the binding
properties of the inhibitors was also studied (Table 4). As was contributions of the P3-tert-butyl and N-Boc moieties, when
observed for the truncation of the P2-aryl substituent, added to the ‘core’ consisting of inhibitor 5, are synergistic and
sequential deletion of elements of the P1 vinylcyclopropyl believe that much of this synergy stems, ultimately, from
side chain13 had very little influence on either the cis/trans conformational factors. Both the tert-butyl and Boc-amino
conformational equilibrium of the P2-P3 amide or the P3 Φ- groups in compound 1 play roles in holding the P3 H-bonding
angle as evidenced by the large coupling constants. The 13C elements in the correct orientation with respect to each other,
H dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX
Journal of Medicinal Chemistry Article

Figure 4. HCV NS3 protease inhibitors that are recently approved drugs for treating HCV infection, and other inhibitors currently in phase II and III
clinical trials. The conserved bonds/atoms derived from DDIVPC or 1 are colored red, whereas new bonds/atoms are colored blue. The bottom
view is a superposition of the complexes involving the above compounds that were either derived from X-ray crystallography or from the docking
exercise described in Materials and Methods. All complexes are superimposed onto the X-ray structure of Faldaprevir (see Materials and Methods)
where the Faldaprevir atoms are all colored blue.

as well as the rest of the molecule. Global stabilization of the number of groups who have sought to mimic the substructure
active site by the P3 to S3 backbone contacts likely strengthen with rigid scaffolds.15
other key remote interactions and vice versa. While favorable The impact of γ-substitution of the P2-prolyl residue on the
interaction between the P3-tert-butyl and the enzyme must conformational behavior of the present molecules is clearly
contribute to binding, the X-ray structure shows that the evidenced by the NMR data. The effect of rigidifying the
hydrophobic contact area is small and that much of the group is pyrrolidine ring to a Cγ-exo pucker was consistent with studies
exposed to solvent. NMR studies also argue against the of simple proline derivatives, which pointed to stereoelectronic
interaction being very strong.2a,d,3a The nonlinear SAR factors being important in electronegative substituents
described above serves to underline the shortcomings of conferring this preferred ring conformation. Expanding the
assessing binding contributions of particular groups outside the aromatic system (B-ring) restricts O-aryl bond rotation to
context of the ligand as a whole. The importance of the conformations similar to that observed in the X-ray structure
extended P3 conformation described above has been noted by a and further rigidifies the molecule as a whole. Understanding
I dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX
Journal of Medicinal Chemistry

■
Article

the conformational effects of γ-proline substitution also clarifies CONCLUSIONS

the role of the P3 tert-butyl group of inhibitor 1. The invariance The present work underlines how optimization of the peptidic
of the P3 JNH‑αH values points to the bulky side chain being the N-terminal product sequence DDIVPC has led to a number of
predominant factor in restricting the P3 Φ-angle to an extended marketed and advanced drug candidates targeting HCV
conformation, as well as rendering the trans-P2-P3 rotamer protease, including Faldaprevir and BILN 2061 (by way of
much more stable than its cis-counterpart. The P2-aryloxy inhibitor 1, BILN 1508). Along the path to the latter,
substituent clearly stabilizes the conformation of the Boc-P3 improvements to binding involved both stabilization of the
moiety as evidenced by the incremental increases in 13C NT1 bioactive conformation, as established by the current results, as
values through the series 1, 6, 8, and 10. This was likely due, for well as the formation of better interactions between ligand and
the most part, to the restriction of the Ψ-angle. The similar target, at times simultaneously. Although one cannot separate
changes in the mobility of both the Boc and P3 tert-butyl these two global factors in an unambiguous manner, we have
carbons are consistent with this view. Although seemingly demonstrated through a carefully designed truncation SAR
remote from each other, the P2-aryloxy and Boc groups (exemplified by the dissection of compound 1) coupled to a
(provided a tert-butyl was present at P3) are in proximity to series of systematic NMR studies that rigidification of the
each other and may restrict each other’s motion. peptide scaffold to the bioactive conformation is of central
importance. Examination of an array of advanced molecules
Unlike the role of the P2-aryl substituent, the vinyl-
targeting HCV protease and which appear to all (present
cyclopropyl moiety of the P1 residue does not appear to
molecules included) bind in a single, canonical mode when one
have a great influence on the dynamic behavior of the examines the conformation of the peptide core through
remainder of the molecule. The fact that complete removal of molecular modeling indicates that stabilization of the bound
the side chain results in only small increases in mobility (NT1 conformation is globally relevant.
data) is consistent with the notion that a bulky P3 side chain The multidisciplinary approach described herein, which
and large proline substituent are the major determinants in combines medicinal chemistry, NMR spectroscopy, molecular
global rigidification of the inhibitors. Thus, the losses in modeling, and X-ray crystallography, clearly demonstrates that
potency observed within the sequence 1 to 11 to 12 are the systematic conformational analysis and investigation of dynamic
result of local increases in flexibility at P1 and the loss of behavior of a carefully chosen series of molecules can certainly
interactions within the S1 pocket of the active site.2c,7b The provide important insights into the nature of small molecule−
alteration of the bond angles about the P1 α-carbon upon protein interactions that can be exploited for drug discovery
introduction of the cyclopropyl ring has been discussed purposes.
elsewhere.3g
Bioactive Conformations of Clinical Drugs. A patent
and literature search of HCV protease inhibitors/drugs showed
■ MATERIALS AND METHODS
Inhibitors. The purity of all inhibitors was determined to be >95%
that most, if not all, of the advanced compounds targeting this by HPLC and high-field NMR. Synthetic procedures and additional
enzyme incorporate the identical peptide core structure derived physical data are provided either in the Supporting Information or
have been described elsewhere.3b,c,8
from DDIVPC (see Appendix 12 of the Supporting Enzyme Assay. The radiometric enzymatic assay13 was performed
Information). Recently approved drugs and other inhibitors in 50 mM Tris-HCl, pH 8.0, 0.25 M sodium citrate, 0.01% n-dodecyl-
currently in phase II and III clinical trials were docked into the β-D-maltoside, 1 mM TCEP at 23 °C. Twenty-five micromolar
enzyme active site and are superimposed at the bottom of concentration of the substrate DDIVPCSMSYTW, ∼1 nM biotin-
Figure 4. Despite the fact that these compounds vary DDIVPCSMSY[125I]TW, and various concentrations of inhibitors
(DMSO stock solutions) were incubated with 6 nM HCV NS3-NS4A
substantially with regard to their C-termini and side chain
(genotype 1b). Under these assay conditions, a KM of 9.1 μM was
appendages (blue-colored atoms at the top in Figure 4), they all found for the substrate DDIVPCSMSYTW. At least two IC 50
have the identical peptide core structure derived from DDIVPC determinations were carried out for each of two weightings for all of
(red-colored atoms in Figure 4). The remainder of the peptidic the inhibitors. Note that the inherent solubility imparted by the
structures all consist of a modified P2 proline and a P3 residue carboxylates for this series of compounds made testing for activity
with either a bulky tert-butyl side chain or one that is tethered feasible at high concentrations.
Ki values were calculated using eq 1:
to P1. The design and benefits of macrocyclization from P1 to
P3 have been addressed elsewhere2c,6,7a,b,9a and resulted in the IC50 = 0.5[Etotal ] + K i(1 + [S]/KM) (1)
breakthrough compound BILN 2061 (Figure 1), the first proof-
The free energies of binding were calculated using eq 2:
of-concept for a new class of HCV antiviral. This inspired
others to employ macrocyclization as a strategy for conforma- ΔG binding = − RT ln(1/K i) (2)
tional restriction (see Figure 4 and Appendix 12 of the
Supporting Information).16 Figure 4 shows how the P1 to P3 X-ray Crystallography. The X-ray structures of compounds
DDIVPC (PDB code 4JMY) and 1′ (PDB code 4K8B) were
and the P2 to P4 macrocycles can bind to the HCV protease determined as described in the Supporting Information.
active site. It is also noteworthy that many compounds with NMR Studies. NMR experiments were performed on a Bruker
proline modifications consist of the attachment of a γ-aryloxy or Avance spectrometer (400 or 600 MHz 1H frequencies) at 27 °C.
similar structure that would be expected to lock the pyrrolidine Samples were prepared by dissolving inhibitors 1−12 in DMSO-d6
in the appropriate pucker (see Figure 4 and Appendix 12 of the (13−16 mM) followed by deoxygenation of the solutions through
several rounds of exposure to vacuum and purges with nitrogen gas.
Supporting Information). The fusion of a second ring, as in the
The glass NMR tubes were subsequently sealed using a flame.
cases of Boceprevir and Telaprevir, is another rigidification Assignment of the 1H resonances were secured by careful analysis of
strategy that accomplishes the same conformational locking of standard homonuclear COSY and ROESY11 experiments. Homo-
the five-membered ring. nuclear ROESY data were acquired with a continuous spin-lock during

J dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry Article

mixing (300 ms), a 2.5 kHz spin-lock field, the carrier set at 4.7 ppm, Awith MOE’s LigX module with receptor, ligand, and solvent atoms
256−400 t1 data points from the addition of 32−64 transients, a 5200 restrained. Energy cutoff range was set between 10 and 12 Å and the
Hz sweep width, and a relaxation delay of 1.2 s. Similar ROESY data Born solvation model was used for implicit solvation.
sets were also collected for comparison purposes where the offset was The binding models for 1, 3, 5, 10, and 12 were constructed based
shifted to 0 or 10 ppm, the spin-lock time varied, and where the spin- on the in-house X-ray structure of Faldaprevir (ref 3f, PDB 3P8O) in
lock was applied by other methods. The existence of cis- and trans- complex with the protease. A complex involving compound 1 was also
rotamers was clearly identified through variable temperature studies derived from an X-ray structure of the highly related analogue 1′ (PDB
that showed the coalescence of resonances at higher temperatures, as code 4K8B) such that the urea was replaced by a carbamate (N atom
well as the observation of chemical exchange crosspeaks in the ROESY to an O) then energy minimized. The X-ray structure of DDIVPC
spectra. The resonances of the trans-rotamer were easily distinguished (PDB code 4JMY) was acquired as described in the Materials and
from those of the cis-rotamer based on comparison of distinctive Methods section. The complex involving BILN 2061 was based on the
ROESY crosspeak intensities (e.g., P2-Hδ/P3-Hα and P2-Hα/P3- modification of an X-ray structure3f,6 of a related macrocyclic
Hα). The 1H chemical shift and J-coupling values were easily extracted compound and energy minimization.
from high-resolution, one-dimensional spectra. In some instances, The binding model for Boceprevir was created from PDB code
decoupling strategies (selective irradiation) were used to derive 3LOX where the cyclohexyl was transformed to the tert-butyl. An in-
coupling constants. 13C resonance assignments were carried out using house X-ray structure was used for Faldaprevir (PDB 3P8O). PDB
1
H/13C correlation HMBC and HMQC experiments. HMBC spectra, code 3KEE was used for Simeprevir, while PDB code 3M5L was used
providing long-range 1H/13C correlation data, were acquired with no for Danoprevir. Vaniprevir was modeled starting from the X-ray
decoupling during acquisition, 256−512 t1 data points from the structure of Danoprevir while the binding mode model of Asunaprevir
addition of 64 transients, a relaxation delay of 1.5 s, and delays for was based on PDB code 3OYP. Telaprevir was modeled based on PDB
evolution of long-range couplings corresponding to 5 or 8 Hz. HMQC code 3LOX with initial cyclopropyl orientation based on PDB code
spectra provided single-bond 1H/13C correlation data and were 3M5L.
acquired with decoupling (GARP) during acquisition, 512 t1 points The superposition of the protein−ligand complexes was performed
from the addition of 16−32 transients, a relaxation delay 1.5 s, and a with MOE’s Superpose module. For Figure 4, the RMSD = 0.62 Å for
delay for evolution of one-bond coupling corresponding to 140 Hz. all complexes. The surface was created with the coordinates of
Samples and experimental data acquisition for transferred NOESY Faldeprevir (PDB code 3P8O) [LigX: unselect: Waters farther than
experiments are described in ref 3a. 4.5 from ligand; select: Solvent restraint; fix atoms farther than 10 Å;
Samples for the 13C T1 relaxation measurements were prepared by refine to 0.1 kcal/mol/Å].
dissolving inhibitors 1−12 in DMSO-d6 (41−77 mM), adding EDTA-
d16 (3 mM) for the purpose of chelating any traces of contaminating
paramagnetic metals, and degassing the solution by vacuum followed
■
*
ASSOCIATED CONTENT
S Supporting Information
by purges with argon gas. The glass NMR tubes were immediately Characterization information for novel compounds reported
sealed using a flame. Relaxation data were acquired at 150 MHz (with
cryoprobe) and 27 °C using the inversion recovery method with
here, the structures of patented compounds that also have the
power-gated proton decoupling (WALTZ16) during acquisition. DDIVPC critical scaffold, NMR resonance assignment tables,
Twelve spectra were acquired corresponding to the τ delays (0.01, free vs bound differences, resistance suppression, free-state
0.1, 0.2, 0.37, 0.6, 0.65, 1.0, 1.5, 2.2, 3.0, 5.0, and 7.5 s). Each spectrum ROESY data and more. This material is available free of charge
via the Internet at http://pubs.acs.org.

■
was typically acquired by adding 3000 transients and using a relaxation
delay of 8.0 s. T1 relaxation curves were calculated based on resonance
intensities using the WinNMR software (Bruker Canada, Milton, AUTHOR INFORMATION
Ontario). Relaxation times reported here are derived from resonances Corresponding Authors
that were clearly resolved (i.e., the cis- and trans-resonances were
distinct) and for which the calculated T1 curves described the
*E-mail: stevenrlaplante@gmail.com.
experimental data with relatively little deviation. Some samples were *E-mail: stephen.kawai@concordia.ca.
subjected to repeated data acquisition and calculations, and less than Present Addresses
§
5−10% variation was generally observed, as a result of good S/N NMX Solutions, 500 boulevard Cartier Ouest, Suite 6000,
obtained with a cryoprobe, and the fact that natural abundance 13C Laval, QC, Canada H7 V 5B7.
nuclei relax predominantly via the covalently attached hydrogen(s). ⊥
Department of Chemistry and Biochemistry, 7141 Sherbrooke
All two-dimensional data sets were processed using XWinNMR Street West, SP 201.00, Montreal, QC, Canada H4B 1R6.
(Bruker Canada, Milton, Ontario) and WinNMR software. Data sets
were typically zero-filled to yield 2048 (f2) × 1024 (f1) points after Notes
Fourier transformation using a shifted sine-bell window function. 1H The authors declare no competing financial interest.
and 13C spectra were chemical shift calibrated relative to the standard
values attributed to the DMSO peaks (2.5 and 39.51 ppm,
respectively). ROESY crosspeak volumes were scaled and converted
■ ACKNOWLEDGMENTS
We are grateful for the valuable contributions and support
to apparent interproton distances. The derived distances were
normalized to the Q-H5 to Q-H6 crosspeak, which corresponds to a
provided by G. Kukolj, J. Gillard, P. Anderson, M. Bailey, M.
fixed distance of 2.4 Å. Only well-resolved ROESY crosspeaks Cordingley, R. Bethell, P. Edwards, D. Lamarre, M.-A. Poupart,
corresponding to the trans-conformers and which did not exhibit P. White, M. Llinas-Brunet, J. Rancourt, A.-M. Faucher, and D.
any visual artifacts stemming from COSY or HOHAHA contributions Thibeault. We also acknowledge the valuable teamwork and
contributions of our colleagues at Boehringer Ingelheim.

■
were used in the analysis. Data similar to that reported below were
collected using other offsets and spin-lock sequences.
Molecular Modeling and Docking. Protein−ligand complexes ABBREVIATIONS USED
were derived in the following manner. If multiple chains were present
Boc, tert-butoxycarbonyl; COSY, correlated spectroscopy;
in the deposited X-ray coordinates then Chain A was used by default.
Crystallographic water molecules were kept. Hydrogen atoms were DMSO, dimethylsulfoxide; DTT, dithiothreitol; EDTA, ethyl-
added using MOE’s Protonate 3D module. The Amber99 force field enediaminetetraacetic acid; HCMV, human cytomegalovirus;
was applied with AM1-BCC charges for the ligands. In order to relieve HCV, hepatitis C virus; HEPES, 4-(2-hydroxyethyl)-1-piper-
any potential major strains between the protein and ligand, energy azineethanesulphonic acid; HMBC, heteronuclear multiple-
minimization was performed to an RMS gradient of 0.1 kcal/mol/ bond correlation; HMQC, heteronuclear multiple quantum
K dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX
Journal of Medicinal Chemistry Article

coherence; MES, 4-morpholineethanesulfonic acid; NMR, LaPlante, S. R.; Maurice, R.; Poirier, M.; Poupart, M.-A.; Thibeault, D.;
nuclear magnetic resonance; NS, nonstructural; ROESY, Wernic, D.; Lamarre, D. Peptide-based inhibitors of the hepatitis C
rotating-frame nuclear Overhauser effect; SAR, structure− virus serine protease. Bioorg. Med. Chem. Lett. 1998, 8, 1713−1718.
activity relationship; TCEP, tris(2-carboxyethyl)phosphine; (c) Llinàs-Brunet, M.; Bailey, M.; Fazal, G.; Ghiro, E.; Gorys, V.;
Goulet, S.; Halmos, T.; Maurice, R.; Poirier, M.; Poupart, M.-A.;
Tris, tris(hydroxymethyl)aminomethane; transferred NOESY,
Rancourt, J.; Thibeault, D.; Wernic, D.; Lamarre, D. Highly potent and
transferred nuclear Overhauser effect; T1, longitudinal selective peptide-based inhibitors of the hepatitis C virus serine
relaxation time protease: towards smaller inhibitors. Bioorg. Med. Chem. Lett. 2000, 10,

■ REFERENCES
(1) (a) Teague, S. Implications of protein flexibility for drug
2267−2270. (d) LaPlante, S. R.; Aubry, N.; Bonneau, P.; Kukolj, G.;
Lamarre, D.; Lefebvre, S.; Li, H.; Llinàs-Brunet, M.; Plouffe, C.;
Cameron, D. R. NMR line-broadening and transferred NOESY as a
medicinal chemistry tool for studying inhibitors of the hepatitis C virus
discovery. Nat. Rev. Drug Discovery 2003, 2, 527−541. (b) Weikl, T.
NS3 protease domain. Bioorg. Med. Chem. Lett. 2000, 10, 2271−2274.
R.; von Deuster, C. Selected-fit versus induced-fit protein binding:
(e) Goudreau, N.; Cameron, D. R.; Bonneau, P.; Gorys, V.; Plouffe,
Kinetic differences and mutational analysis. Proteins: Struct., Funct.,
C.; Poirier, M.; Lamarre, D.; Llinàs-Brunet, M. NMR structural
Bioinf. 2009, 75, 104−110. (c) Hammes, G. G.; Chang, Y. C.; Oas, T.
characterization of peptide inhibitors bound to the hepatitis C virus
G. Conformational selection or induced fit: A flux description of
reaction mechanism. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 13737− NS3 protease: Design of a new P2 substituent. J. Med. Chem. 2004, 47,
13741. (d) Bakan, A.; Bahar, I. The intrinsic dynamics of enzymes 123−132. (f) Lemke, C. T.; Goudreau, N.; Zhao, S.; Hucke, O.;
plays a dominant role in determining the structural changes induced Thibeault, D.; Llinàs-Brunet, M.; White, P. W. Combined X-ray, NMR,
upon inhibitor binding. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 14349− and kinetic analyses reveal uncommon binding characteristics of the
14354. (e) Lange, O. F.; Lakomek, N. A.; Fares, C.; Schroeder, G. F.; hepatitis C virus NS3-NS4A protease inhibitor BI 201335. J. Biol.
Gunnar, G.; Walter, K. F. A.; Becker, S.; Meiler, J.; Grubmueller, H.; Chem. 2011, 286, 11434−11443. (g) Kawai, S. H.; LaPlante, S. R.;
Griesinger, C.; de Groot, B. L. Recognition dynamics up to Llinàs-Brunet, M.; Hucke, O. The effect of the P1 side-chain on the
microseconds revealed from an RDC-derived ubiquitin ensemble in binding of optimized carboxylate and activated carbonyl inhibitors of
solution. Science 2008, 320, 1471−1475. (f) Kern, D.; Zuiderweg, E. R. the hepatitis C virus NS3 protease. Fut. Med. Chem. 2010, 2, 1073−
The role of dynamics in allosteric regulation. Curr. Opin. Struct. Biol. 1081 and references therein.
2003, 13, 748−757. (g) Mittermaier, A.; Kay, L. E. New tools provide (4) (a) LaPlante, S. R.; Aubry, N.; Bonneau, P. R.; Cameron, D. R.;
new insights in NMR studies of protein dynamics. Science 2006, 312, Lagacé, L.; Massariol, M.-J.; Montpetit, H.; Plouffe, C.; Kawai, S. H.;
224−228. (h) Koshland, D. The active site and enzyme action. Adv. Fulton, B. D.; Chen, Z.; Ni, F. Biochemistry 1998, 37, 9793−9801.
Enzymol. Relat. Subj. Biochem. 1960, 22, 45−97. (b) Edwards, P. J.; LaPlante, S. R. Peptides as Leads for Drug
(2) (a) LaPlante, S. R.; Bonneau, P. R.; Aubry, N.; Cameron, D. R.; Discovery, in Peptide Drug Discovery and Development; Castanho, M.,
Déziel, R.; Grand-Maître, C.; Plouffe, C.; Tong, L.; Kawai, S. H. Santos, N. C., Eds.; Wiley: New York, 2011; pp 3−55.
Characterization of the human cytomegalovirus protease as an (5) Nar, H.; Bauer, M.; Schmid, A.; Stassen, J.-M.; Wienen, W.;
induced-fit serine protease and the implications to the design of Priepke, H. W. M.; Kauffmann, I. K.; Ries, U. J.; Hauel, N. H.
mechanism-based inhibitors. J. Am. Chem. Soc. 1999, 121, 2974−2986. Structural basis for inhibition promiscuity of dual specific thrombin
(b) LaPlante, S. R.; Gillard, J. R.; Jakalian, A.; Aubry, N.; Coulombe, and factor Xa blood coagulation inhibitors. Structure 2001, 9, 29−37.
R.; Brochu, C.; Tsantrizos, Y. S.; Poirier, M.; Kukolj, G.; Beaulieu, P. L. (6) Tsantrizos, Y. S.; Bolger, G.; Bonneau, P.; Cameron, D.;
Importance of ligand bioactive conformation in the discovery of potent Goudreau, N.; Kukolj, G.; LaPlante, S. R.; Llinas-Brunet, M.; Nar, H.;
indole-diamide inhibitors of the hepatitis C virus NS5B. J. Am. Chem. Lamarre, D. Macrocyclic inhibitors of the NS3 protease as potential
Soc. 2010, 132, 15204−15212. (c) LaPlante, S. R. Exploiting Ligand therapeutic agents of hepatitis C virus infection. Angew. Chem., Int. Ed.
and Receptor Adaptability in Rational Drug Design Using Dynamics 2003, 42, 1356−1360.
and Structure-based Strategies. Top. Curr. Chem. 2007, 272, 259−296. (7) (a) Lamarre, D.; Anderson, P. C.; Bailey, M.; Beaulieu, P.; Bolger,
(d) LaPlante, S. R.; Aubry, N.; Deziel, R.; Ni, F.; Xu, P. Transferred G.; Bonneau, P.; Bös, M.; Cameron, D. R.; Cartier, M.; Cordingley, M.
13C T1 relaxation at natural isotopic abundance:A practical method G.; Faucher, A.-M.; Goudreau, N.; Kawai, S. H.; Kukolj, G.; Lagace, L.;
for determining site-specific changes in ligand flexibility upon binding LaPlante, S. R.; Narjes, H.; Poupart, M.-A.; Rancourt, J.; Sentjens, R.
to a macromolecule. J. Am. Chem. Soc. 2000, 122, 12530−12535. E.; St George, R.; Simoneau, B.; Steinmann, G.; Thibeault, D.;
(e) Peng, J. W. New probes of ligand flexibility in drug design: Tsantrizos, Y. S.; Weldon, S. M.; Yong, C. L.; Llinas-Brunet, M. An
transferred (13)C CSA-dipolar cross-correlated relaxation at natural NS3 protease inhibitor with antiviral effects in humans infected with
abundance. J. Am. Chem. Soc. 2003, 125, 11116−11130. (f) Peng, W.; hepatitis C virus. Nature 2003, 426, 186−189. (b) LaPlante, S. R.;
Wilson, B. D.; Namanja, A. T. Mapping the dynamics of ligand Llinàs-Brunet, M. Dynamics and structure-based design of drugs
reorganization via 13CH3 and 13CH2 relaxation dispersion at natural targeting the critical serine protease of the hepatitis C virus - from a
abundance. J. Biomol. NMR 2009, 45 (1−2), 171−183. (g) Peng, J. W. peptidic substrate to BILN 2061. Curr. Med. Chem.: Anti-Infect. Agents
Communication breakdown: Protein dynamics and drug design. 2005, 4, 111−132.
Structure 2009, 17, 319−320. (h) Mauldin, R. V.; Carroll, M. J.; Lee, (8) Llinàs-Brunet, M.; Bailey, M. D.; Goudreau, N.; Bhardwaj, P. K.;
A. L. Dynamic dysfunction in dihydrofolate reductase results from Bordeleau, J.; Bös, M.; Bousquet, Y.; Cordingley, M. G.; Duan, J.;
antifolate drug binding: Modulation of dynamics within a structural Forgione, P.; Garneau, M.; Ghiro, E.; Gorys, V.; Goulet, S.; Halmos,
state. Structure 2009, 17, 386−394. (i) Taylor, R. E.; Chen, Y.; Beatty, T.; Kawai, S. H.; Naud, J.; Poupart, M. A.; White, P. W. Discovery of a
A.; Myles, D. C.; Zhou, Y. Conformation−activity relationships in potent and selective noncovalent linear inhibitor of the hepatitis C
polyketide natural products: A new perspective on the rational design virus NS3 protease (BI 201335). J. Med. Chem. 2010, 53, 6466−6476.
of Epothilone analogues. J. Am. Chem. Soc. 2003, 125, 26−27. (9) (a) Hinrichsen, H.; Benhamou, Y.; Wedemeyer, H.; Reiser, M.;
(j) Thoma, G.; Magnani, J. L.; Patton, J. T.; Ernst, B.; Jahnke, W. Sentjens, R. E.; Calleja, J. L.; Forns, X.; Erhardt, A.; Cronlein, J.;
Preorganization of the bioactive conformation of sialyl LewisX Chaves, R. L.; Yong, C. L.; Nehmiz, G.; Steinmann, G. G. Short-term
analogues correlates with their affinity to E-selectin. Angew. Chem., antiviral efficacy of BILN 2061, a hepatitis C virus serine protease
Int. Ed. 2001, 125, 1941−1945. inhibitor, in hepatitis C genotype 1 patients. Gastroenterology 2004,
(3) (a) LaPlante, S. R.; Cameron, D. R.; Aubry, N.; Lefebvre, S.; 127, 1347−1355. (b) Zeuzem, S.; Asselah, S.; Angus, P.; Zarski, J.-P.;
Kukolj, G.; Maurice, R.; Thibeault, D.; Lamarre, D.; Llinàs-Brunet, M. Larrey, D.; Mullhaupt, H.; Gane, E.; Schuchmann, M.; Lohse, A.; Pol,
Solution structure of substrate-based ligands when bound to hepatitis S.; Bronowicki, J.-P; Roberts, S.; Arasteth, A.; Zoulim, A.; Heim, M.;
C virus NS3 protease domain. J. Biol. Chem. 1999, 274, 18618−18624. Stern, J. O.; Kukolj, G.; Nehmiz, H.; Haefner, C.; Boecher, W. O.
(b) Llinàs-Brunet, M.; Bailey, M.; Fazal, G.; Goulet, S.; Halmos, T.; Efficacy of the protease inhibitor BI 201335, polymerase inhibitor BI

L dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry Article

207127, and ribavirin in patients with chronic HCV infection.

Gastroenterology 2011, 141, 2047−2055. (c) Zeuzem, S.; Soriano, V.;
Asselah, T.; Bronowicki, J.-P.; Lohse, A.; Müllhaupt, B.; Schuchmann,
M.; Bourliere, M.; Buti, M.; Roberts, S.; Gane, E.; Stern, J. O.; Kukolj,
G.; Dai, L.; Böcher, W. O.; Mensa, F. J. SVR4 and SVR12 with an
interferon-free regimen of BI201335 AND BI207127, ± ribavirin, in
treatment-naive patients with chronic genotype-1 HCV infection:
Interim results of SOUND-C2. J. Hepatol. 2012, 56 (Suppl. 2), S45.
(d) Zeuzem, S.; Soriano, V.; Asselah, T.; Bronowicki, J.-P.; Lohse, A.
W.; Müllhaupt, B.; Schuchmann, M.; Bourlière, M.; Buti, M.; Roberts,
S. K.; Gane, E. J.; Stern, J. O.; Vinisko, R.; Kukolj, G.; Gallivan, J.-P.;
Böcher, W.-O.; Mensa, F. J. Faldaprevir and Deleobuvir for HCV
Genotype 1 Infection. N. Engl. J. Med. 2013, 369, 630−639.
(10) The peptide/protease subsite nomenclature used is that of
Schechter, I.; Berger, A. Biochem. Biophys. Res. Commun. 1967, 27,
157−162 where the amino acid residues are denoted ...P4-P3-P2-P1-
P1′-P2′..., the scissile bond being situated between P1 and P1′. The
notation ...S4-S3-S2... etc. is used to indicate the corresponding
binding subsites on the enzyme..
(11) (a) Bax, A.; Davis, D. G. Practical aspects of two-dimensional
transverse NOE spectroscopy. J. Magn. Reson. 1985, 63, 207−213.
(b) Bax, A.; Grzesiek, S. In Encyclopedia of Nuclear Magnetic Resonance;
Grant, D. M., Harris, R. K., Eds; Wiley: New York, 1996; pp 4157−
4167. (c) Hwang, T.-L.; Shaka, A. J. Cross relaxation without TOCSY:
transverse rotating-frame Overhauser effect spectroscopy. J. Am. Chem.
Soc. 1992, 114, 3157−3159. (d) Hwang, T.-L.; Shaka, A. J. Reliable
two-dimensional rotating-frame cross-relaxation measurements in
coupled spin systems. J. Magn. Reson., Ser. B 1993, 102, 155−165.
(e) Baleja, J. D.; Moult, J.; Sykes, B. D. J. Magn. Reson. 1990, 87, 375−
384. (f) Neuhaus, D.; Williamson, M. P. The Nuclear Overhauser Effect
in Structural and Conformational Analysis; VCH: New York, NY, 1989.
(12) Wüthrich, K. NMR of Proteins and Nucleic Acids; John Wiley &
Sons: New York, NY, 1986.
(13) Rancourt, J.; Cameron, D. R.; Gorys, V.; Lamarre, D.; Poirier,
M.; Thibeault, D.; Llinàs-Brunet, M. Peptide-based inhibitors of the
hepatitis C virus NS3 protease: Structure−activity relationship at the
C-terminal position. J. Med. Chem. 2004, 47, 2511−2522.
(14) Kawai, S. H.; Bailey, M. D.; Halmos, T.; Forgione, P.; LaPlante,
S. R.; Llinas-Brunet, M.; Naud, J.; Goudreau, N. The use of chemical
double-mutant cycles in biomolecular recognition studies: Application
to HCV NS3 protease inhibitors. ChemMedChem 2008, 3, 1654−1657.
(15) Kawai, S. H.; Aubry, N.; Duceppe, J.-S.; Llinàs-Brunet, M.;
LaPlante, S. R. Dimethylthiazolidine carboxylic acid as a rigid P3 unit
in inhibitors of serine proteases: Application to two targets. Chem. Biol.
Drug Des. 2009, 74, 517−522.
(16) Marsault, E.; Peterson, M. L. Macrocycles are great cycles:
Applications, opportunities, and challenges of synthetic macrocycles in
drug discovery. J. Med. Chem. 2011, 54, 1961−2004.

M dx.doi.org/10.1021/jm401338c | J. Med. Chem. XXXX, XXX, XXX−XXX