Molecular modeling based approach to potent P2-P4 macrocyclic inhibitors of hepatitis C NS3/4A protease

Article abstract of DOI:10.1021/ja711120r

Molecular modeling of inhibitor bound full length HCV NS3/4A protease structures proved to be a valuable tool in the design of a new series of potent NS3 protease inhibitors. Optimization of initial compounds provided 25a. The in vitro activity and select

Full text of DOI:10.1021/ja711120r

Published on Web 03/14/2008
Molecular Modeling Based Approach to Potent P2-P4 Macrocyclic Inhibitors
of Hepatitis C NS3/4A Protease
Nigel J. Liverton,*,‡ M. Katharine Holloway,§ John A. McCauley,‡ Michael T. Rudd,‡
John W. Butcher,‡ Steven S. Carroll,† Jillian DiMuzio,† Christine Fandozzi,| Kevin F. Gilbert,‡
Shi-Shan Mao,† Charles J. McIntyre,‡ Kevin T. Nguyen,‡ Joseph J. Romano,‡ Mark Stahlhut,†
Bang-Lin Wan,| David B. Olsen,† and Joseph P. Vacca‡
Departments of Medicinal Chemistry, AntiViral Research, Drug Metabolism, and Molecular Systems,
Merck Research Laboratories, West Point, PennsylVania 19486
Received December 19, 2007; E-mail: nigel_liverton@merck.com
Worldwide, 170-200 million people are chronically infected
with the hepatitis C virus (HCV).1 The positive RNA strand
FlaViViridae virus replicates primarily in the liver, and while disease
progression is typically a slow process occurring over many years,
ultimately a significant fraction develops serious liver disease
including cirrhosis and hepatocellular carcinoma.2 HCV is currently
a leading cause of death in HIV coinfected patients3 and is the
most common basis for liver transplantation surgery.4 Several
promising antiviral targets for HCV have emerged in recent years,5
with NS3/4A protease inhibitors showing perhaps the most dramatic
Figure 1. NS3/4A protease inhibitors.
antiviral effects.6 Clinical proof of concept has been demonstrated
both for rapidly reversible inhibitors and slowly reversible keto-
amide active site serine trap compounds such as BILN-2061 (1)7
and VX-950 (telaprevir, 2),8 respectively (Figure 1).
Examination of published views of a close analog of BILN-2061
bound to the 1-180 protease domain of NS3 protease9 suggests
that the P2 thiazolylquinoline portion of the inhibitor lies on a
relatively featureless enzyme surface with binding interactions that
provide little apparent basis for the dramatic potency derived from
that moiety (>30 000-fold) in a related series of tripeptide inhibi-
tors.10 In an effort to rationalize this, we chose to model 1 bound
to the full length NS3/4A protein including the significantly larger
helicase domain (Figure 2) to determine the extent of any role the
helicase could play in inhibitor binding. No full length structures
with inhibitors bound are currently available; consequently a
published apoenzyme structure11 was used as the starting point.
To mimic the conformational change required to permit inhibitor
binding, the six C-terminal residues (DLEVVT) of the helicase
Figure 2. Model of 1 (cyan) bound to full length NS3/4A (protease, green;
domain which lie in the active site of the apo X-ray structure were
helicase, purple) with key protein-inhibitor interactions shown.
truncated.
nected, the proposed P2-P4 linker was formed, and a 3-phe-
Analysis of 1 docked in the latter structure indicated that the
nylquinoline P2 was used to facilitate rapid synthesis of a range of
helicase domain can provide a surface over the P2 moiety including
analogs. Modeling scores for the different linker lengths by two
a pocket to accommodate the thiazolyl substituent. Specific
methods (Table 1) predicted that the 5- and 6-carbon linkers would
inhibitor-helicase interactions include His528-carbamate oxygen and
show the greatest activity.
Gln526-quinoline. Also apparent from this study is that there is
The desired compounds were prepared as outlined in Scheme 1
space to accommodate a connection between the carbamate
via a high yielding ring closing metathesis (RCM) strategy.
cyclopentane and the quinoline ring. Re-examination of the helicase
Displacement of the brosylate of proline derivative 4 with bromo-
C-terminus from the apo structure (Figure 3), overlaid with BILN-
hydroxyquinoline 5 yielded ether 6 which was vinylated via
2061, demonstrates that the side chain of Glu628 occupies the same
palladium catalyzed reaction with tributylvinyltin to give 7.
space as the proposed linker. Together, these observations suggested
Subsequent removal of the Boc protecting group and coupling with
that an alternative P4 cyclopentyl-P2 quinoline macrocyclization
the appropriate norleucine carbamate derivatives 8a-d afforded
to form a structurally distinct series of inhibitors was feasible.
key RCM precursors 9a-d. Ring closure was accomplished in
Initially targeted were carbamate derivatives 3a-d (3c docked
excellent yield (84-93%) with 1,3-bis(2,4,6-trimethylphenyl)-4,5-
in Figure 4), in which the P1-P3 macrocyclic linker was discon-
dihydroimidazol-2-ylidene[2-(isopropoxy)-5-(N,N-dimethylamino-
† Department of Antiviral Research.
‡ Department of Medicinal Chemistry.
§ Department of Molecular Systems.
| Department of Drug Metabolism.
sulfonyl)phenyl]methyleneruthenium(II) dichloride (Zhan 1b cata-
lyst) in 1,2-dichloroethane to give macrocyclic olefins 10a-d.
Conversion to final products 3a-d was carried out via hydrogena-
9
10.1021/ja711120r CCC: $40.75 © 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 4607-4609
4607

C O M M U N I C A T I O N S
Scheme 1. Synthesis of Initial Targets 3a-da
Figure 3. Model of 1 bound to full length NS3/4A with helicase C-terminus
(green) restored and Glu628 highlighted with mesh surface.
a Conditions: (a) Cs2CO3, NMP, 40 °C, 86%; (b) Bu3SnCHdCH2,
Pd(PPh3)4, toluene, 100 °C, 79%; (c) 4 N HCl, dioxane; (d) HATU, DIPEA,
DMAP, DMF, 8a-d; (e) Zhan 1b catalyst, DCE, ∼10 mM, 84-93%; (f)
H2, 10% Pd/C, EtOAc; (g) LiOH, THF, MeOH, H2O; (h) HATU, DIPEA,
DMAP, 11, DMF; (j) LiOH, THF, MeOH, H2O.
Figure 5. Additional phenylquinoline analogs.
Table 2. Pharmacokinetic Profiles of Key Compoundsa
Figure 4. Target macrocycle 3c (yellow) overlaid with 1.
Table 1. In Vitro Activitya
Cmax
(nM)
plasma AUC
0-4 h ( M*h)
4 h liver
concn ( M)
compound
µ
µ
15
16
25a
25b
7
6
240
110
<0.01
0.01
0.36
0.2
modeling
1b replicon IC50 (nM)
10% FBS 50% NHS
3.9
18.6
13.4
b
compound
Einter
Xscoreb
1b Ki (nM)
0.27
1
2
0.3
93
3
1100
19
4800
a Compounds dosed at 5 mg/kg P.O. in PEG400 (n ) 2-3).
3a
3b
3c
3d
13
14
15
16
25a
25b
-69.5
-70.5
-71.1
-71.4
8.09
8.26
8.36
8.44
2000
145
8.5
25
4400
40
<0.016
<0.016
0.07
0.18
6100
1150
1200
>100 000
5600
through synthesis of the corresponding 5-substituted derivative 13
by an analogous synthetic route, which proved dramatically less
active (Ki 4400 nM). In addition, synthesis of an acyclic analog 14
demonstrated that potency enhancement, particularly in the replicon
assay, could be achieved through macrocyclization.
9100
4800
>100 000
6.7
13
4.5
8.7
26
25
14
46
Previous work has established that the carboxylic acid function-
ality can be effectively replaced with a cyclopropylacylsulfona-
mide,15 and application of this strategy to 3c afforded 15, with
subnanomolar inhibition of NS3/4A protease (Ki <0.016 nM).
Disappointingly, given the critical need for liver exposure, oral
administration of 15 to rat at 5 mpk P.O. provided low (0.2 µM)
compound levels in liver at 4 h with barely detectable plasma
exposure (Table 2). In contrast, when the P3 n-butyl residue was
replaced with tert-butyl, the resultant inhibitor 16, with a very
similar in vitro activity profile, was effectively partitioned into liver
with a tissue concentration at 4 h of 3.9 µM, although plasma levels
were unimproved. The dramatic impact of this minor structural
change on liver levels strongly suggests that uptake is via an active
transporter mediated process. Having successfully demonstrated that
this macrocyclization approach could yield potent compounds with
a Data are geometric averages of g3 determinations. b Molecular model-
ing details in the Supporting Information.
tion of the styryl olefin, hydrolysis of the proline ester, coupling
with cyclopropylaminoester 1112 to afford 12a-d, and hydrolysis
of the P1 ethyl ester.
While 3a, with the three carbon linker, proved to have very
modest activity of 2000 nM in a genotype 1b NS3/4A enzyme
inhibition assay13 (Table 1), incremental lengthening of the linker
afforded dramatic improvements with optimized activity of 8.5 nM
in the case of the pentyl linker 3c. A corresponding improvement
in genotype 1b cell based replicon activity14 was also observed.
The point of attachment on the quinoline was shown to be critical
9
4608 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

C O M M U N I C A T I O N S
Scheme 2. Synthetic Approach to Isoquinolines 25a and 25ba
advanced development including intramolecular Heck or Suzuki
reactions, proline amide coupling, or RCM, in contrast to P1-P3
compounds such as 1, where an RCM step appears unavoidable.17
While there are reports of compounds employing a related P2-
P4 cyclization strategy,18,19 the inhibitors had modest micromolar
potencies. Synthesis of the cyclopropylacylsulfonamide analog of
one of these compounds had little effect on potency.20
In summary, molecular modeling of inhibitor bound full length
NS3/4A protease structures proved a valuable tool in the design of
a new series of potent NS3 protease inhibitors 3a-d which was
optimized to compound 25a. Compound 25a is accessible via a
9-step longest linear sequence in 35% overall yield leading to a
straightforward preparation of >10 g of material. The in vitro
activity and selectivity as well as the rat pharmacokinetic profile
of 25a compare favorably with the data for other NS3/4A protease
inhibitors currently in clinical development for treatment of HCV
infection.
a Conditions: (a) KOtBu, DMF; (b) HCl, EtOH, 65% (2 steps); (c) 20,
HATU, DIPEA, DMAP, DMF, 74%; (d) Bu3SnCHdCH2, Pd(PPh3)4,
toluene, 100 °C, 87%; (e) Zhan 1b catalyst, DCE, ∼10 mM, 83%; (f) H2,
10% Pd/C, EtOAc, quant.; (g) LiOH, THF, MeOH, H2O; (h) HATU,
DIPEA, DMAP, 24, DMF, 93%.
Supporting Information Available: Synthetic procedures and
characterization data for new compounds, molecular modeling methods,
and complete references for 7a, 7b, 8a, and 14. This information is
available free of charge via the Internet at http://pubs.acs.org.
References
significant liver exposure, we sought to broaden the strategy to
include alternative P2 moieties. The use of an isoquinoline P2 to
generate potent inhibitors has been reported previously,16 and an
unsubstituted form would also offer somewhat reduced molecular
weight inhibitors which might be reflected in improved systemic
exposure. To this end, the isoquinoline analogs 25a and 25b
possessing the optimal five carbon macrocycle linker length were
prepared (Scheme 2). Reaction of trans-hydroxyproline derivative
18 with 7-bromo-1-chloroisoquinoline 17, under basic conditions,
afforded the crude substitution product which was treated with
ethanolic HCl to effect concomitant esterification of the acid and
deprotection of the Boc group providing 19 in 65% overall yield.
Coupling with the pentenyl carbamate of tert-leucine 20 to give
21 (74%) was followed by vinylation to 22 (87%). RCM then
afforded almost exclusively E-olefin 23a in 83% isolated yield.
Hydrolysis and coupling with aminoacylsulfonamide 2416 afforded
25a. The corresponding saturated analog 25b was prepared from
23a analogously, following an initial hydrogenation step. Both
analogs showed very potent inhibition in both in vitro enzyme and
replicon assays, but, more importantly, liver exposure was dramati-
cally improved with 4 h liver concentrations of 18.6 and 13.4 µM,
respectively, for 25a and 25b. Furthermore, both compounds were
now clearly detectable in plasma with AUCs of 0.36 and 0.27
µM*h, respectively. Sustained exposure in liver is clearly an
important factor for any potential treatment for HCV infection, and
high compound levels relative to activity in the replicon assay (500
nM, 35 × replicon EC50 in the presence of 50% NHS) are
maintained in rat liver for 24 h following a 5 mg/kg dose of 25a.
A detailed evaluation of the more potent analog 25a revealed
no significant activity vs other serine proteases (>50 000-fold
selectivity over trypsin and chymotrypsin), hERG binding (IC50 >30
µM), or in a broad based Panlabs screen (>4000-fold selectivity).
While 25a has a number of functionalities that might be viewed as
potentially susceptible to metabolism, 25a is primarily excreted in
bile as an unchanged parent after rat IV dosing.
(1) WHO. Weekly Epidemiol. Rec. 1999, 74, 425-427.
(2) Liang, T. J.; Heller, T. Gastroenterology 2004, 127, S62-S71.
(3) Salmon-Ceron, D.; Lewden, C.; Morlat, P.; Be´vilacqua, S.; Jougla, E.;
Bonnet, F.; He´ripret, L.; Costagliola, D.; May, T.; Cheˆne, G. J. Hepatology
2005, 42, 799-805.
(4) Brown, R. S. Nature 2005, 436, 973-978.
(5) Gordon, C. P.; Keller, P. A. J. Med. Chem. 2005, 48, 1-20.
(6) For recent reviews, see: (a) Thomson, J. A.; Perni, R. B. Curr. Opin.
Drug DiscoVery DeV. 2006, 9, 606-617. (b) Chen, S.-H.; Tan, S.-L. Curr.
Med. Chem. 2005, 12, 2317-2342.
(7) (a) Llinas-Brunet, M. et al. J. Med. Chem. 2004, 47, 1605-1608. (b)
Lamarre, D. et al. Nature 2003, 426, 186-189.
(8) (a) Perni, R. B. et al. Antimicrob. Agents Chemother. 2006, 50, 899-
909. (b) Lin, C.; Kwong, A. D.; Perni, R. B. Infect. Disord.: Drug Targets
2006, 6, 3-16.
(9) Tsantrizos, Y. S.; Bolger, G.; Bonneau, P.; Cameron, D. R.; Goudreau,
N.; Kukolj, G.; LaPlante, S. R.; Llinas-Brunet, Nar, H.; Lamarre, D.
Angew. Chem., Int. Ed. 2003, 42, 1356-1360.
(10) LaPlante, S. R.; Llinas-Brunet, M. Curr. Med. Chem.- Anti-InfectiVe Agents
2005, 4, 111-132.
(11) Yao, N.; Reichert, P.; Taremi, S. S.; Prosise, W. W.; Weber, P. C. Structure
1999, 7, 1353-1363.
(12) Beaulieu, P. L.; Gillard, J.; Bailey, M. D.; Boucher, C.; Duceppe, J.-S.;
Simoneau, B.; Wang, X.-J.; Zhang, L.; Grozinger, K.; Houpis, I.; Farina,
V.; Heimroth, H.; Krueger, T.; Schnaubelt, J. J. Org. Chem. 2005, 70,
5869-5879.
(13) Mao, S.-S.; DiMuzio, J.; McHale, C.; Burlein, C.; Olsen, D. B.; Carroll,
S. S. Anal. Biochem. 2008, 373, 1-8.
(14) Migliaccio, G. et al. J. Biol. Chem. 2003, 278, 49164-49170.
(15) Tu, Y.; Scola, P. M.; Good, A. C.; Cambell, J. A. WO2005/054430, 2005.
(16) Wang, X. A.; Sun, L.-Q.; Sit, S.-Y.; Sin, Y.; Scola, P. M.; Hewawasam,
P.; Good, A. C.; Chen, Y.; Campbell, J. A. US 6,995,174, 2006.
(17) Nicola, T.; Brenner, M.; Donsbach, K.; Kreye, P. Org. Process Res. DeV.
2005, 9, 513-515.
(18) Marchetti, A.; Ontoria, J. M.; Matassa, V. G. Synlett 1999, S1, 1000-
1002.
(19) Chen, K. X.; Njoroge, F. G.; Pichardo, J.; Prongay, A.; Butkiewicz, N.;
Yao, N.; Madison, V.; Girijavallabhan, V. J. Med. Chem. 2006, 49, 567-
574.
(20) Incorporation of the cyclopropylacyclsulfonamide into 2619 (Ki 2100 nM)
afforded 27 (Ki 300 nM).
An additional attraction of this class of macrocycles is the
potential to use a range of macrocyclization strategies in any
JA711120R
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4609