Pyrrolidine-5,5-trans-lactams. 2. The Use of X-ray Crystal Structure Data in the Optimization of P3 and P4 Substituents

Article abstract of DOI:10.1021/ol027014p

(Matrix Presented) In this, the second of two letters, we describe the elaboration of the pyrrolidine-5,5-trans-lactam template to delineate the requirements for optimal substitution of the pyrrolidine and lactam nitrogen atoms. Central to the strategy is the use of rapid iterative synthesis in conjunction with X-ray crystal structure determination of ligand-protein complexes.

Full text of DOI:10.1021/ol027014p

ORGANIC
LETTERS
2002
Vol. 4, No. 25
4479-4482
Pyrrolidine-5,5-trans-lactams. 2. The Use
of X-ray Crystal Structure Data in the
Optimization of P3 and P4 Substituents
David M. Andrews,* Helene Chaignot, Barry A. Coomber, Andrew C. Good,†
S. Lucy Hind, Martin R. Johnson, Paul S. Jones, Gail Mills, J. Ed Robinson,
Tadeusz Skarzynski, Martin J. Slater, and Donald O’N Somers
GlaxoSmithKline Medicines Research Centre, Gunnels Wood Road, SteVenage,
SG1 2NY, UK
daVid.m.andrews@gsk.com
Received October 2, 2002
ABSTRACT
In this, the second of two letters, we describe the elaboration of the pyrrolidine-5,5-trans-lactam template to delineate the requirements for
optimal substitution of the pyrrolidine and lactam nitrogen atoms. Central to the strategy is the use of rapid iterative synthesis in conjunction
with X-ray crystal structure determination of ligand−protein complexes.
Hepatitis C virus (HCV) infects chronically an estimated 3%
of the global human population,1 often leading to cirrhosis,
hepatocellular carcinoma, and liver failure in later life.2
Current therapies are based upon interferon-R, alone or in
combination with ribavirin. Although sustained response rates
are markedly improved using combination therapies, at least
50% of patients fail to show a sustained response. Addition-
ally, current therapies have the disadvantage of frequent and
severe side-effects.3 The development of new therapies for
treating HCV infection effectively is thus of paramount
importance and is currently an intensive area of research.4
HCV is a small, enveloped virus, the genome of which is
a 9.5 kb single-stranded RNA that encodes for a single
polyprotein of 3010-3030 amino acids. Mature nonstructural
replicative proteins are released from this polyprotein by the
action of the viral proteases NS2 and NS3. It has been
established that introducing mutations into the NS3 protease
region of the HCV genome abolishes infectivity,5 demon-
strating that NS3 protease is thus an essential viral function
and should prove to be an excellent target for the develop-
ment of novel anti-HCV agents.
In the preceding Letter,6 we outlined the synthesis and
incorporation of the R,R-disubstituted trans-lactam core into
nonpeptidic inhibitors of the NS3 protease; the optimization
of the P1 substituent was described. Herein, we describe how
initial X-ray structures and molecular models obtained with
1 and 2 (Figure 1) were used to guide the selection of potency
enhancing substituents on the pyrrolidine and lactam nitrogen
atoms.
† Current address: Bristol-Myers Squibb, 5 Research Parkway, P.O. Box
5100, Wallingford, CT 06492
(1) World Health Organization Weekly Epidemiological Record 1997,
72, 65.
(2) (a) Gerber, M. A. J. Hepatol. 1993, 17 (suppl. 3), 108. (b) Alter, M.
J. Hepatology 1997, 26, 62. (c) Hoofnagle, J. H. Hepatology 1997, 26, 15S.
(d) Kwong, A. D. AntiViral Res. 1998, 40, 1. (e) Hoofnagle, J. H. Digestion
1998, 59, 563.
Crystallographic Evaluation. The potency of spirocy-
lobutyl inhibitor 2 can be rationalized by modeling the
inhibitor into previously reported crystal structures;7 the
spirocylobutyl fills the S1 pocket more optimally than the
R-ethyl side chain of 1. Modeling the interaction of the Boc-
(3) (a) Reichard, O.; Scvarcz, R.; Weiland, O. Hepatology 1997, 26, 108S.
(b) Davies, G. L.; et al. N. Eng. J. Med. 1998, 339, 1493. (c) McHutchinson,
J. G.; et al. N. Eng. J. Med. 1998, 339, 1485.
(4) May Wang, Q.; Du, M. X.; Hockman, M. A.; Johnson, R. B.; Sun,
X.-L. Drugs Future 2000, 25, 933.
(5) Kolykhalov, A. A.; Mihalik, K.; Feinstone, S. M.; Rice, C. M. J.
Virol. 2000, 74, 2046.
(6) Andrews, D. M.; et al. Org. Lett. 2002, 4, 4475.
(7) Kim, J. L.; et al. Cell 1996, 8, 344.
10.1021/ol027014p CCC: $22.00 © 2002 American Chemical Society
Published on Web 11/19/2002

positively or negatively. The urethane nitrogen is within 3.4
Å of the carbonyl of Ala157, which is slightly too far to be
considered a viable hydrogen bond, although we considered
that this distance would be more likely in the region of 2.9
Å in the modeled preacylation complex. The carbonyl oxygen
of the urethane is solvent exposed and unlikely to contribute
much to binding; the bridging oxygen may also be within
hydrogen-bonding distance to the oxygen of Ala157. This
oxygen-oxygen interaction is clearly not a hydrogen-
bonding combination, but it suggests that in future analogues,
the urethane should be replaced with a urea moiety. The tert-
butyl group is just above and to the edge of S4, suggesting
that a methylene spacer linking to a small hydrophobic group
may improve binding. At the other end of the molecule, it
is very difficult to rationalize the role of the lactam
substituent because the structure only presents a picture of
the postacylation complex. More useful from the perspective
of molecule design is a model of the preacylation complex,
and thus an attempt was made to construct a model of the
preacylation complex computationally. Figure 3 illustrates
Figure 1. trans-Lactams used in X-ray crystallographic evaluation
and modeling.
valine substituent offers considerably greater challenges,
however, and only a direct determination of an X-ray crystal
structure can unequivocally establish how these inhibitors
bind.
A 2.6 Å resolution X-ray structure of 1 was obtained by
soaking preformed crystals of the truncated NS3 protease
domain: NS4A cofactor (tNS3:NS4A)7 (pH 6.5, compound
concentration 2mM, 4% DMSO, 48 h).8 Although the crystals
were found to be only modestly diffracting, a number of key
features could be established (Figure 2). The core of the
Figure 3. Computational model of preacylation complex of 2 with
HCV NS3 protease. Solvent-accessible protein surface is colored
by electrostatic potential, and residues and subsites referred to in
the text are marked.
Figure 2. Crystal structure of 1 complexed with HCV NS3
protease. Bold structure is the ring-opened form of 1 covalently
bound to Ser139. Key protein residues (referred to in the text) are
drawn with thick bonds, and other NS3 residues are drawn with
thin bonds.
a model built on the basis of compound 2, utilizing the
crystallographic data on the acylation complex of 1. The
rigidity of the trans-lactam core made the reconstitution of
the trans-lactam system a trivial exercise, and this fragment
was then docked into the enzyme active site. Constraining
the spirocyclobutyl to fit into S1 and maintaining the distance
between the side chain oxygen of Serine-139 and the lactam
carbonyl to ca. 2.6 Å ensures correct overall orientation.
Optimization of P3 Substituent. Inspection of the crystal
structures and models (Figures 2 and 3) suggested amino
acid scanning as a means of enhancing binding in this region.
Boc- or Cbz-protected R-amino acids bearing the normal
range of functionality were synthesized as described for 2,
and a selection of the results obtained are summarized (Figure
4, information on rapid array synthesis, compound charac-
terization, and further analogues are described in Supporting
Information).
molecule corresponds to the areas of electron density that
are best defined, and the inhibitor is clearly seen as being
covalently linked to the protein through Ser139. The R-ethyl
side chain is seen in S1, which has Phe154 at the base of
the pocket, accounting for the substrate preference for
cysteine at this position. The isopropyl side chain of valine
overlays a shallow, solvent-exposed indentation on the
protein surface, suggesting that it should be possible to
substitute this position with a variety of groups; however, it
is unlikely that potency will be significantly impacted, either
(8) Atomic coordinates have been deposited in the Protein Data Bank
as 1N1L.
4480
Org. Lett., Vol. 4, No. 25, 2002

Table 1. Optimization of P4 Substituent
Figure 4. Optimization of P3 substituent.
Data are consistent with the X-ray structural and modeling
data. Only the L-R-amino acids are accommodated, with a
strong preference for the more bulky â-branched residues
(cf. Val, Ile, and notably allo-Ile (3c) with Nva; cf. Phg with
Phe). Charged residues are less potent than Valine despite
the known ability of the enzyme to accommodate acidic
residues in the 4A/4B substrate.9 N-Methylated amino acids
are very poorly active, confirming the likely critical role of
the hydrogen bond to the carbonyl of Ala157.
Optimization of the P4 Substituent. Having investigated
the requirements for the P3 substituent, we next turned our
attention to P4. Figures 2 and 3 show the tert-butyl group to
lie just at the lip of the S4 pocket. An array of compounds
was synthesized by rapid array methodology with the
objective of improving both the potency and overall molecule
properties. (Information on rapid array synthesis and com-
pound characterization is described in Supporting Informa-
tion). Several groups containing a basic nitrogen were
synthesized with the objective of enhancing the solubility
(4a-d, Table 1) and interacting with Asp168.
Linear, flexible capping groups were investigated (4a and
4b) along with more compact, cyclic groups connected by a
linker of varying length (4c and 4d). The poor potency of
the groups investigated illustrates that it is difficult to obtain
increases in potency by making interactions with highly
solvent-exposed charged residues. Other substituents bearing
hydrogen-bonding functionality also failed to show any
improvement (4e).
2). However, optimal potency is demonstrated by groups of
similar size to a compact C4-methyl unit (4g and 4i) or C5-
methyl unit (4f), since smaller (4j) or larger (4k) groups are
less potent. In general, replacing the urethane (4l) with the
more flexible carboxamide linker (4m) reduced potency. This
trend was reproduced in the cyclohexyl series: interestingly,
although the cyclohexyl carbamoyl (4n) is suboptimal,
changing the linking oxygen to methylene (4o) is deleterious,
although synthesis of the urea analogue 4p gives a dramatic
enhancement in potency, possibly as a consequence of
formation of the putative hydrogen bond. Future com-
munications will elaborate the extent and limitations of this
interesting observation.
Replicon Cell-Based Assay Results. Representative com-
pounds were examined in a replicon surrogate cellular assay
system.10 Overall, a reasonable correlation between enzyme
and cellular assay data was observed. A number of factors
Compact, hydrophobic groups were also evaluated. The
most important observation was that introducing a spacer
methylene modestly improved potency, presumably by
providing additional conformational flexibility (cf. 4f-i with
(10) (a) Lohmann, V.; Korner, F.; Koch, J.; Herian, U.; Theilmann, L.;
Bartenschlager, R. Science 1999, 285 (5424), 110. (b) Blight, K. J.;
Kolykhalov, A. A.; Rice, C. M. Science 2000, 290 (5498), 1972. (c)
Bartenschlager, R.; Lohmann, V. AntiViral Res. 2001, 52, 1. (d) Parry, N.
R.; Chung, V.; Viner, K. C.; Carroll, A. R. AntiViral Res. 2002, 53, A73,
Abstract 120.
(9) Grakoui, A.; et al. J. Virol. 1993, 67, 2832.
Org. Lett., Vol. 4, No. 25, 2002
4481

yield replicon IC50s falling in a narrow range. Compounds
bearing highly polar groups fail to show activity, e.g., 3p,
4c, 4d, most likely as a result of poor membrane permeability
or insufficient inherent biochemical potency. Compound 3d
appears anomalously active (in the light of its biochemical
potency); however, in-assay cellular toxicity was evident at
concentrations 10 times the apparent IC50 and is the probable
explanation.
Table 2. Replicon Cell-Based Assay IC50 Values for a Range
of P3- and P4-Modified Pyrrolidine-5,5-trans-lactams
IC50
In conclusion, we have shown that rapid array synthesis
can be used in conjunction with crystallographic and model-
ing data to demonstrate that one of the optimal substituents
on the trans-lactam pyrrolidine nitrogen is valine. We have
also shown that the valine nitrogen may be further substituted
with small, hydrophobic carbamates to yield drug-like
molecules in which potency is maintained and in some cases
enhanced.
compd
P4
P3
(uM)
2
Boc
Boc
Boc
Boc
Boc
Boc
Boc
Cbz
Cbz
Val
D-Val
Ile
5.7
3a
3b
3c
3d
3f
3h
3k
3p
4c
4d
4g
4l
>100
10.3
15.6
allo-isoloeucine
cyclohexylglycine 3.71
Leu
Ala
14.8
38
N-Me-Val
Glu
>100
9.311
2411
8.3
8.6
22
Acknowledgment. We thank Drs. Berwyn Clarke and
George Hardy for support and encouragement; Drs. Graham
Baker, Sue Bethell, and Malcolm Ellis for provision of NS3
protease protein and initial assay systems; Derek Evans for
provision of intermediates; Norman M. Gray and Seb J.
Carey for biochemical test data; and Dr. Nigel Parry and
Liz Amphlett for replicon test results.
2-morpholino-ethyloxycarbonyl Val
3-piperidino-propyloxycarbonyl Val
isobutyloxycarbonyl
Cbz
Val
Val
impact cellular potency, but it is clear that prerequisites for
activity are good cellular penetration in conjunction with
kobs/I greater than 150 M-1 s-1. Compounds 2, 3b-f, and
4g, all rather close structural analogues, demonstrate bio-
chemical potency in a very similar range and, reassuringly,
Supporting Information Available: Detailed experi-
mental procedures for synthesis of representative compounds
and characterization data for test compounds. This material
is available free of charge via the Internet at http://pubs.acs.org.
(11) In-assay toxicity observed at concentrations <10-fold greater than
the quoted IC50.
OL027014P
4482
Org. Lett., Vol. 4, No. 25, 2002