Novel purine nitrile derived inhibitors of the cysteine protease cathepsin K

Article abstract of DOI:10.1021/jm0493111

Starting from the high-throughput screening hit 1a, novel cathepsin K inhibitors have been developed based on a purine scaffold. High-resolution X-ray structures of several derivatives have revealed the binding mode of these unique cysteine protease inhib

Full text of DOI:10.1021/jm0493111

J. Med. Chem. 2004, 47, 5833-5836
5833
Table 1. Compounds 1a-c Profiling Data
Novel Purine Nitrile Derived Inhibitors
of the Cysteine Protease Cathepsin K
Eva Altmann,* Sandra W. Cowan-Jacob, and
Martin Missbach*
Novartis Institutes for BioMedical Research,
CH-4002 Basel, Switzerland
Received August 20, 2004
IC50 (µM)a
Abstract: Starting from the high-throughput screening hit
1a, novel cathepsin K inhibitors have been developed based
on a purine scaffold. High-resolution X-ray structures of
several derivatives have revealed the binding mode of these
unique cysteine protease inhibitors.
compd
R
Cath K
Cath L
Cath S
1a
1b
1c
CN
CH3
CH2OH
0.014
.1
.1
0.430
>10
>10
0.340
>10
>10
a Inhibition of rh cathepsins K, L, and S in a fluorescence-based
assay employing Z-Phe-Arg-AMC (cathepsins K and L) and Z-Leu-
Leu-Arg-AMC (cathepsin S) as synthetic substrates. Data repre-
sent the mean of two experiments performed in duplicate.
Cysteine proteases have received increasing interest
as targets for therapeutic intervention in different
disease areas. This has led to considerable efforts aimed
at the discovery, design, and optimization of novel
cysteine protease inhibitors.1-6 Cathepsin K is a lyso-
somal cysteine protease which is considered to be the
key enzyme responsible for resorption of the bone
matrix. Its ability to degrade type I collagen both within
and outside the helical regions make it a unique
mammalian protease.7-10 Deficiency in or inhibition of
cathepsin K results in impaired activity of bone resorb-
ing osteoclasts. In addition, mice lacking cathepsin K
exhibit osteopetrosis, confirming the importance of this
enzyme for bone remodeling.11,12 The crucial role cathe-
psin K plays in bone matrix resorption is further
demonstrated by a rare human skeletal dysplasia,
pycnodysostosis, which is the consequence of mutations
in the cathepsin K gene.13-15 Thus, the development of
specific inhibitors for this protease may offer an effica-
cious treatment for diseases characterized by excessive
bone loss such as osteoporosis.
In this communication we report on a new class of
highly potent heterocyclic inhibitors of cathepsin K. The
starting point for our investigations of these inhibitors
was the low-nanomolar high-throughput screening (HTS)
hit 1a. The compound is based on a purine template,
which is an unprecedented scaffold for cysteine protease
inhibition. Our first objective in the evaluation of this
screening hit was to establish whether the nitrile group
is the active principle of this inhibitor. As can be seen
from Table 1, replacement of the 2-cyano moiety in 1a
by a methyl or hydroxymethyl substituent, as in 1b and
1c, respectively, is detrimental to inhibition, thus
demonstrating the crucial role of this group. These
initial findings indicate that cathepsin K inhibition by
1a probably involves covalent interaction of the nitrile
carbon with the active site Cys-25 sulfhydryl group.
To convert the HTS hit 1a into a truly viable lead,
replacement of the ribose moiety was an essential
prerequisite. On the basis of molecular modeling studies
and prior experience with other classes of cathepsin K
inhibitors, we hypothesized that the 6-aminocyclohexyl
Figure 1.
ring of 1a was likely to bind in the lypophilic S2 subsite
of the enzyme while the sugar moiety would reach
toward the surface of the enzyme. On the basis of this
assumption, a series of purine-derived compounds 4
(Figure 1) incorporating hydrophilic R groups were
prepared. As a representative example, the synthesis
of 4b is outlined in Scheme 1. Thus, commercially
available 2,6-dichloropurine (2) was reacted with cyclo-
hexylamine to afford the 6-aminocyclohexyl-substituted
purine derivative which was then converted into 3 by
treatment with 1-chloro-2-bromoethane. Subsequent
reaction of 3 with 1-methylpiperazine followed by the
displacement of the 2-chloro substituent with cyanide
furnished the target molecule 4b.
As summarized in Table 2 substitution of the sugar
moiety in 1a by a simple cyclopentane ring led to the
highly potent inhibitor 4a, which also showed significant
selectivity over inhibition of cathepsins L and S. Re-
placement of the ribose moiety by a 4-methylpiperazi-
nylethyl or by the linear (2-methoxyethyl)methylamino-
ethyl moiety likewise resulted in low nanomolar cathep-
sin K inhibitors 4b and 4c, respectively. However,
compared to 4a and 4b, 4c exhibits only moderate
selectivity vs the highly homologous cathepsins L and
S.
A high-resolution X-ray crystal structure of inhibitor
4b bound in the active site of cathepsin K (Figure 2)
confirmed the predicted binding mode for these purine
derivatives. As expected, 4b is found to bind covalently
to the Sγ of active site Cys-25, the resulting thioimidate
moiety being coplanar with the central part of the
inhibitor. The purine scaffold packs against the hydro-
phobic wall of the cathepsin K active site, while the
cyclohexyl group is the P2 moiety and occupies the
* To whom correspondence should be addressed. For E.A.: phone,
+41 61 696 1227; fax, +41 61 696 6071; e-mail, eva.altmann@
pharma.novartis.com. For M.M.: phone, +41 61 696 3962; fax, +41
61 696 6071; e-mail, martin.missbach@pharma.novartis.com.
10.1021/jm0493111 CCC: $27.50 © 2004 American Chemical Society
Published on Web 10/23/2004

5834 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 24
Letters
Scheme 1a
a
Reagents: (a) cyclohexylamine, pentan-1-ol, 70 °C, 4 h, 76%;
Figure 3. Potential expansion of purine derivatives 4 into
(b) 1-chloro-2-bromoethane, K2CO3, DMF, 45 °C, 5 h, 53%; (c)
1-methylpiperazine, 50 °C, 7 h, 81%; (d) NaCN, DMA, 160 °C, 20
h, 26%.
S3.
Scheme 2a
Table 2. Selectivity Data on the Inhibition of Homologous
Cathepsins
a Inhibition of rh cathepsins K, L, and S in a fluorescence-based
assay employing Z-Phe-Arg-AMC (cathepsins K and L) and Z-Leu-
Leu-Arg-AMC (cathepsin S) as synthetic substrates. Data repre-
sent the mean of two experiments performed in duplicate.
a Reagents: (a) 1-chloro-2-bromoethane, K2CO3, KI, TBAB,
acetone, 50 °C, 24 h, 41%; (b) 2-nitrophenol, DEAD, Ph3P, THF,
room temp, 4 h, 63%; (c) H2, PtO2, AcOEt, 3 h, 90%; (d) 2,6-
dichloropurine, pentan-1-ol, 80 °C, 5 h, 56%; (e) bromocyclopen-
tane, K2CO3, DMF, 50 °C, 60%; (f) 1-methylpiperazine, 50 °C, 2
h, 21%; (g) NaCN, DMA, 20 h, 150 °C, 56%.
incorporating an ortho-substituted phenyl ring were
prepared. However, it should be emphasized that mod-
eling also indicated the potential for other orientations
of substituents attached to the 2-position of the phenyl
ring that would not involve interactions with the S3
pocket and therefore would not lead to enhanced
specificity. To distinguish these possibilities, we syn-
thesized and investigated compounds 9.
The preparation of compounds 9, exemplified by 9a,
is outlined in Scheme 2. O-Alkylation of 4-hydroxym-
ethylphenol 5 with 1-chloro-2-bromoethane afforded the
benzyl alcohol 6. This intermediate was reacted with
2-nitrophenol under Mitsunobu conditions, followed by
reduction of the resulting nitro-phenyl-benzyl-ether with
platinum oxide to yield the corresponding chloro-ethoxy-
benzyloxy-phenylamine 7. Elaboration to produce the
9-cyclopentyl-6-phenylaminopurine derivative 8 was
achieved by reaction of 7 with 2,6-dichloropurine and
subsequent alkylation of the 9-position of the purine
nucleus with bromocyclopentane. Treatment of inter-
mediate 8 with 1-methylpiperazine followed by the
displacement of the 2-chloro substituent with cyanide
provided the desired target molecule 9a.
Initially we explored a para-substituted benzyloxy
and a substituted propoxy moiety attached to the
2-position of the aminophenyl group. While the aryl part
of the benzyloxy derivatives 9a and 9b was intended to
interact with the phenyl ring of Tyr67 and thus improve
the inhibitory activity and selectivity, the polar exten-
sion in the para position served to improve aqueous
solubility. In contrast, the polar groups attached to the
purine scaffold by a propoxy linker, as in 9c and 9d,
Figure 2. Purine 4b cocrystallized in the cathepsin K active
site. PDB code is 1U9V. Image was produced using Weblab-
Viewer Lite (Accelrys, Inc.).
lypophilic S2 subsite. No direct interactions with the
binding site are observed for the piperazinyl moiety.
On the basis of the above inhibition and structural
data, we expanded our chemistry efforts to synthesize
compounds that not only extend into the S2 pocket but
in addition would have the potential to interact with
amino acid residues of the S3 subsite. Because the S3
subsites show different characteristics between cathe-
psins K, S, and L, we hypothesized that this strategy
would result in better compound selectivity through
additional interaction points provided by a P3 residue.
Modeling studies suggested two possibilities to reach
the S3 subsite: namely either the addition of a sub-
stituent in the 2-position of the aminocyclohexyl group
(which raises stereochemical issues); or the replacement
of the cyclohexyl moiety by an ortho-substituted phenyl
ring (Figure 3). Aspiring for achiral inhibitors 9a-d

Letters
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 24 5835
Table 3. Selectivity Data on the Inhibition of Homologous
Cathepsins
Figure 4. Difference electron density maps contoured at 2.5σ
of purine derivatives 9c and 9d (green) in complex with
cathepsin K (yellow). PDB codes are 1U9W (9c) and 1U9X (9d).
The picture was produced using the program O.17
purine ring in the active site cleft is slightly different
from that observed in the X-ray structure of the cyclo-
hexyl derivative 4b/enzyme complex. In the case of 4b,
this tilt is necessary to enable the bulkier cyclohexyl
moiety to fit into S2. Again, the nitrile forms a covalent
bond to Cys25 of cathepsin K. The difference electron
density map for the 9d/cathepsin K complex is very
similar to that of 9c except that there is some weak
density present for the propylpiperazine part of the
inhibitor. This density was found to be best represented
when two conformations were modeled. Multiple con-
formations were also modeled for the cyclopentane ring.
In summary, we have discovered novel potent cathep-
sin K inhibitors, starting from the purine-based HTS
hit 1a. A representative inhibitor (9d) of this purine
series proved efficacious in a functional in vitro bone-
resorption assay. Moreover, the high-resolution X-ray
structure of 4b was in good agreement with the pre-
dicted binding mode of these heterocyclic inhibitors and
provides valuable information for the design of different
nonpeptidic inhibitors.
a Inhibition of rh cathepsins K, L, and S in a fluorescence-based
assay employing Z-Phe-Arg-AMC (cath K and L) and Z-Leu-Leu-
Arg-AMC (cath S) as synthetic substrates. Data represent the
mean of two experiments performed in duplicate.
were also meant to interact with the side chain of Asp
61, thereby further improving activity and specificity
for cathepsin K. The results displayed in Table 3 show
that 9a-d are generally low-nanomolar inhibitors of
cathepsin K but do not exhibit improved potency
compared to compounds 4. Inhibitors 9a and 9b having
a 2-benzyloxy substituent at the phenyl moiety are
slightly less potent than 9c and 9d, which incorporate
an o-propoxy linker positioned at the 6-aminophenyl
ring. The observation that compounds 9 do not show a
better overall specificity profile than compounds 4
indicated that the substituents attached to the 2-posi-
tion of the aminophenyl moiety apparently do not extend
into the S3 subsite of cathepsin K. However, it is
noteworthy that the selectivity profile of compounds 9
changes depending on the nature of the spacer used to
direct these potential P3 substituents into the S3 pocket.
While 2-benzyloxy-containing compounds 9a and 9b
have better selectivity vs cathepsin L, compounds with
a 2-propoxy substituent, 9c and 9d, prove to be more
specific against cathepsin S. In the functional in vitro
rabbit bone-resorption assay, 9d inhibits bone resorp-
tion with an IC50 of 176 nM.16
X-ray crystal structures were obtained for complexes
of 9c and 9d and cathepsin K. These structures revealed
that the substituents that were intended to bind in the
S3 subsite of the enzyme do not occupy a fixed position
and face toward the solvent rather than bind in the S3
part of the active site cleft. The difference electron
density map for 9c (Figure 4 left) shows very strong
contributions from the ligand except for the imidazolyl
group and its propyl linker, indicating that this part of
the molecule is highly flexible. The density for the
cyclopentane ring also suggests that this group is in
more than one orientation. The phenyl ring binds in the
S2 subsite as expected. Interestingly the tilt of the
Acknowledgment. The authors thank P. Ramage,
S. Geisse, and T. Inaoka for cathepsin K production,18
S. Niwa for performing the bone-resorption assay, and
I. Sigg for technical assistance.
Supporting Information Available: Description of the
inhibition assays and characterization (1H NMR and HRMS)
of 1a-c, 4a-c, and 9a-d and crystallographic structure
determination of 4b, 9c, and 9d in complex with cathepsin K.
This material is available free of charge via the Internet at
http://pubs.acs.org.
References
(1) Hernandez, A. A.; Roush, W. R. Recent advances in the synthe-
sis, design and selection of cysteine protease inhibitors. Curr
Opin. Chem. Biol. 2002, 6, 45-465.
(2) Leung, D.; Abbenante, G.; Fairlie, D. P. Protease inhibitors:
current status and future prospects. J. Med. Chem. 2000, 43,
305-341.
(3) Yamashita, D. S.; Dodds, R. A. Cathepsin K and the design of
inhibitors of cathepsin K. Curr. Pharm. Des. 2000, 6, 1-24.
(4) Veber, D. F.; Thompson, S. K. The therapeutic potential of
advances in cysteine protease inhibitor design. Curr. Opin. Drug
Discovery Dev. 2000, 3, 362-369.
(5) Marquis, R. W. Inhibition of cysteine proteases. Annu. Rep. Med.
Chem. 2000, 35, 309-320.
(6) Deaton, D. N.; Kumar, S. Cathepsin K inhibitors: their potential
as anti-osteoporosis agents. Prog. Med. Chem. 2004, 42, 245-
375.
(7) Inaoka, T.; Bilbe, G.; Ishibashi, O.; Tezuka, K.; Kumegawa, M.;
Kokubo, T. Molecular cloning of human c-DNA for cathepsin K:
novel cysteine protease predominantely expressed in bone.
Biochem. Biophys. Res. Commun. 1995, 206, 89-96.

5836 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 24
Letters
(8) Drake, F. H.; Dodds, R. A.; James, I. E.; Connor, J. R.; Debouck,
C.; Richardson, S.; Lee-Rykaczewski, E.; Coleman, L.; Rieman,
D.; Bathlow, R.; Hastings, G.; Gowen, M. Cathepsin K but not
cathepsins B, L, or S is abundantly expressed in human
osteoclasts. J. Biol. Chem. 1996, 271, 12511-12516.
(9) Bossard, M. J.; Tomaszek, T. A.; Thompson, S. K.; Amegadzie,
B. Y.; Hanning, C. R.; Jones, C.; Kurdyla, J. T.; McNulty, D. E.;
Drake, F. H.; Gowen, M.; Levy, M. A. Proteolytic activity of
human osteoclast cathepsin K. Expression, purification, activa-
tion and substrate identification. J. Biol. Chem. 1996, 271,
12517-12524.
(10) Garnero, P.; Borel, O.; Byrjalsen, I.; Ferreras, M.; Drake, F. H.;
McQueney, M. S.; Foged, N. T.; Delmas, P. D.; Delaisse, J.-M.
The collagenolytic activity of cathepsin K is unique among
mammalian proteinases. J. Biol. Chem. 1998, 273, 32347-
32352.
(11) Saftig, P.; Hunziker, E.; Wehmeyer, O.; Jones, S.; Boyde, A.;
Rommerskirch, W.; Moritz, J. D.; Schu, P.; von Figura, K.
Impaired osteoclastic bone resorption leads to osteopetrosis in
cathepsin-K-deficient mice. Proc. Natl. Acad. Sci. U.S.A. 1998,
95, 13453-13458.
(12) Gowen, M.; Lazner, F.; Dodds, R.; Kapadia, R.; Feild, J.; Tavaria,
M.; Bertoncello, I.; Drake, F.; Zavarselk, S.; Tellis, I.; Hertzog,
P.; Debouck, C.; Kola, I. Cathepsin K knockout mice develop
osteopetrosis due to a deficit in matrix degradation but not
demineralization. J. Bone Miner. Res. 1999, 14, 1654-1663.
(13) Gelb, B. D.; Edelson, J. G.; Desnick, R. J. Linkage of pycnody-
sostosis to chromosome 1q21 by homozygosity mapping. Nat.
Genet. 1995, 10, 235-237.
(14) Polymeropoulos, M. H.; Ortiz De Luna, R. I.; Ide, S. E.; Torres,
R.; Rubenstein, J.; Francomano, C. A. The gene for pycnodys-
ostosis maps to human chromosome 1cen-q121. Nat. Genet. 1995,
10, 238-239.
(15) Hou, W. S.; Bro¨mme, D.; Zhao, Y.; Mehler, E.; Dushey, C.;
Weinstein, H.; Miranda, C. S.; Fraga, C.; Greig, F.; Carey, J.;
Rimon, D. L.; Desnick, R. J.; Gelb, B. D. Characterization of
novel cathepsin K mutations in the pro and mature polypeptide
regions causing pycnodysostosis. J. Clin. Invest. 1999, 103, 731-
738.
(16) Kakudo, S.; Miyazawa, K.; Mano, H.; Mori, Y.; Yuasa, T.;
Nakamaru, Y.; Shiokawa, M.; Nagahira, K.; Tokunaga, S.;
Hakeda, Y.; Kumegawa, M. Isolation of highly enriched rabbit
osteoclasts from collagen gels: A new assay system for bone
resorbing activity of mature osteoclasts. J. Bone Miner. Metab.
1996, 14, 129-136.
(17) Jones, T. A.; Zou, J.-Y.; Cowan, S. W., Kjeldgaard, M. Improved
methods for building protein models into electron density maps
and the location of errors in these models. Acta Crystallogr. 1991,
A47, 110-119.
(18) Manuscript on cathepsin K production is in preparation.
JM0493111