P2-P3 conformationally constrained ketoamide-based inhibitors of cathepsin K

Article abstract of DOI:10.1016/j.bmcl.2005.05.062

An orally bioavailable series of ketoamide-based cathepsin K inhibitors with good pharmacokinetic properties has been identified. Starting from a potent inhibitor endowed with poor drug properties, conformational constraint of the P²-P³ linker and modifications to P1′ elements led to an enhancement in potency, solubility, clearance, and bioavailability. These optimized inhibitors attenuated bone resorption in a rat TPTX hypocalcemic bone resorption model.

Full text of DOI:10.1016/j.bmcl.2005.05.062

Bioorganic & Medicinal Chemistry Letters 15 (2005) 3540–3546
P2–P3 conformationally constrained ketoamide-based
inhibitors of cathepsin K
David G. Barrett,a,ꢀ Virginia M. Boncek,b,ꢁ John G. Catalano,a David N. Deaton,a,*
Anne M. Hassell,c Cynthia H. Jurgensen,b Stacey T. Long,d Robert B. McFadyen,a
Aaron B. Miller,c Larry R. Miller,b J. Alan Payne,b John A. Ray,a Vicente Samano,a
Lisa M. Shewchuk,c Francis X. Tavares,a Kevin J. Wells-Knecht,e,§ Derril H. Willard, Jr.,f,z
Lois L. Wrightg and Hui-Qiang Q. Zhoua
aDepartment of Medicinal Chemistry, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
bDepartment of Molecular Pharmacology, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
cDiscovery Research Computational, Analytical, and Structural Sciences, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
dDepartment of World Wide Physical Properties, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
eDepartment of Research Bioanalysis and Drug Metabolism, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
fDepartment of Gene Expression and Protein Purification, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
gDiscovery Research Biology, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
Received 21 April 2005; revised 16 May 2005; accepted 17 May 2005
Abstract—An orally bioavailable series of ketoamide-based cathepsin K inhibitors with good pharmacokinetic properties has been
identified. Starting0 from a potent inhibitor endowed with poor drug properties, conformational constraint of the P2-P3 linker and
modifications to P1 elements led to an enhancement in potency, solubility, clearance, and bioavailability. These optimized inhibitors
attenuated bone resorption in a rat TPTX hypocalcemic bone resorption model.
Ó 2005 Elsevier Ltd. All rights reserved.
With demographic trends toward increases in the elderly
population in developed nations and concomitant
increases in the number of patients with osteoporosis,
the search for therapeutic interventions to minimize
fractures has intensified. Complementing agents that in-
duce net bone formation, such as parathyroid hormone
(PTH), are anti-resorptive strategies to reduce bone loss
and maintain skeletal integrity.1 Bone resorption results
from secretion of acid and enzymes by specialized cells
called osteoclasts. The acid dissolves the mineral,
hydroxyapatite, while the proteolytic enzymes degrade
the protein components of the bone. The major bone-
resorbing protease in osteoclasts is the C1A family
cysteine protease, cathepsin K.2 It rapidly hydrolyzes
the major component of bone matrix, type I collagen,
in complex with glycosaminoglycans.3 Inhibition of
cathepsin K is therefore expected to decrease bone
resorption. In fact, small molecule inhibitors of cathep-
sin K have been shown to inhibit bone resorption in ani-
mal models of osteoporosis.4 Evidence that cathepsin K
plays a similar role in humans can be found in a rare
autosomal recessive trait known as pycnodysostosis,
which is characterized by short stature, abnormal bone
and tooth development, increased bone mineral density,
and increased bone fragility. Pycnodysostosis results
from deactivating mutations in cathepsin K.5 Thus,
cathepsin K inhibitors could be a novel treatment of
osteoporosis.
Keywords: Cathepsin K; Cysteine protease inhibitors; Ketoamide.
*
Corresponding author. Tel.: +1 919 483 6270; fax: +1 919 315
0430; e-mail: david.n.deaton@gsk.com
ꢀ Present address: Lilly Forschung GmbH, Essener Str. 93, 22419
Hamburg, Germany.
ꢁ Present address: 3040 Cornwallis Road, 111 Hermann Building,
Research Triangle Park, NC 27709, USA.
§ Present address: Schering-Plough Research Institute, 2015 Galloping
Hill, Kenilworth, NJ 07033, USA.
As part of a larger program to develop novel cathepsin
K inhibitors, researchers from these laboratories have
z Deceased.
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.05.062

D. G. Barrett et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3540–3546
3541
recently reported the discovery of ketoamide cathepsin
inhibitors (IC50 = 0.79 nM) and (IC50 =
functional groups of the resulting carbamates were
hydrolyzed to yield acids 8a–f, which were coupled with
cyanomethyltriphenylphosphonium ylide to afford the
phosphoranes 9a–f. Then, oxidative cleavage of the
phosphorus–carbon double bond with ozone generated
an acyl nitrile. In situ displacement of the cyanide
by (R)-a-methylbenzylamine, 5-amino-1-methylpyraz-
ole, (S)-3-amino-2-piperidinone, or 3-amino-2-oxazolid-
inone afforded the desired ketoamides 6a–f, 6l–m, and
6o.
K
1
2
0.77 nM).6–8 To optimize these leads further, which
suffered from poor aqueous solubility and metabolic
instability, a molecular modeling-based strategy was de-
vised to introduce conformational const0raints into the
P2–P3 linker of analog 1. In addition, P1 moieties with
better drug properties, such as the pyrazole group from
compound 2, were introduced to enhance further the
drug properties in a constrained inhibitor. With these
goals in mind, the synthesis of inhibitors, exemplified
by the generic structure 3 was undertaken.
The alcohols 4a–j were prepared, as depicted in Schemes 3
and 4. In the first method, 2-methyl-5-aryl-1,3,4-oxadiaz-
oles 10a–c, available from refluxing the commercially
available hydrazides with triethylorthoacetate in xylenes,
were deprotonated and then reacted with aldehydes 11a–c
to afford the racemic alcohols, which were then resolved
by chiral chromatography (SFC, MeOH in CO2, Chiral-
cel AD) to give alcohols 4a–f.
Two general routes were utilized to produce the a-keto-
amides. As shown in Scheme 1, one procedure involved
joining the chloroformates or para-nitrophenylcarbon-
ates, formed from the alcohols 4f–j, to the b-amino-a-
hydroxyamides 5a–e. The resulting carbamates were
subsequently oxidized to the desired ketoamides 6f–k,
6n, and 6p–r.
The other protocol utilized the Wasserman acyl cyano-
phosphorane oxidative cleavage and amine coupling
procedure to generate the ketoamide moiety.9 As dis-
played in Scheme 2, alcohols 4a–f were coupled to a
known ester-containing isocyanate 7.6 The ester
Scheme 3. Reagents and conditions: (a) LiTMP or n-BuLi, THF,
À78 °C; aldehyde, À78 °C to rt, 11–64%.
Scheme 1. (a) Reagents and conditions: para-Nitrophenyl chlorofor-
mate, pyridine, CH2Cl2, or 1.93 M COCl2 in PhMe, 46–99%; 5a–e,
i-Pr2NEt, DMF, 54–83%; (b) Dess-Martin periodinane, CH2Cl2, 71–
83%.
Scheme 4. Reagents and conditions: (a) H2SO4, NaNO2, H2O, 0 °C,
65%; (b) LiAlH4, THF, 0 °C to rt, 63%; (c) TsCl, pyridine, 0 °C to rt,
85%; (d) NaOH, MeOH, H2O, 0 °C, 83%; (e) 14, NEt3, i-PrOH, 85 °C,
sealed tube, 53–84%; (f) 14, EtOH, 85 °C, sealed tube, 98%; (g) H2/Pd-
C, MeOH; (h) 4-CF3C6H4N@C@O, THF, 47% over 2 steps; (i)
DTAD, Ph3P, THF, 84%.
Scheme 2. Reagents and conditions: (a) PhMe, 85 °C, sealed tube, 77–
99%; (b) LiOHÆH2O, THF, H2O; 1 N HCl; (c) Ph3P@CCN, DMAP,
EDC, CH2Cl2, 29–76% over two reactions; (d) O3, CH2Cl2, À78 °C;
N2; amine, À78 °C to rt; AgNO3, THF, H2O, 11–48%.

3542
D. G. Barrett et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3540–3546
Alcohols 4g–j were prepared via opening of the epoxide
14, as illustrated in Scheme 4. Starting from (S)-L-tert-
leucine 12, diazotization of the amine and trapping with
water provide the a-hydroxyacid, which was subse-
quently reduced to afford the diol 13. Then, selective
conversion of primary alcohol to tosylate, followed by
cyclization with base, gave the epoxide 14. Heating of
pyrazole 15a, pyrazole 15b, or imidazole 15c with the
epoxide 14 in a sealed tube yielded the desired alcohols
4g–i. To synthesize alcohol 4j, the protected ethanol-
amine 16 was alkylated with epoxide 14, and then the
benzyl group was removed by hydrogenolysis to give
the aminodiol 17. Selective reaction of the amine 17 with
4-trifluoromethylphenyl isocyanate provided urea. Sub-
sequent cyclization under Mitsunobu conditions affor-
ded the alcohol 4j.
tion with amines 22a–e and cleavage of the ethylcarbon-
ate afforded the a-hydroxyamides. Then, removal of the
carbamate gave the desired amines 5a–e.
As shown in Table 1, the oxadiazole constrained ketoa-
mide 6a is a picomolar inhibitor of cathepsin K
(IC50 = 0.24 nM) and is 3-fold more potent than inhibi-
tor 1 (IC50 = 0.79 nM). Presumably, this slight increase
in potency results from a smaller entropic loss due to
preconstraining the P2–P3 tether, which more than com-
pensates for a higher enthalpic penalty of desolvating an
oxadiazole ring as compared to an aliphatic hydrocar-
bon chain. This desolvation liability is offset by in-
creased hydrophilicity of 6a, resulting in an increased
aqueous solubility (solubility in phosphate buffer at
pH 6.8, 6a = 0.010 mg/mL vs 1 < 0.001 mg/mL). As de-
picted in Table 2, analog 6a is moderately selective
against cathepsins B (B/K = 50) and L (L/K = 42), with
lower selectivity for cathepsins S(S/K = 3) and V (V/K
= 7). None of the analogs in Table 2 were potent inhib-
itors of cathepsin H. This conformational constraint did
not enhance the selectivity profile of 6a versus 1. Like
other ketoamides from these laboratories, the (S)-epi-
mer 6b (IC50 = 460 nM) is at least 1000-fold less active
than the (R)-analog 6a (6b is contaminated with
ꢀ0.1% of 6a).6,7
The heterocycles 15a–c were prepared, as shown in
Scheme 5. Condensation of acetophenone 18 with ethyl
formate gave the b-ketoaldehyde, which was treated
with hydrazine to afford pyrazole 15a. Phenylacetic acid
19 was heated with the Vilsmeier reagent to give c-di-
methylamino-2-propenal that was also condensed with
hydrazine to provide pyrazole 15b. Finally, heating
a-bromoketone 20 with formamide afforded imidazole
15c.
Amines 5a–e were prepared from a known acid 21, as
shown in Scheme 6.6 First, the acid 21 was activated
by ethyl chloroformate, with concomitant protection
of the alcohol as ethyl carbonate. Subsequent condensa-
To determine if the tert-butyl moiety was still the opti-
mal P2 substituent in this constrained ketoamide series,
the iso-propyl analog 6c and the tert-amyl analog 6d
were prepared. Both 6c (IC50 = 0.83 nM) and 6d
(IC50 = 1.5 nM) were weaker cathepsin K inhibitors
than 6a. In fact, the bulkier tert-amyl derivative 6d
was apparently a more potent cathepsin Sinhibitor
(IC50 = 0.72 nM) than a cathepsin K inhibitor. Given
these results, the P2 group was defined as tert-butyl for
subsequent SAR development.
Since unsubstituted phenyl groups can be prone to
phase I metabolism, fluoro- or trifluoromethyl substitu-
ents were introduced into the P3 phenyl group to steri-
cally and electronically attenuate potential oxidation
by Cyp450 enzymes. The para-fluoro analog 6e
(IC50 = 0.37 nM) and the para-trifluoromethyl analog
6f (IC50 = 0.41 nM) were equipotent to the parent ana-
log 6a, and these substituents were incorporated into
the further design of inhibitors.
Scheme 5. Reagents and conditions: (a) NaH, EtOCHO, THF, 0 °C to
rt; NH2NH2, MeOH, 80%; (b) POCl3, DMF, 0 °C to rt; 19, DMF, rt to
70 °C; K2CO3, 0 °C; NaOH, 50 °C; NH2NH2, MeOH, 80%; (c)
NH2CHO, 175 °C; 1N HCl, reflux, 67%.
0
The P1 aryl group in analogs 6a–f did not aid in aqueous
solubility and was also0 a potential metabolic liability.
Building on previous P1 SAR development,8 the (R)-a-
phenylethylamine was replaced by the potency- and sol-
ubility-enhancing pyrazole, giving inhibitor 6g. Pyrazole
6g (IC50 = 0.029 nM) was ꢀ10-fold more potent and was
slightly more selective than 6f. Moreover, as shown in
Table 3, this pyrazole exhibited excellent permeability
in an in vitro Madin–Darby canine kidney (MDCK) per-
meation assay (PAPP = 290 nm/s)10, as well as improved
solubility in fasted state-simulated intestinal fluid
(FS-SIF = 0.061 mg/mL)11,12 at pH 6.8, relative to inhib-
itor 6a (FS-SIF = 0.019 mg/mL). With these encouraging
predictors of good absorption, the pharmacokinetics of
analog 6g were profiled in male Han Wistar rats. It
Scheme 6. Reagents and conditions: (a) N-methyl piperidine, CH2Cl2,
À40 °C; EtOCOCl, À40 to 5 °C; 22a–e, 5 °C to rt; K2CO3, H2O,
MeOH, 39–77%; (b) HCl, dioxane, rt.

D. G. Barrett et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3540–3546
Table 1. Inhibition of human cathepsin K
3543
R1
R2
a
IC50 (nM)
#
Ar
X
6a
0.24
6b
6c
6d
6e
6f
460
0.83
1.5
0.37
0.41
0.029
0.072
0.032
0.026
0.34
4.9
6g
6h
6i
6j
6k
6l
6m
6n
6o
6p
6q
6r
0.65
2.8
0.13
8.1
5.8
0.42
a Inhibition of recombinant human cathepsin K activity in a fluorescence assay using 10 lM Cbz-Phe-Arg-AMC as substrate in 100 mM NaOAc,
10 mM DTT, and 120 mM NaCl, pH 5.5. The IC50 values are means of two or three inhibition assays, individual data points in each experiment
were within a 3-fold range of each other.

3544
D. G. Barrett et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3540–3546
Table 2. Cathepsin B, L, S, and V inhibition and selectivity
#
Cat K IC50 (nM)
Cat B IC50 (nM)a
Cat L IC50 (nM)b
Cat SIC (nM)c
Cat V IC50 (nM)d
50
6a
6c
6d
6e
6f
0.24
0.83
1.5
12
31
10
17
0.75
3.7
1.6
5.6
56
17
33
0.72
0.53
1.6
2.8
0.37
0.41
0.029
0.072
0.032
0.026
0.34
4.9
9.8
32
34
190
69
5.0
1.2
6g
6h
6i
45
20
0.52
0.21
0.078
0.16
0.54
110
0.81
2.7
8.7
36
0.17
0.43
0.13
1.7
10
27
6j
10
37
6k
6l
60
3500
6.2
18
3800
15
230
1.3
14
6m
6n
6o
6p
6q
6r
0.65
2.8
130
160
240
1000
69
0.13
8.1
5.2
107
870
23
1.2
3.7
7.1
4.5
5.8
0.42
19
36
0.12
0.89
a Inhibition of recombinant human cathepsin B activity in a fluorescence assay using 10 lM Cbz-Phe-Arg-AMC as substrate in 100 mM NaOAc,
10 mM DTT, and 120 mM NaCl, pH 5.5. The IC50 values are means of two or three inhibition assays, individual data points in each experiment
within a 2-fold range of each other.
b Inhibition of recombinant human cathepsin L activity in a fluorescence assay using 5 lM Cbz-Phe-Arg-AMC as substrate in 100 mM NaOAc,
10 mM DTT, and 120 mM NaCl, pH 5.5.
c Inhibition of recombinant human cathepsin Sactivity in a fluorescence assay using 10 lM Cbz-Val-Val-Arg-AMC as substrate in 100 mM NaOAc,
10 mM DTT, and 120 mM NaCl, pH 5.5.
d Inhibition of recombinant human cathepsin V activity in a fluorescence assay using 2 lM Cbz-Phe-Arg-AMC as substrate in 100 mM NaOAc,
10 mM DTT, and 120 mM NaCl, pH 5.5.
Table 3. Pharmacokinetics of combination analogs
Sol. FS-SIFa (mg/mL)
Clc (mL/min/kg)
Fe (%)
b
t1/2 (min)
d
VSS (mL/kg)
#
clogP
MDCK PAPP (nm/s)
6g
6h
6i
3.34
5.55
5.34
5.08
4.80
2.59
3.10
3.53
4.42
5.96
5.33
5.99
290
91
0.061
0.11
530
320
340
460
340
150
140
230
410
210
280
400
11
9.6
9.1
15
55
33
35
11
11
41
15
3.2
1900
1600
2100
2400
5000
3600
1400
1400
3100
1800
1800
1700
50
42
45
55
27
88
17
46
79
34
43
40
120
160
—
0.097
0.13
6j
6k
6l
0.043
0.058
0.62
350
—
6m
6n
6o
6p
6q
6r
280
15
0.49
0.17
180
76
0.38
0.048
0.064
180
a FS-SIF is the equilibrium solubility in a fasted state-simulated intestinal fluid at pH 6.8. The values are means of two measurements.
b t1/2 is the i.v. terminal half-life dosed as a solution in male Han Wistar rats. All in vivo pharmacokinetic values are means of two experiments.
c Cl is the total clearance.
d VSS is the steady-state volume of distribution.
e F is the oral bioavailability.
exhibited a long terminal half-life (t1/2 = 530 min), with a
low clearance (Cl = 11 mL/min/kg) and medium volume
of distribution (VSS = 1900 mL/kg). In confirmation of
the in vitro permeability and solubility measurements,
6g was orally bioavailable (F = 50%).
pharmacokinetic properties in rats (t1/2 = 320–460 min,
Cl = 9.1–15 mL/min/kg, and F = 42–55%). In contrast,
the cyclic urea 6k exhibited a clearance (Cl = 55 mL/
min/kg), approaching hepatic blood flow for the rat
and a lower oral bioavailability (F = 27%), suggesting
extensive metabolism.
To investigate this promising series further, the oxadi-
azole ring constraint was replaced with other five-
membered ring heterocycles. The pyrazoles 6h
(IC50 = 0.072 nM), and 6i (IC50 = 0.032 nM) and the
imidazole 6j (IC50 = 0.026 nM) were essentially equipo-
tent inhibitors of cathepsin K, whereas the cyclic urea
6k (IC50 = 0.34 nM) was less active. As shown in Table
3, the heteroaryl constraints also demonstrated excellent
0
Since replacement of the P1 a-methylbenzyl with pyra-
zole resulted in enhanced potency and improved0 drug
properties, the effect of other heterocycles at P1 was
investigated. As expected based on previous SAR,8 the
N-methyl pyrazole 6l (IC50 = 4.9 nM) was less potent,
but exhibited a better selectivity profile than 6e. It was
highly orally bioavailable (F = 88%), but exhibited a

D. G. Barrett et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3540–3546
3545
higher clearance (Cl = 33 mL/min/kg), relative to the
parent pyrazole, possibly due to N-demethylation of
the pyrazole. This clearance resulted in a diminished ter-
minal half-life (t1/2 = 150 min). Replacement of the pyr-
azole with further heterocycles lacking a hydrogen bond
donor resulted in decreased potency. For example,
thiazole 6q (IC50 = 5.8 nM) was ꢀ200-fold less potent
than pyrazole 6j and homologated isoxazole 6r
(IC50 = 0.42 nM) was ꢀ6-fold less potent than pyrazole
6h. The pharmacokinetics of these still potent cathepsin
K inhibitors were determined in rats. Thiazole 6q
(Cl = 15 mL/min/kg, F = 43%) showed a profile reminis-
cent of pyrazole 6j. The clearance of isoxazole 6r was
extremely low (Cl = 3.2 mL/min/kg) and the oral bio-
availability (F = 40%) was good.
significantly less potent inhibitors of rat cathepsin K
than human cathepsin K (18- to 62-fold). These results
can be explained by differences in the active site of the
rat and human cathepsin K (61Asp ! Tyr in S3 and
133Ala ! Ser in S2).2 As expected based on rat cathepsin
K IC50s, ketoamides 6h–j were significantly less potent
in attenuating deoxypyridinoline crosslink release in an
ex vivo rat calvarial resorption assay14,15 (IC50 = 2900–
6000 nM) than in a human osteoclast resorption assay.
Despite this species-specific loss in potency, inhibitors
6h and j were tested in a thyroid–parathyroidectomized
hypocalcemic rat assay (TPTX).16 In this model, remov-
al of the parathyroid gland eliminated parathyroid hor-
mone (PTH) secretion, altering the maintenance of
calcium homeostasis via PTH-stimulated bone resorp-
tion. Rapid depletion of calcium from serum occurs
when dietary calcium supplementation is prevented.
Infusion of hPTH (1–34) at 30,000 ng/kg/h then induces
a rapid increase in serum calcium levels. As shown in
Table 4, oral administration of cathepsin K inhibitors
6h or j dose-dependently attenuated the PTH-stimulated
calcium increase in this model by blocking collagen ma-
trix degradation. These results reached statistical signif-
icance at a dose of 100 mg/kg p.o. Because cathepsin K
inhibitors do not affect resorption lacunae acidification,
some calcium is released by hydroxyapatite dissolution
before the collagen fibrils are exposed.17 Osteoblastic
MMP-13 and MMP-14 may also partially compensate
for any lack of cathepsin K activity in matrix degrada-
tion. Despite these counteractions, treatment of the rats
with cathepsin K inhibitors clearly inhibited calcium re-
lease from bone.
An X-ray co-crystal structure of ketoamide 2 with
cathepsin K showed that the pyrazole NH forms two
hydrogen bonds with the protein. One with the indole
NH of 184Trp and a second interaction with 19Gln
through bridging water molecules.8 From a modeling
point of view, it has been suggested that the 2-piperi-
done carbonyl of ketoamide 6m (IC50 = 0.65 nM) and
the oxazolidinone carbonyl of ketoamide 6o
(IC50 = 0.13 nM) could serve as hydrogen bond accep-
tors in interactions with the 184Trp indole. However,
these analogs were less potent than the pyrazole 6g. Pre-
sumably, the proposed hydrogen bond interaction could
not compensate for the loss of both the hydrogen bonds
of the pyrazole. The clearance of 2-piperidone 6m
(Cl = 35 mL/min/kg) was approximately half of hepatic
blood flow in the rat, and its oral bioavailability
(F = 17%) was lower than that of pyrazole 6g. In con-
trast, oxadiazolidinone 6o exhibited a low clearance
(Cl = 11 mL/min/kg) and an excellent oral bioavailabil-
ity (F = 79%), despite moderate in vitro permeability
(MDCK PAPP = 15 nm/s).
An X-ray co-crystal structure of cathepsin K with
inhibitor 6o was subsequently solved, providing further
insight into the binding mode of these constrained
ketoamide inhibitors. The active site is shown in
Figure 1. These reversible, time-dependent, tight binding
inhibitors form a covalent hemithioketal intermediate
between the a-keto moiety of the inhibitor and the active
site 25Cys of the enzyme. In contrast to aldehyde and
ketone inhibitors bound to cathepsin K, the hemithiok-
etal hydroxyl does not occupy the oxyanion hole, rather
it is stabilized by hydrogen bonds to the catalytic
In an attempt to improve further the aqueous solubility
of the series, the two pyridine derivatives 6n
(IC50 = 2.8 nM) and 6p (IC50 = 8.1 nM) were prepared.
Both 6n and 6p were significantly less potent cathepsin
K inhibitors than 6g or 6h, respectively, as expected
from the loss of a hydrogen bond donor, and were also
less selective versus other cathepsins. Furthermore, the
improved solubilities did not lead to a better oral expo-
sure for either 6n (F = 46%) or 6p (F = 34%).
Table 4. Rat TPTX hypocalcemic resorption assay
t (h)
Serum calcium
Analogs 6g–j were tested in an ex vivo human osteoclast
resorption assay.13 Human peripheral blood monocyte
cells were placed in media on bone chips and differenti-
ated into osteoclast-like cells with receptor activator of
nuclear factor j B ligand (RANK-L) and macrophage
colony stimulating factor (m-CSF). Then, bone resorp-
tion was measured by calcium release into media in
the presence or absence of a cathepsin K inhibitor. All
four compounds inhibited bone resorption at nanomo-
lar concentrations in this assay, with 6h (IC50 = 120 nM)
and 6i (IC50 = 190 nM) being more potent than 6g
(IC50 = 690 nM) and 6j (IC50 = 570 nM).
Veh.
PTH
6h
6j
30a
100a
30a
100a
0
2
4
6
Avg.
SD
1.07
0.07
1.07
0.06
1.05
0.10
1.07
0.04
1.12
0.08
1.14
0.09
Avg.
SD
0.98
0.03
1.19
0.08
1.14
0.13
1.10
0.03
1.21
0.09
1.16
0.06
Avg.
SD
0.94
0.06
1.41
0.06
1.30
0.07
1.21c
0.02
1.42
0.05
1.31b
0.07
Avg.
SD
0.90
0.04
1.64
0.04
1.57
0.20
1.34c
0.07
1.63
0.04
1.51b
0.07
a Dose, mg/kg.
b p < 0.05 compared to PTH control.
c p < 0.01 compared to PTH control.
Analogs 6g (rat IC50 = 1.8 nM), 6h (rat IC50 = 1.3 nM),
6i (rat IC50 = 2.0 nM), and 6j (rat IC50 = 1.0 nM) are

3546
D. G. Barrett et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3540–3546
lar inhibitors of human cathepsin K attenuate bone
resorption in ex vivo and in vivo animal models of bone
resorption. Additional efforts to improve the selectivity
of these inhibitors versus other cathepsin endoproteases
are warranted.
Acknowledgments
The authors wish to thank Melissa A. Gomez and Man-
on S. Villeneuve for chiral separations of alcohols 4a–f.
References and notes
1. Einhorn, T. A. In Osteoporosis; Marcus, R., Feldman, D.,
Kelsey, J., Eds.; Academic Press: San Diego, California,
1996, p 3.
2. Deaton, D. N.; Kumar, S. Prog. Med. Chem. 2004, 42,
245.
3. Li, Z.; Hou, W.-S.; Escalante-Torres, C. R.; Gelb, B. D.;
Bromme, D. J. Biol. Chem. 2002, 277, 28669.
4. Stroup, G. B.; Lark, M. W.; Veber, D. F.; Bhattacharyya,
A.; Blake, S.; Dare, L. C.; Erhard, K. F.; Hoffman, S. J.;
James, I. E.; Marquis, R. W.; Ru, Y.; Vasko-Moser, J. A.;
Smith, B. R.; Tomaszek, T.; Gowen, M. J. Bone Miner.
Res. 2001, 16, 1739.
Figure 1. Active site of the X-ray co-crystal structure of compound 6o
complexed with cathepsin K. The cathepsin K carbons are colored
magenta, with inhibitor 6o carbons colored cyan. The semi-transparent
white surface represents the molecular surface, while hydrogen bonds
are depicted as yellow dashed lines. The coordinates have been
deposited in the Brookhaven Protein Data Bank, Accession No. 1YT7.
This figure was generated using PYMOL version 0.97 (Delano
Scientific, www.pymol.org).
histidine (162His) and the backbone carbonyl of 161Asn,
consistent with other ketoamide co-crystal structures of
this group.8,18 The oxyanion hole is occupied by the car-
bonyl of the amide, which is stabilized by hydrogen
bonds to the side chain NH of 19Gln and the backbone
NH of 25Cys. The peptide backbone recognition site of
the enzyme stabilizes the inhibitor with two additional
hydrogen bonds between 161Asn and the carbamate
NH and 66Gly and the carbamate carbonyl oxygen.
5. Gelb, B. D.; Shi, G.-P.; Chapman, H. A.; Desnick, R. J.
Science 1996, 273, 1236.
6. Catalano, J. G.; Deaton, D. N.; Long, S. T.; McFadyen,
R. B.; Miller, L. R.; Payne, J. A.; Wells-Knecht, K. J.;
Wright, L. L. Bioorg. Med. Chem. Lett. 2004, 14, 719.
7. Barrett, D. G.; Catalano, J. G.; Deaton, D. N.; Long, S.
T.; Miller, L. R.; Tavares, F. X.; Wells-Knecht, K. J.;
Wright, L. L.; Zhou, H.-Q. Q. Bioorg. Med. Chem. Lett.
2004, 14, 2543.
8. Tavares, F. X.; Boncek, V.; Deaton, D. N.; Hassell, A. M.;
Long, S. T.; Miller, A. B.; Payne, A. A.; Miller, L. R.;
Shewchuk, L. M.; Wells-Knecht, K.; Willard, D. H., Jr.;
Wright, L. L.; Zhou, H.-Q. J. Med. Chem. 2004, 47, 588.
9. Wasserman, H. H.; Petersen, A. K. Tetrahedron Lett.
1997, 38, 953.
10. Irvine, J. D.; Takahashi, L.; Lockhart, K.; Cheong, J.;
Tolan, J. W.; Selick, H. E.; Grove, J. R. J. Pharm. Sci.
1999, 88, 28.
11. Dressman, J. B.; Amidon, G. L.; Reppas, C.; Shah, V. P.
Pharm. Res. 1998, 15, 11.
12. Kostewicz, E. S.; Brauns, U.; Becker, R.; Dressman, J. B.
Pharm. Res. 2002, 19, 345.
13. Nicholson, G. C.; Malakellis, M.; Collier, F. M.; Cam-
eron, P. U.; Holloway, W. R.; Gough, T. J.; Gregorio-
King, C.; Kirkland, M. A.; Myers, D. E. Clin. Sci. 2000,
99, 133.
Apart from these hydrogen bond stabilizing interactions
0
with the protein, the P1 oxazolidinone carbonyl oxygen
forms a hydrogen bond with the indole NH of 184Trp, as
predicted by modeling. Moreover, the norleucine-
derived P1 group of the inhibitor lies along the S1 wall
formed from 23Gly, 24Ser, 64Gly, and 65Gly, with one
face of the n-butyl group forming van der Waals interac-
tions with the protease, while its terminal carbon re-
mains solvent exposed. Furthermore, the P2 tert-butyl
substituent forms significant lipophilic interactions with
the deep S2 pocket composed of 67Tyr, 68Met, 134Ala,
163Ala, and 209Leu. Finally, the P3 para-trifluoromethyl-
phenyloxadiazole moiety interacts with the active site
trough, as well as pointing into the S3 subsite formed
by 60Asn, 61Asp, 65Gly, 66Gly, and 67Tyr.
14. Hahn, T. J.; Westbrook, S. L.; Halstead, L. R. Endocri-
nology 1984, 114, 1864.
15. Conaway, H. H.; Grigorie, D.; Lerner, U. H. J. Endocri-
nol. 1997, 155, 513.
16. Thompson, D. D.; Seedor, J. G.; Fisher, J. E.; Rosenblatt,
M.; Rodan, G. A. Proc. Natl. Acad. Sci. U.S.A. 1988, 85,
5673.
17. Kiviranta, R.; Morko, J.; Alatalo, S. L.; NicAmhlaoibh,
R.; Risteli, J.; Laitala-Leinonen, T.; Vuorio, E. Bone 2005,
36, 159.
In summary, this report details the optimization of
ketoamide 1 to potent, orally bioavailable cathepsin K
inhibitors, such as 6h and j. Conformational constraint
was utilized to reduce entropic losses upon inhibitor
binding. These efforts led to increased cathepsin K
inhibitory activity. Furthermore, the heterocyclic nature
of the constraints led to a reduction in hydrophobicity
and an improvement in aqueous solubility. Combining
18. Barrett, D. G.; Catalano, J. G.; Deaton, D. N.; Hassell, A.
M.; Long, S. T.; Miller, A. B.; Miller, L. R.; Shewchuk, L.
M.; Wells-Knecht, K. J.; Willard, D. H., Jr.; Wright, L. L.
Bioorg. Med. Chem. Lett. 2004, 14, 4897.
0
these P2–P3 changes with more drug-like P1 moieties
afforded potent, orally bioavailable cathepsin K inhibi-
tors with good pharmacokinetics in rats. These picomo-