Identification of a potent and selective non-basic cathepsin K inhibitor

Article abstract of DOI:10.1016/j.bmcl.2005.12.071

Based on our previous study with trifluoroethylamine as a P2-P3 amide isostere of cathepsin K inhibitor, further optimization led to identification of compound 22 (L-873724) as a potent and selective non-basic cathepsin K inhibitor. This compound showed e

Full text of DOI:10.1016/j.bmcl.2005.12.071

Bioorganic & Medicinal Chemistry Letters 16 (2006) 1985–1989
Identification of a potent and selective non-basic
cathepsin K inhibitor
Chun Sing Li,a,* Denis Deschenes,a Sylvie Desmarais,a Jean-Pierre Falgueyret,a
a
b
a
a
´
´ ´
´
Jacques Yves Gauthier, Donald. B. Kimmel, Serge Leger, Frederic Masse,
Mary E. McGrath,c Daniel J. McKay,a M. David Percival,a Denis Riendeau,a
b
a
a
b
´
Sevgi B. Rodan, Michel Therien, Vouy-Linh Truong, Gregg Wesolowski,
Robert Zambonia and W. Cameron Blacka
aMerck Frosst Centre for Therapeutic Research, PO Box 1005, Pointe-Claire-Dorval, Que., Canada H9R 4P8
bDepartment of Bone Biology and Osteoporosis, Merck Research Laboratories, West Point, PA, USA
cCelera Genomics, Inc., 180 Kimball Way, South San Francisco, CA 94080, USA
Received 2 November 2005; revised 16 December 2005; accepted 20 December 2005
Available online 18 January 2006
Abstract—Based on our previous study with trifluoroethylamine as a P2–P3 amide isostere of cathepsin K inhibitor, further optimiza-
tion led to identification of compound 22 (L-873724) as a potent and selective non-basic cathepsin K inhibitor. This compound showed
excellent pharmacokinetics and efficacy in an ovariectomized (OVX) rhesus monkey model. The volumes of distribution close to unity
were consistent with this compound not being lysosomotropic, which is a characteristic of basic cathepsin K inhibitors.
Ó 2006 Elsevier Ltd. All rights reserved.
Cathepsin K (Cat K) is a lysosomal cysteine protease
believed to be the principal enzyme responsible for
degradation of Type I collagen in osteoclastic bone
resorption. Thus, an inhibitor of Cat K may be effec-
tive as an anti-resorptive for treatment of osteoporo-
sis.1 Recently, we have reported trifluoroethylamine
as an excellent amide isostere for the P2–P3 amide
bond of Cat K inhibitors.2 When compared to the
corresponding dipeptide inhibitors, the trifluoroethyl-
amine derivatives provided enhanced potency for
when compound 1 was tested in the whole cell enzyme
occupancy assays,6 we observed much greater potency
against cathepsins B, L, and S than was observed in
the corresponding enzyme assays (Table 1).7 Similar
potency increases in the functional cathepsin K bone
resorption assay8,9 using rabbit osteoclasts were not
observed.5 Therefore, the high selectivity of 1 observed
in enzyme assays is reduced when the compound is
evaluated in these cell-based assays. This reduction
of selectivity could increase the potential for adverse
events in clinic due to inhibition of off-target proteas-
es. Nevertheless, the enhanced potency and selectivity
provided by the trifluoroethylamine amide isostere
offered an opportunity to re-visit SAR at the P3 re-
gion. Our goal was to identify a potent, selective,
and orally bioavailable Cat K inhibitor containing a
non-basic P3 substituent which would be devoid of
the lysosomotropic properties observed with basic
Cat K inhibitors.
cathepsin
K and increased selectivity over other
cathepsins. Compound 1 was identified as a 5 pM
inhibitor of Cat K with a selectivity of >10,000-fold
over other cathepsins. This compound contains a
biphenyl piperazine P3 substituent, which has been
found to be critical for both potency and selectivity
in other series of Cat K inhibitors.3,4 However, we
have also noted that basic inhibitors are often lyso-
somotropic, which results in increased potency in off-
target cell-based enzyme occupancy assays.5 Indeed,
CF3
H
N
CN
Keywords: Bone resorption; Osteoporosis; Cysteine protease; Cathep-
sin K; Cathepsin K inhibitor; Non-basic cathepsin K inhibitor; L-
873724.
N
H
O
*
N
Corresponding author. Tel.: +1 524 428 3217; e-mail: chunsing_li@
merck.com
1
HN
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.12.071

1986
C. S. Li et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1985–1989
Table 1. Functional, whole cell, and enzyme activity of compound 1
ty of this acid could be further improved by trituration
in diethyl ether or by crystallization of the dicyclohexyl-
amine salt. Amide bond formation between the acid
intermediate and aminoacetonitrile then afforded 4.13
IC50 (nM)
Enzyme occupancy assays
Functional Cat K
Rabbit bone res
HepG2
Cat B
HepG2
Cat L
Ramos
Cat S
Early in the SAR study, it was quite clear that simple P3
substituents would not provide sufficient potency and
selectivity (Table 2). Compound 8 lacking a P3 substitu-
ent (R = H) exhibited low potency against all the tested
cathepsins. Introduction of a small alkyl substituent
such as the butyl group in 9 improved the activity, while
the phenyl analog 10 provided a dramatic improvement
in potency against cathepsins K, L, and S. The 4-bromo-
substituted analog 11 showed a further improvement in
potency of Cat K. This beneficial effect of para substitu-
tion on the phenyl group of 10 prompted us to pursue
our SAR studies in this direction.
82
16a
7
12a
Enzyme assays
Cat B
1111
Cat L
47
Cat S
451
Rabbit Cat K
0.01b
a This value likely represents the lower limit of the assay.
b IC50 was determined at a concentration of 10 pM enzyme.
Our initial synthesis of the trifluoroethylamine moiety
was not diastereoselective.2 Subsequently an efficient
diastereoselective method was developed (Scheme 1).
Commercially available (S)-leucinol was converted to
imine 2 and then reacted with alkyl, aryl, or heteroaryl
Our strategy for obtaining a wide range of 4-substitut-
ed analogs was to use rapid analog synthesis based on
a Suzuki coupling reaction with readily available
boronic acids or arylhalides. For these subsequent
studies, the bromide 5 and boronate 6 serve as conve-
nient intermediates for Suzuki coupling reactions to
provide biaryl inhibitors as represented by compound
7. Bromide 5 was obtained from imine 2 as described
in Scheme 1, where the lithium reagent used was de-
rived from 1,4-dibromobenzene (Scheme 2). In the
cases where the substitituent in the Ar group of com-
pound 7 could be further transformed, an appropriate
reaction would be used to complete the conversion
such as methyl sulfide to methyl sulfone by oxidation
reaction.
lithium reagents in
a
diastereoselective manner.10
Deprotection of the silyl group gave the desired (S,S)-
amino alcohol intermediate 3 with diastereoselectivities
ranging from 5:1 to 20:1; the accompanying (R,S)-dia-
stereomer could be partially removed by chromatogra-
phy in many cases. Oxidation with periodic acid in the
presence of catalytic chromium trioxide11 provided the
carboxylic acid intermediate.12 The diastereomeric puri-
c, d
a,b
OTBDMS
OH
F3C
N
H2N
2
The results of representative examples are summarized
in Table 3. The replacement of the bromo substituent
in 11 with a phenyl group (12) resulted in the improve-
ment of Cat K potency and selectivity against Cat L
and Cat S, and thus confirmed the beneficial effect of
para substitution at the proximal phenyl ring. Further
substitutions on the distal phenyl ring were generally
tolerated. However, there was no clear trend for the
SAR on potency and selectivity. Most compounds
showed selectivities of >5000-fold for cathepsin B,
>1000-fold for cathepsin L, and >500-fold for cathepsin
S based on the enzyme assays. Mono-substitution with
halo (13–15), alkyl (18), electron-donating (19 and 21),
CF3
H
N
CF3
e, f
CN
OH
R
N
H
R
N
H
O
4
3
Scheme 1. Reagents and conditions: (a) tert-Butyldimethylsilyl chlo-
ride, Et3N, cat. DMAP, CH2Cl2, room temperature, 2 h; (b) triflu-
oroacetaldehyde methyl hemiacetal, benzene, reflux with a Dean–Stark
trap, overnight, 85% for two steps; (c) R-Li, ꢀ78 °C, 2 h; (d) TBAF,
THF, HOAc, room temperature, 2 h, 15–75% for two steps; (e) H5IO6/
CrO3, Na2HPO4, CH3CN, 0 °C, 1 h; (f) aminoacetonitrile hydrochlo-
ride, HATU, Et3N, DMF, room temperature, 2 h, 10–55% for two
steps.
Table 2. In vitro activity of simple P3 analogs
CF3
H
N
CN
N
R
H
O
a
IC50 (nM)
Compound
R
Cat Kb
Cat B
Cat L
Cat S
Bone resc
8
9
H
Butyl
4123
391
6.7
>10,000
>10,000
>10,000
2993
>10,000
1950
34
6183
1308
36
10
11
Phenyl
4-Bromophenyl
200
111
0.8
13
54
a IC50s are an average of at least two independent titrations.
b Humanized rabbit enzyme (Ref. 3), see Ref. 6 for assay conditions.
c Rabbit bone resorption assay—functional assay, see Ref. 8 for assay conditions.

C. S. Li et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1985–1989
1987
a spacer atom either as an ether linker (25) or a carbonyl
(26) reduced Cat K potency. Among these biphenyl ana-
logs, the most selective compound appeared to be 13
with >6000-fold selectivity for Cat S and over 16,000-
fold for cathepsins B and L. The most potent non-basic
biphenyl analog in the rabbit bone resorption assay is
the methyl sulfone analog 22. The SAR with heterocy-
clic substituents was also studied. Some representative
examples are also shown in Table 3. Pyridine 27, thio-
phene 28, thiazole 29, and indole 32 are very potent
against Cat K, while oxadiazole 30 and pyrimidine 31
show reduced potency. In the functional rabbit bone
resorption assay, the pyridine 27 has excellent potencies,
whereas other heterocyclic analogs are less potent.
However, pyridine 27 shows poorer selectivity than the
biphenyl analog 22 against Cat L (267-fold vs 1320-fold)
and Cat S (206-fold vs 890-fold). The tricyclic P3 ana-
logs 33 and 34 were briefly studied. This modification
resulted in poor pharmacokinetics in rats and therefore
was not pursued further. Nevertheless, the SAR showed
increased selectivity over Cat L and S, but decreased
selectivity over Cat B (33 vs 22 and 34 vs 27). Among
two of the best biphenyl analogs, 13 and 22, preliminary
pharmacokinetic study showed that 22 had superior
CF3
CF3
g
H
H
N
CN
N
CN
N
N
H
H
O
O
Br
Ar
5
7
i
h
CF3
H
N
CN
N
H
O
O
B
6
O
Scheme 2. Reagents and conditions: (g) ArB(OH)2, PdCl2(dppf), aq
Na2CO3, DMF, 80 °C; (h) bis(pinacolato)diboron, PdCl2(dppf),
KOAc; (i) Ar-(Br or I), PdCl2(dppf), aq Na2CO3, DMF, 80 °C.
and electron-withdrawing (20, 22, 23 and 24) substitu-
ents provided compounds with similar potency against
the Cat K enzyme (IC50 0.1–0.9 nM). Disubstitution
was also tolerated, but the substitution patterns were
critical to potency, as the 3,4-difluoro analog 16 has
an IC50 of 1.1 nM and the 3,5-difluoro analog 17 is less
potent with an IC50 of 6 nM against Cat K. Insertion of
Table 3. In vitro activity of representative P3 analogs
CF3
H
N
CN
N
H
O
R
a
IC50 (nM)
Compound
R
Cat Kb
Cat B
1111
Cat L
Cat S
Bone resc
1
12
13
14
15
16
17
18
19
20
21
22f
23
24
25
26
27
28
29
30
31
32
33
34
4-Piperazin-1-yl phenyl
Phenyl
2-Fluorophenyl
60.005d
0.2
0.1
0.5
0.6
1.1
6
47
1473
1670
2792
887
451
303
614
278
449
189
249
282
196
172
153
178
154
265
220
254
62
12e
44
1605
6086
2476
5939
8869
28*
67*
29*
46*
n.d.
27*
n.d.
30*
50*
13e
52
3-Fluorophenyl
4-Fluorophenyl
3,4-Difluorophenyl
3,5-Difluorophenyl
4-Methylphenyl
1001
3002
809
>10,000
1337
1660
0.3
0.1
0.9
0.2
0.2
0.2
0.3
2.3
1.5
0.3
0.2
0.3
2.6
44
4-Methoxyphenyl
4-Cyanophenyl
560
344
7392
1400
4-(Methylthio)phenyl
4-(Methylsulfonyl)phenyl
4-(Aminosulfonyl)phenyl
3-(Methylsulfonyl)phenyl
Phenoxy
524
264
5239
1561
470
6984
1480
771
53*
88*
151*
19
33*
52
>10,000
>10,000
2374
Benzoyl
Pyridin-4-yl
1915
80
3-Thienyl
1,3-Thiazol-2-yl
1155
4818
1475
878
365
240
436
1698
607
653
188
3-Methyl-1,2,4-oxadiazol-5-yl
Pyrimidin-2-yl
4073
5903
314
118*
93*
80*
5e
1857
3047
2422
342
4-(1H-Indol-4-yl)
40-(Methylsulfony)biphenyl-4-yl
4-Pyridin-4-yl-phenyl
0.4
1.2
<0.2
4699
1873
483
14e,*
a IC50s are an average of at least two independent titrations unless specified otherwise.
b Humanized rabbit enzyme (Ref. 3), see Ref. 6 for assay conditions.
c Rabbit bone resorption assay, see Ref. 8 for assay conditions.
d IC50was determined at a concentration of 10 pM enzyme.
e This value likely represents the lower limit of the assay.
f L-873724.
* n = 1. n.d. = not determined.

1988
C. S. Li et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1985–1989
pharmacokinetics to 13 in rats. Therefore, compound
22, now designated as L-873724, was selected for further
profiling.
L-873724 has good pharmacokinetics in three preclinical
animal species (Table 4). The volumes of distribution
close to unity are consistent with this compound not
being lysosomotropic. In the whole cell enzyme occu-
pancy assays (Table 5), L-873724 showed no significant
increases in potency against Cat B, L, and S compared
to isolated enzyme potencies, further evidence support-
ing its non-lysosomotropic character. In addition,
L-873724 had excellent potency in the rabbit bone
resorption assay. As a result, L-873724 is anticipated
to maintain good functional selectivity in vivo.
NS = non significant
SW = Saline washing
Figure 1. Efficacy of compound 22 (L-873724) in the OVX rhesus
monkey model.
The important question of whether a non-basic inhibitor
L-873724 would show efficacy in an animal model of
bone resorption now needed to be addressed. The possi-
bility exists that the lysosomotropic properties associat-
Cat K shows excellent agreement with the X-ray crystal
structure. The CF3 group can clearly be seen directed
out of the active site, toward water. The hydrogen bond
between the NH adjacent to the CF3 group and the
Gly66 carbonyl is well formed and closely resembles
the comparable hydrogen bond found in the P2–P3
amide dipeptide series. The phenyl group adjacent to
the CF3 group is coplanar with the glycine shelf
(Gly66–Gly65) creating a significant hydrophobic inter-
action. Due to the sp3 nature of the trifluoroethylamine
group, the orientations of phenyl ring over the glycine
shelf and the orientation of the hydrogen bond are pre-
ferred. This is in contrast to the P2–P3 amide dipeptide
inhibitors where both of these interactions are subopti-
mal due to the sp2 nature of the linking amide bond.
The sulfonamide substituent at the distal phenyl ring is
largely exposed outside the binding pocket and does
not have significant interaction at the active site. In
addition, one side of the distal phenyl ring is also ex-
posed to solvent and the other side is in the S3 region
of the enzyme. This may help to explain why the substi-
ed with basic Cat
K inhibitors may help the
accumulation of compound in the acidic osteoclast
resorption lacunae where cathespin K digests collagen
and therefore contribute to efficacy. In this context,
one could conclude that a non-basic inhibitor would
show reduced efficacy in vivo. Fortunately, in the ovari-
ectomized (OVX) rhesus monkey model,14 a mean de-
crease of 68% in urinary uNTx/Cr was observed
during the six-day on-treatment phase at 3 mg/kg with
once-daily oral dosing, as indicated in Figure 1. Three
days following the last dose, uNTx/Cr levels returned
to baseline. The response of L-873724 in this model is
comparable to that of the basic P2–P3 amide dipeptide
Cat K inhibitor CRA-013783/L-006235 which has simi-
lar in vitro potency.15
An X-ray crystallographic structure of compound 23,
˚
which is a close analog of 22, with Cat K at 2A resolu-
tion has been obtained (Fig. 2, PDB entry 2FD2). Our
previous model structure2 of trifluoroamine analog into
Table 4. Pharmacokinetics of 22 (1% methocel as dosing vehicle)
Species
po dose
(mg/kg)
% F
Cmax
t1/2
Vd
(lM)
(h)
(L/kg)
Rat
Dog
10
10
5
40
65
24
4.4
18
3
6
2
1.4
0.8
0.5
Rhesus monkey
0.7
Table 5. Functional, whole cell, and enzyme activity of compound 22
IC50 (nM)
Enzyme occupancy assays
Functional Cat K
HepG2 Cat B HepG2 Cat L Ramos Cat S Rabbit bone res
4807
1221
95
13a
Enzyme assays
Cat B
5239
Cat L
264
Cat S
178
Rabbit Cat K
0.8b
Figure 2. X-ray crystal structure of compound 23 bound to Cat K
showing the P2–P3 trifluoroethylamine group and the hydrogen bond
it makes to Gly66 (PDB entry 2FDZ).
a This value likely represents the lower limit of the assay.
b IC50 was determined at a concentration of 10 pM enzyme.

C. S. Li et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1985–1989
1989
Tawa, P.; Wesolowski, G.; Bass, K. E.; Venkatraman, S.;
Percival, M. D. Anal. Biochem. 2004, 335, 218.
7. For enzyme assay conditions, see Ref. 6.
8. Falgueyret, J. P.; Oballa, R. M.; Okamoto, O.; Wesolow-
ski, G.; Aubin, Y.; Rydzewski, R. M.; Prasit, P.;
Riendeau, D.; Rodan, S. B.; Percival, M. D. J. Med.
Chem. 2001, 44, 94.
tution pattern of disubstituted analogs is critical for
potency, 16 versus 17. Overall the binding interactions
with compound 23 are essentially identical to that of
the P2–P3 amide dipeptide inhibitors (pdb entry
1MEM), without the internal destabilization of the li-
gand found in the dipeptides.
9. The rabbit cathepsin K has 94% homology to the human
enzyme. Inaoke, T.; Bilbe, G.; Ishibashi, O.; Tezuka, K.;
Kumegawa, M.; Kokubo, T. Biochem. Biophys. Res.
Commun. 1995, 206, 89.
10. Gosselin, F.; Roy, A.; O’Shea, P. D.; Chen, C.-Y.;
Volante, R. P. Org. Lett. 2004, 6, 641.
In summary, we have identified L-873724 as a non-basic,
potent, and highly selective cathepsin K inhibitor. This
compound is orally bioavailable, non-lysosomotropic,
and efficacious in the rhesus monkey bone resorption
model.
11. Zhao, M.; Li, J.; Song, Z.; Desmond, R.; Tschaen, D. M.;
Grabowski, E. J. J.; Reider, P. J. Tetrahedron Lett. 1998,
39, 5323.
References and notes
12. The presence of water led to a significant amount of
oxidative cleavage of leucine. However, eliminating the 1%
added water as described in Ref. 10 improved yield of the
desired acid product.
13. Diastereomeric purity of final products was usually
greater than 95% de. The other diastereomers were not
detected by 1H NMR in many cases. No loss of
diastereoselectivity was observed with the HATU cou-
pling reaction.
14. See Ref. 3 for assay details. The sequence for the rhesus
monkey enzyme is 98% identical to that of the human with
100% identity within the mature active form of cathepsin K
Guay, J.; Riendeau, D.; Mancini, J. A. Bone 1999, 25, 205.
15. Palmer, J. T.; Bryant, C.; Wang, D.-X.; Davis, D. E.; Setti,
E. L.; Rydzewski, R. M.; Venkatraman, S.; Tian, Z.-Q.;
Burrill, L. C.; Mendonca, R. V.; Springman, E.; McCar-
ter, J.; Chung, T.; Cheung, H.; McGrath, M.; Somoza, J.;
Enriquez, P.; Yu, Z. W.; Strickley, R. M.; Liu, L.; Venuti,
M. C.; Percival, M. D.; Falgueyret, J.-P.; Prasit, P.;
Oballa, R.; Riendeau, D.; Young, R. N.; Wesolowski, G.;
Rodan, S. B.; Johnson, C.; Kimmel, S. B.; Rodan, G. J.
Med. Chem. 2005, 48, 7520.
1. For recent reviews: (a) Grabowska, U. B.; Chambers, T.
J.; Shiroo, M. Curr. Opin. Drug Discov. Devel. 2005, 8,
619; (b) Marquis, R. W. Annu. Rep. Med. Chem. 2004, 39,
79.
2. Black, W. C.; Bayly, C. I.; Davis, D. E.; Desmarais, S.;
´
´
Falgueyret, J.-P.; Leger, S.; Li, C. S.; Masse, F.; McKay,
D. J.; Palmer, J. T.; Percival, M. D.; Robichaud, J.; Tsou,
N.; Zamboni, R. Bioorg. Med. Chem. Lett. 2005, 15, 4741.
3. Robichaud, J.; Oballa, R.; Prasit, P.; Falgueyret, J.-P.;
Percival, M. D.; Wesolowski, G.; Rodan, S. B.; Kimmel,
D.; Johnson, C.; Bryant, C.; Venaktraman, S.; Setti, E.;
Mendoca, R.; Palmer, J. T. J. Med. Chem. 2003, 46, 3709.
4. Robichaud, J.; Bayly, C.; Oballa, R.; Prasit, P.; Mellon,
C.; Falgueyret, J.-P.; Percival, M. D.; Wesolowski, G.;
Rodan, S. B. Bioorg. Med. Chem. Lett. 2004, 14, 4291.
5. Falgueyret, J.-P.; Desmarais, S.; Oballa, R.; Black, W. C.;
´
Cromlish, W.; Khougaz, K.; Lamontagne, S.; Masse, F.;
Riendeau, D.; Toulmond, S.; Percival, M. D. J. Med.
Chem. 2005, 48, 7535.
6. Falgueyret, J. P.; Black, W. C.; Cromlish, W.; Desmarais,
S.; Lamontagne, S.; Mellon, C.; Riendeau, D.; Rodan, S.;