J. G. Catalano et al. / Bioorg. Med. Chem. Lett. 14 (2004) 275–278
277
1
.
Scheme 4. (a) H2, RhCl3 H2O, Aliquat 336, ClCH2CH2Cl, H2O,
(99%).
Scheme 6. (a) CF3CO2Et, NEt3, MeOH; (b) NaH, THF; MeI, 0 ꢀC
to rt.
to have good potency. Cathepsin K inhibitory activity
of phenylglycinal 2w is 51 nM. The phenylalaninal ana-
logue 2y is the least active member of this series
(IC50=110 nM). Activitythen increases with the phen-
ethyl derivative 2aa (IC50=30 nM) and the phenpropyl
analogue 2ac (IC50=15 nM). The cyclohexyl derivatives
are also potent inhibitors of cathepsin K. The cyclo-
hexyl glycine derivative 2x is roughlyas potent as the
propyl analogue 2f. It is interesting to note that the
branching at the beta carbon does not lead to the same
decrease in activityobserved in 2d. The methyl-, ethyl-
and propylcyclohexyl analogues 2z, 2ab, and 2ad,
respectively, are potent inhibitors with IC50s of 36, 26,
and 16 nM. These results appear to show that no
favorable p–p interactions have been picked up bypre-
sence of an aryl group in these analogues.
Scheme 5. (a) iPrOCOCl, NEt3, THF, À10 ꢀC; NaBH4, H2O, 0 ꢀC,
(30%); (b) NEt2, THF; (c) MeOCOCl, iPr2NEt, THF, (43%); (d)
LiOH, THF, H2O.
With the aim of preventing hemiaminal formation, a
series of methylated trifluoroacetamides was synthe-
sized. As shown in Scheme 6, the amines 11 (n=2–4)
were converted into the trifluoroacetamides 5ak (n=2–
4).15 The trifluoroacetamides were then deprotonated
with sodium hydride, followed by alkylation with
methyl iodide to give the tertiary amides 5ai, 5aj, and
5al.16 Moffat oxidation of the alcohols 5 as in Scheme 1
afforded the aldehydes 2.
The electronics of the S1 subsite were probed bysub-
stituting heteroatoms into the P1 group. Incorporation
of polar heteroatoms into the side chain close to the
amino acid backbone leads to decreased cathepsin K
inhibitoryactivity. Ether analogues 2s and 2u are over
10-fold less potent than the norleucinal 2l. Amine deri-
vative 2ag (IC50=370 nM) is also less potent than 2l
(IC50=51 nM). Sulfur is an exception to this trend.
With available d orbitals and more diffuse electron lone
pairs, the ‘soft’ sulfur is closer to carbon than it is to
oxygen or nitrogen. The thioethers 2t and 2v are equi-
potent with the butyl analogue 2l. In contrast to P1 side
chains with ‘hard’ heteroatoms close to the amino acid
backbone like 2ag, the ornithine 2aj and lysine deriva-
tives 2ah, 2ak, and 2al are among the more potent
aldehydes surveyed. Thus, inhibitor solubility could be
enhanced bythese P 1 groups.
With an X-raycrystal structure of cathepsin K unavail-
able at the start of this work, a traditional structure–
activityrelationship studywas employed varying steric
and electronic properties of the P1 moiety. As shown in
Table 1, an n-alkyl side chain of 1–6 carbons leads to an
increase in cathepsin K activityover the glycine deriva-
tive 2a. The alanine derivative 2b is roughly30-fold
more potent than the glycine analogue 2a. The longer
linear P1 side chain aldehydes like 2c, 2f, and 2l have
equivalent potencyand are slightlymore potent than
the alanine derivative 2b. Additional substitution of the
a-carbon of the side chain is detrimental to enzyme
inhibitoryactivity(compare 2c to 2d or 2f to 2g and 2h)
with the fullysubstituted 2e being roughly900-fold less
potent than 2c. Steric branching at the b-carbon of the
P1 side chain appears to be tolerated (compare 2f and
2j) although potencydoes drop off in the fullysub-
stituted case 2k. Incorporation of p-bonds into the P1
side chain gave mixed results. The Z-alkene 2n was
experimentallyindistinguishable from the norleucine
analogue 2l, while the E-alkene 2o is 4-fold less active.
The alkyne 2p is 10-fold less potent than Z-alkene 2n.
During our SAR investigations, an X-rayco-crsytal
structure of inhibitor 2x bound to the active site of
cathepsin K was obtained (Fig. 1). The structure lends
some support to the observed SAR of these P1 ana-
logues. The S1 binding site is more aptlydescribed as
a wall rather than pocket, since one half of the subsite
is solvent exposed. It is created byresidues 23Gly, 24Ser,
64Gly, and 65Gly. All of the polar backbone atoms that
form the wall are hydrogen bonded which results in the
subsite having a veryhydrophobic character, explaining
the preference for linear hydrophobic substituents in
this position. Furthermore, P1 side chains with ‘hard’
heteroatoms close to the amino acid backbone should
have unfavorable interactions with this hydrophobic
enzyme surface. However, the enzyme wall is not extre-
melylarge ( ꢂ5 A in length). Thus, heteroatoms
attached bylonger tethers should eventuallybecome
exposed to solvent. In accord with this, lysine and orni-
Hoping to pick up favorable p–p stacking interactions
with potential aryl groups in the S1 subsite, a phenyl
ring was attached to the amino acid backbone with
varying tether lengths. These P1 side chains were found