5898
D. C. Tully et al. / Bioorg. Med. Chem. Lett. 18 (2008) 5895–5899
O
O
O
NHR2
O
NO2
HO
HO
O
OH
OMe
OMe
OMe
c,d
a
N
N
H
N
N
b
O
O
O
O
HCl
O
O
Cbz N
Cbz
N
O
Cbz
N
H
H
H
38
36
37
O
O
O
NHR2
NHR2
O
OH
O
H
H
N
N
O
O
N
N
f,g
e
O
N
O
N
O
O
Cbz
N
Cbz
N
H
H
HN
H2N
39
26-32
Boc
Scheme 4. Reagents and conditions: (a) Cbz-
(d) LiOHÁH2O, 1,4-dioxane/water 50:50, 85%; (e) 5, HATU, DIEA, CH2Cl2, 73–86%; (f) Dess–Martin periodinane, CH2Cl2, 45–65%; (g) TFA/CH2Cl2, reverse-phase HPLC, 75–85%.
D
-hPhe-OH, HATU, DIEA, CH2Cl2, rt, 84%; (b) para-nitrophenylchloroformate, pyridine, CH2Cl2, 76%; (c) H2NR2, CH2Cl2, 72–80%;
para-position led to modest improvements, with 4-chlorobenzyl
ether 23 contributing to threefold boost in potency
(Ki = 0.012 M) compared to 17. Cycloaliphatic ethers 24 and 25
demonstrate that a cyclohexyl ring (24, Ki = 0.019 M) is preferred
slightly over cyclopentyl (25, Ki = 0.041 M), suggesting that the
ethylcarbamate 32 (Ki = 1.55 lM), which is roughly equipotent to
a
the unfunctionalized proline 14, demonstrates that small hydro-
phobic substituents at the proline 4-position make no significant
contribution to the binding affinity, in stark contrast to the gain
in potency conferred by a methyl substituent at the 3-position of
l
l
l
larger cyclohexyl ring is able to more efficiently occupy the S2
proline (16, Ki = 0.049 lM). This suggests that a possible binding
subsite.
mode for 16 may be similar to that seen for compound 12, in which
the 90’s loop is closed, allowing for this methyl group to derive a
hydrophobic interaction within the narrow hydrophobic cleft.
In summary, we have utilized the substrate specificity profile as
a template for the first small molecule inhibitors of the channel-
activating protease prostasin. Direct translation of the substrate
sequence to an inhibitor sequence resulted only in marginally ac-
tive compounds. Subsequently, guided by the X-ray structure, we
were able to optimize the peptidomimetic scaffold to generate
the first low nanomolar Ki inhibitors of prostasin. The structure–
activity relationships surrounding analogs from this scaffold dis-
play a clear preference for large, flexible hydrophobic substituents
at the 4-position of the P2 proline subunit, whereas at the 3-posi-
tion of proline, there is a preference for smaller hydrophobic sub-
stituents. Structure-based design and medicinal chemistry
optimization eventually led to compound 23, which is a potent,
A second X-ray crystal structure of prostasin with compound 23
bound to the active site is shown in Figure 2. The various interac-
tions between the P1-lysine-ketobenzoxazole moiety and the res-
idues of the catalytic triad, oxyanion hole, and S1 subsite are
essentially identical to the structure with compound 12. The
noticeable difference arises in the S2 subsite, as the loop containing
Glu97, which defines a narrow cleft occupied by the phenethyl
group in the structure of compound 12 (Fig. 1), has undergone a
significant conformational change in this second structure
(Fig. 2). This change accommodates the binding of the para-chloro-
benzylether moiety into a broad hydrophobic compartment de-
fined primarily by Trp215, Met180, and Tyr94. The adjacent S3
subsite, defined principally by Trp215 and Pro172I, accommodates
the P3 phenethyl group. This moiety occupies essentially the same
space as the benzylcarbamate moiety of compound 12, while the
N-terminal benzyl group of 23 is now situated over the small
hydrophobic patch defined by Ala218.
reversible inhibitor of prostasin with Ki = 0.012 lM.
Next, we explored the SAR around a number of carbamates
substituted on the 4-position of the P2 proline subunit (com-
pounds 26–32), which were synthesized according to the proce-
dure outlined in Scheme 4. Both phenyl and benzyl carbamates
26 and 27 both exhibit a loss of potency from benzyl ether 23,
Supplementary data
Supplementary data associated with this article can be found, in
while addition of a polar sulfone (28, Ki = 0.176
lM) has a further
deleterious effect confirming the S2 subsite’s strong preference
References and notes
for non-polar, hydrophobic moieties. Interestingly, the piperidine
carbamate 29 (Ki = 0.029 lM) is roughly equipotent with cyclo-
hexylmethyl ether 24, while the pyrrolidine carbamate results in
a several-fold loss in inhibitory activity, suggesting again that the
six-membered aliphatic rings are able to fill the hydrophobic S2
1. Yu, J. X.; Chao, L.; Chao, J. J. Biol. Chem. 1994, 269, 18843.
2. Shipway, A.; Danahay, H.; Williams, J. A.; Tully, D. C.; Backes, B. J.; Harris, J. L.
Biochem. Biophys. Res. Commun. 2004, 324, 953.
3. Adachi, M.; Kitamura, K.; Miyoshi, T.; Narikiyo, T.; Iwashita, K.; Shiraishi, N.;
Nonoguchi, H.; Tomita, K. J. Am. Soc. Nephrol. 2001, 12, 1114.
4. Donaldson, S. H.; Hirsh, A.; Li, D. C.; Holloway, G.; Chao, J.; Boucher, R. C.;
Gabriel, S. E. J. Biol. Chem. 2002, 277, 8338.
5. Tong, Z.; Illek, B.; Bhagwandin, V. J.; Verghese, G. M.; Caughey, G. H. Am. J.
Physiol. Lung. Cell. Mol. Physiol. 2004, 287, L928.
6. McKenna, E. E.; Luke, C. J.; Frizzell, R. A.; Kleyman, T. R.; Pilewski, J. M. Am. J.
Physiol. Lung. Cell. Mol. Physiol. 2008, 294, L932.
subsite more efficiently. Morpholine carbamate 31 (Ki = 0.510 lM),
which drops off more than 17-fold in inhibitory activity from the
piperidine analog 29, again confirms that even a single heteroatom
is not well tolerated by the S2 binding pocket. Meanwhile, the dim-