Discovery of inhibitors of the channel-activating protease prostasin (CAP1/PRSS8) utilizing structure-based design

Article abstract of DOI:10.1016/j.bmcl.2008.08.029

Structure-based design was utilized to guide the early stage optimization of a substrate-like inhibitor to afford potent peptidomimetic inhibitors of the channel-activating protease prostasin. The first X-ray crystal structures of prostasin with small molecule inhibitors bound to the active site are also reported.

Full text of DOI:10.1016/j.bmcl.2008.08.029

Bioorganic & Medicinal Chemistry Letters 18 (2008) 5895–5899
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of inhibitors of the channel-activating protease prostasin
(CAP1/PRSS8) utilizing structure-based design
a
a
a
a
David C. Tully a, , Agnès Vidal , Arnab K. Chatterjee , Jennifer A. Williams , Michael J. Roberts ,
H. Michael Petrassi a, Glen Spraggon a, Badry Bursulaya a, Reynand Pacoma a, Aaron Shipway a,
Andrew M. Schumacher a, Henry Danahay b, Jennifer L. Harris a
*
a Genomics Institute of the Novartis Research Foundation, 10675 John J. Hopkins Dr., San Diego, CA 92121, USA
b Novartis Institute for Biomedical Research, Wimblehurst Road, Horsham RH12 5AB, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Structure-based design was utilized to guide the early stage optimization of a substrate-like inhibitor to
afford potent peptidomimetic inhibitors of the channel-activating protease prostasin. The first X-ray crys-
tal structures of prostasin with small molecule inhibitors bound to the active site are also reported.
Ó 2008 Elsevier Ltd. All rights reserved.
Received 12 June 2008
Revised 8 August 2008
Accepted 11 August 2008
Available online 14 August 2008
Keywords:
Prostasin
PRSS8
hCAP
Channel-activating protease
ENaC activation
Peptidomimetic
Ketoheterocycle inhibitor
Prostasin (CAP1/PRSS8) is
a
glycosylphosphatidyl-inositol-
known substrate specificity of prostasin2 as
a template. By
linked (GPI) serine protease of the chymotrypsin-fold that is ex-
pressed in the airway epithelium.1,2 Although the physiological
substrates of the catalytic activity are not known, prostasin has
been shown to be an activator of the epithelial sodium channel
(ENaC).3,4 Indeed, coexpression of prostasin and ENaC increases
the sodium current in Xenopus oocytes,3,4 whereas siRNA knock-
down attenuates the amiloride-sensitive sodium current in a cys-
tic fibrosis airway epithelial cell line.5 Recent findings suggest that
a proteolytic cascade involving prostasin regulates Na+ absorption
in the airway, and that abnormal prostasin expression contributes
to the excessive proteolytic activation of ENaC in cystic fibrosis pa-
tients.6 Since ENaC and other ion transport processes control the
luminal osmolarity required for normal lung function, it is be-
lieved that inhibition of upstream ENaC activators such as prosta-
sin may be attractive therapeutic targets in the treatment of cystic
fibrosis.
employing positional scanning combinatorial substrate libraries,
we have shown previously that prostasin exhibits trypsin-like sub-
strate specificity with a strong preference for arginine or lysine in
P1, large hydrophobic amino acids in P2, and basic residues P3.2
Here, we report the discovery of potent, reversible inhibitors of
prostasin that were initially developed using a substrate mimic ap-
proach as a starting point, and then optimized with the aid of the
crystal structure of prostasin to improve binding affinity.
We sought to convert our optimized peptide substrate
Ac-KHYR-acmc 1 into an inhibitor by replacing the labile coumarin
amide with an electrophilic transition-state analog mimic. To
accomplish this, we chose a-ketoheterocycle transition-state ana-
logs, which are well-known reversible warheads extensively
employed in serine protease inhibitors.7–9 The synthesis of the pre-
ferred P1-lysine
a-ketobenzoxazole intermediate 5 is shown in
Scheme 1, starting with commercially available Cbz-Lys(Boc)-OH
2 which was converted to the corresponding aldehyde 3 in two
steps. Grignard addition of benzoxazole to the aldehyde 3 gave
alcohol 4, and the subsequent hydrogenolysis of the Cbz protecting
group afforded the aminoalcohol 5 as a mixture of diastereomers.
A small number of transition-state analog substrate mimics
(compounds 8–13, Table 1) were initially made using the standard
peptide coupling (HATU) and hydrogenolysis conditions shown in
To date, no small molecule inhibitors of prostasin have been
reported. High-throughput screening efforts yielded no viable
starting points for medicinal chemistry, so we instead turned to a
rational design strategy of peptidomimetic inhibitors using the
* Corresponding author. Tel.: +1 8588121559; fax: +1 8588121648.
E-mail address: dtully@gnf.org (D.C. Tully).
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.08.029

5896
D. C. Tully et al. / Bioorg. Med. Chem. Lett. 18 (2008) 5895–5899
R2
OH
O
O
H
N
H
N
H
N
R2
Cbz
O
Cbz
OH
N
H
Cbz
H
Cbz
OH
d,e
O
N
c
N
H
a
a, b
8, 9
O
NH
NH
6
NH2
Boc
2
Boc
3
R2
OH
R2
O
H
N
H
N
OH
R3
OH
O
R3
O
H
N
N
H
N
H
H2N
O
O
Cbz
O
N
O
N
d
d,e
N
N
b,c
NH2
10-18
NH2
7
NH
NH
4
Boc
Boc
5
Scheme 2. Reagents and conditions: (a) 5, HATU, DIEA, DCM, rt, 86%; (b) H2
(40 psi), EtOH, Pd/C, rt, 18 h, 92%. (c) R3COOH, HATU, DIEA, DCM, rt, 81–90%; (d)
Dess–Martin periodinane, CH2Cl2, 45–65%; (e) TFA/ CH2Cl2, reverse-phase HPLC,
75–85%.
Scheme 1. Reagents and conditions: (a) i—iso-BuOCOCl, Et3N, THF; ii—NaBH4, H2O,
63%; (b) Dess–Martin periodinane, CH2Cl2; (c) iso-PrMgCl, benzoxazole, THF,
À20 °C, 30 min, then aldehyde 3, À20 °C to rt, 31–48% yield combined steps b
and c; (d) H2 (40 psi), EtOH, Pd/C, rt, 18 h, 69%.
Asp189 at the bottom of the S1 pocket, while the P3-lysine is
mostly solvent exposed with no obvious interactions. The phen-
ethyl P2 moiety is sandwiched between the Glu97 and Trp215
sidechains, while the N-terminal benzyl group is stacked edge on
face with the indole ring of Trp215.
Several clues from the X-ray structure of compound 12 suggest
possible directions for the optimization of inhibitor design in order
to improve binding affinity. First, the P2 NH clearly does not partic-
ipate in any H-bonding interactions, suggesting that proline could
potentially be tolerated in P2 to add a degree of conformational
restraint to the peptidic scaffold. Second, the natural L-configura-
tion of the P3-lysine directs the sidechain out toward solvent,
and forces the N-terminal benzyl group into the hydrophobic pock-
et defined primarily by Trp215, Pro172I, His172J, and Gly98. Using
this X-ray structure as a guide, modeling suggested that the oppo-
site configuration of the P3 subunit bearing a hydrophobic side-
chain could more efficiently fill this hydrophobic pocket, and
Table 1
Prostasin inhibition constants (Ki) for early substrate-like transition-state analogsa
R2
O
H
N
R3
O
N
H
O
N
NH2
Compound
R3
R2
Prostasin Ki (lM)
8
9
10
11
12
13
Cbz
Cbz
Cbz-Lysine
Cbz-Histidine
Cbz-Lysine
Cbz-Histidine
–CH2CH(CH3)2
–CH2CH2Ph
–CH2CH(CH3)2
–CH2CH(CH3)2
–CH2CH2Ph
>100
>100
15.2
5.72
5.78
2.10
take advantage of the p-stacking interaction with Trp215.
Incorporation of these changes into the initial inhibitor design
led us to compound 14, which has a proline in P2 and the opposite
–CH2CH2Ph
a
Details of the assay conditions can be found in Ref. 2 and Supplementary
material.
configuration D-hPhe in P3 (Table 2). Conformational restraint con-
ferred by the proline in P2 led to a slight improvement in potency
Scheme 2, followed by Dess–Martin oxidation of the penultimate
alcohol 7 and deprotection of the Boc groups in the final step. Uti-
lizing the preferred substrate specificity profile of prostasin,2
hydrophobic residues leucine and homophenylalanine (hPhe) were
initially chosen for P2, while basic residues His and Lys were cho-
sen for P3. Structure–activity relationships in Table 1 clearly dem-
onstrate the requirement for a P3 residue, as truncated analogs 8
and 9 did not inhibit prostasin at concentrations below 100
Introduction of the Cbz-Lys and Cbz-His in P3 (10–13) generated
several prostasin inhibitors with weak (low M Ki) binding affinity.
The X-ray structure of compound 12 (Ki = 5.78 M) bound to
lM.
l
l
the active site of prostasin is shown in Figure 1, and clearly reveals
each of the four subsites (P4-P1) and their respective interactions.
The
a-ketobenzoxazole warhead establishes the contacts in the
oxyanion hole analogous to previously reported X-ray crystal
structures of similar
a-ketoheterocycles bound to the active sites
of related serine proteases.7–10 Key hydrogen bonds along the main
chain of the inhibitor include the Gln192 sidechain to the P2 car-
bonyl and an anti-parallel b-sheet-like conformation between the
Gly216 and the P3-lysine moiety. The inhibitor’s P1-lysine side-
chain establishes a water-mediated H-bond with the sidechain of
Figure 1. Crystal structure of Prostasin with compound 12 at 1.6 Å resolution (RCSB
PDB ID: 3E16). Figure generated using PyMol.

D. C. Tully et al. / Bioorg. Med. Chem. Lett. 18 (2008) 5895–5899
5897
Table 2
able to extend into the S2 subsite. Interestingly, the replacement
of the phenyl group for a simple methyl group at the proline 3-po-
Effect of P2 substituents on prostasin inhibition (Ki)
sition (16) further improved the potency (Ki = 0.049 lM). To gain
R3
R2
additional insight into the SAR at the P2 position, we next chose
to explore the effect of introducing substituents at the syntheti-
cally more accessible 4-position of proline. Benzyl ether 17
O
H
N
N
O
(Ki = 0.040 lM) demonstrated that comparable improvements in
inhibitory activity could be achieved by the incorporation of
O
O
N
Cbz
N
hydrophobic substituents at the proline 4-position, while the 4-hy-
H
droxy analog 18 (Ki = 1.04 lM) confirmed that larger hydrophobic
NH2
substituents were preferred over small polar moieties.
To explore the effect of substituents on the P2 benzyl ring have
on the SAR, a small subset of substituted benzyl ethers (19–25)
were synthesized according to the procedure outlined in Scheme
Compound
R3
R2
Prostasin Ki (lM)
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
–H
–H
–H
–OCH2Ph
–OH
–OCH2(meta-CF3)Ph
–OCH2(para-CF3)Ph
–OCH2(para-Me)Ph
–OCH2(para-F)Ph
–OCH2(para-Cl)Ph
–OCH2-cyclo-C6H11
–OCH2-cyclo-C5H9
–OCONHPh
–H
1.40
–Ph
–Me
–H
–H
–H
–H
–H
–H
–H
–H
–H
–H
–H
–H
–H
–H
–H
–H
0.267
0.049
0.040
1.04
3
from the commercially available starting material Boc-
L-
hydroxyproline. Compound 19 (Ki = 0.041 M) demonstrated that
l
substitution on the meta-position of the benzyl ring has no notice-
able effect on potency, while small hydrophobic substituents at the
0.041
0.028
0.045
0.027
0.012
0.019
0.065
0.130
0.060
0.176
0.029
0.101
0.510
1.55
–OCONHCH2Ph
–OCONHCH2(p-SO2Me)Ph
Piperidinecarbonyloxy–
Pyrrolidinecarbonyloxy–
Morpholinecarbonyloxy–
–OCONMe2
(14, Ki = 1.4 lM) over the previous analogs 10–13. However, com-
pound 14 lacks a P2 substituent capable of extending into the S2
subsite to pick up any favorable hydrophobic interactions. To
achieve this, compound 15 was prepared using the conformation-
ally constrained analog of hPhe in P2 from commercially available
(2R,3S)-3-phenyl-
onto the proline ring resulted in a fivefold improvement in potency
(Ki = 0.267 M) over 14, demonstrating the importance of the
L-proline. Incorporation of a phenyl substituent
l
Figure 2. Crystal structure of Prostasin with compound 23 at 1.7 Å resolution (RCSB
PDB ID: 3EOP). Figure generated using PyMol.
favorable interaction gained by hydrophobic substituents that are
R2
R2
O
R2
HO
O
O
OH
N
OH
a
b,
c
d,e
OH
OMe
N
O
N
N
H
O
Boc
Cbz
N
O
O
Boc
O
O
H
33
34
32
R2
R2
O
O
OH
H
N
H
N
O
O
N
N
f
g,h
O
N
O
N
O
Cbz
N
O
Cbz
N
H
H
HN
19-25 H2N
35
Boc
Scheme 3. Reagents and conditions: (a) i—KOH (8.0 equiv), DMSO, 0 °C, 20 min; ii—R2CH2Br (4.5 equiv), 0 °C, 15 min, then 20 °C, 4 h, 82%; (b) TMS-CHN2 (2.0 M in Et2O), 20%
MeOH in CH2Cl2, 89%; (c) TFA (50%) in CH2Cl2, 100%; (d) Cbz-
-homoPhe-OH, HATU, DIEA, CH2Cl2, rt, 84%; (e) LiOHÁH2O, 1,4-dioxane/water 50:50, 85%; (f) 5, HATU, DIEA,
CH2Cl2, 73–86%; (g) Dess–Martin periodinane, CH2Cl2, 45–65%; (h) TFA/CH2Cl2, reverse-phase HPLC, 75–85%.
D

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
the online version, at doi:10.1016/j.bmcl.2008.08.029.
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-

D. C. Tully et al. / Bioorg. Med. Chem. Lett. 18 (2008) 5895–5899
5899
7. (a) Edwards, P. D.; Meyer, E. F., Jr.; Vijayalakshmi, J.; Tuthill, P. A.; Andisik, D. A.;
Gomes, B.; Strimpler, A. J. Am. Chem. Soc. 1992, 114, 1854; (b) Edwards, P. D.;
Wolanin, D. J.; Andisik, D. W.; Davis, M. W. J. Med. Chem. 1995, 38, 76; (c)
Edwards, P. D.; Zottola, M. A.; Davis, M. W.; Williams, J.; Tuthill, P. A. J. Med.
Chem. 1995, 38, 3972.
Corcoran, T. W.; Giardino, E. C.; Kauffman, J. A.; Lewis, J. M.; deGaravilla, L.;
Haertlein, B. J.; Maryanoff, B. E. J. Med. Chem. 2005, 48, 1984.
A comprehensive review on the subject of a-ketoheterocycle protease
inhibitors: Maryanoff, B. E.; Costanzo, M. J. Bioorg. Med. Chem. 2008, 16,
9.
1562.
8. (a) Costanzo, M. J.; Maryanoff, B. E.; Hecker, L. R.; Schott, M. R.; Yabut, S. C.;
Zhang, H.-C.; Andrade-Gordon, P.; Kauffman, J. A.; Lewis, J. M.; Krishnan, R.;
Tulinsky, A. J. Med. Chem. 1996, 39, 3039; (b) Costanzo, M. J.; Almond, H. R., Jr.;
Hecker, L. R.; Schott, M. R.; Yabut, S. C.; Zhang, H.-C.; Andrade-Gordon, P.;
10. (a) Matthews, J. H.; Krishnan, R.; Costanzo, M. J.; Maryanoff, B. E.; Tulinsky, A.
Biophys. J. 1996, 71, 2830; (b) Recacha, R.; Costanzo, M. J.; Maryanoff, B. E.;
Carson, M.; DeLucas, L. J.; Chattopadhyay, D. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 2000, 56, 1395.