Lipophilic versus hydrogen-bonding effect in P3 on potency and selectivity of valine aspartyl ketones as caspase 3 inhibitors

Article abstract of DOI:10.1016/j.bmcl.2005.05.116

Caspase 3 is a cysteinyl protease that mediates apoptotic cell death. Its inhibition may have an important impact in the treatment of several degenerative diseases. The P₁ aspartic acid residue is a required element of recognition for this enzy

Full text of DOI:10.1016/j.bmcl.2005.05.116

Bioorganic & Medicinal Chemistry Letters 15 (2005) 3886–3890
Lipophilic versus hydrogen-bonding effect in P3 on potency
and selectivity of valine aspartyl ketones as caspase 3 inhibitors
a
a
a
Christophe Mellon,a, Renee Aspiotis, Cameron W. Black, Christopher I. Bayly,
*
´
Erich L. Grimm, Andre Giroux, Yongxin Han, Elise Isabel, Daniel J. McKay,
a
a
a
a
a
´
Donald W. Nicholson,c Dita M. Rasper,a Sophie Roy,c John Tam,a Nancy A. Thornberry,b
John P. Vaillancourt,a Steven Xanthoudakisa and Robert Zambonia
aMerck Frosst Centre for Therapeutic Research, Merck Frosst Canada & Co., P.O. Box 1005, Pointe-Claire-Dorval,
Que., Canada H9R 4P8
bMerck Research Laboratories, Rahway, NJ 07065, USA
cMerck & Co., Inc., MRLSDB1, 3535 General Atomics Court, San Diego, CA 92121, USA
Received 2 March 2005; revised 25 May 2005; accepted 26 May 2005
Available online 14 July 2005
Abstract—Caspase 3 is a cysteinyl protease that mediates apoptotic cell death. Its inhibition may have an important impact in the
treatment of several degenerative diseases. The P1 aspartic acid residue is a required element of recognition for this enzyme that was
maintained constant along with the adjacent natural valine as the P2 group. The thiobenzylmethylketone warhead on the aspartate
was conveniently handled through solid-phase synthesis allowing modification in the P3 region that eventually led to simpler deriv-
atives with increased potency against caspase 3. The key to such an effect is the introduction of hydroxyl group alpha to the P3
carbonyl.
Ó 2005 Elsevier Ltd. All rights reserved.
The human caspases constitute an ensemble of at least
13 cysteinyl-aspartate-specific proteinases that are key
elements in the molecular pathways resulting in the
apoptosis of cells.1 Caspases 1, 4, and 5 mediate
cytokine maturation and are linked to the inflammatory
phenomenon. Caspases 2, 3, and 7 are the major effec-
tors of cell death. Caspases 6, 8, 9, and 10 are activators
of the effector caspases 2, 3, and 7. Finally, caspase 14 is
a heratinocyte-specific caspase. Caspase 3 in particular
appears to be a pivotal player in neuronal apoptosis.
Peptidic inhibitors of caspase 3 have shown some effica-
cy in models such as stroke, traumatic brain/spinal cord
injury, hypoxic brain damage, and cardiac ischemia/
reperfusion injury.2 The P1 aspartic acid residue is a
required element of recognition of the enzyme, and its
scissile amide bond can be replaced with a ketone to
afford reversible inhibitors. Here, we report dipeptidic
aspartyl ketones as potent, selective, and cell-penetrable
inhibitors of caspase 3.
We recently disclosed that an a-methylthiobenzyl aspar-
tyl ketone warhead was a suitable electrophile, replacing
with advantage over the aldehyde warhead in a peptide
structure.3 The ketone itself, in conjunction with the
aspartic acid residue, was a capricious functionality
whose synthesis required tedious protection–deprotec-
tion sequences or convoluted routes. We therefore devel-
oped a very convenient method to rapidly access a variety
of P3 derivatives in that series using solid-phase chemistry.
Hence, commercial Fmoc-aspartic acid b-t-butyl ester 1
(Scheme 1) was activated using the mixed anhydride
method4 and was treated with diazomethane. The result-
ing a-diazoketone was reacted with 48% aqueous HBr at
low temperature to yield the corresponding a-bromoke-
tone 2. At this point, the a-bromoketone 2 was attached
to a Merrifield resin that had previously been elaborated
to unveil hydrasemicarbazone 3.5 The bromide was dis-
placed with benzyl mercaptan to give thioether 4 fol-
lowed by removal of the Fmoc group. The free amine
was then coupled to FmocValine to yield intermediate
5. Again, the Fmoc was removed using piperidine to
liberate the amine of the polymer bound valine
Keywords: Caspase 3 inhibitors ketones hydrogen bond.
*
Corresponding author. Tel.: +1 514 428 3710; fax: +1 514 428
4939; e-mail: christophe_mellon@merck.com
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.05.116

C. Mellon et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3886–3890
3887
N
O
O
H
N
H
H
d, e
SBn
N
N
Br
a, b, c
Fmoc
Fmoc
f,g
Fmoc
N
OH
COOt-Bu
COOt-Bu
COOt-Bu
1
2
4
O
O
H
H
N
f,h,i
Fmoc
N
SBn
SBn
R
N
N
H
H
COOt-Bu
O
COOH
O
O
5
6-35
Linker:
N
H
H
H
N
N
H2N
=Merrifield resin, 2% crosslinked
NH2
3
O
Scheme 1. Synthesis of dipeptidic aspartylthiobenzylmethylketones. Reagents and conditions: (a) i-BuOCOCLI, NMM, À10 °C, CH2CL2;
(b) CH2N2, Et2O, 0 °C; (c) HBr/AcOh, À78 °C; (d) linker-NH2 3, AcOH, THF; (e) PhCH2SH, DMF, (iPr)2NEt; (f) 20% piperidine/DMF;
(g) FMOCVal-H, HATU, DMF, (iPr)2NEt; (h) RCOOH (see 6–35), HATU, DMF, (iPr)2NEt; (i) 9:1 TFA/H2O.
aspartylthiobenzylmethylketone. Amide formation with
a variety of carboxylic acids followed by the treatment
of the resin with wet trifluoroacetic acid afforded dipep-
tides 6–35 on 10–20 mg scale, generally in good to excel-
lent purity.
was used for solid-phase chemistry. This resulted in a
mixture of diastereomers that were separated only when
they were distinct by flash chromatography.
Introduction of a methyl group (compound 7) at this
benzylic position was deleterious to the potency on cas-
pase 3. Attempt to maximize lipophilic interactions with
the enzyme by increasing the size of the substituents
resulted in decreased caspase 3 activity (compounds 8
and 9). This was especially true when introducing steri-
cally more demanding groups (compounds 10 and 11).
However, potency was restored with the narrower sub-
stituents as demonstrated by the allyl (12) and propargyl
groups (13 and 14). In the latter case, potency was mod-
ulated by the stereochemistry of that group (compare 13
versus 14), although the stereochemistry was not estab-
lished at the benzylic position. Overall, compound 14
only matched the potency of the unsubstituted parent
One of the hits against caspase 36 that was obtained
from coupling commercially available reagents was a
2,5-substituted phenylacetic acid (compound 6,
Table 1). At that stage, we decided to introduce substi-
tution into the open aliphatic position using enolate
alkylation chemistry (Scheme 2). Generally, 5-bromo-
2-methoxyphenylacetic acid methyl ester underwent
deprotonation when added into a cold solution of lithi-
um diisopropyl amide via a cannula. The resulting
nucleophile was quenched with an excess of an alkyl
halide. The ester was then purified by flash chromatog-
raphy on silica gel and hydrolyzed to afford the acid that
Table 1. Substituted 5-bromo-2-methoxyphenylacetic acid analogs in P3
Compound
R1
R2
Enzyme IC50 (nm)
C-1a
3040
C-3a
C-7a
C-8a
1885
6
H
H
Me
18
176
274
181
9445
323
80
284
7
H
>10,000
>10,000
5810
3690
5300
5380
4505
8
H
n-Pr
Benzyl
i-Pr
9
H
4200
5830
1320
10
11
12
13
14
H
Cyclopropyl
>10,000
>10,000
6505
>10,000
>10,000
2215
>10,000
1625
4470
H
Allyl
Propargyl
H
H
Propargyl
>10,000
1030
341
17
3625
139
92
a Recombinant human caspases 1, 3, 7, and 8. See Ref. 6.

3888
C. Mellon et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3886–3890
Scheme 2. Alkylation of 5-bromo-2-methoxyphenylacetic acid. Reagents and conditions: (a) i—LDA, À78 °C, 30 min and ii—RX, 2–5 equiv.,
À78 °C to room temperature; (b) LiOHÆH2O, MeOH–THF–H2O, room temperature.
6. Furthermore, the selectivity profile against other
caspases generally worsened with these modifications.
These modifications also resulted in an increase in the
molecular weight as well as more metabolically labile
sites. We thus turned our attention to a different
subclass of phenylacetic acids: the mandelic derivatives.
Table 2. Potency and selectivity from SAR around the OH group
Compound
R
Enzyme IC50 (nm)
C-3a C-7a
The (R) mandelic acid derivative 16 (Table 2) showed a
20-fold increase in potency when compared with the
unsubstituted parent 15. This large gain in potency from
a hydroxyl group suggests a hydrogen bonding interac-
tion with a not well-solvated7 backbone carbonyl group
in the enzyme S3 pocket. As in the case with alkyl
groups, potency varied with stereochemistry but only
slightly within a pair of diastereoisomers (compounds
16 versus 17 and 18 versus 19). Noteworthy is that the
preferred stereochemistry depended on the exact nature
of the substituent when compared across pairs (pair 16
and 17 compared to pair 18 and 19). This suggested that
a powerful hydrogen bond was overriding beneficial
lipophilic interactions. In fact, when the hydroxyl group
was capped with a methyl group (compound 20), a con-
siderable loss in potency was observed, in agreement
with a hydrogen bond formation with the enzyme back-
bone although undesirable steric effects cannot be ruled
out. To reduce the potential for metabolism at the ben-
zylic position, this position was quaternized with a
methyl group (compound 21). This racemic derivative
was almost as potent as the optically pure mandelate
16. This was unexpected since alpha substitution in the
phenylacetic acid derivatives showed a marked decrease
in potency. This observation constituted a second hint
for the existence of a different binding mode driven by
a hydrogen bond in the hydroxyl series. Achiral disub-
stituted a-hydroxy-acids were then incorporated via sol-
id-phase chemistry. Clearly, less sterically demanding
groups afforded more potent derivatives as demonstrat-
ed by the sequence of compounds culminating in a
cyclopropyl derivative (compounds 22–25). Interesting-
ly, the bistrifluoromethyl alcohol 26 showed a marked
increase in potency, despite being larger than the gem-
dimethyl 24. This can be rationalized by invoking a
stronger hydrogen bond due to increased acidity of the
bistrifluoromethyl alcohol 26. Overall, the introduction
of a hydroxyl group was beneficial to the selectivity pro-
file over other caspases. The most potent mandelate
derivatives still offered the tantalizing possibility to gain
further potency via substitution on the phenyl ring.
C-1a
21,200
C-8a
22,050
15
16
746
8360
Ph
4160
41
1050
1150
17
18
4060
100
10,860
3650
7270
>10,000
330 >10,000
19
20
21
22
23
24
25
26
3015
>10,000
18,900
118
3405
7340
9175
8000
425 >10,000
85
2760
Ph
>10,000 1295 >10,000 >10,000
Ph OH
>10,000 2020 >10,000
9580
13,800
12,300
18,400
9590
467
160
10,860
5095
>10,000
146 >10,000 >10,000
a Recombinant human caspases 1,3,7, and 8. See Ref. 6.
ing keto esters were reduced with sodium borohydride to
obtain a racemic mixture of a-alkoxy esters. To intro-
duce electron-poor heteroaromatic groups on the phenyl
ring, palladium-catalyzed elaboration of the halide pro-
duced the desired derivatives uneventfully.8,9 After alka-
line hydrolysis of the a-hydroxyl ester, the acid was
coupled with the amine on solid phase using HATU in
To access racemic 2,5-disubstituted mandelic acid
(Scheme 3), Friedel–Craft reactions were performed on
4-substituted anisole derivatives with ethyl oxalyl chlo-
ride in the presence of aluminum trichloride. The result-

C. Mellon et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3886–3890
R1-5
3889
R1-3
R1-3
O
O
d)
a)
b)
c)
OEt
OEt
N
O
OH
O
O
O
R1-5
O
H
N
h,i
Compounds
28,30,32,34,35
SBn
+
H2N
OH
COOt-Bu
O
O
OH
Scheme 3. Preparation of substituted methoxymandelic acid. Reagents and conditions: (a) Ethyloxayl chloride, ALCL3, À10 °C to 25 °C, (CH2CL)2;
(b) NaBH4, MeOH, 0 °C; (c) R1–3 = OMe, Br, and I: anisoles were commercially available; R4 = oxadiazole: see Ref. 9; R5 = acetyl; see Ref. 8;
(d) LiOHÆH2O, MeOH–THF–H2O, room temperature; (h) HATU, DMF, (iPr)2NEt; (i) 9:1 TFA/H2O.
DMF. Under these conditions, the hydroxyl group was
inert and no ester formation was detected. After acidic
cleavage from the resin and purification on silica gel,
the elaborated a-benzylthioketones were obtained in
10–20 mg amounts.
(35) in the NT2 whole cell assay.10 Moreover, on
top of being easy to synthesize, the iodo derivative of-
fers a great handle to develop a cell-permeable radio-
active tool.11
In summary, the discovery and exploration of a hydro-
gen-bonding effect allowed us to determine that two dis-
tinct and mutually exclusive binding modes existed in
the P3 region. The strong gain in potency upon hydro-
gen-bond formation allowed for the reduction of the
molecular weight of our inhibitors, diminished the num-
ber of potentially metabolically labile sites and increased
their selectivity against other caspases (e.g., compound
26). With the simpler derivatives 15–26, the potency of
these inhibitors bottomed out at 146 nM. In the case
where the lipophilic interactions were maximized by
optimal substitution around the phenyl ring (27–34),
the presence of the hydroxyl group diminished the selec-
tivity of these derivatives. Future in vivo evaluation of
each class of compounds should give different options
for their development.
The SAR around the phenyl ring in the mandelic acid
series paralleled the SAR for the corresponding phen-
ylacetic acid derivatives as shown by their relative
rank order of potencies (see pairs in Table 3). Howev-
er, as the substitution on the ring fit better into the
lipophilic part of the enzymatic pocket (compounds
28–35), the potency advantage initially provided by
the introduction of a hydroxyl group (Table 2) dimin-
ished and eventually vanished. It was thus concluded
that the geometry of binding allowing optimal lipo-
philic interactions was not attainable while maintain-
ing a strong hydrogen bond and vice versa. In the
optimization process, we discovered that a simple
iodo substituent could replace the oxadiazole on the
phenyl ring to afford the most potent compound
Table 3. Hydroxyl group effect in optimally substituted phenyl analogs
Compound
R1
R2
IC50 (nm)
C-1a
C-3a
C-7a
C-8a
NT2b
27
28
29
30
31
32
OMe
OMe
Br
H
7070
4490
3050
1540
6190
4850
53
29
18
5
3240
545
285
91
7910
975
5360
945
OH
H
1890
104
—
710
Br
OH
H
ACc
ACc
13
16
555
290
3880
635
1225
1590
OH
d
33
34
35
H
5450
2460
675
9
12
16
97
63
3290
280
98
415
310
224
{
OH
OH
I
135
a Recombinant human caspases 1, 3, 7, and 8. See Ref. 6.
b NT2 whole cell assay. See Ref. 10.
c Acetyl.
d 3-Methyl-1,2,4-oxadiazole. See Ref. 9.

3890
C. Mellon et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3886–3890
4. Chen, F. M. F.; Lee, Y.; Steinauer, R.; Benoiton, N. L.
Can. J. Chem. 1987, 65, 613.
References and notes
5. Webb, T. R. et al. J. Am. Chem. Soc. 1992, 114, 3156.
6. Roy, S.; Bayly, C. I.; Gareau, Y.; Houtzager, V. M.;
Kargan, S.; Keen, S. L.; Rowland, K.; Seiden, I. M.;
Thornberry, N. A.; Nicholson, D. W. Proc. Natl Acad.
Sci. USA 2001, 98, 6132.
7. The energy gained by the interaction of the backbone
carbonyl with the hydroxyl of the inhibitor would be
cancelled by the energy required to remove a water
molecule solvating that carbonyl: Kangas, E.; Tidor, B.
Phys. Rev. E 1999, 59, 5958.
8. Acetyl: Davies, S. G.; Pyatt, D.; Thomson, C. J. Organo-
met. Chem. 1990, 387, 381.
9. (a) Oxadiazole: N-hydroxy-acetamidine Hamze, A.; Her-
nandez, J.-F.; Martinez, J. Tetrahedron Lett. 2003, 32,
6079; Formation: (b)Young, J. R.; DeVita, R. J. Tetrahe-
dron Lett. 1998, 23, 3931.
1. Nicholson, D. W. Cell Death Differ. 1999, 6, 1028.
2. Nicholson, D. W. Nature 2000, 407, 810, and references
cited therein.
3. (a) Grimm, E. L.; Roy, B.; Aspiotis, R.; Bayly, C. I.;
Nicholson, D. W.; Rasper, D. M.; Renaud, J.; Roy, S.;
Tam, J.; Tawa, P.; Vaillancourt, J. P.; Xanthoudakis, S.;
Zamboni, R. Bioorg. Med. Chem. 2004, 12, 845; (b) Han,
Y.; Giroux, A.; Grimm, E. L.; Aspiotis, R.; Francoeur, S.;
Bayly, C. I.; Mckay, D. J.; Roy, S.; Xanthoudakis, S.;
Vaillancourt, J. P.; Nicholson, D. W.; Rasper, D. M.;
Tam, J.; Tawa, P.; Thornberry, N. A.; Paterson, E. P.;
Garcia-Calvo, M.; Becker, J. W.; Rotonda, J.; Nicholson,
D. M.; Renaud, J.; Zamboni, R. Bioorg. Med. Chem. Lett.
2004, 14, 805; (c) Isabel, E.; Black, C. W.; Bayly, C. I.;
Grimm, E. L.; Janes, M. K.; McKay, D. J.; Nicholson, D.
W.; Rasper, D. M.; Renaud, J.; Roy, S.; Tam, J.;
Thornberry, N. A.; Vaillancourt, J. P.; Xanthoudakis,
S.; Zamboni, R. Bioorg. Med. Chem. Lett. 2003, 13,
2137.
10. Tawa, P.; Tam, J.; Cassady, R.; Nicholson, D. W.;
Xanthoudakis, S. Cell Death Differ. 2001, 8, 30.
11. Manuscript in preparation.