Amino-caprolactam derivatives as γ-secretase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2007.08.047

A series of amino-caprolactam sulfonamides were developed from a screening hit. Compounds with good in vitro and in vivo γ-secretase activity are reported.

Full text of DOI:10.1016/j.bmcl.2007.08.047

Bioorganic & Medicinal Chemistry Letters 17 (2007) 5790–5795
Amino-caprolactam derivatives as c-secretase inhibitors
Michael F. Parker,a,* Joanne J. Bronson,a Donna M. Barten,b Jason A. Corsa,b
Wengsheng Du,b Kevin M. Felsenstein,b, Valerie L. Guss,b Darcy Izzarelli,b
Alice Loo,b Kate E. McElhone,a,à Larry R. Marcin,a Ramesh Padmanabha,b
Roger Pak,b Craig T. Polson,b Jeremy H. Toyn,b Sam Varma,b Jian Wang,c
Victoria Wong,b Ming Zhengd and Susan B. Robertsb
aDepartment of Discovery Chemistry, Bristol-Myers Squibb Research and Development,
5 Research Parkway, Wallingford, CT 06492, USA
bDepartment of Neuroscience Biology, Bristol-Myers Squibb Research and Development,
5 Research Parkway, Wallingford, CT 06492, USA
cDepartment of Bioanalytical Research, Bristol-Myers Squibb Research and Development, Rt. 206, Lawrenceville, NJ 08543, USA
dDepartment of Drug Metabolism and Pharmacokinetics, Bristol-Myers Squibb Research and Development,
5 Research Parkway, Wallingford, CT 06492, USA
Received 24 July 2007; revised 20 August 2007; accepted 23 August 2007
Available online 26 August 2007
Abstract—A series of amino-caprolactam sulfonamides were developed from a screening hit. Compounds with good in vitro and
in vivo c-secretase activity are reported.
ꢀ 2007 Elsevier Ltd. All rights reserved.
Alzheimer’s disease (AD) is a progressive, neurodegen-
erative disorder characterized by memory impairment
and cognitive dysfunction.1 AD is characterized patho-
logically by the accumulation of senile (neuritic) pla-
ques, neurofibrillary tangles, amyloid deposition in
neural tissues and vessels, synaptic loss, and neuronal
death.2a,b AD is the most common form of dementia
and now represents the third leading cause of death after
cardiovascular disorders and cancer.3 Evidence suggests
the accumulation of b-amyloid peptides (Ab) is respon-
sible for the neuronal toxicity that is associated with
AD.4 Ab peptides are generated by sequential proteo-
lytic cleavage of a 695–770 amino acid precursor protein
(APP)5 by the action of b- and c-secretases.
We have been interested in identification of compounds
that inhibit the production of b-amyloid peptide (b-AP)
from b-amyloid precursor protein (b-APP), since such
agents may be useful for the treatment or prevention
of Alzheimer’s disease. A cell-based assay measuring
Ab production6 from membranes isolated from cells
expressing b-APP7 was performed on our internal com-
pound collection and the caprolactam sulfonamide
derivative 18,9 (Fig. 1) was found to be a promising
inhibitor of c-secretase (IC50 = 120 nM). Herein we re-
port the SAR within this series of caprolactam c-secre-
tase inhibitors.
Starting from commercially available D- or L-lysine 2,
amino caprolactam 3 was conveniently made by ring
closure of the trimethylsilyl amino ester10 as shown in
H
N
O
Keywords: c-Secretase inhibitors; Alzheimer’s disease; N-Benzyl amino
caprolactams.
N
*
Corresponding author. Tel.: +1 203 677 6466; fax: +1 203 677
7702; e-mail: parkerm@BMS.com
SO2
O
Present address: Johnson & Johnson PRD, Spring House, PA 19477,
USA.
Br
1
à
Present address: Johnson & Johnson Pharmaceutical Research &
Development, La Jolla, CA 92121, USA.
Figure 1.
0960-894X/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.08.047

M. F. Parker et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5790–5795
5791
Scheme 1. This synthetically useful chiral amine under-
went reductive amination by treatment with an aldehyde
and sodium cyanoborohydride in methanol to provide
intermediate 4. The resulting secondary amines were
treated with a variety of sulfonyl chlorides to produce
the desired analog 5 shown in Scheme 1. Alternatively
the chiral amino caprolactam 3 could first be sulfonylat-
ed to afford intermediate 6. The resulting secondary sul-
fonamides could be preferentially alkylated with an
array of alkyl or benzyl halides in DMF with K2CO3
(method 2A). In the preparation of the 4-hydroxybenzyl
benzenesulfonamide analog 7, synthesis via method 1 or
method 2A was unsuccessful because of decomposition
of the benzylic halide under the usual reaction condi-
tions. The alkylation of the sulfonamide was accom-
plished instead by a standard Mitsunobu reaction on
the sulfonamide caprolactam 6 with the differentially
protected ether alcohol, 4-(tert-butyldimethylsilyloxy)
benzyl alcohol. The crude product of this reaction was
then deprotected with tetrabutyl ammonium fluoride
in THF, to yield the desired phenol 7.
Optimization of the 4-bromosulfonamide caprolactam
screening lead first focused on determination of whether
the enantiomers had differential potency. Comparison of
the enantiomers 8 and 9 clearly showed the R-isomer,
which was derived from D-lysine, to be significantly,
and as with 9, exceptionally more potent.12 Likewise,
with the enantiomer pair 10 and 11, the R-isomer 11
was >300· more potent (Table 1).
H
H
N
N
O
O
c
N
R
R
N
Method 1
H
H
SO2
b
c
4
N
5
O
a
Cl
H2N-(CH2)4CHCO2H
NH2
NH2
Method
2A
3
2
d
H
N
Method 2
O
H
N
O
NH
e,f
SO2
N
Method
2B
SO2
OH
Cl
6
7
Cl
Scheme 1. General synthesis of N-benzyl-N-benzenesulfonamide derivatives. Reagents: (a) hexamethyldisilazane, chlorotrimethylsilane, xylene, 95%;
(b) alkyl aldehyde or benzaldehyde, ZnCl2, NaCNBH4, MeOH, 40–80%; (c) 4-chlorobenzenesulfonyl chloride, Et3N, CH2Cl2, 92%; (d) alkyl or
benzyl halide, K2CO3, DMF, 30–88%; (e) triphenylphosphine, 4-(tert-butyldimethylsilyloxy) benzyl alcohol, diethylazodicarboxylate, THF, 58%; (f)
TBAF, THF, 72%.
Table 1. Inhibitory activities of caprolactam enantiomers
H
N
O
R1
N
O2S
R2
Ex. #
R1
R2
Method
2A
Isomer
D/L
Ab40 IC50 (nM)a
1
120
OCH2CH3
OCH2CH3
Br
Br
Br
Cl
Cl
8
9
2A
2A
1
S
3400
13
R
S
OCH2CH3
O
10
11
3200
9.7
O
O
1
R
O
a Values are means of four experiments, with 12 drug concentrations in each exp.; intra-assay variance <10%.6

5792
M. F. Parker et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5790–5795
We next probed the effect of the lactam ring size11 on
potency. As exemplified by compounds 12, 13, and 14
in Figure 2, both the 5- and 6-membered lactams were
significantly less potent than the corresponding over
the 7-membered ring 14. As a result of these findings,
subsequent analogs were focused on the caprolactam
core (Fig. 2).
tion with an alkyl group as in 18 was quite detrimental
to potency.
Our strategy was then to examine the nature of the sub-
stituents at R1 on the sulfonamide nitrogen while keep-
ing 4-chlorophenylsulfonamide constant. Compounds
shown in Table 3 result from examination of various
benzylic substituents, while compounds in Table 4 illus-
trate a variety of non-benzylic groups. As shown in
Table 3, para-substitution on the benzyl side chain
generally improved potency when compared to the
unsubstituted ring system 22. Strongly electron-donat-
ing para-substituents conferred the greatest potency, as
seen with the methoxyphenyl analog 23 (IC50 = 2.8 nM)
and the phenol 7 (IC50 = 5.0 nM). While electron-rich
systems were generally more potent (e.g., aniline 25,
Modification of the aryl moiety attached to the sulfon-
amide sulfur atom was also undertaken. As shown in
Table 2, the 4-chlorobenzenesulfonamide (e.g., 16 and
21) was optimal for potency. Introduction of a heteroa-
ryl sulfonamide, as in the thiophene analog 17, led to a
50-fold loss in potency. Substitution of the 4-chloro by
fluorine (e.g., 15 and 20) or removal of the chlorine as
in 19 resulted in a modest decrease in potency. Substitu-
H
H
O
N
O
N
O
HN
F
F
F
N
N
N
SO2
SO2
SO2
O
O
O
Cl
Cl
Cl
12
13
14
Aβ 40 IC50 = 1100 nM
Aβ 40 IC50 = 250 nM
Aβ 40 IC50 = 9.1 nM
Figure 2.
Table 2. Modification of the aryl sulfonamide moiety
H
N
O
R1
N
O2S
R2
Ex. #
R1
R2
Method
2A
Ab40 (IC50 nM)a
9
13
OCH2CH3
OCH2CH3
OCH2CH3
OCH2CH3
OCH2CH3
CF3
Br
F
15
16
17
18
19
20
21
1
57
9.0
2A
1
Cl
S
830
7700
180
25
1
1
1
CF3
F
1
21
CF3
Cl
a Values are means of four experiments, with 12 drug concentrations in each exp.; intra-assay variance <10%.

M. F. Parker et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5790–5795
Table 3. Inhibitory activities of R1 benzylic substituents
5793
Table 3 (continued)
R1
Method
2A
Ab40 (IC50 nM)a
H
N
Ex. #
O
OMe
F
R1
34
93
29
64
60
N
SO2
35
36
37
2A
2A
2A
Cl
Ex. #
R1
Method
1
Ab40 (IC50 nM)a
CN
CF3
22
65
23
7
1
2.8
5.0
13
OCH3
38
39
2A
2A
820
660
MeO
2B
2A
1
CF3
OH
24
25
26
27
28
29
30
31
32
21
14
11
33
CF3
F
C(CH3)3
N(CH3)2
40
41
2A
2A
130
550
11
CF3
Cl
2A
2A
2A
2A
2A
2A
2A
1
76
(CH2)3CH3
F
Cl
18
OMe
F
O
F
42
43
2A
2A
130
67
45
OMe
OCF3
F
F
33
N
44
45
46
47
48
49
1
460
120
340
27
12
Cl
N
1
25
SO2CF3
CN
41
1
N
N
21
1
CF3
F
Cl
1
9.1
9.7
15
2B
2B
690
4500
OCH3
O
OMe
1
O
F
Cl
2A
a Values are means of four experiments, with 12 drug concentrations in
each exp.; intra-assay variance <10%.
Cl

5794
M. F. Parker et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5790–5795
IC50 = 11 nM; tert-butylphenyl 24, IC50 = 12 nM),
electron-withdrawing groups such as CF3SO2 (28,
IC50 = 25 nM) and CN (32, IC50 = 41 nM) were also
tolerated. Combination of a para-OR group with a
meta-substituent was also acceptable as seen with
the 3-fluoro-4-methoxy phenyl analog 14 (IC50 =
9.1 nM) and the methylenedioxyphenyl derivative 11
(IC50 = 9.7 nM). By comparison, single meta-substitu-
tion of the benzylic group was generally less favored
than para-substitution as illustrated by comparison of
23/34, 32/36, and 21/37. Ortho-substitution was signifi-
cantly disfavored as seen with compounds 38 and 39.
Disubstituted benzyl groups lacking para-substituents
(e.g., 40–43) were also less potent.
Table 4. Inhibitory activities of R1 alkyl substituents
H
N
O
In order to improve solubility, we examined a series of
pyridylmethyl side chains. The 3-pyridyl analog 45 was
more potent than the 2-pyridyl 44 and the 4-pyridyl 46
isomers. Attenuating the pKa of the pyridyl ring by add-
ing a 4-chloro substitution as in 47 had a positive effect
on binding. With the exception of 47, the pyridyl series
was significantly less potent than the simple methoxy- or
methylenedioxy phenyl compounds. Finally, branching
at the benzylic position was examined. Addition of an
a-methyl group, as shown in examples 48 and 49, re-
sulted in a significant loss of potency.
R1
N
SO2
Cl
Ex. # R1
Method Ab40
(IC50 nM)a
50
2A
1400
51
52
53
54
55
56
57
58
59
60
61
1
620
110
85
Analogs with alkyl side chains on the sulfonamide nitro-
gen were also examined (Table 4). Elimination of the
aryl ring had a negative impact on the potency in this
series. Interestingly, we found that substantial length
was required to impart even modest activity in the sim-
ple alkyl series, as shown with compounds 50–53. In the
case of the phenyl alkyl derivatives, extending chain
length also improved potency, with the phenylpropyl
derivative 55 being the most potent in the series. How-
ever, further extension of the phenyl derivative chain
length led to a decrease in potency. Capping the termi-
nus of the alkyl side chain with functional groups such
as CF3, CN, Cl or CO2H (58–61, respectively) did not
improve potency to the level of benzylic substituents,
although the trifluoromethylpropyl derivative 60 had
better potency than the simple propyl derivative 50.
Even the best compound in the alkyl series was still
10-fold less active than in the benzyl series.
2A
2A
2A
2A
2A
2A
2A
2A
2B
2B
1100
67
470
1000
330
180
150
3900
To further assess the caprolactam series, the methoxy-
phenyl (23) and the dimethylaniline (25) were evaluated
in vivo13 for the reduction of brain concentrations of
Ab. The compounds were selected for further study
based on their potency in the Ab40 in vitro assay.6
Transgenic mice (Tg2576 mice,14 3–6 months of age)
were treated by oral gavage with test compounds at a
dose of 200 lmol/kg. The effects of 23 and 25 on Ab
in brain and plasma concentrations and relative
compound concentrations at 3 h post dose are shown
in Table 5. Both compounds 23 and 25 are highly pro-
tein bound (>97%) but were found to diminish brain
and plasma concentrations of Ab as measured by a stan-
dard ELISA for Ab40.6 Compound concentrations were
analyzed by LC–MS–MS method. Although 23 is ꢀ4-
CF3
CN
Cl
CO2H
*b
a Values are means of four experiments, with 12 drug concentrations in
each exp.; intra-assay variance <10%.
b Synthesized from the bromo ester, followed by saponification to yield
the acid 61.
Table 5. In vivo efficacy of potent caprolactam sulfonamides
Ex. #
Ab40 IC50 (nM)
Ab42 IC50 (nM)
% inhibition brain13 (concn)
% inhibition plasma13 (concn)
B/P ratio
23
25
2.8
11
2.4
7.8
49% (2200 nM)
52% (4200 nM)
47% (280 nM)
61% (1600 nM )
7
2.8

M. F. Parker et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5790–5795
5795
pared by dilution into 40 mM Tris–HCl (pH 7.4) with
0.2% BSA and added to assay plates. For the Ab40
measurements, antibodies specific for the Ab40 neoepitope
(TSD, developed at BMS; conjugated to the Wallac
reagent (Perkin-Elmer)) and 26D6 as described above
were mixed and 20 ll of the mixture was added to the 10 ll
aliquots which had been removed previously from the cell
plate yielding a final concentration of 1.6 ng/well TSD and
17.5 ng/well 26D6. Assay plates containing antibodies
were sealed with aluminum foil and incubated overnight at
4 ꢁC. Signal was determined using a Viewlux counter
(Perkin-Elmer) and IC50 values determined using curve
fitting in CurveMaster (Excel-Fit based data analysis
package). Also, see Smith, D. W.; Munoz, B.; Srinirvasan,
K.; Bergstrom, C. P.; Chaturvedula, P. V.; Deshpande, M.
S.; Keavy, D. J.; Lau, W. Y.; Parker, M. F.; Sloan, C. P.;
Wallace, O. B.; Wang, H. H. WO 0050391.
fold more potent than the aniline 25, 25 has higher rel-
ative brain and plasma exposures. This may explain
why the in vivo potencies of these compounds are
comparable.
In summary, potent c-secretase inhibitors emerged from
SAR studies and optimization of an initial amino capro-
lactam screening hit. The most potent compounds to
emerge from this study were the 4-methoxy benzyl com-
pound 23 and the N,N-dimethyl analog 25. These com-
pounds were approximately 50-fold more potent than
the original screening hit and markedly reduced the con-
centration of Ab in the brain and plasma in transgenic
Tg2576 mice. However, 23 and 25 were also found to
be nanomolar inhibitors of the human CYP450 isoforms
3A4 and 2C19. Further investigations are focused on the
removal of these liabilities from this and similar sulfon-
amide series.
7. Roberts, S. B.; Hendrick, J. P.; Vinitsky, A.; Lewis, M.;
Smith, D. W.; Pak, R. Isolation of functionally active c-
secretase presenilin 1 complex and fluorescence assay for
c-secretase activity and inhibitors. US 2004, 0121411 A1.
8. Purchased from ComGenex Corporation as part of a
collection of sulfonamides.
References and notes
9. Recently disclosed in a patent, Martin, N.; preparation of
N-substituted heterocyclic sulfonamides for treating cog-
nitive disorders. US 2006/0270657A1.
1. Scheff, S. W.; Price, D. A.; Schmitt, F. A.; Mufson, E. J.
Neurobiol. Aging 2006, 27, 1372.
10. Pellegata, R.; Pinza, G.; Pifferi, G. Synthesis 1978, 614.
11. The 3 amino-2-piperidone was synthesized from D-orni-
thine following the method of Pellegata (9). The 3-amino-
2-pyrrolidinone was purchased from Ramidus AB.
12. Possible clarifications for the 10-fold increase in activity
include but are not limited to: antagonism or partial
agonist activity by the inactive enantiomer, chiral ligand
regulated enzyme activity, allosteric binding by one of the
enantiomers or even relative error in the binding assay.
These hypothetical explanations are being explored
further.
2. (a) Clark, C. M.; Karlawish, J. H. Ann. Intern. Med. 2003,
138, 400; (b) Hardy, J.; Higgins, G. A. Science 1992, 256,
184.
3. Nguyen, J.; Yamani, A.; Kiso, Y. Curr. Pharm. Des. 2006,
12, 4295.
4. Walsh, D. M.; Klyubin, I.; Fadeeva, J. V.; Cullen, W. K.;
Anwyl, R.; Wolfe, S.; Rowan, M. J.; Delkow, D. J. Nature
2002, 416, 535.
5. Selkow, D. J.; Podlisny, M. B.; Joachim, C. l.; Vickers, E.
A.; Lee, G.; Fritz, L. C.; ltersdorf, T. Proc. Natl. Acad.
Sci. U.S.A. 1988, 85, 7341.
13. Determination of Ab, full length APP, and APP carboxy-
terminal fragments, brain tissues was performed following
the procedure as described previously. Barten, D. M.;
Guss, V. L.; Corsa, J. A.; Loo, A.; Hansel, S. B.; Zheng,
B.; Munoz, K.; Srinivasan, K.; Wang, B.; Roberson, B. J.;
Polson, C. T.; Wang, J.; Roberts, S. B.; Hendrick, J. P.;
Anderson, J. J.; Loy, J. K.; Denton, R.; Verdoorn, T. A.;
Smith, D. W.; Felsenstein, K. M. J. Pharmacol. Exp. Ther.
2005, 312, 635.
6. H4 cells stably transfected with APP751 containing the
Swedish mutation (H4 APP751 SWE clone 8.20, devel-
oped at BMS) were maintained in log phase through twice
weekly passage at a 1:20 split. For IC50 determinations,
1.5 · 104 cells/well in DMEM containing 0.0125% BSA
(Sigma A8412) were plated directly into 384-well com-
pound plates (Costar 3709) containing 0.1 ll serially
diluted compound in DMSO. Following incubation for
19 h in 5% CO2 at 37 ꢁC, plates were briefly centrifuged
(1000 rpm, 5 min). A 10 ll aliquot from each well was
transferred to a second assay plate (Costar 3709) for Ab1–
40 measurements. Antibody cocktails were freshly pre-
14. Hsiao, K.; Chapman, P.; Nilsen, S.; Eckman, C.; Hari-
gaya, Y.; Younkin, S.; Yang, F.; Cole, G. Science 1996,
274, 99.