Substituted 2-oxo-azepane derivatives are potent, orally active γ-secretase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2007.10.074

A hydroxamic acid screening hit 1 was elaborated to 5,5-dimethyl-2-oxoazepane derivatives exhibiting low nanomolar inhibition of γ-secretase, a key proteolytic enzyme involved in Alzheimer's disease. Early ADME data showed a high metabolic clearance for t

Full text of DOI:10.1016/j.bmcl.2007.10.074

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 304–308
Substituted 2-oxo-azepane derivatives are potent, orally active
c-secretase inhibitors
Eric A. Kitas,a,* Guido Galley,a Roland Jakob-Roetne,a Alexander Flohr,a
a
a
b
c
c
´
Wolfgang Wostl, Harald Mauser, Andre M. Alker, Christian Czech, Laurence Ozmen,
Pascale David-Pierson,d Dieter Reinhardtc and Helmut Jacobsenc
aPharma Division, Discovery Chemistry, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
bPharma Division, Molecular Structure Research, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
cPharma Division, Preclinical CNS Research, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
dPharma Division, Safety Technical Services, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
Received 21 September 2007; revised 18 October 2007; accepted 22 October 2007
Available online 5 November 2007
Abstract—A hydroxamic acid screening hit 1 was elaborated to 5,5-dimethyl-2-oxoazepane derivatives exhibiting low nanomolar
inhibition of c-secretase, a key proteolytic enzyme involved in Alzheimer’s disease. Early ADME data showed a high metabolic
clearance for the geminal dimethyl analogs which could be overcome by replacement with the bioisosteric geminal difluoro group.
Synthesis and structure–activity relationship are discussed and in vivo active compounds are presented.
ꢀ 2007 Elsevier Ltd. All rights reserved.
Table 1. HTS hit 1 and representative early modifications
Alzheimer’s disease (AD) is characterized by the deposi-
tion in the brain of amyloid in extracellular plaques and
intracellular neurofibrillary tangles. The amyloid pla-
ques are composed of Ab peptides (mainly 40- and 42-
amino-acid length) which originate from b-amyloid pre-
cursor protein (APP) by a series of proteolytic cleavage
steps involving enzymes b-secretase and c-secretase.
Whereas b-secretase is a typical aspartyl protease, c-
secretase proteolytic activity consists of several proteins
including the presenilins, nicastrin, aph1, and pen-2.1
Among the known substrates of c-secretase are the
APP and the proteins of the Notch receptor family.2,3
According to the amyloid hypothesis of AD, the pro-
duction and deposition of Ab is the ultimate cause of
the disease.
O
O
O
O
N
S
R
O
Cl
Compound
R
Ab lowering IC50 (lM)
1
2
NHOH
NH2
2.6
0.008
The poor physicochemical properties (e.g., compound 2:
solubility <1 lg/ml; logD > 3; low PAMPA permeabil-
ity5) and relatively high clearance (e.g., compound 2: hu-
man/mouse 34/282 ll/min/mg protein) observed for
primary amides in this class led us to consider ring con-
strained structures (Table 2). Cyclic lactams 3–6 were pre-
pared, thus fusing the vectors corresponding to the
isobutyl and amide groups in 2. The azepinone (n = 3)
ring size was found to be optimal in this series with sulfon-
amide 5 having a modest inhibition of IC50 0.40 lM.6
In our program to identify potent and selective inhibi-
tors of c-secretase, a micromolar screening hit 1 was
identified from our propriatory compound library.
Hydroxamic acid 1 was elaborated to the primary amide
2, a nanomolar inhibitor of amyloid synthesis as deter-
mined by our cellular Ab lowering assay (Table 1).4
Keywords: Alzheimer’s disease; Amyloid Ab lowering; c-Secretase;
Proteolytic activity; Metabolism; In vivo.
*
Alternatives to the piperonyl northern vector were
examined and phenyl ring substitutions (Me, OMe
Corresponding author. Tel.: +41 61 688 76 18; fax: +41 61 688 64
59; e-mail: eric_a.kitas@roche.com
0960-894X/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.10.074

E. A. Kitas et al. / Bioorg. Med. Chem. Lett. 18 (2008) 304–308
305
Table 2. Lactam derivatives 3–16
para-chlorophenyl was a well-established pharmaco-
phore in sulfonamide or sulfone classes of c-secretase
inhibitors.7 Nevertheless, some variations will be dis-
cussed later (vide infra, Table 4).
R
O
O
N
NH
[ ]n
S
O
Cl
Next we considered the SAR at the azepinone ring in an
attempt to improve the modest inhibition seen so far. N-
Methyl substitution of the lactam gave inactive com-
pounds (data not shown).
Compound
n
R
Ab lowering
IC50 (lM)
O
O
O
O
3
1
2
3
4
3
3
3
3
3
3
3
3
3
3
>20
14
An overlay of lactam 12 with the most potent acyclic
analog 2 (Fig. 1) shows the high similarity of both clas-
ses. Interestingly, the structural comparison suggests
that the lactams do not occupy all available space (indi-
cated by the cyan sphere). Thus, additional substitution
in 4- or 5-position was considered for improving binding
affinity.
O
O
O
O
4
5
0.40
9
Indeed the introduction of an isopropyl- or CF3-substi-
tuent (Table 3, compounds 17 and 18) in position 5 led
6
7
0.32
0.23
0.17
17
O
O
O
O
O
8
F
9 [R isomer]
F
F
10 [S isomer]
11
0.15
0.08
0.17
2
F
Figure 1. Overlay of acyclic HTS derivative 2 and lactam 12. The
X-ray structure8 of 15 of the lactam series was taken as reference for a
flexible alignment of 2 using the program Moloc (Gerber, P.;
H2N
O
12 [R isomer]
www.moloc.ch, 2007). For clarity the propionic acid side chain was
replaced by a primary amide leading to compound 12.
H
N
13
O
Table 3. Substituted 2-oxoazepane derivatives
Me2N
O
14
O
N
H
O
3
15 [R isomer]
0.14
0.9
HOOC
N
O
NH
S
O
N
7
Cl
16
O
5
Compound
Lactam subst. Ab lowering Cl (human/mouse
IC50 (lM)
microsomes)a
17
18
19
20
21
22
23
5-i-Pr[Rac]
5-CF3 [Rac]
5-t-Bu[Rac]
5,5-Me2
0.04
0.02
0.12
0.006
>20
51/203
19/39
93/110
45/191
—
OCF3, Cl, F, data not shown) indicated that larger para-
substituents were tolerated (12–15) and inhibitory activ-
ities could be improved by further addition of F to the
phenyl ring (8–11). The preferred R-stereochemistry
was established by preparing compounds 9 and 10
starting from (R)-3-amino-azepan-2-one and (S)-3-ami-
no-azepan-2-one, respectively. Furthermore, phenyl to
pyridyl replacement such as in compound 16 maintained
sub-micromolar affinity for c-secretase. Minimal varia-
tions to the aryl western vector were attempted as
7,7-Me2
6,6-Me2
4,4-Me2
3.5
—
—
0.39
0.015
0.004
0.30
24 [R isomer] 5,5-Me2
32 [R isomer] 5,5-F2
33 [R isomer] 6,6-F2
36/655
1.9/2.2
0.0/5.4
a Microsomal clearance (ll/min/mg protein).

306
E. A. Kitas et al. / Bioorg. Med. Chem. Lett. 18 (2008) 304–308
to stronger inhibition of c-secretase activity compared to
the corresponding unsubstituted derivative 13. This ef-
fect was less pronounced for the larger t-butyl derivative
19. A very potent compound could be obtained with
geminal dimethyl substitution 20 and also simplified
the structure by the removal of one chiral center. Start-
ing from 20 as the least complex structure, a ‘gem.di-
methyl walk’ around the caprolactam (compounds 20–
23) showed that optimal substitution was in fact the 5-
position.
potency of the chiral 5,5-dimethyl analog 24 in the Ab
lowering assay (Table 3).
The synthesis of fluoro ring-substituted 3-amino-aze-
pan-2-ones is described in Scheme 2. (R)-N Boc-allylgly-
cine was coupled to allyl-(2,4-dimethoxybenzyl)amine to
yield the bis-allyl intermediate 34.12 Ring closure
metathesis reaction preferably using the 2nd generation
Grubb’s catalyst (0.1 equiv) proceeded efficiently with-
out the need for high dilution affording 35 in 80% yield.
The chiral integrity of the cyclization product was as-
sessed by its further reaction to yield known compound
36 where respective optical rotations could be com-
pared,13 confirming the mildness of the olefin metathesis
conditions. Alkene 35 was oxidized to the 5,6-oxo
regioisomers using a palladium complex in a saturated
oxygen environment.
The synthesis of 24 (the active R-isomer of 20) is de-
scribed in Scheme 1 and began with the hydrogenation
of a,b-unsaturated cyclohexenone 25 to yield 26 which
was subjected to a Beckmann rearrangement using
hydroxylamine-o-sulfonic acid in formic acid.9 Selective
dibromination in position 3 of the azepinone followed
by monodebromination under hydrogenolytic condi-
tions afforded 30.10 Sodium azide displacement of Br
and reduction of azide to amine was followed by prepa-
ration of sulfonamide 31 as a mixture of 1:1 enantiomers
which were separated by chiral chromatography. Fur-
ther derivatization with 4-chloromethyl-N-cyclopropyl-
benzamide under basic conditions gave compound 24.
The active isomer was tentatively assigned with the
R-configuration by comparing Ab lowering activities
of the enantiomer pair 9 and 10.
The regioisomeric mixture 37a,b was used directly in the
further transformation to geminal difluorinated com-
pounds using DAST (3 equiv) in a modest 45% overall
yield and following silica gel chromatography the pre-
ferred isomer 38 was isolated in a 22% yield. In a
straightforward manner, further homologation of 38
yielded the potent inhibitor 32.14
Compound 32 was profiled in vivo in a transgenic
mouse model for inhibition of c-secretase activity15
and was found to be inactive after 3 · 20 mg/kg po
and ip administration. We attributed this finding to
possible low brain penetration due to the presence
of two amides in its structure. We therefore went back
to methoxyaryl northern substituents that we already
With the first potent c-secretase inhibitors in hand, we
initiated early ADME profiling. Human/mouse micro-
somal clearance was relatively high for 5-alkyl substi-
tuted derivatives 17, 19, 20, when compared to 18,
suggesting that alkyl substituents were labile to micro-
somal oxidation. Our attempts at blocking this liability
led us to the surprising result that the 5,5-difluoro and
6,6-difluoro analogs yielded highly bioavailable inhibi-
tors 32 and 33, respectively, with 32 matching the
O
O
H
O
H
H
N
DMB
O
N
N
DMB
Boc
N
i
S
Boc
NH
N
ii,v
O
Cl
34
35
iii
36
O
O
N
O
H
N
H
O
O
N
DMB
DMB
Boc
Boc
iv
N
NH
+
i
ii
iii
37a+b
O
O
25
26
27
O
O
O
F
O
O
H
N
H
N
H
O
O
O
N
Br
DMB
DMB
DMB
Br
N3
Boc
Boc
S
N
NH
N
N
vi
iv
v
v
NH
NH
Br
+
O
Cl
F
F
F
F
F
30
28
29
40
39
38
O
O
H
N
O
O
N
O
F
H
H
H
S
NH
vii
O
N
N
N
N
S
vi
O
vii
NH
H
O
O
S
O
O
N
Cl
O
S
NH
Cl
24
Cl
F
31
O
Cl
F
41
32
F
Scheme 1. Reagents and conditions: (i) H2, Pd–C in EtOH, 95%; (ii)
hydroxylamine-o-sulfonic acid in formic acid, reflux, 2 h, 55%; (iii)
PCl5, ZnCl2, Br2 4 h, 67 %; (iv) H2 Pd–C, NaOAc in AcOH, 71%; (v)
NaN3 in DMSO, 80 ꢁC, 1.5 h, 21%; (vi) H2 Pd–C then 4-chloro-
benzenesulfonyl chloride, DIEA, 1.5 h 67%; (vii) chiral separation on
Chiralpak AD 20% isopropanol, 80% heptane, then 4-chloromethyl-N-
cyclopropyl-benzamide, K2CO3 KI, 65 ꢁC 2.5 h, 31%.11
Scheme 2. Reagents and conditions: (i) Grubb’s II cat., reflux, 1.5 h,
MeCl2, [0.03 M] 80%; (ii) H2 Pd–C, EtOH then TFA; (iii) O2,
PdCl2,CuCl, DMF/water, 50 ꢁC, 2 d, 72%; (iv) DAST, MeCl2, 5 h, 45%
overall; (v) separation from 39 then HCl/1,4-dioxane, 2 h then 4-
chlorobenzene-sulfonyl chloride, DIEA in MeCl2, 2 h, 60%; (vi) TFA/
triethylsilane/water, 3 d, 40%; (vii) 4-chloromethyl-N-cyclopropyl-
benzamide, K2CO3, KI in DMF 65 ꢁC, 1.5 h.

E. A. Kitas et al. / Bioorg. Med. Chem. Lett. 18 (2008) 304–308
Table 4. 5,5-Difluoro-substituted 2-oxoazepane derivatives
307
R2
O
O
N
S
O
NH
R1
F
F
Compound
R1
R2
Ab lowering IC50 (lM)
Cl (human/mouse microsomes)a
O
42
pCl-Ph
0.02
13/105
19/121
9/72
—
F
F
O
43
44
45
46
pCl-Ph
0.002
0.03
0.15
0.11
F
N
N
N
pCl-Ph
O
O
O
CF3(CH2)2
—
S
Cl
a Microsomal clearance (ll/min/mg protein).
found active with the non-fluorinated caprolactams
(vide supra, Table 2). The methoxyphenyl derivative
42 was less potent than 32 but inhibitory activity
could be further improved by addition of another
fluorine (Table 4). In fact, compounds 43 and 44 were
found to be active in vivo at a minimal effective dose
(MED) of 20 mg/kg po. This relatively high dose may
be explained by the low brain exposure (typically
brain/plasma ratio <0.1) determined from the in vivo
experiments.16
Brun, Vittorio Bona, Christophe Schweitzer, and Mark-
us Haenggi is gratefully acknowledged.
Supplementary data
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.bmcl.2007.
10.074.
A few examples at replacing the para-chlorophenyl moi-
ety were prepared and trifluoropropyl 45 and the 5-chlo-
rothiophene 46 derivatives were identified as potential
alternatives.
References and notes
1. Hardy, J.; Selkoe, D. J. Science 2002, 297, 353.
2. (a) Radtke, R. et al. Nat. Immunol. 2004, 5, 247; (b)
Radtke, F.; Clevers, H. Science 2005, 307, 1904.
3. Shih, Ie-Ming; Wang, Tian-Li Cancer Res. 2007, 67,
1879.
In conclusion, we have identified a series of 5,5-difluoro-
substituted 2-oxoazepane derivatives which are potent
inhibitors of c-secretase. Difluoro-substitution was
found to be a key factor to ensure both potency and
metabolic stability of the compounds. Further variation
of substituents yielded compounds 43 and 44 which were
orally active in a transgenic mouse model for AD. Low
brain exposure possibly related to observed P-glycopro-
tein interactions for representative compounds,17 is a
remaining issue that needs to be addressed for this
otherwise attractive series.
4. Description of Ab lowering assay: HEK293 cells which
express a human bAPP cDNA were treated for 16–18 h
with compound. The supernatant was harvested and the
Ab 40 peptide concentrations were determined. Ninety-six
well ELISA plates (e.g., Nunc MaxiSorb) were coated with
monoclonal antibody which specifically recognizes the C-
terminal end of Ab40 (Brockhaus et al., NeuroReport
1998, 9, 1481). After blocking of non-specific binding sites
with, e.g., 1% BSA and washing, the culture supernatants
were added in suitable dilutions together with a horserad-
ish peroxidase-coupled Ab-detection antibody (e.g., anti-
body 4G8, Senetek, Maryland Heights, MO) and
incubated for 5–7 h. Subsequently the wells of the micro-
titer plate were washed extensively with Tris-buffered
saline containing 0.05% Tween 20 and the assay was
developed with tetramethylbenzidine/H2O2 in citric acid
buffer. After stopping the reaction with one volume 1 N
H2SO4 the reaction was measured in an ELISA reader at
450 nm wavelength. In repeated measurements the stan-
Acknowledgments
The excellent technical assistance of Patrick Studer,
Betty Hennequin, Annick Goergler, Roman Hutter, Pe-
ter Weber, Joseph Wuhrlin, Daniel Zimmerli, Fabienne
Goepfert, Patrick Biry, Remy Wybrecht, Marie Elise

308
E. A. Kitas et al. / Bioorg. Med. Chem. Lett. 18 (2008) 304–308
dard deviation for the calculated IC50 was 30%. Measured
cDNA and a human FAD presenilin 2 cDNA. See also
(a) Richards, J. G.; Higgins, G. A.; Quagazzal, A. M.;
Ozmen, L.; Kew, J. N.; Bohrmann, B.; Malherbe, P.;
Brockhaus, M.; Loetscher, H.; Czech, C.; Huber, G.;
Bluethmann, ; Jacobsen, H.; Kemp, J. A. J. Neurosci.
2003, 23, 8989; (b) Inhibition of c-secretase was assessed
by accumulation of its substrate, the C-terminal frag-
ment (CTFb) of APP released by the previous
b-secretase cleavage. The accumulation of CTFb in
total brain was quantified by Western blot (see reference
above Richards et al. for details). Compounds were first
tested at 20 mg/kg for efficacy followed by a dose–
response with 0.5–3–10–30 mg/kg, three animals per
dose. The well-characterized c-secretase inhibitor
LY411575 was used as reference compound. The CTFb
band was quantified by Chemo-luminescence and nor-
malized to an endogenous control protein band. In the
PS2APP transgenic mouse the expression of the human
APP transgene is driven by a neuron-specific form of
the Thy1 promoter. The peripheral level of APP and the
plasma level of Ab peptide are too low to allow a
meaningful determination of c-secretase inhibition; The
use of CTFb accumulation as a sensitive marker for c-
secretase inhibition has been described by others (c)
Zhao, Guojun; Cui, Mei-Zhen; Mao, Guozhang; Dong,
Yunzhou; Tan, Jianxin; Sun, Longsheng; Xu1, Xuemin
J. Biol. Chem. 2005, 280, 37689; (d) Walsh, D. M.;
Fadeeva, J. V.; LaVoie, M. J.; Paliga, K.; Eggert, S.;
Kimberly, W. T.; Wasco, W.; Selkoe, D. J. Biochemistry
2003, 42, 6664; (e) De Strooper, B.; Saftig, P.; Craessa-
erts, K.; Vanderstichele, H.; Guhde, G.; Annaert, W.;
Von Figura, K.; Van Leuven, F. Nature 1998, 391, 387.
16. Compound 44 was dosed at 20 mg/kg po and the brain/
plasma ratio was determined at 2 h: plasma conc. = 1046 g/
ml, brain conc. = 60 ng/g; b/p = 0.06; N = 3.
data were also in agreement with cell-free-assay data
obtained for selected compounds according to the proce-
dure by Li (Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 6138).
5. For a description of the PAMPA (parallel artificial mem-
brane permeability assay) model, a prediction assay for oral
absorption Kansy, M.; Fischer, H.; Kratzat, K.; Senner, F.;
Wagner, B.; Parrilla, I. Helv. Chim. Acta 2000, 447.
6. Other recent publications on N-(oxoazepanyl) benzene-
sulfonamides and related derivatives as c-secretase inhib-
itors for treating Alzheimer’s disease: Neitzel, M.; Dap-
pen, M. S.; Marugg, J. WO Patent WO2005042489, 2005;
Parker, M. F.; Bronson, J. J.; Barten, D. M. et al. Bioorg.
Med. Chem. Lett. 2007, 17, 5790.
7. (a) Further literature exemplifying the sulfonamide and
sulfone classes of c-secretase inhibitors: Smith, D. W. WO
Patent WO2000050391, 2000; (b) Pineiro J. L.; Churcher
I.; Dinnell K. et al. WO patent WO2002081435, 2002; (c)
Parker M. F.; McElhone K. E.; Mate R. A. et al. WO
patent WO20-03053912, 2003; (d) Josien H. B. WO patent
WO2003013527, 2003; (e) Churcher, I.; Harrison, T.;
Kerrad, S. et al. WO patent WO2004031137, 2004; (f)
Brands K. M.; Davies A. J.; Oakley P. J. et al.
WO2004013090, 2004; (g) Churcher, I.; Beher, D.; Best,
J. D., et al. Bioorg. Med. Chem. Lett. 2006, 16, 280; (h)
Asberom, T.; Zhao, Z.; Bara, T. A. Bioorg. Med. Chem.
Lett. 2007, 17, 511.
8. Full crystallographic data for compound 15 have been
deposited with the Cambridge Crystallographic Data
Center (CCDC Reference No. 661530). Copies of the
data can be obtained free of charge via the internet at
http://www.ccdc.cam.ac.uk.
9. Olah, G. A.; Fung, A. P. Synthesis 1979, 537.
10. Shirota, F.; Nagasawa, H. T.; Elberling, J. A. J. Med.
Chem. 1977, 20, 1623.
11. For detailed experimental conditions, see: Galley, G.;
Jakob-Roetne, R.; Kitas, E. A. WO Patent
WO2006005486, 2006.
12. Analogous RCM in synthesis of 3-aminoazocan-2-one
derivatives Creighton, C. J.; Leo, G. C.; Du, Y.; Reitz, A.
B. Bioorg. Med. Chem. 2004, 12, 4375.
17. The permeability and the P-gp substrate properties of 44
were assessed by a bi-directional transport assay in an
epithelial cell monolayer system, using LLC-PK1 cells
exogenously expressing human P-glycoprotein (P-gp),
MDR1. A high passive permeability of around 280 nm/s
was estimated based on the observed apical to basolateral
and reverse permeabilities in the different experiments
where active transport was inhibited. Directional trans-
port (at a concentration of 1 lM) was observed in cells
expressing the human MDR1 transporter with an export
permeability ratio of 2.2, while in presence of an inhibitor
(Elacridar, 5 lM) this transport was fully inhibited (export
ratio = 1). These results indicate that 44 is a substrate for
human MDR1 P-glycoprotein.
13. Compound 36 [a]D À130.6ꢁ (c = 0.64, CHCl3) was also
prepared using commercial (R)-3-amino-azepan-2-one:
[a]D À130.9ꢁ (c = 0.89, CHCl3).
14. For detailed experimental conditions, see: Flohr, A.;
Galley, G.; Jakob-Roetne, R.; Kitas, E. A.; Wostl, W.
US Patent US20070037789, 2007.
15. Compounds were profiled in vivo in the PS2APP
transgenic mouse which expresses the human APPsw