Novel orally active, dibenzazepinone-based γ-secretase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2007.07.078

Structural modifications of the γ-secretase inhibitor, LY411575, led to a malonamide analogue (S),(S)-1 with potent inhibitory activity in vitro, but disappointing activity in a mouse model of Alzheimer's disease. Identification and replacement of a metab

Full text of DOI:10.1016/j.bmcl.2007.07.078

Bioorganic & Medicinal Chemistry Letters 17 (2007) 5918–5923
Novel orally active, dibenzazepinone-based c-secretase inhibitors
a,
b
b
Jens-Uwe Peters,a,* Guido Galley, Helmut Jacobsen, Christian Czech,
*
Pascale David-Pierson,c Eric A. Kitasa and Laurence Ozmenb
aPharma Division, Discovery Chemistry, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
bPharma Division, Preclinical CNS Research, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
cPharma Division, Safety Technical Services, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
Received 5 July 2007; revised 23 July 2007; accepted 24 July 2007
Available online 22 August 2007
Abstract—Structural modifications of the c-secretase inhibitor, LY411575, led to a malonamide analogue (S),(S)-1 with potent
inhibitory activity in vitro, but disappointing activity in a mouse model of Alzheimer’s disease. Identification and replacement of
a metabolically labile position provided an improved compound (R/S),(S)-13 with high in vitro activity (IC50 = 1.7 nM), and
in vivo activity after oral administration (MED = 3 mg/kg). Further modifications gave an equipotent carbamate analogue 14 with
improved molecular properties.
Ó 2007 Elsevier Ltd. All rights reserved.
Alzheimer’s disease (AD) is an invariably fatal neurode-
generative disease associated with progressive dementia,
and the deterioration of the patient’s memory and per-
sonality.1 Current AD therapies can achieve a limited
improvement of the patient’s declining cognitive func-
tions, but a causal therapy for this devastating disease
is not yet available.
inhibition of c-secretase has proven to be a better trac-
table target for drug discovery.5 One of the first dis-
closed c-secretase inhibitors with high activity and
drug-like properties was LY411575,6 a compound which
efficaciously inhibited Ab production in animal models
of AD after oral administration.7
To evaluate the potential of LY411575 as a lead struc-
ture, we prepared a number of close analogues, which
we tested (a) in a HEK293 reporter-gene assay (‘cellu-
lar’) based on a construct of the c-secretase substrate
Notch 1 fused with a Gal4/VP16-transcription factor
which activates a firefly luciferase reporter-gene,8 and
(b) in a second assay based on an ELISA technique9
measuring Ab40 peptide produced from a recombinant
APP C-terminal fragment by HEK293 derived mem-
brane fractions containing c-secretase activity (‘cell-
free’).10 Among the tested compounds, we identified
malonamide 1 (Fig. 1) as a c-secretase inhibitor with
good potency in both assays (Table 1). 2 was identified
Although AD was first described 100 years ago,2 the ex-
act mechanism of its pathogenesis is still not completely
known. A hallmark of AD is the formation of cortical
plaques consisting mainly of Ab peptides, and it is
widely accepted that the Ab peptides and their oligomers
play a causal role in the pathology of the disease.3 Ab
peptides are produced from b-Amyloid Precursor
Protein (APP) by sequential cleavage by two proteases,
b-secretase and c-secretase. In a first proteolytic step,
b-secretase produces a C-terminal fragment of APP
(CTFb). In a subsequent step, c-secretase cleaves the
CTFb and produces Ab peptides of varying lengths.
Inhibition of b-secretase or c-secretase should thus
provide a causal therapy for AD. Whereas inhibitors
of b-secretase with the potential to cross the blood-
brain-barrier have remained elusive for a long time,4
O
OH
H
N
F
N
N
H
O
Keywords: Alzheimer’s Disease; Gamma secretase; Inhibitor;
LY411575; Dibenzazepinone.
O
F
*
Corresponding authors. Tel.: +41 61 6882636 (J.-U.P.); tel.: +41 61
6880437 (G.G.); e-mail addresses: jens-uwe.peters@roche.com;
guido.galley@roche.com
LY411575
Figure 1. Elan/Lilly’s c-secretase inhibitor, LY411575.
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.07.078

J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5918–5923
5919
Table 1. Malonamide 1 was identified as a potent c-secretase inhibitor
with low microsomal stability
reside mainly in (R/S),(S)-1, in which the stereogenic
centre of the dibenzazepinone ring was of the same con-
figuration as in LY411575. Using preparative HPLC,
(R/S),(S)-1 was then separated into epimers, which were
obtained in a ꢀ1:1 ratio. Both epimers had single-digit
nanomolar activities in the two assays; the malonamide
stereocentres were arbitrarily defined. The dimethyl
substituted or unsubstituted malonamide analogues
were found to be less active in our in vitro assays (data
not shown).
O
O
F
N
N
N
H
H
O
1
F
O
H
We anticipated that the malonic acid stereogenic centre
of (S),(S)-1 and (R),(S)-1 would be susceptible to epi-
merisation, but we found practically no epimerisation
(<0.5%) or racemisation in DMSO solutions of
(S),(S)-1 and (R),(S)-1 during 10 days. This surprising
stability was corroborated by deuteration experiments
with 1 in d6-DMSO, to which we had added a few drops
of D2O or DCl. No deuterium was incorporated into the
malonamide a-position of 1, which suggested that the
kinetics of a keto–enol equilibrium as a prerequisite
for epimerisation are very slow in DMSO, even under
acidic conditions.
N
O
N
N
H
O
O
2
IC50 (cellular) IC50 (cell free) MABa
MABa (human)
(nM)
[nM]
(rat) [%] [%]
1
2
10
1500
22
2700
9
10
Not determined
Carbamate 2 was indentified as another analogue of LY411575 with
however reduced inhibitory activity. Both 1 and 2 were tested as
mixtures of diastereomers.
a Maximal achievable bioavailability estimated from incubations with
liver microsomes.
LY411575 and (S),(S)-1 were evaluated in a transgenic
mouse model of AD which expresses a cDNA of the hu-
man APPsw and a human FAD presenilin 2.11 After
subchronic administration of the test compounds, the
CTFb/APP ratio in the brain homogenate was deter-
mined by Western blot analysis and was used as a read-
out for inhibition of c-secretase activity. Inhibition of
c-secretase leads to an accumulation of CTF, and thus
to an increased CTF–APP ratio (see introductory para-
graphs). Whereas LY411575 increased the CTF–APP
ratio in this experiment after an administration of 5
doses of 20 mg/kg po over 2 days, (S),(S)-1 showed an
only weak activity after 3 doses of 50 mg/kg ip.
as another compound with relatively poor activity,
which should however become important during the fur-
ther course of our studies (vide infra).
To evaluate the importance of the stereogenic centres of
1, and also to assess the stability of the diastereomers to-
wards epimerisation, we prepared (R/S),(S)-1 and (R/
S),(R)-1 (Table 2) from chiral dibenzazepinone building
blocks (vide infra). The biologic activity was found to
The permeability through an artificial membrane was
high for (S),(S)-1 and similar compounds (PAMPA,12
>106 cm sÀ1). The weak activity of (S),(S)-1 in vivo was
thus attributed to a low metabolic stability, as predicted
by a low stability of 1 in rat and human liver microsome
Table 2. A separation of diastereomers revealed that the biologic
activity resides in the (S),(S) and (R),(S) diastereomers of 1
O
O
O
O
putative site of
oxidative metabolism
N
N
Ar
N
H
N
Ar
N
H
N
H
H
O
O
O
O
(R/
S
),(S
)-1
(R/S),( )-1
R
F
N
N
H
N
H
O
1
F
O
O
O
O
rat liver
microsomes
N
N
Ar
N
H
N
Ar
N
H
N
*
*
H
H
O
O
( ),( )-1
(R),( )-1
S
S
S
IC50 (cellular) [nM]
IC50 (cell-free) [nM]
O
O
(R/S),(S)-1
(R/S),(R)-1
(S),(S)-1
8.8
580
1.6
7.1
3.1
620
2.0
8.7
N
N
H2N
H
O
3
(R),(S)-1
Significant epimerisation or racemisation of the diastereomers was not
observed (see text).
*Stereochemistry arbitrarily assigned.
Figure 2. Compound 3 was identified as the main metabolite from an
incubation of 1 with rat liver microsomes. This metabolite suggests
oxidative metabolism in the benzylic position of 1.

5920
J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5918–5923
Table 3. Inhibitory activities and microsomal stabilities of malonamides 3–13 (diastereomeric mixtures)
O
O
R
N
N
N
H
H
O
R–NH@
IC50 (cellular) [nM]
IC50 (cell-free) [nM]
5.3
MABa (rat) [%]
MABa (human) [%]
3
4
>25
H2N
N
H
0.075
0.015
5
6.3
5.7
N
H
6
14
12
N
H
OH
7
>25
20
22
N
H
N
H
8
>20
3.0
9
14
O
9
3.3
N
H
10
11
12
13
0.21
0.042
0.024
0.002
0.09
0.02
0.037
12
37
18
55
N
H
F
N
F
H
F
F
F
F
N
H
F
F
F
N
H
F
F
Attempts to stabilise the metabolically labile benzylic position were not successful (5–8). Fluoroalkyl R groups were finally found as stable
replacements (11–13).
a Maximal achievable bioavailability estimated from incubations with liver microsomes.

J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5918–5923
5921
preparations (Table 1). An HPLC-MS/MS analysis of an
incubation of 1 with rat liver microsomes indicated 3 as
the main metabolite (Fig. 2). This finding points to an
oxidation of the benzylic position of 1 to form an unsta-
ble aminal (not shown) as a precursor of 3. Compound 3
has an only marginal inhibitory activity (Table 3).
From a chiral dibenzazepinone building block (vide in-
fra), we prepared (R/S),(S)-13 (Table 4). In contrast to
(R/S),(S)-1 (Table 2), a further separation of the epi-
meric mixture was unsuccessful. The unsuccessful or te-
dious separation of epimers by HPLC began to emerge
as an obstacle that might seriously hamper our future ef-
forts in this series. We recalled that with 2 (Table 1), we
had found that a carbamate in place of an amide was
tolerated. Although 2 was an only weak c-secretase
inhibitor, it offered the possibility to derive the stereo-
genic centre from the chiral pool, for example, lactic
acid, thus avoiding any separation of diastereomers dur-
ing the assembly of inhibitors from building blocks.
Simply changing malonamide 13 into its carbamate ana-
logue, 14, gave a c-secretase inhibitor that was equipo-
tent in both our in-vitro assays.
In an attempt to improve the low metabolic stability of 1,
we prepared analogues 5–8 (Table 3), with sterically
blocked benzylic positions. However, these compounds
had a much reduced inhibitory activity (compare with 4,
Table 3). Furthermore, when we assayed 8 for microsome
stability, we found that this compound was similarly
unstable as 1, although an oxidation of the quaternary
benzylic position is not possible. We therefore abandoned
the concept of stabilising the benzylic position and rather
looked for replacements of this motif. However, the diflu-
orobenzyl substituent proved quite critical for activity.
We prepared ꢀ50 compounds by parallel chemistry, most
of which were inactive and often had low microsomal sta-
bilities. Compounds 9 and 10 are examples which showed
marginal in-vitro activity. Interestingly, similar benzoaz-
epinone-derived, cyclic malonamides were very recently
disclosed to be potent c-secretase inhibitors with good
pharmacokinetic properties.13 Finally we prepared flu-
oro-alkyl derivative 11, which had an almost restored
inhibitory activity as compared to 1, and moreover a suf-
ficient metabolic stability. Follow-up compound 12 was
similarly active. Increasing the degree of fluorination led
to 13, which now even surpassed the inhibitory activity
of 1. For the ease of synthesis, we prepared and tested
3–13 as diastereomeric mixtures.
In comparison with malonamide 13, carbamate 14 had a
much improved solubility, and a further improved
microsomal stability, which translated into a reduced
clearance in vivo. The carbamate functionality is occa-
sionally encountered as a biodegradable motif, however,
14 was found to be stable with plasma and in buffer at a
pH of 1.7, 4.7 and 7.8. Only at a pH of 12.8, slow deg-
radation was observed (t1/2 = 57 h).
Both (R/S),(S)-13 and 14 were active in our in-vivo AD
model, and dose-dependently increased the CTF–APP
ratio after an oral administration of 3 doses of 3 mg/kg
over 2 days as a minimal effective dose.
Schemes 1–3 outline the syntheses of our key com-
pounds as representative procedures. Dibenzoxazepine
building block 19 was prepared in close analogy to the
method of Wu et al.6 We obtained substantially im-
proved yields when we cyclised the tertiary amide 17
(Scheme 1) instead of a secondary amide as described.
This finding can be rationalised by the assumption that
Table 4. Malonamide (R/S),(S)-13 and carbamate 14 emerged as the
best c-secretase inhibitors from the described studies
O
F
O
F
F
N
N
N
H
H
F
F
O
(R/S),(S)-13
a,b
HN
H2N
O
16
15
F
F
H
N
O
N
H
F
F
N
O
F
O
d
c
14
Cl
N
O
(R/S),(S)-13
14
17
IC50 (cellular) [nM]
IC50 (cell-free) [nM]
solubility [g/L]
1.7
3.0
28
1.7
8.6
290
70
e,f
MABa (human) [%]
MABa (mice) [%]
38
55
72
32
N
N
H2N
CL (mice) [ml/min/kg]
Fb [%]
53
21
O
O
23
0.5–0.9
3
18
19
Brain-to-plasma ratio
MEDc, po [mg/kg]
0.5–1.2
3
Scheme 1. Reagents and conditions: (a) ClCOOEt, K2CO3, THF/
H2O = 1.5/1, overnight rt, 93%; (b) RedAl, THF, 6 h 50 °C, then
overnight rt, 89%; (c) ClCH2COCl, DIPEA, CH2Cl2, 3 h rt, 71%; (d)
AlCl3, 90–180 °C, 2 h, 81%; (e) KHMDS, isoamyl nitrite, toluene, 1 h
0 °C, 65%; (f) H2 (2.4 bar), Pd/C, EtOH, 24 h rt, 83%.
a Maximal achievable bioavailability estimated from incubations with
liver microsomes.
b Mice, 7.8 mg/kg po.
c Transgenic mouse model of AD, see text.

5922
J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5918–5923
In summary, scaffold modifications of the c-secretase
inhibitor, LY411575, led to malonamide and carbamate
analogues. The initially low metabolic stability of these
compounds was greatly improved by the replacement
of the difluorobenzyl for a pentafluoropropyl substitu-
ent, as in malonamide (R/S),(S)-13 and carbamate 14
(Table 4). Both c-secretase inhibitors, (R/S),(S)-13 and
14, had single-digit nanomolar inhibitory activities
in vitro, and were efficacious in a mouse model of AD
after oral administration.
a,b
O
O
N
N
N
H
H2N
HO
O
O
(S)-19
20
c
F
O
O
F
F
N
N
H
N
H
F
F
O
(
R/S),(S)-13
Acknowledgements
Scheme 2. Reagents and conditions: (a) methylmalonic acid mono-
tert-butyl ester, HOBt, EDCI, DIPEA, THF, overnight rt, 48%; (b)
TFA/CH2Cl2 1:1, overnight rt 73%; (c) CF3CF2CH2NH2, TPTU,
DMF, overnight rt.
The excellent technical assistance of Annick Goergler,
´
Markus Haenggi, Roman Hutter, Stephane Kritter,
Dieter Reinhardt, Christophe Schweitzer and Silja We-
ber is gratefully acknowledged.
References and notes
a
O
HO
N
N
N
1. (a) Selkoe, D. J. Arch. Neurol. 2005, 62, 192; (b) Selkoe, D.
J. Ann. Intern. Med. 2004, 140, 627; (c) Plosker, G. L.;
Keating, G. M. Dis. Manage. Health. Outcomes 2004, 12,
55; (d) Ewbank, D. C. Am. J. Public Health 1999, 89, 90.
2. Alzheimer, A. Allgemeine Zeitschrift fu¨r Psychiatrie und
Psychisch-Gerichtliche Medizin 1907, 64, 146.
3. (a) Morgan, C.; Colombres, M.; Nunez, M. T.; Inestrosa,
N. C. Prog. Neurobiol. 2004, 74, 323; (b) Selkoe, D. J.
Science 2002, 297, 353; (c) Hardy, J. A.; Higgins, G. A.
Science 1992, 256, 184.
H2N
H
O
O
(S)-19
21
b
O
F
F
H
O
N
F
F
N
N
H
O
F
O
14
4. Durham, T. B.; Shepherd, T. A. Curr. Opin. Drug Discov.
Devel. 2006, 9, 776.
Scheme 3. Reagents and conditions: (a) L-(+)-lactic acid, HOBt,
EDCI, DIPEA, THF, 0 °C, then overnight rt ꢀquant.; (b) 4-
nitrophenyl chloroformate, Et3N, toluene, overnight rt, then
CF3CF2CH2NH2, overnight rt.
5. (a) Churcher, I.; Beher, D. Curr. Pharm. Des. 2005, 11,
3363; (b) Kornilova, A. Y.; Wolfe, M. S. Annu. Rep. Med.
Chem. 2003, 38, 41; (c) John, V.; Beck, J. P.; Bienkowski,
M. J.; Sinha, S.; Heinrikson, R. L. J. Med. Chem. 2003, 46,
4625; (d) Josien, H. Curr. Opin. Drug Discov. Devel. 2002,
5, 513.
a cis orientation of the amide bond is a prerequisite for
an intramolecular reaction: the preferred trans-orienta-
tion of secondary amide bonds might disfavour a cycli-
sation, whereas tertiary amides adopt no predominant
amide orientation and thus undergo cyclisation more
readily. The tertiary amide 17 was obtained from biphe-
nylamine 15 via 16 in 3 steps and was then further con-
verted by the reported procedure6 via 18 to 19.
Enantiomeric separation of racemic 19 by preparative,
chiral HPLC (chiralpak AD, ethanol/heptane) provided
enantiopure dibenzazepinone building blocks, (S)-19
and (R)-19.
6. (a) Wu, J.; Tung, J. S.; Thorsett, E. D.; Pleiss, M. A.;
Nissen, J. S.; Neitz, J.; Latimer, L. H.; John, V.;
Freedman, S.; Britton, T. C.; Audia, J. E.; Reel, J. K.;
Mabry, T. E.; Dressman, B. A.; Cwi, C. L.; Droste, J. J.;
Henry, S. S.; Mcdaniel, S. L.; Scott, W. L.; Stucky, R. D.;
Porter, W. J. Int. Patent Appl. WO 9828268, 1998; (b)
Audia, J. E.; Hyslop, P. A.; Nissen, J. S.; Thompson, R.
C.; Tung, J. S.; Tanner, L. I. Int. Patent Appl. WO
2000019210, 2000.
7. (a) Wong, G. T.; Manfra, D.; Poulet, F. M.; Zhang, Q.;
Josien, H.; Bara, T.; Engstrom, L.; Pinzon-Ortiz, M.; Fine,
J. S.; Lee, H.-J. J.; Zhang, L.; Higgins, G. A.; Parker, E.
M. J. Biol. Chem. 2004, 279, 12876; (b) Lanz, T. A.;
Hosley, J. D.; Adams, W. J.; Merchant, K. M. J. Pharm.
Exp. Ther. 2004, 309, 49; (c) Best, J. D.; Jay, M. T.; Otu,
F.; Ma, J.; Nadin, A.; Ellis, S.; Lewis, H. D.; Pattison, C.;
Reilly, M.; Harrison, T.; Shearman, M. S.; Williamson, T.
L.; Atack, J. R. J. Pharm. Exp. Ther. 2005, 313, 902.
8. Compounds were routinely checked for cytotoxicity using
a c-secretase-independent luciferase assay.
(R/S),(S)-13 was prepared according to Scheme 2: (S)-19
was coupled with methylmalonic acid mono-tert-butyl
ester, and the tert-butyl protecting group was cleaved
under standard conditions to give 20. In a second cou-
pling step, 20 was condensed with pentafluoropropyl
amine to give the desired product.
9. Brockhaus, M.; Grunberg, J.; Rohrig, S.; Loetscher, H.;
Wittenburg, N.; Baumeister, R.; Jacobsen, H.; Haass, C.
NeuroReport 1998, 9, 1481.
10. Li, Y.-M.; Lai, M.-T.; Xu, M.; Huang, Q.; DiMuzio-
Mower, J.; Sardana, M. K.; Shi, X.-P.; Yin, K.-C.; Shafer,
J. A.; Gardell, S. J. Proc. Natl. Acad. Sci. U.S.A. 2000, 97,
6138.
Compound 14 was prepared from (S)-19 as outlined in
Scheme 3: Coupling with lactic acid gave 21, which
was converted to 14 via a nitrophenyl carbonate as an
activated in-situ intermediate. Full experimental details
can be found in Ref. 14.

J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5918–5923
5923
11. Richards, J. G.; Higgins, G. A.; Ouagazzal, A.-M.;
Ozmen, L.; Kew, J. N. C.; Bohrmann, B.; Malherbe, P.;
Brockhaus, M.; Loetscher, H.; Czech, C.; Huber, G.;
Bluethmann, H.; Jacobsen, H.; Kemp, J. A. J. Neurosci.
2003, 23, 8989.
12. Kansy, M.; Avdeef, A.; Fischer, H. Drug Discovery Today:
Technol. 2004, 1, 349.
13. Yang, M. G.; Shi, J.-L.; Modi, D. P.; Wells, J.; Cochran,
B. M.; Wolf, M. A.; Thompson, L. A.; Ramanjulu, M. M.;
Roach, A. H.; Zaczek, R.; Robertson, D. W.; Wexler, R.
R.; Olson, R. E. Bioorg. Med. Chem. Lett. 2007, 17, 3910.
14. (a) Galley, G.; Goergler, A.; Jacobsen, H.; Kitas, E. A.;
Peters, J.-U. Int. Patent Appl. WO 2004069826, 2004; (b)
Flohr, A.; Galley, G.; Jakob-Roetne, R.; Kitas, E. A.;
Peters, J.-U.; Wostl, W. U.S. Pat. Appl. US 2005054633,
2005; (c) Flohr, A.; Galley, G.; Jakob-Roetne, R.; Kitas,
E. A.; Peters, J.-U.; Wostl, W. U.S. Pat. Appl. US
2005075327, 2005.