Design and synthesis of a novel series of cyanoindole derivatives as potent γ-secretase modulators

Article abstract of DOI:10.1016/j.bmcl.2019.05.023

The discovery, design and synthesis of a new series of GSMs is described. The classical imidazole heterocycle has been replaced by a cyano group attached to an indole nucleus. The exploration of this series has led to compound 26-S which combined high in

Full text of DOI:10.1016/j.bmcl.2019.05.023

Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Design and synthesis of a novel series of cyanoindole derivatives as potent
γ-secretase modulators
⁎
François P. Bischoffa, , Adriana Ingrid Veltera, Garrett Minnea, Serge Pietersa, Didier Berthelota,
Michel De Cleyna, Harrie J.M. Gijsena, Gregor Macdonalda, Michel Surkyna, Sven Van Brandta,
Yves Van Roosbroecka, Chiara Zavattaroa, Marc Merckena, Nigel Austina, Deborah Dhuyvettera,
Herman Borghysa, Ishtiyaque Ahmadb, Swapan Kumar Samantab
a Janssen Research & Development, Turnhoutseweg 30, 2340 Beerse, Belgium
b Jubilant Biosys Ltd, Bangalore, Karnataka, India
A R T I C L E I N F O
A B S T R A C T
Keywords:
The discovery, design and synthesis of a new series of GSMs is described. The classical imidazole heterocycle has
been replaced by a cyano group attached to an indole nucleus. The exploration of this series has led to compound
26-S which combined high in vitro and in vivo potency with an acceptable drug-like profile.
Alzheimer’s disease
γ-Secretase
γ-Secretase modulators
GSMs
An estimated 50 million people worldwide are living with dementia.
Alzheimer’s disease (AD) is, the most common form of dementia, ac-
counting for 60–80 percent of cases, and causing problems with
memory, thinking, and behavior.1 There is a tremendous unmet med-
ical need with no current efficacious treatment for AD. In 1906 the
German physician Alois Alzheimer described the hallmarks of this
progressive, neurodegenerative illness. The defining characteristics of
AD are the deposition in the brain of extracellular amyloid plaques and
intracellular neurofibrillary tangles that are mainly composed of amy-
loid-β (Aβ) peptides and hyper phosphorylated tau protein, respec-
tively. Genetic evidence and histopathological information led to the
definition of the amyloid cascade hypothesis which proposes that ac-
cumulation of Aβ peptides in the brain parenchyma is the event which
triggers AD.2,3 Aβ peptides are produced through the sequential clea-
vage of the Amyloid Precursor Protein (APP) by two secretases: β-se-
cretase (BACE1) first cleaves APP to generate a membrane bound C-
terminal fragment (C99) which in turn is cleaved by γ-secretase (GS),
within the membrane, to release Aβ peptides of lengths varying from 37
to 43 amino acids. Whereas Aβ40 is the most abundant form, it is the
longer and more hydrophobic Aβ42 which is most prone to aggregate
and form neurotoxic oligomers that eventually generate the neurotoxic
amyloid plaques associated with AD. Reducing the production of Aβ42
via BACE1 inhibition and GS inhibition or modulation has been a
strategy over the last two decades to develop disease modifying AD
phase III clinical trial with the non-selective GSI semagacestat has been
halted due to lack of efficacy and serious toxicity including increased
risk of skin and gastrointestinal cancers possibly associated with the
inhibition of the processing of the cell surface receptor Notch, another
GS substrate.5,6 Notch-mediated signaling activities are crucial for
neuronal embryonic development and cell differentiation in adults.7,8
Besides GS inhibition, GS modulation is an attractive approach. GS
modulators (GSMs) selectively reduce the formation of the longer
neurotoxic Aβ peptides, especially Aβ42, by shifting the GS mediated
APP processing towards the shorter and more soluble Aβ peptides such
as Aβ37 and Aβ38.9 The concept of GS modulation was first reported
when a subset of non-steroidal anti-inflammatory drugs (NSAIDs) were
found to selectively lower the production of Aβ42 in favor of Aβ38
without inhibiting Notch1 cleavage.10 Later, another class of com-
pounds, namely the aryl-imidazole derived GSMs, was reported to elicit
similar effects shifting APP towards shorter Aβ peptides. Also, this class
of GSMs did not affect the processing of Notch.11 A recent review covers
the effort by all major research groups that have worked in the field of
GSMs.12
We have previously described our medicinal chemistry effort to-
wards the search of aryl-imidazole derived GSMs. The exploration of a
novel series of bicyclic heterocycles led to the identification of benzi-
midazole 1 (Fig. 1), a very potent in vitro compound that demonstrated
in vivo efficacy across species in lowering Aβ42 and increasing the
shorter isoforms Aβ37 and Aβ38.13 Unfortunately, this series suffered
drugs. Several GS inhibitors (GSIs) have been evaluated in the clinic4
A
⁎ Corresponding author.
E-mail address: fbischof@its.jnj.com (F.P. Bischoff).
https://doi.org/10.1016/j.bmcl.2019.05.023
Received 18 March 2019; Received in revised form 12 May 2019; Accepted 14 May 2019
0960-894X/©2019ElsevierLtd.Allrightsreserved.
Pleasecitethisarticleas:FrançoisP.Bischoff,etal.,Bioorganic&MedicinalChemistryLetters,https://doi.org/10.1016/j.bmcl.2019.05.023

F.P. Bischoff, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Fig 1. Janssen lead compounds from previous series.
Fig. 2. HTS hits with a 4-cyano-aniline moiety.
Fig. 3. Design of indole-based GSMs and replacement of the imidazole ring by a cyano group.
from suboptimal physico-chemical properties such as low solubility,
high lipophilicity and high aromaticity. Later we reported a bicyclic
triazolo-piperidine series exemplified by 214 and a series of anilino-
triazoles, illustrated by 315 (Fig. 1). Although both series combined
potent GSM activity in vitro and in vivo with improved drug-like
properties compared to compounds from previous series, further de-
velopment was not pursued.
contained a N2-phenylpyrimidine-2,4-diamino scaffold also present in
previously described GSMs,17 but interestingly both avoided the clas-
sical imidazole heterocycle, instead replacing it with a nitrile func-
tionality. Hit 6 also contained an amino-benzonitrile motif substituted
in that case by a dihydro-pyrrolo pyrimidine heterocycle. Shortly after
these findings cyano-pyrimidines based-GSMs were reported else-
where.18
A high throughput screening campaign of a limited selection of the
corporate collection was conducted. Compounds were tested for their
ability to increase the production of Aβ38 in vitro,16 and several hits
were found. In a subsequent round of testing, hits were also seen to
inhibit the production of Aβ42, with moderate potency, confirming
their modulatory activity. Two of these compounds (4 and 5) (Fig. 2)
In the meantime, we were exploring other approaches for the design
of new scaffolds and we hypothesized that, the pyridine-NH-benzimi-
dazole in 7 (Fig. 3 and a close analogue of 1) could be replaced by an
indole-triazole system to deliver 8 that turned out to retain moderate
potency (IC50 = 0.085 μM). A similar strategy was applied starting from
9, a potent GSM (IC50 = 0.011 μM) from the patent literature.19 By
2

F.P. Bischoff, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
activity was maintained by converting 10 (IC50 = 0.046 μM) to 14
(IC50 = 0.036 μM) suggesting that the CN group could be an alternative
for the usual imidazole heterocycle. We decided to explore the SAR on
the newly designed indole series with a focus on the cyano-indole de-
rivatives.20
Most of the cyanoindoles were prepared according to the general
methods shown in Scheme 1. Thus, the imidazole derivatives I were
synthesized via condensation of amidines III with chloroacetylindole
derivatives II. On the other hand, condensation of VI with in-
dolylhydrazides V provided the corresponding triazole derivatives IV.
One route for the synthesis of the cyanoindole cores II and V is
described in Scheme 2. Thus, aldehydes 39a–b were submitted to
Knoevenagel condensation conditions with methylazidoacetate to pro-
vide intermediates 40a–c in good yields. 40a–c were cyclized to indoles
41a–c by heating at 200 °C in xylene. Alternatively, the cyclization was
done in the presence of catalytic Rh2(C3F7CO2H)4, in toluene at 60 °C,
albeit in lower yields.21 Intermediates 41a–c were converted to hy-
drazides 42a–b upon treatment with hydrazine hydrate in refluxing
ethanol. Similarly, halogenated indolyl esters 46a–b22 were trans-
formed to hydrazides 47a–b. The indole nitrogen atom in 41c was also
protected with a trimethylsilylethoxymethyl (SEM) group, and then
cyanation of the resulting SEM-protected indole provided intermediate
43. On one hand, 43 was converted to hydrazides 45, used for the
preparation of triazolo-derivatives IV. On the other hand, condensation
of 43 with chloroacetic acid, followed by decarboxylation, provided
indolyl acetyl chlorides 44, used in the preparation of compounds I.
Scheme 1. Retrosynthetic analysis.
analogy with our previous work,13 the double bond could be replaced
by an NH linker, incorporated in an indole nucleus. The resulting
compound 10 showed good potency (IC50 = 0.046 μM). Based on the
results of the high throughput screen, we started by exploring re-
placement of the imidazole heterocycle by a cyano group in several of
these compounds. Starting from 9, a significant drop of activity was
observed for its cyano-methoxyphenyl analogue 13 (IC50 = 0.275 μM)
and this held true when comparing 11 (IC50 = 0.182 μM) to 12
(IC50 > 10 μM). However, we were pleased to observe that potent
Scheme 2. Synthesis of the cyanoindole core. Reagents and conditions: (a) Na, MeOH (EtOH), −10 °C, 2–4 h (50–69%); (b) Xylene, 200 °C; 2 h; (c) Rh2(C3F7CO2H)4,
toluene, 60 °C, 24 h (14–16%); (d) NH2NH2·H2O, 100 °C, 3–4 h (76%–81%); (e) NaH, SEMCl, THF, 0 °C – RT, 12 h (88%); (f) Zn(CN)2, Pd2(dba)3 (75%); (g) i.
ClCH2CO2H, LDA (2 eq), −78 °C, 2 h; ii. AcOH, RT, 15 min (95%); (h) NH2NH2·H2O, EtOH, 80 °C, 12 h (56–86%); (i) (2-MePh)3P, Pd(OAc)2, DIPEA, DMF, 85 °C, 16 h
(68%); (j) TFA, DCM, 0 °C – RT, 16 h (61%).
3

F.P. Bischoff, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Scheme 3. Synthesis of amino-tetrahydropyridine. Reagents and conditions: (a) Cl(CH2)3Br, KOtBu, THF, −10 °C to 0 °C; (b) AcCl, EtOH, 0 °C – RT, 1–3 d; (c) 7 N NH3
(MeOH), RT, 5 d (86%).
Scheme 4. Synthesis of dihydrooxazine amines. Reagents and conditions: (a) Ethylene glycol, PTSA, toluene, 100 °C (37–100%); (b) TMSCN, ZnI2, DCM, −20 °C, 2 h
(complete conversion); (c) TBAF, THF, RT, 1 h (75–93% over two steps); (d) SOCl2, DCE, 60 °C, 2 h (56–89%); (e) AcCl, EtOH, −15 °C – RT, 2.5 h (68–100%); (f) 7 N
NH3 (MeOH), 55 °C, 24 h (28–89%).
Finally, commercially available 4-bromo-2-methoxy-benzonitrile 48
was subjected to Heck reaction conditions with Boc-protected acryla-
mide hydrazide 49, to provide hydrazide 50 after Boc-deprotection.
Hydrazide 50 was further used in the synthesis of prototype 13.
The necessary amidine and imidates right hand sides III and VI used
in the synthesis of triazolo/imidazole piperidines were synthesized
according to Scheme 3. Thus, nucleophilic substitution of arylacetoni-
triles 51a–b with 1-chloro-3-bromopropane gave intermediates 52a–b.
These compounds were then treated with HCl, generated in situ from
acetyl chloride and ethanol, to provide imidates 53a–b. Imidate 53b
was stirred in a methanolic ammonia solution to give amidine 54b.
The imidate and amidine intermediates VI and III (Scheme 1), re-
spectively, necessary for the synthesis of triazolo-and imidazole-mor-
pholines, were prepared according to the synthetic route in Scheme 4.
Thus, commercially available benzaldehydes 55a–i were first reacted
with ethylene glycol to give ketals 56a–i. Addition of trimethylsi-
lylcyanide (TMSCN) to 56a–i, followed by TMS deprotection with
TBAF, provided alcohols 57a–i. These were converted to the
Scheme 5. Preparation of substituted
piperidino-amidines. Reagents and con-
ditions: (a) LiHMDS, THF, −78 °C – RT,
16 h (58%); (b) NH3 (7 N in MeOH),
MeOH, 75 °C, 16 h (94%); (c) i. POCl3,
THF, 100 °C, 3 h; ii. NH3 (7 N in MeOH),
0 °C
– RT, 12 h (75%); (d) NaHSO3,
NaCN, H2O, RT, 16 h (89%); (e) DHP,
PPTS, DCM, 0 °C – RT, 14 h (93%); (f) i.
Na, EtOH, 0 °C – RT, 45 min; ii·NH3, –
40 °C, 20 min; iii. NH4Cl, −40 °C – RT,
18 h (65%).
4

F.P. Bischoff, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Scheme 6. Reagents and conditions: (a) Imidazole, MeOH, 30 °C, 3–14 d (54–100%); (b) ZnCN2, Pd(PPh3)4, DMF, 135 °C, 150 min, MW (13–43%); (c) NaH, MeI, THF,
RT, 16 h (31%); (d) i. TFA, DCM, 0 °C – RT, 2.5 h; ii. NaOH, THF (8/1, v/v), RT, 2 h (11–88%); € Prep SFC on Chiralcel Diacel OD 20 × 250 mm; mobile phase, CO2,
MeOH with 0.2% iPrNH2 (12–92%).
Scheme 7. Reagents and conditions: (a) Na2CO3, NaI,
acetone, 60 °C, 2 h (42–84%), (b) i. TFA, DCM, 0 °C –
RT, 2.5 h; ii. NaOH, THF (8/1, v/v), RT, 2 h (53%);
(c) Prep SFC on Chiralcel Diacel OD 20 × 250 mm or
Chiralcel Diacel OJ 20 × 250 mm or Chiralpak
Diacel AS 20 × 250 mm; mobile phase, CO2, MeOH
with 0.2% iPrNH2 (12–92%).
corresponding chlorides 58a–i by reaction with thionyl chloride in di-
chloroethane at 60 °C. Intermediates 58a–i were then converted to
imidates 59a–i and amidines 60a–i following the above described
procedures.
ammonia, 63 smoothly cyclized to lactam 64. Compound 64 was then
converted into amidine 65 in a one pot, two-step process, first by re-
action with phosphorus oxychloride to the corresponding chlor-
oiminium intermediate, which was treated in situ with a methanolic
solution of ammonia. 2-chlorobenzaldehyde 66 was submitted to
Strecker reaction conditions, to provide the corresponding cyanohy-
drin. Protection of the alcohol moiety provided 67, which was sub-
mitted to Pinner reaction conditions to provide amidine 68, after in situ
reaction of the resulting Pinner salt with ammonia.
The synthesis of the amidine intermediates needed for the pre-
paration of 6-substituted imidazole-piperidines and morpholines 34–38
is described in Scheme 5. In this case, nucleophilic substitution of ethyl
2-(2-chlorophenyl) acetate 61 with bischloromethylethylene 62 pro-
vided intermediate 63. When treated with a methanolic solution of
5

F.P. Bischoff, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Scheme 8. Reagents and conditions: (a) NaHCO3, NaI, acetone, RT – 40 °C, 18 h; (b) H2, Pd/C, EtOAc, RT, 4 h (90%); (c) i. TFA, THF, 0 °C – RT, 3 h; ii. NaOH or KOH,
RT, 5 h (42–94%); (d) OsO4, NMO, acetone, H2O, 0 °C – RT, 4 h (63%); (e) PhSiH3, Mn(THMD)3, iPrOH, DCM, 0 °C, 3 h then RT, 24 h (39%); (f) NaHCO3, THF, H2O,
RT – 80 °C, 16 h (93%); (g) Cs2CO3, DMF, 80 °C, 12 h; (h) PTSA, MeOH, RT, 16 h; (i) PPTS, DCE, 90 °C (MW), 1 h; (j) TFA, 90 °C (MW), 1 h.
60a–c and 61e–i, under Finkelstein reaction conditions, provided the
corresponding cyanoindoles, after SEM deprotection and SFC separa-
tion of enantiomers.
Table 1
2-CF3-phenyl derivatives.
Finally, the synthesis of C(6)-substituted imidazo-piperidines and
morpholines 34–38 is shown in Scheme 8. Cyclization of indolyl
chloroketone 44 with amidine 71 provided cyanoindole 72. Hydro-
genation of the ethylene substituent followed by SEM deprotection
provided compound 34. Alternatively, dihydroxylation of 72, and then
SEM deprotection provided diol 35. Finally, Mn(III)-catalyzed epox-
idation reaction, followed by in situ epoxide reduction with phenylsi-
lane provided compound 36.
On the other hand, reaction of indolylchloroketone 70 with amidine
68 gave the imidazole derivative 73.23 Addition of 73 to either (R) or
(S) 2-methyloxirane, followed by THP deprotection, provided the diols
74a–b, which were then submitted to Williamson ether synthesis to
give imidazo-morpholines 75a–b. Deprotection of the PMB group in
75a–b provided the compounds 37 and 38.
Compd
R1
R2 R3
R4
X
Y
hIC50 (μM)
Aβ42a
mIC50 (μM)
Aβ42a
(cLog P)
14 (RS) (4.7)
15 (RS) (4.9)
H
H
CN OMe
H
H
H
H
H
H
H
Me
H
H
N
CH2 0.036
CH2 0.117
CH2 0.107
0.038
0.186
0.166
0.041
0.042
0.295
8.5
CN
H
H
N
16 (RS) (4.7) OMe CN
N
The initial SAR is described in Table 1. Removal of the 4-methoxy
group of the indole nucleus in 14 (IC50 = 0.036 μM) resulted in a ∼3-
fold decrease in the in vitro activity (15, IC50 = 0.117 μM) whereas a
methoxy substituent on the 6-position (16, IC50 = 0.107 μM) did not
improve the potency compared to 15. According to calculated lipo-
philicity (cLog P), triazolo-piperazine derivatives (14–16) still appeared
highly lipophilic (cLog P around 4.7–4.9) while the introduction of an
oxygen atom provided triazolo-morpholine compounds with a marked
decrease in lipophilicity (17, cLog P = 3.2 vs 14 cLog P = 4.7). There
was a slight drop in activity for racemate 17 (IC50 = 0.072 μM vs
IC50 = 0.036 μM for 14) which was purified by SFC to provide 17-S and
17-R. The absolute configuration was not determined, for these two
compounds. However, by analogy with compounds 24-S and 24-R for
which absolute configuration was determined via VCD and based on the
clear difference in potency of the two enantiomers of 17, the S-con-
figuration was tentatively assigned to the more active enantiomer (17-
S, IC50 = 0.045 μM vs 17-R, IC50 = 0.603 μM). When the cyano group
was moved to the 5-position of the indole (18, IC50 = 3.981 μM), or
when the indole was N-methylated (19, IC50 = 4.467 μM) activity was
almost completely lost. Imidazo-morpholine analogus were also pre-
pared and again one enantiomer turned out to be considerably more
17 (RS) (3.2)
17-S* (3.2)
17-R* (3.2)
H
H
H
CN OMe
CN OMe
CN OMe
N
O
O
O
0.072
0.045
0.603
N
N
18 (RS) (4.9) CN
H
H
H
N
CH2 3.981
19 (RS) (4.8)
20-S* (3.9)
20-R* (3.9)
H
H
H
CN
N
CH2
O
4
> 10
0.019
0.251
CN OMe
CN OMe
CH
CH
0.018
0.347
O
a
IC50 represents the concentration of a compound that is required for re-
ducing the Aβ42 level by 50%. The IC50 values are a mean of at least 2 de-
terminations. hIC50 and mIC50 represent the values in human and mouse, re-
spectively. The human and mouse assays are described in Ref. 13 and in the
Supplementary info section, respectively.
Most of the triazolo derivatives IV were prepared according to
Scheme 6. Thus, condensation of hydrazide 42a–b with imidate 53a,
provided the corresponding triazolopiperidines, which were submitted
to cyanation reaction conditions to provide 14 and 16. Compounds 10,
13, 15, 17–19, 22, 23-S, 25-S, 27-S, 28-S were also prepared in a si-
milar manner.
Most of the imidazole derivatives I were prepared according to
Scheme 7. Cyclization of indolyl chloroketone 44 with amidines 54b,
6

F.P. Bischoff, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Table 2
Exploration of the bicyclic heterocycle.
Compd
X
Y
R
hIC50
(μM)
mIC50
(μM)
Brain
levels
(μM)b
Plasma
Decrease Aβ42
Increase Aβ38 fu,b (%)e Free Brain
Coverage ratio
CRg
(cLog P)
levels (μM)b Vivo (%)c
Vivo (%)d
levels Cub
(μM)f
Aβ42a
Aβ42a
17-S (3.2)
20-S (3.9)
N
O
O
0.045
0.018
0.042
0.019
10.5
16.4
5.4
8.4
46
73
39
10
ndh
nd
nd
CH
0.21
0.034
1.8
21 (RS) (5.4) CH CH2
0.112
0.025
0.032
0.023
0.050
0.028
0.034
0.016
1.9
4.5
51
45
5
17
nd
nd
nd
nd
nd
1
22 (RS) (4.4)
23-S (2.9)
24-S (3.6)
N
CH2
O
0.69
1.71
6.9
0.75
1.52
9.2
−9
−12
122
nd
nd
N
nd
nd
CH
O
45
0.23
0.016
25-S (2.9)
26-S (3.6)
N
O
O
0.031
0.014
0.019
0.017
5.5
12
2.5
8.2
33
66
60
40
0.17
0.22
0.009
0.026
0.47
1.53
CH
27-S (3.4)
N
O
0.059
0.059
9.6
1.7
2
19
0.09
0.009
0.15
28-S (3.4)
29-S (4)
N
O
O
0.098
0.041
0.107
0.074
nd
nd
nd
35
nd
41
nd
nd
nd
nd
nd
nd
CH
16.3
10.6
30-S (4)
CH
CH
O
O
0.036
0.026
0.042
0.023
18.7
23.8
11.3
8.3
40
60
136
61
nd
nd
nd
31-S (3.7)
0.47
0.112
4.8
32-S (2.7)
33
CH
CH
O
O
0.110
0.219
0.085
0.339
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
a
IC50 represents the concentration of a compound that is required for reducing the Aβ42 level by 50%. The IC50 values are a mean of at least 2 determinations.
hIC50 and mIC50 represent the values in human and mouse, respectively.
b
c
d
e
f
g
h
Plasma and brain samples were analyzed for the tested compound using a qualified research LC–MS/MS method.
Mouse in vivo at 30 mpk p.o. after 4 h (n = 6) expressed as a percentage of Aβ42 lowering in brain compared with untreated animals.
Mouse in vivo at 30 mpk p.o. after 4 h (n = 6) expressed as a percentage of Aβ38 increase in brain compared with untreated animals.
Fraction of unbound compound in brain (fu,b).
Free brain concentrations, Cub, calculated from Cub = fu,b × [brain levels, μM].
Coverage ratio, CR, calculated as CR = Cub/mIC50
.
nd – not determined.
potent (20-S, IC50 = 0.018 μM) than the other (20-R, 0.347 μM).
The triazolo-morpholine derivative 17-S was tested to measure ef-
fects on Aβ peptides in mice. An oral dose of 30 mg/kg showed a clear
lowering of Aβ42 (−46%) combined with an increase of Aβ38
(+39%), 4 h after dosing, along with high total brain levels (10.5 μM,
B/P = 1.94). 20-S, its direct imidazo-morpholine analogue, was even
more efficacious in lowering Aβ42 (−73%) but weaker in increasing
Aβ38 (+10%). All compounds from this series, including 20-S, were
inactive (IC50 > 10 μM) in decreasing the total Aβ level. The brain
fraction of unbound compound (fu,b) was low (0.21%), but combined
with the high total brain concentration (16.4 μM) this resulted in a free
brain concentration of 0.034 μM. This exceeded the mouse IC50
(0.019 μM, Coverage Ratio (CR = Cu,b/mIC50 = 1.8).
We further explored the substitution of the distal phenyl ring of the
bicyclic5,6 system (Table 2). Results generated from previous explora-
tions highlighted the importance of a non-planar arrangement between
7

F.P. Bischoff, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Table 3
despite higher total brain concentrations (1.71 μM). The imidazo-mor-
pholine analogue 24-S, somewhat more potent in vitro, demonstrated
similar lowering of Aβ42 (−45%) and strong Aβ38 increase (+122%),
with higher total brain concentration (6.9 μM) and CR = 1. Its en-
antiomer 24-R was weakly active (hIC50 = 1.09 μM for 24-R vs
hIC50 = 0.023 μM for 24-S). The absolute configuration of these two
compounds was determined via VCD and confirmed that activity re-
sided in the S-enantiomer in analogy with previous series of triazolo-
piperidines.14 Although 25-S and 26-S had comparable in vitro activity,
26-S proved to be more efficacious in vivo at lowering Aβ42 due to
higher brain levels, resulting in a higher CR (1.53). 27-S was less potent
in vitro and did not show significant activity in vivo consistent with a
very low CR (0.15). A drop of activity was observed for 28-S and 29-S
(mIC50 = 0.107 and 0.74 μM, respectively) the latter showing moderate
in vivo activity (−35% for Aβ42 and +41% for Aβ38). 30-S
(mIC50 = 0.042 μM) combined high total brain levels (18.7 μM) with
moderate Aβ42 lowering and a strong Aβ38 increase (+136%). 31-S
was one of the most potent compounds in this series demonstrating a
60% decrease of Aβ42 and a 61% increase of Aβ38 with a high CR
(4.8). 33, the most active of the four possible 4-F-benzyl-α-methyl
diastereomers, was weakly active (mIC50 = 0.339 μM).
6-substituted bicyclic imidazoles.
Compd (cLog P)
Y
R1
R2
H
hIC50 (μM) Aβ42a
0.015
0.039
0.022
0.066
34 (S,R) (6)
35 (RS,RS) (3.9)
36 (RS,RS) (4.6)
37 (S,S) (4.2)
CH2
CH2
CH2
O
OH
Me
CH2OH
OH
H
38 (S,R) (4.2)
O
H
0.017
a
IC50 represents the concentration of a compound that is required for re-
ducing the Aβ42 level by 50%. The IC50 values are a mean of at least 2 de-
terminations. hIC50 and mIC50 represent the values in human and mouse, re-
spectively.
The most efficacious compounds in lowering the production of Aβ42
(> 60%) were all imidazo-morpholine derivatives (20-S, 26-S and 32-S).
Their respective Log P24 (measured values: 4.29, 4.19 and 4.14) were
somewhat higher than the calculated values (cLog P), thus comparable to
the ones from lead compounds 2 and 3 from former series.14,15 The less
lipophilic triazolo-morpholine analogs were also less efficacious in low-
ering Aβ42. As discussed previously,15 robust in vivo activity could be
achieved despite the low fraction of unbound compound (fu,b < 0.5%),
but, high total brain levels and high in vitro potency are required, re-
sulting in coverage ratios above 1 (1.8, 1.5, and 4.8, for 20-S, 26-S and
32-S, respectively). However, the poor correlation between in vivo effi-
cacious concentrations and in vitro activity has been already discussed
for GSMs25 and 25-S despite a low coverage ratio (0.47) demonstrated
strong Aβ38 (+60%) increase and moderate Aβ42 decrease (−33%).
Hence, we observed a disconnection, or at least no consistency, between
the level of Aβ42 reduction and the level of Aβ38 increase.
Table 4
Compared drug-like profiles of 26-S and 3.
26-S
3
Log P
4.19
3.66
0.014
0.017
18
4.24
3.48
0.019
0.049
4
LLE
hIC50 (μM)
mIC50 (μM)
hLM (%)a
mLM (%)a
dLM (%)a
Kin. Sol. pH 7.4 (μM)
27
0
29
10
6
39
CYP450 IC50 (μM)
3A4 (M)
2D6
> 10
9.8
Activator
16
12.3
22.1
–
2C9
5.9
The introduction of substituents on the piperidine (or morpholine)
ring was evaluated (Table 3). A small lipophilic substituent such as
methyl was tolerated, whereas bulkier groups as CF3, ethyl, or isopropyl
were detrimental for activity (data not shown). Small polar substituents
such as hydroxy or hydroxymethyl delivered also active compounds.
These data confirmed previous results19 showing that this area allowed
the introduction of polarity. These compounds had a clean profile for
CYP inhibition (data not shown) but were not further progressed.
2C19
1A2
8.3
> 10
1.3
2C8
–
hERG IC50 (μM)b
5.29
> 10
hERG PX (%)c
31
52
fu,b (%)d
0.22
8.25
12.1
1.46
0.026
1.53
66
0.5
Conc mouse plasma 30 mg/kg (μM)
Conc mouse brain 30 mg/kg (μM)
B/P 4 h
17.8
14
0.8
Imidazo-morpholine 26-S was profiled and compared with
3
Conc mouse brain free 30 mg/kg (μM), 4 h
Cover Ratio
0.069
1.41
55
(Table 4). The compound was metabolically less stable and less soluble.
The replacement of the usual imidazole, which helps for the solubility
of GSMs, by the cyano group, resulted in reduced solubility, which was
observed for all compounds across this series. 26-S had a higher pro-
pensity to inhibit CYP450. In general, 26-S demonstrated similar drug-
like properties and efficacy to modulate Aβ production in vivo. 26-S
was more potent in vitro, in mouse, (mIC50 = 0.017 μM) than 3
(mIC50 = 0.049 μM), and it translated in a robust decrease of Aβ42
levels in mouse brain (66%), achieved at a Cub = 12.1 μM, similar with
that of 3 (14 μM) and at a higher B/P ratio (1.46 vs 0.8 for 3). 26-S was
tested orally in the dog at 10 mg/kg and showed a reduction of 37% of
Aβ42 and an increase of 40% of Aβ38 in the cerebrospinal fluid (CSF)
8 h post dosing. The maximum plasma concentration (Cmax = 2.4 μM)
was reached 1.6 h post dosing. No relevant changes in behavior or
appearance were seen and no increase in liver enzymes levels (ALT and
AST) was observed.
Mouse Aβ42 lowering, 30 mg/kg, 4 h (%)e
Mouse Aβ38 increase, 30 mg/kg, 4 h (%)e
40
53
a
% metabolized after 15 min upon incubation with human (hLM), mouse
(mLM) and dog (dLM) liver microsomes at 1 μM concentration.
b
c
d
e
Receptor binding on the human Herg ion channel.
% inhibition of IKr current at 3 μM.
Unbound fraction to rat brain tissue.
% Modulation of Aβ peptides as determined by Meso Scale.26
triazole and distal phenyl, therefore, here we focused particularly on
ortho-substituents.14
Piperidine derivatives 21 and 22 had good activity in vitro, 0.05 μM
and 0.028 μM (mIC50), respectively. These two highly lipophilic com-
pounds (21, cLog P = 5.4 and 22, cLog P = 4.4) showed equivalent
Aβ42 lowering of −45% in vivo (mouse, 30 mg/kg, PO, 4 h post-
dosing) although total brain exposures were relatively low, 1.9 μM and
0.69 μM, respectively. Aβ38 production was not significantly affected.
The less lipophilic (cLog P = 2.9) triazolo-morpholine analogue, 23-S,
had similar in vitro activity but did not translate into in vivo activity
To summarize, we have reported the evolution of classical (hetero)
aryl imidazole GSMs to imidazoindole derivatives followed by the
successful replacement of the imidazole heterocycle by a cyano moiety.
So far, GSMs have suffered from suboptimal physico-chemical
8

F.P. Bischoff, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
properties such as low solubility, high lipophilicity and high ar-
omaticity and the search for structural alternatives to the arylimidazole
moiety has been highly challenging. The successful replacement of the
imidazole by a nitrile moiety has been reported27 or discussed only in a
few papers25 and recently a benzimidazole series without the ar-
ylimidazole chemotype has been disclosed.28 We have described the
design and the synthesis of these new cyano indole derived GSMs as
well as their drug-like properties. We identified 26-S as a potent in vitro
GSM that demonstrated efficacy in vivo in lowering Aβ42 and in-
creasing Aβ38, in both, mouse and dog. No increase in liver enzymes
levels (ALT and AST) was observed after a single dose (10 mg/kg) in the
dog. Hence, the replacement of the ubiquitous imidazole moiety by the
CN did not improve the overall drug-like profile in this case. The GSMs
that have entered clinical trials failed in early phases due to off-target
side effects which can be traced back to the general high lipophilicity of
GSMs and consequently high plasma concentrations needed for
achieving the PD effect. To test GS and the amyloid hypothesis in the
clinic, finding compounds in a novel chemical space, with better drug-
like properties, remains key to success. However, the failure of GSIs, the
recent concerns of BACE inhibition in clinical trials, regarding cogni-
tion, possibly related to substrate selectivity issues,29 combined with
the need for developing agents suitable for multi-decade make it rather
unlikely that these approaches will be further extensively investigated.
References
1. Alzheimer’s Association. https://www.alz.org/alzheimers_disease_what_is_
alzheimers.asp.
2. Hardy JA, Higgins GA. Science. 1992;256:184.
3. Hardy JA, Selkoe DJ. Science. 2002;297:353.
4. Gijsen HJM, Bischoff FP. Annu Rep Med Chem. 2012;47:55.
5. Karran E, Hardy J. Ann Neurol. 2014;76:185.
6. De Strooper B. Cell. 2014;159:721.
7. Imayoshi I, Sakamoto M, Yamaguchi M, Mori K, Kageyama R. J Neurosci.
2010;30:3489.
8. De Strooper B, Annaert W, Cupers P, et al. Nature. 1999;398:518.
9. Oehlrich D, Berthelot DJC, Gijsen HJM. J Med Chem. 2011;54:669.
10. Weggen S, Eriksen JL, Das P, et al. Nature. 2001;414:212.
11. Chavez-Gutiérrez L, Bammens L, Benilova I, et al. EMBO J. 2012;31:2261.
12. Hall A, Velter AI. Comprehensive Medicinal Chemistry III. 2017;7:280–325.
13. Bischoff F, Berthelot D, De Cleyn M, et al. J Med Chem. 2012;55:9089.
14. Oehlrich D, Rombouts FJR, Berthelot D, et al. Bioorg Med Chem Lett. 2013;23:4794.
15. Velter AI, Bischoff FP, Berthelot D, et al. Bioorg Med Chem Lett. 2014;24:5805.
16. The screen is described in the Supplementary Information.
17. See for example: Baumann K, Flohr A, Goetschi E, et al. WO 2010053438.
(a) For BMS-932481, the most advanced GSM with a N2-phenylpyrimidine-2,4-
diamino scaffold: Toyn JH, Boy KM, Raybon J, et al. J Pharmacol Exp Ther.
2016;358:125–137;
(b) Soares HM, Gasior M, Toyn JH, et al. J Pharmacol Exp Ther. 2016;358:138–150
18. Pettersson M, Stepan AF, Kauffman GW, Johnson DS. Expert Opin Ther Pat.
2013;23:1349.
19. Kimura T, Kitazawa N, Kaneko T, et al. US 2009/0062529.
20. The synthesis of 8 and 12 is described in the Supplementary Information.
21. Stokes BJ, Dong H, Leslie BE, Pumphrey AL, Driver TG. J Am Chem Soc.
2007;129:7500.
Acknowledgments
22. Note: Iodoindole 60a was prepared according to procedure described by Beshore DC,
Dinsmore C. J Synth Commun. 2003;33:2423–2427.
23. The synthesis of intermediate 70 is described in the Supplementary Information.
24. Takacs-Novak K, Avdeef A. J Pharm Biomed. 1996;14:1405.
25. Hall A, Patel TR. Prog Med Chem. 2014;53:101–145.
The authors would like the members of Janssen R&D Biology, ADME
and screening departments and the members of the purification and
analysis department.
26. Pan C, Korff A, Galasko D, et al. J Alzheimer Dis. 2015;45:709–719.
27. Gerlach K, Hobson S, Eickmeier C, et al. Bioorg Med Chem. 2018;26:3227.
28. Sekioka R, Honjo E, Honda S, et al. Bioorg Med Chem Lett. 2018;26:435.
29. http://www.ctad-alzheimer.com.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.bmcl.2019.05.023.
9