Anilinotriazoles as potent gamma secretase modulators

Article abstract of DOI:10.1016/j.bmcl.2014.10.024

The design and synthesis of a novel series of potent gamma secretase modulators is described. Exploration of various spacer groups between the triazole ring and the aromatic appendix in 2 has led to anilinotriazole 28, which combined high in vitro and in

Full text of DOI:10.1016/j.bmcl.2014.10.024

Bioorganic & Medicinal Chemistry Letters 24 (2014) 5805–5813
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Anilinotriazoles as potent gamma secretase modulators
a
a
a
a
Adriana I. Velter a, , François P. Bischoff , Didier Berthelot , Michel De Cleyn , Daniel Oehlrich ,
Libuse Jaroskova b, Gregor Macdonald a, Garrett Minne a, Serge Pieters a, Frederik Rombouts a,
Sven Van Brandt a, Yves Van Roosbroeck a, Michel Surkyn a, Andrés A. Trabanco c, Gary Tresadern b,
⇑
Tongfei Wu a, Herman Borghys b, Marc Mercken a, Chantal Masungi b, Harrie Gijsen a,
⇑
a Neuroscience, Janssen Research & Development, Turnhoutseweg 30, 2340 Beerse, Belgium
b Discovery Sciences, Janssen Research & Development, Turnhoutseweg 30, 2340 Beerse, Belgium
c Neuroscience, Janssen Research & Development, Jarama 75, 45007 Toledo, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
The design and synthesis of a novel series of potent gamma secretase modulators is described. Explora-
tion of various spacer groups between the triazole ring and the aromatic appendix in 2 has led to anili-
notriazole 28, which combined high in vitro and in vivo potency with an acceptable drug-like profile.
Ó 2014 Elsevier Ltd. All rights reserved.
Received 29 August 2014
Revised 2 October 2014
Accepted 7 October 2014
Available online 18 October 2014
Keywords:
Gamma secretase modulators
Alzheimer’s disease
Anilinotriazoles
Dementia affects nearly 36 million people worldwide, a number
which is expected to triple by 2050.1 The most common form of
dementia is Alzheimer’s disease (AD),2 a progressive, neurodegen-
erative illness described in 1906 by the German physician Alois
Alzheimer. No treatment exists yet for this condition, the marketed
drugs targeting only the amelioration of the symptoms.3 Charac-
teristic for AD is the deposition of extraneuronal amyloid plaques
and intraneuronal neurofibrillary tangles of hyperphosphorylated
tau protein in the limbic and cortical regions of the brain, which
eventually leads to neurodegeneration. Based on genetic evidence,
Hardy and Higgins formulated the amyloid hypothesis, according
to which accumulation of amyloid peptides in the brain is the pri-
mary driver of the AD.4,5
Formation of the amyloid beta peptides in the brain is the result
of a sequence of proteolytic cleavages of amyloid precursor pro-
tein, APP. This is first cleaved by b-secretase (BACE-1) to form a
membrane bound C-terminal fragment (C99), which is further
cleaved by gamma secretase (GS) to produce Ab peptides of lengths
varying from 37 to 43 aminoacids. Of these, Ab42 and Ab43 are
most prone to aggregate and generate the neurotoxic amyloid pla-
ques. The amyloid cascade theory opened the door to anti Ab ther-
apeutics as strategies for developing anti Alzheimer disease
modifying drugs,6 among which suppression of the Ab42
production via BACE-1 inhibition and GS inhibition or modulation
is intensively pursued.7 In this regard, GS, the intra-membrane
protease complex responsible for the final cleavage step in the
amyloid cascade, represents an attractive yet challenging target.8
Recently, GS inhibitors (GSIs) semagacestat and avagacestat were
discontinued in clinical trials due to side effects such as toxicity
related to inhibition of other GS substrates and decline in cogni-
tion.8,9 GS modulators (GSMs),10,11 a class of compounds which
are able to reduce the level of longer, neurotoxic Ab peptides by
shifting the APP processing of c-secretase towards shorter isoforms
(such as Ab37, Ab38), represent a viable alternative to GS inhibi-
tion.12 Several classes of GSMs are known to date: carboxylic acids
derived from non-steroidal anti-inflammatory drugs (NSAIDs),
such as tarenflurbil, non-NSAID GSMs such as aryl imidazole-
derived GSMs and more recently, triterpene-derived modulators.7
Our own search towards aryl-imidazole derived GSMs led to the
discovery of the benzimidazole derivative 1 (Fig. 1).13 It is one of
the most potent GSMs to date, suffering however from sub-optimal
drug-like properties. More recently, we have described the design
and synthesis of bicyclic triazolo-derivatives, such as triazolopi-
peridine 4 (Scheme 1),14 as potent in vitro and in vivo GSMs with
an improved drug-like profile. These compounds resulted from a
conformational restriction of benzyltriazoles 3, which in turn
derived from the introduction of a methylene spacer in 2 between
the central triazole ring (C) and the aromatic appendix D, in this
case the 4-F-phenyl group. While this work was in progress, an
⇑
Corresponding authors.
E-mail addresses: ivelter@its.jnj.com (A.I. Velter), hgijsen@its.jnj.com (H. Gijsen).
http://dx.doi.org/10.1016/j.bmcl.2014.10.024
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

5806
A. I. Velter et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5805–5813
F
dosing, despite high compound brain levels (5.7 lM, B/P = 0.42).
This compound was however highly bound to brain proteins, with
the fraction of unbound compound in brain (fu,b) under the detec-
tion limit (fu,b < 0.05%).16 It can be hypothesized that the free brain
concentration of 14 was not high enough for in vivo activity.
Nevertheless, in the benzimidazole series of 1 many derivatives
with fu,b < 0.1% demonstrated robust in vivo activity.13 Further
exploration around this hit was aimed at identifying similarly
potent analogs that would also demonstrate in vivo efficacy.
It was found that modifications on the aniline ring (Table 2)
resulted in modulation of both in vitro and in vivo activity. Thus,
substitution at 3- and 3,5-positions (compounds 18–19 in Table 2)
with both electron withdrawing and electron donating substitu-
ents maintained the in vitro potency of the hit, compound 18 being
one of the most potent GSMs in this series. As for 14, this potency
again did not translate in vivo in mice, as Ab42 levels were not sig-
nificantly reduced 4 h after administration. Compounds 18 and 19
were also highly bound to brain tissue and thus, despite the high
compound brain levels, in both cases the free brain concentration
(Cu,b) was probably not high enough for in vivo activity (estimated
as less than 2 nM and 4 nM, respectively).17 In mouse, the free
brain concentration of 18 and 19 was at least 20 fold less than
IC50 = 0.019 μM
N
H
N
Vivo mouse 30 mg/kg:
Aβ42 -65%
N
Aβ38 +144%
N
N
Brain/Plasma 21.9 μM/ 19.5μM
OMe
N
1
Figure 1. Benzimidazole derived GSM.
Note: IC50 represents the concentration of a compound that is required for reducing
the Ab42 level by 50%. IC50 values are a mean of at least 3 determinations. Mouse
in vivo at 30 mg/kg p.o. after 4 h (n = 6) expressed as a percentage of Ab42 lowering
in brain compared with untreated animals. Plasma and brain samples were
analyzed for the tested compound using a qualified research LC–MS/MS method.
alternative approach to improve the potency of 2, while maintain-
ing its physicochemical properties (MW < 400, clogP < 4), was
investigated.15 Considering the influence of torsion between the
C and D rings on potency,14,15 other spacers in 2 that would disrupt
the planarity between these rings were also evaluated (Table 1).
Starting from derivative 3, small substituents were added on
the benzylic position with the intent of pointing the D ring towards
an active conformation, for instance by hindering free rotation, as
in 6 (Table 1). This modification had no effect on potency. Replace-
ment of the linker with a carbonyl function, as in ketone 7 or amide
8, led to a significant decrease in activity. An extended sp2-carbon
linker on the other hand, such as the E-alkene in 9, led to a 10-fold
increase in potency. A saturated 2-atom spacer, such as in aniline
10 and benzylamine 11 was probably too flexible and did not man-
age to bring the IC50 under 100 nM. Changing the C linker in 3 to a
heteroatom (12–14) led to the discovery of the highly potent
aniline 14, significantly more active than its phenol- and sulfide-
analogs (14 vs 12 and 13). This result would indicate that the
H-bond donor capability of 14 may contribute to the increase in
potency (vide infra). Furthermore, the trifluoromethyl substituent
on the 3-position of the aromatic D ring brought a 13-fold increase
in potency versus the non-substituted analog 15. Restriction of the
free rotation of 15, by cyclizing the aniline into the aromatic ring
(16), did not improve the IC50. Swapping the position of the
aromatic and the methyl substituents on the triazole ring in 15
(compound 17, Fig. 2) also had a detrimental effect on potency.
Aniline 14 (MW = 428, clogP = 4.9) was selected for further
evaluation. When 14 was tested orally in mice at 30 mg/kg it
showed no significant effect (ꢀ17%) on Ab peptides levels 4 h after
the IC50 in vitro as indicated by the coverage ratio (CR = Cu,b
/
mIC50, <0.05, see Table 2). Substituents at the 4-position of the ani-
line ring did not bring an improvement in potency (20, not tested
in vivo). A substituent present at the 2-position of the aniline
(21–28) turned out to be essential for in vivo potency. Mono
2-substituted anilines, either with electron withdrawing (21) or
lipophilic electron donating groups (22), had moderate in vitro
activity, but showed good modulation of Ab peptides levels
in vivo. For 21 and 22, the free brain concentrations were higher
than their respective IC50 in vitro (CR = 1.7 and 2.7, respectively).
A polar methoxy substituent was weakly active (23) and it was
not tested in vivo. Additional substitution at the 5-position
(compounds 24 and 25) increased the in vitro potency and
maintained the in vivo activity. Compound 24 had high brain and
plasma levels and improved B/P ratio (close to 1).
More polar substituents introduced on the 5-position of the ani-
line ring (e.g., dimethyl amine in 25) decreased the lipophilicity of
the compounds (4.3 for 25 vs 5.1 for 24), but did not affect their
potency. For instance, 25 was highly potent in vitro and demon-
strated a surprising 51% lowering of Ab42 in mice at 30 mg/kg,
4 h from administration, despite
a poor brain penetration
(B/P = 0.2) and low mouse free brain concentrations (Cu,b = 4 nM,
F
CF3
D
F
D
D
N
N
N
C
N
N
C N
N
C
N
N
B
B
B
N
N
N
N
A
A
A
OMe
OMe
OMe
N
N
N
2, IC50 = 0.62 μM
3, IC50 = 0.85 μM
4, IC50 = 0.037 μM
Vivo mouse 30 mg/kg:
Aβ42 -62%
Aβ38 +66%
D
L
Brain/Plasma 2.8 μM/ 12.3μM
N
C
N
N
B
N
A
5
OMe
N
Scheme 1.
Note: IC50 represents the concentration of a compound that is required for reducing the Ab42 level by 50%. The IC50 values are a mean of at least 3 determinations. Mouse
in vivo at 30 mg/kg p.o. after 4 h (n = 6) expressed as a percentage of Ab42 lowering in brain compared with untreated animals. Plasma and brain samples were analyzed for
the tested compound using a qualified research LC-MS/MS method.

A. I. Velter et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5805–5813
5807
Table 1
Exploration of linkers between C and D rings
D
L
N
C
N
N
B
N
A
OMe
N
a
a
Compd
L-Ar
IC50
(lM)
Compd
L-Ar
IC50 (lM)
F
F
H
N
3
0.85
0.71
11
0.32
0.27
F
CF3
CF3
6
7
12
13
O
F
O
6.92
6.17
0.65
S
CF
CF
3
3
O
8
9
14
15
16
0.015
0.20
N
H
N
H
F
0.085
0.19
N
H
CF
3
N
10
0.26
N
H
a
IC50 represents the concentration of a compound that is required for reducing the Ab42 level by 50%. The IC50 values are a mean of at least 2 determinations.
F3C
NH
N
CF3
N
N
N
N
N
N
17
OMe
N
N
OMe
N
14
hIC50= 0.28 μM
hIC50 = 0.015 μM
mIC50 = 0.087 μM
Mouse 30 mg/kg PO 4 h
Aβ42 -17%
Mouse Brain total concentration 6 μM (B/P = 0.42)
Mouse Brain free concentration, Cu,b < 0.003 μM
Figure 2.
Note: IC50 represents the concentration of a compound that is required for reducing the Ab42 level by 50%. The IC50 values are a mean of at least 2 determinations. Mouse
in vivo at 30 mg/kg p.o. after 4 h (n = 6) expressed as a percentage of Ab42 lowering in brain compared with untreated animals. Plasma and brain samples were analyzed for
the tested compound using a qualified research LC-MS/MS method.
6-fold lower than mIC50). Introduction of a heteroatom on the ani-
line ring was detrimental to potency as exemplified by 26. Intro-
duction of a N-atom in the anisole ring (compounds 27–29,
Table 1) reduced the clogP by 0.5 while maintaining the in vitro
and in vivo potency. This modification led to a reduced brain pen-
etration and a suboptimal B/P ratio for most of the methoxypyridyl
derivatives (27, 29). On the other hand, the fu,b of these compounds
increased versus their anisole analogs (see 27–29 vs 21, 24 and 14,
respectively). An interesting compound was 27 which, although
relatively weak in vitro, was moderately potent in vivo, showing
39% Ab42 lowering in mice at 30 mg/kg, 4 h after administration.
Compounds 28 and 29 had comparable in vitro potencies, both in

5808
A. I. Velter et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5805–5813
Table 2
Exploration of the aniline ring
Ar
HN
N
N
N
X
N
OMe
N
Brain levelsc
M)
Plasma levelsc Inhib Ab42
vivod (%)
fu,b
Free brain levels
Cu,b (lM)
Cover. ratio
CRg
e
Compd
(clogP)
Ar
X
hIC50
(lM)
mIC5o
(lM)
Ab42a
Ab42a
(
l
(l
M)
(%)
f
F C
3
OMe
18 (4.9)
19 (5)
CH 0.005
CH 0.025
0.050
0.078
3.4
8.6
17.1
18.9
9
<0.05
<0.002
<0.004
<0.04
<0.05
OCF
3
28
nd
<0.05
nd
CF3
20 (4.9)
CH 0.027
0.245
ndh
Nd
nd
nd
21 (4.9)
22 (5.2)
23 (3.89)
CH 0.068
CH 0.060
CH 0.178b
0.13
16
9.7
nd
19.4
26.6
Nd
40
40
nd
1.4
2
0.22
0.20
nd
1.7
2.7
nd
F C
3
0.074
0.257
nd
MeO
CF
3
24 (5.1)
CH 0.011
0.059
0.025
11
9.85
3.5
35
51
0.13
0.6
0.014
0.004
0.24
0.17
F
F
NMe
2
25 (4.33)
CH 0.022b
CH 0.603
0.74
N
26 (3.65)
27 (4.5)
nd
nd
Nd
nd
39
nd
nd
nd
F C
3
N
N
0.135
0.019
0.21
2.64
11.7
4.0
0.10
0.48
F C
3
CF
3
28 (4.6)
29 (4.5)
0.049
0.098
14
17.8
29
55
14
0.5
0.3
0.07
1.4
F
F C
3
N
0.008
3.0
0.009
0.09
a
b
c
IC50 represents the concentration of a compound that is required for reducing the Ab42 level by 50%. The IC50 values are a mean of at least 3 determinations.
Mean of 2 determinations.
Plasma and brain samples were analyzed for the tested compound using a qualified research LC–MS/MS method.
Mouse in vivo at 30 mg/kg p.o. after 4 h (n = 6) expressed as a percentage of Ab42 lowering in brain compared with untreated animals.
Fraction of unbound compound in rat brain tissue (fu,b).
d
e
f
Free brain concentrations, Cu,b, calculated from Cu,b = fu,b ꢁ [brain levels,
lM].
g
h
Coverage ratio, CR, calculated as CR = Cu,b/mIC50
.
nd–not determined.
human and in mouse. Aniline 29 was a less brain penetrant (B/
P = 0.1 vs 0.8 for 28), which combined with a low fu,b (0.3%)
resulted in a free brain concentration 10-fold lower than the IC50
(Table 2). This might explain the difference in the in vivo activity
of the two anilines: while 28 reduced Ab42 levels significantly
4 h after dosing in mouse at 30 mg/kg, 29 had no effect on the
Ab peptides. Compound 28 was one of the most potent methoxy-
pyridyl derivative both in vitro and in vivo, and maintained a rela-

A. I. Velter et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5805–5813
5809
F
H
N
HN
N
CF3
CF3
N
N
CF3
N
N
N
N
N
N
N
N
N
N
N
OMe
N
OMe
32, hIC50 = 0.14 μM
N
OMe
N
30, hIC50 = 0.26 μM
31, hIC50 = 0.16 μM
Figure 3. Other modifications. Replacements of imidazole ring.
Note: IC50 represents the concentration of a compound that is required for reducing the Ab42 level by 50%. The IC50 values are a mean of at least 2 determinations.
Table 3
Variations of the B ring
CF3
CF3
F
HN
N
H
N
N
N
N
N
N
N
CF3
N
N
4
N
1
N
B
N
N
OMe
OMe
N
N
N
43, hIC50 = 0.34 μM
44, hIC50 = 0.67 μM
a
a
Compd
B
IC50
(lM) Compd
B
IC50 (lM)
Figure 4. Replacement of the aromatic D ring with aliphatic substituents.
Note: IC50 represents the concentration of a compound that is required for reducing
the Ab42 level by 50%. The IC50 values are a mean of at least 2 determinations.
4
4
1
1
1
1
1
33
0.12
36
0.045
0.79
F
OMe
4
4
Table 5
N
34
35
0.5
37
38
Compared drug-like profiles of 28 and 1.
OCF3
28
1
4
F
4
clogP
LLE
4.6
3.08
0.019
4
5.9
1.8
0.019
6
1
0.036
0.058
IC50 (lM)
hLM (%)a
OMe
CN
mLM (%)a
0
0
a
Kin. sol. pH 7.4(
lM)
39
3
IC50 represents the concentration of a compound that is required for reducing
the Ab42 level by 50%. The IC50 values are a mean of at least 2 determinations.
CYP450 IC50
3A4 (T)
3A4 (M)
2D6
(lM)
25.4
2.9
8.7
3.6
21.9
3.5
Activator
16.00
12.3
2C9
2C19
Table 4
22.1
Variations of the alkyl substituents on the triazole core
b
hHerg IC50
(l
M)
>10
52
2%
0.5%
17.8
14
8.5
30
F
hHerg PXc (%)
H
N
Fu,h Plasmad
0.13%
<0.05%
19.5
22
1.13
<0.011
62%
e
fu,b
CF3
N
Conc mouse plasma 30 mg/kg,
Conc mouse brain 30 mg/kg,
B/P 4 h
l
M
M
N
R
N
l
0.8
N
Conc mouse brain free 30 mg/kg,
l
M, 4 h
0.069
55%
53%
OMe
Mouse Ab42 lowering, 30 mg/kg,f 4 h
Mouse Ab38 increase, 30 mg/kg,f 4 h
N
144%
a
a
Compd
R
IC50
(
lM)
Compd
R
IC50 (lM)
a
% metabolized after 15 min upon incubation with human (hLM) and mouse
(mLM) liver microsomes at 1 M concentration.
Receptor binding on the human Herg ion channel.
CN
OH
l
39
Et
0.026
0.052
41
0.041
0.076
b
OMe
c
d
e
f
% inhibition of IKr current at 3 lM.
Unbound fraction to human plasma proteins.
Unbound fraction to rat brain tissue.
% inhibition of Ab peptides as determined by Meso Scale.
40
42
a
IC50 represents the concentration of a compound that is required for reducing
the Ab42 level by 50%. The IC50 values are a mean of at least 2 determinations.
tively high B/P ratio (0.8). Methylation of the aniline nitrogen of 28
led to a 14-fold loss of potency (30, Fig. 3), which underlined the
importance of the N–H bond.
Other structural modifications were made, seeking to improve
the efficacy in this series of compounds. Replacement of the
imidazole ring (see 31–32, Fig. 3, vs 14) or of the anisole B ring
(see 33–38, Table 3, vs 24) did not improve the in vitro potency.
Compound 35, three times less potent than parent 24, was one of
the most potent compounds resulting from this exercise. Nevertheless
when tested at 30 mg/kg in mice, 35 showed no significant
lowering of Ab42 levels 4 h after dosing. Bioanalysis of 35 in mice
4 h post dosing showed that this compound suffered from low
exposure both in plasma (0.78
lM) and in brain (0.34 lM), which
could be explained by its poor solubility (<1.5 mg/mL in 20%

5810
A. I. Velter et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5805–5813
A 42
A 38
β
β
103
120
100
80
60
40
20
0
167
100
200
150
100
50
160
164
146
160
82
124
62
100
48
50
81
42
39
0
A
B
Figure 5. Time course of Ab modulation of 28 in non-transgenic mouse brain after a single oral administration of 30 mg/kg. A. Reduction of the Ab42 levels. B. Increase of the
Ab38 levels.
Table 6
inhibit Notch processing. Compared with 1, 28 had improved drug-
like properties (reduced aromaticity, lower clogP, better lipophilic
efficiency) which was reflected in the compound’s ADMET profile:
improved kinetic solubility, weak CyP inhibition potential (IC50
Modulation of Ab peptides in CSF dog
28
1
Conc dog plasma 20 mg/kg, 8 h,
CSF Ab42 loweringa
lM
6.7
1.7
56%
54%
31%
48%
36%
84%
>10
lM overall). The moderate inhibition of the potassium mem-
CSF Ab40 loweringa
brane current by 28 (52% at 3
l
M) did not translate in vivo in anes-
CSF Ab 38 increasea
thetized guinea pigs. When tested orally in mice at 30 mg/kg, 28
showed a significant modulation of the Ab peptides, inhibiting
secretion of Ab42 by 55% and stimulating production of Ab38 levels
by 53%—again in line with a typical profile of a GSM. While 28 was
still highly bound to both plasma and brain proteins, it did show an
improvement over benzimidazole 1 (see Table 5).
The duration of action of 28 was examined in more detail in a
time-course experiment after a single oral dose of 30 mg/kg in
non-transgenic mice. Reduction of Ab42 levels (38%) was observed
1 h after dosing, a maximum inhibition of Ab42 levels being
reached 6 h after treatment. 12 h after dosing, Ab42 levels
remained moderately reduced below vehicle values and returned
a
% inhibition of Ab peptides as determined by alphalisa.
captisol). Replacement of the methyl group on the triazole ring
with other aliphatic substituents at best maintained the in vitro
potency (39–42, Table 4), at the expense of the physicochemical
properties, such as MW and lipophilicity.
Attempts to reduce aromaticity, as in 43 and 44 (Fig. 4) resulted
in a loss of potency.
The most interesting compound in this series remained 28, with
a good balance of the in vitro and in vivo potency. As expected for a
GSM, 28 did not affect the total level of Ab peptides and it did not
R2
Br
Br
a - b
N
N
N
R1
N
Br
R1
NH
N
Br
N
Br
N
N
O
46a, R1 = Me, (a)
46b, R1 = iPr, (a)
47a, R1 = Me, R2
47b, R1 = Me, R2
=
=
, (c)
N
H
CF
3
45
CF
3
CF
3
, (d)
46c, R1
=
, (b)
N
H
47c, R1 = Me, R2
47d, R1 = Me, R2
47e, R1 = Me, R2
47f, R1 = iPr, R2
=
=
=
CF , (e)
O
3
CF , (f)
N
3
H
N
, (g)
CF
3
=
, (h)
N
CF3
47g, R1
=
,
R2 = NHMe, (i)
Scheme 2. Synthesis of triazolo derivatives from dibromotriazole. Reagents and conditions: (a) R1I, NaH, DMF, 100 °C, 12 h, 93–98%; (b) 3-(CF3)-PhB(OH)2, Na2CO3, Cu(OAc)2,
Py, tol, 70 °C, 16 h, 68%; (c) i. 46a, nBuLi, THF, CO (dry ice), ꢀ78 °C, 1 h, 82%; ii. BOP-Cl, DIPEA, 3-(CF3)-PhNH2, DCM, 3 h, 80%; (d) i. 46a, nBuLi, DMF, THF, ꢀ78 °C to rt, 1 h; ii. 3-
(CF3)-PhNH2, NaBH(OAc)3, THF, 0 °C–rt, 16 h, 15%; (e) 46a, 3-CF3-PhOH, K2CO3, DMF, MW, 160 °C, 45 min, 88%; (f) 46a, (RS)(RS)-3-(trifluoromethyl)-cyclohexanamine, K2CO3,
DMF, MW, 160 °C, 90 min, 24%; (g) 46b, DL-3-(trifluoromethyl)piperidine, K2CO3, DMF, MW, 160 °C, 45 min, 45%; (h) 46a, tetrahydroquinoline, LiHMDS, THF, rt, 2 h, 16%; (i)
46c, MeNH2, K2CO3, ACN, MW, 140 °C, 20 min, 68%.

A. I. Velter et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5805–5813
5811
R2
N
O
B
F
N
K+
F
F
B–
R1
O
N
a
b
N
N
N
OMe
48
OMe
c
N
N
OMe
N
8, 10, 12, 16, 17, 43, 44
49
Scheme 3. Reagents and conditions: (a) KHF2, acetone, rt, 1 h, 80%; (b) 47a,b,d,f, Pd(PPh3)4, K2CO3, ACN, water, MW, 150 °C, 20 min, 15–22%; (c) 47c,e,g, Pd(PPh3)4, K2CO3,
DMF, water, MW, 150 °C, 1 h, 6–38%.
F
F
O
N
N
N
N
N
H
N
NHNH2
a
b
+
6
+
N
N
N
N
F
F
OMe
OMe
OMe
N
N
N
50
51
52
53
N
a
+
51
9
54
Scheme 4. Reagents and conditions: (a) K2CO3, nBuOH, 150 °C, 12 h, 33%; (b) NaH, MeI, DMF, rt, 1 h, 6%.
NH
OEt
N
R
N
R
R
b
a
X
X
X
N
N
F
N
N
55a-e
56a-f
57a-f
X = CH or N
R = H, 2-OMe; 2-Me;
2-OMe, 6-F
S
R1
C
Y = CH, N
N
R1 = 3-CF3, 5-OMe; 3-OCF3
c
2-CF3 ; 2-F, 5-CF3; 2-F, 5-NMe2
2-OMe.
Y
58a-g
R1
S
R1
H
Y
N
HN
Y
N
R3
d-h
N
R
N
N
R
OEt
X
N
X
N
N
59a-h
14, 15, 18-20, 23-26,
28, 33, 34, 38 (d)
39, (e)
N
41, (f)
40, (g)
42, (h)
Scheme 5. Reagents and conditions: (a) DMSO, 4-Methylimidazole, K2CO3, 80–120 °C, 1 h, 39–47%; (b) HCl, EtOH, 0 °C to rt, 29–96%; (c) 60, CH3CN, 50 °C, 32 h, 26–90%; (d)
MeNHNH2, MeOH, 50 °C, 1 h, 6–81%; (e) EtNHNH2 ꢁ H2C2O4, DIPEA, MeOH, 50 °C, 1 h, 29%; (f) CN(CH2)2NHNH2, MeOH, 50 °C, 1 h, 32%; (g) MeO(CH2)2NHNH2 ꢁ HCl, DIPEA,
MeOH, 50 °C, 1 h, 35%; (h) t-Bu(OH)NHNH2, MeOH, 50 °C, 1 h, 49%.
to normal 24 h after administration (Fig. 5A). Similarly, Ab38 levels
in brain were significantly increased from 1 to 12 h post dosing, a
64% increase being reached 6 h after treatment (Fig. 5B). Total lev-
els of Ab in brain remained unchanged.
slight increase in bilirubin was present, which normalized within
24 h after dosing. This represents an improvement versus 1, which
in the same canine model showed increase in both liver enzymes
and bilirubin at lower exposure levels.19
When tested in dog at 20 mg/kg, 28 showed a reduction of 56%
of Ab42 in the cerebrospinal fluid (CSF) 8 h post doing, comparable
to that of 1 (see Table 6).19 Upon evaluation of compound 28 in a
one week repeated dose study in dog,18 at 10 and 20 mg/kg/day,
no increase in liver enzyme levels (ALT or AST) was observed. Only
Chemistry. Most analogs from Table 1 were prepared according
to the route described in Schemes 2 and 3. N-alkylation or aryla-
tion of dibromotriazole 45 provided triazolo-derivatives 46a–c in
good to excellent yields. Various substituents were then intro-
duced at the 5-position of 46a–c, as depicted in Scheme 2. The

5812
A. I. Velter et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5805–5813
CN
N
R
R
H
N
PhO
OPh
H
R
R
N
NC
NH2
N
N
N
61
b
N
H2N
N
H2N
N
PhO
N
H
N
a
60
62
63
64
R = 2-CF3; 2-F, 5-CF3; 3-CF3
c
O2N
N
R
R
H
N
N
N
+
N
N
Br
Br
N
N
66
65
Scheme 6. Reagents and conditions: (a) Method 1: 60, THF, 100 °C, 12 h or Method 2: 60, LiHMDS, THF, rt, 3 h, 25–72%; (b) MeNHNH2, iPrOH, 90 °C, 2 h, 59–82%; (c) Method
1: tBuONO, CuBr2, acetonitrile, 56–59%; Method 2: NaNO2, CuBr, HBr, acetonitrile, 73%.
58, thioimidates 59 were obtained. Upon treatment of 59 with
hydrazines, anilinotriazoles listed in Scheme 5 were formed in
yields varying from poor to excellent.
R
H
N
Br
N
R
NH2
N
N
N
+
This route could not provide all desired analogs, and thus for-
mation of triazoles from cyanoimidates using a method developed
by Webb and coworkers23 was investigated (Scheme 6). Addition
of anilines to diphenyl cyanocarbonimidate provided cyano isoure-
as 62 in low to moderate yields. This reaction needed to be per-
formed at higher temperatures and proceeded with longer
reaction times than the ones originally reported. The yields were
improved when the addition was performed in the presence of a
base, LiHMDS in particular. When treated with methyl hydrazine,
62 cyclized to anilinotriazoles 63, usually as single regioisomers.
For the 2-CF3 substituted cyanoisourea 62, the regioisomer 64
was also formed in a low (22%) yield. When 63 were subjected to
Sandmeyer reaction conditions with t-butyl nitrite and Cu(II) bro-
mide,24 bromotriazoles 65 were formed in low to moderate yields.
Side product 66 was difficult to separate from the product mix-
tures, but its formation was diminished when NaNO2 and HBr were
used in the Sandmeyer reaction.
This synthetic pathway proceeded with low overall yields.
Formation of the secondary products such as 66 was another
drawback. Given the beneficial effect of LiHMDS to the addition
reaction of anilines to imidate 61, it was questioned whether this
base would also help in adding anilines 60 directly to bromotriaz-
oles such as 46 (Scheme 7). Previous attempts of introducing ani-
lines on bromotriazole 46a under classical Buchwald conditions
Br
N
Br
N
65
46a
60
R = 2-Me; 2-F, 5-CF
3
Scheme 7. Reagents and conditions: LiHMDS (3 equiv), THF, rt, 2 h, 78–81%.
resulting bromotriazoles 47a–g were then submitted to Suzuki
reaction conditions in the presence of either boronate 48 or
potassium trifluoroborate salt 49 to yield analogs 8, 10, 12, 16,
17, 43, 44.20 Compound 7 was prepared in 68% yield from 3, via
oxidation with manganese oxide.21
The synthesis of compounds 6 and 9 is shown in Scheme 4.
Commercially available nitriles 50 and 54 were condensed with
hydrazide 5114 to give 52 and 9, respectively. Methylation of 52
provided 6 and its regioisomer 53, in a disappointing 1:6 ratio.
Initially, the anilinotriazoles were prepared by addition of
methyl hydrazine to thioureas,22 as exemplified in Scheme 5. Thus,
starting
from
commercially
available
4-F-benzonitriles,
nucleophilic aromatic substitution with 4-Me-imidazole provided
imidazo-aryl derivatives 56 as the major regioisomers, in a moder-
ate yield. Exposure of 56 to ethanolic HCl provided iminoesters 57,
isolated as HCl salts. When 57 were treated with arylisothiocyanates
R1
A
O
B
N
N
X
R
N
H
O
N
+
X
X
X
A
A
A
R
R
R
67
21, 22, 27-29,
31, 32, 35-38
N
48
,
A =
, X = CH, R = OMe
, X = N, R = OMe
N
N
48a, A =
48b, A =
N
N
N
, X = CH, R = CN
N
N
48c, A =
48d, A =
, X =CH, R = OMe
, X =CH, R = OMe
N
N
Scheme 8. Reagents and conditions: 65, Pd(PPh3)4, Cs2CO3, DME, H2O, 16 h, 90 °C, 19–75%.

A. I. Velter et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5805–5813
5813
11. Recent work on GSMs includes: (a) Pettersson, M.; Johnson, D. S.;
Subramanyam, C.; Bales, K. R.; Am Ende, C. W.; Fish, B. A.; Green, M. E.;
Kauffman, Gregory W.; Mullins, P. B.; Navaratnam, T.; Sakya, S. M.; Stiff, C. M.;
Tran, T. P.; Xie, L.; Zhang, Longfei; Pustilnik, L. R.; Vetelino, B. C.; Wood, K. M.;
Pozdnyakov, N.; Verhoest, P. R.; O’Donnell, C. J. J. Med. Chem. 2014, 57, 1046; (b)
Kobayashi, T.; Iwama, S.; Fusano, A.; Kato, Y.; Ikeda, A.; Teranishi, Y.; Nishihara,
A.; Tobe, M. Bioorg. Med. Chem. Lett. 2014, 24, 378; (c) Chen, J. J.; Qian, W.;
Biswas, K.; Yuan, C.; Amegadzie, A.; Liu, Q.; Nixey, T.; Zhu, J.; Ncube, M.; Rzasa,
R. M.; Chavez, F., Jr.; Chen, N.; De Morin, F.; Rumfelt, S.; Tegley, C. M.; Allen, J.
R.; Hitchcock, S.; Hungate, R.; Bartberger, M. D.; Zalameda, L.; Liu, Y.; McCarter,
J. D.; Zhang, J.; Zhu, L.; Babu-Khan, S.; Luo, Y.; Bradley, J.; Wen, P. H.; Reid, D. L.;
Koegler, F.; Dean, C., Jr.; Hickman, D.; Correll, T. L.; Williamson, T.; Wood, S.
Bioorg. Med. Chem. Lett. 2013, 23, 6447.
failed. Gratifyingly, by addition of 3 equiv of LiHMDS to a mixture
of 46a and anilines 60, at ꢀ78 °C, bromoanilinotriazoles 65 were
obtained in high yields and as single regioisomers.
When bromotriazoles 65 were reacted with boronic esters 48
under Suzuki reaction conditions,25 the desired anilinotriazoles
were obtained, in moderate to good yields. A major side product of
this reaction was the homocoupled product derived from 48 (67,
Scheme 8). Formation of 67 was drastically decreased when the
boronic ester was added dropwise at reflux to a solution of 65, base
and catalyst. It was also found that using [1,10-bisdiphenylphosphi-
no)ferrocene]dichloropalladium (II) as catalyst provided better
yields than tetrakis(triphenylphosphine)palladium(0).26
12. (a) Li, T.; Huang, Y.; Jin, S.; Ye, L.; Rong, N.; Yang, X.; Ding, Y.; Cheng, Z.; Zhang,
J.; Wan, Z.; Harrison, D.; Hussain, I.; Hall, A.; Lee, D. H. S.; Lau, L.-F.; Matsukoa,
Y. J. Neurochem. 2012, 121, 277.
In conclusion, starting from the weakly potent GSM 3, explora-
tion of various spacer groups between the triazole ring and the aro-
matic appendix led to the discovery of a new series of potent
gamma secretase modulators, with an improved drug-like pro-
file—compared with the originally reported series around 1. In this
anilinotriazole series, there was a disconnection between the
in vitro potency and the in vivo activity of some of the analogs.
The presence of an ortho-substituent on the aniline ring proved
to be crucial for obtaining in vivo activity. This could be the result
of the higher free fraction in brain tissue of most of the ortho-
substituted anilines in comparison to the meta- and or para-
analogs. In order to better understand the discrepancy between
the in vitro and the in vivo activity observed in the anilinotriazole
series, the coverage ratio was examined. This parameter represents
the ratio between the free brain concentration and the potency,
given by IC50. Some compounds already showed a significant effect
in modulating Ab peptide levels in brain at a coverage ratio ꢂ0.1. A
higher coverage ratio routinely translated into robust reduction in
Ab42 peptide levels. Due to the general low fu,b of all analogs in this
series, high exposures of the compounds in the mouse brain were
generally required for in vivo activity. Compound 28 was
identified, combining good in vitro efficacy with in vivo potency
across species. The lower lipophilicity of 28 versus 1 led to an
improvement of the overall ADME profile versus 1. In dog, treat-
ment with 28 resulted in significant changes in Ab levels and, more
importantly, single and repeated dosing did not affect the liver
function. Compound 28 was therefore progressed towards further
evaluation.
13. Bischoff, F.; Berthelot, D.; De Cleyn, M.; Macdonald, G.; Minne, G.; Oehlrich, D.;
Pieters, S.; Surkyn, M.; Trabanco, A. A.; Tresadern, G.; Van Brandt, S.; Velter, I.;
Zaja, M.; Borghys, H.; Masungi, C.; Mercken, M.; Gijsen, H. J. M. J. Med. Chem.
2012, 55, 9089.
14. Oehlrich, D.; Rombouts, F. J. R.; Berthelot, D.; Bischoff, F. P.; De Cleyn, M.;
Jaroskova, L.; Macdonald, G.; Mercken, M.; Surkyn, M.; Trabanco, A. A.;
Tresadern, G.; Van Brandt, S.; Velter, A. I.; Wu, T.; Gijsen, H. J. M. Bioorg. Med.
Chem. Lett. 2013, 23, 4794.
15. Wu, T.; Gijsen, H. J. M.; Rombouts, F. J. R.; Bischoff, F. P.; Berthelot, D. J. -C.;
Oehlrich, D.; De Cleyn, M. A. J.; Pieters, S. M. A.; Minne, G. B.; Velter, A. I.; Van
Brandt, S. F. A.; Surkyn, M. WO2011006903.
16. Maurer, T. S.; DeBartolo, D. B.; Tess, D. A.; Scott, D. O. Drug Metab. Dispos. 2005,
33, 175.
17. For these compounds, the fraction of unbound compound in brain (fu,b) was
under the detection limit (fu,b < 0.05%).
18. Borghys, H.; Tuefferd, M.; Van Broeck, B.; Clessens, E.; Dillen, L.; Cools, W.;
Vinken, P.; Straetemans, R.; de Ridder, F.; Gijsen, H.; Mercken, M. J. Alzheimer’s
Dis. 2012, 28, 809.
19. Gijsen, H. J. M.; Mercken, M. Int. J. Alzheimer’s Dis. 2012, 2012, 1.
20. The preparation of compound 48 was described in Ref. 14.
21. Berner, H.; Reinshagen, H. Monatsh. Chem. 1975, 1059, 106.
22. Zarguil, A.; Bourkhis, S.; El Efrit, M. L.; Souizi, A.; Essaissi, E. M. Tetrahedron Lett.
2008, 49, 5883.
23. (a) Webb, R. L.; Labaw, C. S. J. Heterocycl. Chem. 1982, 19, 1205; (b) Webb, R. L.;
Eggleston, D. S.; Labaw, C. S.; Lewis, J. J.; Wert, K. J. Heterocycl. Chem. 1987, 24,
275.
24. Doyle, M. P.; Siegfried, B.; Dellaria, J. F. J. Org. Chem. 1977, 42, 2426.
25. The compounds 48a–d were prepared in a similar manner to compound 48
(Ref. 14).
26. Example: synthesis of 28. Step 1. Synthesis of 3-bromo-N-[2-fluoro-5-
(trifluoromethyl)phenyl]-1-methyl-1H-1,2,4-triazol-5-amine, 65. LiHMDS (1 M
in THF, 420 mL, 420 mmol) was added dropwise at 0 °C to a cooled (ice bath)
solution of 3-amino-4-fluorobenzotrifluoride (70 mL, 527.8 mmol) and 1H-
1,2,4-Triazole, 3,5-dibromo-1-methyl-(46a, 50 g, 207.6 mmol) in anhydrous
THF (500 mL). The reaction mixture was stirred at room temperature (rt) for
20 h. A saturated aqueous NH4Cl solution was added slowly. The reaction
mixture was extracted with dichloromethane (DCM) and the organic layer was
washed with brine, dried (MgSO4) and concentrated under reduced pressure.
The resulting slurry was triturated in heptane/DIPE and a solid was formed,
filtered and dried under vacuum at 60 °C, to provide 65 in 78% yield; mp
159.2 °C. 1H NMR (600 MHz, CDCl3) d ppm 3.77 (s, 3H), 6.37 (d, J = 3.9 Hz, 1H),
7.20–7.24 (m, 1H), 7.27–7.29 (m, 1H), 8.34 (dd, J = 7.7, 2.2 Hz, 1H). MS (ESI) m/z
339 [M+H].+ Anal. (C10H7BrF4N4) C, H, N.
Acknowledgments
The authors would like to thank Cellzome, the members of Jans-
sen R&D Biology, ADME and screening departments and the mem-
bers of the purification and analysis department.
Step 2. Synthesis of N-[2-fluoro-5-(trifluoromethyl)phenyl]-3-[6-methoxy-5-(4-
methyl-1H-imidazol-1-yl)-2-pyridinyl]-1-methyl-1H-1,2,4-Triazol-5-amine, 28. A
mixture of 65 (47 g, 138.6 mmol) and Cs2CO3 (104 g, 320 mmol) in a mixture of
DME (200 mL) and H2O (200 mL) was flushed with N2 for 5 min. PdCl2(dppf)
(7.88 g, 10.7 mmol) was added and the r.m. was flushed with N2 for an
additional 5 min. The mixture was heated at 80 °C and a solution of 48a (35 g,
107 mmol) in DME (400 mL) was added dropwise over a period of 4 h. After
addition, the reaction mixture was stirred at 80 °C for an additional 30 min.
The reaction mixture was cooled to rt and the layers were separated. The
aqueous layer was extracted with DME (100 mL). The combined organic layers
were concentrated to a volume of approximately 250 mL and a precipitate was
formed. The precipitated was filtered, washed with DME (50 mL) and dried in
vacuo. The solid was dissolved in an aqueous 4 N HCl sol. (600 mL). The
resulting solution was washed with DCM (100 mL) and EtOAc (100 mL). The
aqueous layer was then cooled with ice and brought to pH 8–9 via the addition
of 50% aq NaOH solution. The resulting precipitate was filtered, washed with
water and dried in vacuo. The product was recrystallized from iPrOH. The solid
was filtered, washed with DIPE and dried to provide 28 in 48% yield; mp
230.1 °C. 1H NMR (400 MHz, DMSO-d6) d ppm 2.17 (s, 3H), 3.89 (s, 3H), 4.02 (s,
3H), 7.28 (s, 1H), 7.36–7.43 (m, 1H), 7.52 (dd, J = 11.1, 8.5 Hz, 1H), 7.65 (d,
J = 7.9 Hz, 1H), 7.89–7.98 (m, 2H), 8.58 (dd, J = 7.7, 2.4 Hz, 1H), 9.22 (s, 1H). MS
(ESI) m/z 447 [M+H].+ Anal. (C20H17F4N7O) C, H, N.
References and notes
1. http://www.who.int/mediacentre/news/releases/2012/dementia_20120411/
en/.
2. Berchtold, N. C.; Cotman, C. W. Neurobiol. Aging 1998, 19, 173.
3. Pohanka, M. Curr. Med. Chem. 2014, 21(3), 356.
4. Hardy, J.; Selkoe, D. J. Science 2002, 297, 353.
5. Karran, E.; Mercken, M.; De Strooper, B. Nat. Rev. Drug Discovery 2011, 10, 698.
6. De Strooper, B.; Vassar, R.; Golde, T. Nat. Rev. Neurol. 2010, 6, 99.
7. Gijsen, H. J. M.; Bischoff, F. P. Annu. Rep. Med. Chem. 2012, 47, 55.
8. De Strooper, B.; Iwatsubo, T.; Wolfe, M. S. Cold Spring Harb. Perspect. Med. 2012,
2, a006304.
9. (a) Devi, L.; Ohno, M. Neurobiol. Dis. 2012, 45, 417–424; (b) Tamayev, R.;
Matsuda, S.; Arancio, O.; D’Adamio, L. EMBO Mol. Med. 2012, 4, 171–179; (c)
Tamayev, R.; D’Adamio, L. Mol. Neurodegener. 2012, 7, 19.
10. For reviews, see: (a) Oehlrich, D.; Berthelot, D. J.-C.; Gijsen, H. J. M. J. Med. Chem.
2011, 54, 669–698; (b) Pettersson, M.; Stepan, A. F.; Kauffman, G. W.; Johnson,
D. S. Expert Opin. Ther. Patents 2013, 23, 1349.