Discovery of 4-aminomethylphenylacetic acids as γ-secretase modulators via a scaffold design approach

Article abstract of DOI:10.1016/j.bmcl.2011.10.047

Starting from literature examples of nonsteroidal anti-inflammatory drugs (NSAIDs)-type carboxylic acid γ-secretase modulators (GSMs) and using a scaffold design approach, we identified 4-aminomethylphenylacetic acid 4 with a desirable γ-secretase modulation profile. Scaffold optimization led to the discovery of a novel chemical series, represented by 6b, having improved brain penetration. Further SAR studies provided analog 6q that exhibited a good pharmacological profile. Oral administration of 6q significantly reduced brain Aβ42 levels in mice and rats.

Full text of DOI:10.1016/j.bmcl.2011.10.047

Bioorganic & Medicinal Chemistry Letters 21 (2011) 7277–7280
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of 4-aminomethylphenylacetic acids as
via a scaffold design approach
c-secretase modulators
Zhili Xin a, , Hairuo Peng a, Andrew Zhang a, Tina Talreja a, Gnanasambandam Kumaravel a, Lin Xu b,
Ellen Rohde b, Mi-yong Jung c, Melanie N. Shackett c, David Kocisko c, Sowmya Chollate c,
Anthone W. Dunah c, Pamela A. Snodgrass-Belt c, H. Moore Arnold c, Arthur G. Taveras a,
Kenneth J. Rhodes c, Robert H. Scannevin c
⇑
a Department of Medicinal Chemistry, Biogen Idec Inc., 14 Cambridge Center, Cambridge, MA 02142, USA
b Department of Drug Metabolism and Pharmacokinetics, Biogen Idec Inc., 14 Cambridge Center, Cambridge, MA 02142, USA
c Department of Neurology Research, Biogen Idec Inc., 14 Cambridge Center, Cambridge, MA 02142, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Starting from literature examples of nonsteroidal anti-inflammatory drugs (NSAIDs)-type carboxylic acid
Received 27 September 2011
Revised 11 October 2011
Accepted 13 October 2011
Available online 19 October 2011
c
-secretase modulators (GSMs) and using a scaffold design approach, we identified 4-aminomethylphe-
nylacetic acid 4 with a desirable -secretase modulation profile. Scaffold optimization led to the discov-
c
ery of a novel chemical series, represented by 6b, having improved brain penetration. Further SAR studies
provided analog 6q that exhibited a good pharmacological profile. Oral administration of 6q significantly
reduced brain Ab42 levels in mice and rats.
Keywords:
Ó 2011 Elsevier Ltd. All rights reserved.
c
-Secretase
Modulator
Amyloid
Alzheimer’s disease
Scaffold design
Alzheimer’s disease (AD) is a neurodegenerative disorder associ-
ated with memory loss and behavioral changes. AD affects more
than 26 million people worldwide, and it is estimated that this
number will reach 106 million by 2050.1 Unfortunately, there are
no disease modifying therapies currently available. It was proposed
trials, but failed to show cognitive improvement in AD patients..7
This compound has poor in vitro potency (Ab42 EC50 ꢀ300 M)
and low brain penetration. Therefore, a -secretase modulator
l
c
(GSM) with improved potency and brain permeability is desirable
for further evaluation of this class of GSMs as a potential treatment
for AD. Major efforts have been made by scientists at Chiesi to im-
prove the potency based on flurbiprofen.8 Recently CHF5074 (2,
that the accumulation of amyloid
c-peptides (Ab) in the brain
drives AD pathogenesis.2 Ab is produced by the sequential cleavage
of amyloid precursor protein (APP) by two aspartic proteases
EC50 = 40
Piperidinyl acetic acids, as exemplified by compound 3 (Ab42
EC50 = 0.60
M, Fig. 1)10 and similar derivatives11 have been re-
l
M, Fig. 1) was advanced into phase II clinical trials.9
known as b- and
emerged as a promising target for the treatment of AD. However,
-secretase is known to process several substrates besides APP
(e.g., Notch), and inhibiting its activity was reported to be associ-
ated with undesirable gastrointestinal toxicities.3
-Secretase mod-
c-secretase. Over the last decade, c-secretase
l
c
ported to have sub-micromolar activity for Ab42 reduction and
c
O
ulation is more desirable than inhibition from a therapeutic
perspective and may reduce the risk of mechanism-based toxicities.
In cultured cells, certain nonsteroidal anti-inflammatory drugs
(NSAIDs) preferentially decrease the levels of Ab42, a highly
amyloidogenic peptide, and increase the levels of Ab38, a shorter
and less neurotoxic peptide.4 Compounds showing this Ab modula-
tion effect in vivo include ibuprofen5 and flurbiprofen.6 R-flurbi-
profen (1, Tarenflurbil, Fig. 1) was advanced into phase III clinical
O
O
F
OH
F
OH
F
OH
F
N
CF3
F3C
Cl
Cl
2, CHF5074
EC50 = 40
1, Tarenflurbil
EC50 ~ 300
3, Merck GSM
EC50 = 0.6
µ
M
µ
M
µM
⇑
Corresponding author. Tel.: +1 617 914 1111; fax: +1 617 679 3635.
E-mail address: zhili.xin@biogenidec.com (Z. Xin).
Figure 1. NSAID-type c-secretase modulators.
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.10.047

7278
Z. Xin et al. / Bioorg. Med. Chem. Lett. 21 (2011) 7277–7280
promising in vivo pharmacological profiles. Recent work in this
field has been reviewed extensively.12,13 We have disclosed 4-ami-
nomethyl acetic acid-based GSMs in recent publications.14,15 Here-
in, we describe the discovery of a novel series of GSMs, represented
by 6q, via a scaffold design and optimization approach.
high MW (573) and increased conformational flexibility of com-
pound 5c may have contributed to its decreased brain exposure.
Unfortunately, analogs with
a reduced MW, such as 5d
(MW = 467, EC50 = 5.6 M), were found to be less active. To im-
l
prove brain penetration while maintaining potency, we decided
to increase the molecular rigidity by reducing rotatable bond
count, and explore SAR based on unsubstituted acetic acid 5a with
a relatively lower MW. An initial attempt was to keep the basic
nitrogen at the benzylic position and relocate the isopentyl side
Phenylacetic acid 4 (Ab42 EC50 = 1.5 lM in APP CHO cells, Fig. 2)
was the first compound in the series showing a modulatory profile.
This compound decreased the Ab42 level, increased Ab38 secre-
tion, and had no effect on Ab40 in cultured cells. Introduction of
a benzylic nitrogen likely reduces the lipophilicity of the molecule,
and provides the flexibility for the associated substitutions to
adopt favorable hydrophobic interactions. Benzoic acid analog 4a
chain of 5a to the
provided 6a with a similar Ab42 EC50 value (0.45
a
-position of the benzylamine. This modification
M) to 5a, sug-
l
gesting that substitution is well tolerated at this position. The sub-
sequent replacement of the cyclohexylamine by a cyclic tertiary
amine, such as piperidine, led to a structurally more rigid analog
6b with equipotent cellular activity. Compound 6b demonstrated
enhanced brain penetration and PD response in mice. It achieved
about 35% brain Ab42 reduction (50 mg/kg, po) at the 4 h time
was three- to fourfold less potent (EC50 = 5.4 lM), suggesting that
the extended acid functional group is preferred. Compound 4
was subjected to a pharmacodynamic (PD) study in wild-type mice
(male CF-1 mice, 50 mg/kg, po), where the percent of Ab42 level
reduction versus a vehicle control was evaluated 4 h post-dose.
Unfortunately, this compound showed no significant reduction in
brain Ab42 levels. The concentration of compound 4 was found
point. The brain and plasma exposure levels were 17 and 32 lM,
respectively. Compound 6b exhibited a favorable pharmacokinetic
(PK) profile in mice characterized by its excellent oral bioavailabil-
ity (F = 100%), low clearance (CL = 3.2 ml/min/kg) and a half-life of
2.9 h.
to be low (0.6
sure was significant (15
l
M) in brain samples, even though the plasma expo-
M).
l
It was observed that replacement of the aniline by a cyclohex-
ylamine resulted in analog 5a with an improved brain penetration
Based on the encouraging results associated with compound 6b,
replacement of the 4-trifluoromethylpiperidine moiety to further
reduce the molecular weight while improving or maintaining cel-
lular potency was next examined. A number of piperidines and
pyrrolidines were evaluated as these cyclic amines feature less
conformational mobility. The presence of an electron-withdrawing
substituent on the piperidine ring seemed to be required for po-
tency. As shown in Table 1, the EC50 values steadily increased from
(brain–plasma ratio ꢀ1:3) and cellular potency (EC50 = 0.67
However, introduction of a less lipophilic side chain, such as a tri-
fluoromethylpyridyl group (5b, EC50 = 5.0 M), significantly re-
lM).
l
duced activity. A single dose of compound 5a in mice (50 mg/kg,
po, 4 h) selectively reduced the brain Ab42 levels by 24%. The brain
exposure level was ꢀ7
in vitro EC50 value. To further increase Ab42 potency, introduction
of additional substituents at the -position of the phenylacetic acid
were explored. This work culminated in compound 5c (racemic,
EC50 = 0.12 M), which was about fivefold more potent than 5a.
lM, approximately 10 times higher than the
a
Table 1
l
In vitro activity of GSM 6b–i against Ab42
However, this gain in in vitro potency was offset by a loss of brain
penetration. No Ab42 reduction was observed when compound 5c
was dosed at 50 mg/kg in mice, the brain exposure level was only
O
OH
0.7
lM.
Physicochemical properties such as molecular weight (MW)
and rigidity are known to greatly affect BBB penetration. Com-
pounds with higher MW are generally less brain permeable. The
R1
N
CF3
R2
6b-i
Compd
NR1R2
Ab42 EC50 (l
M)a
O
O
R
OH
OH
n
N
6b
0.62
F3C
N
t-Bu
F3C
N
R'
X
N
CF3
6c
0.52
CF3
t-Bu
N
5a (R = H, X = C, R' = CF3), EC50 = 0.67 µM
5b (R = H, X = N, R' = CF3), EC50 = 5.0 µM
5c (R = i-Bu, X = C, R' = CF3), EC50 = 0.12 µM
5d (R = H, X = C, R' = F), EC50 = 5.6 µM
6d
6e
6f
1.1
10
4 (n = 1), EC50 = 1.5 µM
4a (n = 0), EC50 = 5.4 µM
N
N
O
O
3.3
F
F
OH
OH
N
6g
6h
6i
3.3
5.9
>10
t-Bu
F
N
H
CF3
N
CF3
N
F3C
F
F
N
6b, EC50 = 0.62 µM
6a, EC50 = 0.45 µM
Figure 2. Design and evolution of a phenylacetic acid scaffold as a
modulator.
c-secretase
a
EC50 values were calculated as an average of at least two determinations.

Z. Xin et al. / Bioorg. Med. Chem. Lett. 21 (2011) 7277–7280
7279
CF3 (6b, EC50 = 0.62
ing CH3 group (6e, EC50 = 10
lipophilicity was found to be beneficial. Compared to the methyl
analog 6e, compound 6d (EC50 = 1.1 M), carrying a bulkier t-butyl
group, displayed a ninefold decrease in its EC50 value. The pyrroli-
dine derivatives were found to be less potent in the cell-based as-
l
M) to F (6f, EC50 = 3.3
l
M) to electron-donat-
6n, a low micromolar GSM (EC50 = 1.4
sus 6b). Compounds with
(EC50 = 0.42 M), 6p (EC50 = 0.54
well tolerated in the cellular assay. The brain Ab42 reduction of 6o
and 6p were weak to moderate (15% and 30%, respectively, 4 h,
30 mg/kg) when dosed orally in PD studies. However, 6q reduced
brain Ab42 levels by 50% (30 mg/kg, po, 4 h). Considering its mod-
erate in vitro potency, it was not surprising that the brain exposure
l
M; twofold less potent ver-
lM). On the other hand, increasing
a
bulkier R4 group such as 6o
M) and 6q (EC50 = 0.63 M) were
l
l
l
l
say (6h, EC50 = 5.9
substituents explored, trifluoromethyl substituted piperidines
exhibited the highest potency. Compound 6c (EC50 = 0.52 M), a
lM; 6i, EC50 >10 lM). Among the different
l
levels of 6q in mice were very high (32 lM) with a brain to plasma
regioisomer of 6b, was tested as a mixture of four diastereomers
and showed a similar cellular potency.
ratio of ꢀ1. Compound 6q also showed a dose-dependent lowering
of brain Ab42 (12.5, 25, and 50 mg/kg). At 12.5 mg/kg oral dose, 6q
Based on review of the data, it was decided to explore the SAR of
4-trifluoromethylpiperidine analog 6b, which contains only one
stereogenic center, and to investigate the substitution at R3 posi-
tion for a potential potency gain. As shown in Table 2, compound
reduced Ab42 levels by 25% with a brain exposure level of 9 lM.
In addition to its high brain penetration and good PD/PK correla-
tion in mice, compound 6q demonstrated favorable ADME proper-
ties. It showed good liver microsomal stability (RLM Qh = 31%, HLM
Qh = 39%) and good Caco-2 permeability [Papp(A À B) = 16 Â 10
À6 cm/s, P ratio (B À A/A À B) = 1.9]. Table 3 illustrates the pharma-
cokinetic parameters of 6q in rats and dogs. The clearance of 6q was
low in both species (rat = 2.5 ml/min/kg; dog = 0.6 ml/min/kg).
6j (EC50 = 0.34
potent relative to the parent 6b (EC50 = 0.62
methyl derivative (6k, EC50 = 0.46 M) offered no significant
advantage in cellular potency, whereas the -isobutyl substitution
(6l, EC50 = 0.21 M) enhanced activity by threefold. However, brain
lM), an
a
-methyl analog of 6b, was slightly more
l
M). The gem-di-
l
a
l
Compound 6q showed no significant hERG activity at 10
did it inhibit any of the five major human CYP isoforms assayed
(3A4, 2C9, 2C19, 2D6, and 1A2; IC50 >10 M).
lM nor
concentrations tended to decrease with bulkier substituents. Simi-
lar to an earlier observation, substituted isobutyl analog 6l dis-
l
played a lower drug concentration in the brain (4
to the unsubstituted analog 6b (17 M), or methyl-substituted ana-
log 6j (13 M). These results correlated well with a decreased PD
response of 6l in the brain. We next focused our efforts on the mod-
ification of the isopentyl side chain. Truncation of the isopentyl
group resulted in the complete loss of potency (6m, EC50
l
M) compared
Chiral chromatographic separation of compound 6q was per-
formed to obtain two enantiopure isomers, both of which dis-
l
l
played very similar cellular potency (IC50 ꢀ0.7
lM) and PD
response in rats (25% [enantiomer 1] and 30% [enantiomer 2]
Ab42 reductions in the brain at 30 mg/kg). The isolated enantiomer
2 (absolute configuration not assigned) that achieved a slightly
>10
l
M). Installation of a CF3-substituted alkyl side chain rendered
higher brain exposure level (26
lM) and brain Ab42 reduction
Table 2
SAR summary of GSM 6b, 6j–q
O
R3
OH
N
R4
CF3
F3C
6b, 6j-q
c
R3
R4
Ab42 EC50
(l
M)a
Brain Ab42 reductionb
B/P
(lM/l
M)
6b
6j
H
0.62
35% (50 mpk)
17/32
15/22
13/21
Me
0.34
0.46
35% (30 mpk)
20% (30 mpk)
6k
Me, Me
6l
i-Bu
0.21
>10
21% (30 mpk)
NDd
4/14
NDd
6m
H
H
6n
6o
H
H
1.4
NDd
NDd
CF3
0.42
15% (30 mpk)
10/23
6p
6q
H
H
0.54
0.63
30% (30 mpk)
50% (30 mpk)
18/25
32/37
a
b
c
EC50 values were calculated as an average of at least two determinations.
PD response were measure at 4 h after oral administration of compounds at 30 mg/kg in a solution formulation (ethanol/propylene glycol/10% solutol = 1:1:8).
B/P, brain/plasma concentration.
ND, not determined.
d

7280
Z. Xin et al. / Bioorg. Med. Chem. Lett. 21 (2011) 7277–7280
Table 3
Single-dose PK parameters of 6q
Species
Dose (mg/kg)
AUC/dose ng à h à kg/(ml à mg)
Cmax (ng/ml)
CL (ml/min/kg)
T1/2 (h)
Rat (n = 2)
2
0.5
5495
28712
—
1125
2.5
0.6
9.6
10.1
Doga (n = 2)
a
From enantiomer 2.
O
desirable pharmacokinetic properties in mice, rats and dogs. This
compound also exhibits good brain penetration and a robust PD re-
sponse with selective lowering of brain Ab42 levels in mice and
rats. Furthermore, compound 6q was shown to be low risk in
in vitro hERG and CYP inhibition assessments, and is thus suitable
for further evaluation.
O
O
H
O
OH
OMe
OMe
OMe
OH
7
e, f
d
b, c
t-Bu
t-Bu
CF3
N
CF3
5a
N
H
CF3
O
O
H
a
9
8
O
O
O
Acknowledgments
O
OMe
OH
OMe
OMe
OMe
OH
h
b, c
g
The authors thank the following Biogen Idec personnel: Harriet
Rivera for compound management, Xiaoping Hronowski for HRMS
analysis, the DMPK group for permeability, metabolic stability and
PK studies, and the Neuropharmacology group for in vivo studies.
O
HO
CF3
12
O
CF3
11
10
i
O
O
O
Supplementary data
OH
OMe
OMe
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2011.10.047.
f
j
N
CF3
N
CF3
I
CF3
References and notes
F3C
F3C
6q
14
13
1. Brookmeyer, R.; Johnson, E.; Ziegler-Graham, K.; Arrighi, H. M. Alzheimer’s
Dement. 2007, 3, 186.
2. Hardy, J.; Selkoe, D. J. Science 2002, 297, 353.
3. van Es, J. H.; van Gijn, M. E.; Riccio, O.; van den Born, M.; Vooijs, M.; Begthel, H.;
Cozijnsen, M.; Robine, S.; Winton, D. J.; Radtke, F.; Clevers, H. Nature 2005, 435,
959.
4. Weggen, S.; Eriksen, J. L.; Das, P.; Sagi, S. A.; Wang, R.; Pietrzik, C. U.; Findlay, K.
A.; Smith, T. E.; Murphy, M. P.; Bulter, T.; Kang, D. E.; Marquez-Sterling, N.;
Golde, T. E.; Koo, E. H. Nature 2001, 414, 212.
Scheme 1. Conditions and reagents: (a) MgCl2, Et3N, paraformaldehyde, MeCN,
reflux, 9%; (b) Tf2O, pyridine, CH2Cl2, 78–97%; (c) 4-CF3-phenylboronic acid,
(Ph3P)4Pd, 77–90%; (d) 4-t-butylcyclohexylamine, NaBH(OAc)3, 45%; (e) isovaleral-
dehyde, NaBH(OAc)3, 100%; (f) NaOH, 80–100%; (g) 3,3-dimethylbutanoic acid,
EDCI, Et3N, CH2Cl2; BF3ÁEt2O, 25%; (h) BH3 in THF, 0 °C, 40%; (i) NaI, BF3ÁEt2O,
CH2Cl2; (j) 4-CF3-piperidine hydrochloride, Et3N, CH2Cl2, 40% for two steps.
5. Lim, G. P.; Yang, F.; Chu, T.; Chen, P.; Beech, W.; Teter, B.; Tran, T.; Ubeda, O.;
Ashe, K. H.; Frautschy, S. A.; Cole, G. M. J. Neurosci. 2000, 20, 5709; Ing, R.;
Yoshida, M.; Mariano, P. S. J. Org. Chem. 1996, 61, 4439.
6. Eriksen, J. L.; Sagi, S. A.; Smith, T. E.; Weggen, S.; Das, P.; McLendon, D. C.; Ozols,
V. V.; Jessing, K. W.; Zavitz, K. H.; Koo, E. H.; Golde, T. E. J. Clin. Invest. 2003, 112,
440.
(30%) was evaluated for Notch activity by immunoblot analysis of
endogenous HES1, a Notch intracellular domain (NICD) regulated
transcription product. This enantiomer of 6q showed no inhibitory
activity at 10 lM.
7. Imbimbo, B. J. Alzheimers Dis. 2009, 17, 757.
The syntheses of 5a and 6q are outlined in Scheme 1. Formyla-
tion of methyl 2-(3-hydroxyphenyl)acetate with paraformalde-
hyde in the presence of magnesium chloride provided a low yield
of salicylaldehyde 7. The phenol 7 was transformed to a triflate
intermediate, which was converted to 8 via a Suzuki coupling.
Reductive amination with an excess of cis/trans-4-tert-butylcycloh-
exanamine generated amine 9 as a pure trans-isomer. A second
reductive amination with isovaleraldehyde was followed by hydro-
lysis with NaOH to provide compound 5a in excellent yield.
Synthesis of compound 6q was accomplished via similar proce-
dures starting with acylation of 2-(3-hydroxy-phenyl)acetate to
yield the arylketone 10. Triflation and a Suzuki coupling afforded
key intermediate 11. A three step sequence that involved reduction
to 12, iodination to 13 and displacement to install the amine moi-
ety of 14 was performed. The direct reductive amination of 11 did
not proceed well. Treatment of ester 14 with NaOH proceeded
smoothly to afford the corresponding acid 6q.
8. Peretto, I.; Radaelli, S.; Parini, C.; Zandi, M.; Raveglia, L. F.; Dondio, G.;
Fontanella, L.; Misiano, P.; Bigogno, C.; Rizzi, A.; Riccardi, B.; Biscaioli, M.;
Marchetti, S.; Puccini, P.; Catinella, S.; Rondelli, I.; Cenacchi, V.; Bolzoni, P. T.;
Caruso, P.; Villetti, G.; Facchinetti, F.; Giudice, E. D.; Moretto, N.; Imbimbo, B. P.
J. Med. Chem. 2005, 48, 5705.
9. Evaluation of safety & tolerability of multiple dose regimens of CHF 5074 and
exploration of effects on potential markers of clinical efficacy in patients with
mild cognitive impairment (CT04). http://clinicaltrials.gov/ct2/show/
NCT01303744.
10. Stanton, M. G.; Hubbs, J.; Sloman, D.; Hamblett, C.; Andrade, P.; Angagaw, M.;
Bi, G.; Black, R. M.; Crispino, Jamie.; Cruz, J. C.; Fan, E.; Farris, G.; Hughes, B. L.;
Kenific, C. M.; Middleton, R. E.; Nikov, G.; Sajonz, P.; Shah, S.; Shomer, N.;
Szewczak, A. A.; Tanga, F.; Tudge, M. T.; Shearman, M.; Munoz, B. Bioorg. Med.
Chem. Lett. 2010, 20, 755.
11. Hall, A.; Elliott, R. L.; Giblin, G. M. P.; Hussain, I.; Musgrave, J.; Naylor, A.; Sasse,
R.; Smith, B. Bioorg. Med. Chem. Lett. 2010, 20, 1306.
12. Oehlrich, D.; Berthelot, D. J.-C.; Gijsen, H. J. M. J. Med. Chem. 2011, 54, 669.
13. Pattersson, M.; Kauffman, G. W.; am Ende, C. W.; Patel, N. C.; Stiff, C.; Tran, T.
P.; Johnson, D. S. Expert Opin. Ther. Patents 2009, 21, 205.
14. Peng, H.; Cuervo, J.H.; Ishchenko, A.; Kumaravel, G.; Lee, W.-C.; Lugovskoy, A.;
Talreja, T.; Taveras, A. G.; Xin, Z. WO2010138901, 2010.
15. Peng, H.; Talreja, T.; Xin, Z.; Cuervo, J. H.; Kumaravel, G.; Humora, M. J.; Xu, L.;
Rohde, E.; Gan, L.; Jung, M.; Shackett, M.; Chollate, S.; Dunah, A. W.; Snodgrass-
belt, P. A.; Arnold, H. M.; Taveras, A. G.; Rhodes, K. J.; Scannevin, R. H. Med. Lett.
2011, 2, 786.
In summary, we have discovered a novel series of 4-aminophe-
nylacetic acids as
c-secretase modulators through a scaffold design
approach. This has culminated in compound 6q, which has