Discovery of a novel series of quinolone α7 nicotinic acetylcholine receptor agonists

Article abstract of DOI:10.1016/j.bmcl.2013.01.070

High throughput screening led to the identification of a novel series of quinolone α7 nicotinic acetylcholine receptor (nAChR) agonists. Optimization of an HTS hit (1) led to 4-phenyl-1-(quinuclidin-3-ylmethyl) quinolin-2(1H)-one, which was found to be potent and selective. Poor brain penetrance in this series was attributed to transporter-mediated efflux, which was in turn due to high pK_a. A novel 4-fluoroquinuclidine significantly lowered the pK_a of the quinuclidine moiety, reducing efflux as measured by a Caco-2 assay.

Full text of DOI:10.1016/j.bmcl.2013.01.070

Bioorganic & Medicinal Chemistry Letters 23 (2013) 1684–1688
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of a novel series of quinolone
a7 nicotinic acetylcholine
receptor agonists
⇑
Ivar M. McDonald , Robert A. Mate, F. Christopher Zusi, Hong Huang, Debra J. Post-Munson,
Meredith A. Ferrante, Lizbeth Gallagher, Robert L. Bertekap Jr., Ronald J. Knox, Barbara J. Robertson,
David G. Harden, Daniel G. Morgan, Nicholas J. Lodge, Steven I. Dworetzky,
Richard E. Olson, John E. Macor
Bristol–Myers Squibb Pharmaceutical Research Institute, 5 Research Parkway, Wallingford, CT 06492-7660, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
High throughput screening led to the identification of a novel series of quinolone a7 nicotinic acetylcho-
Received 7 December 2012
Revised 11 January 2013
Accepted 16 January 2013
Available online 29 January 2013
line receptor (nAChR) agonists. Optimization of an HTS hit (1) led to 4-phenyl-1-(quinuclidin-3-
ylmethyl)quinolin-2(1H)-one, which was found to be potent and selective. Poor brain penetrance in this
series was attributed to transporter-mediated efflux, which was in turn due to high pKa. A novel 4-fluor-
oquinuclidine significantly lowered the pKa of the quinuclidine moiety, reducing efflux as measured by a
Caco-2 assay.
Keywords:
Ó 2013 Elsevier Ltd. All rights reserved.
a7 Nicotinic acetylcholine receptor
Schizophrenia
Efflux
Basicity
The
a
7 nicotinic acetylcholine receptor (nAChR) has been a tar-
A high-throughput screen of the Bristol–Myers Squibb com-
pound collection for 7 nAChR agonists was performed using a
Fluorometric Image Plate Reader (FLIPR) that measured calcium
ion influx in HEK293 cells expressing the rat
7 nAChR.8,9 Quino-
lone 1 (Fig. 2) was identified as a novel 7 nAChR agonist (
get of significant and growing interest in neuroscience research for
more than two decades as evidence has accumulated that it may be
a viable target for treating the cognitive deficits and negative
symptoms seen in schizophrenia.1–3 Although current antipsy-
chotic medications are able to manage the positive symptoms of
schizophrenia (e.g., hallucinations, delusions), no treatments exist
which alleviate the negative symptoms (e.g., reduced affect, anhe-
donia, social withdrawal) or cognitive deficits which are also core
a
a
a
a7
EC50 = 1.3
lM). Although the potency was only modest, we were
features of the disorder. Several
a7 nAChR agonists have demon-
N
H
strated improvement in preclinical models of memory and cogni-
tion, and a few examples have shown signs of alleviating both
cognitive deficits and negative symptoms in clinical trials for
N
Me
MeO
N
N
schizophrenia.2
a7 nAChR agonists have also become targets of
OMe
O
interest for alleviating the cognitive deficits seen in Alzheimer’s
disease and as anti-inflammatory agents.2,4
nicotine
GTS-21
O
Several known
tylcholine (ACh) is the natural ligand for the
the known 7 nAChR agonists can be considered ACh mimetics,
a
7 nAChR agonists are shown inFigure 1.5–7 Ace-
N
a7 nAChR, and most of
N
O
a
O
Acetylcholine
(ACh)
containing a carbonyl or carbonyl isostere connected by a three-
atom linker to a basic amine, expected to be highly charged at
physiological pH.
S
Cl
NH
N
N
AZD0328
EVP-6124
⇑
Corresponding author. Tel.: +1 203 677 5658; fax: +1 203 677 7884.
E-mail address: ivar.mcdonald@bms.com (I.M. McDonald).
Figure 1. Examples of known a7 nAChR agonists.
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.01.070

I. M. McDonald et al. / Bioorg. Med. Chem. Lett. 23 (2013) 1684–1688
1685
Table 2
SAR of amine analogs
CN
O
R2
O
N
N
R1
1
N
13-21
Figure 2. Lead compound from high-throughput screen.
R1
R2
a7 nAChR EC50 (lM)
a
Compd
13b
H
>10
N
N
14b
15
H
>10
4.3
9.3
a
b
CN
O
O
NH2
N
H
CN
CN
N
2
16
N
CN
O
c
CN
+
CN
O
17
CN
CN
0.90
0.46
N
H
N
N
O
N
18b
NH
N
Cl
Cl
3
4
5
19
H
0.21
>10
NH
N
Scheme 1. Reagents and conditions: (a) cyanoacetic acid, PCl5,CH2Cl2, reflux
30 min, then NaOH (87%); (b) NaH, 3-chloro-1-iodopropane, DMF, rt, 3.5 h (3:21%,
4:38%); (c) dimethylamine, THF, NaI (cat) 70 °C, 4 h (60%).
20b
21c
22c
CN
H
Table 1
a
7 nAChR FLIPR data for selected compounds
0.26
0.15
NH
N
O
R2
R2
O
R2
O
H
N
R1
1, 8-10
O
N
N
R1
11
R1
a
b
c
All values are averages of at least two independent experiments.
Racemic.
Homochiral, final compounds separated by chiral supercritical fluid chroma-
12
Compd
R1
R2
a
7 nAChR EC50 (lM)
a
tography (SFC). Data reported for more potent enantiomer (absolute configuration
not determined).
1
CN
1.3
2.8
N
N
encouraged by good selectivity (EC50s >100
1b1d ; IC50 = 19 M at the closely related 5HT3 receptor), rela-
tively low molecular weight (331 g/mol) and novelty compared
to other known 7 agonists. Another interesting aspect of quino-
lone 1 was that the 4 atom linker connecting the basic amine
and carbonyl moieties was longer than those present in most
known agonists (see Fig. 1 for comparison).10
lM at a3b4, a4b2 and
a
e
l
8
H
a
9
H
H
>10
>10
N
10
N
The initial route to quinolone 1 and analogs varying at the
amine or in linker length followed along the lines of the previously
reported synthesis (Scheme 1).11 Thus, acylation of o-aminobenzo-
phenone with cyanoacetic acid followed by intramolecular Knoe-
venagel condensation led to cyanoquinolone 2, which was
treated with sodium hydride and iodochloropropane to afford a
separable mixture of the O- and N-alkylated compounds (3, 4).
11
12
CN
CN
>10
>10
N
N
a
All values are averages of at least two independent experiments.

1686
I. M. McDonald et al. / Bioorg. Med. Chem. Lett. 23 (2013) 1684–1688
with the secondary amine 17 showing a modest boost in potency
and the conformationally restricted piperidines (18, 19) and
morpholine (21) showing a significant improvement in potency.
While the cyano analog 1 was somewhat more potent than the
corresponding des-cyano compound 8 in the dimethylaminopropyl
series, for the piperidine analogs this was reversed with the
des-cyano compound 19 showing improved potency. Interestingly,
N-methylation of the piperidine was not tolerated (20), again
highlighting the sensitivity of this moiety toward steric bulk. A
a,b
c
NH2
N
O
N
O
R
R
6
7
8
Scheme 2. Reagents and conditions: (a) RCHO, NaHB(OAc)3, DCE, reflux (32–53%);
(b) phenylpropiolic acid, SOCl2, benzene (52–99%); (c) TfOH (neat), rt, 30 min (18–
45%).
Table 3
a7 nAChR FLIPR data for various piperidine and quinuclidine substituted quinolones
R2
O
OH
O
a
b
N
R1
O
O
O
N
R
O
N
R
O
N
H
O
28-36
23
24
25
Compda
R1
R2
a7 nAChR EC50 (lM)
b
OTf
Ar
d,e
or
d,f
28
1.5
1.3
NH
NH
c
N
R
O
N
R1
O
N
O
26
27
29
30
H
N
NH
NH
0.11
R=
R1=
N
Boc
N
BH3
N
H
O O
S
A
B
A
B
N
31
32
0.077
Scheme 3. Reagents and conditions: (a) R–OH, PPh3, DIAD, THF, rt (58–85%); (b)
EtOAc, LHMDS, THF, then warm in toluene (20–61%); (c) Tf2O, Et3N (55–76%); (d)
ArB(OH)2, PdPPh3, K2CO3; (e) TFA, DCM; (f) 1:1, 3 N HCl/acetone (73–91%, 2 steps).
H
O
NH
NH
N
H
0.10
The desired N-alkylated compound 4 was then treated with a sec-
ondary amine and catalytic sodium iodide to afford the final com-
pound (5). The syntheses of the de-annulated pyridone 11 and the
des-phenyl quinolone 12, were carried out in an analogous manner
from the commercially available heterocyclic analogs of 2.
To readily access analogs bearing more complex amine side-
chains and avoid unproductive O-alkylation, an alternate synthesis
was employed. Reductive alkylation of aniline with an aldehyde
followed by acylation with phenylpropynoyl chloride led to amide
7 which was then rapidly cyclized in the presence of neat triflic
acid to afford the des-cyano quinolone analogs 8.12
33
>10
H
N
34c
0.023
N
N
Results of core modification of quinolone 1 are shown in Table
O O
S
1.13 De-annulating the quinolone ring to the pyridone 11 or remov-
N
35c
0.014
0.032
ing the pendant phenyl ring (12) led to a loss of a7 nAChR potency.
H
However, the cyano group could be removed without substantially
impacting potency (8 vs 1). Altering the chain length linking the
quinolone core to the amine was not tolerated, with the 2-carbon
(9) and 4-carbon (10) linkers showing a significant loss in activity.
Replacing the dimethylamino group with other tertiary amines
also led to significant loss in potency, even with simple analogs
such as ethylmethylamine 15 or pyrrolidine 16 (Table 2). Similarly,
simple substitutions on the chain linking the amine to the quino-
O
N
N
H
36c
a
b
c
All compounds racemic.
All values are averages of at least two independent experiments.
At least 50-fold selective versus 5HT3, at least 100-fold selective vs. other nic-
3b4, 4b2 and 1b1d ).
lone core such as the
not tolerated. Changes to the amine moiety that reduced terminal
bulk or provided conformational restriction were more fruitful,
a- and b-methyl compounds (13, 14) were
otinic receptors (
a
a
a
e

I. M. McDonald et al. / Bioorg. Med. Chem. Lett. 23 (2013) 1684–1688
1687
Table 4
7 Concentration response EP data for selected compounds
further modest boost in potency was observed with the more con-
formationally restricted quinuclidine 22.
a
a
The two best amine groups (piperidine and quinuclidine) were
chosen for further derivitization by changing the functionality and
substitution of the pendant phenyl ring. In order to access such
analogs efficiently an alternate sequence was devised, which al-
lowed for late-stage incorporation of the aryl group. Thus, isatoic
anhydride (23) and either N-Boc-3-hydroxymethyl piperidine or
3-hydroxymethyl quinuclidine-borane complex14 were coupled
under Mitsunobu conditions to afford heterocycles 24 (Scheme
3). Addition of the enolate of ethyl acetate provided the hydrox-
yquinolone 25, which was then activated as the triflate to allow
for Suzuki coupling with a variety of aryl and heteroaryl boronic
acids. Final deprotection was carried out either with TFA for the
Boc deprotection, or with a mixture of acetone and aqueous HCl
for the borane-protected quinuclidines.
Compd
a7 EP EC50
(l
M,
a
7 EP Ymax (area)
a7 EP Ymax,
(peak)
obs
area)
(%)
(%)
1
8.2
2.5
1.2
0.76
0.54
0.26
NFd
44
30
41
15
60
34
60
62
21
21
19b
21b
22b
34c
35c
36c
110
80
97a
84a
a
Observed Ymax
.
b
Homochiral, final compounds separated by chiral supercritical fluid chroma-
tography (SFC). Data reported for more potent enantiomer (absolute configuration
not determined).
c
Racemic.
Curve could not be fit to the data.
d
A broader SAR screen of aryl analogs was carried out on the
more accessible piperidine compounds with the results guiding
target selection in the quinuclidine series (Table 3). In the piperi-
dine series, substitution at the 4-position of the phenyl ring was
not tolerated (see example 33) and substitution at the 2-position
in most cases led to reduction in potency (see example 28). Substi-
tution at the 3-position was better tolerated, with polar function-
ality containing a hydrogen bond donor leading to a significant
boost in potency (30–32). When applied to the quinuclidine series,
the potency was even further improved for indole 34, sulfonamide
35 and acetamide 36.
Table 5
a
7 pKa and penetrance data for selected compounds
Compound PAMPA Pc (pH 7.4;
nm/s)
pKa
Caco-2 efflux
ratio
Brain/
plasma
AZD0328a
21a
510
550
900
400
0
8.9
8.1
10.1 6.9
NDe >18
NDe NDe
NDe NDe
0.9
1.0
1.7–2.7b
0.9b
22a
<0.02c
NDe
34d
35d
NDe
36d
1
380
NDe
43d
7.6
0.6
NDe
Key compounds were further studied in a whole-cell patch-
clamp electrophysiology paradigm using HEK cells stably express-
ing rat a7 along with human Ric3. In these experiments, both the
a
b
c
Homochiral.
Rat, sc, 30 min.
Rat, ip, 30 min.
Racemic.
peak intensity (peak) and area under the curve (area) were
recorded, giving values for both potency and efficacy (Ymax; %
response with respect to acetylcholine). It has been reported that
area under the curve (also referred to as total charge) is a more
accurate determination of the true electrophysiological
response.13,15 Data for selected compounds is shown in Table 4.
As can be seen, the dimethylaminopropyl (1), piperidine (19) and
morpholine (21) compounds were partial agonists while the
quinuclidines (22, 34–36) were either full agonists or high-intrin-
sic activity partial agonists as determined by area under the curve
measures.
d
e
Not determined.
O
F
O
O
O
O
O
F
O
O
a
b,c
N
N
N
O
O
O
Unfortunately, the inclusion of polar functionality in the more
potent compounds led to a loss of permeability, as shown by the
low PAMPA values for the sulfonamide (35) and acetamide (36)
analogs (Table 5). Although good permeability was observed for
quinuclidine 22, it was found to have very poor brain penetration
(B/P <0.02). Follow-up experiments showed that all the quinucli-
dines tested had very high efflux ratios in a bidirectional Caco-2
assay, indicating that they may be substrates for transporter-med-
iated efflux. It has been previously demonstrated that PGP
mediated efflux is often exacerbated by the presence of strongly
basic amines,16,17 and we hypothesized that this was the case here,
since the quinuclidine–quinolones described herein are signifi-
O
37
38
39
O
O
F
N
d
e
N
F
40
41
cantly more basic than other literature quinuclidine-based
a7
O
nAChR ligands. This was attributed to the lack of a heteroatom sub-
stituent on C-3 of the quinuclidine ring for compounds such as 22
(see Fig. 1 for comparison), as a heteroatom substituent on the
quinulidine would be expected to inductively reduce the pKa of
the amine In order to test whether efflux was driven by the pres-
ence of the strongly basic quinuclidine, we attempted to survey
the basicity, efflux and brain penetration relationship of less basic
g-i
f
N
N
F
O
F
N
42
43
analogs. Toward this end,
designed, with the goal of tempering the basicity of the quinucli-
dine nitrogen (Scheme 4).17 Thus,
-fluorination of Boc-isonipeco-
tate 37 followed by deprotection and coupling with ethyl
bromoacetate led to piperidine 39, which then underwent a
a novel 4-fluoroquinuclidine was
Scheme 4. Reagents and conditions: (a) LHMDS, NFSI, THF (32%); (b) TFA/CHCl3; (c)
ethyl bromoacetate, Cs2CO3, THF (62%, 2 steps); (d) KOtBu, then HCl (42%); (e)
dimethyl (1-diazo-2-oxopropyl)phosphonate, K2CO3, MeOH; (f) HClO4, CHCl3 (g)
aniline, NaHB(OAc)3, DCE, reflux (40%, 3 steps); (h) phenylpropiolic acid, SOCl2,
benzene (50%); (i) TfOH (neat), rt, 30 min (62%).
a

1688
I. M. McDonald et al. / Bioorg. Med. Chem. Lett. 23 (2013) 1684–1688
6. For AZD0328, see: Sydserff, S.; Sutton, E. J.; Song, D.; Quirk, M. C.; Maciag, C.; Li,
C.; Jonak, G.; Gurley, D.; Gordon, J. C.; Christian, E. P.; Doherty, J. J.; Hudzik, T.;
Johnson, E.; Mrzljak, L.; Piser, T.; Smagin, G. N.; Wang, Y.; Widzowski, D.; Smith,
J. S. Biochem. Pharmacol. 2009, 78, 880.
7. For GTS-21, see: Kitagawa, H.; Takenouchi, T.; Azuma, R.; Wesnes, K. A.;
Kramer, W. G.; Clody, D. E.; Burnett, A. L. Neuropsychopharmacology 2003, 542.
8. Rudolf, R.; Mongillo, M.; Rizzuto, R.; Pozzan, T. Nat. Rev. Mol. Cell Biol. 2003, 4,
579.
Dieckmann cyclization to give 4-fluoroquinuclidinone 40. Two-
step homologation to the aldehyde 42 was followed with conver-
sion to quinolone 43 according to the prior method (see Scheme 2).
The pKa of fluoroquinuclidine 43 was in fact substantially al-
tered by the introduction of the fluorine, going from 10.1 for quinu-
clidine 22 to 7.6 for fluoroquinuclidine 43 (Table 5). Unfortunately,
along with this drop in basicity, a very substantial loss in potency
9. González, J. E.; Maher, M. P. Receptors Channels 2002, 8, 283.
10. For a recent example of a series of
a7 nAChR agonists with longer tethers
was observed in fluoroquinuclidine 43
(a7 nAChR FLIPR
between the amine and carbonyl moieties, see: Zanaletti, R.; Bettinetti, L.;
Castaldo, C.; Cocconcelli, G.; Comrey, T.; Dunlop, J.; Gaviraghi, G.; Ghiron, C.;
Haydar, S. N.; Jow, F.; Maccari, L.; Micco, I.; Nencini, A.; Scali, C.; Turlizzi, E.;
Valacchi, M. J. Med. Chem. 2012, 4806.
EC50 = 5.8 M). The efflux that was seen in the more basic com-
l
pounds was, however, no longer an issue, as evidenced by a
Caco-2 efflux ratio near unity. This data, along with data for the
morpholine compound 21, showed that, at least for the limited
number of compounds examined, high basicity contributed to poor
brain-penetration, likely by increased transporter-mediated efflux.
In summary, optimization of a lead quinolone led to a series of
11. Turk, C. F.; Krapcho, J. U.S. Patent 4,232,027, 1980.
12. Koltunov, K. Y.; Walspurger, S.; Sommer, J. Eur. J. Org. Chem. 2004, 19, 4039.
13. Although FLIPR provides
a reading of intrinsic activity (Ymax) along with
potency, it has been shown that the intrinsic activity found in a FLIPR assay
may not accurately reflect that which is seen in electrophysiological (EP)
measurements. For this reason, FLIPR Ymax data is not reported. For more
background on the relationship between FLIPR and EP data, see: Dunlop, J.;
Roncarti, R.; Jow, B.; Bothman, H.; Lock, T.; Kowal, D.; Bowlby, M.; Terstappen,
G. C. Biochem. Pharmacol. 2007, 74, 1172.
potent quinuclidine/quinolone
a7 nAChR agonists. Evidence sup-
ports the high pKa of these compounds as reason for their lack of
brain penetration. Less basic compounds in this series were found
to lack transporter-mediated efflux and for morpholine 21, this led
to significantly improved brain penetration (Table 5).
14. Stotter, P. L.; Friedman, M. D.; Dorsey, G. O.; Shiely, R. W.; Williams, R. F.;
Minter, D. E. Heterocycles 1987, 25, 251.
15. Papke, R. L.; Porter-Papke, J. K. Br. J. Pharmacol. 2002, 137, 49.
16. Ecker, G.; Huber, M.; Schmid, D.; Chiba, P. Mol. Pharmacol. 1999, 791.
17. For a similar approach to mitigating PGP efflux by tempering amine basicity via
the installation of proximal fluorine atoms, see: Cox, C. D.; Breslin, M. J.;
Whitman, D. B.; Coleman, P. J.; Garbaccio, R. M.; Fraley, M. E.; Zrada, M. M.;
Buser, C. A.; Walsh, E. S.; Hamilton, K.; Lobell, R. B.; Tao, W.; Abrams, M. T.;
South, V. J.; Huber, H. E.; Kohl, N. E.; Hartman, G. D. Bioorg. Med. Chem. Lett.
2007, 2697.
References and notes
1. Hajos, M.; Rogers, B. N. Curr. Pharm. Des. 2010, 16, 538.
2. Wallace, T. L.; Porter, R. H. P. Biochem. Pharmacol. 2011, 82, 891.
3. Mazurov, A. A.; Speake, J. D.; Yohannes, D. J. Med. Chem. 2011, 54, 7943.
4. Marrero, M. B.; Bencherif, M.; Lippiello, P. M.; Lucas, R. Pharm. Res. 2011, 413.
5. For EVP-6124, see: Prickaerts, J.; van Goethem, N. P.; Chesworth, R.; Shapiro, G.;
Boess, F. G.; Methfessel, C.; Reneerkens, O. A. H.; Flood, D. G.; Hilt, D.; Gawryl,
M.; Bertrand, S.; Bertrand, D.; König, D. Neuropharmacology 2012, 1099.