Quinolinone-based agonists of S1P1: Use of a N-scan SAR strategy to optimize in vitro and in vivo activity

Article abstract of DOI:10.1016/j.bmcl.2011.10.085

We reveal how a N-scan SAR strategy (systematic substitution of each CH group with a N atom) was employed for quinolinone-based S1P₁ agonist 5 to modulate physicochemical properties and optimize in vitro and in vivo activity. The diaza-analog 1

Full text of DOI:10.1016/j.bmcl.2011.10.085

Bioorganic & Medicinal Chemistry Letters 22 (2012) 527–531
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Quinolinone-based agonists of S1P1: Use of a N-scan SAR strategy
to optimize in vitro and in vivo activity
Lewis D. Pennington a, , Michael D. Croghan a, Kelvin K.C. Sham a, Alexander J. Pickrell a,
⇑
Paul E. Harrington a, Michael J. Frohn a, Brian A. Lanman a, Anthony B. Reed a, Matthew R. Lee b, Han Xu c,
Michele McElvain c, Yang Xu d, Xuxia Zhang e, Michael Fiorino e, Michelle Horner f, Henry G. Morrison g,
Heather A. Arnett e, Christopher Fotsch a, Andrew S. Tasker a, Min Wong e, Victor J. Cee a
a Medicinal Chemistry, Amgen, One Amgen Center Drive, Thousand Oaks, CA 91320, USA
b Molecular Structure, Amgen, One Amgen Center Drive, Thousand Oaks, CA 91320, USA
c HTS and Molecular Pharmacology, Amgen, One Amgen Center Drive, Thousand Oaks, CA 91320, USA
d Pharmacokinetics and Drug Metabolism, Amgen, One Amgen Center Drive, Thousand Oaks, CA 91320, USA
e Inflammation Research, Amgen, One Amgen Center Drive, Thousand Oaks, CA 91320, USA
f Toxicology, Amgen, One Amgen Center Drive, Thousand Oaks, CA 91320, USA
g Pharmaceutics, Amgen, One Amgen Center Drive, Thousand Oaks, CA 91320, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
We reveal how a N-scan SAR strategy (systematic substitution of each CH group with a N atom) was
Received 4 October 2011
Revised 24 October 2011
Accepted 26 October 2011
Available online 4 November 2011
employed for quinolinone-based S1P1 agonist 5 to modulate physicochemical properties and optimize
in vitro and in vivo activity. The diaza-analog 17 displays improved potency (hS1P1 RI; 17: EC50 = 0.020
120% efficacy; 5: EC50 = 0.070 M; 5:
M, 110% efficacy) and selectivity (hS1P3 Ca2+ flux; 17: EC50 >25
EC50 = 1.5 M, 92% efficacy), as well as enhanced pharmacokinetics (17: CL = 0.15 L/h/kg, Vdss = 5.1 L/kg,
lM,
l
l
l
T1/2 = 24 h, %F = 110; 5: CL = 0.93 L/h/kg, Vdss = 11 L/kg, T1/2 = 15 h, %F = 60) and pharmacodynamics (17:
1.0 mg/kg po, 24 h PLC POC = ꢀ67%; 5: 3 mg/kg po, 24 h PLC POC = ꢀ51%) in rat.
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
Sphingosine-1-phosphate-1 receptor agonist
Physicochemical properties
Immunosuppression
Multiple sclerosis
N-scan SAR
Sphingosine-1-phosphate (1, S1P) is the endogenous ligand for
the G-protein-coupled receptors S1P1–5 and participates in modu-
lating a wide variety of physiological processes including cell
motility, pulmonary endothelium integrity, and cardiovascular
function ( Fig. 1).1 S1P (1) and the sphingosine-1-phosphate-1
receptor (S1P1) were discovered several years ago to regulate the
trafficking of lymphocytes from the lymphatic system.2 Exposure
to high concentrations of 1 or synthetic agonists of S1P1 induces
receptor internalization (RI) of S1P1 located on the cell surface of
lymphocytes in lymphatic tissues and prevents lymphocyte egress
to the systemic circulation and central nervous system (1: hS1P1 RI
of fingolimod (2b) is believed to arise primarily from the agonism
of S1P1 by 2a (hS1P1 RI EC50 = 0.0020 l
M, 115% efficacy)3 and the
ensuing peripheral lymphocyte depletion.8
Notably, fingolimod (2b) has also been shown to induce dose-
dependent transient bradycardia in preclinical animal models
and human subjects.9 Until recently, the activation of S1P3 by 2a
NH2
NH2
OH
OR
2−
CH3(CH2)12
OPO3
EC50 = 0.018 l
M, 91% efficacy).3,4 Synthetic agonists of S1P1 are
OH
CH3(CH2)7
2−
being investigated intensely as potential immunosuppressive ther-
apies for a variety of conditions, including multiple sclerosis (MS).5
FTY720-P (2a) is the first nonnatural agonist of S1P1 to be re-
ported and the bioactive metabolite of fingolimod (2b, FTY720),6
the first oral therapy for relapsing forms of MS (Fig. 1).7 The efficacy
1 (S1P)
2a
(FTY720-P): R = PO3
2b (FTY720): R = H
O N
N
O
O
CF3
NH
NH
S
O
N
F3C
F3C
3 (SEW2871)
4
⇑
Corresponding author. Tel.: +1 805 447 0048; fax: +1 805 480 1337.
E-mail addresses: lewpenn@me.com, lewisp@amgen.com (L.D. Pennington).
Figure 1. Natural and nonnatural agonists of S1P1.
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.10.085

528
L. D. Pennington et al. / Bioorg. Med. Chem. Lett. 22 (2012) 527–531
(hS1P3 Ca2+ flux EC50 = 0.027
l
M, 35% efficacy)10 was believed to
deleterious effect on potency toward S1P1 (hS1P1 RI EC50 = 4.1,
be the sole cause of these cardiovascular side effects in humans,11
and the search for S1P3-sparing synthetic agonists of S1P1 has gar-
nered significant attention.12 SEW2871 (3) is the first example of
an S1P1-selective agonist lacking a polar head group that is typical
of most synthetic agonists of this receptor (Fig. 1).13 Although the
0.19, 1.1, 2.9, 0.31, and 0.45 lM; and 86%, 100%, 59%, 170%, 110%,
and 110% efficacy, respectively)3 compared to 5. In contrast, the
data for aza-analogs 6, 8, 9, 11, and 14 showed that N atom substi-
tution at positions 2, 6, 7, 30, and 200, respectively, has either a ben-
eficial or neutral effect on potency toward S1P1 (hS1P1 RI
potency of 3 is modest (hS1P1 RI EC50 = 0.27
selectivity for S1P1 versus S1P3 is excellent (hS1P3 Ca2+ flux EC50
>25
M).10,14
l
M, 86% efficacy),3 its
EC50 = 0.041, 0.024, 0.029, 0.042, and 0.060 lM; and 130%, 100%,
120%, 120%, and 110% efficacy, respectively)3 and, with the excep-
tion of the 30 position (aza-analog 11), imparts improved cell
permeability (Papp = 2.3, 5.7, 3.4, <1.0, and 11 ꢁ 10–6 cm/s, respec-
tively) versus 5. In general, the effects of a single N atom substitu-
tion on solubility and microsomal stability were found to be
modest; however, incorporation of a N atom at positions 200, 300, or
400 (compounds 14, 15, and 16, respectively) was observed to im-
l
Our laboratory recently disclosed carbamoylnicotinamide 4 as a
novel example of this atypical chemotype of S1P1 agonist lacking a
polar head group (Fig. 1).15 Carbamoylnicotinamide 4 displays im-
proved potency (hS1P1 RI EC50 = 0.035
pared to SEW2871 (3), while maintaining excellent selectivity
against S1P3 (hS1P3 Ca2+ flux EC50 >25 M).10 The greater cell po-
l
M, 96% efficacy)3 com-
l
prove solubility in 0.010 M aq HCl (Sol = 12, 23, and 6.4
respectively) compared to 5, and aza-analogs 15 and 16 were the
only derivatives found to exhibit >14 L/min/mg clearance in rat
liver microsomes (CLint = 18 and 44 L/min/mg, respectively).
lg/mL,
tency of 4 versus 3 is recapitulated in vivo. Whereas carbamoylnic-
otinamide 4 effects dramatic reduction of circulating lymphocytes
in rat 24 h after a 1 mg/kg oral dose (POC = ꢀ78%; [4]plasma = 63 ng/
mL),15,16 SEW2871 (3) requires a much larger dose (30 mg/kg po)
to achieve similar levels of peripheral lymphocyte depletion at
the 24 h time point (POC = ꢀ66%; [3]plasma = 1100 ng/mL).16 As part
of an effort to discover new and potentially improved atypical
chemotypes of S1P1 agonists lacking a polar head group, we set
out to explore conformationally constrained analogs of 4.
Because a large number of heterocyclic bioisosteres of the acyl
urea moiety of carbamoylnicotinamide 4 have been reported in
S1P1 agonists (e.g., the oxadiazole of 3),5,13 we focused our atten-
tion on the right-hand side of 4 (Fig. 2). We envisioned replacing
the intramolecular hydrogen bond between the carbamoyl NH
group and the nicotinamide OMe substituent of 4 with a covalent
linkage, and substituting the pyridyl N atom of 4 with a CH group
to give quinolinone 5. This cyclization strategy would simulta-
neously serve to impart a more rigid conformational constraint
and remove a hydrogen bond donor.17,18 Furthermore, we desired
to systematically substitute each CH group in 5 with a N atom to
probe the effect on potency and tune molecular and physicochem-
ical properties with minimal impact on molecular weight. This
homologous series of aza-substituted, quinolinone-based S1P1
agonists was anticipated to reveal structure–activity relationships
(SAR) that might lead to further improvement of in vitro and
in vivo activity by the potentially additive effects of multiple N
atom substitution. We expected that the synthesis of these
aza- and poly-aza-analogs of 5 would challenge this N atom scan
(N-scan) SAR strategy.
l
l
We desired next to examine the effect of quinolinone 5 and its
best aza-analogs 6, 8, 9, 11, and 14 on peripheral lymphocyte counts
(PLC) in vivo (1 mg/kg po).16 Notably, aza-analog 8 is the only com-
pound that effects statistically significant (P <0.05 versus vehicle by
the ANOVA/Dunnett’s Multiple Comparison Test) reduction in cir-
culating lymphocytes (POC = ꢀ64%; [8]plasma = 350 ng/mL). Given
their comparable in vitro profiles, it is unclear why derivatives 5,
6, 9, 11, and 14 do not achieve higher plasma levels and realize sta-
tistically significant reduction of peripheral lymphocyte counts. Be-
cause N atom substitution at positions 30 and 200 was shown to have
either a beneficial or neutral effect on potency toward S1P1 (com-
pounds 11 and 14, respectively), diaza-analogs 17 and 18 and tri-
aza-analog 19 were synthesized. The data for diaza-analog 17
showed that N atom substitution at both positions 6 and 30 has a
neutral effect on potency (hS1P1 RI EC50 = 0.020 l
M, 120% efficacy)3
and a positive effect on cell permeability (Papp = 6.7 ꢁ 10–6 cm/s)
compared to aza-analogs 8 and 11. In contrast, the data for diaza-
analog 18 and triaza-analog 19 revealed that having N atom substi-
tution at both the 6 and 20 positions has a negative effect on potency
(hS1P1 RI EC50 = 0.20, 0.21 lM and 100%, 130% efficacy, respec-
tively)3 but a positive effect on cell permeability (Papp = 16 and
15 ꢁ 10–6 cm/s, respectively) and solubility (Sol = 120, <1.0, 6.7
and >200, 3.0, 5.8 lg/mL, respectively) versus aza-analogs 8 and
14 (for 18) and diaza-analog 17 (for 19). Of these three poly-aza-
substituted compounds, only diaza-analogs 17 and 18 effect statis-
tically significant (P <0.05 versus vehicle by the ANOVA/Dunnett’s
Multiple Comparison Test) reduction in circulating lymphocytes
(POC = ꢀ64% and ꢀ32%, respectively; [17]plasma = 130 ng/mL;
[18]plasma = 150 ng/mL).16 The data presented in Table 1 show that
embedded N atoms are tolerated at specific positions of 5, and
the resulting aza-analogs generally have improved physicochemi-
cal properties that can result in markedly enhanced in vivo
pharmacodynamics.19
The results of this study are detailed in Table 1. Pleasingly, quin-
olinone 5 was found to exhibit similar potency toward S1P1 (hS1P1
RI EC50 = 0.070 l
M, 110% efficacy)3 compared to carbamoylnicoti-
namide 4. It is also notable that quinolinone 5 is the only compound
in Table 1 that displays appreciable agonism of S1P3 (hS1P3 Ca2+
flux; 5: EC50 = 1.5 lM, 92% efficacy; 6–19: EC50 >2.5 or 25 l
M).10
Other salient features of this investigation are as follows. The data
for aza-analogs 7, 10, 12, 13, 15, and 16 revealed that N atom sub-
stitution at positions 5, 8, 50, 60, 300, and 400, respectively, has a
Pharmacokinetic profiles of compounds 5, 8, and 17 were under-
taken in preparation for more detailed studies in vivo (Table 2). In
N-scan SAR strategy:
ring formation
N
O
O
O
NH
O
NH
NH
O
N
4
5
F3C
F3C
substitute N
CH
systematically substitute CH
N
Figure 2. Transformation of 4 ? 5 and design of the N-scan SAR strategy.

L. D. Pennington et al. / Bioorg. Med. Chem. Lett. 22 (2012) 527–531
529
Table 1
N-scan SAR study of 5: effects on physicochemical properties, potency, microsomal stability, and lymphocyte count
2
O
N
3'' 2''
6' 5'
8
4''
7
NH
O
3'
5
6
F3C
c
Compd
CH ? N
clogPa
ACDlogPb
(ACDLogD7.4
hS1P1 RI EC50
M (% eff.)
Sol (
HCl, PBS, SIF
l
g/mL)d
Papp (ER)e
CLint, RLMf
L/min/mg)
PLC, POCg
([C]plasma, ng/mL)
)
l
(ꢁ106 cm/s)
(
l
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
–
2
5
6
4.8
4.9
3.5
3.5
3.5
3.5
3.8
3.8
4.0
3.5
3.3
3.3
2.4
2.2
1.2
6.2 (6.2)
5.5 (5.5)
4.7 (4.7)
4.8 (4.8)
4.7 (4.7)
5.3 (5.3)
5.4 (5.4)
5.0 (5.0)
5.8 (5.8)
4.8 (4.8)
4.9 (4.9)
4.8 (4.8)
4.1 (4.0)
3.5 (3.5)
2.7 (2.7)
0.070 (110)
0.041 (130)
4.1 (86)
0.024 (100)
0.029 (120)
0.19 (100)
0.042 (120)
1.1 (59)
1.3, <1.0, 11
<1.0, 1.6, 12
NDi
<1.0, <1.0, 4.9
<1.0, 2.6, 6.1
<1.0, <1.0, 5.9
<1.0, <1.0, 2.3
NDi
<1.0, <1.0, 13
12, <1.0, 5.2
23, <1.0, 6.7
6.4, 7.9, 12
<1.0, 2.4, 10
120, <1.0, 6.7
>200, 3.0, 5.8
1.2 (0.40)
2.3 (1.2)
16 (1.4)
5.7 (0.80)
3.4 (1.0)
2.2 (1.7)
<1.0 (NR)k
<1.0 (NR)k
11 (1.9)
11 (0.80)
18 (1.6)
15 (1.5)
6.7 (1.0)
16 (1.0)
<14
<14
<14
<14
<14
<14
<14
<14
<14
<14
18
–3 (20)h
+6 (9.7)h
NDi
–64 (350)j
–36 (140)h
NDi
7
8
30
+7 (1.3)h
NDi
50
60
2.9 (170)
NDi
200
0.060 (110)
0.31 (110)
0.45 (110)
0.020 (120)
0.20 (100)
0.21 (130)
+25 (5.1)h
NDi
300
400
44
<14
15
NDi
6, 30
6, 200
6, 30, 200
–64 (130)j
–32 (150)j
–32 (140)h
15 (1.0)
<14
a
b
c
d
e
f
Calculated using Daylight Chemical Information Systems software (web site: www.daylight.com).
Calculated using Advanced Chemistry Development (ACD/Labs) software (web site: www.acdlabs.com).
Data represents an average of at least two determinations; see Ref. 15 for experimental details.
Solubility in 0.010 M aqueous hydrogen chloride (HCl), phosphate-buffered saline (PBS) at pH 7.4, or simulated intestinal fluid (SIF).
Apparent permeability (Papp) through porcine proximal tubule cells (LLC-PK1 cell line) and efflux ratio (ER).
Estimated intrinsic clearance (CLint) determined by incubation of test compound with rat liver microsomes (RLM) for 30 min and measurement of % turnover.
Percent-of-control (POC) reduction versus vehicle in peripheral lymphocyte count (PLC) 24 h after a single oral dose (1 mg/kg; vehicle = 20% Captisol, 1% HPMC, 1%
g
Pluronic F68, pH 2.0 with MSA) administered to female Lewis rats (N = 5/group) and total compound concentration in plasma ([C]plasma) at the 24 h time point.
h
The measured POC reduction in PLC is not statistically significant (P >0.05 versus vehicle by the ANOVA/Dunnett’s Multiple Comparison Test).
ND, not determined.
The measured POC reduction in PLC is statistically significant (P <0.05 versus vehicle by the ANOVA/Dunnett’s Multiple Comparison Test).
NR, not reportable.
i
j
k
male Sprague–Dawley rats (N = 3) given a single intravenous dose
(1 mg/kg), diaza-analog 17 displays a more favorable pharmacoki-
netic profile (CL = 0.15 L/h/kg, Vdss = 5.1 L/kg, T1/2 = 24 h) compared
to that for both aza-analog 8 (CL = 0.16 L/h/kg, Vdss = 9.2 L/kg, T1/2
= 41 h) and quinolinone 5 (CL = 0.93 L/h/kg, Vdss = 11 L/kg, T1/2
= 15 h). Additionally, diaza-analog 17 displays enhanced bioavail-
ability (%F = 60, 95, and 110 for 5, 8, and 17, respectively).20
Although it is not perfectly linear, diaza-analog 17 also exhibits bet-
ter dose-proportional exposure at high doses versus 5 and 8 ([AUC
and <0.001 versus vehicle by the ANOVA/Dunnett’s Multiple Com-
parison Test, respectively).21 This dose-dependent decrease in cir-
culating lymphocytes reaching a plateau at 76% maximal reduction
(from the 3.0 mg/kg dose of 17) is consistent with the S1P1 agonist
activity of 5, 8, and 17.13,14
Compounds 5, 8, and 17 were prepared as detailed in Scheme 1,
and their syntheses serve to exemplify the general manner in
which all of the compounds discussed in Table 1 were synthesized.
The preparation of quinolinone 5 was facile. Commercially avail-
able compounds 2-(trifluoromethyl)biphenyl-4-amine (20) and 1-
methyl-4-oxo-1,4-dihydro-3-quinolinecarboxylic acid (21) were
coupled using O-(7-azabenzotriazol-1-yl)-N,N,N0,N0-tetramethyl-
uronium hexafluorophosphate (HATU) and Hünig’s base to give
quinolinone 5 in 62% yield. The synthesis of aza-analog 8 was more
demanding. Methylation of commercially available ethyl 4-oxo-
1,4-dihydro-1,6-naphthyridine-3-carboxylate (22) with Me2SO4/
K2CO3 afforded ethyl 1-methyl-4-oxo-1,4-dihydro-1,6-naphthyri-
dine-3-carboxylate (23). Saponification of 23 (LiOH/H2O) provided
lithium carboxylate 24, which was used in the next step of the syn-
thesis without purification. The union of biphenylamine 20 and 24
was mediated by HATU/i-Pr2NEt to afford aza-analog 8 in 30%
yield. The synthesis of diaza-analog 17 was the most challenging
of these three compounds. Commercially available 6-(trifluoro-
methyl)-2-pyridinamine (26) was regioselectively brominated
using NBS in CHCl3 to give 5-bromo-6-(trifluoromethyl)-2-pyridin-
amine (27) in moderate yield. A phenyl group was appended to 27
by palladium-mediated cross-coupling ((Ph3P)4Pd/K2CO3) with
phenylboronic acid, affording 5-phenyl-6-(trifluoromethyl)-2-
pyridinamine (28) in 84% yield. The lower nucleophilicity of pyri-
dinamine 28 versus biphenylamine 20 demanded recourse to more
forcing coupling conditions than those used for the synthesis of 5
(100 mg/kg po)]/[AUC0?24h (3 mg/kg po)] = 4.7, 6.0, and 11
0?24h
for 5, 8, and 17, respectively). The comparable plasma protein bind-
ing of 5, 8, and 17 (%PPB = 97, 92, and 98, respectively) suggests that
the improved pharmacokinetic profile of 17 is not merely the result
of greater nonspecific binding in this compartment. Because 17 dis-
plays better cell permeability (Papp = 6.7 ꢁ 10–6 cm/s) than carba-
moylnicotinamide 4 (Papp = 2.3 ꢁ 10–6 cm/s),15 and comparable
pharmacokinetics,15 the diminished solubility of 17 (Sol = <1.0,
2.4, 10
6.6, 16
l
l
g/mL in HCl, PBS, and SIF, respectively) versus 4 (Sol = 17,
g/mL in HCl, PBS, and SIF, respectively)15 may be responsi-
ble for its nonlinear exposure at high doses (P100 mg/kg po), a
shortcoming that 4 does not suffer.15
We next sought to evaluate the pharmacodynamic effects of 5,
8, and 17 in more detail (Fig. 3).16 Following a single oral dose in
female Lewis rats (0.30, 1.0, or 3.0 mg/kg), compounds 5, 8, and
17 exhibit dose-proportional plasma exposure ([5]plasma = 5.5, 18,
and 48 ng/mL; [8]plasma = 96, 380, and 1100 ng/mL; and [17]plas-
ma = 47, 180, and 640 ng/mL, respectively), and concomitant
dose-dependent reduction in circulating lymphocytes at the 24 h
time point. Statistically significant lymphocyte depletion is first
realized for the 3.0 mg/kg dose for 5 and 8 and the 1.0 mg/kg dose
for 17 (5: –51%, 8: –64%, and 17: –67%, respectively; P <0.05, <0.01,

530
L. D. Pennington et al. / Bioorg. Med. Chem. Lett. 22 (2012) 527–531
Table 2
Pharmacokinetic profiles of 5, 8, and 17a
Compd
PPB (%)b rat
CL (L/h/kg)c
1 mg/kg iv
Vdss (L/kg)c
1 mg/kg iv
T1/2 (h)c
1 mg/kg iv
AUC0?24h (ng h/mL)d
3 mg/kg po
Fe (%)
AUC0?24h (ng h/mL)f
100 mg/kg po
5
8
17
97
92
98
0.93
0.16
0.15
11
9.2
5.1
15
41
24
1200
5100
12000
60
95
110
5800
31000
130000
a
b
c
In vivo experiments were conducted using male Sprague–Dawley rats (N = 3/group).
Percent rat plasma protein binding (PPB) measured in vitro following separation by ultracentrifugation.
Clearance (CL), volume of distribution (Vdss), and half-life (T1/2) determined following a single intravenous dose (1 mg/kg; vehicle = DMSO).
d
Area under the plasma concentration–time curve (AUC0?24 h) determined following a single oral dose (3 mg/kg; vehicle = 20% Captisol, 1% HPMC, 1% pluronic F68, pH 2.0,
with MSA).
e
Percent bioavailability (F) calculated using AUC0?1 values determined from the 1 mg/kg iv and 3 mg/kg po doses.
Area under the plasma concentration–time curve (AUC0?24 h) determined following a single oral dose (100 mg/kg; vehicle = 20% Captisol, 1% HPMC, 1% pluronic F68, pH
f
2.0, with MSA).
and 8. Lithium carboxylate 24 was treated with SOCl2 in toluene at
reflux to provide acid chloride 25 (quantitative yield, crude). Final-
ly, exposure of a solution of 25 and 28 in dichloroethane to DMAP/
i-Pr2NEt afforded diaza-analog 17 in 63% yield. Comparison of the
synthesis of compounds 5, 8, and 17 illustrates that incorporating
an increasing number of specifically placed N atoms into a scaffold
often results in progressively more lengthy and complex synthetic
sequences.
In conclusion, a set of conformationally constrained, atypical
chemotypes of S1P1 agonists lacking a polar head group was
discovered. This homologous series of aza-substituted, quinoli-
none-based agonists of S1P1 provided insight into SAR trends that
allowed improvement of both in vitro and in vivo activity. This
study illustrates that the conceptually simple N-scan SAR strategy
of systematically incorporating N atoms into a scaffold to tune
molecular and physicochemical properties can lead to significant
improvement in potency, selectivity, pharmacokinetics, and phar-
macodynamics (e.g., quinolinone 5 ? aza-analog 8 ? diaza-analog
17). However, this approach can be rendered challenging by the
progressively more complex synthesis of these aza-substituted
compounds. Finally, although diaza-analog 17 displays improved
pharmacokinetics and pharmacodynamics compared to quinoli-
none 5 and aza-analog 8, it exhibits nonlinear exposure at high
doses required to establish a safety margin (P100 mg/kg po),
limiting the prospects for further development of this compound.
NH2
N
N
HATU
O
F3C
O
20
i-Pr2NEt
NH
5
+
DMF
50 °C
(62%)
O
O
N
F3C
HO
21
R
LiOH⋅H2O
O
X
O
N
4:1:1THF-MeOH-H2O
23 °C
EtO
O
24
N
O
N
(quant., crude)
Me2SO4, K2CO3
DMF, 80 °C
(46%)
SOCl2
22: R = H
23: R = Me
: X = OLi
25: X = Cl
toluene, 100 °C
(quant., crude)
O
N
HATU, i-Pr2NEt
20 + 24
NH
DMF
O
N
23 °C
8
F3C
(30%)
PhB(OH)2
(Ph3P)4Pd, K2CO3
X
NH2
NH2
N
N
1,4-dioxane
100 °C
F3C
F3C
(84%)
NBS
26: X = H
27: X = Br
28
CHCl3, 23 °C
10
9
8
7
6
5
4
3
2
1
0
1200
1000
800
600
400
200
0
(46%)
O
N
DMAP, i-Pr2NEt
28 + 25
NH
ClCH2CH2Cl
83 °C
N
O
N
17
F3C
(63%)
Scheme 1. Chemical synthesis of compounds 5, 8, and 17.
*
**
8
***
17
Efforts to modify the aza-quinolinone scaffold to overcome this
challenge are the subject of a future disclosure.
***
17
Acknowledgments
5
8
17
5
8
17
5
8
5
Vehicle
0.3 mg/kg 1.0 mg/kg 3.0 mg/kg
We thank Valerie Almon, Ronya Primack, and Elizabeth
Tominey (Pharmacokinetics and Drug Metabolism, Amgen) for
in vivo pharmacokinetic studies. All live animal studies were con-
ducted in an AAALAC-accredited facility and husbandry procedures
met all of the recommendations of the Guide for the Care and Use
of Laboratory Animals. All work using research animals was con-
ducted under Institutional Animal Care and Use Committee (IA-
CUC) approved protocols.
Figure 3. Female Lewis rats (N = 5/group; vehicle = 20% Captisol or HPBCD, 1%
HPMC, 1% Pluronic F68, pH 2.0 with MSA) administered a single oral dose of 5, 8, or
17 (0.30, 1.0, or 3.0 mg/kg) showed dose-proportional plasma exposure (black
circles represent average plasma concentration SE) and dose-dependent reduction
in circulating lymphocytes (gray bars represent average blood lymphocyte
counts SE) at the 24 h time point (statistical significance: ⁄P <0.05, ⁄⁄P <0.01, or
⁄⁄⁄P <0.001 versus vehicle by the ANOVA/Dunnett’s Multiple Comparison Test); see
Ref. 15 for experimental details.

L. D. Pennington et al. / Bioorg. Med. Chem. Lett. 22 (2012) 527–531
531
Itano, A.; McElvain, M.; Middleton, S.; Schrag, M.; Vargas, H. M.; Xu, H.; Xu, Y.;
Zhang, X.; Siu, J.; Bürli, R. W. ACS Med. Chem. Lett. 2011, 2, 102. and references
cited therein.
References and notes
1. For a review, see: Rosen, H.; Gonzalez-Cabrera, P. J.; Sanna, M. G.; Brown, S.
Annu. Rev. Biochem. 2009, 78, 743.
2. For a review, see: Cyster, J. G. Annu. Rev. Immunol. 2005, 23, 127.
3. By measurement of RI of an hS1P1–eGFP fusion protein in U2OS cells (% efficacy
is reported relative to known hS1P1 agonist 1-(4-(6-benzylbenzofuran-2-yl)-3-
13. Hale, J. J.; Lynch, C. L.; Neway, W.; Mills, S. G.; Hajdu, R.; Keohane, C. A.;
Rosenbach, M. J.; Milligan, J. A.; Shei, G.-J.; Parent, S. A.; Chrebet, G.; Bergstrom,
J.; Card, D.; Ferrer, M.; Hodder, P.; Strulovici, B.; Rosen, H.; Mandala, S. J. Med.
Chem. 2004, 47, 6662.
14. Sanna, M. G.; Liao, J.; Jo, E.; Alfonso, C.; Ahn, M.-Y.; Peterson, M. S.; Webb, B.;
Lefebvre, S.; Chun, J.; Gray, N.; Rosen, H. J. Biol. Chem. 2004, 279, 13839. S1P1
agonists have been shown to effect ꢂ70% maximal reduction in circulating
lymphocytes; see Ref. 13.
15. Pennington, L. D.; Sham, K. K. C.; Pickrell, A. J.; Harrington, P. E.; Frohn, M. J.;
Lanman, B. A.; Reed, A. B.; Croghan, M. D.; Lee, M. R.; Xu, H.; McElvain, M.; Xu,
Y.; Zhang, X.; Fiorino, M.; Horner, M.; Morrison, H. G.; Arnett, H. A.; Fotsch, C.;
Wong, M.; Cee, V. J. ACS Med. Chem. Lett. 2011, 2, 752.
16. Female Lewis rats (N = 5/group) were administered a single oral dose of each
compound; after 24 h, lymphocyte counts in blood and compound
concentrations in plasma were measured (the 24 h time point was selected
for these studies, as it was found to mitigate the effects of dosing on
lymphocyte counts; see Ref. 15 for experimental details).
17. For a review, see: Hitchcock, S. A.; Pennington, L. D. J. Med. Chem. 2006, 49,
7559.
18. For a review, see: Ursu, O.; Rayan, A.; Goldblum, A.; Oprea, T. I. WIREs Comp.
Mol. Sci. 2011, 1, 760.
19. Comparison of the ACDLogP and ACDLogD7.4 values for each compound in
Table 1 indicates that all are calculated to be neutral at physiological pH, and
analysis of the ACDlogP and clogP values reveals that each CH ? N
substitution typically effects ꢂ1 unit reduction in calculated logP.
20. Calculated using the AUC0!1 values determined from the 1 mg/kg iv and 3 mg/
kg po doses.
fluorobenzyl)azetidine-3-carboxylic acid at a concentration of 1.0 lM); see:
Cee, V. J.; Frohn, M.; Lanman, B. A.; Golden, J.; Muller, K.; Neira, S.; Pickrell, A.;
Arnett, H.; Buys, J.; Gore, A.; Fiorino, M.; Horner, M.; Itano, A.; Lee, M. R.;
McElvain, M.; Middleton, S.; Schrag, M.; Rivenzon-Segal, D.; Vargas, H. M.; Xu,
H.; Xu, Y.; Zhang, X.; Siu, J.; Wong, M.; Bürli, R. W. ACS Med. Chem. Lett. 2011, 2,
107. and Ref. 15 for experimental details.
4. For a review, see: Rivera, J.; Proia, R. L.; Olivera, A. Nat. Rev. Immunol. 2008, 8,
753.
5. For a review, see: Bolli, M. H.; Lescop, C.; Nayler, O. Curr. Top. Med. Chem. 2011,
11, 726.
6. Albert, R.; Hinterding, K.; Brinkmann, V.; Guerini, D.; Müller-Hartwieg, C.;
Knecht, H.; Simeon, C.; Streiff, M.; Wagner, T.; Welzenbach, K.; Zécri, F.;
Zollinger, M.; Cooke, N.; Francotte, E. J. Med. Chem. 2005, 48, 5373.
7. Kappos, L.; Radue, E.-W.; O’Connor, P.; Polman, C.; Hohlfeld, R.; Calabresi, P.;
Selmaj, K.; Agoropoulou, C.; Leyk, M.; Zhang-Auberson, L.; Burtin, P. N. Engl. J.
Med. 2010, 362, 387.
8. Mandala, S.; Hajdu, R.; Bergstrom, J.; Quackenbush, E.; Xie, J.; Milligan, J.;
Thornton, R.; Shei, G.-J.; Card, D.; Keohane, C.; Rosenbach, M.; Hale, J.; Lynch, C.
L.; Rupprecht, K.; Parsons, W.; Rosen, H. Science 2002, 296, 346.
9. Schmouder, R.; Serra, D.; Wang, Y.; Kovarik, J. M.; DiMarco, J.; Hunt, T. L.;
Bastien, M.-C. J. Clin. Pharmacol. 2006, 46, 895. and references cited therein.
10. By measurement of Ca2+ mobilization in CHO-K1 cells ex-pressing hS1P3 and a
chimeric Gq/i5 G-protein (% efficacy is reported relative to S1P at
a
21. For aza-analog 8, the higher dose required to achieve statistically significant
reduction in circulating lymphocytes in the dose–response study (3.0 mg/kg
po; Fig. 3) versus the single dose experiment (1 mg/kg po; Table 1) may arise
from the lower peripheral lymphocyte counts and greater variability in the
concentration of 0.20 M; >[highest concentration tested] is reported for
l
compounds that do not achieve >10% of control activity); see Ref. 15 for
experimental details.
11. Hamada, M.; Nakamura, M.; Kiuchi, M.; Marukawa, K.; Tomatsu, A.; Shimano,
K.; Sato, N.; Sugahara, K.; Asayama, M.; Takagi, K.; Adachi, K. J. Med. Chem.
2010, 53, 3154.
12. Lanman, B. A.; Cee, V. J.; Cheruku, S. R.; Frohn, M.; Golden, J.; Lin, J.; Lobera, M.;
Marantz, Y.; Muller, K. M.; Neira, S. C.; Pickrell, A. J.; Rivenzon-Segal, D.; Schutz,
N.; Sharadendu, A.; Yu, X.; Zhang, Z.; Buys, J.; Fiorino, M.; Gore, A.; Horner, M.;
vehicle group for this study (6.1 2.5 ꢁ 103/
l
L
vs 8.4 1.5 ꢁ 103/
lL,
respectively), as the peripheral lymphocyte counts and variability
(4.3 0.84 ꢁ 103/
l
L
vs 3.1 1.2 ꢁ 103/
lL, respectively), as well as total
compound concentration in plasma and variability (380 55 ng/mL vs
350 23 ng/mL) were similar for the 1.0 mg/kg po dose in both studies.