I. Discovery of a novel series of CXCR3 antagonists. Multiparametric optimization of N,N-disubstituted benzylamines

Article abstract of DOI:10.1016/j.bmcl.2016.10.035

N,N-Disubstituted benzylamine derivatives have been identified as CXCR3 antagonists. Compounds were optimized to improve affinity and selectivity, to increase metabolic stability in human and mouse liver microsomes, to increase Caco-2 permeability. Optimization was supported by monitoring physico-chemical properties using both experimental and computational means. Several compounds with double-digit nanomolar CXCR3 affinity, favorable selectivity, microsomal stability, Caco-2 permeability and human hepatocyte clearance have been identified.

Full text of DOI:10.1016/j.bmcl.2016.10.035

Bioorganic & Medicinal Chemistry Letters 26 (2016) 5418–5428
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
I. Discovery of a novel series of CXCR3 antagonists. Multiparametric
optimization of N,N-disubstituted benzylamines
Imre Bata ,y, Zsuzsanna Tömösközi, Péter Buzder-Lantos, Attila Vasas à, Gábor Szeleczky, Sándor Bátori,
⇑
⇑
,§
Veronika Barta-Bodor, László Balázs, György G. Ferenczy
Sanofi Co. Ltd, H-1045 Budapest, Tó u. 1-5, Hungary
a r t i c l e i n f o
a b s t r a c t
Article history:
N,N-Disubstituted benzylamine derivatives have been identified as CXCR3 antagonists. Compounds were
optimized to improve affinity and selectivity, to increase metabolic stability in human and mouse liver
microsomes, to increase Caco-2 permeability. Optimization was supported by monitoring physico-
chemical properties using both experimental and computational means. Several compounds with
double-digit nanomolar CXCR3 affinity, favorable selectivity, microsomal stability, Caco-2 permeability
and human hepatocyte clearance have been identified.
Received 26 July 2016
Revised 8 October 2016
Accepted 12 October 2016
Available online 15 October 2016
Keywords:
Ó 2016 Elsevier Ltd. All rights reserved.
Chemokine receptor
CXCR3 antagonist
Inflammatory disease
Microsomal stability
Caco-2 permeability
CXCR3 is an inflammatory chemokine receptor which is
predominantly expressed on activated immune cells such as
CD4+ T-cells, effector CD8+ T cells and innate-type lymphocytes.1
CXCR3 is selectively activated by three interferon-inducible
chemokines, CXCL9 (also termed as MIG), CXCL10 (IP-10) and
CXCL11 (ITAC). Activation of CXCR3 by these endogenous agonists
elicits intracellular Ca2+ mobilization via phospholipases C (PLC)
and activation of both mitogen-activated protein kinase (MAP-
kinase) and PI3-kinase. These intracellular events finally result in
stimulation of lymphocyte migration and proliferation. CXCR3
plays a key role in selective recruitment of activated immune cells
to the site of inflammation. At the site of inflammation the
recruited Th1 and CTL cells release IFN-gamma that stimulates
epithelial cells and macrophages to further release of CXCR3
agonists that leads to a persistent inflammatory activation. Clinical
evidence has shown significant overexpression of CXCR3 receptor
and/or its endogenous agonists (CXCL10, CXCL11) in multiple
autoimmune or inflammatory diseases.2
Blocking the activation of CXCR3 by antagonists represents a
possible approach for the treatment of diseases such as COPD, pso-
riasis, organ transplant rejection, coeliac disease, inflammatory
bowel disease (IBD), type-1 diabetes and other neuroinflammatory
diseases.3
Our objective was to find an orally active CXCR3 antagonist.
According to our knowledge no small molecule CXCR3 antagonist
is in clinical development.4 AMG-487 (1, Fig. 1) has formerly
reached phase II studies where its development was abandoned
owing to the lack of efficacy in psoriasis.2a,5
The CXCR3 antagonist program was started by a FLIPR-based
high throughput screening6 of 500,000 compounds in Sanofi’s pro-
prietary collection followed by the radioligand binding (IP-10)
test7 as secondary assay. The hit-to-lead process identified com-
pound 2a as a Lead compound (Fig. 1).
Compound 2a exhibited satisfactory activity in FLIPR in human
cell lines (IC50 = 74 nM) and its binding value was also promising
(IC50 = 403 nM vs IC50 of 1 = 121 nM). The S-enantiomer of 2a
showed higher activity (IC50 = 154 nM vs 790 nM of the R-enan-
tiomer). Other parameters like clogP,8 solubility and penetrability
(Caco-2: 56.2 nm/s)9 were also acceptable therefore our efforts
have to be focused on the improvement of activity, the increase
of metabolic stability10 (53% and 52% of remaining parent com-
pound on human and mouse liver microsomes) and the increase
⇑
Corresponding authors.
E-mail addresses: batai@richter.hu (I. Bata), ferenczy.gyorgy@med.semmelweis-
univ.hu (G.G. Ferenczy).
y
Present address: Gedeon Richter Plc., Budapest 10, PO Box 27, H-1475, Hungary.
Present address: Servier Research Institute of Medicinal Chemistry, Záhony u. 7.,
à
H-1031, Budapest, Hungary.
of selectivity (radioligand displacement over 65% at 10 lM for
§
Present address: Department of Biophysics and Radiation Biology, Semmelweis
}
University, 1094 Budapest, Tuzoltó u. 37-47, Hungary.
http://dx.doi.org/10.1016/j.bmcl.2016.10.035
0960-894X/Ó 2016 Elsevier Ltd. All rights reserved.

I. Bata et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5418–5428
5419
O
O
N
N
O
N
O
O
N
N
N
F
N
O
O
O
F
O
F
1 (AMG-487)
2a
Figure 1. Structure of 1 (AMG-487) and Lead compound (2a).
O
O
OH
O
Br
O
O
O
a
N
N
b
+
+
O
NH2
O
O
O
3a
4a
5a
6a
O
O
O
O
O
O
O
c
N
N
H
HCHO
+
N
N
O
O
O
7a
8a
2a
Scheme 1. Synthesis of Lead compound 2a. Reagents and conditions: (a) acetonitrile, K2CO3, reflux; (b) NaBH(OAc)3, AcOH, THF, rt; (c) NaBH(OAc)3, AcOH, THF, rt.
adrenergic alpha 1A, 2A, histamine H1, muscarinic M2 and mu-
opioid receptors).
structures 2g and 2h in Table 1. However, these compounds
exhibited considerably lower binding affinity than does 2a. Thus
affinity improvement by rigidifying the O-ethyl linker could not
be achieved.
In the hit-to-lead process the compounds were synthesized on
solid phase applying a combinatorial chemistry method that
allowed rapid progress. On the other hand, solution phase synthe-
sis was preferred in the lead optimization (Scheme 1) so to facili-
tate scale-up. The synthesis was started with the alkylation of
vanillin (3a) by the N-(bromo-ethyl) succinimide (4a). This reac-
tion resulted in aldehyde 5a that was reacted with amine 6a to
yield the secondary amine 7a. Its reductive amination with
formaldehyde (8a) led to the N,N-disubstituted benzylamine 2a.11
First the optimal linker to the succinimide group was investi-
gated. Various substitutions of the benzylic group were tested
(Table 1). In the synthesis according to Scheme 1 aldehydes (A-
CHO, 5) formed from aromatic aldehydes (3) and N-(bromoalkyl)-
succinimide derivatives (4) were used. Here and later on IP-10
radioligand binding7 was used to measure CXCR3 affinity. (cAMP
functional assay was performed for selected compounds to check
antagonistic effect). Microsomal metabolism10 was also monitored.
An interchange of the relative ortho-substituents at the benzyl
group (2b) drastically reduced the activity. The meta-substitution
(2e and 2f) also led to very weak activity. Similarly, the replace-
ment of the 2 carbon chain connecting the succinimide ring and
O-atom at para-position of the benzyl group with either shortened
(2c) or lengthened (2d) chain resulted in weakly active
compounds.
It was also investigated if the O-ethyl chain linking the succin-
imide moiety and the benzyl group can be made more rigid in
order to achieve higher activity. A conformational analysis12 of this
part of the molecule revealed that the O-ethyl linker can adopt var-
ious low energy conformations. The lowest energy conformation
shown on Figure 2 does not seem to offer the possibility of rigidi-
fying either by anellating a ring to the central phenyl ring or by
introducing double bonds or rings in the chain. On the other hand,
some of the conformers whose energy does not exceed the mini-
mum energy by more than 20 kJ/mol are compatible with rigidified
Structure–activity studies in the hit-to-lead process revealed
that the cyclic 2,5-dioxo-pyrrolidine (i.e., succinimide ring) moiety
at the end of the O-ethyl linker is required for the activity (activity
was completely lost with 2-oxo-pyrrolo group—data not shown).
Replacing a CH2 group of the succinimide ring by NH or N-Me
groups or S-atom (hydantoine, N-methyl-hydantoine and dioxo-
thiazolidine ring, respectively) the activity increased significantly,
but the metabolic stability did not improve or even decreased
(Table 1, 9a–c). The replacement of the O-ethyl linker by NH-ethyl
linker reduced the activity (9d). When the methoxy group on the
central phenyl group was replaced by methyl group or chlorine
atom the activity increase was accompanied by a decrease in
metabolic stability (9e, 9f).
The effect of the basicity of the central N-atom and the substi-
tution of the neighboring C-atom was then investigated (Table 2).
Methyl substitution of the aliphatic C-atom of the benzylic group
or the replacement of the central N-atom by O-atom dramatically
reduced the activity (data not shown).
Other substituents on the benzylic group, namely the 3-meth-
oxy and 4-[2-(2,5-dioxopyrrolidin-1-yl)ethoxy] groups were kept
fixed in these studies. The synthesis was performed according to
Scheme 1 with varying benzofuranylamines 6 (leading to different
R1 and R2) and aldehydes 8 (to introduce R3 see Table 2) while
aldehyde 5a was kept constant.
Secondary benzylamines were found to have improved
metabolic stability but their activity was weaker than that of the
Lead (cf. compounds 7a with IC50 = 607 nM and Lead 2a with
IC50 = 403 nM in Table 2) Double methyl substitution of the C-atom
adjacent to the N-atom further diminished the activity (compound
7b, IC50 = 780 nM).
Converting the secondary benzylamine into tertiary benzy-
lamine is beneficial for the activity. Interestingly, the effect of the

5420
I. Bata et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5418–5428
Table 1
Exploration of the substituents, linker and hetero rings on the right hand side of 2a
O
A
N
Compound
Structure (A)
See Figure 1
hCXCR3a (IC50 nM or inhibition)
121
Met. stab.b (h/m; %)
77/70
AMG-487
O
O
N
N
2a
2b
403
53/52
N.D.
O
O
O
O
18% at 10 lM
O
O
O
O
O
N
2c
8% at 10
l
M
N.D.
N.D.
O
O
N
O
2d
39% at 10
l
M
O
O
O
O
N
2e
4% at 10
lM
N.D.
O
O
O
N
O
2f
15% at 10
36% at 10
l
M
N.D.
N.D.
O
O
O
O
O
N
2g
lM
O
O
O
N
2h
7% at 10
lM
N.D.
N
O
O
O
O
O
O
O
N
N
N
N
N
9a
9b
9c
9d
9e
240
202
120
443
230
60/55
28/26
4/7
NH
N
O
O
O
O
O
O
O
O
O
O
S
H
N
N.D.
O
O
29/19
CH3

I. Bata et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5418–5428
5421
Table 1 (continued)
Compound
Structure (A)
hCXCR3a (IC50 nM or inhibition)
Met. stab.b (h/m; %)
O
O
N
9f
187
14/17
Cl
O
N.D. means not determined.
a
Displacement of 125I-labeled IP-10 from human recombinant CXCR3 receptors.7
Human/mouse liver microsomes.10
b
This bent conformation does not correspond to the lowest
energy for Model 1 (unsubstituted) and Model 2 (alpha-substi-
tuted). The excess energy of a similar bent conformation with
respect to the minimum energy conformations is 4 and 6 kJ/mol
for Model 1 and 2, respectively. This is in line with the observed
order of affinities and also with their relative magnitude. (We note
that ꢀ6 kJ/mol free energy difference corresponds to a 10 fold
difference in the population of states.)
Summarizing the results so far, tertiary benzylamines were
identified whose activity can be modulated by substitutions on
the N-atom and on the neighboring C-atom. The effect of the sub-
stituents can be rationalized by the conformational preference of
the compounds. Double digit nanomolar CXCR3 compounds were
found by appropriately chosen substituents (2i, 2k). Although their
activities were significantly better than AMG-487, these tertiary
amines are metabolically unstable (Table 2).
The effect of the variation of the tertiary amine N-substituent on
the activity and metabolic stability was then further investigated
(Table 4). The introduction of hydroxalkyl-group decreased the
IC50 below 100 nM, but the metabolic stability decreased (10a,
10b). Similar effects were observed with keto group (10d) and with
cyclohexyl group (10e). The basic aminoethyl group somewhat
increased the human metabolic stability but the activity was also
significantly reduced (10c). Neither aromatic nor heteroaromatic
groups gave satisfactory activity and metabolic stability simultane-
ously (10f–i). Introduction of aliphatic (10j) and aromatic acids
(10k) resulted in compounds with somewhat lower than optimal
activity but with appropriate metabolic stability. They have,
Figure 2. Minimum energy conformation of the model compound corresponding to
a fragment of 2a.
substituents on the N-atom (R3) and on the neighboring C-atom (R1
and R2 in Table 2) was found to be interdependent. N-substitution
is beneficial and the affinity increases with increasing the size of
the substituent (cf. 2a and 2i). However, increasing the size of
the C-substituent of the secondary benzylamines (i.e. without
N-substitution, R3 = H) tends to decrease the affinity (cf. 7a–f).
On the other hand, C-substitution of the tertiary amines resulted
in affinities comparable to compounds substituted on the N-atom
only. As these phenomena were observed for various substituents
on both atoms, it is reasonable to assume that the substitutions
partially exert their effects on the affinity through altering the con-
formational preference of the compounds. In order to examine this
assumption conformational analysis of model compounds in
Table 3 was performed.12 It was found that Model 3 (N-substituted)
and Model 4 (N- and C-substituted) adopt highly similar shape in
their lowest energy conformation as it is shown in Figure 3.
Table 2
Exploration of the substitutions around the central nitrogen atom
O
O
O
R3
N
N
O
O
R1 R2
Compound
R1
R2
R3
hCXCR3a (IC50 nM)
Met. stab.b (h/m; %)
2a
2i
2j
2k
2l
2m
2n
7a
7b
7c
7d
7e
7f
Me
Me
Et
H
H
H
H
H
H
H
H
Me
H
H
H
H
H
Me
i-Bu
Et
i-Bu
Me
Me
i-Bu
H
H
H
H
H
403
55
110
38
53/52
2/9
5/5
19/19
9/8
N.D.
7/11
61/62
N.D.
N.D.
N.D.
N.D.
N.D.
61/67
Et
n-Pr
i-Bu
H
Me
Me
Et
n-Pr
i-Bu
c-Penthyl
H
110
330
70
607
780
1060
1200
1800
3260
290
H
H
7g
N.D. means not determined.
a
Displacement of 125I-labeled IP-10 from human recombinant CXCR3 receptors.7
Human/murine liver microsomes.10
b

5422
I. Bata et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5418–5428
Table 3
Model compounds used in the conformational analysis
Model compound
Name
Representative CXCR3 ligand
hCXCR3a (IC50 nM)
Compound
O
O
O
N
N
Model 1
290
7g
H
N
O
O
O
O
O
O
O
O
O
O
N
N
N
N
Model 2
Model 3
Model 4
607
70
7a
2n
2a
H
N
O
O
O
O
N
N
N
N
O
O
403
O
a
Displacement of 125I-labeled IP-10 from human recombinant CXCR3 receptors.7
however, available data suggest that such a range exists but it is
smaller and shifted towards higher values with respect to that of
the succinimide derivatives.
Compounds with free carboxyl group and logD between 1 and 3
were designed in order to retain activity while to improve ADME
Table 4
Exploration of substitution of the central N-atom
R3
N
O
O
O
O
N
O
Figure 3. Minimum energy conformation of Model
4 of Table 3 (N-atom
protonated).
Compound R3
hCXCR3a (IC50 nM) Met. stab.b (h/m; %)
2a
CH3
403
82
53/52
3/4
however, both low Caco-2 permeability9 and low oral bioavailabil-
ity (F) in mice13 (Table 4).
OH
OH
10a
In order to facilitate the design of compounds with appropriate
ADME parameters relationships were sought between various
easily computed parameters (like logD, logP, polar surface area)8
and measured permeability and metabolism values. It was found
that permeability tends to increase with logD corresponding to
pH = 7.4. This is in line with the general observation that absorp-
tion increases linearly with lipophilicity, at least in a certain range
of the latter.14
10b
10c
10d
77
2/0
NH2
O
2200
202
48/13
2/0
10e
10f
52
35/27
19/29
N
190
Figure 4a shows, that a logD >1 is required to achieve appropri-
ate absorption (Papp >20 nm/s). On the other hand, metabolic sta-
bility exhibits a slight decrease with the increasing lipophilicity
that sets a limit to logD. Figure 4b suggests that it is advantageous
to keep logD around 3 or below. These two limits of logD >1 set by
permeability and logD <3 set by metabolic stability were used as
guidelines in the design of new compounds.
It is important to note that both permeability and metabolism
studies discussed above are restricted to compounds with terminal
succinimide. It was found, that when succinimide was replaced by
N-methyluracil then both the logD-permeability and the logD-
metabolism relationships changed; higher logD values were calcu-
lated at the same pH which resulted in higher permeability and
lower metabolic stability. The lower number of compounds with
N-methyluracil does not allow defining a logD range where both
permeability and metabolism are expected to be acceptable,
N
10g
10h
130
97
41/42
24/19
F
10i
10j
10k
340
28/24
100/95
61/63
O
210
OH
Caco-2 = 0c;
F = 2%d
O
OH
120
Caco-2 = 0c;
O
F = 9%d
a
b
c
Displacement of 125I-labeled IP-10 from human recombinant CXCR3 receptors.7
Human/mouse liver microsomes.10
Permeability on Caco-2 cell assay.9
Bioavailability in mice.13
d

I. Bata et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5418–5428
5423
Figure 4. (a) Caco-2 permeability (Papp in 10ꢁ7 cm/s) and (b) metabolic stability (%) in mouse liver microsomes vs. calculated logD at pH = 7.4. Acceptable permeability and
metabolic stability are most probably achievable in the shaded region.
Table 5
Simultaneous optimization of three parameters (activity, metabolic stability and permeability)
O
R3
N
O
N
O
B
O
Compound
B
R3
hCXCR3a (IC50 nM)
clogDb
Met.stab.c (h/m; %)
61/63
Caco-2d (nm/s)
0
O
120
F = 9%
COOH
COOH
10k
1.45
O
11a
160
2.0
60/78
14.6
O
H
N
N
11b
11c
11d
126
345
138
1.9
51/77
98/100
84/87
2
N
N
Cl
COOH
2.1
37
13.4
O
COOH
1.81
O
O
COOH
11e
11f
4200
3.0
52/N.D.
92/94
N.D.
32.1
COOH
382
225
1.67
O
O
COOH
COOH
11g
11h
1.6
1.2
71/79
98/92
65.6
9.8
60
F = 17.3%
N.D. means not determined.
a
Displacement of 125I-labeled IP-10 from human recombinant CXCR3 receptors.7
Calculated LogD.8
b
c
Human/mouse liver microsomes.10
Permeability on Caco-2 cell assay.9
d

5424
I. Bata et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5418–5428
Table 6
Optimization of cyclohexanecarboxylic acids
Compound
Structure
hCXCR3a (IC50 nM)
Met.stab.b (h/m; %)
CaCo-2c (nm/s)
Developabilityd
HO
HO
HO
HO
HO
HO
HO
HO
HO
F = 17.3%
M2 = 54%
O
O
O
O
O
O
O
O
O
O
O
O
11h
60
98/92
9.8
O
O
O
N
N
N
N
N
N
N
N
N
N
N
N
O
O
O
O
O
O
12
13
14
15
16
17
18
19
760
37
55
35
44
19
76
38
N.D.
N.D.
29.3
11.6
39.5
22.2
52.8
80.1
195.2
O
O
O
O
O
O
M2 = 72%
MOP = 94%
73/73
88/93
56/92
70/87
91/91
62/76
73/87
O
O
S
pH >6.5 chem. unstable
O
O
O
M2 = 98%
O
Cl
F = 4.1
M2 = 99%
MOP = 78%
O
O
F = 13.3%
MOP = 74%
O
O
O
O
O
Cl
N
N
O
O
F = 26.7%
HEP = 0.159
O
O
O
Cl
Cl
N
N
N
O
N
F = 93.5%
HEP = 0.138
a1A = 71%
O
M2 = 97%
O
N
O

I. Bata et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5418–5428
5425
Table 6 (continued)
Compound
Structure
hCXCR3a (IC50 nM)
Met.stab.b (h/m; %)
CaCo-2c (nm/s)
Developabilityd
HO
F = 63.3%
HEP = 0.234
O
O
20
21
99
63/100
157.3
O
Cl
N
N
N
O
HO
a
1A = 52%
MOP = 63%
O
M2 = 0%
O
24
66/78
89.8
O
N
Cl
N
O
O
N.D. means not determined.
a
Displacement of 125I-labeled IP-10 from human recombinant CXCR3 receptors.7
Human/mouse liver microsomes.10
b
c
Permeability on Caco-2 cell assay.9
d
F: Absolute oral bioavailability in mice.13 Radioligand displacement on different receptors at 10
lM (M2: muscarinic; a1A: adrenergic; MOP: mu opioid); HEP: human
hepatocyte clearance.15
the oral bioavailability decreased (F = 13.3%). The oral bioavailabil-
ity improved by replacing the succinimide ring of this compound
by N-methyluracil (18), but the human hepatocyte clearance
(HEP; mL/h/106 cells) became too high.15 The methoxy—methyl
replacement on the benzyl group of 17 led to excellent oral
bioavailability in mice (19, F = 93.5%), but it was associated with
selectivity problems and somewhat higher than desirable human
hepatocyte clearance. Changing the succinimide group of this latter
compound to N-methylhydantoin (20) corrupted both oral
bioavailability and hepatocyte clearance.
Figure 5. Minimum energy conformations of 1-phenylethyamine and indan-1-
ylamine.
The introduction of 5-chloroindanyl group in place of the
benzofuranylethyl moiety led to significant improvement in
permeability while the compounds retained advantageous activity
(cf. 21 and 11h). The preservation of activity by this new structural
element is in line with the conformational analysis performed to
understand the coupled effect of substituents on the N- and adja-
cent C-atoms (see above). These calculations suggested that the
binding conformation is bent and the CAN bond is close to being
perpendicular to the aromatic ring of the terminal group (e.g.
phenylethyl or benzofuranylethyl) as it is shown in Figure 5. The
direction of the alpha methyl group of this conformation fits well
to the direction of the corresponding bond in the minimum energy
structure of 1-indanamine. This suggests that closing the alpha-
methyl substituent in a ring and thus forming an indane-amine
is in line with the bioactive conformation.
Compound 21 exhibited mild affinity towards the adrenergic
alpha 1A and mu opiate receptors (it has no effect on muscarin
M2 receptor). It is worth mentioning that the 6-chloroindane
derivative showed a two order of magnitude lower affinity towards
the CXCR3 receptor than did the 5-chloroindane compound (data
not shown).
As a first step in ‘fine-tuning’ the structure of 21 (Table 7) the
activity of the cis isomer (22) was measured. The former had a
25-fold higher binding affinity than did the latter. This is in line
with the relative activity of 11h and 12 (see above). The separation
of the racemic 21 resulted in a highly active S-enantiomer (24,
IC50 = 13 nM) and a 15 fold less active R-enantiomer (23). In addi-
tion, 24 showed better selectivity and higher metabolic stability.
Since the final step in the synthesis of 24 is the hydrolysis of the
ester function (Scheme 2) the activity of the 25 ester was also mea-
sured and was found lower by almost two orders of magnitude, so
it did not work as a prodrug either.
parameters and ultimately oral bioavailability (Table 5). Increasing
the number of C-atoms in the substituents neighboring to the
N-atom was beneficial to the permeability but decreased the activ-
ity (11a, 11e, 11f). The bioisosteric replacement of the carboxylic
group by tetrazole (11b) did not improve the properties. The
cinnamic acid derivative (11d) had slightly better permeability
and it was further improved with bicyclo[2,2,2]octanecarboxylic
acid (11g). This latter, however, showed decreased activity, just
as the compound obtained with the replacement of benzofuran
by 4-chlorophenyl group (11c). The introduction of the cyclohex-
anecarboxylic acid moiety (11h) advantageously changed all
investigated parameters; activity and permeability increased while
metabolism decreased. These improvements associated with
higher bioavailability in mice (F = 17.3%).
It was found during the further investigation of compounds
possessing a cyclohexane carboxylic acid group (Table 6) that the
trans isomer (11h) is ten-fold more active than the cis isomer
(12). Therefore, the trans isomer was used in all subsequent stud-
ies. In an attempt to improve oral bioavailability the succinimide
group was replaced first by dioxo-thiazolidine (13) and then by
dioxo-oxazolidine (14). Both activity and permeability improved,
but 13 showed some affinity towards mu-opioid (MOP) and
muscarin (M2) receptors, while 14 was found to be chemically
unstable above pH = 6.5 (ring opening occurred). The introduction
of a chlorine atom or a methyl group instead of the methoxy group
on the benzyl group was also accompanied by the appearance of
selectivity issues (15 and 16). Replacement of the benzofuran
moiety by chlorophenyl (17) was highly beneficial for the activity
(IC50 = 19 nM), but affinity towards opiate receptor appeared and

5426
I. Bata et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5418–5428
Table 7
Fine-tuning of the compound 21
Compound
Structure
hCXCR3* (IC50 nM)
121
Met. stab.* (h/m; %)
Caco-2* (nm/s)
251
Developability*
AMG-487
See Figure 1
77/70
HO
HO
HO
HO
a
1A = 52%
MOP = 64%
O
O
O
O
M2 = 0%
O
O
O
O
21
22
23
24
24
66/78
89.8
N.D.
66.2
75.2
O
O
N
N
N
N
Cl
Cl
Cl
Cl
N
N
N
N
O
O
O
O
1660
200
13
N.D.
O
O
54/55
70/79
O
O
F = 24%
HEP = 0.112
a1A = 25%
MOP = 43%
O
O
O
HO
HO
HO
O
O
O
O
O
O
O
O
25
840
N.D.
N.D.
O
O
N
N
N
N
Cl
Cl
Cl
Cl
N
N
N
N
O
O
O
O
F = 69%
HEP = 0.090
M1 = 82%
M2 = 92%
MOP = 87%
26
27
28
53
78/91
94/96
36/24
81.2
80.2
O
O
HEP = 0.069
M2 = 37%
176
1000
165.3
O

I. Bata et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5418–5428
5427
Table 7 (continued)
Compound
Structure
hCXCR3* (IC50 nM)
Met. stab.* (h/m; %)
Caco-2* (nm/s)
Developability*
HO
HO
HO
O
O
O
O
O
29
30
23
15/3
72.4
S
N
N
Cl
N
N
O
O
O
HEP = 0.083
M2 = 0%
25
75/81
57
O
O
Cl
Cl
F = 47%
HEP = 0.031
M2 = 16%
O
31
26
73/81
29.7
O
O
N
Cl
F
N
O
*
See Table 6 for the abbreviations used.
O
NH2
O
O
O
N
b
O
O
a
Cl
H
N
N
+
O
O
Cl
Cl
O
O
7h
6b
5a
COOMe
COOH
O
O
O
O
O
6
N
c
N
5
Cl
N
N
O
O
1
4
2
3
25
24
Scheme 2. Synthesis of compound 24. Reagents and conditions: (a) NaBH(OAc)3, THF, rt; (b) trans 4-formylcyclohexanecarboxylic acid methyl ester (8b), NaBH(OAc)3, THF, rt.
(c) HCl, dioxane, 60 °C.16
Replacing the methoxy group on the central benzyl moiety by
methyl- (26) or ethyl-group (27) the affinity towards the CXCR3
receptor decreased. The replacement of the O-atom in the 4th posi-
tion on the central benzyl moiety by CH2-group (28) decreased the
affinity by two orders of magnitude. By contrast, the affinity was
retained with an S-atom (29) instead of the O-atom, however,
the metabolic stability was significantly reduced. These results
suggest that the lone-pair present on both the O- and S-atoms
but missing on the CH2-group contributes advantageously to the
activity. Owing to the expected lower metabolic stability of the
compound containing S-atom, the properties are optimal with
the O-atom. Exploring the possible substitutions on the indane-
ring it was found that the activity only tolerates substitution in
the 4th position in addition to the 5-chloro-substituent: 4-Cl (30)
and 4-F (31) derivatives both exhibit slightly weaker affinities than
does compound 24.
In summary, a new series of potent CXCR3 antagonists was
discovered. Starting with an HTS campaign, the Lead compound
was extensively optimized for potency, in vitro metabolic stability,
permeability and pharmacokinetic profile. Calculated and mea-
sured physico-chemical properties together with conformational
analysis supported the optimization. Several compounds have
highly advantageous profile and they are currently subject to
further evaluation.
Acknowledgements
The authors greatly appreciate Sanofi R&D management for the
strong support, the CXCR3 project team for their contributions, the
Sanofi analytical department for their support on analytical studies
in confirmation of the structures and for chiral separation and

5428
I. Bata et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5418–5428
7. The CXCR3 binding affinity was measured by displacement of 125I-CXCL10 (IP-
10) from human recombinant CXCR3 receptors generated by transfection of
Flp-In-CHO host cells with plasmid construct of pCDA5-FRT-TO_IRES-
Sanofi DMPK and Toxicology Department for PK and in vitro/
in vivo safety assessments.
G
ai4qi4_DEST. IC50’s are the mean values (n = 2) with SD of ꢀ13%. See Ref.
17 for detailed assay protocol.
Supplementary data
8. Calculated logP, logD, PSA were determined by ACD/LogD Suite. Version 8,
Advanced Chemistry Development, Inc.,Toronto, ON, Canada.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2016.10.
035. These data include MOL files and InChiKeys of the most
important compounds described in this article.
9. The Caco-2 cell assay was carried out in 24 and 96-well plates (Millipore). After
verification of cell monolayer integrity, high and low permeability controls and
test compound were added. Samples from both apical (A) and basal (B) sides
were taken 2 h later. Compound transport was measured under four different
conditions: A-B flux, pH 6.5/7.4 (0% or 0.5% BSA in A and 5% BSA in B); A-B flux
and B-A flux pH 7.4/7.4 (0.5% BSA in A and B respectively). The amount of
compound in each compartment was measured by LC/MS. Caco-2 value of the
rapidly transported compounds was Papp >20 nm/s (=20 ꢂ 10ꢁ7 cm/s).
10. Metabolic stability of test compounds was performed using human and mouse
liver microsomes in presence of NADPH at 37 °C. Parent compound loss was
followed using LC–MS–MS technique. Classification: low: <40%; medium:
between 40% and 60%; high: >60%.
11. Recently published low molecular weight CXCR3 inhibitors were different
from our tertiary amine library; see: Wijtmans, M.; de Esch, I. J. P.; Leurs, R. In
Chemokine Receptors as Drug Targets; Smit, M. J., Ed.; Wiley-VCH, 2011; p 302.
Chapter 13.
12. MacroModel, version 9.9; Schrödinger, LLC: New York, NY, 2011.
13. Absolute oral bioavailability was measured following a single oral (10 mg/kg)
and intravenous (3 mg/kg) administration to male Balb/c mouse (4 and 4
animals for individual profiles in each animal or using spare sampling in 24
mice, and 3 mice per time point for mean profiles).
References and notes
1. Groom, J. R.; Luster, A. D. Immunol. Cell Biol. 2011, 89, 207.
2. (a) Wijtmans, M.; Verzijl, D.; Leurs, R.; de Esch, I. J. P.; Smit, M. J. Chem. Med.
Chem. 2008, 3, 861; (b) Muller, M.; Carter, M. J.; Hofert, J.; Campbell, I. L.
Neuropathol. Appl. Neurobiol. 2010, 36, 368; (c) Liu, M.; Guo, S.; Hibbert, J. M.;
Jain, V.; Singh, N.; Wilson, N. O.; Stiles, J. K. Cytokine Growth Factor Rev. 2011, 22,
121.
3. (a) Hansel, T. T.; Barnes, P. J. Lancet 2009, 374, 744; (b) Krueger, J. G.; Bowcock,
A. Ann. Rheum. Dis. 2005, 64, 30; (c) Hancock, W. W.; Lu, B.; Gao, W.; Csizmadia,
V.; Faia, K.; King, J. A.; Smiley, S. T.; Ling, M.; Gerard, N. P.; Gerard, C. J. Exp. Med.
2000, 192, 1515; (d) Lammers, K. M.; Khandelwal, S.; Chaudhry, F.; Kryszak, D.;
Puppa, E. L.; Casolaro, V.; Fasano, A. Immunology 2011, 132, 432; (e) Nishimura,
M.; Kuboi, Y.; Muramato, K.; Kawano, T.; Imai, T. Ann. N.Y. Acad. Sci. 2009, 1173,
350 (Contemp. Challenge Autoimm.); (f) Shimida, A.; Oikawa, Y.; Yamada, Y.;
Okubo, Y.; Narumi, S. Rev. Diab. Studies 2009, 6, 81; (g) Pease, J. E.; Horuk, R.
Expert Opin. Ther. Pat. 2009, 19, 199; (h) Breser, M. L.; Motrich, R. D.; Sanchez, L.
R.; Marckern-Oberti, J. P.; Rivero, V. E. J. Immun. 2013, 190, 3121.
14. Waring, M. J. Exp. Opinion Drug Discov. 2010, 5, 235.
15. Intrinsic hepatic clearance (HEP) was determined using primary or
cryopreserved human hepatocytes and compound concentration of
5 lM,
in vitro Clint value is mL/h/10ꢁ6 cells. Classification (mL/h/10ꢁ6 cells): low:
<0.04; intermediary: between 0.04 and 0.12; high: >0.12.
4. Andrews, S. P.; Cox, R. J. J. Med. Chem. 2016, 59, 2894.
5. Henne, K. R.; Tran, T. B.; VandenBrink, B. M.; Rock, D. A.; Aidasani, D. K.;
Subramanian, R.; Mason, A. K.; Stresser, D. M.; Teffera, Y.; Wong, S. G.; Johnson,
M. G.; Chen, X.; Tonn, G. R.; Wong, B. K. Drug Metab. Dispos. 2012, 40, 1429.
6. Bioactivity was tested using engineered CHO cells expressing the human
16. (S)-5-Chloroindan-1-ylamine (6b) and trans 4-formylcyclohexanecarboxylic
acid methyl ester (8b) were commercially available materials.
17. Bata, I.; Buzder-Lantos, P.; Vasas. A.; Bartáne-Bodor, V.; Ferenczy, Gy.;
Tömösközi, Zs.; Szeleczki, G.; Batori, S.; Smrcina, M.; Patek, M.; Weichse, A.;
Thorpe, D. US 9073853 B2.
CXCR3A coupled to a chimeric G-protein (Gai4qi4). In the FLIPR instrument,
following basal readings, IP-10 is added and fluorometric images are read.
IC50’s are the mean values (n = 2) with SD of ꢀ15%.