CXCR3 antagonists: Quaternary ammonium salts equipped with biphenyl- and polycycloaliphatic-anchors

Article abstract of DOI:10.1016/j.bmc.2011.04.035

Small-molecule ligands for the CXCR3 chemokine receptor receive considerable attention as a means to interrogate the roles of CXCR3 in vitro and in vivo and/or to potentially treat various conditions such as inflammatory diseases and cancer. We have synthesized and explored a novel class of small-molecule antagonists for CXCR3. A medium-throughput screen revealed an adamantane-guanidine as a hit. The guanidine unit was replaced by a small quaternary ammonium group, leading to ca. 5-fold increase in affinity. Substitution of the adamantane group by a myrtenyl moiety further increased affinity, while the benzyl group was decorated with an additional (substituted) aryl ring. This led to the identification of several bisaryl-based ligands with CXCR3 affinities of around 100 nM and with the ability to antagonize the functional activity of CXCL10.

Full text of DOI:10.1016/j.bmc.2011.04.035

Bioorganic & Medicinal Chemistry 19 (2011) 3384–3393
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
CXCR3 antagonists: Quaternary ammonium salts equipped with biphenyl-
and polycycloaliphatic-anchors
Maikel Wijtmans, Dennis Verzijl, Serge Bergmans, Michael Lai, Leontien Bosch, Martine J. Smit,
⇑
Iwan J. P. de Esch, Rob Leurs
Leiden/Amsterdam Center for Drug Research, Division of Medicinal Chemistry, Faculty of Sciences, VU University Amsterdam, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands
a r t i c l e i n f o
a b s t r a c t
Article history:
Small-molecule ligands for the CXCR3 chemokine receptor receive considerable attention as a means to
interrogate the roles of CXCR3 in vitro and in vivo and/or to potentially treat various conditions such as
inflammatory diseases and cancer. We have synthesized and explored a novel class of small-molecule
antagonists for CXCR3. A medium-throughput screen revealed an adamantane-guanidine as a hit. The
guanidine unit was replaced by a small quaternary ammonium group, leading to ca. 5-fold increase in
affinity. Substitution of the adamantane group by a myrtenyl moiety further increased affinity, while
the benzyl group was decorated with an additional (substituted) aryl ring. This led to the identification
of several bisaryl-based ligands with CXCR3 affinities of around 100 nM and with the ability to antago-
nize the functional activity of CXCL10.
Received 18 February 2011
Revised 4 March 2011
Accepted 17 April 2011
Available online 24 April 2011
Keywords:
CXCR3
Antagonist
Suzuki coupling
Ammonium salt
Myrtenyl
Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction
diseases.3,4 A massive medicinal chemistry effort, mostly driven
by the pharmaceutical industry, has revealed an increasing number
Chemokines are small endogenous peptides most notably in-
volved in the recruitment of immune cells. They exert their action
through activation of chemokine receptors, which belong to the
superfamily of GPCRs. The group of reported chemokine receptors
has steadily grown over the years and 19 members are currently
known.1,2 Blocking chemokine receptors by antagonists is thought
to be associated with a large therapeutic potential, as a distorted
immune system is involved in, for example, autoimmune diseases,
inflammatory conditions and viral infections. A recent report re-
views the development of chemokine receptor antagonists in more
detail.2
CXCR3 is a prominent member of the class of chemokine recep-
tors. CXCR3 is a 368-amino acid protein found on, for example,
activated Th1 lymphocytes and on a small proportion of B cells
and natural killer cells. The endogenous agonists for CXCR3 are
the CXC chemokines CXCL9, CXCL10 and CXCL11, with the latter
having the highest potency and efficacy. Based on expression levels
and knockout studies, potential therapeutic roles for CXCR3 have
been postulated in areas such as cancer and several autoimmune
of chemotypes as CXCR3 antagonists.3–6 A selection of published
antagonist structures is collected in Figure 1.7–12 Reports on pre-
clinical studies with such small-molecule antagonists (most nota-
bly 1, AMG487) have disclosed positive results for some of the
postulated roles of CXCR3 (e.g., in cancer and certain autoimmune
diseases) whereas a controversy has emerged on the role of CXCR3
in organ transplant rejection.4,13–15
With the aim of obtaining tool compounds for our ongoing
CXCR3 research,11,16,17 we set out to develop and investigate a
new CXCR3 antagonist scaffold. Toward this end, synthetic accessi-
bility should be high and straightforward introduction of chemical
diversity should be possible. In this paper, we report on such a
scaffold class. Starting with a hit from a medium-throughput
screening campaign, strategic changes in core structure were ap-
plied. This was followed by SAR studies which eventually led to
the inception of a class of ammonium salts, all within excellent
reach by the use of robust reactions.
2. Results and discussion
2.1. Screening campaign and design of a core structure
Abbreviations: IPAG, 1-(4-Iodophenyl)-3-(2-adamantyl)guanidine; DCE, 1,2-
Dichloroethane; TEA, Triethylamine; hCXCR3, Human CXCR3; cLogD, Calculated
Log D; GPCR, G protein-coupled receptor; TFA, Trifluoroacetic acid; HEK, Human
embryonic kidney; SAR, Structure–activity relationship; rt, Room temperature.
A
proprietary library, containing 3360 pharmacologically
active compounds, was screened at 10 M for the ability to
displace 125I-CXCL10 from membranes of HEK293 cells stably
l
⇑
Corresponding author. Tel.: +31 20 59 87579; fax: +31 20 598 76 10.
E-mail address: Leurs@few.vu.nl (R. Leurs).
0968-0896/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2011.04.035

M. Wijtmans et al. / Bioorg. Med. Chem. 19 (2011) 3384–3393
3385
OEt
N
N
O
N
NH
H
N
H
N
N
CF3
O
N
N
O
N
N
F
O
OCF3
Cl
1
4
2
3
O
Cl
N
Cl
O
N
O
Cl
H
NH
N
N
N
H
N
Cl
Cl
O
N
N
HN
5
Figure 1. Structures of selected CXCR3 antagonist chemotypes.
expressing hCXCR3. The screening campaign had a hit-rate of 2.7%:
90 compounds out of 3360 gave >50% displacement at 10 M. Fig-
ure 2 shows the results for a particularly rich plate. Compounds
which inhibited specific binding of 125I-CXCL10 for more than
50% were further evaluated by generation of full concentration-dis-
placement curves.
tion with a CXCR3 residue, we decided to replace the guanidine
with two different units capable of engaging in ionic interactions:
a tertiary amine or a quaternary ammonium cation (Fig. 3). The
distance between the aryl and adamantanyl units was kept similar
to IPAG by inserting two CH2 spacers. The resulting core scaffolds
were used to introduce diverse chemical decoration, both at the
polycycloaliphatic as well as at the aromatic anchor.
l
Several interesting hits emerged from this medium-throughput
screen. We were intrigued by 1-(4-iodophenyl)-3-(2-adaman-
tyl)guanidine (IPAG, 6), a sigma-receptor ligand (see Fig. 2).18 IPAG
gave a moderate CXCR3 affinity of pKi = 5.4 in the 125I-CXCL10 dis-
placement assay. The structure of IPAG can be divided into three
structurally different parts: (1) a highly lipophilic adamantane
group, (2) a basic guanidine unit and (3) a 4-I-substituted phenyl
group. The notion of the adamantane group coincided with our
general interest for a potential CXCR3 pocket capable of accommo-
dating polycycloaliphatic groups. Indeed, the finding of the ada-
mantane being adjacent to a basic center in IPAG was integrated
into a merging approach on which we have published, eventually
giving rise to antagonists such as 7a, b.17 In the current study,
we followed a conceptually different approach. The reported syn-
thesis of compounds like IPAG involves reaction of the aniline
hydrochloride with an adamantyl-cyanamide (by heating in the
melt or refluxing for extended periods of time), the latter of which
in turn has to be prepared by reaction with CNBr.19 Such a syn-
thetic approach is deemed non-optimal if chemical diversity is to
be pursued. Hypothesizing that the basic guanidine unit may act
(after protonation) as an important contributor to an ionic interac-
2.2. Synthesis
The syntheses of the tertiary amines and of the ammonium salts
were highly integrated, with the former being the precursor for the
latter. Thus, salts 8a–g, 9a–j and 10a–u were all prepared in a sin-
gle step from the corresponding tertiary amines 11a–g, 12a–j and
13a–u (with 10p being the sole exception, vide infra). The case of
the exploration of the group is illustrative for this general synthetic
approach (Scheme 1).
A reductive amination between any of the aldehydes 14–19
(syntheses shown in Supplementary data) and 4-iodobenzylamine
hydrochloride using NaBH(OAc)3 delivered secondary amine 20
(31–69%).20 Subsequent Eschweiler–Clark methylation afforded
the tertiary amine 11 (74%-quant.). Last, methylation of 11 with
an excess of MeI in the dark followed by cold precipitation with
DCM/tBuOMe afforded the iodide salt 8 (15–92%). Although con-
versions were always high in the MeI-reactions described here
and in following paragraphs, isolated yields varied considerably
due to, for example, varying solubilities of the product salts in
125
100
75
50
25
0
Figure 2. A part of the screening campaign, displaying one 96-well plate (80 compounds). Each bar represents a compound (10 lM) from the library. Data are presented as
percentage specific 125I-CXCL10 binding to hCXCR3. Non-specific binding was determined in the presence of the CXCR3 antagonist 1 (10 M). Inhibition of 125I-CXCL10
l
binding by IPAG is indicated.

3386
M. Wijtmans et al. / Bioorg. Med. Chem. 19 (2011) 3384–3393
approach required bromo-containing building blocks 12g and 23.
H
N
H
N
Z
Compound 23 was obtained by the sequence starting from 22 as
discussed for 12b. The same held true for the synthesis of 12g,
which was scaled up to 10 g without the need for chromatography
using this alternative route. Gratifyingly, reaction of 12g or 23 with
many arylboronic acids under microwave Suzuki conditions
smoothly afforded the tertiary bisaryl-amines 13a, f–o, q–u and
24 (30–79%). No interference of the myrtenyl double bond was
encountered in these reactions. Precursors 13b–e were prepared
from 22 and the corresponding commercially available aldehydes
(39–90%). Once again, all syntheses were completed by reaction
of 13a–o, q–u with MeI to the salts 10a–o, q–u (29–89%). For an
aniline substituent, compatibility problems were foreseen and we
opted to use Boc-protected anilines 24 and 25 as surrogates
throughout the sequence. Target anilines 13p and 10p were ob-
tained from 24 and 25 by standard TFA-mediated deprotection
and by convenient acid-free microwave heating of an aqueous mix-
ture,21 respectively.
I
NH
I
6 (IPAG)
=
Z
H
N
H
H
N
Cl
N
R
6:
N
NH
F
N
R=
R=
7a
7b
N
Figure 3. Design strategy based on the hit IPAG (6).
The large majority of ammonium salts comprised solids that
were readily soluble in DMSO, MeCN or CHCl3. The stability of se-
lected salts in solution was evaluated and it was shown that these
salts possess good stability in DMSO or MeOH/H2O (see Supple-
mentary data).
the precipitation medium. Introduction of one ethyl group was
accomplished by reductive amination between 20f and acetalde-
hyde to 11g, followed by a similar methylation to salt 8g. A second
ethylation of 11g with EtI failed under several conditions, most
likely due to the steric crowding imposed upon the nitrogen atom.
The synthesis of the majority of compounds with varying aro-
matic side chains (Scheme 2) followed a strategy which is similar
to the one depicted in Scheme 1. That is, myrtenal (19) was reduc-
tively aminated with various commercially available amines to
21a, c, e–j (20–39%). Compounds 21a, c, e–j were subsequently
methylated under Eschweiler–Clark conditions to tertiary amines
12a, c, e–j (43–97%). A more convenient and efficient approach
for the syntheses of compounds like 12 was initiated at the stage
of the 2-iodo-compound 12b. That is, large-scale imine formation
between 19 and H2NMe followed by NaBH4-based reduction affor-
ded building block 22, which was reductively alkylated by 2-iodo-
benzaldehyde to yield 12b. Cyclic tertiary amine 12d was prepared
by reductive alkylation of tetrahydroisoquinoline with 19. All syn-
theses were finalized by methylation of 12a–j to salts 9a–j as de-
scribed (31–91%).
2.3. Binding affinities
All synthesized compounds were tested for their affinity for
hCXCR3 by measuring the displacement of 125I-CXCL10 binding
to membranes of HEK293 cells stably expressing hCXCR3.16 In this
assay, the reference CXCR3 antagonist AMG487 (1) gives a pKi
value of 7.4 0.1 (n = 19), while CXCL10 binds CXCR3 with a pKd
value of 9.7 0.1 (n = 4).
As shown in Figure 3, the design strategy involved replacement
of the guanidine in IPAG by a tertiary amine and subsequently by a
quaternary ammonium cation. Since the secondary amine was an
intermediate en route to these two compounds, it was also in-
cluded in our initial pharmacological analysis. As Table 1 shows,
this secondary amine (20a) gave a slightly higher affinity com-
pared to IPAG, but the tertiary amine (11a) led to 0.5 log unit drop
in affinity. More interestingly, methylation to a permanent cation
(8a) lifted the pKi substantially above that of IPAG (from 5.4 to
6.1). While it is known that increasing lipophilicity, for example
A Suzuki reaction was envisioned as the key reaction in the syn-
theses of functionalized bisaryls 13a, f–u (Scheme 3). This
O
O
O
O
O
O
14
15
16
17
18
c)
19
I
I
I
a)
b)
H
14-19
N
N
N
R
R
R
I
20a-f
11a-f
8a-f
I
I
I
c)
d)
H
N
N
N
I
11g
8g
20f
Scheme 1. Synthesis of compounds with varying cycloaliphatic groups and of N-ethyl analogues. Reagents and conditions: (a) 4-iodobenzylamine hydrochloride,
NaBH(OAc)3, DCE, TEA, rt, 1–2 days; (b) HCOOH, 37% aq H2CO, 2–4 h, reflux; (c) [1] DCM, MeI, 1–3 days, rt [2] DCM/tBuOMe, 0 °C, filtration; (d) MeCHO, NaBH(OAc)3, DCE,
2.5 h, tr, 65%.

M. Wijtmans et al. / Bioorg. Med. Chem. 19 (2011) 3384–3393
3387
a)
c)
c)
b)
H
N
N
N
R
I
R
R
R-NH2
19
O
21a,c,e-j
12a,c,e-j
9a-c, e-j
I
d)
a)
e)
c)
NH
N
12b
9d
19
19
O
O
22
N
N
HN
12d
I
Scheme 2. Synthesis of compounds with varying aromatic groups. Reagents and conditions: (a) NaBH(OAc)3, DCE, (TEA), rt, 1–2 d; (b) HCOOH, 37% aq. H2CO, 2–4 h, reflux; (c)
[1] DCM, MeI, 1–3 d, rt [2] DCM/tBuOMe, 0 °C, filtration. (d) [1] H2NMe (33% in EtOH), MeOH, 1 d, rt [2] NaBH4, 3 h, rt, 89%. (e) NaBH(OAc)3, 2-iodobenzaldehyde, DCE, rt, 20 h,
36%.
NH2
NH2
N
N
I
I
13p
10p
From 24
e)
From 25
f)
Ar
Br
Ar
a)
c)
c)
d)
d)
N
N
NH
N
22
12g
13f-o, q-u
10f-o, q-u
24
25
:Ar=3-(BocNH)-Ph
:Ar=3-(BocNH)-Ph
b)
N
b)
N
N
Ph
Br
Ph
I
23
13a
10a
R2
R3
R2
R3
d)
N
N
I
13b-e
10b-e
Scheme 3. Synthesis of compounds with varying bisaryl groups. Reagents and conditions: (a) [1] 4-BrPhCHO, NaBH(OAc)3, DCE, rt, 20 h [2] HCl, Et2O [3] recrystallisation, 71%
as HCl salt; (b) substituted benzaldehyde, NaBH(OAc)3, DCE, rt, 1–2 days; (c) Pd(PPh3)4, ArB(OH)2, Na2CO3, iPrOH, toluene, H2O, W, 1.5 h; (d) [1] DCM, MeI, 1–3 days, rt [2]
DCM/tBuOMe, 0 °C, filtration; (e) TFA, DCM, rt, 1 h, 88%; (f) H2O, W, 2 h, 88%.
l
l
by N-methylation, can increase the affinity of protein ligands, the
calculated LogD values for homologues 20a, 11a and 8a (Table 1)
clearly suggest that lipophilicity is not the major contributor to
their affinities. Studies with the tetra-alkyl-ammonium salt TAK-
779 and the CCR5 chemokine receptor have shown that the nature
of the binding of a tetra-alkyl-ammonium cation could involve a
subtle mix of ionic interactions with a negatively charged residue
precursors, we were well suited to test a variety of such tertiary
amines, the affinities of which will also be reported throughout
the manuscript. This enabled the investigation of the general valid-
ity of our hypothesis dictating (an) electrostatic interaction(s) to be
a key contributor to CXCR3 affinity.
Having established the quaternary ammonium cation as a suit-
able and synthetically accessible core, the stage was set for decora-
tion of the peripheral groups. As mentioned, the notion of the
adamantane in IPAG coincided with our search for a potential
‘polycycloaliphatic’ CXCR3 pocket17 and this led us to first probe
the ‘left-hand’ side (Table 2).
and cation–
p
interactions with aromatic residues.22 In analogy,
we believe that the ability of our core to engage in (an) electro-
static interaction(s) plays a dominant role in the observed affini-
ties. Indeed, sequential N-methylation has been used before as a
probe for electrostatic interactions in other proteins.23,24 Since all
of our ammonium salts synthetically required the tertiary amine
A shift of the adamantane toward the 1-position (8b) led to low-
er affinity than for 8a, but it could be restored by chain elongation

3388
M. Wijtmans et al. / Bioorg. Med. Chem. 19 (2011) 3384–3393
Table 1
the polycycloaliphatic head of the current IPAG-derived set is tar-
geting the receptor in a different manner compared to the 4-ami-
no-piperidine and 1-aryl-3-piperidin-4-yl-urea series.
hCXCR3 affinities of key compounds involved in the design strategy based on the IPAG
hit
Compound
Structure
cLogDa
pKi SEMb
With the myrtenyl group having lifted the pKi to attractive val-
ues, we decided to briefly return to the nitrogen center and inves-
tigate the effect of N-substituents larger than Me. We had already
encountered steric crowding during our synthetic attempts toward
N,N-diethylation (vide supra). In line with this observation, the ste-
ric bulk imposed upon the N-atom by a Me and Et group in 8g may
shield the cation from fully interacting with the protein resulting
in a drop in pKi (compare 8f and g).
Attention was then devoted to the aromatic ‘right-hand’ side of
8f (Table 3). An iodine-scan (8f, 9a–b) established the para-posi-
tion as the preferred anchor for substitution. The advantage of a
suitable para-substituent was also evident from the inferior affinity
of benzyl-analogue (9c). A constrainment exercise on 9c by incor-
poration of a tetrahydroisoquinoline substructure (giving 9d) pro-
vided no obvious surplus value and this approach was not pursued
further. A para-halogen-scan (8f, 9e–g) revealed that all halogens
except F were tolerated. Upon further elaboration with the syn-
thetically more accessible Cl-derivatives, we found that neither
double Cl-substitution (9h) nor chain elongation (9i) improved
affinity. The data in Table 3 suggests that there is still room for
growth beyond a halogen at the para-position of the benzyl substi-
tuent. This inspired us to apply a strategic switch toward a biphe-
nyl scaffold. The affinity of 9j (pKi = 6.5) revealed that this switch
was allowed, opening the door for SAR efforts on the newly added
phenyl ring (Table 4).
H
H
N
N
6 (IPAG)
1.98
5.4 0.0
NH
I
I
I
N
H
20a
11a
2.22
2.85
5.6 0.1
5.1 0.1
N
N
8a
1.02
6.1 0.1
I
I
a
Calculated for pH 7.4 using the standard protocol of ChemAxon.
Measured by the displacement of 125I-CXCL10 binding to membranes of
b
HEK293 cells stably expressing hCXCR3 (n = 2–4, each in duplicate).
(8c). Introduction of a cyclooctenyl (8d) or a bornyl group (8e) had
no advantage. However, a significant finding was the beneficial
effect exerted by the myrtenyl group8 in compound 8f (pKi = 6.6).
Two considerations are of interest with respect to this finding.
First, the advantageous effect of a myrtenyl group over a 2-ada-
mantanylmethylene unit is opposite of what we observed with
4-amino-piperidines.17 Second, while N-methylation adds 1.1 log
unit to the tertiary base in our case (compare 11f and 8f), N-meth-
ylation on myrtenyl-urea 2 has been reported by others not to
increase the potency.8 Collectively, this evidence suggests that
Just as with the case of iodides 8f versus 9a, the preference for a
para-attachment over meta-attachment was reinforced within the
biphenyl system (9j vs 10a). The relative orientation (i.e., torsion
angle) and interactions between the biphenyl rings may also play
Table 2
hCXCR3 affinities of compounds with varying cycloaliphatic groups
R
Z
I
Y
N
Me
Y
R
Y
pKi SEMa
pKi SEMa
=
N
=
Z
Z
I
Me
11a
5.1 0.1
8a
6.1 0.1
Me
Me
Me
11b
11c
11d
n.d.b
8b
8c
8d
5.7 0.3c
6.2 0.1
6.0 0.1
5.3 0.2
5.1 0.2
Me
Me
Et
11ed
11f
5.1 0.0
5.5 0.1
5.4 0.1
8ed
8f
5.8 0.2
6.6 0.1
6.2 0.1
11g
8g
a
b
c
For assay conditions, see footnotes of Table 1.
Affinity could not be accurately measured due to poor solubility resulting from high crystallinity.
The relatively large SEM results from poor solubility due to high crystallinity.
Mixture of four stereomers.
d

M. Wijtmans et al. / Bioorg. Med. Chem. 19 (2011) 3384–3393
3389
Table 3
hCXCR3 affinities of compounds with varying aromatic groups
R
Z
Me
N
Me Me
N
R
=
Z
pKi SEMa
pKi SEMa
=
Z
I
11f
5.5 0.1
5.1 0.2
5.3 0.0
8f
6.6 0.1
5.7 0.2
5.8 0.1
I
I
12a
9a
9b
12b
I
12c
4.8 0.0
4.9 0.2
9c
5.4 0.1
5.6 0.2
12db
9d
b
12e
5.2 0.1
5.6 0.2
5.5 0.1
9e
9f
5.9 0.2
6.7 0.2
6.5 0.1
F
12f
Cl
12g
9g
Br
Cl
12h
5.5 0.1
5.3 0.1
9h
9i
6.4 0.1
6.0 0.1
Cl
12i
Cl
12j
5.5 0.1
9j
6.5 0.2
a
For assay conditions, see footnotes of Table 1.
No N-Me (12d) or only one N-Me (9d) substituent is present, as the N is part of a tetrahydroisoquinoline ring.
b
important roles. A few approaches were unleashed on 9j to inves-
tigate this further. Insertion of an O-atom to reduce inter-ring
interactions (10b) led to almost a log unit decrease in affinity. Con-
strainment of 9j and 10b (affording 10c–e) did not surpass the
affinity of 9j, although the 0.6 log unit increase from 10b to e is
of note. Clearly, the unconstrained biphenyl system lent itself best
to subsequent studies and the remainder of the efforts involved
singly linked bisaryl systems.
An array of substituents with varying characteristics was in-
stalled on the ‘outer’ phenyl ring to elucidate the SAR around this
moiety. A comparison between a 3-F and 2-F group (10f vs g) as
well as between a 3-CF3 and a 4-CF3 group (10h vs i) pointed to-
ward the meta-position as the preferred anchor. Nonetheless,
meta-substitutions with carbonyl-containing groups (10j, k) were
not of surplus value. The same held true for a selected meta,
para-disubstituted compound (10l). The fact, however, that 10f
had an affinity similar to 9j caused us to shift our attention toward
a broader set of small meta-substituents (10m–q). Significant pro-
gress was made in this area. For example, the 3-Me- and 3-NO2-
derivatives (10m, n) afforded good affinities of pKi = 6.4 and 6.7,
respectively. On the other hand, protic polar substituents such as
–OH and –NH2 (10o, p) reduced the affinity, most notably in the
case of the –NH2 group. This could be (partially) the result of a
higher enthalpic penalty for desolvation of 10o, p from an aqueous
environment. Indeed, move towards chlorine substituents
a
restored the affinity, with a meta-chloro pattern (10q) providing
an excellent affinity (pKi = 6.9). For 10q, the meta-position was
reinforced as the better anchor than the para-position (compare
10q–r). Addition of an extra chloro-group to 10q (i.e., 10s) led to
a drop in affinity. Last, two selected heteroaromatic rings were
tried (10t, u), resulting in good affinities (pKi = 6.6 and 6.7, respec-
tively) for these heteroaromatic bisaryl compounds.
Collectively, the presented affinity evidence shows the advanta-
geous effect of specific polycycloaliphatic groups on affinity for
CXCR3.17 The data also reveal that the unconstrained biphenyl-
moiety is
a versatile scaffold for SAR studies on CXCR3. A
meta-group on the outer ring of this moiety is beneficial, with an
inspection of the best compounds revealing that a non-protic,
medium-sized meta-substituent works best. Compound 10q
emerges as the best ligand, but it is fair to say that the biphenyl
moiety of the current class of ammonium salts generally exerts
subtle CXCR3 effects on overall CXCR3 affinity, both from a steric
and electronic point of view. A consistent picture that emerges is
that of a clear increase in affinity upon methylation of the tertiary
amine to a quaternary ammonium salt. This can be deduced from
three different series: 11 ? 8, 12 ? 9 and all measured examples

3390
M. Wijtmans et al. / Bioorg. Med. Chem. 19 (2011) 3384–3393
Table 4
hCXCR3 affinities of compounds with varying bisaryl groups
Ar
Z
Me
N
Me Me
= N
Ar
pKi SEMa
pKi SEMa
6.5 0.2
5.5 0.2
5.7 0.2
=
Z
Z
I
12j
5.5 0.1
n.d.d
9j
b
b
13ab
13b
10ab
10b
n.d.d
O
3
13cc
13dc
n.d.d
n.d.d
10cc
10dc
6.3 0.2
6.0 0.1
4
3
N
O
4
3
4
13ec
13f
n.d.d
10ec
10f
6.3 0.1
6.5 0.1
F
5.6 0.1
F
13g
n.d.d
10g
6.2 0.0
CF3
13h
13i
n.d.d
n.d.d
10h
10i
6.3 0.0
5.9 0.2
CF3
O
13j
n.d.d
n.d.d
10j
6.0 0.2
O
O
13k
13l
10k
10l
6.2 0.1
6.3 0.0
OMe
OMe
d
n.d
13m
13n
13o
13p
13q
n.d.d
10m
10n
10o
10p
10q
6.4 0.1
6.7 0.1
5.8 0.2
5.5 0.1
6.9 0.1
NO2
OH
NH2
Cl
5.7 0.1
n.d.d
5.2 0.1e
5.4 0.1
13r
n.d.d
10r
6.3 0.0
Cl

M. Wijtmans et al. / Bioorg. Med. Chem. 19 (2011) 3384–3393
3391
Table 4 (continued)
Me
N
Me Me
= N
Ar
pKi SEMa
pKi SEMa
=
Z
Z
I
Cl
13s
n.d.d
10s
6.4 0.1
Cl
O
13t
5.2 0.1
5.4 0.0
10t
6.6 0.1
6.7 0.2
13u
10u
S
a
For assay conditions, see footnotes of Table 1.
Ar-group on meta-position.
This is a tricyclic ‘biphenyl’-system, with the enumerations 3 and 4 for the two linkages referring to their position with respect to the aromatic carbon which is connected
b
c
to the NCH2 carbon.
d
Affinities for tertiary amine precursors, belonging to salts having a pKi <6.5, were not determined.
Although the corresponding salt 10p had a pKi <6.5, compound 13p had been prepared as a final product and was therefore tested.
e
from 13 ? 10. Such an increase not uncommonly involves more
125
than 0.9 log unit (see Fig. S1). This underscores the importance of
a quaternary cationic center in interacting with the chemokine
1
8f
receptor22 and also strongly suggests that all presented com-
100
9f
pounds bind CXCR3 by very similar means.
10q
10u
2.4. 125I-CXCL11 displacement and functional activity
75
50
25
0
A subset of compounds was inspected for additional pharmaco-
logical characteristics on hCXCR3. The set consisted of two inter-
esting members of each different subclass: 4-chlorophenyl salt
9f, 4-iodophenyl salt 8f, chlorobiphenyl salt 10q and thienylphenyl
salt 10u.
Since CXCL11 is another endogenous ligand for CXCR3, the se-
lected compounds were tested in an assay using 125I-CXCL11 with
standards being 1 (pKi = 6.8 0.2, n = 3) and CXCL11 (pKd = 10.3
0.0, n = 2). The results show that all four compounds were able to
displace 125I-CXCL11 from hCXCR3 (Table 5). The CXCR3-binding
ability of initial hit IPAG (6) was reconfirmed in this assay as well
(pKi = 5.4, Table 5).
-10
-9
-8
-7
-6
-5
Log [compound] (M)
Figure 4. Grouped functional curves for antagonistic activities of representative
compounds on hCXCR3. Footnote c of Table 5 contains assay details.
The activity (expressed as pKb values) of the compounds in a
functional assay was measured by their inhibitory effect on [3H]-
inositolphosphates levels upon stimulation of HEK293T-hCXCR3/
variations exist, the pKi and pKb values generally fall within the
same range.
Ga
qi5 cells with 50 nM CXCL10, as reported by us before.16 The re-
sults in Table 5 and Figure 4 clearly show that all four compounds,
as well as reference 1, were able to inhibit the activity of CXCL10
on CXCR3 in a concentration-dependent manner. Thus, these com-
pounds can be classified as CXCR3 antagonists. Although slight
3. Conclusion
This paper describes SAR explorations based on an initial guani-
dine hit (IPAG) found by a medium-throughput screen on hCXCR3.
Strategic replacement of the guanidine core by amine- and ammo-
nium-moieties led to two postulated guanidine surrogates. Of
these, a quaternary ammonium core clearly surpassed the affinity
of IPAG and provided a new benzyl-ammonium CXCR3-scaffold.
Versatile synthetic routes collectively provided access to ꢀ50
members, allowing efficient SAR studies on many structural fea-
tures of this scaffold (measured by inhibition of 125I-CXCL10 bind-
ing). Most notably, installing a polycycloaliphatic myrtenyl group
and expansion of the benzyl group to a (substituted) biphenyl unit
led to subtle effects in binding and provided affinities up to
ꢀ100 nM (10q). Representative members inhibited 125I-CXCL11
binding to CXCR3 and prevented activation of CXCR3 by CXCL10
in a functional assay.
Table 5
Additional pharmacological data on hCXCR3 of representative compounds
Compound
pKi,
SEMa
pKi,
SEMb
pKb SEMc
CXCL10
CXCL11
6 (IPAG)
8f
9f
10q
10u
1 (AMG487)
5.4 0.0
6.6 0.1
6.7 0.2
6.9 0.1
6.7 0.2
7.4 0.1
5.4 0.0
6.4 0.2
6.3 0.2
6.5 0.0
6.5 0.1
6.8 0.2
n.d.
6.6 0.2
6.3 0.1
6.3 0.2
6.4 0.1
8.0 0.0
a
As shown in Tables 1–4.
Measured by the displacement of 125I-CXCL11 binding to membranes of
b
HEK293 cells stably expressing hCXCR3 (n = 2–4, each in duplicate).
c
The consistent increase in affinity upon methylation of the ter-
tiary amine precursors to the quaternary ammonium salts suggests
that electrostatic interactions contribute significantly to the
Measured by the dose-dependent inhibitory effect on [3H]-inositolphosphates
levels after stimulation with 50 nM CXCL10 in HEK293T-hCXCR3/G
4, each in duplicate). n.d. = not determined.
aqi5 cells (n = 2–

3392
M. Wijtmans et al. / Bioorg. Med. Chem. 19 (2011) 3384–3393
affinities of the ammonium salts. Furthermore, our results rein-
force our previously disclosed notion that certain polycycloaliphat-
ic-anchors are beneficial for CXCR3 affinity. The described
biphenyl-scaffold may serve as a novel tool in CXCR3 research.
dried (Na2SO4), filtered and concentrated. If required, the crude
product was purified by column chromatography.
4.3. Procedure B: General procedure for Eschweiler–Clark N-
methylation
4. Experimental
The secondary amine (0.13 mmol) was heated at reflux for 2–
4 h in a mixture of formic acid (2 mL) and 37% formaline (2 mL).
The reaction mixture was allowed to cool to room temperature,
and volatiles were removed by evaporation. The crude product
was mixed with aq NaOH-solution (ca. 10 mL, 2.0 M), and ex-
tracted with DCM (3Â). The organic layers were dried (Na2SO4), fil-
tered and concentrated. If required, the product was purified by
column chromatography.
4.1. General notes on synthetic procedures
4.1.1. Chemicals
Unless reported otherwise, all chemicals were from Aldrich. Hit
compound IPAG (6) was obtained from Tocris Bioscience (HR-MS:
calcd for M+H+(C17H23IN3): 396.0931, found: 396.0939). 9-Methyl-
9H-carbazole-2-carbaldehyde was from Matrix. AMG487 (1) has
been prepared as reported by us before.25 THF, toluene and CH2Cl2
were distilled freshly from CaH2, all other solvents were used as
received.
4.4. Procedure C: General procedure for methylation to a
quaternary ammonium salt
In
a round bottom flask, the tertiary amine precursor
4.1.2. Synthesis
(0.25 mmol) was dissolved in dry DCM (6.4 mL). The flask was
placed under a nitrogen atmosphere, closed with a septum and
covered with aluminum foil. Then, iodomethane (31 lL,
Unless indicated otherwise, all reactions were carried out under
an inert atmosphere. TLC analyses were performed with Merck
F254 Alumina Silica Plates using UV visualization or staining. Col-
umn purifications were carried out manually using Silicycle Ultra
Pure Silica Gel or automatically using the BiotageÒ equipment.
Microwave reactions were carried out on a BiotageÒ Initiator. All
HR-MS spectra were recorded on Bruker micrOTOF MS using ESI
in positive ion mode. The 1H-, 13C- and 2D-NMR spectra were re-
corded on a Bruker 200, 250, 400 or 500 MHz spectrometer. Melt-
ing points were taken using the Standford Research Systems
Optimelt apparatus and values given are uncorrected. Systematic
names for molecules were generated using Chemdraw Version
11. Unless reported otherwise, all compounds have a purity
P95% as measured by LC–MS. Analytical LC–MS analyses were
conducted using a Shimadzu LC-8A preparative liquid chromato-
graph pump system with a Shimadzu SPD-10AV UV–vis detector
with the MS detection performed with a Shimadzu LC–MS-2010 li-
quid chromatograph mass spectrometer. The column used is an
0.50 mmol) was added to the solution via a syringe. The reaction
mixture was allowed to stir at room temperature overnight in
the dark. TLC analysis was used to monitor the progress of the reac-
tion. If required, additional quantities of iodomethane (1–2 equiv)
were added followed by prolonged stirring. Such cycles were ap-
plied until completion of the reaction. Next, the reaction mixture
was cooled in an ice bath and 3 volume equivalents of tBuOMe
were added to the reaction mixture slowly via a dropping funnel
while stirring. This usually caused precipitation of the salt. The pre-
cipitate was filtered and washed with a precooled solution of DCM/
tBuOMe (1:3), after which the salt was dried. If required, the salt
was recrystallised. This whole isolation procedure proved robust
and delivers the product salt in high purity (although formation
of solvates sometimes occurs). Isolated yields varied due to, for
example, varying solubilities in DCM/tBuOMe (1:3). Occasionally,
in the initial precipitation with tBuOMe the salt does not precipi-
tate as a solid but as a sticky oil. In those cases, the solvent is dec-
anted from the oil after tBuOMe addition is complete. The
remaining oil was washed with a precooled solution of DCM/
tBuOMe (1:3) by stirring, settling and decanting. All ammonium
salts were stored in the dark.
Xbridge (C18) 5
l
m column (100 mm  4.6 mm). Compound puri-
ties were calculated as the percentage peak area of the analyzed
compound by UV detection at 230 nm. The buffer mentioned is a
0.4% (w/v) NH4HCO3 solution in water, adjusted to pH 8.0 with
NH4OH.
4.1.3. Typical elution mode for quaternary ammonium salts
Solvent B (90% MeCN–10% buffer) and solvent A (90% water–
10% buffer), flow rate of 2.0 mL/min, start 5% B, linear gradient to
90% B in 10 min, then 15 min at 90% B, then 10 min at 5% B, total
run time of 35 min. The iodide counter ion is also visible by UV.
4.5. Procedure D: General procedure for reductive amination of
myrtenal
The amine (3.3 mmol) and myrtenal (19, 0.5 g, 3.3 mmol) were
mixed with dry DCE (10 mL). If the amine is a salt, TEA (460 lL,
3.3 mmol) was added to liberate the parent amine. Next,
Na(AcO)3BH (1.06 g, 5.00 mmol) was added and the mixture was
stirred at rt for 48 h. Aq NaOH-soln (2 M) was added and the mix-
ture was stirred vigorously for 10 min, after which DCM was
added. The organic layer was separated and the aq layer was
back-extracted with DCM (1Â). The combined organic layers were
dried (Na2SO4), filtered and concentrated. The crude product was
purified by column chromatography.
4.1.4. Typical elution mode for tertiary amines
Solvent B (90% MeCN–10% buffer) and solvent A (90% water–
10% buffer), flow rate of 2.0 mL/min, isocratic mode at 10% solvent
A and 90% solvent B, total run time of 30 min.
4.2. Procedure A: General procedure for reductive alkylation of
4-iodobenzylamine hydrochloride
The aldehyde (1 mmol) and 4-iodobenzylamine hydrochloride
(0.270 g, 1 mmol) were mixed with dry DCE (25 mL). If indicated,
4.6. Procedure E: General procedure for reductive alkylation of
compound 22
TEA (139 lL, 1 mmol) was added to liberate the parent amine.
Next, Na(AcO)3BH (0.318 g, 1.5 mmol) was added and the mixture
was stirred at rt for 24–36 h. Aq NaOH-soln (2 M) was added and
the mixture was stirred vigorously for 10 min, after which DCM
was added. The organic layer was separated and the aq layer was
back-extracted with DCM (1Â). The combined organic layers were
Amine 22 (0.3 g, 1.8 mmol) and the aldehyde (1.0 mmol) were
mixed with dry DCE (4 mL). Next, Na(AcO)3BH (0.53 g, 2.5 mmol)
was added and the mixture was stirred at rt for 20 h. Aq NaOH-soln
(1.0 M, ca. 10 mL) was added and the mixture was stirred vigor-
ously for 10 min, after which DCM was added. The organic layer

M. Wijtmans et al. / Bioorg. Med. Chem. 19 (2011) 3384–3393
3393
was separated and the aq layer was back-extracted with DCM (1Â).
The combined organic layers were dried (Na2SO4), filtered and con-
centrated. The crude product was purified by column
chromatography.
Saskia Hulscher, Hans Custers, Cindy van Dam, Eman Mohamed
and Benjamin Faasse for technical assistance. Chris de Graaf, Danny
Scholten and Meritxell Canals are acknowledged for helpful
discussions.
4.7. Procedure F: General Suzuki procedure
Supplementary data
A microwave vial was charged with the arylbromide (12g or 23,
300 mg, 0.90 mmol), the boronic acid (ArB(OH)2, 1.80 mmol), tolu-
Supplementary data (figure for
DpKi between quaternary
ammonium salts and tertiary amines; detailed syntheses and char-
acterization of all compounds; stability data; copies of representa-
tive 1D/2D NMR spectra and chromatogram) associated with this
article can be found, in the online version, at doi:10.1016/
j.bmc.2011.04.035.
ene (3 mL), i-PrOH (0.8 mL), and
a 2.0 M Na2CO3-solution
(0.50 mL). This mixture was degassed for 15 min by bubbling N2
through the mixture using a needle. Next, Pd(PPh3)4 (52 mg,
0.045 mmol) was added to the mixture and the vial was immedi-
ately capped and again degassed for 5 min. The vial was heated
in the microwave for 90 min at 130 °C. After this, the reaction mix-
ture was filtered over Celite and most of the organic solvents were
evaporated in vacuo. The pH of the residue was adjusted to pH >11
with a 1.0 M aq NaOH solution and the mixture was extracted with
DCM (3Â). The combined DCM layers were washed with brine,
dried over Na2SO4, filtered and concentrated in vacuo. This
afforded the crude product, which was purified by column
chromatography.
References and notes
1. Murphy, P. M.; Baggiolini, M.; Charo, I. F.; Hebert, C. A.; Horuk, R.; Matsushima,
K.; Miller, L. H.; Oppenheim, J. J.; Power, C. A. Pharmacol. Rev. 2000, 52, 145.
2. Horuk, R. Nat. Rev. Drug. Discov. 2009, 8, 23.
3. Wijtmans, M.; Verzijl, D.; Leurs, R.; de Esch, I. J.; Smit, M. J. ChemMedChem
2008, 3, 861.
4. Wijtmans, M.; de Esch, I. J. P. R. Leurs. In Chemokine Receptors as Drug Targets;
Smit, M. J., Lira, S. A., Leurs, R., Eds.; Wiley-VCH: Weinheim, 2011; pp 301–315.
5. Medina, J. C.; Johnson, M. G.; Collins, T. L. Annu. Rep. Med. Chem. 2005, 40, 215.
6. Collins, T. L.; Johnson, M. G.; Medina, J. C. In Chemokine Biology-Basic Research
and Clinical Application; Neote, K., Letts, G. L., Moser, B., Eds.; Birkhauser: Basel,
2007; pp 79–88. Volume II.
4.8. Procedures for pharmacological assays
7. Johnson, M.; Li, A.-R.; Liu, J.; Fu, Z.; Zhu, L.; Miao, S.; Wang, X.; Xu, Q.; Huang, A.;
Marcus, A.; Xu, F.; Ebsworth, K.; Sablan, E.; Danao, J.; Kumer, J.; Dairaghi, D.;
Lawrence, C.; Sullivan, T.; Tonn, G.; Schall, T.; Collins, T.; Medina, J. Bioorg. Med.
Chem. Lett. 2007, 17, 3339.
8. Allen, D. R.; Bolt, A.; Chapman, G. A.; Knight, R. L.; Meissner, J. W. G.; Owen, D.
A.; Watson, R. J. Bioorg. Med. Chem. Lett. 2007, 17, 697.
4.8.1. Cell lines
HEK293 cells stably expressing human CXCR3 were a kind gift
from Dr. Erik Boddeke en Dr. Knut Biber (University Medical Center
Groningen).
9. Hayes, M. E.; Wallace, G. A.; Grongsaard, P.; Bischoff, A.; George, D. M.; Miao,
W.; McPherson, M. J.; Stoffel, R. H.; Green, D. W.; Roth, G. P. Bioorg. Med. Chem.
Lett. 2008, 18, 1573.
10. Cole, A. G.; Stroke, I. L.; Brescia, M. R.; Simhadri, S.; Zhang, J. J.; Hussain, Z.;
Snider, M.; Haskell, C.; Ribeiro, S.; Appell, K. C.; Henderson, I.; Webb, M. L.
Bioorg. Med. Chem. Lett. 2006, 16, 200.
11. Storelli, S.; Verdijk, P.; Verzijl, D.; Timmerman, H.; van de Stolpe, A. C.; Tensen,
C. P.; Smit, M. J.; De Esch, I. J. P.; Leurs, R. Bioorg. Med. Chem. Lett. 2005, 15, 2910.
12. McGuinness, B. F.; Carroll, C. D.; Zawacki, L. G.; Dong, G.; Yang, C.; Hobbs, D.
W.; Jacob-Samuel, B.; Hall, J. W., 3rd; Jenh, C. H.; Kozlowski, J. A.; Anilkumar, G.
N.; Rosenblum, S. B. Bioorg. Med. Chem. Lett. 2009, 19, 5205.
13. Pradelli, E.; Karimdjee-Soilihi, B.; Michiels, J. F.; Ricci, J. E.; Millet, M. A.;
Vandenbos, F.; Sullivan, T. J.; Collins, T. L.; Johnson, M. G.; Medina, J. C.;
Kleinerman, E. S.; Schmid-Alliana, A.; Schmid-Antomarchi, H. Int. J. Cancer
2009, 125, 2586.
4.8.2. Library screen using 125I-CXCL10
The binding experiment was performed as described before,16
with the exception that membranes were prepared from HEK293
cells stably expressing human CXCR3.26 10
was used.
lM of test compound
4.8.3. Affinity measurements on hCXCR3 using 125I-CXCL10
pKi values at hCXCR3 were determined as described before,16
with the exception that membranes were prepared from HEK293
cells stably expressing human CXCR3.26
14. Cambien, B.; Karimdjee, B. F.; Richard-Fiardo, P.; Bziouech, H.; Barthel, R.;
Millet, M. A.; Martini, V.; Birnbaum, D.; Scoazec, J. Y.; Abello, J.; Al Saati, T.;
Johnson, M. G.; Sullivan, T. J.; Medina, J. C.; Collins, T. L.; Schmid-Alliana, A.;
Schmid-Antomarchi, H. Br. J. Cancer 2009, 100, 1755.
15. Halloran, P. F.; Fairchild, R. L. Am. J. Transplant. 2008, 8, 1578.
16. Verzijl, D.; Storelli, S.; Scholten, D.; Bosch, L.; Reinhart, T. A.; Streblow, D. N.;
Tensen, C. P.; Fitzsimons, C. P.; Zaman, G. J. R.; Pease, J. E.; De Esch, I. J. P.; Smit,
M. J.; Leurs, R. J. Pharmacol. Exp. Ther. 2008, 325, 544.
17. Wijtmans, M.; Verzijl, D.; van Dam, C. M.; Bosch, L.; Smit, M. J.; Leurs, R.; de
Esch, I. J. Bioorg. Med. Chem. Lett. 2009, 19, 2252.
18. Whittemore, E. R.; Ilyin, V. I.; Woodward, R. M. J. Pharmacol. Exp. Ther. 1997,
282, 326.
4.8.4. Affinity measurements on hCXCR3 using 125I-CXCL11
pKi values at hCXCR3 were determined as described before,16
with the exception that membranes were prepared from HEK293
cells stably expressing human CXCR3.26
4.8.5. Functional activity through phospholipase C (PLC)
activation
The pKb values were determined as described before.16
19. Wilson, A. A.; Dannals, R. F.; Ravert, H. T.; Sonders, M. S.; Weber, E.; Wagner, H.
N., Jr. J. Med. Chem. 1991, 34, 1867.
4.9. Data analysis
20. AbdelMagid, A. F.; Carson, K. G.; Harris, B. D.; Maryanoff, C. A.; Shah, R. D. J. Org.
Chem. 1996, 61, 3849.
Non-linear regression analysis of the data and calculation of pKd
and pKi values was performed using Prism 4.03. The pKb values in
the PLC activation assay were calculated using the Cheng–Prusoff
equation pKb = IC50/(1 + [agonist]/EC50).
21. Wang, G.; Li, C.; Li, J.; Jia, X. Tetrahedron Lett. 2009, 50, 1438.
22. Dölker, N.; Deupi, X.; Pardo, L.; Campillo, M. Theor. Chem. Acc. 2007, 118, 579.
23. Salonen, L. M.; Bucher, C.; Banner, D. W.; Haap, W.; Mary, J. L.; Benz, J.; Kuster,
O.; Seiler, P.; Schweizer, W. B.; Diederich, F. Angew. Chem., Int. Ed. 2009, 48, 811.
24. Hughes, R. M.; Benshoff, M. L.; Waters, M. L. Chem. Eur. J. 2007, 13, 5753.
25. Storelli, S.; Verzijl, D.; Al-Badie, J.; Elders, N.; Bosch, L.; Timmerman, H.; Smit,
M. J.; De Esch, I. J.; Leurs, R. Arch. Pharm. (Weinheim) 2007, 340, 281.
26. Dijkstra, I. M.; Hulshof, S.; van der Valk, P.; Boddeke, H. W.; Biber, K. J. Immunol.
2004, 172, 2744.
Acknowledgments
This work was financially supported by the Dutch Top Institute
Pharma (project number D1.105: the GPCR Forum). We thank