Insight into structural requirements for selective and/or dual CXCR3 and CXCR4 allosteric modulators

Article abstract of DOI:10.1016/j.ejmech.2018.05.013

Based on the previously published pyrazolopyridine-based hit compound for which negative allosteric modulation of both CXCR3 and CXCR4 receptors was disclosed, we designed, synthesized and biologically evaluated a set of novel, not only negative, but also positive allosteric modulators with preserved pyrazolopyridine core. Compound 9e is a dual negative modulator, inhibiting G protein activity of both receptors. For CXCR4 receptor para-substituted aromatic group of compounds distinguishes between negative and positive modulation. Para-methoxy substitution leads to functional antagonism, while para-chloro triggers agonism. Additionally, we discovered that chemotaxis is not completely correlated with G protein pathways. This is the first work in which we have on a series of compounds successfully demonstrated that it is possible to produce selective as well as dual-acting modulators of chemokine receptors, which is very promising for future research in the field of discovery of selective or dual modulators of chemokine receptors.

Full text of DOI:10.1016/j.ejmech.2018.05.013

Accepted Manuscript
Insight into structural requirements for selective and/or dual CXCR3 and CXCR4
allosteric modulators
Anja Kolarič, Urban Švajger, Tihomir Tomašič, Regine Brox, Theresa Frank, Nikola
Minovski, Nuska Tschammer, Marko Anderluh
PII:
S0223-5234(18)30417-3
10.1016/j.ejmech.2018.05.013
EJMECH 10424
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 10 January 2018
Revised Date: 18 April 2018
Accepted Date: 9 May 2018
Please cite this article as: A. Kolarič, U. Švajger, T. Tomašič, R. Brox, T. Frank, N. Minovski, N.
Tschammer, M. Anderluh, Insight into structural requirements for selective and/or dual CXCR3
and CXCR4 allosteric modulators, European Journal of Medicinal Chemistry (2018), doi: 10.1016/
j.ejmech.2018.05.013.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Insight into Structural Requirements for Selective
and/or Dual CXCR3 and CXCR4 Allosteric
Modulators
a,b
c
a
d
d
Anja Kolarič , Urban Švajger , Tihomir Tomašič , Regine Brox , Theresa Frank , Nikola
b
d,1
a,
Minovski , Nuska Tschammer and Marko Anderluh *
a
Faculty of Pharmacy, University of Ljubljana, Aškerčeva 7, 1000 Ljubljana, Slovenia
b
National Institute of Chemistry, Ljubljana, Slovenia
c
Blood Transfusion Center of Slovenia, Šlajmerjeva 6, 1000 Ljubljana, Slovenia
d
Department of Chemistry and Pharmacy, Medicinal Chemistry, Emil Fischer Center, Friedrich-
Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany
Corresponding author:
*
1
Prof. Dr. Marko Anderluh, University of Ljubljana, Faculty of Pharmacy, Aškerčeva 7, SI –
000 Ljubljana, Slovenia; Tel.: +386 1 476 96 39; fax: +386 1 425 80 31.
1
NanoTemper Technologies, Flößergasse 4, 81369 München, Germany.
Abbreviations: CXCR3, CXC chemokine receptor 3; CXCR4, CXC chemokine receptor 4;
CXCL9, CXC chemokine ligand 9; CXCL10, CXC chemokine ligand 10; CXCL11, CXC
chemokine ligand 11; CXCL12, CXC chemokine ligand 12; CD4, immune cells (cluster of
differentiation 4); Boc, tert-butyloxycarbonyl; IFN-γ, interferon gamma; Th, helper T cell;
CD25, interleukin 2 receptor alpha; FITC, Fluorescein isothiocyanate; SDF-1, stromal cell-
derived factor 1.

ACCEPTED MANUSCRIPT
ABSTRACT
Based on the previously published pyrazolopyridine-based hit compound for which negative
allosteric modulation of both CXCR3 and CXCR4 receptors was disclosed, we designed,
synthesized and biologically evaluated a set of novel, not only negative, but also positive
allosteric modulators with preserved pyrazolopyridine core. Compound 9e is a dual negative
modulator, inhibiting G protein activity of both receptors. For CXCR4 receptor para-substituted
aromatic group of compounds distinguishes between negative and positive modulation. Para-
methoxy substitution leads to functional antagonism, while para-chloro triggers agonism.
Additionally, we discovered that chemotaxis is not completely correlated with G protein
pathways. This is the first work in which we have on a series of compounds successfully
demonstrated that it is possible to produce selective as well as dual-acting modulators of
chemokine receptors, which is very promising for future research in the field of discovery of
selective or dual modulators of chemokine receptors.
Keywords: chemokine receptors, CXCR3, CXCR4, allosteric modulators
2

ACCEPTED MANUSCRIPT
1
. Introduction
G protein – coupled receptors (GPCRs) orchestrate diverse biological functions.
Numerous diseases and disorders are associated with the malfunction in their signaling, which
makes them interesting and promising drug targets [1]. Approximately 30% of marketed drugs
target GPCRs [2]. Historically, drug discovery efforts in the field of GPCRs were oriented in
targeting GPCRs orthosteric binding site and thus competing with the endogenous ligands.
Today the focus shifted to the targeting of allosteric sites of GPCRs, which are distinct from the
orthosteric site where the endogenous ligands bind. Addressing allosteric sites has several
advantages over targeting orthosteric sites: a) they are less conserved among related GPCRs and
thus offer the potential for the design of highly selective synthetic ligands and b) because they do
not compete with the orthosteric ligand, the effects like probe dependence and insurmountable
effects are observed. In general, allosteric ligands can act in three different ways; positive
allosteric modulators enhance the action, while negative allosteric modulators decrease the
affinity and/or efficacy of the orthosteric agonist. Thirdly, and neutral allosteric ligands do not
affect affinity and/or efficacy of the orthosteric agonist, but prevent binding of other allosteric
modulators [3,4]. Additionally, the phenomenon of biased signaling complicates the outcome of
GPCRs modulation; biased signaling is the ability of ligands to stabilize different conformations
of GPCRs, resulting in the stimulation of selected (but not all) signaling pathways that can be
activated by the receptor and thus evoking different functional outcomes [5,6]. Therefore,
allosteric modulators provide a range of novel opportunities in modulation of GPCR function.
Especially in the case of chemokine receptors that are the key regulators of our immune
3

ACCEPTED MANUSCRIPT
response, the development of allosteric drugs like maraviroc (allosteric inhibitor of the CCR5
receptor) indicates the power of this approach [7].
Chemokine receptors are representatives of family A of GPCRs [8]. Endogenous ligands
for these receptors are chemokines (chemoattractant cytokines), small proteins playing essential
role in the development, homeostasis, and function of the immune system [9]. Among
chemokine receptors, CXCR3 is mostly expressed on activated Th1 cells, but is also located on
circulating blood T cells, B cells and natural killer cells [10-12]. Three distinct CXCR3 isoforms
have been identified, including CXCR3-A, CXCR3-B and less known alternative CXCR3-alt
[13, 14]. Receptor’s endogenous ligands are CXCL9, CXCL10 and CXCL11, of which CXCL11
is the dominant ligand followed by CXCL10 [15]. Additionally, endogenous CXCL4 was
discovered as a high affinity ligand for CXCR3-B isoform. [13].On the other hand, CXCR4
receptor is expressed on several immune cells like monocytes, B cells and naive T cells, and
binds endogenous chemokine CXCL12 only [15,16]. Overexpression of CXCR3 and its
endogenous ligands has been found to play an important role in a variety of inflammatory
diseases including autoimmune diseases (e.g., multiple sclerosis, psoriasis, and rheumatoid
arthritis) and transplant rejection [15]. It is also responsible for the promotion of melanoma,
osteosarcoma and for metastasis of colon, breast, lymph nodes and lung metastasis. Likewise,
CXCR4 is often overexpressed in various cancers (e.g., breast cancer, prostate cancer, ovarian
cancer and melanoma) [17] and serves as a co-entry receptor for HIV virus [18]. CXCR3 and
CXCR4 co-involvement in metastasis of colorectal cancer into other tissues, especially lymph
nodes, liver and lungs has been discovered and accordingly higher pharmacological effect
obtained by blocking both receptors instead of antagonism of each individual receptor was
4

ACCEPTED MANUSCRIPT
proven for treatment of colorectal cancer [19]. The two receptors act synergistically, meaning
that activation of CXCR3 strengthens binding of CXCL12 to CXCR4 [19]. All these findings
lead to the conclusion that compounds that would bind and inhibit the function of both receptors
thereby acting as non-selective dual CXCR3 and CXCR4 antagonists/negative allosteric
modulators would have interesting potential in treatment of the aforementioned diseases.
Structurally diverse small-molecule ligands with ability to individually modulate the
function of either CXCR3 or CXCR4 have been disclosed including (aza)quinazolinones
(AMG487, Figure 1), piperazinylpiperidines (VUF11211, Figure 1), pyrido[1,2-a]pyrimidin-4-
ones, 1-aryl-3-piperidin-4-yl-ureas, 4-N-aryl[1,4]diazepanylureas, 2-iminobenzimidazoles,
bispiperidines and ergolines for CXCR3 [20,21]. These ligands are presumed to be allosteric
modulators [22]. AMG487 (Figure 1), the most studied member of azaquinazolines, was the only
reported CXCR3 small molecule antagonist that progressed to Phase II clinical trials for
psoriasis, but failed due to lack of efficacy [23]. Different structural classes of CXCR4 small-
molecule antagonists have been designed, but only a few of them have been studied in detail:
tetrahydroquinolines (AMD070, Figure 1), N-substituted indoles, 1,4-phenylenebis(methylene)
derivatives and N-containing heterocycles are among them [24]. AMD070 (Figure 1) is one of
the most known representatives of tetrahydroquinolines, an orally active CXCR4 negative
allosteric modulator for the prevention of T-tropic HIV infection [25-27].
5

ACCEPTED MANUSCRIPT
Figure 1. Known small molecule CXCR3 and CXCR4 antagonists/allosteric modulators.
Despite of sharing 36.7% and 63.9% sequence identity and similarity, respectively, small
molecule ligand chemical spaces of CXCR3 and CXCR4 differ markedly. Using ChEMBL22
(version 17), 858 and 484 tested ligands were found for CXCR3 and CXCR4, respectively, but
unfortunately no of it seems to be active on the opposite receptor at a threshold of 10 mΜ [28].
Schmidt et al. described the hit compound 7h (re-synthesized, Figure 2) as a dual negative
allosteric modulator, potently inhibiting the activation of CXCR3 and CXCR4 with the pKB
3
5
value of 8.82 and 8.72, respectively, as determined in the [ S]GTPγS assay [28]. The
cooperativity factor αβ of the hit compound 7h was 0.12 and 0.04 for CXCR3 and CXCR4,
respectively, clearly demonstrating that this compound acts as a negative allosteric modulator.
Starting from this dual negative allosteric modulator, our goal was to design and
synthesize a series of novel potent dual allosteric modulators of CXCR3 and CXCR4. With the
35
intention to avoid the limitations of [ S]GTPγS assay, which is performed with isolated
membrane preparations of cells expressing a given receptor, we decided to characterize new
series of compound in a whole cell-based in vitro assays using transiently transfected HEK293
and primary peripheral blood mononuclear cells (PBMCs) or CD4+ T cells. Predicted binding
poses and interactions identified as crucial for negative allosteric modulation studied by
6

ACCEPTED MANUSCRIPT
molecular docking gave us a new insight into structure-activity relationship (SAR) providing
progress in the on-going design and optimization of novel small molecule dual CXCR3 and
CXCR4 negative allosteric modulators.
7

ACCEPTED MANUSCRIPT
2
. Results and discussion
2
.1.Design
By examining the structural features of known small molecule CXCR3 and CXCR4
antagonists (Figure 1), we can conclude that in general, small molecule ligands consist of
aromatic heterocyclic system connected through a cyclic or linear alkyl linker to a lipophilic
moiety of one or two substituted or un-substituted aromatic/heterocyclic rings. Our hit compound
(7h) [28] is in accordance with these findings, where pyrazolopyridine scaffold represents
aromatic heterocyclic system connected with 1,3-diaminopropane linker to lipophilic moiety
represented by benzene and ethyl group as depicted in Figure 2.
Figure 2. Investigation of structural features of our hit compound (7h) discovered by Schmidt et
al. [28] and design of novel compounds.
We have designed a series of compounds with preserved pyrazolopyridine scaffold,
connected with different (cyclic and linear) linkers to lipophilic moieties. A previous study of
Schmidt et al. [28] anticipated optimal interactions of pyrazolopyridine scaffold with receptor,
8

ACCEPTED MANUSCRIPT
which was the reason for its preservation. As presented in Figure 2, we have introduced four
different linkers including linear 1,3-diaminopropane, cyclic 4-aminopiperidine (in two different
orientations) and piperazine linker, in which we varied flexibility, length and influence of the
cationic center. A comparison of the properties of ligands between CXCR3 and CXCR4 shows
that CXCR4 tends to bind ligands with a stronger positive charge compared to CXCR3 [28].
Also, these ligands differ in the distribution and the number of hydrogen bond donors and the
predicted lipophilicity. Our prediction was that modification of the linker drives selectivity and
influences biased signaling of individual or even both receptors. According to the structure of
known small molecule ligands (Figure 1), it seems that those acting on CXCR3 have more linear
shape, while those acting on CXCR4 have two lipophilic parts that point in different directions.
Therefore, we connected linker to one or two aromatic lipophilic moieties. To provide stronger
interactions with receptor, the influence of substitution pattern was determined by introduction of
one or two substituents with different electronic effects.
Complete novel series of compounds was characterized in detail in the whole-cell based
assays, including their ability to elicit biased signaling at the receptors of interest. The
PathHunter β-arrestin 2 recruitment assay (DiscoverX) detects the recruitment of β-arrestin 2 to
activated receptor by complementing two β-galactosidase (β-gal) enzyme fragments, while
BRET (bioluminescence resonance energy transfer) based cAMP assay measures the
concentration of cAMP. Moreover, the transwell migration assay using PBMC or CD4+ T cells
was employed to measure chemokine induced cell migration.
2
.2. Chemistry
9

ACCEPTED MANUSCRIPT
Synthetic strategy for construction of the pyrazolopyridine scaffold 2 is described in
Scheme 1. Commercially available pyrazole and ethyl 6-chloronicotinate reacted in the presence
of sodium hydride in anhydrous N,N-dimethylformamide (DMF) under inert atmosphere at room
temperature to give ester 1, which was hydrolyzed to carboxylic acid 2 upon treatment with 3 M
sodium hydroxide in ethanol.
a
Scheme 1. Synthesis of pyrazolopyridine scaffold.
a
Reagents and conditions: (i) NaH (60% in mineral oil), DMF, rt; (ii) 3M NaOH, EtOH, rt.
Synthesis of the derivatives with 1,3-diaminopropane linker (Scheme 2) started with the
formation of compound 3 from phthalimide potassium salt and 1-bromo-3-chloropropane in
anhydrous DMF at 70 °C. N-Alkylation with various aniline derivatives (aniline, 4-chloroaniline
and 4-methoxyaniline) in DMF at 70 °C afforded compounds 4a, 4b and 4c respectively.
Following deprotection of the phthalimide group with hydrazine hydrate in DMF at 90 °C gave
the desired amines 5a, 5b and 5c. In the next step coupling of 2 with 5a, 5b or 5c using 1-ethyl-
3
-(3-dimethylaminoisopropyl) carbodiimide (EDC), 1-hydroxybenzotriazole (HOBt) and
triethylamine (Et N) in dry DMF at 0 °C provided 6a, 6b and 6c. Only compound 6a was used
3
for further synthesis of the derivatives 7a-7h with substituted 1,3-diaminopropane linker.
Various reagents were used for N-alkylation with catalytic amount of benzyltriethylammonium
10

ACCEPTED MANUSCRIPT
chloride (BTEAC) and potassium iodide in acetonitrile under inert atmosphere at 60 °C to form
final compounds 7a-7g. Ester of 7f was hydrolyzed to carboxylic acid 7i with 3 M sodium
hydroxide in ethanol. Compound 7h was synthesized in microwave reactor (100 °C, 15 bar, 20
min) using BTEAC and potassium carbonate.
a
Scheme 2. Synthesis of the compounds with 1,3-diaminopropane linker.
11

ACCEPTED MANUSCRIPT
a
2 3
Reagents and conditions: (iii) 1-bromo-3-chloropropane, DMF, 70 °C; (iv) K CO , DMF, 60
°
C; (v) H NNH × H O, EtOH, 90 °C; (vi) HOBt, NMM, EDC × HCl, DMF, 0 °C; (vii) BTEAC,
2 2 2
K CO , KI, CH CN, 60 °C; (viii) bromoethane, BTEAC, K CO , CH CN, MW: 100 °C, 20 min.
2
3
3
2
3
3
The synthetic strategy for compounds with piperazine (8-9e) and 4-aminopiperidine (10-
3f) linker is shown in Scheme 3. The first step was the same for all three series of compounds;
coupling of 2 with three different linkers (1-Boc piperazine, 4-amino-1-Boc-piperidine and 4-(N-
Boc-amino)piperidine) using O-(benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium
1
tetrafluoroborate (TBTU) in the mixture of DCM and DMF (1:1) in the presence of N-
methylmorpholine (NMM) under inert atmosphere at room temperature. Removal of Boc
protection by TFA was performed in dry DCM at 40 °C in inert atmosphere to give the desired
amines 8, 10 and 12. Final compounds 9a-9e and 11a-11d were formed by N-alkylation in
acetonitrile with various benzyl halides and potassium carbonate at 55 °C. The problem of
reaction selectivity has been encountered in the synthesis of 13a, where we first tried to achieve
monosubstitution with N-alkylation in the presence of BTEAC, potassium carbonate, potassium
iodide in acetonitrile at 55 °C, but the major final product was always disubstituted 13f,
independent of equivalents of aryl halide used in the reaction. Reductive amination using sodium
cyanoborohydride gave monosubstituted 13a.
a
Scheme 3. Synthesis of the compounds with 4-aminopiperidine and piperazine linker.
12

ACCEPTED MANUSCRIPT
13

ACCEPTED MANUSCRIPT
a
Reagents and conditions: (ix) TBTU, NMM, DMF, DKM, r.t; (x) 1. TFA, DKM, 40 °C; (xi)
various benzyl halides, K CO , CH CN, 55 °C; (xii) 3,4-difluorobenzaldehyde, CH COOH,
2
3
3
3
NaCNBH , MeOH, r.t.; (xiii) benzyl chloride, BTEAC, K CO , KI, CH CN, 55 °C; (xiv)
3
2
3
3
benzaldehyde, CH COOH, NaCNBH , MeOH, r.t; (xv) various benzyl halides, BTEAC, K CO ,
3
3
2
3
KI, CH CN, 80 °C.
3
2
.3. Functional characterization
Upon binding of the endogenous chemokines to CXCR3 and CXCR4, the G-protein
dependent and independent signaling pathways are activated. Both receptors are the G -coupled
αi
receptors, thus the G protein-dependent signaling can be monitored as a decrease in the
intracellular cAMP level [29]. In the G protein-independent signaling, β-arrestin 2 is involved
[30]. The activation of these pathways leads to the chemotaxis of the cells expressing given
receptors.
14

ACCEPTED MANUSCRIPT
Figure 3. Graphical illustration of functional assays used to study the binding of novel
compounds and their ability to modulate chemokine mediated activation of CXCR3 and CXCR4.
In β-arrestin 2 recruitment assay, all novel compounds were assessed for their ability to
suppress chemokine mediated β-arrestin 2 recruitment to activated CXCR3 receptor. The
compounds 7a, 7h, 9e, 13b, 13d, 13e and 13f were able to suppress CXCL11-mediated β-
arrestin recruitment for at least 20%, which was chosen as a limit (Figure 4). Known antagonist
(±)-NBI-74330 (for structure see Figure S1 in Supplementary data) [31], which served as a
control compound, could fully inhibit β-arrestin 2 recruitment at 10 µM. Concise determination
of the IC values for the novel compounds was unfortunately not possible due to small assay
5
0
windows, resulting only up to 30% of inhibition of CXCL11-mediated β-arrestin 2 recruitment.
15

ACCEPTED MANUSCRIPT
Figure 4. The ability of the tested compounds at 10 µM to inhibit the β-arrestin 2 recruitment,
representing the percents of chemiluminescence in comparison with CXCL11 (100% activation)
at 50 nM and (±)-NBI-74330 at 10 µM (full inhibition) [31] in single point β-arrestin 2
recruitment assay. The values represent the mean ± S.E.M. of three experiments performed in
triplicate.
All the compounds were also screened for agonism, the ability to activate CXCR3 and
cause β-arrestin 2 recruitment in the absence of the parent chemokine, and none was found to
possess direct agonistic activity (Figure S3 in Supplementary data).
Our efforts to detect the CXCL12-mediated recruitment of the β-arrestin2 to CXCR4
failed for unknown reasons, although the expression of receptor was adequate (Figure S2 in
Supplementary data). The detailed testing of various CXCR4 constructs containing either
ProLink 1 (PK1) or ProLink 2 (PK2) (PathHunter assay, DiscoverX) remained without success,
however we believe that the literature underscores the importance of β-arrestin 2 in CXCR4
signaling and further efforts will be needed to elucidate it [32,33].
16

ACCEPTED MANUSCRIPT
Next, we investigated the ability of our novel compounds to reverse the chemokine
mediated modulation of adenylate cyclases, which are the main effectors of CXCR3 and CXCR4
coupled G -proteins [34]. Because coupling of CXCR3 and CXCR4 receptor to G leads to a
i
αi
negative regulation of the adenylate cyclase, which decreases cAMP levels, the adenylate cyclase
has to be pre-stimulated with forskolin. Therefore, the measured activity actually represents
inhibition of forskolin-stimulated cAMP production. All of the compounds were first screened in
a single point BRET based cAMP experiment. Compounds 9b, 9e, 11a, 11e and 13c inhibited
the CXCL11 induced CXCR3 activation for more than 50% (Figure 5).
Figure 5. The chart of the tested compounds at 10 µM for CXCR3 antagonism in BRET based
cAMP assay, representing the percents of netBRET signal in comparison with CXCL11 at 50
nM (100%) and cRAMX3 at 10 µM (full inhibition, receptor activation reduced to 0%) [35]. The
values represent the mean ± S.E.M. of three experiments performed in triplicate.
Surprisingly, the only compound that appeared active in both assays (β-arrestin 2
recruitment and BRET based cAMP assay) on CXCR3 is compound 9e. Compounds 9b and 9e
with best efficacy of more than 60%, were further tested in dose-response assay (Figure 6). For
17

ACCEPTED MANUSCRIPT
further characterization of allosteric profile for these novel compounds, the data of dose-response
assays were analyzed with a ternary complex model of allosterism, using the algorithms in Prism
5
.0 (GraphPad Software, San Diego, CA, USA) to determine pK and α. pK measures the
B B
affinity of allosteric modulator for its allosteric binding site and α denotes the cooperativity
between allosteric and orthosteric ligand and it characterizes the affinity of orthosteric ligand in
the presence of allosteric modulator [5]. Values of α > 1 denote positive cooperativity (increase
in orthosteric ligand binding), whereas α < 1 denotes negative cooperativity (reduction of
orthosteric ligand binding). α = 0 cannot be distinguishable from competitive antagonism and K
B
value approaches K value. If α = 1, the result is unaltered ligand affinity [5].
i
18

ACCEPTED MANUSCRIPT
Figure 6. The ability of novel negative allosteric modulators 9b and 9e to inhibit the CXCL11
(50 nM) mediated CXCR3 G protein activation in BRET based cAMP assay with representative
dose-effect curves. The dose-dependent receptor activity is plotted for known allosteric
antagonist (cRAMX3) as a reference [35]. The values represent the mean ± S.E.M. of three
experiments performed in triplicate.
pK values of both compounds are comparable (pK = 6.35 ± 0.10 for 9b and pK = 6.15
B
B
B
±
0.15 for 9e) and are one order of magnitude weaker than the known allosteric antagonist
cRAMX3 (Figure 6) [35]. We assume that both 9b and 9e act as negative allosteric modulators,
19

ACCEPTED MANUSCRIPT
with the α value equaling zero marring their behavior is undistinguishable from competitive
antagonism. While compound 9b was not identified with β-arrestin 2 recruitment assay as a
promising inhibitor of recruitment, but it was significant negative allosteric modulator of
CXCR3 in the G protein-dependent pathway, we assume that it is capable of biased CXCR3
signaling. Comparing the structures of both compounds, the only difference lies in the
substitution of benzyl moiety; para-chloro for 9b and para-methoxy for 9e. The substituents
differ in electronic effects, but mostly in H-bonding acceptor ability of the methoxy group, which
indicates that a single H-bond could be the determining factor for biased signaling. To confirm
biased CXCR3 signaling of compound 9b, Student t-test was performed. Unfortunatelly no
significant difference was observed between compounds 9b and 9e, therefore 9b can not be
considered as biased. Dose-effect curves also indicate that compounds could fully inhibit
activation of G proteins. None of the compounds demonstrated intrinsic agonist properties when
i
tested alone (Figure S4 in Supplementary data).
On CXCR4, the compounds showed significantly different profile. For CXCR4, the
ability of novel compounds to inhibit CXCL12-mediated modulation of cAMP production was
monitored in a single point experiment with the known inhibitor IT1t (Figure 1) as a control [36].
Compounds 7a, 7g, 9c, 9e and 13e inhibited CXCL12 induced CXCR4 activation for more than
5
0% (Figure 7B). Strikingly, several compounds (13a being the most potent) enhanced the
efficacy of CXCL12, which suggests that these compounds act as positive modulators of CXCR4
Figure 7B).
(
20

ACCEPTED MANUSCRIPT
Figure 7. (A) The results of CXCR4 agonism screening of compounds at 10 µM with CXCL12
at 20 nM as a positive control. (B) The chart of the tested compounds at 10 µM for CXCR4
antagonism in BRET based cAMP assay, representing the percents of netBRET signal in
comparison with CXCL12 at 20 nM (100%) and IT1t at 10 µM (full inhibition, receptor
activation reduced to 0%) [36]. The values represent the mean ± S.E.M. of three experiments
performed in triplicate.
To determine potential intrinsic agonist properties of these compounds, the compounds
were tested for direct agonism in the cAMP assay. The compounds 6a, 6b and 11a clearly acted
as agonists, meaning they activated the receptor in the absence of the chemokine (Figure 7A).
21

ACCEPTED MANUSCRIPT
For the further characterization of allosteric profile for these promising compounds, the
activity profile of these compounds was characterized in the presence of CXCL12 and the data
were analyzed with a ternary complex model of allosterism [5].
22

ACCEPTED MANUSCRIPT
23

ACCEPTED MANUSCRIPT
Figure 8. The ability of novel negative modulators 7a, 7g, 9c, 9e and 13e to inhibit and positive
modulator 13a to activate the CXCL12 (20 nM) mediated CXCR4 protein activation in BRET
based cAMP assay with representative dose-effect curves. Compounds 6a, 6b and 11a, showing
direct agonistic effect were also tested for allosteric profile. The dose-dependent receptor activity
is plotted for known antagonist (IT1t) as a reference [36]. The values represent the mean ±
S.E.M. of three experiments performed in triplicate.
Compounds 7a, 7g, 9c and 9e have comparable affinities (Figure 8). Closer look into the
structure-activity relationship revealed that cationic center is not relevant for activity. Two types
of compounds seem to favor inhibition of chemokine induced activation of G protein;
compounds with longer, flexible linker, connected to two unsubstituted aromatic rings, such as
7
9
a and 7g, and compounds with short, rigid linker and substituted aromatic group, such as 9c and
e. This is supported by 13e where additional substituted aromatic group on more rigid linker
decreased the activity. Compounds 6a, 6b, 11a and 13a with one aromatic group act as agonists,
regardless of linker’s properties. Mono-substitution pattern seems to play the major role as a
molecular switch between agonistic and antagonistic behavior having in mind that compounds
most likely act as modulators and not direct agonists/antagonists. Para-Chloro substitution seems
to favor agonism (6b and 11a), while para-methoxy favors antagonism (9e and 13e).
Disubstitution seems to exhibit weak negative cooperativity, by which the affinity of chemokine
is poorly decreased and results in the lack of efficacy as in 9c. Only 7g displays strong negative
cooperativity, while the cooperativity of 7a, 9e and 13e cannot be distinguished from
competitive antagonism. Therefore, only 9c and 7g can be determined as allosteric modulators,
while for the rest we can only assume allosteric binding. Compounds 6a, 6b and 11a could
24

ACCEPTED MANUSCRIPT
partially enhance the activity of CXCL12 on CXCR4 and thus, positive allosteric agonists with
intrinsic agonist activity, which is observable from positive cooperativity value α of compounds
6
b and 11a. In terms of agonism, none of the compounds acted as allosteric agonist on CXCR3,
but only on CXCR4. From dose-effect curves of compounds 7a, 9e and 13e, we can conclude
that the compounds would be able to fully inhibit activation of G protein and act as good
i
negative allosteric modulators of chemokine mediated activation. Due to poor signal-to-noise
ratio, the data obtained for compounds 6a and 13a could be fitted only by non-linear regression
curve fixing the Hills slope at 1 to obtain approximate pEC50 values.
To confirm negative modulatory effect of novel compounds in an ex vivo assay, we
examined the influence of compounds 9b, 9e, 11a, 11e and 13c for CXCR3 and compounds 7a,
7
g, 9c, 9e and 13e for CXCR4 mediated migration of PBMC or CD4+ T with transwell
migration assay. The activation of CXCR3 has been shown to induce migration by activation of
G through multiple pathways, including phosphoinositide-3 kinase (PI3K) [37]. For CXCR4 it
αi
was published that G subunit interacting with ion channels is responsible for activation of
βγ
PI3K, which leads to chemotaxis [38]. Till now no direct studies on CXCR3 migration induced
by β-arrestin 2 have been published [39]. However, Fong and co-workers pointed out the
involvement of β-arrestin 2 in chemotaxis regulation [40].
Two types of cells were used for evaluation of compounds influence on chemokine
induced migration; PBMCs in the case of CXCR3 receptor [41] and CD4+ T cells for CXCR4
[42]. Firstly, phenotypic characterization was done, using flow cytometric analysis in order to
confirm sufficient expression of CXCR3 and CXCR4 receptors on selected cells. Moreover, the
cells were pre-treated with selected compounds afterwards to determine the negative modulatory
25

ACCEPTED MANUSCRIPT
effect. All investigated compounds (9b, 9e, 11a, 11e and 13c) were able to decrease CXCL11
induced migration of cells expressing CXCR3 (Figure 9A). At lower concentration (30 µM) the
decrease was more than 10% and at higher concentration (150 µM) more than 20% compared to
control. With approximately 80% decreased migration, 11a was the most potent migration
inhibitor.
Figure 9. Inhibition of cell migration using transwell assay. Selected compounds were tested for
their capacity to inhibit migration of human PBMCs from upper to lower chamber in a transwell
+
system, as described in Experimental. Briefly, PBMCs (A) or isolated CD4 T cells (B) were
either left untreated (black) or pre-treated with CXCR3 (A) or CXCR4 (B) antagonists and their
migratory capacity toward endogenous chemokine ligands for CXCR3 (CXCL11) or CXCR4
(CXCL12) was evaluated. After the incubation period, the migrated cells from the lower
chamber were thoroughly collected and counted on a flow cytometer using 60s counts. Using bar
graphs, mean ± SD is shown from four independent experiments. Statistical significance between
individual pairs (migration of untreated vs. treated with all individual compounds) was
26

ACCEPTED MANUSCRIPT
calculated using Student’s unpaired t test (ns – non-significant; p<0.05 - *; p<0.01 - **; p<0.001
-
***).
The compounds 7a, 7g, 9c, 9e and 13e were tested for their ability to suppress the
CXCL12 induced migration of CXCR4-expressing cells (Figure 9B). All of the compounds were
able to decrease migration for more than 80% at higher (150 µM) concentration in comparison to
control, with compound 13e being able to inhibit migration almost completely. Interestingly,
compounds 11a (for CXCR3) and 13e (for CXCR4) were identified as modest negative allosteric
modulators in the BRET based cAMP assay, but did cause the most potent inhibition of
migration. This suggests that quantitative effect on inhibition of chemotaxis is not necessarily in
a strong correlation with the inhibition of G protein-dependent pathways. Other signaling
pathway(s) may be involved in chemotaxis, which alone, or in combination with cAMP-pathway
triggers the full chemotaxis. The very same observation was also confirmed by Milanos et al.
[39].
Surprisingly, compound 7h, which was our starting hit showed only weak negative
allosteric effect on CXCR3 in β-arrestin 2 recruitment assay. Furthermore, it did not have any
effect on G protein activation in BRET based cAMP assay, neither on CXCR3 nor on CXCR4
receptor. This discrepancy between the published and our observations might be explained by
different assay used for the initial characterization of compounds from the virtual screening,
which detects GDP/GTP exchange rather than cAMP concentration as the final consequence of
GPCR activation [28]. According to Kleemann et al. [43], CXCR4 couples to various isoforms
of G
i
proteins and to G
o o
. The activation of G does not lead to the inhibition of adenylyl cyclase
[
44]. For example, the activation of G
o
in the dopamine D2 receptor does not inhibit adenylyl
27

ACCEPTED MANUSCRIPT
35
cyclase [45]. Thus, it is possible that in the [ S]GTPγS assay the compound 7h successfully
inhibits the activation of G proteins, but to lesser extend than that of G . Accordingly, the lack of
o
i
efficacy of the compound 7h to prevent chemokine-induced inhibition of cAMP production by
the G proteins results in no detectable effect of this compound in the cAMP assay. The same
i
applies to CXCR3. Since β-arrestin 2 and cAMP signaling pathways are truly important for
functional agonism/antagonism, we believe that the assays used in this work offer more relevant
information to decipher compounds’ role in CXCR activation.
Overall in our work, we have synthesized and characterized novel negative and positive
allosteric modulators of the chemokine receptors CXCR3 and CXCR4, which are both prominent
drug targets. Importantly, we identified the compound 9e as dual negative allosteric modulator
with comparable functional affinity to both receptors. The compound 9e was able to suppress the
migration of CXCR4 expressing CD4+ T cells.
2
.4. Proposed binding modes of most potent negative allosteric modulators
To propose the binding mode of the most promising negative allosteric modulators as
confirmed by the cAMP BRET assay, a series of molecular docking calculations were performed
using the crystal structure of CXCR4 (PDB code: 3ODU) and its derived CXCR3 homology
model (Supplementary data). The proposed binding modes are shown in Figures 10 and 11 by
representatives 9b and 9e for CXCR3 and 7a and 9e for CXCR4.
1
.39
Figure 10 shows that methoxy substituted benzyl moiety of 9e forms H-bond with Y60
(superscript numbers denote Ballesteros-Weinstein numbering) [46], while chloro substituent
1
.39
cannot. This is in line with previous reports where Y60
has been reported as H-bond
2
.63
interaction partner important for binding of known CXCR3 ligands [22]. Furthermore, D112
28