Discovery of a Locally and Orally Active CXCL12 Neutraligand (LIT-927) with Anti-inflammatory Effect in a Murine Model of Allergic Airway Hypereosinophilia

Article abstract of DOI:10.1021/acs.jmedchem.8b00657

We previously reported Chalcone-4 (1) that binds the chemokine CXCL12, not its cognate receptors CXCR4 or CXCR7, and neutralizes its biological activity. However, this neutraligand suffers from limitations such as poor chemical stability, solubility, and oral activity. Herein, we report on the discovery of pyrimidinone 57 (LIT-927), a novel neutraligand of CXCL12 which displays a higher solubility than 1 and is no longer a Michael acceptor. While both 1 and 57 reduce eosinophil recruitment in a murine model of allergic airway hypereosinophilia, 57 is the only one to display inhibitory activity following oral administration. Thereby, we here describe 57 as the first orally active CXCL12 neutraligand with anti-inflammatory properties. Combined with a high binding selectivity for CXCL12 over other chemokines, 57 represents a powerful pharmacological tool to investigate CXCL12 physiology in vivo and to explore the activity of chemokine neutralization in inflammatory and related diseases.

Full text of DOI:10.1021/acs.jmedchem.8b00657

Subscriber access provided by University of South Dakota
Article
Discovery of a Locally and Orally Active CXCL12
Neutraligand (LIT-927) with Anti-inflammatory Effect in
a Murine Model of Allergic Airway Hypereosinophilia
Pierre Regenass, Dayana Abboud, François Daubeuf, Christine Lehalle, Patrick Gizzi,
Stéphanie Riché, Muriel Hachet-Haas, François Rohmer, Vincent Gasparik, Damien
Boeglin, Jacques Haiech, Tim Knehans, Didier Rognan, Denis Heissler, Claire Marsol,
Pascal Villa, Jean-Luc Galzi, Marcel Hibert, Nelly Frossard, and Dominique Bonnet
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b00657 • Publication Date (Web): 14 Aug 2018
Downloaded from http://pubs.acs.org on August 15, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Pharmacie/Laboratoire d'Innovation Thérapeutique
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Discovery of a Locally and Orally Active CXCL12 Neutraligand (LIT-927) with Anti-
inflammatory Effect in a Murine Model of Allergic Airway Hypereosinophilia
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
,4
2,4
1,4
1,4
3,4
Pierre Regenass , Dayana Abboud , François Daubeuf , Christine Lehalle , Patrick Gizzi ,
1
,4
2,4
1,4
1,4
1,4
Stéphanie Riché , Muriel HachetꢀHaas , François Rohmer , Vincent Gasparik , Damien Boeglin ,
1,4
1,4
1,4
1,4
1,4
3,4
Jacques Haiech , Tim Knehans , Didier Rognan , Denis Heissler , Claire Marsol , Pascal Villa ,
2
,4
1,4
1,4*
1,4*
JeanꢀLuc Galzi , Marcel Hibert , Nelly Frossard , Dominique Bonnet
1
Laboratoire d’Innovation Thérapeutique, Faculté de Pharmacie, UMR7200 CNRS/Université de
Strasbourg, 74 route du Rhin, 67401 Illkirch, France.
2
Biotechnologie et Signalisation Cellulaire, Ecole Supérieure de Biotechnologie de Strasbourg, UMR
7
242 CNRS/Université de Strasbourg, Bld Sébastien Brant, 67412 Illkirch, France.
3
Plateꢀforme de chimie biologique intégrative de Strasbourg, UMS 3286 CNRS/Université de
Strasbourg, 67412 Illkirch, France.
4
Labex MEDALIS, Université de Strasbourg, 67000 Strasbourg, France.
1
ACS Paragon Plus Environment

Page 3 of 62
Journal of Medicinal Chemistry
1
2
3
ABSTRACT
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
We previously reported Chalconeꢀ4 (1) that binds the chemokine CXCL12, not its cognate receptors
CXCR4 or CXCR7, and neutralizes its biological activity. However, this neutraligand suffers from
limitations such as poor chemical stability, solubility and oral activity. Herein, we report on the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
discovery of pyrimidinone 57 (LITꢀ927), a novel neutraligand of CXCL12 which displays a higher
solubility than 1 andis no longer Michael acceptor. While both 1 and 57 reduce eosinophil recruitment
in a murine model of allergic airway hypereosinophilia, 57 is the only one to display inhibitory activity
following oral administration. Thereby, we here describe 57 as the first orally active CXCL12
neutraligand with antiꢀinflammatory properties. Combined with a high binding selectivity for CXCL12
over other chemokines, 57 represents a powerful pharmacological tool to investigate CXCL12
physiology in vivo, and to explore the activity of chemokine neutralization in inflammatory and related
diseases.
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
INTRODUCTION
Chemokines are small proteins with critical roles in the development and function of various tissues in
1
,2,3
vertebrates.
As a rather general rule, chemokines and their G proteinꢀcoupled receptors display
4
,5
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
redundancy and binding promiscuity, i.e. one chemokine may bind to different receptors, whereas
chemokine receptors may be activated by various chemokines. A few chemokines play a pivotal and
nonꢀredundant homeostatic role, as is the case for CXCL12. In the adult, they regulate the directional
6
migration of leukocytes under normal and pathological conditions. They are associated with an
6
extraordinary high number of diseases, including chronic inflammation, autoimmune diseases (lupus
7
8,9
10,11
12
erythematosus), cancer atherosclerosis
or AIDS ; their receptors have been considered as
1
3
druggable targets. Indeed, classical drug design strategies aim at discovering chemokine receptor
14,15
ligands, mainly antagonists, in order to regulate the associated functions.
Yet, many antagonists have
disappointingly failed in clinical trials, which may be related either to chemokine receptor redundancy,
1
6,17
or to the difficulty of designing specific chemokine receptor antagonists.
We have recently opened a
novel avenue for drug development in reporting a small molecule, “Chalconeꢀ4” (1), that displays an
original mechanism of action as it binds to the chemokine CXCL12, not to its two cognate receptors
1
8
CXCR4 and CXCR7, and neutralizes its biological activity. Chalconeꢀ4 and related molecules have
been termed “neutraligands” by analogy with neutralizing antibodies, and proved to have therapeutic
potential. Indeed, 1 inhibits binding of CXCL12 to CXCR4 and CXCR7, reduces intracellular calcium
+
responses, blocks chemotaxis of human peripheral blood CD4 lymphocytes, and prevents CXCR4
1
8
internalization in response to CXCL12. This chemical compound is active in vivo in a mouse model of
allergic airway eosinophilic inflammation, where it inhibits eosinophil infiltration in response to the
allergen. It is also active in other disease models involving the CXCL12/CXCR4 axis such as the WHIM
1
9,20
syndrome or carcinogenesis.
As the poor solubility of 1 prevented its local administration in the
3
ACS Paragon Plus Environment

Page 5 of 62
Journal of Medicinal Chemistry
1
2
3
airways, we have previously successfully developed simple analogues of 1 acting as prodrugs to
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
1
improve solubility, and showed that a significant local bronchial activity was reached at low dose. We
have also developed an antedrug active when administered locally, and inactivated when passing
22
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
systemically, thereby preventing distant adverse events, and thus optimizing the specificity of action.
Therefore, neutralizing CXCL12 by small compounds has proven to be a promising strategy to treat
inflammatory diseases, in particular asthma. However, all these neutraligands encompass a chalcone
chemotype and display major limitations including Michael acceptor character and poor oral activity.
Very few other small organic compounds able to bind CXCL12 have been reported to date. Recently, in
2
3
silico screening led to the identification of ZINC 310454, obtained following a fragmentꢀbased
2
4
structureꢀactivity relationship (SAR) analysis, with some activity on cell migration in vitro at µM
concentrations. Other small molecule ligands have been identified for different binding sites on
2
5
CXCL12, but neither in vitro nor in vivo activity have been reported yet. Another example is the
2
6
aptamer NOXꢀA12 that shows anti CXCL12 activity in vitro and in vivo, but is not orally bioavailable.
Here, we describe a systematic SAR study around 1 and the discovery of the first selective and orally
active CXCL12 neutraligand 57 (LITꢀ927) derived from a pyrimidinone scaffold in an allergic airway
eosinophilic inflammation in the mouse.
4
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
RESULTS AND DISCUSSION
The physiological and pathophysiological importance of CXCL12 and CXCR4 has prompted us to
launch drug discovery programs for various diseases. Chalconeꢀ4 (1), previously identified following a
1
8
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
high throughput screening approach, binds to CXCL12, thereby preventing interaction with CXCR4.
However, 1 displays also some major drawbacks for drug development: 1) its solubility in aqueous
media is low (9 µM in PBS buffer); 2) the presence of the α,βꢀunsaturated carbonyl group (often
27
referred to as Michael acceptors) allows 1,4ꢀadditions with thiols; 3) it is not active in vivo in murine
models of allergic airway hypereosinophilia by the oral route.
To overcome these limitations, 1 was selected as the starting point for further structural optimization.
The modular nature of this compound made it particularly well suited for utilization of array synthesis to
efficiently explore the SAR of the chemotype. Optimization efforts went along four directions: i)
exploration of the substitution on ring A, ii) on ring B, iii) modification of the central core (CC) of 1,
and iv) design of rigid isosteres (Figure 1).
Figure 1. SAR strategy around Chalconeꢀ4 (1).
5
ACS Paragon Plus Environment

Page 7 of 62
Journal of Medicinal Chemistry
1
2
3
Synthesis. Chalcones 2ꢀ31 were prepared from an acetophenone and a benzaldehyde by a Claisenꢀ
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
8
Schmidt reaction using barium hydroxide hydrate as a base. When necessary, methoxymethyl (MOM)
deprotection was carried out with concentrated hydrochloric acid in THF (Scheme 1). In some cases, the
condensation was advantageously performed under acidic conditions by bubbling hydrogen chloride
through a dichloromethane solution of the reagents (32 and 33) or using a mixture of aqueous hydrogen
chloride and sulfuric acid (34). Interestingly, compound 1 was best synthesized by carefully adding
thionyl chloride to an ethanolic solution of the two carbonyl partners, thus generating hydrogen chloride
in situ. In some cases, additional steps were required. Thus, amino derivative 36 was prepared by
reduction of the corresponding nitro precursor 28 in the presence of iron in acidic ethanol, whereas 35
was obtained by demethylation of the methoxy substituents with boron tribromide in dichloromethane.
Chalcone 37 was prepared from 1ꢀ(4ꢀchlorophenyl)butꢀ2ꢀynꢀ1ꢀone and (4ꢀbromoꢀ2ꢀmethoxyphenoxy)
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
tertꢀbutyl)dimethylsilane using nꢀbutyllithium and copper(I) bromide, followed by deprotection of
TBDMS with sodium hydroxide. Chalcone 38 was obtained by treatment of the corresponding carbonyl
compounds with pyridinium acetate/piperidine.
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Scheme 1. Synthesis of Compounds 1-38
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Reagents and conditions: (a) Ba(OH) .H O, EtOH , rt, 20 h, 19ꢀ97%; for preparation of 14 : NaOH,
2
2
EtOH, 80 °C, 16 h, 38%; 15: KOH, EtOH/H O (1/1), rt, 16 h, 95% ; (b) Aqueous HCl, THF, rt, 16 h, 53ꢀ
100%; (c) 32ꢀ33: HCl, MeOH, rt, 2 h, 12ꢀ30%; 34: aqueous HCl, H SO , rt, 16 h, DMF, 2%; (d) BBr ,
CH Cl , 0 °C to rt, 16 h, 10%; (e) Fe, aqueous HCl, EtOH, 80 °C, 2 h, 25%
2
2
4
3
2
2
In order to increase waterꢀsolubility of the target compounds, we planned to introduce a hydrophilic
chain on ring B. Scheme 2 describes the synthesis of these modified chalcones. Treatment of 3,4ꢀ
dihydroxybenzaldehyde in the presence of an excess of sodium hydride to deprotonate both hydroxyl
groups and of an azido sulfonate led to alkylation at the more reactive 3ꢀphenate site. Subsequent
protection of the 4ꢀhydroxyl moiety as a methoxymethyl ether gave compounds 39ꢀ41 that were
28
condensed with pꢀchloroacetophenone as previously described. Finally, MOM deprotection followed
by Staudinger reduction of azides 42ꢀ44 in THFꢀwater enabled the access to 45ꢀ47 as their
hydrochloride salts.
7
ACS Paragon Plus Environment

Page 9 of 62
Journal of Medicinal Chemistry
1
2
3
a
Scheme 2. Synthesis of Chalcones with Increased Water-solubility
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Reagents and conditions: (a) NaH, DMSO, Ts(OCH CH ) N for 39 and 40 or Ms(OCH CH ) N for
1, 0 °C, 1 h, then rt, 16 h, 68ꢀ76%; (b) MOMCl, K CO , acetone, rt, 16 h, 95ꢀ100%; (c) 4ꢀClꢀ
2 3
2
2 n
3
2
2 3
3
4
C H C(O)Me, Ba(OH) .H O, EtOH, rt, 4 h, 30ꢀ56%; (d) 12 M HCl, THF, rt, 16 h, 100%; (e) PSꢀPPh
3
for 45 and 46 or TCEPꢀHCl for 47, THF/H O, rt, 16 h, 50ꢀ70%
6
4
2
2
2
Several other analogues were prepared from 1 as shown in Scheme 3. Thus, 1,3ꢀdiarylpropanꢀ1ꢀone 48
was obtained by hydrogenation on platinum(IV) oxide and oxime 49 by reaction with hydroxylamine.
Treatment with hydrazine hydrate in ethanol afforded the pyrazoline 50 or its Nꢀacetyl analogue 51
when acetic acid was added to the reaction, both compounds being obtained as racemic mixtures.
Several pyrimidines could be obtained by condensation of 1 either with imidamides (52ꢀ54), guanidines
(
55ꢀ56) or urea (57). The pyrazole 60 could be prepared by hydrazine hydrate treatment and MOM
deprotection of diketone 59 prepared from 4ꢀchloroacetophenone and MOMꢀprotected methyl vanillate
8.
5
8
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Scheme 3. Synthesis of Compound 1 Derivatives with Central Core Modifications.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Reagents and conditions: (a) PtO , H , EtOAc, rt, 20 min, 46%; (b) NH OH, pyridine, rt, 16 h, 43%; (c)
NH NH .H O, EtOH, 80 °C, 4 h, 47%; (d) NH NH .H O, AcOH, EtOH, 80 °C, 18 h, 69%; (e) H Nꢀ
CY=NH (Y = H, Me, Ph, NH , NHMe), DMF, base, 20ꢀ53%; (f) urea, HCl, dioxane/EtOH, 100 °C, 2 h,
3%; (g) NaH, 4ꢀClꢀC H C(O)Me, THF, 0 °C, 1 h, 70 °C, 16 h, 32%; (h) NH NH .H O, EtOH, 80 °C,
6 4 2 2 2
2
2
2
2
2
2
2
2
2
2
2
4
1
0 h, 84%; (i) 12M HCl, THF, rt, 16 h, quant.
Constrained chalcones, incorporating an additional 5ꢀ or 6ꢀmembered ring, were also synthesized by
reacting 6ꢀchloroꢀ3(2H)ꢀbenzofuranone, 6ꢀchlorotetralꢀ1ꢀone and 7ꢀchlorochromanꢀ4ꢀone with 4ꢀ
hydroxyꢀ3ꢀmethoxybenzaldehyde under the conditions already mentioned in Scheme 1. Due to the poor
solubility of 61 in aqueous media, 6ꢀchlorobenzofuranꢀ1ꢀone was reacted with benzaldehyde 40 bearing
a hydrophilic ethylene glycol chain to provide after Staudinger reduction the more soluble compound
6
3. Better solubility may also be obtained if the additional ring is an oxygenated heterocycle. We
therefore prepared the benzofuranꢀ3ꢀone 67 by treating 66 with mercury(II) acetate in anhydrous
1
13
pyridine (Scheme 4). All compounds were fully characterized by H, C NMR and MS analysis and
were obtained with a purity >95%.
9
ACS Paragon Plus Environment

Page 11 of 62
Journal of Medicinal Chemistry
1
2
3
a
Scheme 4. Synthesis of Conformationally Constrained Analogues
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Reagents and conditions: (a) Ba(OH) .H O, MeOH/EtOH, rt, 16 h, 28ꢀ48%; (b) SOCl , EtOH, 53ꢀ
2
2
2
100%; (c) 12 M HCl, THF, rt, 16 h, quant.; (d) PSꢀPPh , THF/H O (4/1), 70 °C, 4 h, 20%; (e) 1 M HCl,
rt, 30 min, quant.; (f) Ba(OH) .H O, EtOH, rt, 48 h, 28%; (g) Hg(AcO) , pyridine, 110 °C, 2 h, 25%
3
2
2
2
2
Structure-activity relationship and structural optimization. Pharmacomodulation of 1, the first small
organic compoundꢀbased chemokine neutraligand, led to a structure–activity relationship analysed in a
systematic way by studying the consequences on CXCL12 binding of structural modifications of the 4ꢀ
chloro aromatic ring (ring A; Table 1), of the phenolic part (ring B; Table 2) and of the central core of 1
(
Table 3). Finally, rigid isosteres were prepared and tested to characterize the active conformation of 1
and to potentially improve its affinity and bioavailability (Table 4). The ability of each novel compound
to inhibit CXCL12ꢀCXCR4 interactions was carefully evaluated by using FRETꢀbased binding
1
8,29,30
experiments as previously described.
In this assay, binding of Texas redꢀlabeled CXCL12
(
CXCL12ꢀTR) to CXCR4 fused to EGFP expressed at HEK293 cell membrane induces fluorescence
resonance energy transfer (FRET) between the EGFP and the Texas red, which can be monitored by the
reduction of EGFP fluorescence emission at 510 nm. This reduction of EGFP fluorescence is doseꢀ
dependently inhibited by molecules that bind either CXCR4 or CXCL12 in a competitive manner.
10
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 1. Effect of the Substituents of Ring A on CXCL12-TR Binding
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Solubility ,
µM
>5
Inhibition of CXCL12ꢀTR binding,
Entry Compd
Ar
4ꢀClꢀC H ꢀ
K [nM]
53 ± 31
107 ± 35
i
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
32
33
6
4
>5
2ꢀClꢀC H ꢀ
6 4
b
>5
>5
>5
>5
>5
>5
>5
1.92
0.95
>5
>5
>5
>5
>5
>5
>5
3ꢀClꢀC H ꢀ
25 % of inhibition at 30 µM
430 ± 214
6
4
4ꢀFꢀC H ꢀ
6 4
4ꢀBrꢀC H ꢀ
44 ± 17
107± 17
>10 000
285 ± 17
6
4
4ꢀIꢀC H ꢀ
6
4
Phꢀ
4ꢀMeꢀC H ꢀ
6
4
4ꢀCF ꢀC H ꢀ
1 070 ± 230
25 ± 10
5% of inhibition at 30 µM
1 780 ± 714
30% of inhibition at 30 µM
5 350 ± 750
3
6
4
1
0
1
2
3
4
4ꢀiꢀPrꢀC H ꢀ
6 4
b
1
1
1
1
2ꢀnaphthylꢀ
4ꢀMeOꢀC H ꢀ
6
4
b
4ꢀNO ꢀC H ꢀ
2
6
4
4ꢀHO CꢀC H ꢀ
2 6 4
b
15
16
4ꢀHSO ꢀC H ꢀ
6% of inhibition at 30 µM
107 ± 11
3
6
4
2ꢀFꢀ4ꢀClꢀC H ꢀ
6 3
1
1
1
2
2
7
8
9
0
1
2,3ꢀCl ꢀC H ꢀ
357 ± 20
48 ± 17
53 ± 11
71 ± 21
714 ± 285
> 10 000
> 10 000
2
6
3
2,4ꢀCl ꢀC H ꢀ
2
6
3
>5
>5
>5
>5
2,5ꢀCl ꢀC H ꢀ
2
6
3
2,6ꢀCl ꢀC H ꢀ
2
6
3
3,4ꢀCl ꢀC H ꢀ
2
6
3
22
23
3ꢀPyꢀ
4ꢀPyꢀb
>5
a
Solubility of compounds in HEPESꢀBSA buffer. Limited by the solubility of the compound in HEPESꢀ
BSA buffer
Influence of ring A substitution on CXCL12 binding. The potency of series 1 analogues obtained by ring
A modification to inhibit CXCL12ꢀTR binding to the EGFPꢀtagged CXCR4 receptor (K ) are listed in
i
Table 1. The solubility of all compounds at a 5 ꢀM concentration was validated before binding
assessment. Interestingly, the unsubstituted phenyl derivative 7 is completely inactive at 10 ꢀM (entry
7
), indicating that the chalcone chemotype is not sufficient on its own for binding the chemokine, as
11
ACS Paragon Plus Environment

Page 13 of 62
Journal of Medicinal Chemistry
1
2
3
18
already reported. Substitution at position 4 with halogens results in a dramatic increase in affinity with
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
an optimum observed for chlorine (1) and bromine (5) with K = 53 nM and K = 44 nM, respectively
i
i
(entries 1 and 5). A twoꢀfold decrease is observed with iodine (6) while a 4ꢀfluoro substituent displays a
0ꢀfold decrease in affinity (4) (entries 6 and 4). Chlorine and bromine may represent the best
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
compromise in terms of size, electronegativity and polarisability. The influence of position 4
substituents was further explored with limited success. Hence, methylation (8) provides a neutraligand
with a K = 430 nM, whereas its electron withdrawing trifluoromethyl isostere 9 is even less potent (K =
i
i
1
070 nM) with an affinity comparable to the electron donating 4ꢀmethoxy derivative 12 (K = 1 780
i
nM) (entries 8, 9 and 12). Interestingly, the hydrophobic and bulky 4ꢀisopropyl group (10) further
improves the potency down to K = 25 nM (entry 10). However, its limited water solubility renders
i
difficult its further in vivo evaluation. Finally, a series of three compounds with electron withdrawing
substituents in position 4 were tested (13, 14 and 15). None of them shows a significant affinity for
CXCL12 or CXCR4 (entries 13, 14 and 15). In summary, the chloro, bromo and isopropyl substituents
display the highest affinities, suggesting that bulkiness and polarisability are more important structural
determinants of activity than electron withdrawing or electron donating properties. The 3ꢀchlorophenyl
analogue 3 shows a drastically reduced potency while the 2ꢀchloro isostere 2 retains a good affinity (K =
i
1
07 nM), with only a twoꢀfold decrease compared to 1. The detrimental role of the 3ꢀchloro substitution
is confirmed with the disubstituted compounds 17 and 21: the 3,4ꢀdichloro derivative 17 is indeed 14ꢀ
fold less potent than 11 while the 2,4ꢀdichloro analogue 18 is equipotent (K = 48 nM) (entries 17, 11
i
and 18). The 2ꢀfluoroꢀ4ꢀchloro compound 16 as well retains a significant potency (K = 107 nM) (entry
i
1
6). Similarly, if one compares the potency of the 2ꢀchloro derivatives with a second chlorine in
positions 3, 5 and 6 (17, 19 and 20, respectively), one observes the worst potency when a chlorine is in
position 3 and an affinity comparable to 1 for the 2,5ꢀdichloro analogue 19 (K = 53 nM) (entries 17, 19
i
12
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
and 20). Noteworthy, the replacement of the phenyl moiety by a naphthyl (11), a 3ꢀ or a 4ꢀpyridinyl (32ꢀ
3) failed to provide active compounds (entries 11, 22 and 23).
3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 2. Effect of the substituents of ring B on CXCL12-TR binding
a
1
2
Solubility , Inhibition of CXCL12ꢀTR binding,
Entry Compd
R
R
µM
K (nM)
i
1
2
3
4
5
6
7
8
9
1
ꢀOMe
ꢀOH
ꢀH
ꢀOCH Oꢀ
ꢀOCH CH Oꢀ
ꢀOMe
ꢀOMe
ꢀOMe
ꢀH
ꢀOH
ꢀOMe
ꢀH
>5
>5
53 ± 31
>10 000
18% of inhibition at 1 µMb
25% of inhibition at 3 µM
1 380 ± 357
22
23
24
25
26
27
28
29
30
34
35
36
45
46
47
b
1.17
2.31
>5
0.36
>5
>5
>5
4.72
>5
>5
2
2
2
ꢀH
ꢀOMe
ꢀNO2
ꢀOH
ꢀOH
ꢀOH
ꢀOH
ꢀNH2
ꢀOH
ꢀOH
ꢀOH
>10 000
>10 000
>10 000
625 ± 3
b
10
ꢀOEt
52% of inhibition at 5 µM
11
12
13
14
15
ꢀCO H
>10 000
350 ± 85
>10 000
357 ± 121
1 071 ± 321
475 ± 71
2
ꢀOH
ꢀOMe
>5
>5
>5
>5
+
3
–
ꢀOCH CH NH Cl
2
2
+
–
–
ꢀ(OCH CH ) NH Cl
2
2 2
3
+
16
ꢀ(OCH CH ) NH Cl
2
2 3
3
a
b
Solubility max of compound in HEPESꢀBSA buffer; Limited by the solubility of the compound in
HEPESꢀBSA buffer
Influence of ring B substitution on CXCL12 binding. In order to study the contribution of substitutions
on ring B (Table 2), a series of analogues with a 4ꢀchlorophenyl as ring A was synthesized. The water
solubility of these analogues and their potency to inhibit the chemokine binding are listed in Table 2.
Two ring B substituents are present in the original hit 1: a methoxy group in position 3 and a hydroxyl in
position 4. The derivative 23 with an unsubstituted aromatic ring B is only weakly active and has a low
13
ACS Paragon Plus Environment

Page 15 of 62
Journal of Medicinal Chemistry
1
2
3
solubility in water (entry 2). When the methoxy group on ring B is absent (22), a tenꢀfold decrease in
affinity is observed (entry 2) whereas a loss of affinity occurs when only the methoxy group is preserved
(26) (entry 6). These findings suggest that the 4ꢀhydroxy is more important than the 3ꢀmethoxy for the
affinity towards CXCL12. When the hydroxyl and methoxy groups of 1 are inverted as in 22, no activity
is detected (entry 2). Compound 35 with a 3,4ꢀdihydroxyphenyl ring B is almost two times more active
than 29 bearing only a 4ꢀhydroxyl group (entries 12 and 9). These results indicate that the presence and
the position of the methoxy and hydroxyl substituents on ring B is crucial and contributes greatly to the
affinity of compounds to CXCL12. Furthermore, when the hydroxyl group of 1 is replaced either by a
methoxy (27), a nitro (28) or an amino group (35), the activity is very low (entries 7, 8 and 12).
Accordingly, the cyclic dioxo analogues 24 and 25 are almost inactive (entries 4 and 5). These results
clearly demonstrate the crucial role of the hydroxyl group at the 4 position on ring B.
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
In contrast, the nature of the R alkoxy group can be fineꢀtuned without drastic loss of affinity (30)
(entry 10). Thus, a 2ꢀ(polyethoxy)ethanꢀ1ꢀaminium hydrophilic sideꢀchain was introduced at this
position to improve solubility properties, the best results being obtained with compounds 45 and 47
(entries 14 and 16).
Table 3. Effect of modification of central core (CC) on CXCL12-TR binding
a
Solubility ,
Entry
1
Compd
Central core
Inhibition of CXCL12ꢀTR binding, K (nM)
i
µM
31
>5
> 10 000
> 10 000
2
37
2.6
14
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
3
4
5
38
48
49
>5
nd
>5
> 5 000
> 10 000
> 10 000
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
7
50
51
>5
>5
107 ± 71
> 10 000
b
8
9
52
53
4.29
>5
30% of inhibition at 5 µM
1 680 ± 71
b
1
0
54
55
56
0.53
>5
20% of inhibition at 0.3 µM
> 10 000
11
12
>5
> 10 000
1
1
3
4
57
60
>5
267 ± 71
> 10 000
19.5
a
b
Solubility of compounds in HEPESꢀBSA buffer; Limited by the solubility of the compound in HEPESꢀ
BSA buffer
Influence of central part substitution and stiffening. Having determined the most suitable substituents on
rings A and B, we next examined the influence of the central part of 1 on the affinity for CXCL12
(Tables 3 and 4). The reduction of the olefinic bond leads to the inactive compound 48 (Table 3, entry
4
), likely due to the modification of both the electron distribution and the conformation of the molecule.
The oxime 49, the chalcone 31 where ring A and ring B have switched position or 37 and 38 with a
15
ACS Paragon Plus Environment

Page 17 of 62
Journal of Medicinal Chemistry
1
2
3
trisubstituted double bond have no significant activity on CXCL12 binding (K > 5 µM) (Table 3, entries
i
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
, 2, 3 and 5). In our hands, only the αꢀcyano derivative (CNꢀChalcone 4, Figure 2) retains a good
activity (K = 53 nM). This compound was previously reported active in vivo, behaving as an antedrug
i
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
29
with potential therapeutic application.
Figure 2. Structure of the αꢀcyano derivative of CNꢀChalcone 4.
Table 4. Effect of Stiffening by Forming a 5- or 6-membered Ring Fused With Ring A on
CXCL12-TR binding
a
1
Solubility , Inhibition of CXCL12ꢀTR binding,
Entry Compd
X
n
R
µM
0.88
>5
>5
>5
K (nM)
107 ± 2
496 ± 285
i
1
2
3
4
5
61
63
64
65
67
ꢀCH2ꢀ
ꢀCH2ꢀ
ꢀCH2ꢀ
ꢀOꢀ
0
0
1
1
0
ꢀOMe
+
ꢀ
ꢀ(OCH ) NH Cl
2 2
3
ꢀOMe
ꢀOMe
ꢀOMe
70% of inhibition at 10 µM*
357 ± 114
ꢀOꢀ
>5
203 ± 71
a
Solubility of compounds in HEPESꢀBSA buffer
16
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
We then attempted to rigidify the central oxopropenyl moiety of 1 in two ways: 1) by building a 5ꢀ or 6ꢀ
membered heterocycle on this oxopropenyl connector (Table 3); 2) by bridging the ortho position of
ring A with Cꢀ2 of the oxopropenyl linker by one or two atoms (Table 4). In the first series, the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
pyrazoline 50 is particularly interesting since it retains a good affinity (K = 107 nM) and shows that the
i
α,βꢀunsaturated ketone moiety is not mandatory in this case, compared to 48 (entries 4 and 6). The
acetylated derivative 51 however has a drastically reduced activity (entry 7). Surprisingly enough, the
pyrazole 60 is an inactive analogue though the conjugated system has been restored, thus showing the
subtlety of the docking of these molecules (entry 14). Conjugated sixꢀmembered ring analogues,
pyrimidines 52 to 56, have a poor activity (entries 8 and 12). The simplest, unsubstituted pyrimidine 52
shows a very modest inhibition (30%) at 5 ꢀM (entry 8). The activity is slightly improved by
substitution with a methyl (53) showing the possibility to reach a hydrophobic pocket (entry 9).
However, substitution by a phenyl group (54) or by the more polar amino groups (55, 56) erases the
activity (entries 10ꢀ12). In contrast, the oxo analogue 57 retains a very significant activity (K = 267 nM)
i
though fiveꢀfold lower than 1 (entry 13). Noteworthy, 57 presents a more rigid core than 1 that freezes
the central chain in the allꢀtrans conformation, indicating that this corresponds to the active
conformation of 1. It is in agreement with crystal structures of both 1 and 57 that are very well
superimposable (Supplementary, Figure S1 and Tables S1 and S2). Interestingly, the pyrimidinone
nucleus in structure 57 is different from the overexploited chalcone chemotype. Furthermore, it contains
more heteroatoms than 1, which results in an increased waterꢀsolubility (36 versus 9 ꢀM; Table 5, entry
5
). As a consequence, 57 represents an interesting lead compound that was studied in vivo as detailed
below.
17
ACS Paragon Plus Environment

Page 19 of 62
Journal of Medicinal Chemistry
1
2
3
Table 5. Solubility and Stability of Selected Neutraligands
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
b
Solubility,
µM
Stability,
%
Entry Compd
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
PBS
9
PBS/Cdx
>300
PBS
100
100
100
78
PBS/Cdx
100
1
2
3
4
5
6
7
1
6
1.2
4.2
34.6
36.4
0.5
1.9
319
100
45
50
57
65
67
>5 000
>3 000
>5 000
933
100
98
100
100
100
100
100
330
100
a
b
Measured at pH 7.4 in PBS or PBS/Cdx, 24 h of incubation, 22 °C. Percentage of remaining compound
monitored by RPꢀHPLC after 24 h at 37 °C
The second series of rigidified compounds proved to be interesting as well (Table 4). Linking ring A
with the α,βꢀunsaturated ketone via a methylene bridge as in benzofuranone 61 halves the activity (K =
i
107 nM) relative to 1 (entry 1). This twoꢀfold decrease in affinity might stem from a higher
hydrophobicity. Bridging with two methylene units as in tetralone 64 (entry 3) leads to a very low
activity that may result from a less favorable conformation or from an excessive hydrophobicity. An oxy
bridge leads to 67 that is slightly less active than 61 (K = 203 nM) but has a better water solubility (>5
i
ꢀ
M vs 0.88 ꢀM; entries 1 and 5) whereas the oxymethylene derivative 65 is another less potent but still
active compound (K = 357 nM; solubility >5 ꢀM; entry 4). In summary, two out of three rotatable
i
bonds of the central core of compound 1 can be incorporated into rigid systems while retaining the
ability to inhibit CXCL12 binding to its CXCR4 receptor in the nanomolar range. This supports the idea
18
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
that the active conformation of 1 is probably very close to its conformation in its crystal structure.
However, very subtle changes in the structure may lead to dramatic loss of activity, suggesting that this
series of ligands interacts with a wellꢀdefined and stringent binding site. At this point it was necessary to
demonstrate that these derivatives of molecule 1 are indeed neutraligands, that is to say that they
interfere with CXCL12ꢀCXCR4 binding in actually binding to the chemokine and not to its receptor.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Neutraligand characterization. For the most potent compounds 6, 45, 50, 57, 65 and 67 arising from
the above SAR studies, confirmation of their neutralizing properties was achieved using the procedure
3
1
previously described (Figure 3). Briefly, using the FRETꢀbased binding assay, the characterization of
CXCL12 neutraligands was evaluated in two conditions, differing by the sequence of addition of the
molecules. Hence, reduction of EGFP fluorescence emission is more pronounced when neutraligands are
preincubated (30 min) with the cells expressing CXCR4 receptor than when preincubated with CXCL12
prior to the addition to the cells, indicating that preliminary binding to the chemokine is the mechanism
of action, and the compounds are neutraligands and not receptor antagonists. Following this procedure
data obtained for all six compounds are consistent with a binding to the chemokine and not to the
receptor (Figure 3). Consequently, all selected compounds are indeed CXCL12 chemokine neutraligands
and behave as 1.
19
ACS Paragon Plus Environment

Page 21 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Inhibition of CXCL12ꢀTR binding when neutraligands are incubated with the CXCR4ꢀEGFP
fused receptorꢀexpressing cells prior to addition of CXCL12 (black bars; neutraligand protocol) or with
CXCL12 prior to addition to CXCR4ꢀEGFP fused receptorꢀexpressing cells (grey bars; antagonist
protocol). The compounds were tested at a concentration of 10×K (inhibition constant). The CXCR4
i
antagonist T134 (20 ꢁM) was used as a control in both incubation protocols. Each column represents the
mean (block) ± SD (bars) of n=3 experiments.
In vivo evaluation of new neutraligands. The six compounds identified as CXCL12 neutraligands were
selected for in vivo studies. The iodo analogue 6 is slightly less potent than 1 in vitro (K = 107 nM vs 53
i
nM, respectively) but was selected since it may represent a valuable imaging probe for in vivo
1
23
pharmacokinetic studies by single photonꢀemission computed tomography once I ꢀradiolabeled.
Compound 45 is a direct derivative of 1 branched with a waterꢀsolubilizing side chain. Compounds 50,
57, 65 and 67 are the most potent and soluble cyclic neutraligands identified in vitro, and are
representative of four novel chemotypes: pyrazoline, pyrimidinone, benzofuranone and chromanone.
First, their solubility in assay media was evaluated (Table 5). (2ꢀHydroxypropyl)ꢀβꢀcyclodextrin (Cdx)
was used as the excipient because of its low toxicity and capacity to increase the bioavailability of
3
2
chemical compounds. Solubility studies in PBS/Cdx (10% w/w) showed that all compounds are
20
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
soluble at least at a 300 µM concentration (Table 5). The compound stability was evaluated in PBS or in
PBS/Cdx. Only 50 was unstable in PBS, but the degradation could be avoided when Cdx was added
(entry 4). Next, the in vivo activity of the selected compounds was evaluated in the 8ꢀday murine model
2
1,33
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
of allergic airway eosinophilic inflammation previously described (Figure 4A).
Briefly, Balb/c mice
were sensitized to ovalbumin (OVA) in presence of alum, and challenged 3 times at 24 h intervals with
OVA in saline or saline alone. Treatment with each compound (300 nmol/kg in 10% PBS/Cdx) was
administered by the intranasal route 2 h before each challenge. A bronchoalveolar lavage was performed
34
2
4 h after the last challenge, and inflammatory cell number quantified.
A
12
9
6
3
0
*
OVA
1
6
45 50 57 63 66
B
100
1, IC50 > 500 nmol/kg
57, IC50 = 310 nmol/kg
80
60
40
20
0
1
0
100
1000
10000
[nmol/kg]
Figure 4. In vivo antiꢀinflammatory activity of CXCL12 neutraligands administered topically (i.n.) in
the 8ꢀday mouse model of allergic airway hypereosinophilia. (A) Topical treatment with compounds 1,
6, 45, 50, 57, 65 and 67 (300 nmol/kg) in PBS/CdX 10% (vehicle) administered 2 h before each allergen
challenge. Absolute numbers of eosinophils, neutrophils, and lymphocytes in bronchoalveolar lavage
21
ACS Paragon Plus Environment

Page 23 of 62
Journal of Medicinal Chemistry
1
2
3
fluid (BAL) are shown. Blocks are means, and error bars are S.E.M. values (n = 6/group). Statistical
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
#
###
analysis has been protected by a Bonferroni test. P≤0.05 and P≤0.001 vs. vehicleꢀtreated saline group
and *P≤0.05, **P≤0.01; ***P≤0.001 vs. vehicleꢀtreated OVA group. (B) Doseꢀresponse effect of topical
treatment with compounds 1 and 57 (i.n.). Compounds 1 (gray line) or 57 (black line) solubilized in
PBS/Cdx 10% were administered 2 h before each allergen challenge. The percentage of inhibition of
eosinophil recruitment in the bronchoalveolar lavage is shown. Data points are means and bars are SEM
values (n=6/group).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Interestingly, despite a low solubility, the iodoꢀchalcone 6 retained a very good in vivo activity since it
reduced eosinophil infiltration by 36% while the hit compound 1 reduced it by 47% at the same dose.
Compounds 45, 65 and 67 were almost inactive in our assay and were not further investigated. More
interestingly, the two other rigidified analogues 50 and 57 that had similar in vitro potency displayed
very good to excellent in vivo activity, inhibiting eosinophil infiltration by 39% and 48%, respectively
(Figure 4A). Noteworthy, the recruitment of macrophages and lymphocytes was not affected by
treatment with the CXCL12 neutraligands in this model, when neutrophil numbers were reduced with a
trend towards significance (not shown). The chemical instability of 50 observed during both the
synthesis and storage precluded the separation of both isomers and their further in vitro and in vivo
evaluation. We therefore decided to select 57 as lead neutraligand, and characterized its activity (IC ) in
5
0
vivo in a doseꢀresponse efficacy study versus 1 on eosinophil recruitment in our mouse model. The IC50
of the reference compound 1 could not be determined precisely because the solubility threshold was
reached at 500 nmol/kg even when solubilized in PBS/Cdx. In contrast, the soluble novel pyrimidinone
5
7 gave an IC calculated as 310 nmol/kg (Figure 4B).
50
We thereafter characterized further 1) the Michaelꢀacceptor reactivity of 57, 2) its binding properties to
CXCL12, and 3) its selectivity towards other chemokines.
Michael-Acceptor Reactivity. The Michaelꢀacceptor character of 57 and 1 was evaluated in the
22
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
presence of glutathione (GSH, 2.5 eq). As determined by LCꢀMS analysis, after 2 h at 37 °C, more than
0% of 1 were converted into the GSHꢀ1 adduct while 57 was found much more stable without any
3
formation of GSHꢀ57 adduct (Supplementary, Figure S2).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Biophysical characterization. To support the neutraligand character of 57 reported above from FRETꢀ
based binding assay, the binding to CXCL12 was examined by monitoring changes in its emission of
tryptophan fluorescence intensity (Figure 6). Indeed, as previously reported, CXCL12 contains a single
18
tryptophan residue (Trpꢀ57) whose intrinsic fluorescence can be modulated by neutraligands. Thus, we
observed that tryptophan fluorescence intensity at 340 nm (excitation at 285 nm) declined monotonically
when CXCL12 (2 µM) was incubated with increasing concentrations of 57 (Figure 6C). The resulting
tryptophan fluorescence inhibition curve was satisfactorily fitted according to a 1:1 stoichiometry
interaction mode and the deduced dissociation constant (K = 780 ± 320 nM) was in the same range than
d
that determined by FRETꢀbased assay (K = 267 ± 71 nM).
i
As a control from the whole strategy, the structurally close pyrimidinone 68 (Figure 5A) was
synthesized (supplementary for the synthesis) and tested in the FRETꢀbased assay described above
(
Figure 5B). Thereby, 68 inhibited CXCL12 binding to CXCR4 by less than 5% at 300 µM. In the
tryptophan fluorescence assay, 68 nonꢀspecifically decreased Trp fluorescence intensity at 340 nm
Figure 5C). Additionally, 68 was inactive in the mouse model of allergic airway eosinophilic
inflammation at 300 nmol/kg (data not shown).
(
23
ACS Paragon Plus Environment

Page 25 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. Interaction between 57 and CXCL12. (A) Chemical structure of pyrimidinone 68 as compared
to 57. (B) Inhibition of CXCL12ꢀTR binding when 57 and 68 (10 µM) are incubated with the CXCL12
chemokine prior to addition to CXCR4ꢀEGFP fused receptorꢀexpressing cells. (C) Titration of binding
of pyrimidinones 57 (squares) and 68 (circle) to CXCL12 determined by monitoring changes in the
maximal emission of Trp fluorescence of CXCL12 (measured at 340 nm). Increasing amounts of
molecule were added to CXCL12 solution (2 µM in Hepes buffer). The equation used to fit experimental
2
data is the root of the second order equation: (RL) + (RL) × (ꢀRo ꢀ Lo ꢀ K ) + Ro × Lo = 0 where (RL)
D
2
1/2
=
((Ro + Lo + K ) ± ((ꢀRo ꢀ Lo ꢀ K ) ꢀ 4 × Ro × Lo) )/2 and where Ro, the concentration of
D
D
chemokine, is set to 2 µM; Lo is the initial concentration of molecule; K is the dissociation constant of
D
molecule for the chemokine; and RL is the fractional concentration of receptor and ligand complex. Data
are means (dots) and SD (bars) values of three independent experiments.
24
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Molecular modeling of compounds 1 and 57 bound to CXCL12. Physicochemical properties as well
as the herein presented structureꢀactivity relationship suggest a common binding mode of compounds 1
2
4
and 57 to CXCL12 that completely differs from that observed with the inhibitor recently described.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The later compound is bound to the wide and accessible pocket, which is incompatible with the
observed nanomolar binding affinities of our neutraligands. We therefore envisaged the possibility of an
alternative and mostly apolar pocket that would be just large enough to tightly host compounds 1 or 57.
2
4
Systematic scanning of the CXCL12 chemokine dimeric structure with the inꢀhouse developed VolSite
35
program clearly indicates the presence of a small but structurally druggable pocket (Figure 6A) at the
vicinity of Trp57, whose importance to compound 1 binding has previously been evidenced by
1
8
fluorescence spectroscopy. Docking of compounds 1 and 57 to this pocket is possible at the condition
to allow side chain flexibility for three residues (Arg20, Trp57 and Tyr61, Figure 6) that would control
3
6
37
the entry to the Trp57 pocket. Two independent docking algorithms, Plants, and SurflexꢀDock agree
to similarly dock both compounds in a manner that is almost fully consistent with the herein described
structureꢀactivity relationships, assuming a common binding mode for compounds 1 and 57 (Figure
6
B,C). First, the 4ꢀchlorophenyl ring A is deeply buried in an apolar subpocket (Val23, Leu26, Trp57)
whose size is nicely complementary to that of a 4ꢀchlorophenyl substituent. The observed lower affinity
of the 3ꢀchlorophenyl analog 3 can be easily explained by close proximity of Leu26 or Trp57 backbone
atoms that would prevent a complete entry into the pocket. Second, the proposed binding mode also
explains the crucial role of a 4ꢀhydroxyphenyl ring B. The aromatic ring is stacked between Pro32 and
the ammonium moiety of Lys64 to which a πꢀcation interaction is observed in our model. The hydroxyl
group in ring B donates a hydrogen bond to the nearby Gln37 side chain, thereby explaining the higher
affinity of the hydroxyle with respect to the methoxy analog. Third, the ketone moiety of compound 1 is
hydrogenꢀbonded to Arg20 side chain (Figure 6C). In case of compound 57, the bioisosteric pyrimidinꢀ
25
ACS Paragon Plus Environment

Page 27 of 62
Journal of Medicinal Chemistry
1
2
3
2
6
(1H)ꢀone core is ideally located to accept a bidentate hydrogen bond from Arg20 side chain (Figure
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
B). Contrarily to compound 57, compound 1 is predicted to adopt another possible binding mode
(Figure 6D) in which positions of aromatic rings A and B had been interchanged, the ketone being still
anchored to Arg20. Depending on the substituents at rings A and B, the binding mode may readily
switch from one to the other possible orientations.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In the proposed binding mode, the absolute binding free energy of compounds 1 (preferred binding
3
8
ꢀ1
mode) and 57, computed by the Hyde scoring function, is estimated to ꢀ18.60 kJ.mol and ꢀ14.92
ꢀ1
ꢀ1
kJ.mol , close to the experimentally observed values of ꢀ17.78 and ꢀ16.06 kJ.mol , respectively.
Figure 6. Hypothesized binding mode of compounds 1, 57 to the CXCL12 homodimer. The two
monomers are displayed by a transparent surface and backbone ribbons (monomer 1, tan; monomer 2,
sienna). (A) Detection of a druggable cavity (colored spheres) nearby Trp57 at the surface of the
3
5
3
CXCL12 homodimer (PDB ID 4UAI). The volume of the cavity, detected by VolSite is 184 Å . Cavity
26
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
points are colored by pharmacophoric properties (red, hydrogen bond acceptor; blue, hydrogen bond
donor; gray, hydrophobic/aromatic). (B) PLANTS docking pose of compound 57 to CXCL12. Three
side chains (indicated by an arrow) are allowed to freely move, thereby widening the pocket volume up
3
to 371 Å . Hydrogen bonds are indicated by yellow broken lines. Aromatic and πꢀcation interactions are
represented by dark green broken lines. Residues contributing to the binding cavity are labelled at their
Cꢀα atom (tan, monomer 1; sienna, monomer B). (C) PLANTS preferred docking pose of compound 1 to
CXCL12. (D) PLANTS alternative docking pose of compound 1 to CXCL12.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Binding selectivity towards other chemokines. The selectivity of 57 was evaluated towards four other
chemokines, namely CCL17, CCL22, CCL5 and CCL2 by using the neutraligand protocol of the
3
9
recently reported TimeꢀResolved Intracellular Calcium recording (TRICꢀr) assay (Figure 7). This
generic assay enables the discovery of chemokine neutralizing molecules (neutraligands) by
differentiating them from molecules that block the receptor (antagonists). Initially described for the
3
9
identification of CCL17 and CCL22 neutraligands, the assay has been setꢀup for CXCL12. Thereby, as
depicted in Figure 7, pyrimidinone 57 at 10 µM is able to inhibit the increase in intracellular calcium
+
concentration in EGFPꢀCXCR4 HEK cells in response to CXCL12, while it has no effect on calcium
+
+
responses triggered by either CCL17 or CCL22 on EGFPꢀCCR4 HEK cells, CCL5 on EGFPꢀCCR5
+
HEK cells, or by CCL2 on EGFPꢀCCR2 HEK cells. These results demonstrate the high selectivity of
5
7 towards CXCL12 vs other chemokines also involved in asthma.
27
ACS Paragon Plus Environment

Page 29 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
50
00
50
0
solvent
1
5
7
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
CXCL12 CCL17 CCL22
CCL5
CCL2
Figure 7. Selectivity of compounds 1 and 57 towards 5 chemokines. Effect of DMSO (white bars) and
compound 1 or 57 (10ꢁM) on calcium responses triggered by either 5nM CXCL12 on EGFPꢀCXCR4
HEK cells, by 5nM CCL17 and CCL22 on EGFPꢀCCR4 HEK cells, 5nM CCL5 on EGFPꢀCCR5 HEK
cells, or by 5nM CCL2 on EGFPꢀCCR2 HEK cells. Each column represents the mean (block)±SD
+
+
+
+
(
bars) of n=3 experiments.
In vivo evaluation of neutraligand 57 activity and safety. At this stage of the study, pyrimidinone 57
turned out to be a CXCL12ꢀselective neutraligand with markedly improved solubility and no longer
Michael acceptor character. In addition, we demonstrated its antiꢀinflammatory activity at low dose after
local administration in a murine model of allergic airway hypereosinophilia. These results prompted us
to further evaluate 57 in vivo. We first compared the activity and safety of compounds 1 and 57 versus
the main commercialized comparators, the CXCR4 antagonist AMD3100 and the glucocorticoid
dexamethasone (Dex) in the mouse model of hypereosinophilia. The compounds were administered i.p.
2
h before the allergen challenge. Neutraligands 1 and 57 (350µmol/kg) inhibited eosinophil infiltration
by 54 and 55%, and AMD3100 (12.6 µmol/kg) and Dex (2.5 µmol/kg) by 33 and 59%, respectively
(Figure 8A). The antiꢀinflammatory activity of AMD3100 and Dex was accompanied with a statistically
reliable decrease in body and spleen weight, indicating major adverse eventꢀrelated perturbation by these
compounds. In contrast, neutraligands 1 and 57 did not exhibit any side effects (Figure 8B). Even after
28
ACS Paragon Plus Environment