Design, synthesis, and biological characterization of a new class of symmetrical polyamine-based small molecule CXCR4 antagonists

Article abstract of DOI:10.1016/j.ejmech.2020.112410

CXCR4, a well-studied coreceptor of human immunodeficiency virus type 1 (HIV-1) entry, recognizes its cognate ligand SDF-1α (also named CXCL12) which plays many important roles, including regulating immune cells, controlling hematopoietic stem cells, and directing cancer cells migration. These pleiotropic roles make CXCR4 an attractive target to mitigate human disorders. Here a new class of symmetrical polyamines was designed and synthesized as potential small molecule CXCR4 antagonists. Among them, a representative compound 21 (namely HF50731) showed strong CXCR4 binding affinity (mean IC₅₀ = 19.8 nM) in the CXCR4 competitive binding assay. Furthermore, compound 21 significantly inhibited SDF-1α-induced calcium mobilization and cell migration, and blocked HIV-1 infection via antagonizing CXCR4 coreceptor function. The structure-activity relationship analysis, site-directed mutagenesis, and molecular docking were conducted to further elucidate the binding mode of compound 21, suggesting that compound 21 could primarily occupy the minor subpocket of CXCR4 and partially bind in the major subpocket by interacting with residues W94, D97, D171, and E288. Our studies provide not only new insights for the fragment-based design of small molecule CXCR4 antagonists for clinical applications, but also a new and effective molecular probe for CXCR4-targeting biological studies.

Full text of DOI:10.1016/j.ejmech.2020.112410

Journal Pre-proof
Design, synthesis, and biological characterization of a new class of symmetrical
polyamine-based small molecule CXCR4 antagonists
Xiong Fang, Qian Meng, Huijun Zhang, Boqiang Liang, Siyu Zhu, Juan Wang,
Chaozai Zhang, Lina S. Huang, Xingquan Zhang, Robert T. Schooley, Jing An, Yan
Xu, Ziwei Huang
PII:
S0223-5234(20)30381-0
DOI:
https://doi.org/10.1016/j.ejmech.2020.112410
EJMECH 112410
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 21 February 2020
Revised Date: 14 April 2020
Accepted Date: 27 April 2020
Please cite this article as: X. Fang, Q. Meng, H. Zhang, B. Liang, S. Zhu, J. Wang, C. Zhang,
L.S. Huang, X. Zhang, R.T. Schooley, J. An, Y. Xu, Z. Huang, Design, synthesis, and biological
characterization of a new class of symmetrical polyamine-based small molecule CXCR4 antagonists,
European Journal of Medicinal Chemistry (2020), doi: https://doi.org/10.1016/j.ejmech.2020.112410.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. 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.
©
2020 Published by Elsevier Masson SAS.

Design, synthesis, and biological characterization of
a new class of symmetrical polyamine-based small
molecule CXCR4 antagonists
a
a
a
b
a
b
Xiong Fang , Qian Meng , Huijun Zhang , Boqiang Liang , Siyu Zhu , Juan Wang , Chaozai
c
c
c
c
c
a, d, **
Zhang , Lina S. Huang , Xingquan Zhang , Robert T. Schooley , Jing An , Yan Xu
and
a, *
Ziwei Huang
a
Tsinghua-Peking Joint Center for Life Sciences, School of Life Sciences, Tsinghua University,
Beijing 100084, China.
b
Nobel Institute of Biomedicine, Zhuhai 519080, China
c
Department of Medicine, University of California at San Diego, La Jolla, CA 92037, USA
d
School of Life and Health Sciences, Chinese University of Hong Kong, Shenzhen, China
Corresponding Authors
*
*
Dr. Ziwei Huang: phone, +86-010-62794798; E-mail, zwhny@yahoo.com
*Dr. Yan Xu: phone, +86-755-23519651; E-mail, yanxu@cuhk.edu.cn
1

Abstract
CXCR4, a well-studied coreceptor of human immunodeficiency virus type 1 (HIV-1) entry,
recognizes its cognate ligand SDF-1α (also named CXCL12) which plays many important roles,
including regulating immune cells, controlling hematopoietic stem cells, and directing cancer
cells migration. These pleiotropic roles make CXCR4 an attractive target to mitigate human
disorders. Here a new class of symmetrical polyamines was designed and synthesized as
potential small molecule CXCR4 antagonists. Among them, a representative compound 21
(namely HF50731) showed strong CXCR4 binding affinity (mean IC50 = 19.8 nM) in the
CXCR4 competitive binding assay. Furthermore, compound 21 significantly inhibited SDF-1α-
induced calcium mobilization and cell migration, and blocked HIV-1 infection via antagonizing
CXCR4 coreceptor function. The structure-activity relationship analysis, site-directed
mutagenesis, and molecular docking were conducted to further elucidate the binding mode of
compound 21, suggesting that compound 21 could primarily occupy the minor subpocket of
CXCR4 and partially bind in the major subpocket by interacting with residues W94, D97, D171,
and E288. Our studies provide not only new insights for the fragment-based design of small
molecule CXCR4 antagonists for clinical applications, but also a new and effective molecular
probe for CXCR4-targeting biological studies.
Keywords: chemokine receptor; small molecule; CXCR4 antagonists; polyamine.
2

1
. Introduction
The CXC chemokine receptor type 4 (CXCR4) that belongs to the family of the seven
transmembrane helical G-protein-coupled receptors (GPCRs) is well known for its function as an
important coreceptor of HIV-1 infection [1, 2]. The natural chemotactic ligand of CXCR4 is the
stromal cell-derived factor (SDF)-1α, also known as C-X-C motif chemokine 12 or CXCL12.
The binding of SDF-1α/CXCL12 to CXCR4 triggers divergent downstream signaling pathways,
resulting in a variety of physiological responses including cell migration, intracellular calcium
flux, and gene transcription [3, 4]. In addition to the wide expression of CXCR4 in hematopoietic
stem cells (HSCs) and immune cells, CXCR4 overexpression has been found in many types of
cancer cells (>40 types) [5-7]. In the bone marrow (BM) microenvironment, CXCR4-CXCL12
interaction regulates the retention and homing of both normal HSCs and cancer stem cells [8, 9].
In tumor microenvironment, CXCR4-CXCL12 axis promotes not only tumor progression, but
also evasion of immune surveillance by recruiting suppressive immune cells, such as regulatory
T cells (T ), plasmacytoid dentritic cells, and myeloid-derived suppressor cells (MDSCs) [10,
reg
1
1]. Particularly, CXCR4-CXCL12 axis may mediate the migration of CXCR4-expressing
cancer cells to CXCL12-rich BM or other health tissues, eventually leading to tumor metastases
12, 13]. Collectively, these pleiotropic roles make CXCR4 an attractive therapeutic target for
[
mobilizing HSCs, inhibiting tumor metastasis, and synergizing tumor immunotherapy, in
addition to preventing HIV-1 infection. Therefore, the development of new and effective small
molecule CXCR4 antagonists to block the CXCL12-CXCR4 interaction may lead to a variety of
CXCR4-targeting therapeutic agents.
To date, only one small molecule CXCR4 antagonist, AMD3100 (also commercially known
as Plerixafor) was clinically approved for the use in combination with granulocyte-colony
3

stimulation factor (G-CSF) for autologous HSCs mobilization and transplantation in patients
with non-hodgkin’s lymphoma (NHL) and multiple myeloma (MM) [14]. Initially, AMD3100
was developed as an anti-HIV agent but later terminated in clinical trials due to its poor oral
bioavailability, lack of clinical benefit, and cardiac disturbance [15-17]. Currently, the potential
clinical applications of AMD3100 are also being examined for the treatment of leukemia and
solid tumors as the blockade of CXCL12-CXCR4 by AMD3100 can mobilize leukemia cells out
of the protective BM microenvironment and sensitize them to chemotherapy, block growth and
+
metastasis signals of cancer cells, or reduce immunosuppressive cells or enhance CD8 cytotoxic
T cells in tumor microenvironment [18, 19]. However, the clinical usage of AMD3100 in long-
term treatment is still limited by dose-limiting cardiotoxicity. Thus, safer and more efficacious
CXCR4 antagonists are still needed to address the issues of the current therapeutic.
The cocrystal structures of CXCR4 in complex with three types of ligands including small
molecule IT1t, cyclic peptide CVX15, and viral macrophage inflammatory protein II (vMIP-II)
reveal that the large and open pocket of CXCR4 for ligand binding appears to be divided into a
minor and major subpockets [20, 21]. Small molecule IT1t binds to the minor subpocket
surrounded by Y45 (TM1), W94 (TM2), D97 (TM2), Y116 (TM3), H113 (TM3), E288 (TM7),
etc., whereas peptide CVX15 recognizes the major subpocket formed by H113 (TM3), D171
(TM4), R188 (TM4), H203 (TM5), D262 (TM6), Y255 (TM6), E288 (TM7), etc. [20, 21]. These
structural information prompted us to make some attempts in combining two fragments to bind
to these two subpockets, following a fragment-based drug design strategy. In this study, we
designed and synthesized a new class of polyamine-based small molecule CXCR4 antagonists.
Through competitive binding assays and biological studies, compound 21, also named HF50731,
with only four nitrogen atoms and no metal-chelating property compared to AMD3100, was
4

found out of a collection of synthetic small molecules to possess potent CXCR4 binding and
antagonist activity in blocking CXCR4 cellular function and HIV-1 infection via this coreceptor.
The mechanism of action of compound 21 was analyzed by a combination of structure-activity
relationship (SAR) analysis of a series of analogs of compound 21, site-directed mutagenesis of
CXCR4 and molecular docking simulation of compound 21 and CXCR4 interaction.
2
. Results
2
.1. Design
Employing the fragment-based drug design approach and also being intrigued by the
symmetrical feature of the clinically approved drug AMD3100 in which a 1,4-
phenylenebis(methylene) linker tethers two large cyclam rings, we attempted the design of using
a central linker to combine two 1,3-diamine fragments (F01) that are reminiscent of the cyclam
rings of AMD3100 with the rings opened (Figure 1). Different linker variants (L01-L16) were
introduced to probe the effect of the central linker on fragment combination. In addition, the
importance of F01 was also investigated by substituting it with two truncated monoamine
fragments (F02 and F03). Herein, we designed and synthesized a new class of symmetrical
polyamines.
5

Figure 1. The fragment-based combination for designing a new class of symmetrical
polyamines.
2
.2. Chemistry
The general synthetic routes for all polyamine compounds are depicted in Scheme 1. Starting
from 3-aminopropan-1-ol 1, intermediate 2 was prepared in 71% yield via NaBH -mediated
4
reductive amination and Boc protection in a one-pot synthesis. Then, alcohol group of 2 was
activated with MsCl and then attacked by NaN to generate intermediate 3 in 69% yield over two
3
steps, which further underwent the reduction of azido group in the presence of 10% Pd-C to yield
intermediate 4 in 97% yield. Finally, intermediate 4 was coupled with terephthaloyl dichloride,
followed by Boc deprotection using 4 N HCl solution to afford compound 5. Similarly, 4-
(diethoxymethyl)benzaldehyde 6 was subjected to NaBH
4
-mediated reductive amination with N-
(3-aminopropyl)cyclohexylamine, followed by Boc protection to produce intermediate 7 in 41%
yield over two steps. The acetal protecting group of 7 was hydrolyzed using TsOH∙H O to yield
2
6

4
the corresponding aldehyde intermediate, which underwent KMnO -mediated oxidation to
generate intermediate 8 in 64% yield over two steps. Next, intermediate 8 reacted with
intermediate 4 in the presence of the common coupling reagent (HATU and DIPEA), followed
by removing Boc group to afford compound 9.
Additionally, through the one-pot cascade reaction of NaBH -mediated reductive amination
4
and Boc protection, tetra-Boc-protected intermediates 10-20 can also be prepared starting with
commercially available dialdehydes (terephthalaldehyde, isophthalaldehyde, and 4,4’-
biphenyldicarboxaldehyde) and synthetic aldehydes (naphthalene-1,4-dicarbaldehyde S-2, 2-
nitroterephthalaldehyde S-3, 5-nitrosiophthalaldehyde S-4, benzene-1,3,5-tricarbaldehyde S-6,
pyridine-2,5-dicarbaldehyde S-9, and pyridine-2,6-dicarbaldehyde S-10, see Supplemental
Scheme S1 for synthesizing these aldehyde materials). Stepwise reductive amination between
these aforementioned aldehydes and amines (cyclohexanamine, 2-picolyamine, and N-(3-
aminopropyl)cyclohexylamine) generated the corresponding target compounds, followed by
direct purification to obtain compounds 21-25 and 35-36, or subsequent Boc protection yielded
tetra-Boc-protected intermediates 10-14 and 17. Among these Boc-protected intermediates, 14
and 17 were converted into intermediates 15 and 18 via Raney-Ni-catalyzed hydrogenation
reduction, subsequent methylation of 15 and 18 yielded Boc-protected intermediates 16 and 19-
2
0 using formaldehyde as the methylating agent. Finally, these intermediates 10-20 were
removed Boc protection under acidic condition to afford compounds 21 and 26-34 (Scheme 1).
All newly synthesized compounds were tested for purity on an analytical HPLC system
(Supplemental Table S1).
7

Scheme 1. Synthetic routes for symmetrical polyamine-based compounds. Reagents and
conditions: (a) 2: cyclohexanone; 7, 10-14, 17, 21, 24, and 25-36: N-(3-
aminopropyl)cyclohexylamine; 22: cyclohexanamine; 23: 2-picolylamine, MeOH, 4Å MS, 70°C,
6
-12 h, and then NaBH , rt, 3-6 h; (b) (Boc) O, 2 N Na CO , tert-butanol/water, rt, 12 h; (c)
4 2 2 3
8

MsCl, TEA, DCM, 0°C -rt, 5 h; (d) NaN
6 h; (f) terephthaloyl dichloride, TEA, THF, 0°C-rt, 12 h; (g) 4 N HCl in 1,4-dioxane solution,
DCM, rt, 2-4 h; (h) TsOH∙H O, THF, 0-5°C, 1 h; (i) KMnO , acetone, 0°C-rt, 1 h; (j) HATU,
3 2
, DMF, 60°C, 16 h. (e) 10% Pd-C, H , EtOH/EtOAc, rt,
1
2
4
DIPEA, DMF, rt, 12 h; (k) Raney-Ni, H , THF, rt, 6-12 h; (l) 37% HCHO aqueous solution,
2
AcOH, NaBH CN, rt, 12 h.
3
2
.3. Competitive CXCR4 binding assay and structure-activity relationship (SAR) analysis
To evaluate the CXCR4 binding affinities of the synthesized compounds, we performed 12G5
competitive binding assay as described in our previous publications [22-24]. In addition, to
ensure that the observed binding activity of a compound was not due to any cytotoxicity effect of
the compound on the CXCR4 expressing cells, we tested all compounds in cell cytotoxicity
assays using two concentrations (10 µM and 50 µM) of the compounds. All compounds
displayed either no cytotoxic effect on cell viability at concentrations up to 50 µM or only
moderate cytotoxicity between 10 and 50 µM which are almost an order of magnitude higher
concentrations than their effective receptor binding concentrations (i.e., compounds 21 and 27
which are active in CXCR4 binding in the low nanomolar level) (Supplemental Table S2). This
demonstrated that the observed binding activity of the compounds was not due to any
cytotoxicity. Notably, when two [3-(cyclohexylamino)propyl]amino fragments (F01) were
tethered with three phenyl linkers including terephthaloyl linker (L01), para-methylbenzoyl
linker (L02), and para-phenylenebis(methylene) linker (L03), respectively, resulted in
compounds 5, 9, and 21 with significantly different binding affinities (Table 1). Of these,
compound 21 showed the most active CXCR4 binding (mean IC50 = 19.8 nM), which was much
better than compounds 5 and 9. Furthermore, replacement of fragment F01 with two short
fragments (cyclohexylamino fragment F02 and pyridine-2-ylmethanamino F03) completely
9

abolished CXCR4 binding of compounds 22 and 23 (IC50 > 10 µM). When installing three same
fragments (F01) onto a benzene-1,3,5-triyl linker (L04), the activity of compound 24 did not
further increase relative to compound 21, but display moderate CXCR4 binding (mean IC =
5
0
3
12.3 nM). Thus, it seems to be important for the bioactivity of symmetrical polyamines to
combine two same 1,3-diamine fragments (F01) with the phenylenebis(methylene) linker (L03).
Next, the effect of linker L03 was further explored by incorporating various orientations,
aromatic ring systems, and substituents into the central linkers (L05-L16). When changing para
orientation of L03 to meta orientation of L05, compound 25 was found to be approximately 8-
fold less potent than 21 (Table 1). By contrast, replacing L03 with large aromatic ring systems,
such as 4,4’-biphenylenebis(methylene) linker (L06) and 1,4-naphthalenebis(methylene) linker
(L07), compounds 26 and 27 just decreased slightly a 2-3-fold activities compared with
compound 21. When introducing the electron-withdrawing substituent (-NO ) and electron-
2
donating substituents (-NH , -NH(CH ), -N(CH ) ) onto para-/meta-phenylenebis(methylene)
2
3
3 2
linker (L03/L05), two linker variants (L08 and L11) with a nitro group caused significant loss in
the binding activity of compounds 28 and 31 compared with 21. When the nitro group of L08 (or
L11) was reduced to amine group of L09 (or L12), and subsequent methylation generated linker
variants (L10, L13-L14), these variations led to compounds 28 to 30 with a gradual increase in
CXCR4 binding. This trend was also observed in the corresponding meta-oriented analogs 31 to
3
3 apart from compound 34 as an unexpected exception. Additionally, when replacing the
aromatic linker with a heteroaromatic linker, compound 36 with meta-oriented pyridine linker
L16) was more active than compound 35 with para-oriented pyridine linker (L15), but both of
(
them were approximately 6 to 10-fold less active than compound 21. Taken together, these
findings suggested that small or large aromatic linkers are superior to heteroaromatic linkers to
10

tether the 1,3-diamine fragment (F01), therein unsubstituted para-phenylenebis(methylene)
moiety is an optimal linker (L03). Moreover, the incorporation of electron-withdrawing
substituents might reduce the electron density of the central phenyl ring and then impair the
CXCR4 binding ability of symmetrical polyamine compounds. In contrast, electron-donating
substituents on the central phenyl ring are tolerable for retaining excellent CXCR4 binding
probably because they may increase the electron density. Among these polyamine-based
compounds, compound 21 was further examined in other biological activities due to its excellent
CXCR4-binding activity.
Table 1. CXCR4 receptor binding activities of symmetrical polyamines.
a
Compound Fragment
Linker
Fragment Binding activities
IC50 (nM)
5
9
F01
F01
F01
F01
F01
F01
1697.7 ± 548.8
699.4 ± 249.4
19.8 ± 9.4
21
11

22
23
24
F02
F03
F01
F02
F03
F01
>10000
>10000
312.3 ± 126.9
2
5
6
F01
F01
F01
F01
155.5 ± 73.3
67.5 ± 18.6
L05
2
2
7
8
F01
F01
F01
F01
40.8 ± 7.3
2
354.0 ± 71.1
2
9
0
F01
F01
F01
F01
68.9 ± 19.9
48.0 ± 9.2
3
N
L10
12

31
32
33
F01
F01
F01
NO2
F01
F01
F01
147.3 ± 1.8
81.2 ± 27.7
67.6 ± 14.0
L11
3
4
F01
F01
119.3 ± 26.7
193.6 ± 74.3
3
5
6
F01
F01
F01
F01
N
L15
3
121.6 ± 95.0
325.3 ± 71.9
Positive control
AMD3100
a
Anti-CXCR4 mAb 12G5 competitive binding assay in CXCR4 expressing CHO Cells.
Competitive CXCR4-binding assays were performed in duplicate and the results were presented
as means ± SD. Experimental data were generated from at least three independent experiments.
2
.4. Calcium mobilization assay
The activation of CXCR4 by SDF-1α results in an important physiological response which is
2
+
the transient increase of intracellular Ca concentration. Inhibition assay of SDF-1α-induced
calcium signal is commonly used to evaluate the antagonism of CXCR4. We performed the
calcium mobilization assay to measure the antagonistic property of compound 21. AMD3100
13

was used as a positive control in a parallel assay. While compound 21 was more potent in
competing with CXCR4 binding of mAb 12G5 than AMD3100 as described above (mean IC of
5
0
1
9.8 nM vs. 325.3 nM), compound 21 was weaker in antagonizing the calcium mobilization
(
mean IC of 119.2 nM vs. 2.9 nM, Figure 2A-B). This difference in the ability of compound 21
5
0
in blocking the CXCR4 binding of anti-CXCR4 mAb 12G5 versus the calcium mobilization
triggered by SDF-1α was probably due to the difference in the receptor binding sites of 12G5
versus SDF-1α and compound 21 preferentially recognizing the CXCR4 binding site of 12G5
whereas AMD3100 would act in the opposite manner. Taken together, these results suggest that
compound 21 is an effective CXCR4 antagonist.
2
+
2+
Figure 2. Effects of compound 21 on calcium (Ca ) mobilization induced by SDF-1α. Ca
influx was measured in SUP-T1 cells treated with various concentrations of AMD3100 (A) or
compound 21 (B) in the presence of 50 nM of SDF-1α. The results were from at least two
independent experiments.
2
.5. Cell migration assay
To further investigate the antagonistic activity of compound 21, SDF-1α-induced cell
migration assay was performed using SUP-T1 cells to evaluate the effect of compound 21 on
SDF-1α/CXCR4-mediated cell migration. As shown in Figure 3A-B, compound 21 inhibited
14

SDF-1α-induced cell migration (mean IC = 621.4 nM) at the equivalent level compared with
5
0
the positive control AMD3100 (mean IC = 608.2 nM). Again it was noted that compound 21
5
0
had a differentiating ability in blocking 12G5 binding versus SDF-1α-induced effect likely
owing to the distinctive binding sites of 12G5 and SDF-1α on CXCR4. Therefore, this result
further supported that compound 21 has significant CXCR4 antagonistic activity.
Figure 3. SDF-1α-induced cell migration assays. Cell migration rate was expressed as the
percentage of SUP-T1 cells migrated from the upper chamber to the lower chamber. (A)-(B)
inhibitory effects of AMD3100, and compound 21 on cell migration in a dose-response manner.
Experimental results are the mean values generated from three independent experiments.
2
.6. Anti-HIV-1 activity assay
Compound 21 was shown to be a potent CXCR4 antagonist in the cellular functional studies
described above, we then investigated whether this compound could prevent HIV-1 entry via the
co-receptor CXCR4. By using a CEM-GEP reporter gene system to monitor HIV-1 infection,
compound 21 was found to inhibit the HIV-1 infection with a mean IC value of 1.5 µM (Figure
5
0
4
1
A), while showing no toxic effect on cell viability in the assay even at the high concentration of
0 µM (Figure 4B). These results suggested that compound 21 acting as a CXCR4 antagonist
can be used for blocking HIV-1 entry.
15

Figure 4. Anti-HIV-1 activity of compound 21 was detected in the CEM-GFP reporter system.
A) Compound 21 showed a dose-dependent inhibition in blocking X4-tropic HIV-1 infection
(
and replication. (B) The effect of compound 21 on CEM-GEP cells at different concentrations
used for blocking HIV-1 infection. Experimental data were generated from three independent
experiments.
2
.7. CXCR4 mutagenesis studies and binding site analyses
To determine the binding mode of compound 21, we carried out site-directed mutagenesis
studies to analyze the key binding sites of compound 21 on CXCR4. In previous studies, three
acidic residues (D97, D187, and E288) have been identified to be critical for SDF-1α/CXCR4
interaction and activation [25-27]. More importantly, the IT1t-CXCR4 and CVX15-CXCR4
cocrystal structures have confirmed that two ligands can form several salt bridges with
negatively charged residues D97, D171, D262, E288, etc. [20]. Recently, a mutation study has
also demonstrated that the binding assay using antibody 12G5 is more preferable for probing the
impact of certain mutants (W94A, D97A, D262, E288A, etc.) [28]. As such, we chose
deliberately these mutations in transmembrane helix (TM) other than extracellular loops (ECLs),
in order to avoid the possible effect on 12G5 binding, because ECLs are mainly the recognition
16

epitopes of CXCR4-specific 12G5. These CXCR4 mutants including W94A, D97A, D171A,
D262A, and E288D were selected to perform the 12G5-based competitive binding assays as
descried in our previous publications [24, 29]. In addition, since these selective mutants were
shown to have stable expression levels comparable to that of wild-type (WT) CXCR4 with no
significant effect on the binding to 12G5, they were deemed to be suitable for the binding site
analysis studies, as previously discussed in the publications by us and other groups [24, 28-30].
As depicted in Figure 5, we have assessed the binding ability of compound 21 on the CXCR4
mutants. Only W94A (TM2) and D171A (TM4) mutations reduced remarkably the binding
affinity of compound 21, just as they increased an excess of 60% 12G5 binding in comparison
with that of WT CXCR4. Likewise, the D97A (TM2) and E288D (TM7) decreased the potency
of compound 21 by 40% and 30%, respectively. The two mutations D193A and W195A located
at the junction of TM5 and ECL2 and showed weak effect on the potency of compound 21 (10-
2
0% reduction), and the other mutations at the N-terminus of CXCR4 (R30A), TM3 (H113A),
and TM6 (D262A), had little or no effect on the compound binding. These results indicated that
the four residues (W94, D97, D171, E288) could be important binding sites for compound 21 on
CXCR4. In view of the small molecular size of compound 21, we speculated that compound 21
might not only primarily bind the minor subpocket of CXCR4 by forming key interactions with
residues W94, D97, and E288, but also extend into the major subpocket of CXCR4 by
interacting with residue D171.
17

Figure 5. Binding of compound 21 (200 nM) to wild-type (WT) CXCR4 and mutants was
measured in 12G5 competitive binding assays. Experimental data represent the mean values of
three independent experiments (the error bars indicating the standard deviations).
2
.8. Molecular docking and predicted binding mode
Finally, to further understand the binding mode of compound 21, we carried out computational
docking study using the SYBYL-X 2.1 package and the co-crystal structure of CXCR4-vMIP-II
complex (PDB code 4RWS) [21]. Through various attempts, a protomol docking pocket in the
Surflex-Dock program was generated based on amino acid residues. Taking into the
consideration of the structural flexibility of compound 21 and accordance with mutational data,
an optimal pose was chosen from the 20 generated poses and presented in the protein structure of
CXCR4. As illustrated in Figure 6, molecular docking simulation showed that two secondary
amino groups in [3-(cyclohexylamino)propyl]amino fragment (F01) of compound 21 could form
hydrogen-bond interactions with two key acidic residues D97 and E288, and that the terminal
cyclohexyl ring of fragment (F01) generated hydrophobic interaction with W94 and then
18

anchored at the hydrophobic pocket surrounded by a set of residues W94, W102, V112, H113,
and T116. Of particular note was that the central phenyl ring (linker L03) along with fragment
(F01) favorably occupied the minor subpocket comprising of W94, D97, W102, V112, H113,
Y116, and E288. Another side fragment (F01) of compound 21 may naturally stretch into the
major subpocket via the formation of hydrogen-bonding or polar interactions with residues D171
and R188. These docking results showed that compound 21 could primarily occupy the minor
subpocket by interacting with key residues W94, D97, and E288, while partially binding to the
major subpocket by contacting with residue D171 in the major subpocket. This is in agreement
with our mutational data. To sum up, the molecular docking analysis explained the binding
mode of compound 21 and its potent activity, which would assist further structural design of
more effective analogs.
19

Figure 6. The predicted binding mode of compound 21 in the CXCR4 crystal structure (PDB
code 4RWS). Hydrogen bonds between docking pose of compound 21 (in green) and amino acid
residues are showed with dashed lines (in red). The docking result was visualized by PyMOL.
3
. Discussion
Here we report a new class of symmetrical polyamines (5, 9, 21-36) consisting of central
aromatic linkers (L01-L16) and two small fragments on the sides (F01-F03). Most compounds
showed favourable CXCR4-binding activities, with compound 21 being among the most potent
CXCR4 ligands possibly owing to its unique structural features. The mechanism of action and
SAR of compound 21 were studied by an integrated approach combining competitive receptor
binding, site-directed mutagenesis, and computational docking calculation. Among the structural
components of compound 21, the [3-(cyclohexylamino)propyl]amino fragment (F01) could be
deemed as a privileged fragment that frequently appears in known small-molecule CXCR4
antagonists [19]. The para-phenylenebis(methylene) linker (L03) is an optimal linker that has
also been reflected in the AMD3100 series of analogs and other reported symmetrical CXCR4
antagonists [31-37]. Herein, the recombination between two same fragments (F01) and linker
(L03) confers compound 21 with excellent CXCR4-binding ability attributing to these
interactions with key residues W94, D97, D171, and E288. This is supported by mutation and
molecular docking experiments (Figures 5-6). From the viewpoint of SAR analysis and binding
mode of compound 21, when two side fragments (F01) of compound 21 were simultaneously
replaced with two truncated monoamine fragments (F02 and F03), this resulted in compounds 22
and 23 with no CXCR4 binding activities (Table 1), probably due to the loss of these key
interactions with the aforementioned residues. Compared to linker (L03) in compound 21,
terephthaloyl linker (L01) and para-methylbenzoyl linker (L02) in compounds 5 and 9 (Table 1)
20

restricted the conformational flexibility of the central linker which did not benefit these
compounds to form interactions with key residues in the minor and major subpockets. Next,
benzene-1,3,5-triyl linker (L04) and meta-phenylenebis(methylene) linker (L05) in compounds
2
4-25 changed the orientation of linking fragment, showing relatively weaker CXCR4-binding
ability because this change impaired the active conformations of the compounds. Furthermore,
the incorporation of 4,4’-biphenylenebis(methylene) linker (L06) and 1,4-
naphthalenebis(methylene) linker (L07) increased the length and bulkiness of the central linker,
respectively. Introducing electron-withdrawing/electron-donating groups (-NO , -NH , -
NH(CH ), -N(CH ) and heterocyclic nitrogen into the linker L03 yielded different linker
2
2
3
3 2
)
variants (L08-L16), causing the reduction or enhancement in the electron density and steric
hindrance of the central linker. These alterations may exert different degrees of influence on the
binding poses of the two side fragments (F01) in the minor and major subpockets, thereby
making various levels of decrease in the CXCR4 binding affinities in compounds (26-36).
Overall, the optimal combination between fragment (F01) and linker (L03) facilitates the
binding of compound 21 to the minor and major subpockets through the compound’s basicity
and flexibility. These features endow compound 21 with significant antagonistic activity in both
calcium mobilization and cell migration mediated by SDF-1α/CXCR4 axis. A number of
symmetrical small-molecule CXCR4 antagonists reported to date, such as dipyridine-based
compounds [33, 37], dipyrimidine-based compounds [35], and methylpiperidine-based
compounds [34, 36], have exhibited various bioactivities. We think that compound 21 developed
here in this work may serve as a new important complement to the evolving symmetrical
CXCR4 antagonists.
21

The only clinically approved CXCR4 antagonist at this time, AMD3100 contains two bulky
,4,8,11-tetraazacyclotetradecane (cyclam) rings possessing high positive charge at physiological
1
pH and metal-chelating property, which is probably related to the dose-limiting cardiotoxicity
and poor oral bioavailability of the drug [15, 38, 39]. In comparison, compound 21 with two 1,3-
diamine fragments may be regarded as a downsizing congener of AMD3100 that has only half of
nitrogen atoms and positive charges, such that compound 21 may have no metal-chelating
ability. Another notable difference between AMD3100 and compound 21 is that mutational
studies and molecular modeling reveal AMD3100 interaction with three key residues D171,
D262, and E288 in the major subpocket [40-42], whereas compound 21 primarily occupies the
minor subpocket by interacting with W94, D97, and E288 (Figure 6). The CXCR4 binding 12G5
is shown to be a conformation-dependent mAb recognizing the domains of the extracellular
loops 1 and 2 of CXCR4 [43-45]. On the other hand, SDF-1α binding to CXCR4 is thought to be
mediated by two distinct and sequential interactions, the first involving the N loop and 40s loop
of SDF-1α recognizing the N terminus of CXCR4 and the second involving the flexible N-
terminal domain of SDF-1α inserting into the TM binding pocket of CXCR4 and resulting in
receptor activation and downstream signaling [46-48]. Compound 21 with a relatively small
molecular dimension is able to bind to the minor subpocket located at the deeper TM domain
whereas AMD3100 featuring a larger molecular size is inclined to target the major subpocket
located at the shallower TM domain [42]. Compound 21 in the minor subpocket might possibly
change CXCR4 conformation, thereby competing effectively with 12G5 to bind CXCR4. In a
possible different manner, AMD3100 in the major subpocket might partially shield the 12G5
recognition site on CXCR4 while more preferably block SDF-1α-CXCR4 interaction. These
differences may further explain why compound 21 and AMD3100 possess divergent abilities in
22

the 12G5 binding assay versus the SDF-1α-induced functional assay. Although compound 21 did
not display a stronger inhibition than AMD3100 in calcium mobilization assay, the SDF-1α-
induced cell migration assay revealed that compound 21 had a comparable anti-migration ability
to that of AMD3100 (Figures 2-3). Furthermore, cell-based HIV-1 assay of compound 21
showed that it could block HIV-1 infection mediated by CXCR4 at the low micromolar range
(
Figure 4). These results suggested that compound 21 bearing simple structural feature is capable
of retaining potent CXCR4 antagonistic activity. Recently, utilizing AMD3100 as a CXCR4-
targeting warhead, a growing number of CXCR4-targeted nanocarriers and imaging agents have
been developed for tumor-targeting diagnosis and therapy [49, 50]. In view of this and also
considering the concise and efficient character of the structure of compound 21 and the clinical
limitations of long-term use of AMD3100, we think that compound 21 may provide another
alternative small-molecule template for developing new CXCR4-targeting probes and
therapeutics.
4
. Conclusion
In this work, we designed and synthesized a new class of small molecule CXCR4 antagonists
featuring structurally simple and symmetrical polyamines (5, 9, 21-36). Among them, compound
1 stands out for its remarkable binding potency (mean IC50 = 19.8 nM) and displays no
2
cytotoxicity up to 10 µM that is almost an order of magnitude above its effective receptor
binding concentrations. For the functional evaluation, compound 21 significantly inhibits the
calcium mobilization (mean IC = 119.2 nM) and cell migration (mean IC = 621.4 nM). In
5
0
50
vitro HIV-1 assay also shows that compound 21 can prevent HIV-1 infection via CXCR4. These
results suggest that compound 21 is an effective small-molecule CXCR4 antagonist. SAR
analysis, together with site-directed mutagenesis and molecular docking study reveal the binding
23

mode of compound 21 where it primarily occupies the minor subpocket involving key
interactions with residues W94, D97, and E288, and partially stretches into the major subpocket
by contacting with residue D171. This binding model for compound 21 may provide new
insights for designing novel small molecule CXCR4 antagonists through fragment-based
strategy. Taken together, our findings demonstrate that compound 21 may merit further
development into a new class of therapeutic agents for the clinical interventions of pathological
and physiological processes where CXCR4 plays a role.
5
. Experimental section
5
.1. Chemistry
All chemical materials, reagents, and anhydrous solvents were commercially available and
used directly without further purification. Unless otherwise noted, all reactions were carried out
under an nitrogen atmosphere, and monitored by analytical thin layer chromatography (TLC)
using silica gel GF254 thin-layer plates (Qingdao Marine Chemical Factory, China). TLC plates
were observed by UV light (254 nm) or stained with alkaline KMnO solution. The flash column
4
chromatography was filled with 100-200 mesh (or 200-300 mesh) silica gel (Qingdao Marine
Chemical Factory, China) and then performed on CombiFlash Rf+ Flash Chromatography
System (Teledyne Isco, USA). Preparative reversed-phase liquid chromatography (RP-LC) was
also performed on CombiFlash Rf+ System equipped with two types of C18 columns (RediSep
Rf columns, Teledyne Isco, USA), such as 13 g C18 column with 30 mL/min flow rate, 26 g C18
column with 35 mL/min flow rate, and the mobile phase was composed of acetonitrile (contains
1
13
0
.1%TFA) and water (contains 0.1%TFA). H NMR and C NMR spectra were recorded on a
Bruker AV 400 instrument using tetramethylsilane (TMS δ 0.00) as an internal standard, and
deuterium oxide (D O), or dimethyl sulfoxide- (DMSO- ), or Chloroform-d (CDCl ) were
2
d
6
d
6
3
24

used as the deuterated solvents. High-resolution mass spectrometry (HR-MS) data were obtained
using positive mode ESI on a Waters Xevo G2 QTof instrument (the Drug Analysis Facility of
Tsinghua University).
5
.1.1. Syntheses of intermediates 2-4 and compound 5
A mixture of 3-aminopropan-1-ol (1, 5.00 g, 66.57 mmol), cyclohexanone (7.84 g, 79.88
mmol) and 4Å molecular sieves (several particles) in anhydrous MeOH was stirred for 6 h at
7
1
0°C. After being filtered through a celite pad, the filtrate was treated with NaBH
33.14 mmol) at 0-5°C and stirred continuously for 3 h at room temperature. The resulting
Cl solution and concentrated directly under
4
(5.04 g,
reaction mixture was quenched with saturated NH
4
reduced pressure to dryness. The residue was suspended in methanol and the insoluble solid was
filtered out. The collected filtrate was reconcentrated under reduced pressure. The residue was
resuspended in tert-butanol/water (60 mL/20 mL), and then 2 N Na CO aqueous solution (30
2
3
mL) and (Boc) O (18.90 g, 86.54 mmol) were added in batches to the suspension. The reaction
2
mixture was stirred for 12 h at room temperature. The resulting mixture was diluted with water
(200 mL) and extracted with EtOAc (200 mL). The organic layer was separated and dried over
anhydrous Na SO , filtered, and concentrated under reduced pressure. The residue was purified
2
4
by flash column chromatography (eluent, 0-5% methanol in dichloromethane) to obtain the
desired intermediate 2 (12.0 g, 71% yield over two steps) as pale yellow oil.
To a mixture of intermediate 2 (4.60g, 17.87 mL) and TEA (7.50 mL, 53.61 mmol) in
anhydrous DCM (80 mL) was slowly added MsCl (2.1 mL, 26.81 mmol) at 0°C. After the
reaction mixture was stirred for 1.0 h, the ice-bath was removed and the mixture was
continuously stirred for 5.0 h at room temperature. The resulting mixture was quenched and
2 4
washed with water. The separated organic layer was dried over anhydrous Na SO and
25

concentrated under reduced pressure to yield the red brown oil, which was directly dissolved in
DMF (80 mL), NaN (3.40 g, 52.77 mmol) was added to the solution. Then the mixture was
3
stirred for 16 h at 60°C, water (150 mL) and EtOAc (150 mL) were added to the resulting
reaction mixture. The organic layer was separated and washed with saturated NaCl solution,
dried over anhydrous Na SO , filtered, and concentrated under reduced pressure. The residue
2
4
was purified by flash column chromatography (eluent, 0-50% dichloromethane in petroleum
1
ether) to provide the intermediate 3 (3.5 g, 69% yield over two steps) as pale yellow oil. H NMR
(
400 MHz, CDCl
H), 1.46 (s, 9H), 1.41 – 1.25 (m, 4H), 1.14 – 1.02 (m, 1H).
0%Pd-C (100 mg) was added to a mixed solution of intermediate 3 (1.00 g, 3.54 mmol) in
3
) δ 3.88 (br-s, 1H), 3.30 (t, J = 6.7 Hz, 2H), 3.15 (br-s, 2H), 1.84 – 1.60 (m,
7
1
EtOH/EtOAc (15 mL/15 mL), and then the mixture was stirred for 16 h at room temperature
under H atmosphere. After being filtered through a celite pad, the filtrate was concentrated
2
1
under reduced pressure to obtain the desired intermediate 4 (885 mg, 97% yield) as brown oil. H
NMR (400 MHz, CDCl3)
δ
3.80 (br., 1H), 3.06 (br, 2H), 2.64 (t,
J = 6.9 Hz, 2H), 1.78 – 1.52 (m,
7
H), 1.40 (s, 9H), 1.34 – 1.22 (m, 4H), 1.10 – 0.93 (m, 1H). HRMS (ESI, m/z) calcd. for
+
C
14
H
28
N
2
O
2
257.2151 [M + 1] , found 257.2238.
N1,N4-bis(3-(cyclohexylamino)propyl)terephthalamide hydrochloride salt (compound 5). A
solution of terephthaloyl dichloride (100 mg, 0.49 mmol) in anhydrous THF (3 mL) was added
to a stirred mixture of intermediate 4 (251 mg, 0.98 mmol) and TEA (204 µL, 1.47 mmol) in
anhydrous THF (3 mL) at 0°C, and then the reaction mixture was warmed to room temperature
and stirred for 12 h. The resulting mixture was quenched with water and extracted with EtOAc.
2 4
The organic layer was dried over anhydrous Na SO , filtered, and concentrated under reduced
pressure to yield the colorless oil, which was dissolved in DCM (3.0 mL), 4 N HCl in 1,4-
26

dioxane solution (3.0 mL) was added to the DCM solution. Then the solution was stirred for 2.0
h and concentrated under reduced pressure. The obtained solid was triturated and washed with
dichloromethane/diethyl ether (1:1, v/v), after vacuum drying, compound 5 was obtained as a
1
white solid (121 mg, 48% yield). H NMR (400 MHz, D O)
δ
7.78 (s, 4H), 3.45 (t,
= 12.4 Hz, 2H),
170.35 (s), 136.54 (s),
J = 6.6 Hz,
2
4
1
1
H), 3.14 – 2.96 (m, 6H), 2.07 – 1.88 (m, 8H), 1.83 – 1.69 (m, 4H), 1.59 (d, J
1
3
.34 – 1.17 (m, 8H), 1.16 – 1.05 (m, 2H). C NMR (101 MHz, D O)
δ
2
27.48 (s), 57.27 (s), 41.83 (s), 36.71 (s), 28.89 (s), 25.83 (s), 24.48 (s), 23.87 (s). HRMS (ESI,
+
m/z) calcd. for C26
H
42
N O
4 2
443.3308 [M + 1] , found 443.3380.
5
.1.2. Syntheses of intermediates 7-8 and compound 9
Following the same NaBH -mediated reductive amination and Boc protection procedure
4
described in the synthesis of intermediate 2, intermediate 7 (10.6 g, 41% yield) was obtained
1
starting with 4-diethoxymethyl)benzaldehyde (6) and
NMR (400 MHz, CDCl3) 7.42 (d,
H), 3.85 (br., 1H), 3.65 – 3.57 (m, 2H), 3.56 – 3.47 (m, 2H), 3.23 – 2.96 (m, 4H), 1.80 – 1.59
m, 8H), 1.54 –1.43 (m, 18H), 1.35 – 1.26 (m, 2H), 1.23 (t, = 7.1 Hz, 6H), 1.09 – 1.01 (m, 1H),
.90 – 0.82 (m, 1H).
TsOH∙H O (1.00 equiv.) was added to a solution of intermediate 7 (1.00 equiv.) in THF, and
N
-(3-aminopropyl)cyclohexylamine. H
δ
J = 7.9 Hz, 2H), 7.22 (br-s, 2H), 5.48 (s, 1H), 4.43 (br-s,
2
(
J
0
2
then the mixture was stirred for 1 h at 0-5°C. The resulting mixture was quenched with saturated
NaHCO aqueous solution, and extracted with EtOAc. The collected organic layer was washed
3
with brine, dried over anhydrous Na SO , filtered, and concentrated under reduced pressure to
2
4
obtain an aldehyde intermediate, which was dissolved in acetone, and KMnO (1.50 equiv.) was
4
added. Then the mixture was stirred for 1.0 h at room temperature, quenched with sat. Na
2 2 5
S O
solution, diluted with EtOAc, and filtered through a celite pad. The organic layer was separated
27

and dried over anhydrous Na
2
SO
4
, filtered, and concentrated under reduced pressure to yield
1
intermediate 8 (802 mg, 64% yield over two steps) as an off-white fluffy solid. H NMR (400
MHz, DMSO- 6
d
)
δ
7.87 (br-s, 2H), 7.17 (d,
J = 5.9 Hz, 2H), 4.38 (s, 2H), 3.64 (br., 1H), 3.15 –
2
1
.86 (m, 4H), 1.75 – 1.48 (m, 8H), 1.43 – 1.33 (m, 18H), 1.26 – 1.14 (m, 3H), 1.09 – 1.00 (m,
-
H). HRMS (ESI, m/z) calcd. for C H N O 489.3043 [M - 1] , found 489.2953.
2
7
42
2
6
N-(3-(cyclohexylamino)propyl)-4-(((3-(cyclohexylamino)propyl)amino)methyl)benzamide
hydrochloride salt (compound 9). A mixture of intermediate 8 (1.00 equiv.), intermediate 4 (1.00
equiv.), HATU (1.50 equiv), and DIPEA (2.00 equiv) in anhydrous DMF was stirred for 12 h at
ambient temperature. The resulting mixture was diluted with water and extracted with EtOAc.
The collected organic layer was dried over anhydrous Na SO , filtered, and concentrated under
2
4
reduced pressure. The residue was purified by preparative TLC (developing solvent, petroleum
ether/ethyl acetate) to get the corresponding intermediate which was dissolved directly in DCM,
and the solution was treated with 4 N HCl in 1,4-dioxane solution for 2-4 h. The resulting
solution was evaporated to dryness under reduced pressure. The residue was triturated and
washed with dichloromethane/diethyl ether (1:1, v/v) to afford the title compound 9 (39 mg, 36%
1
yield over two steps) as a white solid. H NMR (400 MHz, D
2
O)
= 6.5 Hz, 2H), 3.14 (t,
m, 6H), 2.09 – 1.90 (m, 8H), 1.81 – 1.71 (m, 4H), 1.59 (d, = 12.4 Hz, 2H), 1.33 – 1.18 (m,
170.53 (s), 134.54 (s), 134.33 (s), 130.15
δ
7.76 (d,
J = 8.2 Hz, 2H), 7.53
(d,
J
= 8.2 Hz, 2H), 4.27 (s, 2H), 3.45 (t,
J
J
= 6.5 Hz, 2H), 3.11 – 2.97
(
J
13
8
H), 1.16 – 1.05 (m, 2H). C NMR (101 MHz, D O) δ
2
(s), 127.91 (s), 57.45 (s), 57.27 (s), 50.67 (s), 44.22 (s), 41.84 (s), 41.20 (s), 36.69 (s), 28.89 (s),
2
8.80 (s), 25.85 (s), 24.48 (s), 24.43 (s), 23.87 (s), 23.85 (s), 22.83 (s). HRMS (ESI, m/z) calcd.
+
for C H N O 452.3515 [M + H + Na] , found 452.3394.
2
6
44
4
5
.1.3. General synthetic procedure for intermediates 10-20 and compounds 21-36
28