Function-Oriented Development of CXCR4 Antagonists as Selective Human Immunodeficiency Virus (HIV)-1 Entry Inhibitors

Abstract

Motivated by the pivotal role of CXCR4 as an HIV entry coreceptor, we herein report a de novo hit-to-lead effort on the identification of subnanomolar purine-based CXCR4 antagonists against HIV-1 infection. Compound 24, with an EC₅₀ of 0.5 nM against HIV-1 entry into host cells and an IC₅₀ of 16.4 nM for inhibition of radioligand stromal-derived factor-1α (SDF-1α) binding to CXCR4, was also found to be highly selective against closely related chemokine receptors. We rationalized that compound 24 complementarily interacted with the critical CXCR4 residues that are essential for binding to HIV-1 gp120 V3 loop and subsequent viral entry. Compound 24 showed a 130-fold increase in anti-HIV activity compared to that of the marketed CXCR4 antagonist, AMD3100 (Plerixafor), whereas both compounds exhibited similar potency in mobilization of CXCR4⁺/CD34⁺ stem cells at a high dose. Our study offers insight into the design of anti-HIV therapeutics devoid of major interference with SDF-1α function.

Graphical abstract

INTRODUCTION

The battle against HIV/AIDS remains a formidable task, and it is of paramount importance to develop efficacious therapies that result in minimal resistance. With significant progress in elucidating molecular insights into HIV pathogenesis and druggable targets, the highly active antiretroviral therapy (HAART) arsenal has evolved beyond the traditional drugs that target three main viral enzymes: reverse transcriptase, integrase, and protease.^1,2 In particular, the HIV entry mechanism involving multiple conformational changes has provided potential targets and propelled a plethora of therapeutic developments for the disruption of HIV viral attachment, co-receptor binding, and fusion.^3–8 HIV-1 infection is initiated by the association of viral glycoprotein 120 (gp120) with CD4 cell receptor, which, in turn, triggers a conformational change in gp120, exposing the third variable loop (V3 loop) of gp120 and allowing it to bind to chemokine receptors including T cell-tropic CXCR4 or macrophage-tropic CCR5. The subsequent conformational change in gp41 leads to a fusion of the viral envelope and host cell membrane.⁹ Indeed, enfuvirtide, a peptidomimetic of gp41, was approved by the FDA in 2003 to block HIV-1 viral fusion.^10,11 A decade after uncovering the critical roles of chemokine receptors CXCR4 and CCR5 in mediating HIV entry, the first-in-class CCR5 chemokine receptor antagonist maraviroc was approved in 2007 to provide an addition to the anti-HIV treatment arsenal.¹² The clinical observation of the predominant CXCR4-utilizing strains in HIV-1 infected patients after maraviroc administration¹³ suggests a mixed-tropic viral population and thus the necessity for the development of CXCR4 antagonists for complete viral suppression.

The induced chemotactic signaling mediated by the chemokine SDF-1α (also known as C-X-C motif chemokine 12α, CXCL12α) and its receptor CXCR4 is of significant biological importance and is involved in tumor metastasis, angiogenesis, progression, and survival.^14–16 Interestingly, this pathway is also exploited by pathogens to alter the signaling patterns of the hosts in the progression of various diseases.^17,18 The need for selective CXCR4 antagonists is imperative not only for therapeutic applications to block HIV-1 entry but also to modulate the SDF-1α/CXCR4 axis’ involvement in tumor metastasis, angiogenesis, progression, and survival.^14–16 Notably, in about 50% of late-stage HIV-1-infected patients, HIV-1 uses either CXCR4 alone or in combination with CCR5 to facilitate viral entry into host cells and accelerate disease progression.¹⁹ The emergence of CXCR4-utilizing strains also coincides with the onset of immune deficiency, accompanied by a marked drop in the CD4⁺ T-cell count, thereby facilitating HIV-1 replication.^20–22 It was shown that SDF-1α, at low concentration, could effectively disrupt the association of HIV-1 with CXCR4 and therefore reduce the infection.^23–25 The disruption comes from the steric hindrance caused by the binding of SDF-1α to CXCR4. Furthermore, SDF-1α-mediated downregulation of cell surface CXCR4, by inducing its endocytosis, could inhibit HIV infection.²⁶ CXCR4 has been validated as a viable target because selective CXCR4 antagonists, such as AMD3100 (Plerixafor)²⁷ or orally bioavailable AMD0070,²⁸ could significantly reduce the viral load in T-tropic (X4) HIV-infected patients. However, AMD3100 failed as an HIV drug in phase II clinical trials due to a lack of oral availability and cardiac disturbance.

The recently reported co-crystal structures of CXCR4 with its antagonists,²⁹ IT1t, a small molecule, and CVX15, a cyclic peptide, have provided key design features for new compounds. Both SDF-1α and the V3 loop of HIV-1 gp120 complement a substantial portion of the CXCR4 acidic extracellular domain, forming multiple salt-bridge contacts predominantly with the aspartate and glutamate residues.^18,29,30 In turn, a myriad of highly basic CXCR4 antagonists were developed to exploit these charge–charge interactions because they play pivotal roles in binding to the CXCR4 receptor.^31,32 To identify efficacious agents against T-tropic (X4) HIV-1 infections, herein, from the discovery of quinazoline-based polyamine CXCR4 antagonists as HIV-1 entry inhibitors, we will describe the design, synthesis, and structure–activity relationships (SAR) culminating in a novel series of HIV-1-selective, CXCR4-specific, purine-based antagonists with a broad therapeutic window. Our study provides tantalizing insights into developing antagonists that selectively target key CXCR4 residues that govern the HIV-1 entry process.

CHEMISTRY

Side chains A–H in Figure 1 and test compounds 1−8 in Table 1 were prepared according to a general synthetic route shown in previous literature.³³ Test compounds 9−11 were prepared according to a general synthetic method shown in Scheme 1 using compound Ia and its corresponding 2,4-diamino quinazoline 9, respectively, as a typical example. The commercially available 2-amino-5-methoxy benzoic acid (Ia) was coupled with urea to provide 6-methoxy-quinazoline-2,4-diol (IIa) in 85% yield. Treatment of IIa with phosphorus oxychloride in the presence of 2-ethyl-pyridine as a base gave 2,4-dichloro-6-methoxy-quinazoline (IIIa), which, without purification, was first coupled with 4-amino-1-Boc-piperidine in a chemoselective manner to give intermediate IVa in 59% yield over two steps, followed by a second coupling with protected side chain D under microwave irradiation to afford Va in 63% yield. Subsequent acidic deprotection afforded desired compound 9 in 92% yield. Compounds 12−14 and 15−17 were synthesized, respectively, from IVa–IVc by coupling with side chains E and F and deprotecting with HCl/ether following a similar synthetic procedure as that for Va and 9.

Figure 1.

Side chains A–H.

Table 1.

Biological Evaluation of Quinazoline Core Polyamines Derivatives on CXCR4 Binding and HIV Inhibitory Profiles

Compound	R1	R2	R3	CXCR4 IC50 (nM)a	Anti-HIV EC50 (nM)	Cytotoxicity CC50 (μM)		Selective Index CC50/EC50
						CEM	TZM-bl	CEM	TZM-bl
1	H	H		877.5±7.8	161.2±22.9	2±0.8	ND	12	ND
2	H	H		50.0±14.3	111.1±11.2	5.2±1.3	4.3±0.3	47	39
3	H	H		83.5±10.8	48.9±8.0	22.6±1.2	16.9±1.9	462	346
4	H	H		36.2±1.4	68.4±8.0	>50	>50	>731	>731
5	H	H		44.8±1.4	58.7±7.6	23±4.8	42.2±3.7	392	719
6	H	H		35.2±3.0	71.5±29.0	3.2±0.8	3.4±0.1	45	48
7	H	H		3149.0±362.0	>500	37.5±7.0	24.6±0.5	ND	ND
8	H	H		>10000	>500	32.2±3.0	24.2±0.6	ND	ND
9	OMe	H		8.7±1.1	47.4 ±18.3	8±1.7	8.4±1.0	169	177
10	H	OMe		6.8±1.8	8.6±5.3	1.6±0.6	10.8±3.3	186	1256
11	OMe	OMe		31.5±3.0	21.7±3.0	17.5±2.4	23.1±7.0	806	1065
12	OMe	H		7.9±0.6	42.6±7.0	6.5±1.1	21.7±7.0	153	509
13	H	OMe		12.8±4.3	45.4±8.4	6.1±1.1	10.8±0.8	134	238
14	OMe	OMe		78.4±19.2	48.4±3.4	18.2±4.5	19.3±0.88	376	399
15	OMe	H		66.0±28.3	93.9±14.7	18.4±5.5	24.5±2.5	196	261
16	H	OMe		47.9±23.1	27.0±4.2	13.8±1.9	32.2±0.2	511	1193
17	OMe	OMe		87.8±15.0	26.2±4.4	17.6±3.0	28.1±1.5	672	1073
AMD3100	–	–	–	213.1±26.2	66.6±23.1	–	–	–	–
IT1t	–	–	–	100.6±23.1	>70	–	–	–	–

^aDetermined by 50% inhibition of radioligand [¹²⁵I]SDF-1α binding to hCXCR4-transfected HEK293 membrane; values represent the mean ± SD of at least three independent experiments.

Scheme 1. Synthetic Procedures for Quinazoline Compounds 9−11^a.

^aReagents and conditions: (a) urea, 200 °C, 2 h, 80–85%; (b) POCl₃, 2-ethyl-pyridine, 110 °C, 3 h; (c) 4-amino-1-Boc-piperidine, Et₃N, DCM, −5 °C to RT, 16 h, 54–59% over two steps; (d) side chain D, 1-pentanol, microwave, 120 °C, 15 min, 59–63%; (e) 1 N HCl in diethyl ether, DCM, 16 h, 83–92%.

Test compounds 18 and 21−27 were prepared according to a general synthetic method shown in Scheme 2, using compound 18 as a typical example. Commercially available 2,6-dichloropurine was protected with 3,4-dihydro-2H-pyran to provide 18-I in quantitative yield, which, in turn, was coupled with side chain D in a chemoselective manner to give intermediate 18-II in 62% yield, followed by a second coupling with piperazine under microwave irradiation (15 min, 100 °C) or heating at 100 °C in 1-pentanol for 15 h to afford 18-III in 64% yield. After acidic deprotection, compound 18 was obtained in 94% yield. Compounds 22 and 25−27 were synthesized, respectively, from 18-I by coupling with different side chains followed by a second coupling with piperazine and deprotection with HCl/ether in a procedure similar to that for 18. The synthesis of compound 19 is shown in Scheme 3.

Scheme 2. Synthetic Procedures for Compound 18^a.

^aReagents and conditions: (a) 3,4-dihydro-2H-pyran, 25 °C, 15 h, 100%; (b) side chain D, TEA, DCM, 50 °C, 4 h, 62%; (c) piperazine, 1-pentanol, 100 °C, 15 h, 64%; (d) 1 N HCl in diethyl ether, DCM, 16 h, 94%.

Scheme 3. Synthetic Procedure for Compound 19^a.

^aReagents and conditions: (a) side chain D, TEA, t-BuOH, 50 °C, 4 h, 89%; (b) MeI, K₂CO₃, DMF, 25 °C, 3 h, 97%; (c) piperazine, ethylene glycol monomethyl ether, 120 °C, 15 h, 69%; (d) 1 N HCl in diethyl ether, DCM, 16 h, 91%.

RESULTS AND DISCUSSION

Quinazoline-Based Antagonists

Our group has previously reported the ability of a series of potent quinazoline-based CXCR4 antagonists to mobilize stem cells.³³ In the present study, we evaluated the potential of these quinazoline-based CXCR4 antagonists (Table 1) in the blockade of HIV-1 entry by performing luciferase activity assays using TZM-bl cells (see Experimental Section). The binding affinities of compounds 1−5 toward the CXCR4 receptor increased significantly by 10- to 20-fold through terminal ring-size expansion from a three- to seven-membered ring, which also showed a relatively good correlation with their anti-HIV-1 activity. Despite the nitrogen atoms at the R³ position of compound 4 and 6’s terminal six-membered rings displaying distinct binding modes toward CXCR4, they both harnessed similar CXCR4 binding affinities and anti-HIV-1 activities. Compounds 7, with a terminal piperazine ring, and 8, with an exposed tertiary amine, both showed dramatic decreases in CXCR4 binding and anti-HIV activities, suggesting the existence of a hydrophobic pocket in CXCR4 to facilitate the terminal binding of the antagonists.

We further addressed the influence of the quinazoline core in CXCR4 binding by synthesizing a series of compounds, 9−17. We installed an electron-donating methoxy group at either R¹, R², or both R¹ and R² in combination with modifications at the R³ position by a terminal cyclohexyl, cycloheptyl, or piperidine ring, respectively (Table 1). Compared to compound 4, the addition of a methoxy group either at the R¹ or R² position led to a 4−5-fold increase in CXCR4 binding affinity for both compounds 9 and 10 with a terminal cyclohexyl ring and compounds 12 and 13 with a cycloheptyl ring at the R³ position. The terminal ring could situate in a hydrophobic cleft of CXCR4, and mounting a methoxy group to the quinazoline core might induce an allosteric conformational change in the compound to facilitate its interaction with CXCR4. Moreover, the increased electron density by the introduced methoxy group might play a role in the enhanced binding of the compound to CXCR4. However, double modifications at both the R¹ and R² positions with methoxy groups did not further increase, but only maintained (compound 11) or even decreased (compounds 14 and 17), the binding affinity of the compounds to CXCR4, suggesting that hydrophobic interactions could be involved in the binding. Moreover, the introduction of two methoxy groups could affect the binding orientations of the compounds by shifting the larger cycloheptyl ring (compound 14) out of position or extending the piperidine tertiary amine to an unfavorable binding mode (compound 17). Interestingly, such an effect was tolerated for compound 11 with the cyclohexyl ring, as CXCR4 binding affinity was maintained.

Despite the strong CXCR4 binding by compounds 9, 10, 12, and 13 (IC₅₀ ∼ 10 nM), only compound 10, with the terminal cyclohexyl ring, showed a significant 8-fold increase in the blockade of HIV-1 infection (EC₅₀ = 8.6 nM). The addition of a methoxy group at the R² position presumably allowed the cyclohexyl ring to adopt a favorable binding orientation, resulting in better interactions with key residue(s) on CXCR4 for HIV-1 entry. From the SAR studies, we identified that compound 10, a quinazoline-based CXCR4 antagonist, had a higher CXCR4 binding affinity and potent anti-HIV activity when compared to that of documented CXCR4 antagonists AMD3100 and IT1t (Table 1). However, with the exception of compound 4, the quinazoline-based antagonists had moderate cytotoxicities to CEM and TZM-bl cells and were deemed to be appropriate with regard to their replacement of the core and other functional groups. Taken together, we based the next design on the incorporation of a different core scaffold while maintaining the terminal cyclohexyl ring on the projected arm.

Purine-Based Antagonists

In an effort to increase the therapeutic index of the CXCR4 antagonists and maintain the extended two-arm projections of the quinazoline-based structures, we employed purine as a new core and synthesized a series of potent CXCR4 antagonists, 18−27 (Table 2). Gratifyingly, there was a significant increase of the therapeutic index, as many of the purine-based antagonists exhibited no toxicities at concentrations up to 50 μM toward CEM and TZM-bl cells. With the purine scaffold, we first tested the effects of side chain modifications. Compound 18A was first synthesized to carry two similar side chains as those in compound 4; however, its CXCR4 binding affinity was dramatically decreased to 230 nM (Figure 2). This is likely due to the loss of hydrophobic interactions of the purine scaffold to the CXCR4 cleft as compared to that mediated with the quinazoline core. Surprisingly, the CXCR4 binding affinity of compound 18 was restored to 18.1 nM by swapping the orientation of the two side chains, and its potency for anti-HIV-1 activity was improved to 2.0 nM (Table 2). The binding orientation of compound 18 may form strong interactions with residues in the extracellular domain of CXCR4 that are important for binding to the HIV-1 gp120 V3 loop.

Table 2.

Biological Evaluation of Purine Core Polyamines Derivatives on CXCR4 and HIV Profiles

Compound	R1	R2	n, m	CXCR4 IC50 (nM)a	Anti-HIV EC50 (nM)	Cytotoxicity CC50 (μM)		Selective Index CC50/EC50
						CEM	TZM-bl	CEM	TZM-bl
18	H	H	1, 1	18.1±2.8	2.0±0.1	>50	>50	>25000	>25000
19	Me	H	1, 1	86.2±28.8	3.4±0.2	20.2±1.5	18.6±2	5941	5470
20	Me	Me	1, 1	59.5±11.9	6.5±0.2	13.8±1.8	11.2±2.1	2123	1723
21	–	–	–	51.6±11.8	84.3±15.8	>50	>50	>593	>593
22	–	–	–	45.9±0.1	4.2±0.3	>50	>50	>11905	>11905
23	H	H	2, 1	45.1±3.7	3.1±0.7	46.7±3.2	>50	15065	>16129
24	H	H	1, 2	16.4±3.5	0.51±0.02	>50	>50	>98039	>98039
25	–	–	–	4.2±0.4	0.61±0.27	>50	>50	>83333	>83333
26	–	–	–	32.9±2.3	2.5±0.1	>50	>50	>20000	>20000
27	–	–	–	44.1±17.3	5.7±1.0	>50	>50	>8772	>8772

^aDetermined by 50% inhibition of radioligand [¹²⁵I]SDF-1 binding to hCXCR4-transfected HEK293 membrane; values represent the mean ± SD of at least three independent experiments.

Figure 2.

IC₅₀ of compounds 4, 18, and 18A. The enhancement of anti-HIV activity is illustrated by the fold change between the anti-HIV activity of AMD3100 to that of compounds 4 and 18.

Methyl group incorporation either at the R¹ position of the purine imidazole ring in compound 19 or at both the R¹ and piperazine ring-linked R² positions in compound 20 resulted in a 4−5-fold decrease in CXCR4 binding affinity. The potency of HIV-1 inhibition was generally maintained in the low nanomolar range, with a 2-fold decrease in anti-HIV activity as compared to that of compound 18. However, these methyl group modifications also led to unfavorable cytotoxicities. The maintained anti-HIV activities for compounds 18−20 implied the importance of the orientation and binding modes of the two side chains. The elongation of the lower side chain by 4-methylpiperidine in compound 21 led to a 2−3-fold decrease in CXCR4 binding and yet a 40-fold decrease in anti-HIV activity, suggesting a small binding pocket in the lower side chain that allows for limited substitutions.

We next turned our attention to the appropriate placement of the peripheral functionalities in the upper side chain of the purine-based antagonists (Table 2). Compound 22, with a pyridine substituted for the phenyl ring moiety, showed a 2-fold decrease in both CXCR4 binding and anti-HIV activities as compared to that of compound 18, suggesting that the N atom in the pyridine ring might not significantly change the binding role of the original phenyl ring moiety. To further access the possible binding motifs and to increase the anti-HIV activity, we investigated the effects of varying chain length by installing extra methylene groups at positions n and m of the upper functional chain. With an extra methylene group at position n, compound 23 showed an around 2-fold decrease in both CXCR4 binding and anti-HIV activities, suggesting that the orientation of the π–π stacking interactions arising from neighboring aromatic rings might be important in CXCR4 binding. However, with the addition of another methylene group at position m, compound 24 showed a 4-fold enhancement of anti-HIV activity (EC₅₀ = 0.5 nM), whereas its CXCR4 binding affinity was not changed compared to that of compound 18. The results suggested that the degree of rotation and flexibility of the terminal chain from the m position allow particular interactions with residues on the CXCR4 receptor that can closely control HIV-1 entry. Encouraged by the above findings, we envisioned that compound 25 with meta-substitution would harness a similar projection of the peripheral upper side chain as that of compound 24. Indeed, compound 25 exhibited a comparable anti-HIV activity as that of compound 24, whereas it demonstrated a 4-fold increase in CXCR4 binding affinity. The reduced correlation between CXCR4 binding affinity and anti-HIV activity for compounds 18, 24, and 25, as examples, suggested that additional mechanisms are involved in determining the anti-HIV activity. As compared to compound 25, compounds 26 and 27, with the same projection of the peripheral upper side chain, exhibited a 4−10-fold decrease in both CXCR4 binding and anti-HIV activities, suggesting that the presence of a tertiary amine in the pyridine moiety would either hinder the π–π stacking interactions between neighboring aromatic rings of the compound or interfere with the hydrophobic cleft of CXCR4. Interestingly, compounds 26, bearing a pyridine substitution and a meta-substitution of the peripheral upper side chain, showed similar CXCR4 binding and anti-HIV activities as that of compound 22. The above findings implied that the projection of the meta-substituted upper side chain from a phenyl ring in compound 25 produced a favorable binding orientation with CXCR4.

In comparison to the initially quinazoline-based compound 4, our hit-to-lead generation effort resulted in the identification of compound 24 with improved CXCR4 binding affinity and HIV-1 entry inhibition by 2.2- and 134-fold (Table 1 vs Table 2), respectively. Furthermore, we greatly reduced the cytotoxicity of the quinazoline-based CXCR4 antagonists by substitution with a purine core. Intriguingly, compound 18A, with a purine core swapped for the quinazoline core in compound 4, resulted in a significant loss in CXCR4 binding. However, compound 18, in which the two peripheral appendages in compound 18A were switched, showed restored CXCR4 binding activity. Moreover, such employment increased the anti-HIV-1 entry activity by 34.2-fold (Figure 2 vs Table 2). Taken together, from the SAR studies of the CXCR4 antagonists, we have shown (1) the importance of a terminal cyclohexyl ring group, (2) the cytotoxic effects of the quinazoline core, which can be resolved by replacement with a purine core, (3) the involvement of the orientation of the side chains with projected peripheral functionalities in CXCR4 binding and anti-HIV activities, and (4) the effects of hydrophobic interactions between the antagonists and CXCR4. Furthermore, the SAR studies could lead to the exploration of other potential binding sites on the CXCR4 receptor for HIV-1 entry by introducing flexibility and optimization of the upper side chain of the purine-based antagonists.

Time-of-Addition Assays

To validate whether these CXCR4 antagonists are HIV-1 entry inhibitors through interactions with CXCR4, we performed time-of-addition assays over one HIV replication cycle by monitoring the effects of compound 24 in TZM-bl cells. Compound 24, at various concentrations ranging from 0 to 20 nM, was added 3 h before, simultaneously with, or 3 h after HIV-1 infection and showed a dose-dependent inhibition of viral replication (Figure 3). When added 3 h before or simultaneously with HIV-1 infection, compound 24 maintained its potent anti-HIV activity at EC₅₀ concentrations, consistent with the observation in Table 2. In contrast, the anti-HIV activity of compound 24 was significantly reduced when added 3 h after HIV-1 infection. The pre- and cotreatments with compound 24 provided more protection against HIV-1 infection than did the post-treatment, confirming that the CXCR4 antagonist indeed protected against viral entry in the early phase of HIV-1 infection.

Figure 3.

Cell-based time-of-addition assay over one HIV replication cycle. Compound 24, at different concentrations, was added 3 h before, simultaneously with, or 3 h after HIV-1 infection. Inset is a magnified view of indicated portion of the main chart.

Molecular Dynamics Simulation Study

We then carried out molecular dynamics (MD) simulations to investigate the CXCR4 (PDB code 3OE0)²⁹ residues that are interacting with 24 to gain insights into the possible binding mode of compound 24 (Figure 4 and Figure S1). Intriguingly, compound 24 docked toward the N-terminal domain of CXCR4 and formed hydrogen bonds with CXCR4 residues Asp193, His281, and Glu288. As shown in Figure 4A, from the purine core to the lower piperazine ring of compound 24, two hydrogen bonds formed with CXCR4: the N3 of the purine core interacted with the side chain of His281 and the terminal piperazine nitrogen with the side chain of Glu288. In particular, Glu288 of CXCR4 has been shown to be one of the most important residues for HIV-1 co-receptor activity. A significant loss (>50%) of this activity was observed by the substitution of Glu288 with either alanine or even aspartic acid, which preserves the physicochemical properties.^34,35 Moreover, the nitrogens from the projected upper side chain of compound 24 formed two hydrogen bonds with the side chain of Asp193. From a simulation study, Asp193 was found to form a highly interacting salt bridge with V3 loop residue Lys10.³⁰ Moreover, alanine substitution of Asp193 resulted in a dramatic reduction of the HIV-1 co-receptor activity of CXCR4.³⁶ While alanine substitutions at CXCR4 residues Asp187 and Phe189 also experimentally impaired >60% of the HIV-1 co-receptor activity, site-directed mutagenesis studies demonstrated the crucial roles of Arg188, Tyr190, and Pro191 in HIV-1 binding.^35,37–40 Interestingly, the terminal cyclohexyl ring and the upper side chain of compound 24 picked up several significant hydrophobic interactions with CXCR4 residues Phe189, Tyr190, and Pro191. The phenyl ring moiety linked to the purine core in compound 24 was involved in hydrophobic interactions with Tyr190, Gln200, and Leu266. The alanine substitution of Gln200 was found to significantly affect HIV-1 co-receptor binding.³⁹ Moreover, Gln200 is calculated to form a strong nonpolar interaction with the aromatic ring of the V3 loop residue Trp20.³⁰ Taken together, the MD results provided great insight into the molecular interactions between compound 24 and CXCR4 receptor. We reasoned that the potent anti-HIV activity of compound 24 could come from its interactions with the critical CXCR4 residues in the N-terminal domain that are exploited by the HIV-1 V3 loop for viral entry. To provide further insight into the potent HIV entry inhibition by these purine-based antagonists, compound 25 was subjected to a MD study (Figure S2). Gratifyingly, compound 25, similar to that of compound 24, docked to the N-terminal extracellular loop II domain and interacted with the majority of the above-mentioned CXCR4 residues. MD simulations showed that compound 25 hydrogen bonded to several CXCR4 residues that were reported to be critical for HIV entry. By modifying the orientation of the upper side chain in compound 24 from para to meta in compound 25, compound 25’s upper side chain gained additional strong hydrogen-bond interactions with Asp187 and Glu277 and hydrophobic interaction with Arg188.^35,37,39 The terminal piperazine nitrogens of both compounds 24 and 25 were anchored toward the interior of CXCR4 by interacting with the side chain of Glu288, an important residue for HIV co-receptor activity. In all, by targeting a distinct domain of CXCR4 that could be pivotal for HIV-1 entry, we significantly improved the anti-HIV activity of compounds 24 and 25, into the subnanomolar range, a 130-fold enhancement in potency relative to that of AMD3100. Furthermore, we envision that a lower dose of compound 24 could be employed for the treatment of chronic HIV-1 infection to minimize adverse effects associated with disrupting the normal physiological functions of the SDF-1α/CXCR4 axis.

Figure 4.

Molecular dynamics simulations between compound 24 and CXCR4 (PDB code 3OE0) receptor. (A) Structure of the CXCR4–24 complex after 20 ns of molecular dynamics (MD) simulations. The hydrogen-bonding network (yellow lines) reveals strong hydrogen bonding around compound 24 (green) with residues Asp193, His281, and Glu288 (cyan). (B) Ligplot diagram showing hydrophobic interactions between the indicated CXCR4 amino acid residues and compound 24 after 20 ns of MD simulation.

Binding Specificity and Functional Activity Tests

Since compounds 24 and 25 were identified as having strong binding affinity for CXCR4 and potent anti-HIV entry activity in the subnanomolar range, further studies were conducted to address their binding specificities and functional activities. As shown in Table 3, compounds 24 and 25 exhibited >609- and >2380-fold selectivities for CXCR4 binding, respectively, versus a panel of closely related chemokine receptors including CXCR2, CCR2, CCR4, and CCR5, whose binding affinities were >10 000 nM (as IC₅₀). Furthermore, compounds 24 and 25 showed specific inhibition of CXCR4. Recent reports suggested that CXCR7 is the most closely related chemokine receptor to CXCR4.⁴¹ Using the established SDF-1α-dependent β-arrestin assay for CXCR7,⁴² compound 24 did not antagonize SDF-1α binding to CXCR7 or the triggering of β-arrestin recruitment even at a 40-fold higher SDF-1α concentration, providing a >1960-fold functional selectivity over CXCR4-dependent HIV entry inhibition by compound 24 (Figure S3). These results, together with several reported CXCR4 antagonists,³¹ strongly suggest that the introduction of nitrogen-containing appendages with a purine core or other skeletons might be pivotal in designing potent and specific CXCR4 antagonists. The nitrogen-containing fragments could mimic the Lys/Arg-rich nature of SDF-1α, the natural ligand involved in the binding and activation of CXCR4.⁴³ Moreover, since TZM-bl cells express both CXCR4 and CCR5 co-receptors, the lack of binding of compounds 24 and 25 to CCR5 from the above-mentioned results provides strong support that the observed potent anti-HIV entry effects (EC₅₀ ∼ 0.5 nM) are correlated with their binding to CXCR4 expressed on the surface of TZM-bl cells.

Table 3.

Specificity of Compounds 24 and 25 against Related Chemokine Receptors

compd	parameter	CXCR4	CXCR2	CCR2	CCR4	CCR5
24	inhibition (%)a	100	16	10	11	15
	IC50 (nM)	16.4	>10 000	>10 000	>10 000	>10 000
	selective index		>609	>609	>609	>609
25	Inhibition (%)a	100	12	2	0	14
	IC50 (nM)	4.2	>10 000	>10 000	>10 000	>10 000
	selective index		>2380	>2380	>2380	>2380

^aPercent inhibition was determined at 10 μM; weak inhibition was observed for all tested chemokine receptors except CXCR4.

Despite the significant therapeutic potential of SDF-1α in blocking CXCR4 in the HIV field, growing evidence has accumulated that the SDF-1α/CXCR4 axis is involved in tumor progression, angiogenesis, and metastasis.^14,15 Recent findings have demonstrated that cancer cells expressing CXCR4 migrate to metastatic target tissues that release SDF-1α.^14,15 A question concerning whether our newly developed CXCR4 antagonists overlapped with the SDF-1α binding sites on the CXCR4 receptor was raised. We addressed this issue by investigating the inhibitory activity of compound 24 on SDF-1α-induced cell migration. The chemotaxis inhibition assay was performed using CCRF-CEM cells that express endogenous human CXCR4. As shown in Figure 5A, while there was no apparent toxicity to CEM cells (CC₅₀ > 50 μM), compound 24 exerted an IC₅₀ of 3.1 nM to block SDF-1α-induced CEM migration. This was fairly comparable to the IC₅₀ (16.4 nM) for the inhibition of radioligand [¹²⁵I]SDF-1α binding by compound 24 to the membrane of hCXCR4-transfected cells (Table 2). AMD3100 was measured at 24.6 nM in this assay system. Thus, compound 24 was 8-fold more potent in CEM chemotaxis inhibition and 13-fold more potent in CXCR4 binding than that of AMD3100 (Table 1 vs Table 2). Interestingly, the IC₅₀ (213.1 nM) for 50% inhibition of [¹²⁵I]SDF-1α binding by AMD3100 to hCXCR4 (Table 1) was not comparable to its IC₅₀ value for chemotaxis inhibition. This observation suggested that compound 24 might partially overlap with the SDF-1α binding sites on the CXCR4 receptor and that AMD3100 might access different binding sites.

Figure 5.

Functional studies of compound 24. (A) SDF-1α-induced chemotaxis assay. Inhibition of CEM cell chemotaxis at various concentrations of compound 24 and AMD3100. (B) C57BL/6 mice were subcutaneously administered with the vehicle control, AMD3100, or compound 24 at different doses, respectively; 2 h after dosing, peripheral blood was harvested to analyze the cell type of interest.

Several observations on AMD3100’s disruption of the homing of stem and progenitor cells had led to its serendipitous development as a mobilization agent of stem cells (CD34⁺). We were interested in finding out if the significant increase in CXCR4 binding affinity by compound 24 would greatly enhance its stem cell mobilization ability. Using C57BL/6 mice for a stem cell mobilization assay, as previously described,³³ to our surprise, compound 24 was found to mobilize only CXCR4⁺/CD34⁺, the stem cells of interest, as efficiently as that of AMD3100 in the linear range (Figure 5B). CXCR4⁺/CD34⁺ cells were isolated in almost equal number from collected peripheral blood 2 h after the indicated dosing (0.1 mg per kilogram (mpk), 1 mpk, and 5 mpk). Although compound 24 showed a 13- and 130-fold increase in CXCR4 binding affinity and anti-HIV-1 activity, respectively, it did not translate into an increased mobilization of stem cells. An explanation for this could be that compound 24 targeted residues of distinct CXCR4 domains crucial for viral entry mediated by the HIV-1 V3 loop since studies have shown that binding and signaling domains in CXCR4 are possibly distinct and separate.^30,35,37 This observation strongly suggests the decoupling of two CXCR4-mediated biological processes, and the functional studies of compound 24 demonstrated the possibility of designing potent and selective CXCR4 antagonists.

CONCLUSIONS

In summary, we successfully designed and synthesized of a novel series of selective CXCR4 antagonists with a purine core scaffold that potently inhibited HIV-1 infection (18−20 and 22−27). Compound 24, the most active inhibitor, displayed a strong binding affinity for CXCR4 (IC₅₀ = 16.4 ± 3.5 nM) and a potent anti-HIV-1 activity as a viral entry blocker (EC₅₀ = 0.51 ± 0.02 nM), with no cytotoxicity up to 50 μM. Although more detailed structural studies are required to dissect the interactions between compound 24 and CXCR4, our MD simulations showed a high degree of molecular complementarity of compound 24 with CXCR4 residues essential for HIV-1 entry mediated by the HIV-1 V3 loop. Compound 24, when added to cells prior to or at the same time as HIV-1 infection in a time-course experiment, showed potent inhibition, confirming that it is an HIV-1 entry inhibitor. Compared to AMD3100, the subnanomolar and function-oriented increase in HIV entry inhibition (>130-fold) of compound 24 should afford a significant dosage decrease in anti-HIV treatment. Since a much higher dose (5 mpk) of compound 24 is needed to mobilize stem cells, the lower dosage of compound 24 may improve the therapeutic window of inhibiting HIV entry while reducing the disruption to stem cells homing in the bone marrow. Furthermore, these results strongly suggest that structural insights into the conformational states of CXCR4 toward compound 24 could aid future therapeutic design to minimize interference with the normal physiological functions of the SDF-1α/CXCR4 axis.

EXPERIMENTAL SECTION

General

Unless otherwise stated, all materials used were commercially obtained and used as supplied. Reactions requiring anhydrous conditions were performed in flame-dried glassware and cooled under an argon or nitrogen atmosphere. Unless otherwise stated, reactions were carried out under argon or nitrogen and monitored by analytical thin-layer chromatography performed on glass-backed plates (5 × 10 cm) precoated with silica gel 60 F254, as supplied by Merck (Merck & Co., Inc., Whitehouse Station in Readington Township, NJ). Visualization of the resulting chromatograms was performed by looking under an ultraviolet lamp (λ = 254 nm) followed by dipping in an ethanol solution of vanillin (5% w/v) containing sulfuric acid (3% v/v) or phosphomolybdic acid (2.5% w/v) and charring with a heat gun. Solvents for reactions were dried and distilled under an argon or nitrogen atmosphere prior to use as follows: THF, diethyl ether (ether), and DMF, from a dark blue solution of sodium benzophenone ketyl; toluene, dichromethane, and pyridine, from calcium hydride. Flash chromatography was used routinely for purification and separation of product mixtures using silica gel 60 of 230−400 mesh size as supplied by Merck. Eluent systems are given in volume/volume concentrations. Melting points were determined using a KRUSS KIP1N melting point meter. ¹H and ¹³C NMR spectra were recorded on a Varian Mercury-300 (300 MHz) and a Varian Mercury-400 (400 MHz). Chloroform-d or dimethyl sulfoxide-d₆ was used as the solvent, and TMS (δ 0.00 ppm), as an internal standard. Chemical shift values are reported in ppm relative to TMS in delta (δ) units. Multiplicities are recorded as s (singlet), br s (broad singlet), d (doublet), t (triplet), q (quartet), dd (doublet of doublets), dt (doublet of triplets), and m (multiplet). Coupling constants (J) are expressed in hertz. Electrospray mass spectra (ESMS) were recorded as m/z values using an Agilent 1100 MSD mass spectrometer. All test compounds displayed more than 95% purity, as determined by an Agilent 1100 series HPLC system using a C18 column (Thermo Golden, 4.6 mm × 250 mm). The gradient system for HPLC separation was composed of a MeOH (mobile phase A) and H₂O solution containing 0.1% trifluoroacetic acid (mobile phase B). The starting flow rate was 0.5 mL/min, and the injection volume was 10 μL. During first 2 min, the percentage of phase A was 10%. At 6 min, the percentage of phase A was increased to 50%. At 16 min, the percentage of phase A was increased to 90% over 9 min. The system was operated at 25 °C. Peaks were detected at 254 nm. IUPAC nomenclature of compounds was determined with ACD/Name Pro software.

N²-{4-[(3-Cyclohexylamino-propylamino)-methyl]-benzyl}-6-methoxy-N⁴-piperidin-4-yl-quinazoline-2,4-diamine Hydrochloride Salt (9)

To a magnetically stirred solution of Va (0.53 g, 0.64 mmol) in CH₂Cl₂ (10 mL) was added 1 N HCl/diethyl ether (20 mL, 20 mmol) dropwise at 25 °C under an atmosphere of argon. The resulting mixture was stirred at 25 °C for 15 h and concentrated by removing the solvent to afford of compound 9 (0.38 g, 92%). ¹H NMR (300 MHz, D₂O) δ 7.56−7.50 (m, 4H), 7.15−6.93 (m, 3H), 4.71 (s, 2H), 4.26 (m, 2H), 4.24 (m, 1H), 3.73 (s, 3H), 3.51 (m, 2H), 3.20−3.05 (m, 6H), 2.13−1.70 (m, 10H), 1.56 (m, 1H), 1.33−1.11 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 158.69, 155.42, 151.48, 140.27, 132.41, 130.21, 129.52, 127.50, 124.55, 117.57, 108.96, 103.77, 57.33, 56.12, 50.61, 46.92, 44.02, 43.79, 43.18, 41.13, 28.72, 27.19, 24.32, 23.77, 22.75; ESMS m/z: 532.3 (M + 1); HPLC purity = 95.35%, t_R = 14.35 min.

N²-{4-[(3-Cyclohexylamino-propylamino)-methyl]-benzyl}-7-methoxy-N⁴-piperidin-4-yl-quinazoline-2,4-diamine Hydrochloride Salt (10)

Compound 10 was synthesized from Vb (0.56 g, 0.67 mmol) following a similar synthetic procedure as that for 9 and was obtained as a white solid (0.36 g, 83%). ¹H NMR (300 MHz, D₂O) δ 7.65 (d, J = 9.0 Hz, 1H), 7.55−7.48 (m, 4H), 6.50 (dd, J = 9.0, 2.1 Hz, 1H), 6.38 (d, J = 2.1 Hz, 1H), 4.67 (s, 2H), 4.26 (s, 2H), 4.21 (m, 1H), 3.74 (s, 3H), 3.53 (m, 2H), 3.17−3.01 (m, 6H), 2.13−1.71 (m, 10H), 1.56 (m, 1H), 1.33−1.13 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 164.03, 158.85, 152.18, 140.29, 140.11, 130.20, 129.53, 127.53, 125.11, 113.62, 102.28, 97.72, 57.34, 55.96, 50.60, 46.78, 44.05, 43.75, 43.15, 41.15, 28.62, 27.33, 24.34, 23.77, 22.74; ESMS m/z: 532.3 (M + 1); HPLC purity = 96.25%, t_R = 14.26 min.

N²-{4-[(3-Cyclohexylamino-propylamino)-methyl]-benzyl}-6,7-dimethoxy-N⁴-piperidin-4-yl-quinazoline-2,4-diamine Hydrochloride Salt (11)

Compound 11 was synthesized from Vc (0.53 g, 0.61 mmol) following a similar synthetic procedure as that for 9 and was obtained as a white solid (0.37 g, 90%). ¹H NMR (300 MHz, D₂O) δ 7.57−7.48 (m, 4H), 7.26 (s, 1H), 6.65 (s, 1H), 4.75 (s, 2H), 4.26 (s, 2H), 4.23 (m, 1H), 3.83 (s, 6H), 3.48 (m, 2H), 3.19−3.00 (m, 6H), 2.13−1.96 (m, 6H), 1.93−1.71 (m, 4H), 1.65 (m, 1H), 1.33−1.13 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 158.55, 154.62, 151.75, 146.11, 140.12, 134.66, 130.15, 129.39, 127.30, 103.61, 101.54, 97.33, 57.38, 56.40, 56.22, 50.67, 46.89, 44.03, 43.92, 43.19, 41.18, 28.69, 27.34, 24.33, 23.78, 22.78; ESMS m/z: 562.3 (M + 1); HPLC purity = 95.27%, t_R = 14.23 min.

N²-{4-[(3-Cycloheptylamino-propylamino)-methyl]-benzyl}-6-methoxy-N⁴-piperidin-4-yl-quinazoline-2,4-diamine Hydrochloride Salt (12)

Compound 12 was synthesized from IVa by coupling with side chain E and deprotection with HCl/diethyl ether following a similar synthetic procedure as that for 9 and was obtained as a white solid (0.37 g, 56% over 2 steps). ¹H NMR (300 MHz, D₂O) δ 7.56−7.50 (m, 4H), 7.14 (d, J = 2.1 Hz, 1H), 7.06 (dd, J = 9.3, 2.1 Hz, 1H), 6.94 (d, J = 9.3 Hz, 1H),, 4.71 (s, 2H), 4.26 (m, 2H), 4.24 (m, 1H), 3.72 (s, 3H), 3.53 (m, 2H), 3.20−2.96 (m, 6H), 2.13−1.80 (m, 9H), 1.68−1.31 (m, 10H); ¹³C NMR(75 MHz, D₂O) δ 158.75, 155.47, 151.61, 140.30, 132.50, 130.23, 129.35, 127.58, 124.61, 117.65, 109.08, 103.81, 59.59, 56.15, 50.59, 46.93, 44.02, 43.70, 43.19, 41.58, 30.20, 27.22, 27.12, 23.12, 22.77; ESMS m/z: 546.4 (M + 1); HPLC purity = 95.36%, t_R = 15.00 min.

N²-{4-[(3-Cycloheptylamino-propylamino)-methyl]-benzyl}-7-methoxy-N⁴-piperidin-4-yl-quinazoline-2,4-diamine Hydrochloride Salt (13)

Compound 13 was synthesized from IVb by coupling with side chain E and deprotection with HCl/diethyl ether following a similar synthetic procedure as that for 9 and was obtained as a white solid (0.36 g, 57% over 2 steps). ¹H NMR (300 MHz, D₂O) δ 7.89 (d, J = 9.0 Hz, 1H), 7.55−7.49 (m, 4H), 6.94 (dd, J = 9.0, 2.1 Hz, 1H), 6.82 (d, J = 2.1 Hz, 1H), 4.78 (s, 2H), 4.35 (m, 1H), 4.26 (s, 2H), 3.91 (s, 3H), 3.47 (m, 2H), 3.17−2.99 (m, 6H), 2.13−1.77 (m, 7H), 1.90−1.39 (m, 12H); ¹³C NMR (75 MHz, D₂O) δ 164.04, 158.87, 152.19, 140.34, 140.12, 130.23, 129.36, 127.63, 125.16, 113.63, 102.32, 97.73, 59.59, 55.97, 50.67, 46.78, 44.05, 43.65, 43.16, 41.58, 30.20, 27.36, 27.11, 23.13, 22.77; ESMS m/z: 546.4 (M + 1); HPLC purity = 96.98%, t_R = 15.07 min.

N²-{4-[(3-Cycloheptylamino-propylamino)-methyl]-benzyl}-6,7-methoxy-N⁴-piperidin-4-yl-quinazoline-2,4-diamine Hydrochloride Salt (14)

Compound 14 was synthesized from IVc by coupling with side chain E and deprotection with HCl/diethyl ether following a similar synthetic procedure as that for 9 and was obtained as a white solid (0.36 g, 52% over 2 steps). ¹H NMR (400 MHz, D₂O) δ 7.54−7.49 (m, 4H), 7.22 (s, 1H), 6.58 (s, 1H), 4.73 (s, 2H), 4.25 (s, 2H), 4.24 (m, 1H), 3.80 (s, 6H), 3.48 (m, 2H), 3.23−3.00 (m, 6H), 2.14−1.96 (m, 7H), 1.93−1.77 (m, 2H), 1.71−1.59 (m, 2H), 1.55−1.37 (m, 8H); ¹³C NMR (75 MHz, D₂O) δ 160.99, 157.14, 154.21, 148.56, 142.75, 137.13, 132.75, 131.95, 129.92, 105.68, 103.95, 99.80, 62.20, 59.02, 58.82, 53.20, 49.47, 46.61, 46.46, 45.79, 44.22, 32.81, 29.94, 29.70, 25.70, 25.39; ESMS m/z: 576.4 (M + 1); HPLC purity = 95.59%, t_R = 14.80 min.

6-Methoxy-N⁴-piperidin-4-yl-N²-{4-[(3-piperidin-1-yl-propy-lamino)-methyl]-benzyl}-quinazoline-2,4-diamine Hydrochloride Salt (15)

Compound 15 was synthesized from IVa by coupling with side chain F and deprotection with HCl/diethyl ether following a similar synthetic procedure as that for 9 and was obtained as a white solid (0.33 g, 51% over 2 steps). ¹H NMR (300 MHz, D₂O) δ 7.57−7.52 (m, 4H), 7.31 (s, 1H), 7.23−7.21 (m, 2H), 4.75 (s, 2H), 4.29 (m, 1H), 4.27 (s, 2H), 3.82 (s, 3H), 3.56−3.47 (m, 4H), 3.20−2.92 (m, 8H), 2.20−2.10 (m, 2H), 2.05−1.65 (m, 9H), 1.47 (m, 1H); ¹³C NMR (75 MHz, D₂O) δ 158.70, 155.45, 151.50, 140.24, 132.40, 130.20, 129.48, 127.40, 124.56, 117.54, 108.99, 103.80, 61.61, 56.18, 53.29, 50.74, 46.96, 44.05, 44.00, 43.22, 27.21, 22.71, 20.97, 20.69; ESMS m/z: 518.3 (M + 1); HPLC purity = 96.36%, t_R = 13.88 min.

7-Methoxy-N⁴-piperidin-4-yl-N²-{4-[(3-piperidin-1-yl-propy-lamino)-methyl]-benzyl}-quinazoline-2,4-diamine Hydrochloride Salt (16)

Compound 16 was synthesized from IVb by coupling with side chain F and deprotection with HCl/diethyl ether following a similar synthetic procedure as that for 9 and was obtained as a white solid (0.31 g, 49% over 2 steps). ¹H NMR (300 MHz, D₂O) δ 7.66 (d, J = 9.0 Hz, 1H), 7.57−7.51 (m, 4H), 6.17 (dd, J = 9.0, 2.1 Hz, 1H), 6.37 (d, J = 2.1 Hz, 1H), 4.66 (s, 2H), 4.27 (s, 2H), 4.21 (m, 1H), 3.70 (s, 3H), 3.56−3.51 (m, 4H), 3.23−2.91 (m, 8H), 2.27−2.20 (m, 2H), 2.05−1.65 (m, 9H), 1.49 (m, 1H); ¹³C NMR (75 MHz, D₂O) δ 164.06, 158.81, 152.19, 140.24, 140.06, 130.18, 129.46, 127.82, 125.14, 113.66, 102.25, 97.73, 61.59, 55.97, 53.33, 53.29, 50.72, 44.06, 44.00, 43.18, 27.33, 22.72, 20.97, 20.69; ESMS m/z: 518.3 (M + 1); HPLC purity = 99.05%, t_R = 12.73 min.

6,7-Dimethoxy-N⁴-piperidin-4-yl-N²-{4-[(3-piperidin-1-yl-propylamino)-methyl]-benzyl}-quinazoline-2,4-diamine Hydrochloride Salt (17)

Compound 17 was synthesized from IVc by coupling with side chain F and deprotection with HCl/diethyl ether following a similar synthetic procedure as that for 9 and was obtained as a white solid (0.32 g, 45% over 2 steps). ¹H NMR (400 MHz, D₂O) δ 7.54−7.44 (m, 4H), 7.29 (s, 1H), 6.68 (s, 1H), 4.76 (s, 2H), 4.35 (m, 1H), 4.27 (s, 2H), 3.86 (s, 6H), 3.57−3.49 (m, 4H), 3.25−3.16 (m, 4H), 3.10−2.91 (m, 4H), 2.27−2.20 (m, 2H), 2.05−1.65 (m, 9H), 1.51 (m, 1H); ¹³C NMR (75 MHz, D₂O) δ 158.49, 154.61, 151.71, 146.01, 140.06, 134.60, 130.15, 129.43, 127.25, 103.13, 101.45, 97.27, 57.35, 56.43, 56.23, 53.30, 50.72, 46.92, 44.08, 44.02, 43.22, 27.35, 22.72, 20.95, 20.68; ESMS m/z: 548.3 (M + 1); HPLC purity = 95.99%, t_R = 13.84 min.

N-Cyclohexyl-N′-{4-[(2-piperazin-1-yl-9H-purin-6-ylamino)-methyl]-benzyl}-propane-1,3-diamine Hydrochloride Salt (18)

A solution of 1 N HCl/diethyl ether (4.8 mL) was added to a solution of 18-III (240 mg, 0.31 mmol) in CH₂Cl₂ (9.6 mL). The reaction mixture was stirred for 15 h and concentrated by removing the solvent to afford of compound 18 (174 mg, 94%). ¹H NMR (400 MHz, D₂O) δ 8.40 (s, 1H), 7.52−7.44 (m, 4H), 4.82 (s, 2H), 4.26 (s, 2H), 4.01 (m, 4H), 3.29 (m, 4H), 3.21−3.08 (m, 4H), 2.18−2.02 (m, 4H), 1.82 (m, 2H), 1.63 (m, 1H), 1.40−1.12 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 154.39, 151.63, 147.61, 139.56, 139.26, 130.13, 129.51, 127.95, 105.36, 57.39, 50.78, 44.06, 43.99, 42.56, 41.73, 41.20, 28.73, 24.37, 23.79, 22.77; ESMS m/z: 478.3 (M + 1); HPLC purity = 96.8%, t_R = 13.07 min.

N-Cyclohexyl-N′-{4-[(9-methyl-2-piperazin-1-yl-9H-purin-6-ylamino)-methyl]-benzyl}-propane-1,3-diamine Hydrochloride Salt (19)

A solution of 1 N HCl/diethyl ether (3 mL) was added to a solution of 19-III (156 mg, 0.23 mmol) in CH₂Cl₂ (6 mL). The reaction mixture was stirred at 25 °C for 15 h and concentrated by removing the solvent to afford 19 (123 mg, 91%). ¹H NMR (400 MHz, D₂O) δ 8.75 (s, 1H), 7.55−7.45 (m, 4H), 4.82 (s, 2H), 4.27 (s, 2H), 4.04 (m, 4H), 3.85 (s, 3H), 3.26−3.11 (m, 8H), 2.18−2.04 (m, 4H), 1.84 (m, 2H), 1.66 (m, 1H), 1.40−1.12 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 159.02, 151.60, 149.92, 140.24, 137.51, 130.12, 129.40, 127.95, 104.67, 57.42, 50.83, 44.00, 43.91, 42.83, 41.38, 41.24, 30.78, 28.78, 24.40, 23.84, 22.80; ESMS m/z: 492.3 (M + 1); HPLC purity = 97.1%, t_R = 13.34 min.

N-Cyclohexyl-N′-(4-{[9-methyl-2-(4-methyl-piperazin-1-yl)-9H-purin-6-ylamino]-methyl}-benzyl)-propane-1,3-diamine Hydrochloride Salt (20)

Compound 20 was synthesized from 19-II following a similar synthetic procedure as that for 19 in and was obtained as a white solid (143 mg, 65% over 2 steps). ¹H NMR (400 MHz, D₂O) 8.75 (s, 1H), 7.53 (d, J = 7.2 Hz, 2H), 7.48 (d, J = 7.2 Hz, 2H), 4.82 (s, 2H), 4.27 (s, 2H), 3.85 (s, 3H), 3.54 (m, 2H), 3.33 (m, 2H), 3.22−3.08 (m, 6H), 3.01 (m, 2H), 2.92 (s, 3H), 2.18−2.04 (m, 4H), 1.85 (m, 2H), 1.66 (m, 1H), 1.40−1.12 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 159.49, 152.14, 150.30, 140.74, 137.95, 130.58, 129.85, 128.38, 105.21, 57.88, 53.25, 51.27, 44.45, 44.32, 43.38, 42.17, 41.67, 31.22, 29.22, 24.86, 24.28, 23.24; ESMS m/z: 506.3 (M + 1); HPLC purity = 97.0%, t_R = 13.39 min.

N-{4-[(2-[4,4′]Bipiperidinyl-1-yl-9H-purin-6-ylamino)-methyl]-benzyl}-N′-cyclohexyl-propane-1,3-diamine Hydrochloride Salt (21)

Compound 21 was synthesized from 18-I following a similar synthetic procedure as that for 18 and was obtained as a white solid (186 mg, 36% over 3 steps). ¹H NMR (400 MHz, D₂O) δ 8.06 (s, 1H), 7.52−7.44 (m, 4H), 4.82 (s, 2H), 4.34 (m, 2H), 4.25 (s, 2H), 3.46 (m, 2H), 3.22−2.90 (m, 8H), 2.18−1.76 (m, 10H), 1.65 (m, 1H), 1.61−1.10 (m, 12H); ¹³C NMR (75 MHz, D₂O) δ 152.01, 151.05, 146.37, 140.88, 139.51, 130.17, 129.54, 128.18, 105.98, 57.39, 50.77, 46.66, 45.70, 44.17, 43.96, 41.19, 39.19, 37.59, 28.75, 27.99, 25.58, 24.39, 23.81, 22.77; ESMS m/z: 560.4 (M + 1); HPLC purity = 99.3%, t_R = 13.69 min.

N-Cyclohexyl-N′-{6-[(2-piperazin-1-yl-9H-purin-6-ylamino)-methyl]-pyridin-3-ylmethyl}-propane-1,3-diamine Hydrochloride Salt (22)

Compound 22 was synthesized from 18-I following a similar synthetic procedure as that for 18 and was obtained as a white solid (150 mg, 29% over 3 steps). ¹H NMR (400 MHz, D₂O) δ 8.91 (s, 1H), 8.72−7.64 (m, 2H), 8.17 (d, J = 8.4 Hz, 1H), 5.27 (s, 2H), 4.57 (s, 2H), 3.93 (m, 4H), 3.39−3.12 (m, 8H), 2.24−2.06 (m, 4H), 1.86 (m, 2H), 1.68 (m, 1H), 1.42−1.18 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 156.44, 155.40, 151.71, 149.16, 147.43, 143.02, 138.67, 129.21, 125.82, 105.08, 57.50, 47.19, 44.95, 42.64, 42.48, 41.38, 41.19, 28.80, 24.42, 23.84, 22.86; ESMS m/z: 479.3 (M + 1); HPLC purity = 98.6%, t_R = 13.69 min.

N-Cyclohexyl-N′-{4-[2-(2-piperazin-1-yl-9H-purin-6-ylami-no)-ethyl]-benzyl}-propane-1,3-diamine Hydrochloride Salt (23)

Compound 23 was synthesized from 18-I following a similar synthetic procedure as that for 18 and was obtained as a white solid (197 mg, 38% over 3 steps). ¹H NMR (400 MHz, D₂O) δ 8.28 (s, 1H), 7.41−7.38 (m, 4H), 4.23 (s, 2H), 4.06 (m, 4H), 3.93 (s, 2H), 3.40 (m, 4H), 3.21−3.03 (m, 6H), 2.16−2.04 (m, 4H), 1.83 (m, 2H), 1.64 (m, 1H), 1.40−1.12 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 154.29, 152.14, 147.64, 140.88, 139.31, 129.86, 129.80, 128.44, 105.93, 57.38, 50.79, 43.83, 43.55, 41.67, 41.43, 41.16, 34.32, 28.73, 24.37, 23.79, 22.75; ESMS m/z: 492.3 (M + 1); HPLC purity = 95.4%, t_R = 13.31 min.

N-Cyclohexyl-N′-(2-{4-[(2-piperazin-1-yl-9H-purin-6-ylami-no)-methyl]-phenyl}-ethyl)-propane-1,3-diamine Hydrochloride Salt (24)

Compound 24 was synthesized from 18-I following a similar synthetic procedure as that for 18 and was obtained as a white solid (195 mg, 38% over 3 steps). ¹H NMR (400 MHz, D₂O) δ 8.39 (s, 1H), 7.41 (d, J = 8.1 Hz, 2H), 7.32 (d, J = 8.1 Hz, 2H), 4.72 (s, 2H), 4.03 (m, 4H), 3.39−3.24 (m, 6H), 3.23−3.08 (m, 4H), 3.02 (m, 2H), 2.18−2.06 (m, 4H), 1.83 (m, 2H), 1.64 (m, 1H), 1.41−1.12 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 154.21, 151.36, 147.53, 139.24, 136.81, 135.45, 129.05, 127.91, 105.19, 57.43, 48.57, 44.49, 44.13, 42.58, 41.75, 41.23, 31.25, 28.76, 24.40, 23.82, 22.77; ESMS m/z: 492.3 (M + 1); HPLC purity = 99.6%, t_R = 13.34 min.

N-Cyclohexyl-N′-{3-[(2-piperazin-1-yl-9H-purin-6-ylamino)-methyl]-benzyl}-propane-1,3-diamine Hydrochloride Salt (25)

Compound 25 was synthesized from 18-I following a similar synthetic procedure as that for 18 and was obtained as a white solid (178 mg, 34% over 3 steps). ¹H NMR (400 MHz, D₂O) δ 8.34 (s, 1H), 7.52−7.36 (m, 4H), 4.82 (s, 2H), 4.21 (s, 2H), 3.99 (m, 4H), 3.27 (m, 4H), 3.21−3.09 (m, 4H), 2.18−2.02 (m, 4H), 1.81 (m, 2H), 1.63 (m, 1H), 1.40−1.12 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 154.40, 151.56, 147.74, 139.30, 139.01, 130.92, 129.54, 128.81, 128.55, 128.40, 105.37, 57.44, 51.04, 44.20, 44.11, 42.60, 41.77, 41.25, 28.78, 24.42, 23.84, 22.81; ESMS m/z: 478.3 (M + 1); HPLC purity = 95.3%, t_R = 13.30 min.

N-Cyclohexyl-N′-{6-[(2-piperazin-1-yl-9H-purin-6-ylamino)-methyl]-pyridin-2-ylmethyl}-propane-1,3-diamine Hydrochloride Salt (26)

Compound 26 was synthesized from 18-I following a similar synthetic procedure as that for 18 and was obtained as a white solid (152 mg, 30% over 3 steps). ¹H NMR (400 MHz, D₂O) δ 8.49 (s, 1H), 8.04 (t, J = 7.6 Hz, 1H), 7.63 (d, J = 7.6 Hz, 1H), 7.58 (d, J = 7.6 Hz, 1H), 4.98 (s, 2H), 4.47 (s, 2H), 3.92 (m, 4H), 3.24−3.10 (m, 6H), 3.08 (m, 2H), 2.18−2.02 (m, 4H), 1.77 (m, 2H), 1.60 (m, 1H), 1.36−1.08 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 156.81, 155.07, 151.68, 148.50, 148.15, 141.16, 139.04, 123.50, 123.16, 105.30, 57.44, 49.90, 44.89, 44.63, 42.57, 41.61, 41.19, 28.75, 24.37, 23.81, 22.77; ESMS m/z: 479.3 (M + 1); HPLC purity = 98.5%, t_R = 13.11 min.

N-Cyclohexyl-N′-{5-[(2-piperazin-1-yl-9H-purin-6-ylamino)-methyl]-pyridin-3-ylmethyl}-propane-1,3-diamine Hydrochloride Salt (27)

Compound 27 was synthesized from 18-I following a similar synthetic procedure as that for 18 and was obtained as a white solid (154 mg, 30% over 3 steps). ¹H NMR (400 MHz, D₂O) δ 8.96−8.93 (m, 2H), 8.79 (s, 1H), 8.56 (s, 1H), 5.14 (s, 2H), 4.57 (s, 2H), 4.02 (m, 4H), 3.38−3.26 (m, 6H), 3.10 (m, 2H), 2.23−2.06 (m, 4H), 1.84 (m, 2H), 1.67 (m, 1H), 1.40−1.15 (m, 6H); ¹³C NMR (75 MHz, D₂O) δ 156.26, 152.33, 149.09, 148.44, 142.15, 141.80, 140.20, 139.39, 131.28, 105.92, 57.93, 47.77, 45.42, 43.18, 41.99, 41.62, 41.51, 29.22, 24.86, 24.28, 23.29; ESMS m/z: 479.3 (M + 1); HPLC purity = 99.0%, t_R = 12.98 min.

Establishment of Human CXCR4 Stable Cell Line and Membrane Purification

hCXCR4 cDNA was subcloned into the pIRES2-EGFP vector (Clontech Laboratories, Inc., Mountain View, CA). Transfected HEK-293 cells stably expressing hCXCR4 (HEK-293 CXCR4) were selected by EGFP and 1 mg/mL G418 sulfate. The selected clone was maintained in DMEM supplemented with 10% fetal bovine serum and 0.5 mg/mL G418 sulfate with 5% CO₂ at 37 °C in a humidified incubator. For membrane purification, cells were homogenized in ice-cold buffer A (50 mM Tris-HCl, pH 7.4, 5 mM MgCl₂, 2.5 mM EDTA, 10% sucrose) with freshly prepared 1 mM PMSF. The homogenate was centrifuged at 3500g for 15 min at 4 °C. The pellet was removed, and the supernatant then was centrifuged at 43 000g for an additional 30 min at 4 °C. The final pellet was resuspended in buffer A and stored at −80 °C.

Radioligand Binding Assay

An amount of 2−4 μg of purified membrane with CXCR4 was incubated with 0.16 nM [¹²⁵I]SDF-1α and compounds of interest in the incubation buffer (50 mM HEPES-NaOH, pH 7.4, 100 mM NaCl, 5 mM MgCl₂, 1 mM CaCl₂, 0.5% BSA). Nonspecific binding was defined in the presence of 50 μM AMD3100 (plerixafor). The reaction mixtures were incubated for 1.5 h at 30 °C and then transferred to a 96-well GF/B filter plate (Millipore Corp., Billerica, MA). The reaction mixtures were terminated by manifold filtration and washed with ice-cold wash buffer (50 mM HEPES-NaOH, pH 7.4, 100 mM NaCl) four times. The radioactivity bound to the filter was measured by Topcount (PerkinElmer Inc., Waltham, MA). IC₅₀ values were determined by the concentrations of compounds required to inhibit 50% of the specific binding of [¹²⁵I]SDF-1α and calculated by nonlinear regression (GraphPad software, San Diego, CA).

Luciferase Activity Assay

Using TZM-bl cells, luciferase activity assays were performed to study the efficacy of the CXCR4 antagonists on HIV-1 infectivity. TZM-bl is a HeLa-derived cell line expressing high levels of the introduced CD4, CCR5, and CXCR4 receptors and contains HIV-1 long-terminal repeat-driven β-galactosidase and luciferase reporter cassettes that are activated by HIV-1 Tat expression.⁴⁴ For the luciferase reporter experiments, 5 × 10³ cells/well were plated in 96-well plates and cultured at 37 °C under 5% CO₂ in a humidified incubator. After incubation for one or two days and usually 30 min prior to HIV infection, TZM-bl cells were treated with serially diluted CXCR4 antagonists and then infected with HIV-1 IIIB at a multiplicity of infection (M.O.I.) of 1. A FARCyte machine (Amersham Pharmacia) was used to record the data, and at least three independent experiments were carried out to obtain the standard deviation.

Cytotoxicity Assay

CEM, a human T-lymphoblast leukemia, and TZM-bl cells were used for cell cytotoxicity assays. 10⁴ cells/well were plated in 24-well plates. After overnight incubation, cells were treated with CXCR4 antagonists for 72 h. The cells were then fixed and stained with 0.5% methylene blue in 50% ethanol for 2 h at room temperature, followed by washing with tap water to remove excess dye. Plates were dried, resuspended in 1% sarkosyl, and incubated for 3 h at room temperature. Methylene blue was oxidized by living cells to a colorless product, whereas dead cells remained blue. Cell growth was quantitated based on the amount of methylene blue adsorbed into cells measured by a spectrophotometer (Molecular Devices) at 595 nm. All experiments were performed in triplicate wells and repeated at least three times to obtain the standard deviation.

Docking Analysis of Compounds with CXCR4

The protein structure of CXCR4 (Protein Data Bank ID: 3OE0)²⁹ was applied for this study. All calculations were performed using Discovery Studio 2.1 (DS 2.1) (Accelrys, Inc., San Diego, CA). The docking analysis was conducted using the DS/LigandFit program with the CHARMm force field.⁴⁵ The number of docking poses was set as 100 with default parameters. The decision of the best pose was made according the binding information from Wu et al.²⁹

Molecular Dynamics Simulations

The molecular dynamics (MD) simulations were carried out using GROMACS v4.5.4 to refine the docked structures.^46,47 The topology of docked ligand was generated by PRODRG serve.⁴⁸ The force field for the whole system was GROMOS 43a1.⁴⁹ The protein–ligand complex was restrained in a box of cubic shape whose edges were placed at 1 nm from the complex, and SPC/E water model was performed. The system was electrically neutralized by adding 11 Cl⁻ ions. A two-step energy minimization was performed using steepest descent and conjugate gradient algorithms to converge the system up to 10 kJ mol⁻¹ nm⁻¹. After a short energy minimization step, the system was subjected to NVT (300 K) and NPT (1 bar) equilibration with 100 ps running, and LINCS algorithm was used to constrain the hydrogen-bond lengths.⁵⁰ The time step was kept at 2 fs for the simulation. A cutoff distance of 10 Å was used for all short-range nonbonded interactions, and 12 Å Fourier grid spacing in PME, for long-range electrostatics. Finally, the restraints of the complex structure were removed and a 20 ns MD calculation was performed (Supporting Information).

Chemotaxis Assay

CCRF-CEM (T-cell acute lymphoblastic leukemia) cells were suspended in RPMI 1640 containing 10% FBS and then preincubated with the indicated concentrations of compounds for 10 min at 37 °C. The assay was performed in Millicell hanging cell culture inserts (pore size, 5 μm; 24-well plate; Millipore, Bedford, MA, USA). Compounds containing 10 nM SDF-1 were plated in the lower chamber, and cells with compounds were plated in the upper chamber at a density of 2.5 × 10⁵ cells/well. After 2.5 h incubation at 37 °C, cells in both chambers were measured by flow cytometry (Guava Technologies, Hayward, CA).

Flow Cytometry Analysis for Stem/Progenitor Cell Counting

C57BL/6 male mice were treated with potential CXCR4 antagonists individually by subcutaneous injection, and then blood samples containing mobilized stem/progenitor cells were collected 2 h later. After labeling with specific antibodies, including APC-conjugated anti-CXCR4 (clone 2B11; eBioscience), FITC-conjugated anti-CD34 (clone RAM34; eBioscience), PE-conjugated anti-CD133 (clone 13A4; eBioscience), and anti-KDR (clone Avsa12a1; eBioscience), anti-c-Kit (clone 2B8; eBioscience), anti-Sca-1 (clone D7; eBio-science), anti-lineage (mouse hematopoietic lineage biotin panel, eBioscience), and Streptavidin PE-Cy7 (eBioscience), cells were washed, characterized, and quantified by flow cytometry (Guava Technologies, Hayward, CA). Each data point included at least 60 000 events for analysis of mobilized cells.

Use of Animal Subjects

All experimental protocols were approved and performed in accordance with the guidelines defined by the Institutional Animal Care and Use Committee (IACUC) of National Health Research Institutes (NHRI), Taiwan, R.O.C.

ABBREVIATIONS USED

HAART

highly active anti-retroviral therapy

gp120

glyco-protein 120

SDF-1α

stromal-derived factor-1α

SAR

structure–activity relationships

MD

molecular dynamics

mpk

milligram per kilogram

V3 loop

third variable loop