Synthesis, Biological Evaluation, and Molecular Modeling Studies of Chiral Chloroquine Analogues as Antimalarial Agents

In a focused exploration, we designed, synthesized, and biologically evaluated chiral conjugated new chloroquine (CQ) analogues with substituted piperazines as antimalarial agents. In vitro as well as in vivo studies revealed that compound 7c showed potent activity (in vitro 50% inhibitory concentration, 56.98 nM for strain 3D7 and 97.76 nM for strain K1; selectivity index in vivo [up to at a dose of 12.5 mg/kg of body weight], 3,510) as a new lead antimalarial agent.

ABSTRACT

In a focused exploration, we designed, synthesized, and biologically evaluated chiral conjugated new chloroquine (CQ) analogues with substituted piperazines as antimalarial agents. In vitro as well as in vivo studies revealed that compound 7c showed potent activity (in vitro 50% inhibitory concentration, 56.98 nM for strain 3D7 and 97.76 nM for strain K1; selectivity index in vivo [up to at a dose of 12.5 mg/kg of body weight], 3,510) as a new lead antimalarial agent. Other compounds (compounds 6b, 6d, 7d, 7h, 8c, 8d, 9a, and 9c) also showed moderate activity against a CQ-sensitive strain (3D7) and superior activity against a CQ-resistant strain (K1) of Plasmodium falciparum. Furthermore, we carried out docking and three-dimensional quantitative structure-activity relationship (3D-QSAR) studies of all in-house data sets (168 molecules) of chiral CQ analogues to explain the structure-activity relationships (SAR). Our new findings specify the significance of the H-bond interaction with the side chain of heme for biological activity. In addition, the 3D-QSAR study against the 3D7 strain indicated the favorable and unfavorable sites of CQ analogues for incorporating steric, hydrophobic, and electropositive groups to improve the antimalarial activity.

INTRODUCTION

Malaria is a parasitic disease with high rates of mortality and morbidity which sternly influences the socioeconomic development of affected countries (1). It is one of the most prevalent parasitic diseases, affecting nearly 300 million people around the globe annually (2). Five species of Plasmodium (Plasmodium falciparum, P. vivax, P. ovale, P. malariae, and P. knowlesi) are known to infect humans, and P. falciparum causes the most severe form of infection. In the erythrocytic stage, the malaria parasite utilizes host hemoglobin for its growth and proliferation and converts toxic heme into a nontoxic insoluble crystalline form called hemozoin (3). However, in infected erythrocytes (RBC), the mechanism of the conversion of heme to hemozoin through biochemical factors is not fully understood (4). Inhibition of heme-to-hemozoin conversion in parasites is one of the most effective ways to treat malaria. Among the antimalarial agents, the quinoline scaffold has immense significance, as it forms a complex with heme in its monomer and µ-oxo dimer forms. Also, one of the 4-aminoquinoline-based derivatives (chloroquine [CQ]) is most widely used for malaria treatment (5, 6). Nevertheless, the widespread resistance of P. falciparum to CQ and the artemisinin class of antimalarials has hampered the efforts to combat this deadly disease (7, 8). Resistance is a consequence of the decreased accumulation of the drug in the food vacuole of the parasite owing to enhanced efflux and reduced uptake (9). Therefore, the need to modify the structure of CQ for the development of alternative drug-like molecules that can circumvent the problem of P. falciparum resistance is urgent.

Studies of the structure-activity relationship (SAR) of CQ analogues revealed that compounds having shorter chain lengths are active against CQ-resistant parasite strains (9–11). By considering this scenario, a number of research groups have synthesized new short-chain analogues of 4-aminoquinolines, which were found to be considerably more potent than CQ against a resistant strain of P. falciparum in vitro (12). However, Roepe and colleagues recently concluded that these observations hold good only for quinoline derivatives that contain a diethyl substituent on the terminal nitrogen (13). It is believed that N-dealkylation reduces the lipid solubility of these drugs and their derivatives, leading to reduced antiplasmodial activity, facilitating drug resistance (14). Towards this objective, present researchers are making efforts to modify the existing chemical moieties to get new leads for antimalarial activity. Previously, our group investigated 4-aminoquinolines, synthesized by modulating the lipophilicity and basicity of the lateral side chain attached to the 4-amino group of the CQ side chain (15–21). Many of these analogues were found to form a complex with hematin and inhibit β-hematin formation, suggesting that this class of compounds acts on a heme dimerization target. Also, many of these analogues have shown significant activity against CQ-resistant parasites.

Encouraged by these findings, we thought that fine-tuning of the terminal side chain with N-methylpiperazine could modulate the antimalarial activity (22). Furthermore, we developed in our laboratories several chiral chain-based chloroquine analogues which showed excellent in vitro activity compared to that of CQ (23). As a part of our antimalarial drug discovery program, the above-mentioned results offered an ideal opportunity to better design molecules with a chirally defined architecture and tailored pendant groups. Therefore, we designed target-specific compounds through a rational method. This rational method was used with a special emphasis on avoiding metabolic N-dealkylation by incorporating bulkier substituents at an adjacent part of the side chain, and the compounds were evaluated for their antimalarial activity.

It was found that the majority of analogues had excellent activities against CQ-sensitive parasite strain 3D7. Nevertheless, no three-dimensional (3D) quantitative structure-activity relationship (QSAR) study with this in-house set of compounds has been reported earlier. Therefore, we focused on a 3D-QSAR study of all in-house chiral chain-based CQ analogues to explain the biological activity against the 3D7 strain. Hence, in the present work we developed our new findings through docking and 3D-QSAR studies to get an insight into the importance of chiral chains in CQ analogues. The presence of H-bond interactions with heme in conjunction with favorable steric, hydrophobic, and electropositive groups paves the way for a novel method to design more potential analogues having antimalarial activity against the CQ-resistant (CQ^r) parasite strain.

RESULTS AND DISCUSSION

In vitro antiplasmodial activity.

All compounds synthesized in the present study (compounds 6a to 6d, 7a to 7h, 8a to 8d, 9a to 9d, 10a, 10b, 11a to 11c, and 12a) were evaluated for their antiplasmodial activity against the 3D7 chloroquine-sensitive (CQ^s) and K1 chloroquine-resistant (CQ^r) strains of P. falciparum according to the procedure reported by Kondaparla et al. (24). Selected compounds (compounds 6b, 6d, 7c, 7h, 8c, 8d, 9a, and 9c) showing activity superior to that of CQ against the K1 strain were evaluated for their in vivo activity against the innately chloroquine-resistant P. yoelii N-67 strain in albino mice of the Swiss strain. All tested compounds displayed moderate activity, with 50% inhibitory concentrations (IC₅₀s) ranging from 22.61 to >5,000 nM against the 3D7 strain. Among all tested compounds, two compounds (compounds 7d and 7h) showed promising activity with IC₅₀s of 24.55 and 22.61 nM, respectively, and three compounds (compounds 6a, 7c, and 9a) with IC₅₀s ranging from 56.98 nM to 108.39 nM showed moderate activity. Moreover, eight compounds of the series were discerned to be more active (IC₅₀s, 46.50 to 161.80 nM) than CQ (IC₅₀ = 255 nM) against the CQ-resistant strain of P. falciparum. Three compounds (compounds 6a, 7e, and 9b) exhibited IC₅₀s of 276.90, 352.30, and 362.28 nM, respectively, which indicates that they were less active than CQ. Moreover, a low resistance index was exhibited by the majority of the synthesized compounds (Table 1). The in vitro biological activity data for the synthesized compounds are summarized as shown in Fig. 1.

TABLE 1.

Biological and biophysical data for the synthesized compounds

Compound	IC50 (nM)a		Resistance factorb	SIc	Log Kdd	IC50 (µM)e
	3D7	K1
6a (M144)	108.39	276.90	2.54	979.10	5.85 ± 0.03	0.38 ± 0.02
6b (M146)	899.23	161.80	0.17	69.6	4.72 ± 0.02	0.41 ± 0.01
6c (M149)	993.50	4,231	4.26	61.5	6.33 ± 0.03	0.69 ± 0.03
6d (M161)	521.20	46.50	0.08	43.28	4.23 ± 0.03	0.23 ± 0.01
7a	123.53	>1,000	>8.09	174.70	6.63 ± 0.02	0.73 ± 0.01
7b	153.90	>1,000	>6.49	292.90	6.43 ± 0.01	0.69 ± 0.02
7c (M140)	56.98	97.76	1.73	3,510.0	4.41 ± 0.02	0.27 ± 0.01
7d (M164)	24.55	797.67	32.49	3,671.69	5.52 ± 0.03	0.79 ± 0.03
7e (M141)	460.05	352.30	0.76	90.90	6.71 ± 0.02	0.44 ± 0.02
7f (M142)	857.68	412.40	0.48	76.60	6.31 ± 0.01	0.45 ± 0.02
7g (M143)	211.10	452.80	2.14	153.50	6.53 ± 0.02	0.48 ± 0.03
7h (M165)	22.61	46.52	2.05	582.04	4.32 ± 0.02	0.21 ± 0.02
8a (M150)	>1,000	1,100	1.10	724.03	6.82 ± 0.01	0.51 ± 0.01
8b (M151)	>5,000	>5,000	1.00	ND	7.81 ± 0.02	0.73 ± 0.03
8c (M152)	427.89	70.82	0.16	66.81	4.51 ± 0.01	0.23 ± 0.01
8d (M153)	757.38	89.0	0.11	61.47	4.48 ± 0.02	0.29 ± 0.02
9a (M154)	108.10	71.39	0.65	117.94	4.53 ± 0.03	0.26 ± 0.01
9b (M155)	126.54	362.28	2.87	>1,580.52	5.98 ± 0.03	0.42 ± 0.03
9c (M156)	404.80	84.01	0.20	112.69	4.42 ± 0.02	0.27 ± 0.01
9d (M157)	538.59	783.73	1.45	40.34	5.12 ± 0.03	0.48 ± 0.02
10a (M158)	4,150.08	>5,000	1.20	ND	6.97 ± 0.03	0.81 ± 0.03
10b (M160)	596.52	528.72	0.88	32.77	6.23 ± 0.02	0.49 ± 0.01
11a (M159)	>5,000	>5,000	1.00	ND	7.13 ± 0.02	0.83 ± 0.02
11b (M162)	3,099.19	>5,000	1.61	ND	6.92 ± 0.03	0.79 ± 0.03
11c (M163)	2,659	>5,000	1.88	ND	6.96 ± 0.03	0.75 ± 0.02
12a (M145)	113.76	502.10	4.44	1,093.10	5.84 ± 0.02	0.49 ± 0.03
CQ	5	255	51	8,983	5.52 ± 0.02	0.17 ± 0.02

^aIC₅₀, the concentration corresponding to 50% growth inhibition of the parasite.

^bThe resistance factor is equal to the IC₅₀ for K1/IC₅₀ for 3D7.

^cSI, selectivity index, which is equal to CC₅₀ for Vero cells/IC₅₀ for 3D7 for antiplasmodial activity.

^dLog K_d, log dissociation constant for 1:1 (compound-to-hematin) complex formation in 40% aqueous DMSO, 20 mM HEPES buffer, pH 7.5, at 25°C. Data are expressed as the mean ± SD from at least three different experiments performed in triplicate.

^eThe IC₅₀ represents the millimolar equivalents of the test compounds relative to the amount of hemin required to inhibit β-hematin formation by 50%. The data are expressed as the mean ± SD from at least three different experiments performed in triplicate. ND, not done.

FIG 1.

Histogram showing the number of compounds evaluated for activity against P. falciparum in the present study.

The structure-activity relationship studies on these compounds suggested that the activities were greatly influenced by the type of substitutions at the 4th position of the piperazine moiety, the structural diversity in the amino acid side chain, and a trifluoromethyl substitution at the C-7 position of the quinoline ring. More importantly, a substitution at the piperazine moiety showed a remarkable influence on the antiplasmodial activity.

The results presented in Table 1 indicate that the compounds having an ethyl, benzyl, and piperonyl group at the 4th position of piperazine (irrespective of the amino acid side chain) exhibited mild inhibition of the 3D7 strain and showed activity against the K1 strain of P. falciparum superior to that of CQ. However, compound 7c showed 2.6-fold more activity than CQ against the K1 strain with an IC₅₀ of 97.76 nM. Further, the replacement of chlorine in compound 7c with trifluoromethyl decreased the activity of compound 7d against the K1 strain manyfold (IC₅₀ = 797.67 nM). This decrease in activity may have been due to an increase in hydrophobicity. Again, the activity against the K1 strain was increased substantially in the case of compound 7h (IC₅₀ = 46.52 nM for K1), which has a piperonyl methyl substitution in the side chain. It is important to note that this compound was found to be 5.5-fold more active than CQ against the K1 strain. When the amino acid side chain was changed to the hydrophobic benzyl group, it furnished the compounds with a considerable decrease in activity against the K1 strain of P. falciparum. However, compound 8c (IC₅₀ = 70.82 nM), which had a substitution similar to that in the case of compound 7c (IC₅₀ = 97.76 nM), exhibited a slight increase in activity against the K1 strain. A similar activity profile was observed in the case of compound 8d with a piperonyl methyl substitution (IC₅₀ = 89.0 nM). Moreover, compound 6d (IC₅₀ = 46.50 nM), which had a glycine at 4th position of the quinoline ring, showed better activity than compound 9a (IC₅₀ = 71.39 nM), which had an alanine at the same position. More importantly, these two compounds exhibited activities 5.5-fold and 3.6-fold superior to the activity of CQ, respectively, against the K1 strain of P. falciparum. The chance that the parasite would develop resistance to a particular class of compounds was calculated as the ratio of the IC₅₀ for CQ-resistant strains to the IC50 for CQ-sensitive strains, called the resistance factor (RI). Therefore, a smaller resistance factor for a given compound corresponds to less of a chance of resistance development. Interestingly, the majority of the tested compounds in this series exhibited low resistance factors of between 0.08 and 32.49, whereas the resistance factor for CQ was 51. Furthermore, the compounds of the present study had better antiplasmodial activity against both CQ^s and CQ^r strains than previous antimalarial activity which was reported from our laboratories. In light of the findings of the current study with the tested compounds, further lead optimization to obtain molecules active against drug-resistant parasites seems to be suggested.

In vitro inhibition of β-hematin formation.

Newly synthesized 4-aminoquinolines (compounds 6a to 6d, 7a to 7h, 8a to 8d, 9a to 9d, 10a, 10b, 11a to 11c, and 12a) were evaluated for their mode of action using a previously reported method (19), and the results are given in Table 1. The heme binding assay results revealed that all synthesized 4-aminoquinoline derivatives interacted with the heme. Further, association constants in the range of 4.23 to 7.81 were observed. These results indicate that the interaction between the quinoline ring and the porphyrin ring system might be a π-π stacking interaction. These results are complemented by the inhibition of β-hematin formation, for which good IC₅₀s in the range of 0.21 to 0.83 µM were also shown. In the present series, the most active compound, compound 7h, exhibited an IC₅₀ of 0.21 µM against the CQ^r strain.

In vitro cytotoxicity.

A 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay in the Vero cell line was used to determine the cytotoxicity of all synthesized compounds (compounds 6a to 6d, 7a to 7h, 8a to 8d, 9a to 9d, 10a, 10b, 11a to 11c, and 12a) (Table 1). The selectivity index (SI) of nearly all compounds in the series was good and ranged from 32.77 to 3,671.69. Compounds 7c, 7h, and 9a of the series showed enhanced antiplasmodial activities against the K1 strain and also exhibited good SI values of 3,510, 582.04, and 117.94, respectively. However, compounds 6a, 9b, and 12a, which displayed promising activities against the K1 strain, also showed considerable SI values of 979.10, >1,580.52, and 1,093.10, respectively. Moreover, within the series of compounds, compounds 7d and 7h showed significant activity against the 3D7 strain with good SI values of 3,671.69 and 582.04, respectively. In general, with their promising activity against the K1 strain, less cytotoxic effects, and fairly high selectivity indexes, most of the 4-aminoquinoline derivatives of this series are very healthy candidates for further lead optimization.

In vivo antimalarial activity.

The in vivo antimalarial activity of the compounds against an inherently chloroquine-resistant P. yoelii strain (the N-67 strain) in albino mice of the Swiss strain was assessed. Compounds 6b, 6d, 7c, 7h, 8c, 8d, 9a, and 9c, which had IC₅₀s ranging from 31.19 nM to 252.28 nM, were selected on the basis of their in vitro antiplasmodial activity. In the beginning, the in vivo activities of selected compounds administered at a dose of 100 mg/kg of body weight once daily for four consecutive days via the oral route were examined. The parasitemia reduction and survival of the animals were recorded until day 28 postinfection (Table 2).

TABLE 2.

In vivo antiplasmodial activity of selected compounds against CQ-resistant P. yoelii (N-67) in albino mice of the Swiss strain

Compound	Dose (mg/kg)c	% suppression on day 4	Survivala	Cureb	Survival after rechallenged
6b (M146)	100	100	0/5	0/5
6d (M161)	100	100	2/5	0/5	2/5
7c (M140)	100	100	5/5	5/5	0/5
	50	100	5/5	5/5	0/5
	25	100	5/5	5/5	0/5
	12.5	100	5/5	5/5	0/5
	6.25	100	2/5	0/5	2/5
7h (M165)	100	100	5/5	5/5	0/5
	50	100	5/5	5/5	0/5
	25	100	0/5	0/5
8c (M152)	100	100	5/5	5/5	0/5
	50	50	2/5	0/5	2/5
8d (M153)	100	100	3/5	0/5	3/5
9a (M154)	100	100	2/5	0/5	2/5
9c (M156)	100	100	1/5	0/5	1/5
CQ	20	99.0	5/5	0/5	5/5

^aSurvival results are given as the number of mice that survived until day 28 postinfection/total number of mice in the group.

^bCure results are given as the number of mice without parasitemia (cured) through day 28 postinfection/total number of mice in the group.

^cEach dose was administered orally for 4 days.

^dMice surviving until day 28 were inoculated with 1 × 10⁶ P. yoelii-parasitized erythrocytes to monitor susceptibility to rechallenge. The results are given as the number of mice that survived until day 28 postinfection/total number of mice in the group.

A dose of 100 mg/kg compounds 6b, 6d, 8d, 9a, and 9c showed 100% parasitemia suppression on day 4, but no mice in the group were cured up to day 28 of treatment. Further, compound 8c was administered at two different doses, 100 and 50 mg/kg, respectively. At a dose of 100 mg/kg, it displayed 100% parasitemia suppression on day 4 with 100% survival and curative rates up to day 28, whereas at a dose of 50 mg/kg it exhibited 50% parasitemia suppression on day 4 with a 40% survival rate up to day 28 of treatment. Moreover, compound 7c, one of the potent compounds in the series initially tested at a dose of 100 mg/kg, produced 100% survival as well as curative rates up to day 28. When it was administered at lower doses of 50, 25, 12.5, and 6.25 mg/kg, it displayed 100% parasitemia suppression on day 4 with a 100% curative rate up to day 28 at doses of 50, 25, and 12.5 mg/kg, while at a dose of 6.25 mg/kg it showed 100% parasitemia suppression on day 4, but none of the mice survived. Reinoculation with the infective inoculum after day 28 showed that all the mice in the cured group developed fulminating parasitemia and succumbed to infection, thus validating the absence of any latent parasites in these animals. In contrast, the surviving animals, which were not designated to be cured, survived the rechallenge (Table 2).

Molecular docking studies.

Recent evidence strongly leans toward an interaction of CQ analogues with the surface of the hemozoin crystal rather than free heme (25, 26). Understandably, it is difficult to make a model of the hemozoin structure for docking. Therefore, the in-house database created in the present study (168 molecules) was used to dock against the heme instead of against hemozoin (extracted from the structure with PDB accession number 4D6U) to investigate the probable interactions between the CQ analogues and heme. Docking analysis of these molecules clearly showed the importance of the total score, crash score, and polar score (Table 3).

TABLE 3.

Binding scores (total score, crash score, and polar score) of selected molecules from the in-house database

Compound code	Total score	Crash score	Polar score
M32	1.1368	−0.4986	1.4845
M62	1.6703	−0.8912	0.4415
M64	1.6657	−0.4918	1.4768
M87	1.9119	−0.9051	1.4469
M96	4.2653	−0.5227	1.4153
M97	2.2337	−0.5357	1.3241
M100	2.8436	−1.0254	0
M113	2.3441	−0.6026	1.4784
M114	2.6098	−0.7997	1.512
M134	2.2171	−0.6227	1.2771
M158	3.1888	−0.8168	2.0496
M165	2.3102	−0.9484	0

Most of the compounds showed moderate to good binding scores with the heme. Compounds M64, M96, M97, and M100 were confirmed to have a good binding affinity with the heme due to strong H-bond interactions with the free carboxylic group of the heme moiety (Fig. 2), whereas compounds M32, M158, and M165 showed poor binding affinity toward the heme due to improper binding conformations. This finding is in agreement with the observed biological activity.

FIG 2.

Molecular docking of the most active compounds, compounds M96 (A), M64 (B), and M62 (C), and the less active compounds, compounds M32 (D), M158 (E), and M165 (F), with heme. Red indicates an H-bond interaction.

Additionally, excellent electrostatic and hydrophobic (π-π stacking) interactions between the quinoline ring and the porphyrin ring of the heme moiety contributed significantly to improve the binding affinities (Fig. 3; Table 4), which are directly linear with their biological activity against the 3D7 strain (Tables 1 and 3). Furthermore, from the observation made above, it was concluded that compounds with more bulky groups in the side chains (compounds M153, M161, and M165) poorly bind with the heme, which is reflected through their maximum crash score values and unfavorable docked conformations due to the lack of π-π stacking interactions. In addition, docking analysis further revealed that a low polar score may reduce the formation of a good heme-ligand complex (compounds M37, M153, and M157), leading to reduced activity (Fig. 2). These results reveal the importance of the side chain modifications to improve the binding affinity of CQ analogues to heme.

FIG 3.

Non-H-bond interactions between heme and M96. Yellow indicates electrostatic interactions, whereas pink indicates hydrophobic interactions.

TABLE 4.

Different types of non-H-bond interactions between heme and CQ analogue M96

No. of non-H-bond interactions	Distance (Å)	Category	Type
1	3.39	Electrostatic	π-cation
2	4.11	Electrostatic	π-anion
3	4.72	Hydrophobic	π-π stacked
4	5.00	Hydrophobic	π-alkyl
5	3.78	Hydrophobic	π-alkyl
6	4.32	Hydrophobic	π-alkyl
7	4.32	Hydrophobic	π-alkyl

Statistical analysis by CoMFA and CoMSIA.

The best possible 3D-QSAR model was derived from the partial least-squares (PLS) statistical data from comparative molecular field analysis (CoMFA) and comparative molecular similarity index analysis (CoMSIA) of the CQ analogues, which is shown in Table 5. The optimal value of Q² (>0.5) was used as the criterion to find a new significant model. The ideal CoMFA model with five PLS components along with a cross-validated Q² value of 0.641, a non-cross-validated r² value of 0.924, and a predicted r² (r²_pred) value of 0.656 was recognized (Table 5). In the case of CoMSIA, five components were established to ideally express the antimalarial activity of the compounds. It revealed a cross-validated Q² value of 0.680 and a non-cross-validated r² value of 0.885. In the CoMSIA, a good predictive value (r²_pred = 0.608) was also identified with the test sets.

TABLE 5.

Goodness of fit of CoMFA and CoMSIA models for the antiplasmodial activities of CQ analogues against the 3D7 strain

PLS statistic	Value for:
	CoMFA model	CoMSIA model
r2 nCVa	0.924	0.885
SEEb	0.237	0.353
F testc	463.97	114.714
$r_{\text{cv}}^2$d	0.641	0.680
SEPe	0.596	0.574
r2predf	0.656	0.608
PLS components	5	5
Contribution
Steric	0.627	0.452
Electrostatic	0.373	0.207
Hydrophobic		0.341
H-bond donor
H-bond acceptor
r2 bootg	0.946	0.933
SEE booth	0.237	0.263
r2 LHOi	0.479	0.535
SD LHOj	0.073	0.051
r2 5cvk	0.532	0.564
SD 5cvl	0.045	0.039

^ar² nCV, the predictable value correlation coefficient. For all models, the optimal number of PLS components is 5.

^bSEE, standard error of the estimate.

^cRatio of r² for explained to unexplained = r²/(1 − r²).

^d$r_{\text{cv}}^2$, leave-one-out predicted cross-validated correlation coefficient.

^eSEP, standard error of the prediction.

^fr²_pred, external predicted correlation coefficient for the test set of compounds.

^gr² boot, average correlation coefficient for 100 samplings determined using the bootstrapped method.

^hSEE boot, average standard error of the estimate for 100 samplings determined using the bootstrapped method.

ⁱr² LHO, average cross-validated correlation coefficient for 50 runs determined using the leave-half-out (LHO) group.

^jSD LHO, standard deviation for average cross-validated correlation coefficient for 50 runs.

^kr² 5cv, average cross-validated correlation coefficient for 50 runs using five cross-validation groups.

^lSD 5cv, standard deviation of the average cross-validated correlation coefficient for 50 runs.

The 3D-QSAR model derived from this experiment showed that all five fields of CoMSIA do not equally play a vital role to explain the activity. Therefore, a single grouping and multiple groupings of CoMSIA fields were measured to choose the important fields with the best cross-validated Q² value. Finally, from this experiment the most important fields responsible for activity, namely, the steric, electrostatic, and hydrophobic fields, were recognized (Fig. 4).

FIG 4.

Graphical plots of Q² values of CoMSIA models which indicated single as well as multiple field combinations for CQ analogues as antiplasmodial inhibitors of the 3D7 strain. In this plot, the x axis indicates the fields of CoMSIA with single letters, as follows: A, H bond acceptor; D, H bond donor; E, electrostatic; H, hydrophobic; S, steric. The field combinations which led to the model with an ideal Q² (y axis) is marked with a circle.

3D models constructed from the results of CoMFA and CoMSIA were found to be effective in bootstrapping and other cross-validation experiments (Table 5). Moreover, the derived model satisfactorily predicted the test set of compounds. The scattered graphical plots of observed versus predicted values of the biological activities of the training and test groups resulting from the CoMFA and CoMSIA study are revealed in Fig. 5.

FIG 5.

Scatter plots of experimental pIC₅₀ values versus predicted pIC₅₀ values derived from CoMFA/CoMSIA training and the test set of CQ derivatives for the Plasmodium falciparum 3D7 strain.

Contour map analysis.

(i) CoMFA steric contour maps. The CoMFA steric contour maps of the activities of the CQ derivatives against the 3D7 strain are shown in Fig. 5. Here, the green and yellow color contours correspondingly specify the steric regions favored as well as unfavored for activity. The green contours present at the linker region (Fig. 6) identify the sterically preferred groups in this area to get improved antimalarial activity, as was observed with compound M96. In addition, the green contour close to the end of the linker site (Fig. 6) indicates the positivity of this location for steric groups. All active analogues (compounds M64, M96, M98, and M126) satisfied these criteria with different steric groups, such as tert-butyl, methylpiperazine, and methylpiperidine moieties. Similarly, yellow contours are situated at the C-7 position of the quinoline ring (Fig. 6). Another yellow contour enclosed more bulky groups at the linker region (Fig. 6, compounds M32, M34, and M68). This speaks in favor of the presence of the least-steric groups at the C-7 position.

FIG 6.

Two different contour maps, including steric field contour maps (A and B) and electrostatic field contour maps (C and D) of CoMFA 3D-QSAR, are shown for CQ analogues as antiplasmodial inhibitors of the 3D7 strain. The poses of compounds M96 (active) and M32 (inactive) are used to demonstrate the contour maps. The color code specifies favorable and unfavorable regions. Here, green contours indicate regions sterically favorable for biological activity and yellow contours shows regions sterically unfavorable for biological activity (A and B). Similarly, blue contours indicate electropositive regions and red contours point out electronegative regions (C and D).

(ii) CoMFA electrostatic contour maps. The electrostatic contour map built for CQ derivatives from the CoMFA model is shown in Fig. 6. In this figure, the large blue contour adjacent to the electropositive 4-amino of the quinoline group indicates the presence of electron-rich nitrogen, which may increase the activity (compounds M60, M62, M96, and M97). Additionally, another small blue contour near the linker site specifies its requirement for electropositive groups. Apart from that, red contours are observed near the linker site in the case of less active compounds (Fig. 6, compounds M146, M149, and M161). Therefore, electronegative groups can be replaced at the linker region with electropositive moieties to improve their activity. Moreover, a blue contour is positioned in front of the quinoline moiety, representing the prominence of electropositive groups for biological activity.

(iii) CoMSIA steric contour maps. In CoMSIA, the favorable areas for steric groups are indicated in the green contour and the disfavored regions are represented in the yellow contour. At this point, two large green contours occurred in proximity to linker (chain) sides, indicating the fitness for large groups in these positions. The presence of tert-butyl, methylpiperazine, and like moieties appears to be favorable for good activity (e.g., compounds M64, M96, and M97). The substituted piperazine moiety has fulfilled the steric requirements of the green contour parallel to the linker region (Fig. 7, compound M96). Most of the active compounds in the data set contribute to this feature (e.g., compounds M96, M100, and M109). Besides this, two big yellow contours were found over the green contour, suggesting the presence of large groups in this area which can diminish the activity (compounds M32, M158, and M165).

FIG 7.

Three different contour maps, including steric field contour maps (A and B), electrostatic field contour maps (C and D), and hydrophobic contour maps (E and F), of CoMSIA 3D-QSAR are shown for CQ analogues as antiplasmodial inhibitors of the 3D7 strain. The poses of compounds M96 (active) and M32 (inactive) are used to demonstrate the contour maps. The color code specifies regions favorable and unfavorable for biological activity. Here, green contours indicate regions sterically favorable for biological activity and yellow contours show regions sterically unfavorable for biological activity (A and B). Similarly, blue contours indicate electropositive regions and red contours point out electronegative regions (C and D). Finally, yellow contours designate hydrophobic regions favorable for biological activity, whereas white contours indicate hydrophobic regions unfavorable for biological activity (E and F).

(iv) CoMSIA electrostatic contour maps. In the electrostatic map, blue contours specify the electropositive favorable area and red contours point out electronegative favorable sites. The blue contour exists close to the linker region (Fig. 7), which suggests that electropositive groups are suitable for activity (compounds M60, M62, M96, and M97). In the case of active compounds, this position is fulfilled by methyl ethyl amine and methylpiperazine groups. Incorporation of alkyl chains in this electropositive region improves the activity, but the red contour map near this region drastically reduces the biological activity (compounds M146, M149, and M161).

(v) CoMSIA hydrophobic contour maps. The CoMSIA hydrophobic contour maps are shown in Fig. 7. In these contours, yellow specifies hydrophobic favorable regions, whereas gray specifies hydrophobic unfavorable sites. The yellow contour near the linker position showed the positive site for hydrophobic groups (compounds M63, M64, M96, and M100). In highly active compounds, dimethyl amine, tert-butyl, methylpiperazine, and other related groups were incorporated in this region. Perhaps the hydrophobic part of these groups along with other characteristics may be appropriate for this position. Apart from this, a large gray contour and a small gray contour occupied the central area of the linker region and the C-7 position of the quinoline ring, respectively. This denoted an adverse environment for hydrophobic moieties in the surrounding area of the middle part of the linker position as well as the C-7 position of the quinoline ring system (compounds M32, M108, and M167).

Conclusion.

In summary, we have synthesized a new series of chiral 4-aminoquinoline derivatives in order to search for more potent molecules active against strains of P. falciparum both in vitro and in vivo Among all synthesized compounds, eight compounds (compounds 6b, 6d, 7c, 7h, 8c, 8d, 9a, and 9c) exhibited excellent antimalarial activity against the K1 strain compared to the activity of CQ, and these compounds were also found to be active against P. yoelii in a mouse model of infection in vivo. It may be inferred from the above-mentioned results that the compound with less hindered amino acids (namely Gly, Ala, and Leu) at the 4th position of the quinoline ring and the ethyl-substituted piperazine in the side chain exhibited potent activity as a new lead and requires further optimization for drug development. The docking as well as 3D-QSAR studies were performed to determine SAR on our in-house 4-aminoquinoline derivative data set. After exploration, it is clear that H-bond as well as hydrophobic π-π stacking interactions between the quinoline ring and the porphyrin moiety of the heme is crucial for antimalarial activity. Also, the CoMFA/CoMSIA outcomes mostly pointed out the influence of steric, hydrophobic, and electropositive environment-creating groups adjacent to the linker attached to the quinoline moiety for their potential activity against P. falciparum. The steric field around the linker site is most crucial to enhance the biological activity. The end part of the linker region is suitable for steric, hydrophobic, and electropositive groups as well. Apart from that, the C-7 position of the quinoline ring specified the significance of minimum hydrophobicity and the least steric influence for the activity. The present in silico experiments with the newly synthesized compounds offer support for their biological activities. Therefore, it is expected that the present study will provide a new way to design and synthesize novel CQ analogues having antimalarial activity against drug-resistant malaria parasites.

MATERIALS AND METHODS

Synthesis of compounds 6a to 6d, 7a to 7h, 8a to 8d, 9a to 9d, 10a, 10b, 11a to 11c, and 12a.

The synthesis of compounds 6a to 6d, 7a to 7h, 8a to 8d, 9a to 9d, 10a, 10b, 11a to 11c, and 12a involves the following steps starting from the preparation of the tert-butyloxycarbonyl (Boc) amino acids (Fig. 8). Compounds 1a to 1g were converted to the corresponding Boc derivatives (compounds 2a to 2g) in quantitative yields. The Boc-protected amino acids were converted to the corresponding methyl esters by using K₂CO₃-MeI in dimethylformamide solvent. Further, these esters were reduced to alcohols using sodium borohydride (27). Boc amino alcohols were subjected to mesylation to afford compounds 5a to 5g in good yields. The mesylated products were treated with N-substituted piperazines under a nitrogen atmosphere in acetonitrile to get the desired compounds in good yields. Further, Boc deprotection was accomplished using 20% HCl-dioxane at room temperature in quantitative yields as the corresponding hydrochloride salt. Hydrochloride salts were converted to free bases using triethylamine. Finally, the intermediates (6AD to 6DD, 7AD to 7CD, 7ED to 7GD, 8AD to 8DD, 9AD to 9DD, 10AD to 10BD, 11AD to 11CD, and 12AD) obtained were fused to 4,7-dichloroquinoline and/or 4-chloro-7-(trifluoromethyl)quinoline in the presence of phenol to obtain the title compounds (28). All compounds were purified by silica gel column chromatography and characterized by mass spectrometry, high-resolution mass spectrometry, ¹H nuclear magnetic resonance (NMR), and ¹³C NMR spectroscopy.

FIG 8.

Synthesis of compounds 6a to 6d, 7a to 7h, 8a to 8d, 9a to 9d, 10a, 11a to 11c, and 12a.

Biological methods.

(i) In vitro antiplasmodial assay. The compounds were evaluated for antiplasmodial activity against the 3D7 (CQ-sensitive) and K1 (CQ-resistant) strains of Plasmodium falciparum using a malaria SYBR green I nucleic acid staining dye-based fluorescence (MSF) assay as described by Kondaparla et al. (24). The stock solution (10 mM) was prepared in dimethyl sulfoxide (DMSO), and test dilutions were prepared in culture medium (RPMI 1640, fetal bovine serum). Chloroquine diphosphate was used as the reference drug. For assessment of antimalarial activity, 50 μl of culture medium was dispensed in a 96-well plate, followed by addition of 50 μl of the highest concentration (<0.5% DMSO) of the test compounds (in duplicate wells) in row B. Subsequently, 2-fold serial dilutions were prepared in culture medium, and finally, 50 µl of a 2.0% parasitized cell suspension containing 0.8% parasitemia (an asynchronous culture containing more than 80% ring stages) was added to each well, except that 4 wells in row A received a nonparasitized erythrocyte suspension. The plates were incubated at 37°C in a CO₂ incubator in an atmosphere of a 5% CO₂-and-air mixture for 72 h. After 72 h, 100 µl of lysis buffer containing a 2× concentration of SYBR green I (Invitrogen) was added to each well and the plate was incubated for 1 h at 37°C. The plates were examined at 485 ± 20 nm of excitation and 530 ± 20 nm of emission for determination of the number of relative fluorescence units (RFUs) per well using a fluorescence plate reader (FLX800; BioTek). Data were transferred into a graphics program (Excel), and IC₅₀s were obtained by logit regression analysis of the dose-response curves using a preprogrammed Excel spreadsheet. Three replicates were carried out for each compound.

(ii) Determination of hematin 4-aminoquinoline derivative association constant. The association constants for hematin and 4-aminoquinoline derivative complex formation were determined by a spectrometric titration procedure in aqueous DMSO at pH 7.5 (29). Under this assay condition, hematin is strictly in the monomeric state and interpretation of the results is not complicated by the need to consider the hematin disaggregation process. The association constant calculated by this technique is a good refection of the interaction that would occur in the acidic food vacuole of the parasite, and pH 7.5 improves the stability of hematin solutions and the quality of the data.

(iii) In vitro inhibition of β-hematin formation assay. The ability of the 4-aminoquinoline derivatives to inhibit β-hematin formation was induced by 1-oleoyl-rac-glycerol. Spectroscopic measurements were done using a UV spectrophotometer at a λ_max of 405 nm and at pH 5 (30). The IC₅₀s obtained from the assay are expressed as percent inhibition relative to the β-hematin formation in a drug-free control. The IC₅₀s for the compounds were obtained from the sigmoidal dose-response curves using nonlinear regression curve-fitting analyses with GraphPad Prism (v.3.00) software (GraphPad Software, San Diego, CA).

(iv) In vitro assay for evaluation of cytotoxic activity. The cytotoxicity of the compounds was carried out using a Vero cell line (C1008; monkey kidney fibroblast) following the method described by Sinha et al. (23). The cells were incubated with compound dilutions for 72 h, and MTT was used as the reagent for the detection of cytotoxicity. The 50% cytotoxic concentration (CC₅₀) was determined using nonlinear regression analysis of the dose-response curves and the preprogrammed Excel spreadsheet. The selectivity index (SI) was calculated as CC₅₀/IC₅₀.

(v) In vivo antimalarial assay. During in vivo study, animal experimentation was approved by CSIR-Central Drug Research Institute's Institutional Animal Care and Use Committee recognized by the Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA), Government of India. The in vivo drug response was evaluated in Swiss mice infected with the P. yoelii N-67 strain, which is innately resistant to CQ (31). The mice (weight, 22 ± 2 g) were inoculated with 1 × 10⁶ parasitized RBC on day 0, and treatment was administered to a group of five mice from days 0 to 3 once daily. The aqueous suspensions of compounds were prepared with a few drops of Tween 80. The efficacy of the test compounds was evaluated at 100 mg/kg/day, and the required daily dose was administered in a 0.5-ml volume via the oral route. Parasitemia levels were recorded from thin blood smears at regular intervals of 4 days throughout the period of the experiment. The mean value determined for a group of five mice was used to calculate the percent suppression of parasitemia with respect to the level of suppression for the untreated control group. Treatment was considered curative when no parasites were detected until day 28. Mice surviving at day 28 were inoculated with 1 × 10⁶ P. yoelii-parasitized RBC to monitor their susceptibility to rechallenge (32). Mice treated with CQ served as reference controls.

Molecular modeling study.

(i) Database preparation. A molecular docking study was carried out for in-house CQ analogues by targeting heme. The SYBYL-X (v.1.3; Tripos Inc., St. Louis, MO, USA) molecular modeling package software was used for the molecular modeling studies. To execute the docking experiment, an in-house database containing 168 CQ derivatives, including the recently synthesized molecules which are reported on in this paper, was prepared. All compounds were drawn by using the sketch module in SYBYL-X (v.1.3) (33). Structures were minimized further by adding Gasteiger-Huckel charges along with the distance-dependent dielectric and the Powell conjugate gradient algorithms with a convergence criterion of 0.001 kcal/mol. All prepared structures were put into a new database and finally aligned with the most active molecule (compound M96) of the present database by the Fit Atom method.

(ii) Protomol-based docking. The docking study of the synthesized compounds was performed using the Surflex-Dock module with standard protocols in SYBYL-X (v.1.3). Since heme is the primary target of CQ analogues, we extracted the heme from the protein of P. falciparum (PDB accession number 4D6U) to continue the docking experiment (34). Additionally, CQ was taken as a reference to validate our docking experiment. Extracted heme was further prepared through the Surflex-Dock protocol. Hydrogen atoms were added to the heme to define the correct configuration and tautomeric states. Gasteiger-Huckel charges were also added to it, followed by energy minimizations using the Tripos force fields along with the distance-dependent dielectric and the Powell conjugate gradient algorithms with a convergence criterion of 0.001 kcal/mol. After the preparation of heme, an automated protomol generation process was used to identify the grid for docking. The docking algorithm implemented in Surflex uses an idealized active-site ligand called a protomol. This protomol is made up of probes representing sites of potential hydrogen bonds and favorable hydrophobic interactions with the docking target.

Molecular docking of 168 molecules was performed by placing the molecules, including the reference compound CQ, into the protomol-based grid by use of an empirical scoring function to score the ligand- and protomol-guided docking. Finally, the protomol-based method and empirically derived scoring function (e.g., total score, crash score, polar score) were used to calculate the binding affinities. The scoring function comprised hydrophobic, polar, repulsive, entropic, crash, and salvation terms. The total score is expressed in −log dissociation constant (K_d) units to represent total binding affinities. A higher total score represents good binding affinity. Crash is the degree of inappropriate penetration by the ligand into the protein and of the interpenetration (self-clash) between ligand atoms that are separated by rotatable bonds. A smaller crash value (near 0) indicates a better ability to exclude the false positives screened. The polar score represents the contribution of the polar interactions to the total score. The polar score is useful for excluding docking results that make no hydrogen bonds.

(iii) Data set preparation for 3D-QSAR. All docked molecules were further used to prepare a 3D-QSAR model to identify important features for antiplasmodial activity. The in-house database of 168 molecules (Table 1) along with their in vitro antiplasmodial activity (against the P. falciparum 3D7 strain) in the form of the logarithm of the inverse of the inhibitory concentration (−log IC₅₀) was used. For the purpose of CoMFA/CoMSIA, the data set was randomly divided into a training set (111 compounds) and a test set (57 compounds) (Table 6).

TABLE 6.

Distribution pattern of 4-aminoquinoline derivatives in training and test sets

Data set	No. of compounds	Activity (pIC50) distributiona
		Min	Max	Avg	SD
Training	111	5.30	8.33	6.85	0.87
Test	57	5.30	8.48	6.84	0.86

^aMin, minimum; Max, maximum.

(iv) Molecular alignment. The 3D structure-building and all modeling studies were performed using the SYBYL program package (v.7.3) (35) on a silicon graphics fuel workstation with an IRIX (v.6.5) operating system. In this study, the conformations of all molecules (n = 168) were taken from a previous docking experiment. Out of this, one best conformation (based on the total score), that of compound M96, was identified to execute the molecular alignment of all molecules. Before alignment, energy minimizations were performed on docked molecules to add charges using the Tripos force field (36) with a distance-dependent dielectric function and the Powell conjugate gradient algorithm with a convergence criterion of 0.05 kcal/mol. Partial atomic charges were calculated using the Gasteiger-Huckel method. The most potent compound, compound M96, was chosen as the template molecule to fit the remaining training and test compounds by using the database align function in SYBYL. The reference atoms in compound M96 were used for alignment via the Fit Atom method, which is shown in Fig. 9.

FIG 9.

General structures of the all synthesized compounds. The asterisks mark the atoms which are common to all structures and used for the CoMFA/CoMSIA alignment.

(v) CoMFA and CoMSIA. In deriving the CoMFA and CoMSIA descriptor fields, a 3D cubic lattice with a grid spacing of 1 Å extending to 4 Å units beyond the aligned molecules in all directions was created to encompass the aligned molecules. CoMFA descriptors were calculated using an sp³ carbon probe atom with a van der Waals radius of 1.52 Å and a charge of +1.0 to generate steric (Lennard-Jones 6-12 potential) field energies and electrostatic (Coulombic potential) fields with a distance-dependent dielectric at each lattice point. The steric and electrostatic fields generated were scaled by the CoMFA standard method in SYBYL with a default cutoff energy of 30 kcal/mol. The minimum column filtering was set to 2.0 kcal/mol to improve the signal-to-noise ratio by omitting those lattice points whose energy variation was below this threshold. CoMSIA similarity indices were derived as described by Klebe (37) with the same lattice box used for the CoMFA calculations. CoMSIA similarity indices (A_F) for a molecule j with atoms i at grid point q were calculated using equation 1, as follows:

$$A_{F,K}^q(j)=-{\displaystyle \sum _{i=1}^n{\text{ω}}_{\text{probe},k}}{\text{ω}}_{ik}{e}^{-\text{α}{r}_{iq}^2}$$ (1)

where r² is coefficient of determination.

The CoMSIA method involves computation of five different physicochemical properties (k), namely, steric, electrostatic, hydrophobic, hydrogen bond donor, and hydrogen bond acceptor properties, as molecular similarity indices (equation 1). The computational use of Gaussian type distance-dependent function was performed between the grid point q and each atom i in the molecule. A default value of 0.3 was used as the attenuation factor (α). In CoMSIA, the steric indices are related to the third power of the atomic radii; the electrostatic descriptors are derived from partial atomic charges; the hydrophobic fields are derived from atom-based parameters (38).

(vi) PLS analysis. The CoMFA and CoMSIA 3D-QSAR models were derived using the PLS regression procedure of SYBYL (39). The predictive ability of the model was measured in terms of leave-one-out predicted cross-validated r² ($r_{\text{cv}}^2$ or Q²), as shown in equation 2.

$$r_{\text{cv}}^2=1-\frac{\sum {\left(Y_{\text{predicted}}-Y_{\text{observed}}\right)}^2}{\sum {\left(Y_{\text{observed}}-Y_{\text{mean}}\right)}^2}$$ (2)

where Y_predicted, Y_observed, and Y_mean are the predicted, actual (observed), and mean values of the target property (pIC₅₀), respectively. The number of components leading to the lowest standard error of prediction (SEP) was used as the optimum number of components (ONC) to generate the final PLS regression models. The models were validated through the bootstrapping analysis for 100 runs and the cross-validation analysis (leave-half-out and leave 20% out; 50 runs each) (39). The CoMFA and CoMSIA equations were plotted as contour maps to express the percent contribution of the respective fields to the activity. This was done by considering the field energies at each lattice point as a multiple of the regression coefficient and the corresponding standard deviation (SD). Here, the contour maps help in identifying the important regions where changes may affect the binding preference and thereby facilitate the recognition of key features contributing to the interactions between the ligand and the active site of a receptor. Furthermore, the developed CoMFA and CoMSIA models were validated by predicting the activity of the external test set compounds. A model with a predicted r² value (r²_pred) of more than 0.5 may be considered statistically significant.