Mangiferin, an Anti-HIV-1 Agent Targeting Protease and Effective against Resistant Strains

The anti-HIV-1 activity of mangiferin was evaluated. Mangiferin can inhibit HIV-1ⅢB induced syncytium formation at non-cytotoxic concentrations, with a 50% effective concentration (EC50) at 16.90 μM and a therapeutic index (TI) above 140. Mangiferin also showed good activities in other laboratory-derived strains, clinically isolated strains and resistant HIV-1 strains. Mechanism studies revealed that mangiferin might inhibit the HIV-1 protease, but is still effective against HIV peptidic protease inhibitor resistant strains. A combination of docking and pharmacophore methods clarified possible binding modes of mangiferin in the HIV-1 protease. The pharmacophore model of mangiferin consists of two hydrogen bond donors and two hydrogen bond acceptors. Compared to pharmacophore features found in commercially available drugs, three pharmacophoric elements matched well and one novel pharmacophore element was observed. Moreover, molecular docking analysis demonstrated that the pharmacophoric elements play important roles in binding HIV-1 protease. Mangiferin is a novel nonpeptidic protease inhibitor with an original structure that represents an effective drug development strategy for combating drug resistance.

Mangiferin is a small compound with a molecular weight of 422. It is interesting that the PR gene mutants HIV-1 RF/V82F/184V and HIV-1 L10R/M46I/L63P/V82T/I84V are all mangiferin sensitive. To identify essential chemical features of mangiferin, a pharmacophore model was generated using Discovery Studio 2.0 software (Accelrys, Inc.). The pharmacophore model of mangiferin is described by four pharmacophore elements: two H-bond donors and two H-bond acceptors. A reference pharmacophore model was also extracted from approved PIs. After pharmacophore alignment, three pharmacophore elements of mangiferin displayed a high degree of match with the reference pharmacophore model, and one novel element was obtained. The CDOCKER algorithm [25] was employed to investigate a possible binding manner of mangiferin in HIV-1 PR. These results revealed that the pharmacophore binding characters of mangiferin and the PIs were not completely the same. Because we have already seen inhibitory activities of mangiferin against resistant strains, it represents a novel small molecule PI with exciting prospects for combating drug resistance.

Cytotoxicity of Mangiferin
The effect of mangiferin on the viability of cells was examined (Figures 2A and 2C). The results indicated that this compound exhibited low cytotoxicity on C8166, MT-4, chronically-infected H9 cells and PBMC, and the 50% cytotoxic concentration (CC 50 ) values were all above 1,000 μg/mL (2369.67 μM). As control，IDV showed cytotoxicity on C8166 with the CC 50 value above 500 μM.  The results are expressed as the mean ± SEM for at least three independent experiments.

Anti-HIV-1 Activity and Mechanism of Mangiferin
The dose-dependent inhibitory activity of mangiferin on HIV-1 induced syncytium formation (cytopathic effects) is shown in Figure 2B. The EC 50 values of mangiferin in HIV-1 IIIB and HIV-1 RF for inhibiting cytopathic effects were 7.13 μg/mL (16.90 μM) and 15.45 μg/mL (36.61 μM), respectively. The therapeutic indexes (TI) of this compound were above 140 and 65, respectively. Moreover, the activity of mangiferin in the protection of MT-4 cells from lysis induced by HIV-1 IIIB is shown in Figure 2C, in which the TI value was above 26.26.
The antiviral activities of mangiferin were not limited to prototypic viruses. The EC 50 values from the p24 antigen assay for the primary HIV-1 isolate HIV-1 KM018 and the EC 50 values from the syncytium assay for the nonnucleoside reverse transcriptase inhibitors (NNRTIs) resistant strain HIV-1 A17 were 14.94 μg/mL (35.40 μM) and 9.60 μg/mL (22.75 μM), respectively (Table 1). We used a time-of-addition assay to determine the stage of the HIV-1 replication cycle with which mangiferin interfered. After HIV-1 IIIB infected C8166 cells, mangiferin added at various times blocked p24 antigen production ( Figure 3). The results demonstrated that mangiferin might have inhibited HIV-1 PR activity because it blocked HIV-1 p24 antigen production after 12 h. While in a parallel assay, AZT and IDV blocked HIV-1 p24 antigen production after 4 h and 12 h, respectively. Mechanism studies revealed that mangiferin had no activity on HIV-1 entry, IN or RT, but had some inhibition of HIV-1 PR with an EC 50 of 144.66 μg/mL (342.80 μM) ( Table 1). The results showed that HIV-1 PR might the most important target.

The Sensitivity of Mangiferin to HIV-1 Protease Gene Mutants
Two HIV-1 PR gene mutants (HIV-1 L10R/M46I/L63P/V82T/I84V and HIV-1 RF/V82F/184V ) resistant to PIs were used to confirm the action of mangiferin. It was interesting that mangiferin was effective in these two mutants ( Figure 4), with EC 50 values of 6.97 μg/mL (16.52 μM) and 14.14 μg/mL (33.51 μM), respectively. The inhibition activity was similar in HIV-1 IIIB and HIV-1 RF . This was significant, as these viral lines are resistant to conventional chemotherapy (IDV EC 50 values of 166.75 μM and 197.13 μM, respectively, while the EC 50 of IDV in wild-type HIV-1 IIIB is only 2.72μM). The result implied that mangiferin had different binding mode with IDV. Since IDV was one of the typical PIs, it is of interest to find out the difference.

Pharmacophore Elements of Mangiferin
The pharmacophore model was generated and the top 10 hypotheses were exported. The first hypothesis (Hypo1), was output ranked automatically as the best pharmacophore hypothesis, based on the highest ranking score and good fit values of the inner ranking function in Discovery Studio 2.0 software (Accelrys, Inc.). The pharmacophore models of the reference commercial drugs were described by three pharmacophore elements: two H-bond acceptors (R_HBA2.11, R_HBA3.11) and one H-bond donor (R_HBD1.11). The pharmacophore model of mangiferin contained the same two H-bond acceptors (M_HBA4.11, M_HBA3.11) and one H-bond donor (M_HBD1.11) seen in the reference models. Moreover, one novel H-bond donor (M_HBD2.11) was observed in mangiferin.
Pharmacophores are a set of structural features in a molecule that are recognized at a receptor site and are responsible for that molecule's biological activity [26]. The pharmacophoric elements exhibited by mangiferin were largely similar to those of drugs currently on the market. Similar pharmacophoric feature is common for molecules with the same biological activity. The similar pattern indicated that mangiferin possessed the basic structural foundation for anti-HIV-1 activity. Pharmacophore model analysis prompted us to investigate the binding model of mangiferin in the HIV-1 RF PR complex.

Binding Mode of Mangiferin in HIV-1 Protease
Ten HIV-1 RF mutants were generated, and refinement was carried out at a high optimization level. The protein structures were evaluated using the DOPE [27] (Discrete Optimized Protein Energy) scoring function. The highest DOPE score mutant (the DOPE Score value was -24336.89) was selected for subsequent analysis. From the mangiferin 3D structure, 255 best-quality diverse conformations were generated as a ligand set. Each ligand produced 100 random conformations during docking. The final results exported 400 refined poses. The pose with the highest -CDOCKER_INTERACTION_ ENERGY value (61.2) indicated the most favorable binding and was selected for HIV-1 RF PR and mangiferin interaction analysis.
The binding mode of mangiferin in the active site of HIV-1 RF PR is shown in Figure 5B. As can be seen, there are many H-bonds between the mangiferin hydroxyl group and various HIV-1 RF PR residues in this complex structure. Mangiferin pharmacophoric element M_HBD4.11, corresponding to the 6'-hydroxyl group of the sugar moiety, forms three H-bonds with the amino acid side chain (Asp30, Lys45) of HIV-1 RF PR and the 3'-hydroxyl group of mangiferin. Moreover, the 2'-hydroxyl group of mangiferin's sugar moiety interacts with the amino acid main chain (Gly48) of HIV-1 RF PR through H-bonds. The novel pharmacophoric element was observed to play a key role in binding HIV-1 RF PR: the 6-hydroxyl group of mangiferin forms a hydrogen bond with the catalytic residue Asp25. Furthermore, other interactions were also observed in the complex structure, including charged, polar and van der Waals interactions ( Figure 5B).
Docking results showed that the modeled pharmacophore hydrogen bonding groups located in the 'head' and 'tail' of mangiferin interacted with HIV-1 RF PR. Furthermore, novel pharmacophoric features were observed in mangiferin that played a key role in binding HIV-1 RF PR catalytic residue Asp25 and may offer clues to understanding its effectiveness against resistant strains.

Cells and Viruses
Human T-cell lines (C8166, MT-4) and chronically infected H9/HIV-1 IIIB were kindly donated by the Medical Research Council (MRC), the AIDS Reagent Project, U.K. All of the cell lines and viruses were maintained at 37 °C under 5% CO 2 in RPMI-1640 medium (Gibco) supplemented with 10% (v/v) heat-inactivated newborn calf serum (NCS). Peripheral blood mononuclear cells (PBMCs) from healthy donors were isolated by Ficoll-Hypaque centrifugation and incubated in complete medium containing 5 μg/mL PHA for 72 h prior to use in antiviral assays.
The laboratory-derived strains HIV-1 IIIB and HIV-1 RF , The NNRTIs resistant strain HIV-1 A17 , PR Gene Mutants HIV-1 RF/V82F/184V and HIV-1 L10R/M46I/L63P/V82T/I84V were kindly donated by NIH. The clinically isolated strain HIV-1 KM018 was obtained from a naïve HIV-1 infected individual in Yunnan province of China. The 50% HIV-1 tissue culture infectious dose (TCID 50 ) in C8166 cells was calculated by the Reed and Muench method. Virus stocks were stored in small aliquots at −70 °C.

Cytotoxicity Assay
The cytotoxicity of the compounds on C8166 cells was determined by a MTT colorimetric assay as described previously [29]. Aliquots of 100 μL/well (4 × 10 5 /mL) C8166 cell suspension were seeded on a microtiter plate. Next, 100 μL/well of various concentrations of compounds was added and the plates were incubated at 37 °C and 5% CO 2 for 3 days. The MTT reagent was added and incubated for 4 h, 100 μL of supernatant was discarded and then 100 μL of 50% DMF-20% SDS was added. The plates were read on an ELISA reader (Elx800, Bio-Tek) at 595/630 nm. The CC 50 was calculated from the dose response curve.

Antiviral Activity Against Acute HIV-1 Infection
The inhibitory activities of mangiferin against the HIV-1 B Ⅲ and HIV-1 RF , HIV-1 KM018 , HIV-1 A17 , HIV-1 L10R/M46I/L63P/V82T/I84V and HIV-1 RF/V82F/184V were determined as previously described [29][30][31]. Briefly, 4 × 10 4 C8166 cells were infected with different HIV isolates at a multiplicity of infection (M.O.I.) of 0.06-0.1 for 2-4 h. Then the plates were incubated in the presence or absence of 100 μL of various concentrations of mangiferin. NVP, IDV and AZT were used as positive drug controls. After 3-7 days of culture, the percentage inhibition of syncytia formation was scored or the level of p24 was measured by ELISA and the EC 50 was calculated.

Protection for HIV-1 Induced Lytic Effects
The activities of the compound against acute HIV-1 infection were based on the inhibition of HIV-1 induced cytopathogenicity in MT-4 cells, as described previously [31]. Uninfected or HIV-1 IIIB -infected (MOI = 0.1) MT-4 cells (3 × 10 5 cells/mL) were seeded in microtiter plates with 100 μL of different concentrations of mangiferin. AZT was used as the control drug. After a 7-day incubation, the viability of both HIV-1 and mock-infected cells were assessed using the MTT method.

Co-culture Assay
Cell-to-cell fusion between normal C8166 cells and H9 cells chronically infected withHIV-1 IIIB was quantified under an inverted microscope. 3 × 10 4 C8166 cells were co-cultured with 1 × 10 4 H9/HIV-1 IIIB cells in the presence or absence of mangiferin at varying serial concentrations. T-20 was used as positive drug control. After an 8 h incubation, the number of syncytia was counted under an inverted microscope [31].

Time of Addition Assay
To determine the stage of the HIV replication cycle with which this anti-HIV compound interfered, a time-of-addition experiment was carried out. C8166 cells were exposed to HIV at M.O.I of 0.2. To ensure that the virus replication steps were synchronized in the whole-cell population, infected cells were incubated for 2 h at 4 °C. After allowing adequate time for adsorption, the unabsorbed virus was removed by washing twice with complete medium. The temperature was then shifted to 37 °C, and mangiferin was added at different times (0, 2, 4, 6, 8, 12, 24 and 48 h) after adsorption. AZT and IDV were used as positive drug control. After 3 days of culture, HIV-1 p24 expression was detected by ELISA.

HIV Reverse Transcriptase, Integrase and Protease Assay
HIV-1 reverse transcriptase (RT) activity was measured by ELISA using a commercially available kit (Roche) according to the instructions of the manufacturer [30]. PFA was used as positive drug control. The interaction between compounds and HIV-1 integrase (IN) was determined by SPR biosensor technology using a BIAcore 3000 biosensor system (Biacore Inc., Piscataway, NJ). HIV-1 IN were immobilized on the surface of the chip, and 200 μg/mL of compounds diluted in HBS-EP was applied. Dissociation of compounds from the IN was monitored after washing the chip and the kinetic rate constants for dissociation (Kd) was obtained by fitting the real-time data using BIA evaluation software [32].
The recombinant HIV-1 PR was expressed and purified as previously described [33]. HIV-1 PR was diluted in reaction buffer and the compounds were added and incubated for 30 min at room temperature. Fluorescent substrate DABCYL-γ-Abu-Ser-Gln-Asn-Tyr-Pro-Ile-Val-Gln-EDANS (Anaspec, San Jose) was added to initiate the reaction. The mixture was allowed to react for 90 min and the change of the fluorescent signal was monitored. IDV was used as positive drug control. The percentage of inhibition was calculated from the change in fluorescence.
The reference pharmacophore features were generated based on ten drugs currently on the market and were verified by virtual screening of the database, which consists of diverse conformers from 10 commercially available drugs. Using the HipHop algorithm [36] internal scoring function, the best pharmacophores were output first. To find a consensus pharmacophore or distinguishable feature, the Pharmacophore Comparison Protocol aligned mangiferin's pharmacophores to reference pharmacophores.

Molecular docking
Using the known structure as a template (PDB code: 3D3T), several mutants were constructed from its sequence by MODELER [37] to generate the HIV-1 RF structure. The MODELER uses different scoring functions and optimization protocols [38] to optimize all the atoms of the mutated residues. The structures of mangiferin came from the ACD3D database (Symyx, Inc.). Afterwards, best-quality diverse mangiferin conformations were generation to ensure the best coverage of conformational space. The set of diverse conformers of mangiferin was used for docking into the HIV-1 RF active pocket. Docking was performed using CDOCKER [25]. The models were energy minimized with the CHARMm force field before performing docking. CDOCKER reports scores based on internal ligand strain energy and receptor-ligand interaction energy. All docking and pharmacophore analysis were performed using Discovery Studio 2.0 software (Accelrys, Inc.).

Conclusions
Mangiferin exhibited low cytotoxicity and good inhibitory activity on HIV-1 replication in a dosedependent manner. The antiviral activities of mangiferin were observed to primary HIV-1 isolate and resistant strains also. Mechanism studies revealed that mangiferin might inhibit the HIV-1 protease, and is still effective against HIV peptidic protease inhibitor resistant strains. Furthermore, three pharmacophore elements of mangiferin displayed a high degree of matching with the reference pharmacophore model from approved PIs and as a key role in binding HIV-1 RF PR. Though the anti-HIV-1 activities were not as good as approved PIs, our results suggest that mangiferin may be a novel NPPI with an original structure that represents an effective drug development strategy for combating drug resistance.