Anti-Human Immunodeficiency Virus Type 1 Activity of Novel 6-Substituted 1-Benzyl-3-(3,5-Dimethylbenzyl)Uracil Derivatives

Abstract

Nonnucleoside reverse transcriptase (RT) inhibitors (NNRTIs) are important components of current combination therapies for human immunodeficiency virus type 1 (HIV-1) infection. In screening of chemical libraries, we found 6-azido-1-benzyl-3-(3,5-dimethylbenzyl)uracil (AzBBU) and 6-amino-1-benzyl-3-(3,5-dimethylbenzyl)uracil (AmBBU) to be highly active and selective inhibitors of HIV-1 replication in vitro. To determine the resistance profiles of these compounds, we conducted a long-term culture of HIV-1-infected MT-4 cells with escalating concentrations of each compound. After serial passages of the infected cells, escape viruses were obtained, and they were more than 500-fold resistant to the uracil derivatives compared to the wild type. Sequence analysis was conducted for RT of the escape viruses at passages 12 and 24. The amino acid mutation Y181C in the polymerase domain of RT was detected for all escape viruses. Docking studies using the crystal structure of RT showed that AmBBU requires the amino acid residues Leu100, Val106, Tyr181, and Trp229 for exerting its inhibitory effect on HIV-1. Four additional amino acid changes (K451R, R461K, T468P, and D471N) were identified in the RNase H domain of RT; however, their precise role in the acquisition of resistance is still unclear. In conclusion, the initial mutation Y181C seems sufficient for the acquisition of resistance to the uracil derivatives AzBBU and AmBBU. Further studies are required to determine the precise role of each mutation in the acquisition of HIV-1 resistance.

INTRODUCTION

Human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) is responsible for synthesizing a double-stranded integrative cDNA from the single-stranded viral genomic RNA in the early virus life cycle. HIV-1 RT is a heterodimer composed of two subunits, p66 and p51, and p51 is generated by the proteolytic processing and removal of C terminus (amino acids 441 to 560) from p66 (32). The p66 subunit is composed of two spatially distinct domains: polymerase (residues 1 to 426) and RNase H (residues 427 to 560). The polymerase domain is composed of four subdomains: fingers (residues 1 to 85 and 118 to 155), palm (residues 86 to 117 and 156 to 236), thumb (237 to 318), and connection (319 to 426) (46). The polymerase domain creates a copy of the viral genome, while the RNase H domain promotes RNA degradation from the DNA/RNA duplex during reverse transcription. Other RNase H functions include the removal of tRNA₃^Lys, and the polypurine tract, which are primers for minus- and plus-strand DNA synthesis (11, 47, 58).

The polymerase domain is currently targeted by two classes of antiretroviral drugs, nucleoside RT inhibitors (NRTIs) and non-nucleoside RT inhibitors (NNRTIs). NNRTIs are important components of current antiretroviral therapies for HIV-1. To date, more than 50 structurally diverse classes of compounds have been identified as genuine NNRTIs (60). Earlier NNRTIs include nevirapine (NVP), delavirdine (DLV), and efavirenz (EFV), and recent NNRTIs include etravirine (ETR). Several other NNRTIs, such as emivirine (MKC-442), underwent clinical trials, and yet they were not approved due to unfavorable pharmacokinetics, insufficient efficacy, and/or safety concerns (50). Recently, rilpivirine (RPV; formerly TMC278) has been formally licensed for clinical use in treatment-naive adult patients (56), while other NNRTIs, including IDX899, RDEA-428, and lersivirine, are currently under clinical development (1, 12, 30).

Although NNRTIs are generally well tolerated, a major limitation for all currently available NNRTIs is the low genetic barrier to resistance, which allows rapid emergence of drug resistance caused by a small number of amino acid mutations in the target region. HIV-1 drug resistance mutations in RT are extensively characterized for NRTIs and NNRTIs (29). NNRTIs inhibit HIV-1 by allosteric binding to a hydrophobic pocket in the RT about 10 Å behind the catalytic site (48). The positions associated with NNRTI resistance that make up the central NNRTI binding pocket are L100, K101, K103, V106, V108, V179, Y181, Y188, G190, F227, and W229. Additional positions that make up the pocket include E138, which is contributed by the p51 subunit, and M230, L234, P236, K238, and Y318, which form part of an extended pocket. Additional accessory NNRTI resistance abutting positions that form the NNRTI binding pocket include A98 and P225.

Mutations that are selected after failure of treatment with NNRTIs are located in the enzyme hydrophobic pocket, and they reduce the binding affinity of the inhibitors to the enzyme (7). A single mutation in the NNRTI-binding pocket may confer high-level resistance to one or more NNRTIs. Since NVP, DLV, and EFV have similar binding modes to RT, viruses resistant to one of these compounds develop cross-resistance to the others. ETR maintains its activity against NVP-, DLV-, or EFV-resistant mutants due to its ability of multiple binding modes to RT.

In order to improve the design of novel NNRTIs for future clinical development, in vitro isolation and analyses of drug-resistant viruses are necessary to obtain valuable information on the resistance patterns of novel compounds. We previously synthesized and evaluated nine novel uracil analogues as NNRTIs, including four 1-substituted 3-(3,5-dimethylbenzyl)-5-fluorouracils and five 6-substituted 1-benzyl-3-(3,5-dimethylbenzyl)uracils (28). Two of these compounds—6-azido-1-benzyl-3-(3,5-dimethylbenzyl)uracil (AzBBU) and 6-amino-1-benzyl-3-(3,5-dimethylbenzyl) uracil (AmBBU)—were found to be highly active and selective inhibitors of HIV-1 replication in vitro.

In the present study, we conducted a long-term culture experiment with HIV-1-infected MT-4 cells with escalating concentrations of AzBBU and AmBBU. After serial passages of the infected cells, escape viruses were obtained, which displayed complete resistance to these compounds. Sequence analysis of RT from escape viruses was performed to determine the mutations related to the acquisition of resistance.

MATERIALS AND METHODS

Compounds.

AzBBU and AmBBU were synthesized as previously described (28). The lead compound 1-benzyl-3-(3,5-dimethylbenzyl) uracil (BBF-29), 6-benzyl-1-ethoxymethyl-5-isopropyluracil (MKC-442), and the nucleoside analog 2′,3′-didehydro-3′-deoxy-4′-ethynylthymidine (4′-Ed4T) were prepared according as previously described (25, 34, 52). All compounds were dissolved in dimethyl sulfoxide at 100 mM and stored at −20°C until use. The chemical structures of AzBBU, AmBBU, BBF-29, MKC-442, and 4′-Ed4T are shown in Fig. 1.

Fig 1.

Structures of BBF-29, AzBBU, AmBBU, MKC-442, and 4′-Ed4T.

Cells and viruses.

MT-4 and M8166 cells were maintained in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum, 100 U of penicillin G/ml, and 100 μg of streptomycin/ml. The HIV-1 strain III_B, the HIV-1 resistance strain III_B-R, and the HIV-2 strain ROD were used throughout the antiviral experiments (Table 1). III_B-R is a NNRTI-resistant mutant established by a serial passage of infected cells in the presence of increasing concentrations of MKC-442 (5). Viruses were propagated and titrated in MT-4 cells (HIV-1) or M8166 cells (HIV-2). Virus stocks were stored at −80°C until use. Escape viruses obtained after long-term culture with AzBBU and AmBBU were used for the following anti-HIV-1 experiments (Table 2).

Table 1.

Antiviral activity of AzBBU and AmBBU against HIV-1 and HIV-2

Compound	Antiviral activity (mean EC50 [μM] and CC50 [μM] ± SD)a against:
	HIV-1 IIIB		HIV-1 IIIB-R		HIV-2 ROD
	EC50	CC50	EC50	CC50	EC50	CC50
AzBBU	0.088 ± 0.009	40.5 ± 6.7	>45.1	45.1 ± 0.5	>40.0	40.0 ± 2.7
AmBBU	0.060 ± 0.011	50.1 ± 1.1	>50.5	50.5 ± 6.3	>45.9	45.9 ± 1.2
BBF-29	0.26 ± 0.02	43.2 ± 9.1	13.9 ± 6.0	>100	>100	>100
MKC-442	0.015 ± 0.002	>100	6.2 ± 1.4	>100
NVP	0.057 ± 0.005	>100	52.1 ± 23.6	>100
4′-Ed4T	0.029 ± 0.008	>100	0.053 ± 0.038	>100	0.019 ± 0.007	>100

^aEC₅₀, 50% effective concentration that inhibits virus-induced CPE of infected cells by 50%; CC₅₀, 50% cytotoxic concentration that reduces the viability of uninfected cells by 50%.The data represent the means for three independent experiments.

Table 2.

Anti-HIV-1 activity of AzBBU and AmBBU against escape viruses

Compound	Anti-HIV-1 activity (mean EC50 [μM] and CC50 [μM] ± SD)a
	Passage 12				Passage 24
	IIIB-AZ12		IIIB-AM12		IIIB-AZ24		IIIB-AM24
	EC50	CC50	EC50	CC50	EC50	CC50	EC50	CC50
AzBBU	>44.5* (>500)	44.5 ± 0.3	>42.6* (>480)	42.6 ± 3.0	>43.6* (>500)	43.6 ± 0.5	>44.0* (>500)	44.0 ± 1.0
AmBBU	>60.7* (>1,000)	60.7 ± 6.5	>56.2* (>940)	56.2 ± 5.7	>48.1* (>800)	48.1 ± 4.8	>47.0* (>780)	47.0 ± 7.6
BBF-29	>46.4* (>180)	46.4 ± 9.5	>42.5* (>160)	42.5 ± 3.9	>49.4* (>190)	49.4 ± 3.0	>45.9* (>170)	45.9 ± 7.2
MKC-442	3.8 ± 1.2* (257)	>100	2.5 ± 0.6* (168)	>100	5.4 ± 3.5 (369)	>100	3.8 ± 1.2* (257)	>100
NVP	49.9 ± 10.9* (875)	>100	32.6 ± 4.6* (572)	>100	37.2 ± 3.5* (652)	>100	33.6 ± 6.7* (589)	>100
4′-Ed4T	0.038 ± 0.029 (1.3)	>100	0.092 ± 0.011* (3.2)	>100	0.028 ± 0.012 (1)	>100	0.023 ± 0.003 (0.8)	>100

^aEC₅₀, 50% effective concentration that inhibits virus-induced CPE of infected cells by 50%; CC₅₀, 50% cytotoxic concentration that reduces the viability of uninfected cells by 50%. Fold changes, based on the EC₅₀ for the wild type, are indicated in parentheses. The data represent the means for three independent experiments. Statistical analysis (Student t test) was performed for each EC₅₀ compared to that of the wild type (*, P < 0.01).

Antiviral assays.

The antiviral activity of test compounds against HIV-1 and HIV-2 was determined by the inhibition of virus-induced cytopathic effect (CPE) (4). Briefly, MT-4 or M8166 cells (10⁵ cells/ml) were infected with HIV-1 or HIV-2, respectively, at a multiplicity of infection (MOI) of 0.1 and were cultured in the presence of various concentrations of the tested compounds. After a 4-day incubation at 37°C, the number of viable cells was monitored by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) method (39). The cytotoxicity of the compounds was evaluated in parallel with their antiviral activity, based on the viability of mock-infected cells, as determined by the MTT method. All experiments were performed in triplicate.

Long-term culture of infected MT-4 cells.

MT-4 cells were infected with the HIV-1 strain III_B and incubated at 37°C for 4 days in the presence of compounds. The initial concentration of AzBBU and AmBBU corresponded to 2-fold higher than their 50% effective concentrations (EC₅₀s), i.e., 0.175 and 0.119 μM, respectively. As control cultures, exactly identical passages of the infected MT-4 cells in the absence of the compounds were carried out in parallel with the cultures exposed to the compounds. At each passage, virus-induced CPE of the cells was monitored to confirm virus replication. The concentration of each compound was escalated 2-fold, when the CPE in the compound-treated culture exceeded 70%. The escape viruses as well as the control viruses were propagated once in MT-4 cells, and they were used for further experiments.

Sequence analysis of RT.

Genomic DNA was extracted from the infected MT-4 cells with a DNA extraction kit (Wako, Tokyo, Japan). The extracted DNA was quantified using a NanoDrop spectrophotometer ND-1000 (NanoDrop Technologies, Wilmington, DE) and subjected to nested PCR. Two regions of RT were amplified (RT1 and RT2). The first PCR consisted of an initial denaturation at 95°C for 2 min, followed by 40 cycles (95°C for 1 min, 46°C for 1 min, and 72°C for 1 min) and a final extension at 72°C for 5 min. The primers rt-1-f (5′-AGGGGGAATTGGAGGTTT-3′) and rt-1-r (5′-TCCCACAACTTCTGTATGTC-3′) were used to amplify the region RT1, and the primers rt-2-f (5′-ATGAACTCCATCCTGATAAATG-3′) and rt-2-r (5′-TGTACAATCTAGTTGCCATAT-3′) were used to amplify the region RT2. The second PCR consisted of an initial denaturation at 95°C for 2 min, followed by 40 cycles (95°C for 1 min, 48°C for 1 min, and 72°C for 1 min) and a final extension at 72°C for 5 min. The primers rt-12-f (5′-CCAGTAAAATTAAAGCCAG-3′) and rt-12-r (5′-TCCCACTAACTTCTGTATGTC-3′) were used to amplify the region RT1, and the primers rt-22-f (5′-CCAGAAAAAGACAGCTGGACT-3′) and rt-22-r (5′-TGGCAGGTTAAAATCACTAGCC-3′) were used to amplify the region RT2. The second PCR generated fragments encompassing nucleotides 2120 to 2881 (RT1) and nucleotides 2834 to 3862 (RT2) of the RT gene corresponding to the HIV-1 complete genome (GenBank accession number AF033819). The amplified products were confirmed by capillary gel electrophoresis using the Agilent Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA). The PCR products were sequenced directly with a cycle sequence kit BigDye Terminator version 3.1 (Applied Biosystems, Foster City, CA), using both forward and reverse primers on an automated DNA analyzer model 3730 (Applied Biosystems), according to the manufacturer's instructions.

Docking study of compounds against HIV-1 RT.

All in silico studies were performed using Molecular Operating Environment (MOE) software (Chemical Computing Group, Montreal, Quebec, Canada). The X-ray crystal structure of HIV-1 RT (PDB code 3m8p) (40) was downloaded from PDB at the Research Collaboration for Structural Bioinfomatics (http://www.rcsb.org/pdb/home/home.do) and optimized for the docking study by removing ligand and water, adding hydrogen atoms, assigning atomic charges, and minimizing using the Merck molecular force field 94X (MMFF94x) (22, 23, 55). Based on this structure, the structure of the HIV-1 RT (Y181C) was constructed. The docking site at the HIV-1 RT structure was searched by Alpha Site Finder, a function of MOE. Partial charges were added to the compound and a maximum of 250 conformers were generated using MMFF94x. MOE-ASEDock 2005 (Ryoka Systems, Tokyo, Japan) was then used for the docking of the compound to HIV-1 RT, and docking scores were calculated (19).

Statistical analysis.

Statistical analysis for the EC₅₀s of the test compounds against the wild-type and resistant viruses was performed using an unpaired two-tailed Student t test. P values of <0.01 were considered statistically significant.

RESULTS

Antiviral activity of AzBBU and AmBBU against HIV-1 and HIV-2.

AzBBU and AmBBU were tested for their inhibitory effects on the replication of HIV-1 III_B, HIV-1 III_B-R, and HIV-2 ROD. The NNRTIs BBF-29, MKC-442, and NVP, as well as the nucleoside analog 4′-Ed4T, were also tested for comparison. Their activities are given in Table 1. AzBBU and AmBBU showed high activity against HIV-1 III_B with similar EC₅₀s (0.088 ± 0.009 and 0.060 ± 0.011 μM, respectively) and 50% cytotoxic concentrations (CC₅₀s) (40.5 ± 6.7 and 50.1 ± 1.1 μM, respectively). AmBBU showed a higher selectivity index (SI) than AzBBU (SI = 835 versus SI = 460). These results are in accordance with those in the previous report (28). Although these compounds showed higher anti-HIV-1 activity against III_B compared to the lead compound BBF-29 (0.26 ± 0.02 μM), they were not active against the NNRTI-resistant HIV-1 stain III_B-R. In addition, AzBBU and AmBBU did not show any activity against HIV-2 ROD. In contrast, NRTI 4′-Ed4T was equally active against HIV-1 III_B, HIV-1 III_B-R, and HIV-2 ROD.

Isolation of escape viruses.

Long-term cultures of HIV-1 (III_B strain)-infected MT-4 cells were started in the absence or presence of AzBBU and AmBBU (Fig. 2). The concentration of each compound was escalated 2-fold, when the CPE in the compound-treated culture exceeded 70%. At passage 24, the concentrations of AzBBU and AmBBU could reach 256-fold their EC₅₀s (22.4 and 15.2 μM, respectively). Viruses were isolated from culture supernatants at passages 12 and 24 (Fig. 2, points a, b, c, and d) and subjected to phenotypic and genotypic analyses.

Fig 2.

Long-term culture of infected MT-4 cells with escalating concentrations of AzBBU and AmBBU. MT-4 cells were infected with HIV-1 (III_B strain) and passaged every 4 days. Viral replication was monitored by determining the CPE of the cells at each passage. Culture supernatants from passages 12 and 24 were used for further experiments (CPE > 70%). Isolated viruses (points a, b, c, and d) were subjected to phenotypic and genotypic analyses. Points: a, passage 12, AzBBU (III_B-AZ12); b, passage 24, AzBBU (III_B-AZ24); c, passage 12, AmBBU (III_B-AM12); d, passage 24, AmBBU (III_B-AM24).

Anti-HIV-1 activity of AzBBU and AmBBU against escape viruses.

When AzBBU and AmBBU were examined for their activity against the escape viruses obtained at passages 12 and 24, the compounds did not show any significant inhibition at their nontoxic concentrations (Table 2). Thus, the isolates were more than 500-fold resistant to AzBBU and AmBBU compared to the wild type. The lead compound BBF-29 was also inactive against the escape viruses. The viruses had partial cross-resistance to MKC-442, probably due to its structural similarity (Fig. 1 and Table 2). NVP marginally inhibited the replication of the escape viruses. In contrast, 4′-Ed4T was equally inhibitory to the replication of the escape viruses and the wild type (Table 2).

Amino acid changes of escape viruses.

To determine what amino acid changes are associated with resistance to AzBBU and AmBBU, sequence analysis of full-length RT genes from escape viruses (III_B-AZ12, III_B-AM12, III_B-AZ24, III_B-AM24), as well as the wild-type control without treatment (III_B), was further performed at passages 12 and 24. The sequences of the escape viruses, III_B-AzBBU and III_B-AmBBU, were deposited in a public database (GenBank accession numbers JQ070415 and JQ070416). In addition, the sequence of the resistant strain III_B-R was also analyzed for comparison (GenBank accession number JQ070417). Figure 3 shows the RT (subunit p66) amino acid sequences of escape viruses. Several synonymous mutations were observed in the RT gene. Overall, 5 amino acid changes were observed. One was in the polymerase domain, and four were in the RNase H domain. The same amino acid changes were detected at passages 12 and 24 (Fig. 3). Mutation Y181C in the polymerase domain was identified in all escape viruses. The resistant strain III_B-R displayed the mutation V108I in addition to Y181C. Four additional amino acid changes (K451R, R461K, T468P, and D471N) were detected in the RNase H domain of all escape viruses.

Fig 3.

RT (subunit p66) amino acid sequences of the escape viruses. The control sequence without treatment was used as reference (WT). Asterisks (*) indicate the positions for synonymous mutations. Black arrows indicate common NRTI resistance mutations. White arrows indicate common NNRTI resistance mutations.

Docking studies of AmBBU.

Docking of the metabolically relevant derivative AmBBU to the binding-pocket of HIV-1 RT was performed. Figure 4 shows the proposed interactions between HIV-1 RT and AmBBU for HIV-1 RT wild type (Fig. 4A) and the mutant Y181C (Fig. 4B). AmBBU interacted with the amino acid residues Leu100, Val106, Tyr181, and Trp229 of HIV-1 RT (wt) through arene-H interactions (arene-H) (Fig. 4A), such as the hydrogen in the side chain of Leu100 with the central benzene ring (2-pyrimidine) of AmBBU, the hydrogen in the side chain of Val106 with 1-benzyl of AmBBU, the phenyl ring in the side chain of Tyr181 with hydrogen (3-methyl) at 3-(3,5-dimethylbenzyl) of AmBBU, and the indol ring in the side chain of Trp229 with 4′-hydrogen of 3-(3,5-dimethylbenzyl) of AmBBU (D2 representation in Fig. 4A). The compound lost the interaction at position 181 when replaced by cysteine (Y181C) (Fig. 4B). According to these data, the docking score between HIV-1 RT (wt) and AmBBU (−13.5110 kcal/mol) was higher than that between HIV-1 RT (Y181C) and AmBBU (−11.1648 kcal/mol).

Fig 4.

Docking of AmBBU to the binding pocket of HIV-1 RT. Docking structures between HIV-1 RT and AmBBU are shown for HIV-1 RT wt (A) and the mutant Y181C (B). The backbone is represented by white ribbons, and the side chains (residues 100, 106, 181, and 229) are represented by colored wire style. AmBBU is represented by magenta wire style. Dotted lines show the interactions between HIV-1 RT and AmBBU. A D2 representation of the interactions is presented below the docking.

DISCUSSION

Initially, we evaluated the antiviral activities of AzBBU and AmBBU against the HIV-1 strains III_B and III_B-R and the HIV-2 strain ROD. AzBBU and AmBBU were highly active against III_B, and the EC₅₀s were similar for both compounds (Table 1). This is explainable, since the 6-azido uracil derivative (AzBBU) may be reduced metabolically to its 6-amino congener (AmBBU) in cell cultures (28). These compounds showed higher activity against III_B than the 1-substituted 3-(3,5-dimethylbenzyl)uracils previously reported (33). In contrast, AzBBU and AmBBU were not active against III_B-R, although the lead compound BBF-29 had weak activity. This may be due to the azido and amino groups introduced at 6-position of the 1-benzyl moiety, which is not present in BBF-29 (Fig. 1). Thus, 6-azido and 6-amino substitutions on the uracil ring increased the antiviral activity against III_B but decreased the antiviral activity against III_B-R. In addition, AzBBU and AmBBU did not show any activity against HIV-2 ROD. This is consistent with previous reports showing that HIV-2 is intrinsically resistant to most NNRTIs (14, 57). As expected, the nucleoside analog 4′-Ed4T was almost equally active against HIV-1 III_B, III_B-R, and HIV-2 ROD.

We isolated two HIV-1 strains highly resistant to AzBBU and AmBBU through long-term culture of HIV-1 III_B-infected MT-4 cells (Fig. 2). The phenotypic analysis revealed that escape viruses had partial cross-resistance to MKC-442, probably due to its structural similarity (Table 2 and Fig. 1). A similar finding has been reported for 1-[(2-hydroxyethoxy)methyl]-6-(phenylthio)thymine (HEPT) derivatives, where the HEPT-resistant virus also displayed cross-resistant to virtually all of other HIV-1-specific inhibitors, including NVP (6). 4′-Ed4T was equally active against the AzBBU- and AmBBU-resistant viruses compared to the wild type because of its different mechanism of action.

Our genotypic analysis revealed amino acid changes associated with resistance to AzBBU and AmBBU (Fig. 3). Several mutations were observed in the RT gene of resistant viruses, but some of them corresponded to synonymous mutations (Fig. 3). Amino acid changes were accumulated within a short period of cultivation (from passage 12, 48 days). Five amino acid changes were identified in all escape viruses at passages 12 and 24 (Fig. 3). They were attributable to the selection pressure by the compounds and were not the consequences of in vitro passage of infected cells, since these changes could be identified only for the escape viruses but not for the corresponding control viruses (data not shown).

We firstly identified the mutation Y181C within the polymerase domain of RT. Mutations responsible for NNRTI resistance occur in the hydrophobic inhibitor-binding pocket of RT, which includes the residue Y181 (46). A single mutation in this pocket can lead to high-level resistance to earlier NNRTIs, including EFV and NVP. However, two or more mutations are required to cause high-level resistance to ETR, RPV, and other recent NNRTIs (3, 35, 53, 61). Y181C alone can cause multidrug resistance to NNRTIs, such as high-level resistance to NVP and DLV, and low-level resistance to EFV (18). In addition, Y181C provides the mutational foundation for the development of higher levels of ETR resistance (49). The Y181C mutation has emerged as an initial mutation in previous studies in patients failing an NNRTI-containing regimen, including NVP monotherapy (15, 43) and DLV monotherapy (17). K103N is frequently observed among NNRTI resistance mutations, but it is usually followed by Y181C and G190A (27). Resistance to HEPT derivatives is generated by mutations at multiple sites in the HIV-1 RT (8), while 1-(3-cyclopenten-1-yl)methyl-6-(3,5-dimethylbenzoyl)-5-ethyl-2,4-pyrimidinedione (SJ-3366) selected for a virus with the mutation Y181C after five tissue culture passages (9). In the present study, the initial mutation Y181C was sufficient for the acquisition of HIV-1 resistance to AzBBU and AmBBU. Other common NNRTI resistance mutations, such as K101E, K103N, and Y188C, were not identified.

Using docking studies, we examined the binding sites for the metabolically relevant derivative AmBBU on the allosteric pocket of HIV-1 RT (Fig. 4). According to the docking, AmBBU binds the allosteric pocket through arene-H interactions with the amino acid residues Leu100, Val106, Tyr181, and Trp229 (Fig. 4A). When the RT mutant Y181C was used for docking, AmBBU lost the interaction at position 181 (Fig. 4B). The docking score between HIV-1 RT (wt) and AmBBU was higher than that between HIV-1 RT (Y181C) and AmBBU, indicating the importance of the residue Tyr181 in the binding to the allosteric pocket. Most NNRTIs are engaged in the H-bond with the backbone of the residues Lys101 and/or Lys103 of RT (13, 26, 41). Previous docking studies suggested an H-bond between the amide of Lys101 and nitrogen of the cyanomethyl and picolyl group of 1-substituted 3-(3,5-dimethylbenzyl)uracils (33). In our previous study, we showed that 6-subtitutions on the uracil ring resulted in elevation of the anti-HIV-1 activity of the uracil derivatives (28). The structure-activity relationship among these uracil derivatives suggested that the strong anti-HIV-1 activity of the 6-amino derivative AmBBU is due to the H-bond formed between the 6-amino group of AmBBU and the amide group of the residue Lys101, as well as hydrophobic interactions (28). Here, our docking showed arene-H interactions of AmBBU with Leu100, Val106, Tyr181, and Trp229 (Fig. 4A). However, we could not identify the precise role of the 6-amino substitution in the binding to the allosteric pocket and how it leads to an increased anti-HIV-1 activity. Thus, although several interactions occur within the allosteric pocket, the interaction with the Tyr181 residue appears essential for docking of 6-substituted 1-benzyl-3-(3,5-dimethylbenzyl)uracils. In addition, Y181C loses important aromatic ring interactions in the core of the NNRTI-binding pocket, decreasing binding of NNRTIs (42). Taken together, the results of sequence analysis and docking studies indicate that AmBBU requires the interaction(s) with Tyr181 for its inhibitory effect on HIV-1.

In addition to Y181C, we identified four mutations in the RNase H domain of RT: K451R, R461K, T468P, and D471N (Fig. 3). Since we do not know exactly when these mutations emerged in relation to Y181C, it is difficult to elucidate their precise role in acquisition of HIV-1 resistance to AzBBU and AmBBU. One possibility is that mutations in the RNase H domain are merely the result of polymorphisms of RT and are not related to Y181C, as indicated by several HIV-1 (subtype B) prototype strains containing some of the identified mutations in the RNase H domain (Table 3). The strains MCK1_LAI_III_B and PV22_LAI_III_B, for instance, display all four amino acid changes but are not associated with any common NNRTI-resistant mutation. However, some strains display one or more NNRTI-resistant mutations in association with the mutation T468P/S, suggesting that mutations in the RNase H domain could act in coordination with NNRTI-resistant mutations to favor drug resistance. It has been postulated that drug resistance mutations reduce the replication fitness of HIV-1 (10). NNRTI-resistant RTs with the Y181C mutation have been shown to alter the rate of one or both modes of RNase H cleavage with no significant effects on RNA- or DNA-dependent DNA polymerization, and a decrease in RNase H activity has been associated with greater reductions in replication fitness (2).

Table 3.

NNRTI-resistant mutations in the polymerase domain and amino acid changes in the RNase H domain of escape virus

GenBank accession no.	Name	RT domaina
		Polymerase					RNase H
		A98	K102	K103	V179	Y181	K451	R461	T468	D471
A04321	IIIB LAI	–	–	–	–	–	–	–	P	–
A07867	LAI-J19	–	–	–	–	–	–	–	–	–
AB221005	Ba-L	Q	–	–	–	–	–	–	S	–
AF033819	HXB2_copy LAI	–	–	–	–	–	–	–	–	–
AF070521	NL43E9 LAI IIIB_NY5	–	Q	–	–	–	–	–	P	–
D86068	MCK1 LAI IIIB	–	–	–	–	–	R	K	P	N
D86069	PM213 LAI IIIB	–	–	–	–	–	–	–	–	–
EU541617	pIIIB	–	Q	–	–	–	–	–	P	–
HB388803	MN patent seq	–	–	–	–	–	–	–	S	–
K02007	SF2 LAV2 ARV2	–	–	–	–	–	–	–	S	–
K02013	LAI BRU	–	–	–	–	–	–	–	–	–
K02083	PV22 LAI IIIB	–	–	–	–	–	R	K	P	N
K03455	HXB2-LAI-IIIB-BRU	–	–	–	–	–	–	–	–	–
M17449	MNCG MN	–	–	–	–	–	–	–	S	–
NC_001802	HXB2-LAI-HXB2R	–	–	–	–	–	–	–	–	–
U26942	NL4_3 LAI_NY5 pNL43 NL43	–	Q	–	–	–	–	–	P	–
U63632	JRFL JR_FL	–	–	R	I	–	–	–	S	–
X01762	REHTLV3 LAI IIIB	–	–	–	–	–	–	K	P	N
JQ070415	AzBBU IIIB-AZ	–	–	–	–	C	R	K	P	N
JQ070416	AmBBU IIIB-AM	–	–	–	–	C	R	K	P	N

^aThe default amino acids and position numbers are specified in each column subheading. –, No change.

In 2005, Nikolenko et al. suggested that mutations in the RNase H domain could significantly contribute to an increase of RT resistance to NRTIs (37). Later, the same group proposed the RNase H-dependent NNRTI-resistant model, which suggests that mutations in the RNase H domain that reduce RNase H cleavage will allow more time for the NNRTIs to dissociate from the NNRTI-RT-template/primer complex, allowing the reinitiation of polymerization and thereby resulting in enhanced NNRTI resistance (36). Thus, combining mutations in RT that reduce NNRTI affinity with mutations that reduce RNase H cleavage should further increase NNRTI resistance (24, 36). In an RNase H-independent mechanism, NNRTIs themselves can increase RNase H activity, so that mutations reducing RNase H activity are selected in response to NNRTI therapy, because they restore the balance between RNase H activity and polymerization (16). An alternative explanation is that NNRTIs may inhibit HIV-1 replication by increasing RT dimer stability (51). Thus, NNRTI-binding pocket mutants that confer drug resistance should decrease the stability of RT heterodimers. In general, the nucleic acid structure-dependent interplay between polymerase and RNase H domains is likely to affect overall efficacy of NNRTIs against HIV-1 replication, as well as the selection of mutations in the NNRTI-binding site associated with NNRTI resistance.

The effects of mutations in the RNase H domain on NNRTI resistance have been confirmed in vitro, and yet their clinical impact is still unclear. Current HIV-1 genotypic analyses of patients generally focus on the N terminus of the polymerase domain, thus missing important information on mutations in the thumb-connection subdomains and RNase H domain that might be related to resistance either alone or in combination with other RT mutations. Yap et al. showed that N348I in the connection subdomain was highly prevalent in a patient cohort and was highly associated with thymidine analogue-associated mutations (TAMs) and the NNRTI mutations K103N and Y181C (59). Hachiya et al. also examined N348I in treatment-experienced clinical isolates from Japan and found that N348I was prevalent in AZT and/or ddI therapy and that several mutations in the connection subdomain and RNase H domain typically acted as pretherapy polymorphisms (20, 21). Waters et al. found N348I prevalent in treatment-experienced patients (54). In addition, these researchers found that the genotypic profiles of patients with or without the K451R mutation within a treatment-experienced group showed a higher incidence of NNRTI mutations in patients with the K451R mutation. Santos et al. analyzed 450 sequences from Brazilian subtype B isolates and public databases and found nine mutations in the connection subdomain and six mutations in the RNase H domain that were associated with NRTI therapy (45). Positions K451 and D471 were less conserved in NRTI-experienced patients, while R461 and T468 were equally variable in both naive and experienced patients. A comparison of RNase H sequences in naive versus NRTI-experienced patients in a French cohort showed that mutations L469T/I/M/H, T470P/S/E/K, A554T/L/K, and K558R/G/E were more prevalent among treatment-experienced patients (44). However, Ntemgwa et al. analyzed RNase H mutations in NRTI-experienced patients from a Canadian and an Italian cohort and found positions D460, P468, H483, K512, and S519 to be extensively polymorphic in both naive and experienced patients (38). Recently, an analysis of patient sequences from databases showed that several mutations in the connection subdomain were significantly higher for sequences that contained one or more RTI resistance mutations compared to sequences without RTI resistance mutations (16). Moreover, subtype B-infected patient database analysis showed that RNase H mutations, including K451R, increased in frequency with the number of TAMs in a dose-dependent fashion (31). That study demonstrated that distinct RT C-terminal mutations can act as primary or secondary drug resistance mutations and are associated in a complex array of phenotypes with RT polymerase domain mutations.

In this in vitro study, we identified four mutations in the RNase H domain that might be related to the NNRTI resistance mutation Y181C. Biochemical studies are needed to understand the molecular mechanism of the associations and interactions between mutations within the polymerase and RNase H domains of RT. Further experiments are under consideration to validate the role of these mutations in the acquisition of resistance to AzBBU and AmBBU. First, viral strains containing the identified mutations in the RNase H domain, such as MCK1_LAI_IIIB and PV22_LAI_IIIB, will be used for testing the uracil derivatives, since these strains have all four amino acid changes but are not associated with any common NNRTI-resistant mutation. Second, recombinant RT enzymes containing the identified mutations will be used to determine the inhibitory effects of AzBBU and AmBBU on their catalytic activity. Third, an experiment on site-directed mutagenesis will be performed to elucidate the precise role of the RNase H mutations in the acquisition of resistance.

In terms of drug development, we found useful information for the future design of 6-substituted uracil derivatives with enhanced chemical properties that improve anti-HIV-1 activity and resistance profiles. Although the chemical properties of AzBBU and AmBBU suggest a good drug-likeness profile, further studies are required to assess the toxicity and pharmacokinetics of 6-substituted uracil derivatives in vivo. A limitation of our study relates to the use of laboratory strains. However, the mutations that we identified have been reported in clinical isolates (38, 45, 54). Another limitation of our study relates to the lack of information on the mutations in the RNase H domain of NNRTI-treated patients, since these mutations have been reported mostly in NRTI-treated patients. Furthermore, in a clinical setting, NNRTIs must be used in combination therapy, which may alter the pattern for resistance. Thus, the emergence of drug resistance should be further investigated and confirmed in clinical trials.

In conclusion, our results provide important information on the acquisition of resistance to the novel uracil derivatives AzBBU and AmBBU. Although the initial mutation Y181C can be sufficient in the acquisition of HIV-1 resistance, additional mutations in the RNase H domain of RT could additionally be associated to the mechanisms of resistance. Docking studies using the crystal structure of RT showed that AmBBU requires the amino acid residues Leu100, Val106, Tyr181, and Trp229 for its inhibitory effect on HIV-1. Further studies are necessary to determine the precise role of each mutation in the acquisition of HIV-1 resistance to the present compounds.