1,1,3-Trioxo-2H,4H-Thieno[3,4-e][1,2,4]Thiadiazine (TTD) Derivatives: a New Class of Nonnucleoside Human Immunodeficiency Virus Type 1 (HIV-1) Reverse Transcriptase Inhibitors with Anti-HIV-1 Activity

Abstract

We report the development of a new group of nonnucleoside reverse transcriptase inhibitors (NNRTIs). One of the most active congeners of this series of 1,1,3-trioxo-2H,4H-thieno[3,4-e][1,2,4]thiadiazine (TTD) derivatives, i.e., 2-(3-fluorobenzyl)-4-cyanomethylen-1,1,3-trioxo-2H,4H-thieno[3,4-e][1,2,4]thiadiazine) (QM96639) was found to inhibit human immunodeficiency virus (HIV) type 1 [HIV-1 (III_B)] replication in MT-4 cells at a concentration of 0.09 μM. This compound was toxic for the host cells only at a 1,400-fold higher concentration. The TTD derivatives proved effective against a variety of HIV-1 strains, including those that are resistant to 3′-azido-3′-deoxythymidine (AZT), but not against HIV-2 (ROD) or simian immunodeficiency virus (SIV/MAC251). HIV-1 strains containing the L100I, K103N, V106A, E138K, Y181C, or Y188H mutations in their reverse transcriptase (RT) displayed reduced sensitivity to the compounds. Their cross-resistance patterns correlated with that of nevirapine. 2-Benzyl-4-cyanomethylen-1,1,3-trioxo-2H,4H-thieno[3,4-e][1,2,4]thiadiazine (QM96521) enhanced the anti-HIV-1 activity of AZT and didanosine in a subsynergistic manner. HIV-1-resistant virus containing the V179D mutation in the RT was selected after approximately six passages of HIV-1 (III_B) in CEM cells in the presence of different concentrations of QM96521. From structure-activity relationship analysis of a wide variety of TTD derivatives, a number of restrictions appeared as to the chemical modifications that were compatible with anti-HIV activity. Modelling studies suggest that in contrast to most other NNRTIs, but akin to nevirapine, QM96521 does not act as a hydrogen bond donor in the RT-drug complex.

The first compounds shown specifically to inhibit human immunodeficiency virus type 1 (HIV-1) (but not HIV-2) replication were 1-(2-hydroxyethoxymethyl)-6-(phenylthio)thymine (HEPT) (3, 34) and tetrahydroimidazo [4,5,1-jk][1,4] benzodiazepin-2(1H)-one and -thione (TIBO) (16, 39). The specificity of the TIBO derivatives was attributed to a specific interaction with the HIV-1 reverse transcriptase (RT) (16, 39). Follow-up studies of HEPT congeners revealed HIV-1 RT as the principal target for their antiviral action as well (4, 5). Subsequently to TIBO and HEPT, several other classes of specific HIV-1 RT inhibitors were discovered, including, for example, nevirapine (BI-RG-587) (33, 52), pyridinone derivatives (L-696,229 and L-697,661) (27, 28), bis(heteroaryl)piperazine (BHAP) (44, 45), 2′,5′-bis-O-(tert-butyldimethylsilyl)-3′-spiro - 5" - (4" - amino - 1",2", - oxathiole - 2",2" - dioxide)pyrimidine (TSAO) derivatives (7, 8), α-anilinophenylacetamides (α-APA) (40), phenylethylthioureathiazole (PETT) (1), oxathiin carboxanilides (6), and quinoxaline derivatives (30). All of these compounds have been commonly referred to as nonnucleoside RT inhibitors (NNRTIs). Since then, at least 25 classes of NNRTIs have been reported (for an overview, see references 17 and 18). The most potent and selective congeners among the NNRTIs, i.e., α-APA R89439 (loviride), thiocarboxanilide UC781, and quinoxaline HBY 097, inhibit HIV-1-induced cytopathicity at nanomolar concentrations with selectivity indices of as much as 100,000.

Here, we report a new class of NNRTIs, the thieno[3,4-e][1,2,4]thiadiazines (TTDs). We examined their activities against different HIV-1 strains, including strains that were resistant to 3′-azido-3′-deoxythymidine (AZT) and certain NNRTIs. One of the most active congeners, QM96521, was further investigated. Its anti-HIV-1 activity in combination with the nucleoside analogs AZT and didanosine (ddI) was determined. QM96521-resistant virus was selected following in vitro passage of HIV-1 in the presence of QM96521. A model for the complex between HIV-1 RT and QM96521 was constructed based on the RT-nevirapine structure (43).

MATERIALS AND METHODS

Compounds.

The detailed synthesis of the test compounds will be reported elsewhere. Nevirapine (BI-RG-587; Viramune) was kindly provided by P. Ganong (Boehringer Ingelheim, Ridgefield, Conn.).

Cells.

MT-4 (35) and CEM (25) cells were grown and maintained in RPMI 1640 medium, supplemented with 10% heat-inactivated fetal calf serum, 2 mM l-glutamine, 0.1% sodium bicarbonate, and 20 μg of gentamicin per ml. Peripheral blood mononuclear cells (PBMCs) were isolated from HIV-seronegative donor buffy coats by using Lymphoprep (Nycomed, Oslo, Norway), stimulated for 3 days in phytohemagglutinin (2 μg/ml; Sigma, Bornem, Belgium)- and interleukin-2 (10 U/ml; Boehringer, Mannheim, Germany)-containing medium, washed, and resuspended in RPMI 1640 supplemented with 2 mM l-glutamine, gentamicin (50 μg/ml), 15% heat-inactivated fetal calf serum, and recombinant human interleukin-2 (10 U/ml; Boehringer).

Viruses.

The origin of the virus stocks was as described previously, i.e., HIV-1 (III_B and RF) (42), and HIV-1 (NDK) (48). HIV-1 (MN) and HIV-1 (ADP/141), a recombinant AZT-resistant HIV-1 strain (RT mutations D67N, K70R, T215F, and K219Q), were obtained through the Medical Research Council’s AIDS Reagent Project, National Institute for Biological Standards and Control, London, United Kingdom, and were contributed by R. C. Gallo and M. Popovic (26) and B. Larder and S. Kemp (32), respectively. HIV-1 (HE) represents a clinical isolate from a Belgian patient with AIDS [for HIV-1 (HE), virus stocks were prepared from the supernatants of MT-4 cells which had been infected with the supernatants of the fifth passage of cocultures of the patient’s PBMCs with cord blood lymphocytes (at a ratio of 1:1)]. HIV-1 and HIV-2 (ROD) (14) stocks were prepared from the culture supernatants of HIV-1- or HIV-2-infected cell lines (37, 47). Simian immunodeficiency virus (SIV) strain MAC251 was originally isolated by Daniel et al. (15) and was obtained from C. Bruck (SmithKline-RIT, Rixensart, Belgium); SIV stocks were prepared from the supernatants of SIV-infected MT-4 cells.

Antiviral activity assays.

The inhibitory effects of the test compounds on HIV-1, HIV-2, and SIV replication were monitored by the inhibition of virus-induced cytopathicity and syncytium formation in MT-4 and CEM cells as described by Pauwels et al. (38) and Balzarini et al. (10), respectively. Virus-induced cytopathicity was recorded at 5 days postinfection for HIV and SIV.

To determine the cross-resistance of HIV to different NNRTIs, CEM cells were suspended at 250,000 cells per ml of culture medium and were infected with 100 50% cell culture infective doses (CCID₅₀) of HIV-1 (III_B) or mutant HIV-1 strains selected for resistance to TIBO R82150 (HIV-1 [L100I]) (12), TIBO R82913 (HIV-1 [K103N]) (12), TSAO-m³T (HIV-1 [E138K]) (13), nevirapine (HIV-1 [V106A]) (11), the pyridinone L697,661 (HIV-1 [Y181C]) (12), or HEPT (HIV-1 [Y188H]) (9). Then, 100 μl of the infected cell suspensions was added to 200-μl microtiter plate wells containing 100 μl of an appropriate dilution of the test compounds (i.e., 100, 20, 4, 0.8, 0.16, 0.032, and 0.006 μg/ml). The inhibitory effects of the test compounds on HIV-1-induced syncytium formation in CEM cells were examined on day 4 postinfection, as previously described (11, 12).

PBMCs (10⁶ cells/ml) were plated in the presence of different concentrations of the test compound and infected with HIV-1 (III_B) at 1,000 CCID₅₀ per ml. At 4 days postinfection, half of the supernatant of the infected cultures was removed and replaced with fresh medium containing the test compound at the appropriate concentration. At 7 days after plating of the cells, p24 antigen was detected in the culture supernatant by an enzyme-linked immunosorbent assay (DuPont, Dreieich, Germany) (51).

RT assay.

Poly(rC) · oligo(dG) and [³H]dGTP and poly(rA) · oligo(dT) and [³H]dTTP were used as the template-primer and radiolabeled substrate, respectively. The final [³H]dGTP and [³H]dTTP concentrations in the reaction mixture were both 2.5 μM.

Analysis of combination effect.

Checkerboard combinations of various concentrations of the test compounds were examined for their combined inhibitory effect on the HIV-1-induced cytopathic effect (CPE). When the combination consisted of two effective compounds, the combined effect was analyzed by the isobologram method as described previously (2). In this analysis, the 50% effective concentration (EC₅₀), which is the concentration of compound required to protect 50% of MT-4 cells against an HIV-1-induced CPE, was used to calculate the fractional inhibitory concentration (FIC). When the minimum FIC index, which corresponds to the FICs of compounds combined (e.g., FIC_x + FIC_y), is equal to 1.0, the combination is additive; when the FIC index is between 1.0 and 0.5, the combination is subsynergistic; and when the FIC index is <0.5, the combination is synergistic. On the other hand, when the minimum FIC index is between 1.0 and 2.0, the combination is subantagonistic, and when it is >2.0, the combination is antagonistic (20).

Selection of QM96521-resistant virus strains.

HIV-1(III_B) was subjected to six passages in CEM cell cultures (4 × 10⁵ cells/ml) in the presence of 15 or 300 μM QM96521. Passages were performed every 3 to 4 days by adding 0.5 ml of the infected cultures to 4.5 ml of 4 × 10⁵ uninfected CEM cells/ml. As soon as virus-induced syncytium formation became fully prominent in the cell culture, supernatant was frozen in aliquots at −70°C and the reverse transcriptase gene of the virus was characterized.

Determination of the amino acid sequence of the QM96521-resistant HIV-1 RT strain.

The procedures for CEM cell infection with QM96521-resistant HIV-1, preparation of the samples for the PCR assays, amplification of the proviral DNA, and sequencing of the 727-bp fragment covering amino acid residues 50 to 270 have been reported elsewhere (9, 12).

Modelling of RT-QM96521 complex structure.

NNRTIs bind to RT in a largely hydrophobic pocket whose shape varies only slightly for different NNRTIs. Thus, complementarity with the pocket shape, rather than with the pocket charge, is the crucial factor in determining the binding mode of many NNRTIs. The structure of QM96521, which was determined by X-ray crystallography, was compared with those of NNRTIs for which the structure of the RT-NNRTI complex was known, and this suggested that the RT-nevirapine complex structure was the most appropriate starting model (43). Based on this structure (Protein Data Bank code 1VRT), a model of the RT-QM96521 complex was built with MIDASPLUS (24), which optimized the volume overlap between the two NNRTIs. Parameter and topology files for AMBER (41) were created to complement the PARM94 parameter set for the protein by using the partial charges assigned by the AMBER* force field in MACROMODEL (36). With AMBER, a restrained energy minimization and molecular dynamics simulation protocol for the RT-QM96521 complex was followed (23). These simulations identified several possible conformations for the inhibitor in the complex, ut the model in which the protein structure was least disturbed from the RT-nevirapine structure was chosen as the preferred model. Figures for the RT-QM96521 complex were produced with BOBSCRIPT (21, 31).

RESULTS

Structure-activity relationship (SAR) for TTDs.

A huge series of TTDs was evaluated for their abilities to inhibit the CPEs of HIV-1 (III_B) in human T4 lymphocyte cells, i.e., MT-4 cells. (Tables 1 and 2 and data not shown). The main structure of this series is given in Table 1. Congeners of the series consisted of 2- and/or 4-substituted TTDs, and from these it was possible to identify the features necessary for anti-HIV-1 activity. The complete structure-activity relationship will be described elsewhere.

TABLE 1.

Structural formulae of TTDs

Compound	X	R4
QM96171	H	CH3
QM96215	H	CH2CH3
QM96440	3-Cl	CH2CH3
QM96517	2-Cl	CH2CH3
QM96519	H	Benzyl
QM96521	H	CH2CN
QM96537	H	CH2C CH
QM96639	3-F	CH2CN

TABLE 2.

Inhibitory effects of TTD derivatives on the replication of different (mutant) strains of HIV-1 in cell culture

Compound	EC50 (μM)a for the following strains (cell lines):													CC50 (μM)b (MT-4)
	IIIB (MT-4)	IIIB (CEM)	RF (MT-4)	MN (MT-4)	NDK (MT-4)	HE (MT-4)	ADP/141c (MT-4)	L100I (CEM)	K103N (CEM)	V106A (CEM)	E138K (CEM)	Y181C (CEM)	Y188H (CEM)
QM96171	7.1	2.3	32.1	65.8	3.9	106.0	4.9	13.0	>324	>324	≥324	>324	>324	690
QM96215	8.7	2.8	28.2		15.5	20.5	3.1	10.2	>310.2	>310.2	310.2	>310.2	>310.2	512
QM96440	2.5	1.1	3.6	0.8	4.8	3.1	0.6	3.6	>56	33.6	9.2	>56	≥11.2	498
QM96517	8.7		10.9			11.5	5.0							540
QM96519	2.6	1.8	5.5	2.3	2.1	8.3	1.8	≥10.4	>10.4	≥10.4	≥10.4	>10.4	7.8	510
QM96521	0.9	1.2	2.4	1.5	6.0	4.2	1.2	7.8	>300	156	7.2	>300	≥60	504
QM96537	0.2		0.6	0.3										157
QM96639	0.09	0.6	0.3	0.2	0.6			2.9	28.5		2.9	28.5		129
Nevirapine	0.03	0.02	0.04	0.02	0.08	0.02	0.03	0.3	5.6	3.0	0.4	25.5		683

^aConcentrations of each compound required to inhibit the CPE of different HIV-1 strains in cell culture by 50%. 

^bCC₅₀, 50% cytotoxic concentration, or concentration of the compound required to reduce MT-4 cell viability by 50%. 

^cADP/141, an AZT-resistant HIV-1 strain (mutations D67N, K70R, T215F, and K219Q). 

Several compounds bearing a benzyl group at the 2-position were synthesized. None of the derivatives without a 4-position substituent displayed any anti-HIV activity. Therefore, further alkylation was performed at this position. Methyl (QM96171), ethyl (QM96215), and benzyl (QM96519) groups were introduced. These N-4-alkylated compounds displayed similar anti-HIV-1 activities. By the introduction of a 4-cyanomethylene substituent (QM96521), the activity increased and a selectivity index of 560 was obtained. As an isostere for the cyanomethylene group, it was replaced by a propynyl group (QM96537). This retained anti-HIV-1 activity, indicating that the polarization of the rod-like terminal part of the cyanomethylene group is not necessary for activity. TTD derivatives without a benzyl group at the 2-position were inactive and/or toxic. A halogen is allowed on the ring of the N₂-benzyl substituent in the ortho (QM96517) or meta (QM96440) positions, but a chlorine atom at the para position leads to a complete loss of the anti-HIV activity. Introduction of two or more halogens also leads to a complete loss of anti-HIV activity. When a fluorine atom was introduced at the meta position of the N₂-benzyl substituent together with a cyanomethylene substituent at position 4 (QM96639), the selectivity index increased to 1,400.

In vitro anti-HIV activity.

Selected compounds were tested against a variety of HIV-1 strains (III_B, RF, MN, and NDK), a clinical HIV-1 isolate (HE), an AZT-resistant HIV-1 strain (ADP/141), HIV-2 (ROD), and SIV (MAC251) in MT-4 cell cultures. All compounds showed comparable activities against all HIV-1 strains tested (Table 2), with QM96639 being the most active congener with a 50% effective concentration (EC₅₀) of 0.09 μM against HIV-1 (III_B) in MT-4 cells. This inhibitory effect was comparable to the anti-HIV activity of nevirapine (EC₅₀, 0.03 μM). The EC₅₀ of QM96521 against the replication of HIV-1 (III_B) in PBMC was 0.3 μM. None of the compounds were inhibitory against HIV-2 (ROD) or SIV (MAC251) at subtoxic concentrations (data not shown).

Anti-HIV-1 RT activity.

QM96521 and QM96639 were also evaluated for their inhibitory effects against HIV-1 RT. QM96521 and QM96639 inhibited HIV-1 RT activity by 50% at 26.0 and 15.1 μM, respectively (data not shown). QM96521 and QM96639 inhibited HIV-1 RT activity at approximately 30- to 150-fold higher concentrations than the concentrations (0.9 and 0.09 μM, respectively) required to inhibit HIV-1 (III_B) replication in MT-4 cells, respectively.

Combined inhibitory effect of QM96521 with AZT or ddI.

When QM96521 was combined with the nucleoside RTI AZT or ddI and the inhibitory effect on HIV-1-induced CPE was evaluated by the isobologram method, all FIC_QM96521 + FIC_NRTI values were between 0.5 and 1.0 (data not shown), defining these combinations as subsynergistic.

Inhibitory activity of TTDs against NNRTI-resistant mutant HIV-1 strains.

The compounds were evaluated for their inhibitory effects on a variety of mutant HIV-1 strains (Table 2). Whereas the TTDs lost only 3- to 10-fold activity against the L100I RT mutant virus, they showed at least a 50-fold reduction in their inhibitory potency against the K103N and Y181C RT mutant viruses. QM96171, QM96215, and QM96521 showed at least 50-fold reduction in potency against the V106A and Y188H RT mutant virus strains, whereas QM96519 and QM96440 were only 10-fold less active. All of the TTDs tested showed a moderate (4- to 40-fold) cross-resistance to the E138K RT mutant virus. The cross-resistance pattern of the TTDs correlated well with that of nevirapine.

Selection of an QM96521-resistant mutant HIV-1 strain in CEM cell cultures.

When HIV-1-infected CEM cell cultures were exposed to different concentrations of QM96521, QM96521-resistant HIV-1 strains emerged after approximately six passages. The mutant HIV-1 strains showed a single amino acid change of Val→Asp at position 179 (V179D) of their RT. These mutant HIV-1 strains proved 15- to more than 140-fold resistant to the TTD derivatives QM96171, QM96215, QM96440 and QM96519, with the exception of QM96521, which (in common with nevirapine) lost only 3- to 10-fold activity (Table 3). Of the other NNRTIs tested, only 8-chloro-TIBO (R86183) showed a 10-fold reduction in potency, whereas α-APA (R89439), BHAP U-90152, and UC-781 (together with the nucleoside analogs 3TC and AZT) retained full activity against the V179D mutant strains.

TABLE 3.

Inhibitory effects of TTD derivatives, other NNRTIs, and NRTIs against QM96521-selected virus strains

Compound	EC50 (μM)a
	WTb	V179Dc	V179Dd
QM96171	2.3	>324	>324
QM96215	2.8	>310	>310
QM96440	1.1	280	56
QM96519	1.8	>10.4	>10.4
QM96521	1.2	5.6	9
Nevirapine	0.02	0.06	0.1
TIBO R86183	0.003	0.03	0.02
α-APA R89439	0.01	0.02	0.02
BHAP U-90152	0.02	0.02	0.03
UC-781	0.006	0.006	0.007
3TC	0.04	0.06	0.04
AZT	0.03	0.07	0.02

^aConcentration required to inhibit the CPE of HIV-1 in CEM cell cultures by 50%. 

^bWT, wild-type HIV-1 (III_B). 

^cV179D mutant virus strain selected in the presence of 15 μM QM96521. 

^dV179D mutant virus strain selected in the presence of 300 μM QM96521. 

Model for the RT-QM96521 complex.

Our model (Fig. 1) appears to be broadly consistent with the activity and resistance mutation data for QM96521 and its congeners (Table 2). However, the variety of NNRTI conformations observed in RT complexes and the lack of hydrogen bonding possibilities between the RT and QM96521 make identification of the correct conformation particularly difficult, and, thus, at this stage, the model should be considered tentative.

FIG. 1.

A model structure for the RT-QM96521 complex. (a) Stereodiagram of QM96521 in the NNRTI-binding pocket of RT. The inhibitor is indicated as an atom-colored ball-and-stick representation (black, C; dark gray, O; mid-gray; N; light-gray, S), the surrounding protein structure is indicated by thin sticks, and water molecules within the pocket are indicated by gray spheres. Residues for which mutations are discussed in the text are labeled. (b and c) Orthogonal views of the relative positions and conformations for QM96521 (in the model structure [black]) and nevirapine (in the X-ray structure 1VRT [43] [gray]) in their respective complexes with RT. The complex structures were superimposed based on the surrounding protein residues in the manner described by Ren et al. (43).

The modelled conformation of QM96521 is similar to the conformation of nevirapine in its complex with RT (43) (Fig. 1). Not only do the positions of the fused rings overlap, but the cyclopropyl group of nevirapine (which has a spatial analog in all RT-NNRTI structures reported) is mimicked by the N₄-cyanomethyl group of QM96521. In line with this model, it is the size rather than the chemical nature of the 4-position substituent that is important for anti-RT activity (e.g., QM96521 and QM96537). The third ring of the inhibitors is not fused with the other two (as with nevirapine) but is attached by a methylene linkage. In our model, this ring is orientated to point toward the proposed entrance of the NNRTI binding pocket (22) in a conformation reminiscent of the HEPT analog TNK-651 in the RT–TNK-651 complex (29). This conformation does not seem to fit comfortably with the shape of the pocket, however, and either removing this ring or adding an extra methylene unit to the linkage (to allow increased conformational freedom) may improve binding of the inhibitor.

The nitrogen atom of the N₄-cyanomethyl group is positioned near to the side chain of V179. The resistance mutation V179D would bring two partial negative charges close together and, thus, would be expected to confer resistance (Table 3). Data for cross-resistance to other NNRTI-induced mutations also supports our model (Table 2), which was based on shape similarity with nevirapine (Fig. 1). As with nevirapine, almost complete resistance to QM96521 is afforded by the K103N, Y181C, and Y188H mutations, but only partial resistance is conferred by the L100I mutation. The N₂-benzyl group of QM96521 bends around the side chain of V106, forming several close contacts; thus, it is not surprising that the V106A mutation (Table 2) also confers resistance. The mechanism of action of the E138K mutation (Table 2) is not clear from our model. However, in many RT-NNRTI complex structures, there is a water molecule in the NNRTI binding pocket which forms hydrogen bonds to the side chain of E138 and other residues. Disruption of this hydrogen bonding pattern and water structure may give rise to resistance by an indirect route.

DISCUSSION

TTDs represent a promising new class of NNRTIs. The most active congener yet discovered, 2-(3-fluorobenzyl)-4-cyanomethylen-1,1,3-trioxo-2H,4H-thieno[3,4-e][1,2,4]thiadiazine(QM96639), proved inhibitory to HIV-1 (III_B) replication at 0.09 μM, without being toxic for the host cells at concentrations of as much as 129 μM (a selectivity index of 1,400). By analyzing the anti-HIV-1 results of a large series of TTDs, we could identify the structural requirement for anti-HIV activity, i.e., a benzyl group at position 2 that might be substituted with a halogen atom at the ortho or meta position and a rather small hydrophobic component at position 4.

Based on structure-activity data together with molecular modelling, we are now attempting to synthesize second-generation TTDs in which the benzyl group at position 2 will be replaced in order to obtain a higher potency and/or selectivity index against HIV-1. The cross-resistance pattern of the TTDs against other NNRTI-resistant mutant HIV-1 strains correlated well with that of nevirapine. This observation is consistent with the modelling data, since the modelled conformation of QM96521 is similar to the conformation of nevirapine in its complex with RT.

Mutant HIV-1 strains have been selected in cell culture in the presence of different concentrations of QM96521 within six subcultivations, which is similar in speed to the emergence of mutant HIV-1 strains resistant to NNRTIs such as nevirapine, TIBO (R82150), pyridinone, BHAP, and TSAO (11, 12). The mutant strains contain the V179D mutation, which has previously been observed to occur in response to the NNRTIs DMP-266, pyridinone (L-697,661), TIBO (R82913), trovirdine, and UC-10 (46). In the modelled conformation of QM96521, we observed that the resistance mutation V179D would bring two partial negative charges close together and thus would be expected to confer resistance. Surprisingly, the V179D mutant viruses and especially the mutant virus strain selected in the presence of 300 μM QM96521 lost only 3- to 10-fold of their sensitivity to QM96521 and nevirapine, while they were highly resistant to the other TTD derivatives evaluated (Table 3). The reason for the anomalous behavior of QM96521 and nevirapine against these mutant virus strains is unclear. Also, mutations at positions 106 and 181 of the RT were found in virus strains that had been passaged in the presence of other TTD derivatives (data not shown). Based on our model, substitutions at position 7 of the TTD may induce a mutation at position 229 of the RT in order to escape to the pressure of the substituent at this site. The selection of this mutation would be interesting to study, since the occurrence of some mutations may lead to suppression of other mutations or to phenotypical conversion of resistance to sensitivity. For instance, when the 3TC-specific M184V mutation occurs in combination with the Y181C mutation, AZT resistance is completely reverted to AZT sensitivity (50). Moreover, the BHAP resistance mutation P236L increases the sensitivity of HIV-1 RT to TIBO, nevirapine, and pyridinone, even if the HIV-1 RT has been mutated at position 181 (Tyr→Cys) (19). Also, the ddI resistance mutation L74V has been reported to suppress AZT resistance (49). A rational approach toward drug combination may be based on the choice of drugs that lead to mutually antagonistic drug resistance mutations (17).

An important benefit of this new class of NNRTIs is the great possibility for modifications of these chemical structures. Due to the large number of TTD representatives, we had the opportunity to study the SAR of these molecules thoroughly. From these results, we were able to extract the features necessary for their anti-HIV activity. Since TTD derivatives show an advantageous biological profile that can even be improved, we will continue the search for more anti-HIV active congeners.