Amino acid substitutions in HIV-1 reverse transcriptase with corresponding residues from HIV-2. Effect on kinetic constants and inhibition by non-nucleoside analogs.

Nevirapine is a highly potent and specific inhibitor of human immunodeficiency virus type 1 (HIV-1) polymerase, but is inactive against HIV-2 and other polymerase. Previous studies demonstrated that residues 176-190 of HIV-1 reverse transcriptase (RT) can confer nevirapine sensitivity to HIV-2 RT. To better characterize the role of this sequence in HIV-1 RT, we have progressively substituted residues 176-190 of HIV-2 RT for those of HIV-1 RT and monitored the impact on the kinetic properties; inhibitory activity of nevirapine (11-cyclopropyl-5,11-dihydro-4-methyl-6H-dipyrido[2,3-b:2',3'-e] [1,4]diazepin-6-one), E-BPU (5-ethyl-1-benzyloxymethyl-6-(phenylthio)-uracil), and TIBO-R82150 ((+)-S-4,5,6,7-tetrahydro-5-methyl-6-(3-methyl-2-butenyl)imidazo[4,5,1-j k] [1,4]benzodiazepin-2(1H)-thione); and inhibitor-induced fluorescence changes of the mutant enzymes. The study revealed that in addition to Try-181 and Tyr-188, a new amino acid residue (Gly-190) plays an important role in determining susceptibility to nevirapine and E-BPU, but not to TIBO-R82150. These data argue that these non-nucleoside inhibitors fit differently, even though they share a common binding pocket. Nevirapine was seen to exert inhibitory activity by altering the interaction of the enzyme with the template-primer. Kinetic parameters were modulated by the template (DNA versus RNA) as well as by some of the mutations.

I To whom correspondence should be addressed Boehringer Ingelheim Pharmaceuticals, Inc., Dept. of Molecular Biology, R & D, 900 Ridgebury Rd., P. 0. Box 368, Ridgefield, CT  genomic template (see review by Goff (1990)). Human immunodeficiency virus type 1 reverse transcriptase (HIV-1 RT)' is an asymmetric heterodimer with subunits of 51 (p51) and 66 (p66) kDa. p51 results from a proteolytic cleavage that removes the RNase H domain, which is located at the COOH terminus in the p66 polypeptide. HIV-1 RT shares common features with the Klenow fragment of Escherichia coli polymerase I, including a large cleft that accommodates the template and the spatial localization of 3 acidic residues, Asp-110,  in the Klenow fragment, respectively), believed to be part of the polymerase active site (Delarue et al., 1990;Kohlstaedt et al., 1992).
There has been an intense effort to inhibit HIV-1 RT in an attempt to develop chemotherapeutic agents for AIDS treatment (see review by Mitsuya (1992)). Inhibitory compounds identified thus far fall into two categories: nucleoside analogs, of which 3'-azido-3'-deoxythymidine is a member, and nonnucleoside analogs (see review by De Clercq (1992)). Nevirapine is a member of this latter group (Merluzzi et ai., 1990;Hargrave et al., 1991;Klunder et al., 1992) and is being tested in clinical trials. Previous chemical cross-linking and mutational analyses have demonstrated that nevirapine binds near Tyr-181 and Tyr-188 in HIV-1 RT and that mutation of either of these residues causes reduced sensitivity to nevirapine in vitro (Cohen et al., 1991;Shih et al., 1991). More recently, the mutation Tyr-181-Cys of HIV-1 RT has also been reported to cause loss of drug susceptibility in cell culture studies (Richman et al., 1991;Nunberg et al., 1991;Mellors et al., 1992).
In the crystal structure of HIV-1 RT complexed with nevirapine, Tyr-181 and Tyr-188 are seen in proximity to the inhibitor and appear opposite to each other in an antiparallel @-sheet that forms one face of the binding pocket (Fig. 1). Significantly, the presumed catalytic residues, Asp-185 and Asp-186, are situated in the @-turn that connects the 6-strands containing these Tyr residues (Kohlstaedt et al., 1992).
HIV-2 RT is closely related to HIV-1 RT in amino acid sequence (-60% identity), and both enzymes contain several aliphatic residues that flank the 6-hairpin formed by   Table I For Gly-190, the a-hydrogens are indicated. Side chains of Asp-185 and Asp-186, which are critical for polymerase activity, are also indicated. This model is based on the coordinates of the CYcarbons and of nevirapine ( N V P ) that were provided by T. Steitz (Kohlstaedt et al., 1992). Side chains were generated by molecular modeling techniques. For a more complete description of the nevirapine-binding pocket, see Tong et al. (1993). respectively, and, as expected, does not bind nevirapine (see Table I). Even though mutation of Ile-181 and Leu-188 to Tyr in HIV-2 RT does not confer complete sensitivity to nevirapine, a chimeric HIV-2 RT containing amino acids 176-190 of HIV-1 RT recovers most of the susceptibility to nevirapine inhibition (Shih et al., 1991;Condra et al., 1992). This region thus appears to be of considerable interest because of its ability to bind non-nucleoside inhibitors, close proximity to the catalytic center, and sequence homology to other polymerases.
In this study, we progressively substituted residues 176-190 of HIV-2 RT for the corresponding residues from HIV-1 RT.
We monitored susceptibility to nevirapine inhibition and to two other non-nucleoside reverse transcriptase (NNRT) inhibitors and the effects of these mutations on the kinetic parameters under steady-state conditions. The mechanism of inhibition by nevirapine was also studied by inhibitor-induced fluorescence changes.

Site-directed Mutagenesis of HIV-I RT-To obtain mutant clones
of HIV-1 RT, we used commercial protocols (Bio-Rad) and in-house synthesized oligonucleotides following procedures described previously (Shih et al., 1991;Kunkel et at., 1987). Recombinant clones were grown in E. coli strain JM109.
Fluorescence Spectroscopy and Calculation of Dissociation Consfants-The interaction of nevirapine with the enzymes was followed by measuring the fluorescence change of ethidium bromide-poly(rA)oligo(dT)12-ls-enzyme complexes upon addition of nevirapine. Solutions containing 50 nM enzyme, 50 nM poly(rA).oligo(dT)12-1R, and 3 pM ethidium bromide in 50 mM Tris-HCI, pH 7.8, 100 mM KCI, 2 mM MgC12, 0.02% CHAPS were prepared for fluorescence measurements. CHAPS was added to prevent the enzyme from adhering to the sides of the fluorescence cuvette during the course of the experiment and had no adverse effects on enzyme activity. Intercalated ethidium bromide was a t a concentration below saturation. A nevirapine titration solution was prepared from a concentrated stock in dimethyl sulfoxide at 100 p~ in the ethidium bromide-poly(rA). oligo(dT)12.1R-enzyme complex to avoid dilution corrections to the fluorescence intensity measurements. Aliquots of this solution were titrated into ethidium bromide-poly(rA) .oligo(dT)12-1R-enzyme samples, and fluorescence intensity was measured in the slow time base mode after 2 min of stirring. In all cases, the final concentration of dimethyl sulfoxide was <I%. Fluorescence measurements were made on an SLM-8000C photon-counting spectrofluorometer equipped with a magnetically stirred and temperature-controlled cell. All measurements were made a t 23 "C in 1-cm fluorescence cuvettes. Ethidium bromide fluorescence intensity was measured a t 595 nm using an excitation wavelength of 360 nm to optimize detection of bound over free fluorophore ( L e Pecq, 1971). Excitation and emission bandwidths were set at 4 nm.
For the calculation of dissociation constants, the fluorescence intensity a t [nevirapine] = 0 was normalized to 1 for convenience. Assuming a one-binding site model, the data were then fit to Equation 1 by a nonlinear least-squares algorithm: where AF = measured fluorescence change at given total nevirapine concentration, AF,.. = maximum fluorescence change at saturation of binding site, E, = total enzyme concentration, L, = total ligand concentration, and Kd = dissociation constant. Kd and were determined from the fitting procedure.
Student's t Test-Probability values were calculated from Student's two-tailed f test performed against wild-type reverse transcriptase (wtRT-1).

Inhibition by Nevirapine, TIBO-R82150, and E-BPU-
RNA-directed DNA polymerase activity of the purified heterodimeric enzymes (Fig. 2) was assayed in the presence of nevirapine using poly(rC) -oligo(dG)12-la as template-primer ( Table I). Mutants in which Tyr-181 and/or Tyr-188 were substituted with Ile and Leu, respectively (mutants A-E), were not inhibited by concentrations of nevirapine up to 250 p~. One exception was mutant F, which showed an IC, of -250 p~. Back-substitution to Tyr-181 and Tyr-188 in the otherwise HIV-2 R T sequence (mutant G) restored most, but not all, of the susceptibility to nevirapine inhibition. This mutant, in fact, had an IC, that was still 20-fold higher than that of wtRT-1 (1.37 versus 0.07 p~) . These data confirmed that Tyr-181 and Tyr-188 are necessary (Cohen et al., Shih et al., 1991), but not sufficient, to mediate full inhibition by nevirapine. Mutation of Lys-176 back to Pro (to generate mutant H) did not have a significant effect (0.95 p~) .
However, when Ala-190 was further mutated back to Gly (to generate mutant I), the susceptibility approached that of wtRT-1 (0.14 p~) .
Other back-mutations to wtRT-1 in mutants J-L did not substantially change the IC, values. These data demonstrated that besides Ile-181 + Tyr and Leu-188 + Tyr, conversion of Ala-190 back to Gly contributed to the recovery of nevirapine susceptibility. We next tested whether the Gly-190 + Ala mutation would confer resistance to nevirapine in the absence of other substitutions. The Gly-190 -Ala mutation conferred an increase in IC, of -24-fold (1.68 p~) , thus confirming that Gly-190 plays a critical role in mediating inhibition of HIV-1 R T by nevirapine. In another mutant containing the Gly-190 + Val substitution in crude bacterial lysate, we observed an increase in ICso of -1000-fold over that of wtRT-1 (data not shown).
The Gly-190 + Ala mutant and mutant H, along with the respective control enzymes wtRT-1 and mutant I, were then tested for their susceptibility to inhibition by TIBO-R82150 and E-BPU. Both mutants displayed a 30-fold increase in IC, with E-BPU, but were sensitive to TIBO-R82150 (Table  11). This suggested that TIBO-R82150 does not interact with all of the same residues as do nevirapine and E-BPU.

Interaction of Nevirapine with Mutant
HIV-1 RT

Fifty percent inhibition of RNA-directed DNA polymerase activity by nevirapine on mutant enzymes of HIV-1 RT
The amino acid sequence of HIV-1 RT from residues 176 to 190 is shown. In boldface type are the residues of the HIV-2 RT sequence, where dots indicate residues identical to HIV-1 RT. The underlined sequence is conserved between HIV-1 and HIV-2 RTs and includes Asp-185 and Asp-186 residues critical for polymerase activity (see text). Values were obtained using poly(rC) .oligo(dG) as the template-primer. See "Experimental Procedures" for details.

Enzyme
Amino acid sequence Tyr-ldl+Ile Tyr-181-C~~ Leu-187-Ala  Fluorescence-Nevirapine binding to the poly(rA) . oligo (dT)lP-ls-enzyme complex in the presence of ethidium bromide resulted in changes in the emission intensity of the fluorophore, indicating a perturbation in the interaction between the enzyme and the polynucleotide. Titration with different concentrations of inhibitor were used to calculate dissociation constants. As this method relies on the fluorescence change associated with the polynucleotide, and not the enzyme directly, we refer to the dissociation constants obtained as apparent Kd values. Fluorescence intensity was unchanged with enzymes containing mutations at Tyr-181 and/or Tyr-188 (mutants A-F), confirming that no nevirapine binding occurred (Table 111). On the other hand, mutants in which  (mutants I-L as well as Leu-187 + Ala) were present demonstrated Kd values comparable to that of wtRT-1. Mutants containing the Gly-190 + Ala mutation (mutants G, H, and Gly-190 + Ala) displayed a moderate increase in Kd; however, a poor fit of the data to  "The nonlinear least-squares fit to a one-binding site model was poor; the K d reported for these enzymes is the best fit to the data.
' p < 0.010. Equation 1 was observed. Fig. 3 shows an example of a typical fit of the data for wtRT-1.
Michaelis Constants and Rate Constants- Table IV summarizes the results of steady-state kinetics for dGTP and dTTP with RNA templates. For dGTP, the K, for wtRT-2 was higher than for wtRT-1 (1.75 uersus 0.80 p~) , confirming previous data (Hizi et al., 1991), although kCat/Km ratios were similar for the two enzymes. In most cases, K , was higher for the mutants than for wtRT-1. The only enzymes whose K , values were comparable to that of wtRT-1 were mutants J, L, and Gly-190 + Ala, indicating that mutations at residues 178 (Ile + Val), 179 (Val -+ Ile), 189 (Val 3 Ile), and 190 (Gly -Ala) did not influence K,. Inspection of the two pairs of mutants L uersus K and J versus I indicated that mutation at residue 187 (Leu 3 Ile) increased K,. The effect of this residue on K,,, was confirmed by the Leu-187 "+ Ala mutant, in which the less conservative substitution Leu -+ Ala resulted in an even greater K,,,. Such a perturbation might not be surprising in view of the fact that Leu-187 is immediately adjacent to aspartic acid residues 185 and 186. Since the postulated role of these residues involves chelating the M$+ ions that help orient and bind dNTP molecules, a slight effect on their conformation could be expected to affect binding. Substitution of Tyr-181 with either Ile or Cys also resulted in a significant increase in K,. When A-D) were pooled and the mean was compared to that from other mutants (mutants E-L), the difference (0.33 k 0.05 uersw 0.54 & 0.10 s") had a very high probability value ( p < 0.005), confirming that Tyr-181-Ile affected kcat. The k,,,/K, ratio in these enzymes was therefore dramatically impaired (4-lo-fold lower than that of WtRT-1).
For dTTP, the K, for wtRT-2 was also higher than that for wtRT-1. Mutation of Tyr-181 in WtRT-1 to either Ile or Cys (mutants Tyr-181 3 Ile, Tyr-181 -+ Cys, and A-D) increased K,, mimicking the pattern observed with dGTP. While the mutation Leu-187 -+ Ile did not affect K,, the substitution Leu-187 3 Ala did affect K,, suggesting that Tyr-181 and Leu-187 affect the binding of both purine and pyrimidine bases. As was observed for dGTP, kcat was lower when Ile was substituted for Tyr-181 in the Tyr-181 + Ile mutant, but remained normal when Cys was present instead.
Most of the other mutants also had low kc,,, values, indicating that each of the mutations resulted in a less efficient enzyme. Most interestingly, for mutants in which these substitutions were present simultaneously (mutants A-C), kcat for dTTP was restored to normal values. This suggests that these residues interact in a concerted fashion and that the subtle differences in amino acid sequence within this region may be sufficient to impair this interaction and alter the kinetic values. As with dGTP, the mutation Leu-187 + Ala did not affect kc,, for dTTP, and Gly-190 * Ala did not affect kinetic parameters.
We then tested whether mutations Leu-187 .--, Ile and Tyr-181 + Ile would also increase K, for dGTP on a DNA