Inhibition of HIV-1 Reverse Transcriptase by a Quinazolinone and Comparison with Inhibition by Pyridinones DIFFERENCES IN THE RATES OF INHIBITOR BINDING AND IN SYNERGISTIC INHIBITION WITH NUCLEOSIDE ANALOGS"

(L-738,372) representa-tive of a novel structural class of nonnucleoside inhibi- tors of human immunodeficiency virus, strain 1 (HIV-l), reverse transcriptase (RT), the quinazolinones. L-738,372 is a reversible inhibitor of HIV-1 RT and is noncompetitive against dTTP with a Ki of 140 n~ with poly(rA)-oligo(dT) as primer-template. Mixed noncom- petitive inhibition by L-738,372 was observed against poly(rC).oligo(dG) as primer-template. This quinazolin- one binds to RT at a site that overlaps the binding site of other nonnucleoside inhibitors as evidenced by the abil- ity of L-738,372 to displace bound radiolabeled L-696,229, a member of the pyridinone class of inhibitors of HIV-1 RT, from complexes of RT and primer-template. Inhibition by L-738,372 shows slow binding characteristics in reactions with all of the primer-templates employed. in

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Clinical trials for the efficacy of the pyridinone, L-697,661 (structure shown in Fig. IA), were hampered by the rapid appearance of resistant strains containing the Y181C mutation in the RT gene (Saag et aZ., 1993). The ease of development of resistance to inhibition by pyridinones created the need for another class of inhibitors that would be effective against the resistant virus and lead to the investigation of the quinazolinones, of which L-738,372 is a member (Fig. 1B) (Tucker et aZ., 1994). L-738,372 is a potent inhibitor of the form of RT containing the Y181C mutation: but does not inhibit HIV-2 RT.' The development of resistance in strains of HIV-1 to inhibition by L-738,372 is under investigation.
As a basis for comparison with the other structural classes of nonnucleoside inhibitors, this work investigates the mechanism of inhibition of RT activity by L-738,372. L-738,372 shares many mechanistic features with the other classes of nonnucleoside inhibitors, including the same or overlapping binding sites and the same mode of inhibition against dNTP. However, differences in the inhibition of RT activity by members of the group of nonnucleoside inhibitors are apparent in the areas of synergistic inhibition in combination with nucleoside analogs and rates of association of the inhibitors and RT. These differences between structural classes of nonnucleoside inhibitor may impact on the effects of the use of these compounds in clinical trials.
Primer-Template-DNA and RNA substrates were prepared as described previously (Carroll et al., 1993).
Steady-state Kinetic Assays with Poly(rA).OZigo(dT)-Reactions with poly(rA).oligo(dT) as primer-template were carried out at 25 "C, except where indicated otherwise, in the presence of poly(rA).oligo(dT) reaction buffer (50 m M Tris, pH 7.8,80 m M KC1,l m M DTT, 6 m M MgCl,, 0.2% PEG 8000, 100 p~ EGTA, 50 p g h l poly(rA).oligo(dT)), 0.5 n M RT, 2-15 p~ [a-32PldTTP, and 0-750 n~ L"738,372 to give a final concentration of 4% dimethyl sulfoxide in all reactions. Reactions were initiated by the addition of dTTP after preincubation of the other components for at least 1 h. Aliquots of the reaction mixture were quenched after 15 min into 0.5 M EDTA solution, pH 8, prior to product analysis by filter binding using DE-81 filters (Whatman) as described previously (Bryant et al., 1983).
Steady-state Kinetic Assays with Poly(rC).Oligo(dG)-Reactions with poly(rC).oligo(dG) were carried out under the same conditions as for reactions with poly(rA).oligo(dT) except that MgCl, was present at 30 mM, poly(rC).oligo(dG) was varied from 5 to 75 pgt'ml, and [ C T -~~P I~G T P was present at 3 p~. Aliquots of the reaction were quenched in gel load buffer, and product analysis was carried out with gel electrophoresis and a PhosphorImager (Molecular Dynamics).
Yonetani-Theorell Plots-Analysis for synergistic inhibition in combinations of L-738,372 and L-696,229 was carried out using reaction conditions as described for poly(rA).oligo(dT), except that the concentration of dTTP was 5 p. Analysis for synergistic inhibition in combinations of L-738,372 and AZTTP was carried out using poly(rA).oligo(dT) as primer-template and the buffer conditions described above except the concentration of dTTP was 5 PM, and DTT was omitted. Reactions included L-738,372 (0-600 n~) and either 0,6, or 12 n M AZTTP. L-738,372 was allowed to bind to RT and poly(rA).oligo(dT) during a 1-h preincubation prior to initiation of the reaction by the addition of a mixture of dTTP and AZTTP. Analysis of synergistic inhibition by combinations of L-697,661 and AZTTP was carried out using the same procedure.
Analysis for synergistic inhibition by combinations of L-738,372 and ddITP was carried out essentially as described for the combination of L-738,372 and AZTTP except that the primer-template used was poly(rC).oligo(dG), dGTP was at 5 p~, and the preincubation time was 2 h. Reactions included L-738,372 (0-40 n~) and 0, 0.5, or 1 m M ddITP. Analysis for synergistic inhibition by combinations of L-738,372 and ddCTP was carried out as described above except that the primertemplate used was poly(rI).oligo(dC), and the reactions included L"738,372 (0-600 m) and 0 , 5 0 , or 100 n M ddCTP.
Inhibition of RT activity by the combination of L-697,661 and AZlTP was also determined by the method of fractional inhibitory concentrations (Elion et al., 1954). Reactions contained standard reaction buffer except DTT was omitted, the concentration of dTTP was 4 p~, and 0-20 PM L-697,661 and 0-200 n~ AZTTP were included. The concentration of each inhibitor used individually that was required to give a certain fraction of inhibition in the range of 80-100% was determined from a Hill plot. The fractional inhibitory concentration of each inhibitor for each reaction was calculated as the concentration of the inhibitor present in the reaction divided by the concentration required to give the same degree of inhibition when the inhibitor was used alone.
Reversibility of Inhibition by L-738,372-RT (25 nM) was preincubated for 1.5 h at 25 "C with L-738,372 (0 or 1 p~) in the standard poly(rA).oligo(dT) reaction buffer except that dTTP was omitted and the concentration of poly(rA).oligo(dT) was increased to 500 pg/ml. An aliquot of the incubation solution was diluted 50-fold into poly(rA).oligo(dT) reaction buffer without dTTP that contained either 0, 20 nM, or 1 PM L-738,372. The enzymatic activity of the diluted solution was monitored over a 9-h period by initiating a reaction with 45 pl of the diluted RT solution and 5 pl of 50 p [CT-~~PI~TTP. After 15 min, the reaction was quenched, and product analysis was carried out by filter binding as described above.
Displacement of Enzyme-bound pHlL-696,229 by L-738,372-[3HlL-696,229 (100 nM) was preincubated with 7 n M RT and 30 pg/ml poly(rC).oligo(dG) in 50 m M Tris, pH 7.8, 80 m M KCl, 6 m M MgCl,, 1 m M DTT, 40 p~ dGTP, and 0.2% PEG 8000 for 20 min. The solution was divided into three aliquots. To one was added L-738,372 to a final concentration of 1 PM. To another was added unlabeled L-696,229 to the same concentration. To the third aliquot was added buffer as a control. Aliquots of each reaction were removed at intervals up to 30 min and centrifuged through a 1-ml Sephadex G-50 column, and the eluate was monitored for radioactivity by scintillation counting.
Analysis of Slow Binding of L-738,372-Reactions included the same buffer conditions as described above. Reactions were initiated either by the addition of RT or by the addition of dTTP to the reaction after preincubation of the other buffer components for 3 h. Aliquots were withdrawn after reaction times up to 30 min, quenched, and analyzed for the amount of product formed by filter binding as described.
Steady-state Reactions with Heteromeric Primer-Templates-Reactions were carried out using the same buffer conditions as described for assays with poly(rA).oligo(dT). Reactions included 5-10 n M RT and 500 n M 5'-32P-end labeled primer-template. Reactions containing L-738,372 were initiated by the addition of dNTP to 5 after preincubation of the enzyme and inhibitor for 3 h. Aliquots were quenched after various reaction times. Product analysis was carried out with gel electrophoresis and autoradiography, as described previously (Carroll et al., 19931, or with a PhosphorImager. Data Analysis-All nonlinear regression calculations were performed as described previously (Carroll et al., 1993). For determinations of IC,, values, inhibitor saturation data were fitted iteratively to Equation 1, (Eq. 1) where Y is the fraction of activity inhibited, a is the residual activity at saturating concentrations of inhibitor, and [I,] is the total concentration of inhibitor.

RESULTS
Slow Binding Inhibition-Slow binding was observed in the inhibition of RT activity on poly(rA).oligo(dT) by L-738,372, as indicated by curvature in a reaction progress curve in the presence of inhibitor. The curvature could be eliminated by preincubation of RT, primer-template, and inhibitor. The rate constant for association of L-738,372 and the complex of RT and poly(rA).oligo(dT) was determined by initiating reactions by addition of RT in the presence of L-738,372 and analyzing for product formation as a function of reaction time. The observed rate of association was determined by fitting the data from a plot of product versus time, an example of which is shown in Fig. 2, to the integrated form of the rate equation, where p(t) is the concentration of product formed at time t , v i is the final rate of the inhibited reaction, u, is the rate of the uninhibited reaction, and k is the observed rate of inhibitor binding (Morrison, 1982). The value of vi was determined by preincubating inhibitor, RT, and primer-template in reaction buffer for a time sufficient to allow the complete association of the inhibitor and then initiating the reaction by addition of dNTP. The replot, shown in the inset of Fig. 2, of the observed rate of association versus the concentration of L-738,372 was linear over the range of inhibitor concentrations used and was indicative of a one-step mechanism for binding of the inhibitor  Table I) was also observed using the same method. The rates of association are shown in Table 11. The rates of dissociation of L-738,372 are either determined from the intercept of the plot of kobs versus [L-738,3721 or are calculated from the measured K,, as determined from the replot of the slopes of a Lineweaver-Burk plot (Fig. 31, and the measured rate of association using 3) The rate of dissociation of L-738,372 from complexes of RT and poly(rC).oligo(dG) is similar to the rate of dissociation of L-738,372 from complexes of RT and poly(rA).oligo(dT). Therefore the change in inhibitory potency is mainly due to the change in the rate of association of the inhibitor to complexes of RT and the two homopolymers. The rate of association of L-738-372 to complexes of RT and poly(rA).oligo(dT) increased by a factor of 2.2 with a n increase in the reaction temperature from 25 to 37 "C.
Slow binding was not observed in the inhibition of RT activity on poly(rA).oligo(dT) by L-697,661 (data not shown), but it was observed in the inhibition by L-697,661 with poly(rC).oligo(dG) as primer-template . The inhibitory potencies of L-697,661 and L-738,372 are comparable with poly(rA).oligo(dT) as the primer-template. However, since L-697,661 does not show slow binding characteristics with poly(rA).oligo(dT) as the primer-template, both the rates of association and dissociation of L-697,661 must be higher than the corresponding rates for binding of L-738,372 to complexes of RT and poly(rA).oligo(dT).
Reversibility of Inhibition-Reversibility of inhibition by L-738,372 was investigated using two methods. A plot of activity against RT concentration at different fixed concentrations of L-738,372 (Ackermann and Potter, 1949) gave a set of lines that intersected at the origin, indicating reversibility of inhibition The more complex two-step mechanism of binding where the initial collision complex undergoes a tightening of binding cannot be excluded at inhibitor concentrations higher than those used, although the relevance of a two-step mechanism of binding, if it exists, is dubious since K, is 140 nM and the one-step mechanism applies up to 600 nM.   The K, was determined as the ratio, k,dk,. Table 1. (data not shown). The second method involved the preincubation of RT, poly(rA).oligo(dT), and L-738,372 for 1.5 h to allow the binding of the inhibitor to take place. The complex was then diluted 50-fold, and the level of RT activity in the diluted solution was monitored. The final level of activity in the diluted reaction equaled the activity of a solution with the same final L-738,372. Reactions were initiated by the addition of dTTP after preincubation of the other components for 1 h and were quenched at specified time points by the addition of 20 p1 of 0.5 M EDTA, pH 8, and the amount of product was analyzed by filter binding as described under "Experimental Procedures." Data were fit to a simple noncompetitive inhibition pattern.  Fig. 4 is therefore 2.2 x lo4 s-l with f = 0.15, in reasonable agreement with the rate of dissociation of 3 x s-* calculated from the measured k,, and K,, but not in agreement with the rate of dissociation determined from the intercept of the plot in Fig. 2 (1.3 x s-'). This discrepancy is probably due to the uncertainty in determining the intercept by extrapolation from the data points to the y axis. We therefore favor the value of 3 x s-l for the dissociation of L-738,372 from complexes of RT and poly(rA).oligo(dT). In contrast, when the reversibility experiment was carried out using L-697,661, maximal activity was regained within 10 min after the dilution of the complex of RT, primer-template, and L-697,661.

Sequences shown in
Mode of Inhibition against dNTP-As shown in Fig. 3, L-738,372 was determined to be a linear, noncompetitive inhibitor against dTTP in reactions that employed poly(rA).oligo(dT) as primer-template and that allowed the equilibration of inhibitor binding by preincubation of L-738,372, RT, and poly(rA).oligo(dT) for 1 h. The I C j for L-738,372, as determined from the replot of the slopes (Fig. 3, inset), was 140 m.
Mode of Inhibition against Primer-Template-L-738,372 was determined to be a mixed noncompetitive inhibitor against poly(rC).oligo(dG) as the variable substrate in reactions that allowed the equilibration of the binding of L-738,372 during a 2-h preincubation (Fig. 5).
Mutually Exclusive Znhibition by L-738,372 a n d Pyridinones--Two experimental methods were used to determine whether L-738, 372  .oligo(dT) reaction buffer (as defined under "Experimental Procedures") except that dTTP was omitted and the concentration of poly(rA).oligo(dT) was increased to 500 pg/ml. An aliquot of the incubation solution was diluted 50-fold into poly(rA).oligo(dT) reaction buffer without d'M'P that contained either 0 (O), 20 n M ( O ) , or 1 p~ (0) L-738,372. The enzymatic activity of the diluted solution was monitored as described under Experimental Procedures. The change in activity following dilution ofthe solutions was fit to an exponential equation to determine the observed rate of approach to the new equilibrium, yielding 2.4 x lo4 s-l( 0 ) and 2.8 x lo4 s-' (0). re11 (Yonetani and Theorell, 19641, which showed parallel lines indicating mutually exclusive inhibition by the two compounds (data not shown). Both L-738,372 and unlabeled L-696,229 displaced radiolabeled L-696,229 from RT.primer-template complexes in experiments implementing a spin-column technique  for resolving enzyme-bound L-696,229 from free compound (data not shown). In the absence of added  (Fig. 6A). Mutually exclusive inhibition by the combination of AZTTP and L-697,661 in the range of inhibition tested (activity <92% inhibited) was indicated in a Yonetani-Theorell plot (Fig.  6B). Similar results were obtained with the combinations of each nonnucleoside inhibitor and either ddITP or ddCTP. The possibility of synergistic inhibition by combinations of L-697,661 and AZTTP at higher concentrations of both inhibitors was investigated using the method of fractional inhibitory concentrations (Elion et al., 19541, and the results are shown in Fig. 7. Synergistic inhibition, when the enzyme was inhibited by more than 92%, was evident because the data points fall below and to the left of the line of additivity. In reactions in which RT activity is inhibited by less than 92%, the data points fall close to the line of additivity. Inhibition of AZT-resistant RT-The AZT-resistant RT was inhibited by L-738,372 with an IC,, on poly(rA).oligo(dT) of 60 m, under conditions that allowed for the equilibration of inhibitor binding by preincubating RT, poly(rA).oligo(dT), and L-738,372. Under the same conditions, the wild-type RT was inhibited with an IC,, of 150 nM. L-697,661 showed the same ratio of inhibitory potencies against the wild-type and AZTresistant RTs.

Inhibition by L-738,372 on Heteromeric Primer-Templates-
The inhibitory potency of L-738,372 was determined using the heteromeric primer-template systems shown in Table I. The sequence of the synthetic 32-mer RNA template is taken from the HIV-1 genome (Alizon et al., 1986). Reactions were limited to the incorporation of a defined number of nucleotides by including only one dNTP that allowed the evaluation of inhibitory potency at a specific region of the template. To allow for the complete association of L-738,372 and complexes of RT and primer-template, the inhibitor was preincubated with the other reaction components in the absence of dNTP. Preincubation times of over 1 h were required for complete binding of L-738,372, whereas L-697,661 was completely bound in less than 10 min.
As previously reported for inhibition of HIV-1 RT activity by L-696,229 (Carroll et al., 1993), the potency of inhibition by L-738,372 is dependent on the position of the template that is being transcribed. For example, the IC,, for inhibition on substrate p22/tC5U is 6-fold greater than the IC,, on p15/tC5U. L-738,372 inhibits the DNA-dependent DNA polymerase activ-ity of RT on substrate p22/tD32 with a potency similar to that with which it inhibits the RNA-dependent DNA polymerase activity of RT on substrate p22/tC5U, which has the same sequence as p22ltD32 with uracil substituting for thymidine. At saturating concentrations, L-738,372 inhibits completely all of the reactions on the heteromeric RNA sequences catalyzed by RT shown in Table I. DISCUSSION Several different structural classes of nonnucleoside inhibitors of HIV-1 RT have been discovered in screening programs (for review, see De Clercq (1993)). Clinical studies of pyridinone compounds have revealed the rapid emergence of viral strains resistant to these inhibitors and have pointed out the need for compounds that are more effective against the broad range of mutants of RT. The appearance of virus resistant to L-697,661 lead to the development of L-738,372, the lead compound of another structural class of inhibitors of HIV-1 RT, the quinazolinones.
The inhibition of RT activity by L-738,372 has several features in common with inhibition by other nonnucleoside inhibitors of HIV-1 RT. L-738,372 is a linear, noncompetitive inhibitor against dNTP. Other nonnucleoside inhibitors have also been reported to be noncompetitive against dNTP (Pauwels et al., 1990;Merluzzi et al., 1990;Romero et al., 1991;Carroll et al., 1993). L-738,372 is a mixed noncompetitive inhibitor against primer-template. Other nonnucleoside inhibitors have been reported to be either noncompetitive, mixed, or uncompetitive against primer-template (Debyser et al., 1991;Tramontano and Cheng, 1992;Althaus et al., 1993). Nonnucleoside inhibitors of HIV-1 RT, including L-738,372 (Tucker et al., 1994) have been reported to be highly specific, with no detectable activity against cellular polymerases or HIV-2 RT.
An unusual feature of L-738,372 compared with most other nonnucleoside inhibitors is the slow binding nature of inhibition by L-738,372. The inhibition of RT activity on poly(rA).oligo(dT) by L-738,372 exhibited slow binding characteristics with an association rate constant of 5 x lo3 s-l at 37 "C. The pyridinone class of nonnucleoside inhibitor has also been reported to have slow binding characteristics with poly(rC).oligo(dG) as primer-template but not with .oligo(dT) as primer-template . Although the mechanistic basis for the slow binding inhibition of L-738,372 has not been established, the structural basis for the difference in association rate constants with poly(rA).oligo(dT) as the primer-template for the two nonnucleoside inhibitors may be related to the greater conformational rigidity of L-738,372. The observation that inhibition by TIBO of the presteady-state burst amplitude of synthesis by HIV-1 RT required preincubation of inhibitor and enzyme was suggested to be due to a requirement of TIBO to bind to the free enzyme (Gopalakrishnan and Benkovic, 1994). Preincubation of L-738,372 and RT for 10 s prior to initiation of the reaction with primer-template and dNTP did not eliminate the slow binding, indicating that binding to the free enzyme is also slow, although that may be the preferred mode of binding. Another important difference between inhibition of RT activity by L-738,372 and by L-697,661 is their synergistic inhibition in combination with nucleoside analogs. Synergistic inhibition of RT activity by the combination of L-738,372 and AZTTP was evident in the nonparallel lines in a Yonetani-0.4 0.8 1 22 1.6 2 [L-697,6611 (PM) Theorell plot. It is important to note that synergistic inhibition does not require synergy in the binding of the two inhibitors but only that it be possible operationally for both inhibitors to bind to the enzyme simultaneously (Segel, 1975). Therefore it is possible for both L-738,372 and whatever form of AZT that causes the inhibition, which may be the AZTMP-terminated primer (Heidenreich et aZ., Reardon and Miller, 19901, to bind simultaneously t o RT. Similar results were found with the combinations of L-738,372 and either ddITP or ddCTP. In contrast, inhibition by the combination of L-697,661 and AZTTP appeared to be additive at less than 92% inhibition of activity, but synergistic inhibition became apparent in plots of fractional inhibitory concentrations at greater than 92% inhibition. The most likely explanation for the requirement of a high fraction of inhibition in order to observe synergistic inhibition is that binding of the first inhibitor decreases the affinity of the second inhibitor for the complex of enzyme and first inhibitor. Therefore higher concentrations of both inhibitors are required to populate the doubly inhibited enzyme complex required for synergistic inhibition. The observation of synergis-