Dominance of the E89G Substitution in HIV-1 Reverse Transcriptase in Regard to Increased Polymerase Processivity and Patterns of Pausing*

The substitution of a glycine for glutamic acid at position 89 in human immunodeficiency virus-1 (HIV-1) reverse transcriptase (RT) (E89G) confers resistance to several nucleoside and non-nucleoside inhibitors of RT. As residue 89 contacts the template strand, it has been suggested that this mutation may modulate the conformation of the RT·template/primer complex. In addition, certain mutations in RT that confer resistance to nucleoside analogs, such as M184V, are located near the polymerase active site. To characterize further these substitutions, we performed processivity assays alongside an analysis of pausing profiles with wild-type (wt) RT and recombinant RTs containing substitutions at E89G, M184V, or both. We now show that E89G RT has higher processivity than wt enzyme as well as a different pattern of pausing sites. Similar findings were obtained with the doubly mutated RT, although enzyme containing only the M184V mutation had lower processivity than wt. Consistent with these observations, and from a mechanistic standpoint, both E89G-containing as well as doubly mutated RT had decreased dissociation constants from a complex consisting of RT and template-primer, in comparison with either wt RT or M184V-containing RT. No significant differences were observed among the various enzymes in regard to K m values for the heteropolymeric RNA template used in these studies. Viruses containing the E89G mutation synthesized longer strand DNA products than either wt viruses or viruses containing only the M184V mutation in endogenous RT assays. Thus, the E89G substitution is a dominant determinant in regard to each of the k offvalues from an RT·template/primer complex, RT processivity, and specific patterns of pausing during DNA polymerization.

Drug resistance-conferring mutations may also cause changes in other enzymatic properties of RT. For example, certain mutated enzymes, including those with the K65R or E89G substitutions, have decreased affinity for substrate dNTPs in in vitro assays (17,18). In some cases, viruses containing certain of these mutations, e.g. M184V, have been shown to replicate less efficiently than wt virus in primary cells that possess low levels of dNTP (19). Drug resistant RT containing the M184V change also has reduced processivity (19,20). In contrast, recombinant RTs containing four mutations associated with AZT resistance (i.e. M41L, D67N, T215Y, and K219Q) or the K65R mutation associated with ddC resistance (21) had increased processivity; viruses containing the former changes had a replication advantage over wild-type (wt) viruses in tissue culture (22).
The substitution of glutamic acid (E) by glycine (G) at residue 89 confers multiple resistance to ddNTPs in cell-free assays, but resistance is less evident in tissue culture (16). In addition, this mutation confers resistance to non-nucleoside RT inhibitors and foscarnet both in vivo and in vitro (15,16). E89G-containing RT has also been reported to have increased polymerase fidelity (23) and a decreased affinity for natural dNTP substrates (18). Template repositioning has also been suggested for the E89G mutation that presumably affects the interaction between RT and the template strand located 3-4 nt downstream of the polymerase-active site (24,25). However, the effect of this substitution on RT processivity has not previously been described.
We now report that E89G recombinant RT as well as a double mutant containing both E89G and M184V possessed higher processivity than either wt RT or RT containing the M184V substitution alone. The former two RTs also had different patterns of pausing on an RNA template than wt in primer extension assays, although M184V RT shared the same pausing patterns with wt while being less processive. As dNTP concentrations were decreased, E89G-containing RTs were less active but still more processive than the wt enzyme. E89Gcontaining RT as well as doubly mutated enzyme also dissociated from the RT⅐template/primer (RT⅐T/P) complex at a decreased rate in comparison with wt enzyme. In support of these findings, viruses containing E89G RTs also synthesized longer DNA strand products than other viruses in endogenous reactions.

MATERIALS AND METHODS
Molecular Cloning-The proviral recombinant clone HxB2D was used as a wild-type HIV-1 plasmid. Mutations in RT, i.e. M184V, E89G, and M184V-E89G, were introduced into HxB2D by site-directed mu-* This work was supported by grants from the Medical Research Council of Canada. 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.
Briefly, the mutations E89G, M184V, and E89G,M184V were introduced into the His-tagged RT expression plasmid pRT6H-PROT (gift of Dr. S. F. J. Le Grice, Case Western University, Cleveland, OH). The RT coding sequence of a BamHI fragment (between nucleotides (nt) 121 and 1968) of pRT6H-PROT was subcloned into the corresponding site in pGEM-3Z (Promega, Madison, WI). The RT mutation at amino acid position 184 (M184V) was introduced by replacing the BstXI fragment (nt 580 -808) with that excised from the M184V-mutated proviral plasmid. The E89G mutation was introduced into wt or M184V-containing plasmids separately, by replacing the BsrGI-EcoRV fragment (nt 244 -560) with one from E89G-containing proviral plasmid. The BamHI fragments were then reintroduced into plasmid pRT6H-PROT, and the presence of the mutations was confirmed by sequencing.
Recombinant RT Purification-Recombinant wt and mutant RTs were expressed in E. coli and purified according to the method of Le Grice and Gruninger-Leitch (27). Briefly, RT expression in bacteria was induced by isopropyl-␤-D-thiogalactopyranoside. The RT molecules were processed into heterodimers by an HIV-1 viral protease coexpressed in the bacteria (28). The bacteria were lysed, sonicated, and centrifuged, and supernatants were applied to a nickel-nitrilotriacetate-Sepharose column (Qiagen, Mississauga, Canada). The column was washed, and histidine-tagged RT was eluted using an imidazole gradient. RT-containing fractions were pooled, passed through DEAE-Sepharose (Amersham Pharmacia Biotech, Montreal, Canada), and directly loaded onto SP-Sepharose (Pharmacia). Fractions containing purified RT were pooled, concentrated on a bed of sucrose, and dialyzed overnight against storage buffer (50 mM Tris (pH 7.0), 25 mM NaC1, 1 mM EDTA, and 50% glycerol). The final product, p66/p51 RT heterodimer, was aliquoted and stored at -20°C.
Quantification of RT DNA Polymerase Activity-RT reaction buffer containing 50 mM Tris (pH 7.8), 5 mM MgCl 2 , 60 mM KCl, 10 mM DTT, 5 M dTTP with 2.5 Ci of [ 3 H] dTTP (70 -90 mCi/mM), 10 units of template/primer, i.e. poly (rA)/(oligo(dT) [12][13][14][15][16][17][18] (Pharmacia), and variable amounts of wt or mutant RT were included in 100-l reaction volumes that were incubated at 37°C for 5 min and quenched with 0.5 ml of 10% cold trichloroacetic acid and 20 mM sodium pyrophosphate. After 30 min on ice, the precipitated products were filtered on Whatman glass fiber filters and sequentially washed with 10% trichloroacetic acid and 95% ethanol. The radioactivity of incorporated products was analyzed by liquid scintillation spectrometry. An active unit of RT was defined as the amount of enzyme that incorporates 1 pM of dTTP in 10 min at 37°C.
Ϫk off⅐t (Eq. 1) In these reactions k off values are not affected by the RT activity of each individual enzyme since each reaction is normalized and internally controlled. Each experiment was performed at least three times; standard deviations are indicated.
Templates and Primers-HIV RNA template was prepared in vitro using the MEGAscript TM transcription kit (Ambion, Austin, TX) from linearized plasmid pHIV-PBS, which consists of a 497-base pair HIV-1 sequence spanning the R region of HIV-1 long terminal repeat and a portion of the gag region (30). Single-strand HIV DNA template was prepared by polymerase chain reaction and enzyme digestion. Briefly, an HIV DNA fragment (from a BglII site in the R region to the PstI site in the gag region) was amplified by polymerase chain reaction from plasmid pHIV-PBS with primer pair PS (sense, nt 15-35)/PST(antisense, nt 964 -942). Primer PST was phosphorylated by T4 kinase and ATP before use in polymerase chain reaction. After recovery by ethanol precipitation, the minus (Ϫ) strand of the DNA fragment was digested with exonuclease (Life Technologies, Inc., Montreal, Canada). Then, the single (ϩ) strand which served as a template was extracted with phenol/chloroform, precipitated with ethanol, and quantified.
The 18-nt DNA primer dPR and the 20-nt DNA primer PA are complementary to the HIV-1 primer-binding site (PBS) and the 5Ј end of the gag sequence, respectively. The primers were [␥-32 P]ATP-labeled and filtered by Sephadex G-25 column. The template/primer complex was prepared as follows: the template and primer were mixed at a ratio of 1:2, denatured at 85°C for 5 min, and then sequentially cooled to 55°C for 8 min and 37°C to allow for specific annealing of primer to the template.
Affinity of RT for the RNA Template/Primer-The affinity of HIV-1 RT for the RNA template/primer was determined in reactions that contained fixed concentrations of enzyme and increasing concentrations of both template and primer. Heterologous HIV-1 RNA template and unlabeled primer, dPR, were annealed at a ratio of 1:2 prior to initiation of RT reactions performed in 50 l containing 50 mM Tris (pH 7.8), 5 mM MgCl 2 , 60 mM KCl, 10 mM DTT, 10 M [ 3 H] TTP, variable amounts of pre-annealed RNA template/primer, and 4.3 nM wt or mutant RT. After 10 min at 37°C, reactions were analyzed as described above under quantification. For determination of K m values, RT activity was plotted against concentrations of template/primer; K m values were calculated according to manufacturer's directions using GraphPad Prism 2.0 (GraphPad Software Inc, San Diego, CA).
Primer Extension Assay-The reaction mixture contained 50 mM Tris (pH 7.8), 5 mM MgC1 2 , 60 mM KCl, and 10 mM DTT. 50 nM HIV RNA or DNA template and 100 nM primer were included in the reaction. Twenty units of RT (defined by quantitation, see above) were included in a total volume of 20 l. Fig. 1 illustrates the templates, primers, and the products of DNA polymerization. Concentrations of dNTPs varied in different assays. Reaction mixtures were incubated for 30 min at 37°C. Products were extracted with phenol/chloroform, boiled for 5 min, and electrophoresed in a 5% denaturing polyacrylamide gel. Sequencing reactions were performed with the same primers and homologous DNA to determine the size of RT products. We used the same amount of enzyme activity in each reaction rather than equivalent protein concentrations, since the levels of DNA polymerase activities in wt versus mutated RTs may differ. Were the same molar concentrations of enzymes to have been used, then results in the presence versus absence of trap would no longer be comparable, since differences could result from altered catalytic activity or processivity of the enzymes used.
Endogenous RT Reactions-Mutated and wt viruses were prepared by transfection of 2 g of proviral DNA into 5 ϫ at 39°C and terminated by adding an equal volume of stop buffer (1% SDS, 50 mM EDTA, and 0.2 M NaCl). Reaction mixtures were digested with 20 g of protease at 56°C for 30 min and then extracted with phenol/chloroform followed by ethanol precipitation. Products were resuspended in 20 l of loading buffer, boiled, and separated on 1% denaturing agarose gels (20 mM NaOH, 1 mM EDTA). After drying, gels were visualized by autoradiography.

RT Activities of wt and Mutant Virions in Endogenous
Reactions-Permeabilized virions, which contain RT, genomic RNA template, tRNA Lys.3 primer, and other viral components, can initiate reverse transcription in vitro in the presence of dNTPs. The endogenous RT reaction showed that E89G and E89G,M184V-containing viruses synthesized longer DNA strand products than did wt and M184V viruses over 6 h at 39°C (Fig. 2). Although the E89G singly-mutated virus possessed higher RT activity than the E89G,M184V double mutant, both viruses generated long DNA strand products, i.e. at least 50% of products were Ͼ6 kilobase pairs. In contrast, products generated by the M184V single mutant and wt virus were mostly Ͻ6 kilobase pairs. The endogenous reaction did not show significant differences between wt and M184V-containing viruses. Similar results were obtained in four separate experiments.
RNA-dependent DNA Synthesis by wt and Mutant RTs-The above result suggested that E89G RT might have increased processivity. To test this directly, we performed primer extension assays, in the presence or absence of a trap, using high concentrations of dNTPs (i.e. 200 M), HIV RNA template, and labeled dPR primer (Fig. 3). Since levels of DNA polymerase activities in wt versus mutated recombinant RTs may differ, we used the same amount of enzyme activity in each reaction (quantified as above) rather than equivalent protein concentrations. Under nonprocessive conditions, i.e. in the absence of a trap, the levels of DNA polymerase activity in wt and mutated recombinant RTs were comparable (Fig. 3, left panel), although molecular image analysis showed that the mutant RTs made about 10 -30% more full-length product than wt (not shown). Both the E89G and E89G,M184V RTs yielded fewer intermediate products than did the wt and M184V RTs (Fig. 3, left  panel) and also had different pausing patterns (see below). No differences in pausing patterns were found between wt and M184V-RTs, although M184V RT generated more intermediate products (Fig. 3, left panel).
In contrast, the behavior of the enzymes was markedly different under conditions of single processive cycle polymeriza- tion. For this purpose, a poly(rA)/(dT) [12][13][14][15][16][17][18] trap was added at the time of initiation of reverse transcription to bind to both free RT as well as RT that had dissociated from the template/ primer. Under these conditions, both E89G RT and the double mutant produced more full-length (Ϫ)-strand strong-stop DNA than wt RT (Fig. 3, right panel). In contrast, M184V RT generated only about half as much product as wt RT. E89G-containing RTs also produced more intermediate products than did wt RT, whereas M184V RT generated mostly intermediate products (Fig. 3, right panel). These results are consistent with the data generated in the endogenous reactions discussed above and show an increased processivity of RTs containing the E89G substitution.
Relationship between dNTP Concentration and RT Activities of wt and Mutant Enzymes-E89G RTs have been reported to have decreased affinity for dNTP substrates (18). We therefore compared E89G RT with wt RT in primer extension experiments in the presence of varying concentrations of dNTPs.
At low concentrations of dNTP (i.e. 1-5 M) and in the absence of trap, we found that wt RT possessed the highest level of activity, followed by E89G RT, M184V RT, and the E89G,M184V double mutant (Fig. 4A, left and middle panel). Under these conditions, polymerase activity was eliminated when reactions were performed in the presence of a trap (data not shown). Differences in activity were less pronounced when dNTP concentrations were increased to 15 M (Fig. 4A, right  panel), although wt RT still generated more full-length DNA product than did either M184V or E89G RT, with the double mutant producing the least. However, the effect of processivity at 15 M dNTP was identical to that observed at high dNTP concentrations. In the presence of trap, both E89G RT and E89G,M184V RT produced more products of long strand DNA (although not full-length) than wt and M184V RTs in the presence of 15 M dNTPs (Fig. 4B).
Comparison of Mutant and wt RT Activities on RNA or DNA Templates-We next studied the frequency of pausing and whether mutant and wt RTs paused at the same or different sites. This was tested using primer PA which initiates DNA synthesis on an RNA template from the 5Ј end of the gag region. The reason for switching from primer dPR to PA is that the latter gives rise to longer full-length products than the former, thereby giving RT molecules more opportunity to pause. In fact, the full-length products primed by PA were twice as long as those primed by dPR (Fig. 1). The results show that all four RTs produced similar amounts of full-length products under nonprocessive conditions, i.e. without RT trap (Fig.  5A), but the wt and M184V RTs generated more intermediate products than the mutant enzymes containing the E89G substitution (see below). The pausing pattern of each RT on the RNA template remained the same when DNA synthesis was initiated at different places, i.e. an RT-pausing site primed by dPR would be shared when primed by PA (compare nt positions in Fig. 5A and Fig. 3, left panel). These findings are summarized in Fig. 6.
The character of the template used, i.e. DNA or RNA, may also affect DNA polymerase activity. To test this notion, we employed a single-stranded DNA template that shares sequence homology with the HIV RNA template except for being several nt shorter at the 5Ј end. When reactions were initiated with PA primer, the results again demonstrated that E89G RT and the double mutant produced fewer intermediate products than the other two enzymes (Fig. 5B). However, all four RTs shared the same strong pause site at CCAATA immediately after TTTTT runs on the DNA template, which was different than the pause sites seen with the RNA template.
Comparison of Pausing Patterns between Mutant and wt RTs on RNA or DNA Templates-As stated above, pausing patterns of E89G-containing RTs were different from those of wt and M184V RTs that were similar to each other. Fig. 6 presents a summary of pause sites observed during reverse transcription. At high concentration of dNTPs, i.e. 200 M, E89G RT and the double mutant RT yielded fewer products at most wt and M184V-RT pausing sites. This phenomenon was even more pronounced in the reaction primed with PA than that primed with dPR. When PA was used as primer, only infrequent pausing was seen with the E89G RT and the double mutant RT (Fig.  6A, indicated with closed triangle), but frequent pausing was seen with wt and M184V enzymes. The wt and M184V-RT pause sites were most frequent at G and C residue(s). However, the E89G and E89G,M184V RTs paused between nt 320 and 331 more often than wt RT (Fig. 5A, large open triangle). The pause site seen with these mutated RTs at nt 207 was not seen at all with either wt or M184V RT (Fig. 5A, small open  triangle).
As expected, when the concentration of dNTPs was decreased to 15 M, all four RTs displayed more pausing sites. However, E89G-containing RTs generated fewer full-length products than wt RT; fewer intermediate pause products were also detected at some pause sites, corresponding to an increase in synthesis of full-length product. Both E89G and the double mutant RT produced more intermediate products than either wt or M184V RTs at each of two regions, i.e. GGGAGCUC between nt 24 and 32 and UUGCC between nt 74 and 78 (Fig.  4A, right panel). When DNA polymerization was carried out with a DNA template, all four RTs shared similar pausing patterns (see CCAATA sequence between nt 291 and 296 (Fig.   5B, closed triangle)). Consistent with the increased processivity of E89G RT, this enzyme produced fewer intermediate products than wt RT.
Affinity of Mutant and wt RT for HIV-1 RNA Template/ Primer-The observed differences in processivity between wt and E89G-mutated RT may suggest that the E89G substitution affected enzyme interactions with the template/primer. We therefore assessed both K m values of the various enzymes employed to monitor affinity and also evaluated the dissociation of enzyme from the template/primer (see below). The K m values Ϯ S.D. of each of wt RT, M184V RT, E89G RT, and doublemutated enzyme for heterologous HIV-1 RNA template were After incubation for 30 min at 37°C, reaction products were extracted, boiled, electrophoresed on a denaturing 5% polyacrylamide gel, and visualized by autoradiography. Reactions were performed under nonprocessive conditions in which initiation was by addition of RNA template/primer (sequencing reaction is shown alongside the gel) (A) or in which DNA template/primer, RT, and dNTPs were added simultaneously (B). Sizing of sequences was as described above. Numbers identify nt positions. 5.10 Ϯ 0.38, 4.75 Ϯ 0.46, 6.06 Ϯ 0.59, and 5.29 Ϯ 0.41 nM, respectively. Thus, the various mutated and wt RTs shared similar affinities for the RNA template/primer.
Dissociation Rate of the RT⅐Template/Primer Complex-The above results (i.e. Fig. 3), obtained in single processive cycle DNA polymerization reactions in the presence of an RT trap, suggest that the various enzymes tested might dissociate from the RT⅐T/P complex at differential rates. This was directly studied as described under "Materials and Methods" through use of an RT trap to prevent rebinding of dissociated RT molecules as per Equation 2. Fig. 7 presents data pertaining to the RT⅐T/P complex remaining in reactions after various times in the presence of trap in comparison to the amount of complex that was present at time 0. Based on these results, the k off values for each of wt, M184V, E89G, and E89G,M184V RTs were calculated as 0.25 Ϯ 0.03, 0.28 Ϯ 0.03, 0.17 Ϯ 0.01, and 0.14 Ϯ 0.01 per min, respectively. Thus, both the E89G and double-mutated enzymes had k off values that were significantly reduced in comparison to both wt enzyme and that containing the M184V substitution. These results are consistent with those obtained in the single cycle DNA synthesis experiments presented above.

DISCUSSION
Processivity of a DNA polymerase refers to the number of dNMPs added to a nascent DNA strand before the enzyme dissociates from the template/primer complex. Processive synthesis of DNA polymerase enzymes involves several factors such as template sequence and structure and the nature of the enzyme itself. HIV-1 RT pauses frequently during runs of Gs and Cs on RNA templates (32)(33)(34). Our results are consistent with this observation. However, the profile and frequency of pausing seen with the E89G and E89G,M184V RTs differed significantly from wt enzyme (Figs. 5 and 6).
When DNA templates were used, runs of Ts and As severely affected HIV-1 RT processivity (34,35). Here we show that fewer differences in pause sites were found on DNA than RNA templates regardless whether wt or mutated RTs were used. In addition, fewer pause sites were seen with E89G RT than wt RT in regard to DNA templates. Although many RT enzymes are blocked in all reactions related to synthesis of DNA at a CCAATA sequence immediately after TTTTT runs, E89G-containing RTs made more full-length but fewer short products than wt and M184V RT (Fig. 5B). RT pausing may result from primer-template slippage in homopolymeric nucleotide runs (35,36) or because of local DNA bending caused by dA⅐dT base pair runs (37).
The processivity of an RT reaction is also related to the concentrations of dNTP substrates used because RT molecules have the potential to stall and dissociate at certain sites on any given template (i.e. pausing sites). Since the catalytic speed of RT is higher at high than at low dNTP levels, it follows that less stalling of RT in the T/P⅐RT complex should occur when higher dNTP concentrations are employed (40). This may explain the data of Figs. 3 and 4; notably, when tested at 200 M dNTPs in the absence of trap, E89G-containing RTs generated similar levels of product as did wild-type enzyme but fewer products when lower dNTP concentrations were employed. This finding is consistent with previous results that E89G RT had a decreased affinity for dNTPs (18).
The determinants of processivity in regard to DNA polymerase are unclear, and it is not known why HIV-1 RT is less processive than other DNA polymerases (32,38,39), although comparisons between the crystal structures of certain of these enzymes have suggested a role for the size of the thumb and the fingers subdomains (40,41). This is supported by the fact that T7 RNA polymerase, the Klenow fragment, and HIV-1 RT have high, moderate, and low processivity, respectively. In T7 RNA polymerase, dissociation of the DNA template is presumably inhibited by the large thumb and fingers subdomains that are wrapped around the nucleic acid substrate. In contrast, the distances between these two subdomains is greater in both HIV-1 RT and the Klenow fragment than in T7 RNA polymerase; thus, the fact that the former two enzymes are less processive that T7 RNA polymerase may facilitate dissociation of the primer/template. In addition, certain amino acid substitutions in RT that confer drug resistance can also affect processivity, e.g. both an AZT-resistant RT containing M41L, D67N, T215Y, and K219Q mutations and a ddC-resistant RT containing a K65R change had enhanced processivity (21,22), whereas the opposite result was obtained in regard to M184V RT (19,20). We have now shown that E89G-containing RT also had higher polymerase processivity than wt and displayed different patterns of pausing, especially during C runs on an RNA template (Fig. 5A, closed triangle). In contrast, M184V RT shared the same pausing site patterns as wt RT. Long DNA products synthesized by E89G RT in endogenous reactions may result from this enhanced processivity. The distinct pausing pattern of E89G RT may also reflect a distinct manner of interaction with template/primer. Of course, other factors may also impact on this situation, including the possibility that E89G-containing RTs may have an advantage over wt enzyme in regard to strand transfer events.
We had considered the possibility of repeating our endogenous viral assay experiments with virus derived from transfections of CD4-negative cells lines in order to provide more uniform sampling. Unfortunately, these transfections did not yield enough virus for use in the endogenous assay. However, consistent results were obtained when the same or different batches of virus were tested in the endogenous assay on multiple occasions. The stability and processing of wt and mutant enzymes were similar as determined by Western blot assay (results not shown).
The crystal structure of HIV-1 RT reveals that amino acid 89 is located within strand ␤5␣, which forms part of the enzyme template grip, whereas amino acid 184 is located in the ␤9-␤10 turn, which, in turn, constitutes part of the enzyme-active site (42)(43)(44)(45)(46)(47). Substitution of an acidic Glu by a neutral Gly at position 89 is thought to alter enzyme interaction with the template/primer (24,25,48). The resultant template repositioning could induce overall conformational changes of the RT⅐T/P complex. Thus, the E89G mutation may indirectly influence the ability of mutated RTs to bind to dNTPs as well as to both ddNTP and non-nucleoside inhibitors, helping to explain the multiple resistance conferred by this substitution (16,18,49). The fact that E89G also affects RT⅐T/P interactions could influence enzyme processivity.
Our results also indicate a potential mechanism for the differences in patterns of DNA synthesis reported in this paper among the different enzymes tested and the dominance of the E89G substitution in regard to processivity. Notably, we found that both E89G mutated RT as well as doubly mutated enzyme, i.e. E89G,M184V both had k off values that were significantly below those of wt or M184V-containing RTs (Fig. 7). This helps to explain why the E89G-containing enzymes also showed consistent diminished pausing during DNA polymerization, despite the fact that the affinities of the various mutated enzymes for template/primer were not significantly different from that of the wt enzyme or each other. These results are consistent with the notion that transcriptional pausing mainly results from the dissociation of RT from the RT⅐T/P complex (33).
Viruses containing the E89G substitution in RT have been shown to be less replication competent than wt viruses (16). This result was not unanticipated, since E89G-containing viruses are not commonly found in untreated patients. One reason that E89G viruses do not predominate despite increased RT processivity is that E89G RT has a lower affinity than wt enzyme for dNTPs (18). Results in our lab have now shown that E89G-containing viruses replicate marginally slower than wt viruses, even in cell lines that maintain high dNTP pool concentrations (results not shown).
The low fidelity of HIV-1 RT is thought to be related to its pausing sites (32, 35, 50 -52). The diminished pausing of E89Gcontaining RT on template may explain the enhanced fidelity of this mutated enzyme (23). However, increased processivity occurs independently of any drug resistance associated with E89G RT. We have also shown that E89G RT has reduced sensitivity to both AZTTP and 3TCTP in primer extension assays although diminished sensitivity to either AZT or 3TC is not conferred by this change in tissue culture (not shown). Of course, other factors such as the viral nucleocapsid protein can also impact on reverse transcription within cells (53).
In addition, the M184V substitution in RT has been shown to FIG. 7. Dissociation of RT from the RT⅐T/P complex. Equal concentrations of RT and template/primer, i.e. poly (rA)/ oligo(dT) [12][13][14][15][16][17][18] , were preincubated for 5 min at 37°C, and the dissociation of RT from the resultant complex was measured by trapping dissociated RT molecules, as described under "Materials and Methods." The inset to the figure describes k off values Ϯ S.D. for each enzyme tested.
increase polymerase fidelity in regard to both the DDDP and RDDP steps of the RT reaction (23, 54 -56), while simultaneously causing diminished processivity. Thus, it is possible to distinguish among mutations in RT in regard to effects on fidelity versus processivity; this subject merits further investigation.
Interestingly, the E89G,M184V double mutant RT behaved like E89G but not like M184V RT, showing the dominance of E89G over M184V in regard to each of processivity, pausing, and dissociation from the RT⅐template/primer complex. These findings may lead to better understanding of the roles played by these individual amino acids within RT.