Introduction

Top Summary Introduction Materials & Methods Results Discussion References

The RT of HIV-1 is essential for HIV-1 replication because it generates the viral DNA that integrates into the cellular genome. Currently there are seven approved drugs for the treatment of AIDS that specifically target HIV-1 RT; five of these are nucleoside analogs: AZT, ddC, ddI, 3TC, and d4T. These RTIs block RT function both as competitive inhibitors with regard to dNTP substrate and as chain terminators of viral DNA synthesis. However, RTI-resistant HIV-1 has developed in patients receiving these treatments (Larder et al., 1989; Rooke et al., 1989; Gu et al., 1992, 1994b; Kozal et al., 1994; Lin et al., 1994; Zhang et al., 1994; Wainberg et al., 1995) and this has been shown to result in increased viral loads (Schuurman et al., 1995; Zazzi et al., 1996). As viral load is strongly correlated with disease progression (Mellors et al., 1996; O'Brien et al., 1997), such increases are likely to reduce long-term clinical benefits. Therefore, new antiretroviral drugs with unique genetic resistance profiles need to be developed.

PMEA (adefovir) is an acyclic nucleoside phosphonate analog that functions as an RTI and is active against multiple retroviruses, including HIV-1 (Pauwels et al., 1988; Cherrington et al., 1996a), as well as other DNA viruses, including herpes viruses (De Clercq et al., 1986; De Clercq et al., 1987) and hepadnaviruses (Yokota et al., 1994). PMEA has also shown potent antiviral activity as a prophylaxis in the simian immunodeficiency model of AIDS (Tsai et al., 1994). An orally bioavailable prodrug, adefovir dipivoxil, has shown anti-HIV activity in phase I/II clinical trials (Deeks et al., 1997) and is presently in phase III clinical trials for the treatment of AIDS and phase II trials for hepatitis B infections. PMEA is a nucleotide analog that requires only two phosphorylation steps by cellular enzymes to become the active metabolite (PMEApp) in cells (Balzarini et al., 1995; Robbins et al., 1995). This novel phosphorylation requirement permits its activity in a wide variety of cell types, including resting T cells and cells of the monocyte/macrophage lineage (Shirasaka et al., 1995; Perno et al., 1996).

In vitro selection of HIV-1 in the presence of PMEA has resulted in the identification of two different mutations in RT with reduced sensitivity to PMEA, K65R, and K70E (Gu et al., 1995; Cherrington et al., 1996b). The K65R mutant of RT showed greater resistance to PMEA than the K70E mutant and was also cross-resistant to 3TC, ddC, and ddI in vitro. The K65R enzyme has been extensively characterized in vitro and showed decreased affinity for most RTIs, in agreement with antiviral susceptibility data (Gu et al., 1994a). We now report that the K70E mutant enzyme shows only minor decreases in affinity for PMEApp, correlating with its 9-fold reduction in PMEA sensitivity in cell culture. Moreover, we observe that the K70E mutant RT is less active and less processive than wild-type RT, correlating with its reduced in vitro replication capacity (Cherrington et al., 1996b). Of these two in vitro PMEA-selected mutations, only the K70E mutation has been observed to develop in two of 29 patients treated with adefovir dipivoxil for up to 1 year (Mulato et al., 1998).

	Materials and Methods

Top Summary Introduction Materials & Methods Results Discussion References

Cloning, expression, and purification of wild-type and mutant RT. The wild-type RT expression construct pRT66 was a gift from M. Wainberg and has been described by Gu et al. (1994a). Briefly, the pol sequences were polymerase chain reactions amplified from the HXB2D molecular clone of HIV-1 and then transferred into an expression vector pKK223-3 (Pharmacia Biotech, Piscataway, NJ). Appropriate initiation and stop codons were included in the polymerase chain reaction primers. Mutant expression vectors corresponding to K65R, K70E, and M184V RTs were subsequently generated by oligonucleotide-based site-directed mutagenesis of the pRT66 vector. All constructs were sequenced to verify correct nucleotide sequences. Escherichia coli JM109 were transformed with the wild-type or mutant constructs and then induced with 1 nM isopropyl -D-thiogalactopyranoside. Purification was performed according to Hansen et al. (1987) using, sequentially, DEAE cellulose, phosphocellulose, and poly(rC)-agarose column chromatography.

K_i/K_m determination. The enzyme kinetic analyses were performed as described in Cherrington et al. (1995). The reaction mixtures for the DNA-dependent DNA polymerase function contained 50 mM Tris·HCl, pH 8.0, 5% glycerol, 1 mM DTT, 500 µg/ml BSA, 5 mM MgCl₂, 200 µg/ml activated calf thymus DNA (Pharmacia), 60 µM of each dNTP, and various concentrations of the appropriate [³H]dNTP (30 Ci/mmol; Amersham, Arlington Heights, IL). For RNA-dependent DNA polymerase activity, a defined 86-bp RNA template was annealed to a 15-bp DNA oligonucleotide primer (Cherrington et al., 1995). For all reactions, approximately 10⁴ units of enzyme were used per 60-µl reaction (unit = incorporation of 1 nmol of [³H]dNTP/hr at 37°). Kinetic constants were determined by fitting the initial rate data to Lineweaver-Burk plots using the KinetAsyst program (Think Technologies).

Recombinant RT quantification. The concentrations of the recombinant RT preparations were determined by quantitative immunoblots using a commercially available recombinant RT (Worthington Biochemical, Freehold, NJ) as a standard. The standard was diluted to a 1 ng/µl concentration and 2, 4, 8 and 16 µl were electrophoresed with 1-15 µl of the RT preparations of unknown concentration in a 10% sodium dodecyl sulfate polyacrylamide gel. A nitrocellulose immunoblot was prepared, blocked with 5% nonfat dry milk in PBS-T, incubated overnight with an anti-HIV-1 RT monoclonal antibody (Intracell, Cambridge, MA; 1:400 in PBS-T/1% BSA), and then incubated with Cy5-conjugated donkey antimouse IgG (Jackson ImmunoResearch, West Grove, PA; 1:300 in PBS-T/1% BSA). After washing, the blot was scanned at 600 V using red fluorescence in a Storm 860 PhosphorImager (Molecular Dynamics, Sunnyvale, CA). The density of the heterodimeric RT bands was within the linear range of the instrument and a plot of fluorescence intensity versus nanograms of the RT standard yielded a straight line. Values of the unknown wild-type and mutant RT concentrations were determined from the linear regression analysis by interpolation. This experiment was performed twice and the average values used for the specific activity studies.

Specific activity determination. The polymerase activity of 1-3 ng of each recombinant RT preparation was evaluated in duplicate using a synthetic poly(rA)/p(dT)_12-18 template/primer (Pharmacia Biotech). Each 50-µl reaction contained 15 µg/ml poly(rA)/p(dT)_12-18, 10 nM DTT, 50 mM Tris·HCl, pH 6.8, 60 mM KCl, 1 mM EDTA, and 10 mM MgCl₂. The reactions were begun by adding [-³²P]dTTP (500 Ci/mmol, Amersham) to a final concentration of 180 nM. Aliquots of 15 µl were removed at 5, 10, and 20 min and applied directly onto Whatman 3-mm filter paper disks. The disks were washed three times for 10 min. each time in 5% trichloroacetic acid/1% sodium pyrophosphate, twice in 95% ethanol, then dried and counted in liquid scintillation fluid (Ready Safe; Beckman, Palo Alto, CA). The incorporated [-³²P]dTTP was plotted as cpm versus time and the initial velocities determined from the slopes of the linear regression analyses. All values are presented as a percentage of the initial velocity of the wild-type recombinant RT with the percentage standard deviation of the duplicate samples also indicated.

Processivity assays. Heteropolymeric HIV RNA template was prepared from a XhoI linearized pHIV-PBS plasmid using the Ribo Max kit (Promega, Madison, WI) according to the manufacturer's instructions. pHIV-PBS contains the 970-bp BglII-SphI fragment of the HXB2D molecular clone of HIV-1, corresponding to nucleotides 472-1442, and includes the R, U5 and 5' gag portions of the genome. This plasmid is similar to that used by Arts et al. (1994). A second RNA template contained the truncated A/T rich 5'-untranslated RNA4 region of the alfalfa mosaic virus together with the 5'-end of the coding sequence from HCMV DNA polymerase. The 600-base RNA template was prepared by transcription of SpeI linearized pUL54-4 plasmid (Cihlar et al., 1997). Double-stranded template/primer was prepared in batch by incubating the RNA templates (200 nM) with a DNA oligonucleotide primer (400 nM) for 10 min at 85°, then 10 min at 55° for annealing. The sequence for the HIV-1 oligonucleotide was 5'-GTC CCT GTT CGG GCG CCA-3' and corresponded to the natural tRNA primer binding site. The HCMV pol oligonucleotide sequence was 5'-CCG CGA CCG CAC CGC CGG TCA-3'. The homopolymeric poly(rA)/oligo(dT)₁₈ template was prepared by annealing 25 nM poly(rA) (6000-bp average length; Boehringer/Mannheim, Indianapolis, IN) with 400 nM oligo(dT)₁₈, resulting in a dT primer for approximately every 375 bases of rA template. The processivity assays were carried out essentially as described by Arion et al. (1996). Briefly, 2 ng of the wild-type or mutant RT was preincubated at 37° for 15 min with 5 pmol of heteropolymeric or 2 pmol of homopolymeric template/primer (calculated in moles of primer) in a 10 nM DTT, 50 mM Tris·HCl, pH 6.8, 60 mM KCl, 1 mM EDTA, and 10 mM MgCl₂ reaction buffer. The 50-µl reactions were begun by adding the dNTPs with or without a quenching template/primer at a final concentration of 50 µM dATP, dCTP, and dGTP; 180 nM [-³²P]dTTP (500 Ci/mmol) and 33 µg/ml poly(rC)/p(dG)_12-18 (Pharmacia Biotech). This concentration of the quenching poly(rC)/p(dG)_12-18 is in 18-fold molar excess to the heteropolymeric template/primer. For the poly(rA)/oligo(dT)₁₈ processivity reactions, only 16.7 nM [-³²P]dTTP (500 Ci/mmol) and 33 µg/ml poly(rC)/p(dG)_12-18 was added. After 1 hr at 37°, the total incorporation was assessed in a filter-based assay as described above and the remainder of the reaction was stopped by adding a 4× formamide loading buffer and heating for 2 min at 95°. Samples were electrophoresed in a 7 M urea-6% polyacrylamide gel. The image was visualized and quantified by PhosphorImager analysis using the Storm 860 and Image Quant analysis software. Line profiles drawn down the lane were used to quantify all the product bands with the HIV RNA template and the poly(rA)/oligo(dT)₁₈ template, whereas box/volume analysis was utilized for the single bands generated from the HCMV RNA template. The efficiency of quenching, as determined by the addition of the quenching template/primer before the test template/primer, was >93%.

	Results

Top Summary Introduction Materials & Methods Results Discussion References

Recombinant enzyme K_m/K_i analysis. The K_m values for three dNTPs and the K_i values for a panel of RT inhibitors were determined for wild-type E. coli-expressed recombinant HIV-1 RT using activated calf thymus DNA as template/primer (Table 1). The K_i/K_m ratios for each inhibitor, indicative of their relative inhibitory capacity, were similar to our previously published data using virion-associated wild-type RT (Cherrington et al., 1996a). Therefore, E. coli-produced recombinant HIV-1 RT expressing the site-directed mutations K65R, K70E, and M184V were used for the subsequent enzymatic studies. In addition to the in vitro PMEA-selected K65R and K70E mutations, the M184V mutation was included in this analysis as it demonstrates notable 3TC resistance and has been shown to be deficient in processivity (Back et al., 1996). As shown in Table 1, the K_m values for dATP, dCTP, and dTTP substrates were not significantly different among all four enzymes. The K_m results shown here with the K65R and M184V mutants are in agreement with results published previously that described the use of an RNA template (Gu et al., 1994b; Ueno and Mitsuya, 1997). Seven different inhibitors were analyzed with the four enzymes and the corresponding K_i values were calculated (Table 1). In agreement with previous results, increased K_i values for most inhibitors with the K65R mutant were observed (Gu et al., 1994b). In contrast, the K70E mutant showed only slight increases in K_i values for PMEApp and 3TCTP. The M184V mutant exhibited a strong and selective increase in K_i for 3TCTP, as expected from previous observations (Quan et al., 1996). Kinetic analyses using a heteropolymeric RNA template with PMEApp and 3TCTP as inhibitors were also performed. In these studies, the increases in K_i values with PMEApp were quite similar to those observed with the DNA template. However, these increases were more pronounced with 3TCTP on the RNA template for the K65R RT (3.5-fold) and the M184V RT (35-fold) compared with wild-type. Thus, with regard to the K70E mutant, no significant changes in affinity for the natural dNTP substrates and minor decreases in affinity for the inhibitors PMEApp and 3TCTP were observed.

Specific activity analysis of recombinant enzymes. To determine the specific activity of the recombinant enzyme preparations, quantitative immunoblot analyses were performed as described in Materials and Methods. The RT concentrations ranged from 1-17 ng/µl in the wild-type and various mutant preparations. The DNA polymerase activity was measured using a synthetic poly(rA)/(dT)_12-18 template/primer over a 20-min initial rate reaction. The calculated initial velocities were then divided by the concentration of enzyme used in the assay to determine the specific activity of the recombinant RT preparations (Fig. 1). All three mutant enzymes were significantly impaired in their specific activity compared with wild-type enzyme, with M184V exhibiting only 35% of wild-type activity and the K70E exhibiting 70% of wild-type activity. In agreement with these results, a second preparation of the K70E RT (K70E #2) containing a 17-fold higher enzyme concentration was tested; it also exhibited only 74% of wild-type activity. Interestingly, the K65R enzyme also showed diminished specific activity, 57% of wild-type.

View larger version (45K): [in this window] [in a new window]   Fig. 1.   Specific activity of recombinant enzymes. RNA-dependent DNA polymerase activity was assessed using a synthetic poly(rA)/p(dT)12-18 template/primer for wild-type RT and the M184V, K70E and K65R mutants of RT. Two independent E. coli preparations of the K70E mutant (#1 and #2) having a 17-fold difference in enzyme concentration were analyzed. All specific activities are expressed as a percentage of wild-type specific activity, as measured in duplicate, with the corresponding percent standard deviations indicated.

Processivity of recombinant enzymes on heteropolymeric RNA templates. Reverse transcription by HIV-1 RT proceeds with multiple pauses and repeated cycles of association/dissociation of the enzyme from the template/primer (Klarmann et al., 1993). To assess the DNA polymerase activity of RT in a single cycle of processivity, an enzyme activity assay was performed in the presence of an excess of a quenching template/primer that cannot incorporate labeled dNTP (Arion et al., 1996). The single-cycle processivity of wild-type, M184V, K65R, and K70E RT was measured using two different heteropolymeric RNA templates, one derived from HIV-1, which generates the (-) strand strong-stop DNA, and the other derived from the HCMV pol gene. In both cases, a DNA oligonucleotide was used as a primer and the products of the reactions were separated by electrophoresis. As shown in Fig. 2A, RT processivity on the HIV RNA template was highly abortive, with a ladder of bands punctuated by preferential bands. However, full-length product of 191 nucleotides was achieved in a minority of reverse transcripts. To normalize for total enzyme activity, each band in each lane of the gel was quantified using a PhosphorImager. In this way, the proportion of product achieving full-length size relative to the sum total of all the product bands was determined and this value ranged from 1.3 to 3.1% for the various enzymes. Results from such analyses were averaged from three experiments and are presented in Fig. 3A as the percentage of wild-type. These results demonstrated that the M184V mutant was notably impaired in processivity, the K70E mutant slightly impaired, and the K65R mutant slightly enhanced in single-cycle processivity.

View larger version (42K): [in this window] [in a new window]   Fig. 2.   Single-cycle processivity of recombinant enzymes using a heteropolymeric HIV RNA (A) or heteropolymeric HCMV RNA (B) template. The resulting single-cycle polymerase products from reactions containing the wild-type RT, M184V, K65R, and two independent preparations of K70E mutant RT (#1 & #2) are shown. Processivity reactions utilized an excess of unlabeled quenching template/primer to assure single-cycle processivity in the labeled products. Full-length product for the HIV RNA template is 191 nucleotides (A) and for the HCMV RNA template is 98 nucleotides (B). Extended primer lengths were determined by simultaneously electrophoresing dideoxy chain terminating DNA sequencing reactions.

View larger version (27K): [in this window] [in a new window]   Fig. 3.   Activity normalized single-cycle processivity results. A, As described in the results section, the quantity of full-length, 191-nucleotide product in Fig. 2A was divided by the sum of all products in that lane to determine the percent full-length product, thereby internally normalizing for enzyme activity. For the HIV RNA template, the M184V, K70E, and K65R mutants processed to full-length product at 42%, 80%, and 106% of wild-type, respectively. B, For the HCMV RNA template, to normalize for enzyme activity, the quantity of full-length, 98-nucleotide product in Fig. 2B was divided by the total quantity of unquenched product, as determined in parallel reactions performed in the absence of quenching template/primer (not shown). For this template, the M184V, K70E, and K65R mutants of RT processed to full-length product at 49%, 84%, and 114% of wild-type RT, respectively. All values are expressed as a percentage of wild-type RT processivity, with the two K70E preparations averaged together, and the percent standard deviation from three separate experiments indicated. The K70E mutant RT #1 processed slightly better than K70E #2 preparation. However, their combined values averaged over the three independent experiments as shown here demonstrated a significant decrease in processivity compared with wild-type, as indicated by the standard error bars.

Processivity of the K70E mutant on a poly(rA)/ oligo(dT)₁₈ template. Analyses of single-cycle processivity were performed with a homopolymeric poly(rA)/oligo(dT)₁₈ template/primer under conditions of limited dNTP to quantify extension length differences in processivity. The products of this primer extension assay were separated on a DNA sequencing gel and quantified by PhosphorImager analysis. The distribution of the cDNA product lengths for the wild-type and K70E RT mutant are shown in Fig. 4. The median cDNA length for these distributions was 25.5 nucleotides for the wild-type RT and 22 nucleotides for the K70E RT mutant. This median decrease of 3.5 nucleotides for the K70E mutant is similar to the reported decrease of 2-10 nucleotides for the M184V mutant under similar experimental conditions (Back et al., 1996). These analyses using a synthetic homopolymeric poly(rA)/oligo(dT)₁₈ template/primer agree with the observations using the heteropolymeric RNA template/primers and demonstrate a reduced processivity for the K70E RT mutant.

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Single-cycle processivity using a homopolymeric poly(rA)/(dT)18 template. The products of a primer extension processivity assay for the wild-type and K70E mutant RT (#1) were quantified with a PhosphorImager and normalized to 100%. The percentages of cDNA product at the various extension lengths, ranging from 19 to 50 nucleotides, are shown. The median extension length for the wild-type and the K70E mutant were 25.5 and 22 nucleotides, respectively. Similar results were observed in three independent experiments under the same conditions.

	Discussion

Top Summary Introduction Materials & Methods Results Discussion References

The K70E mutant of HIV-1 RT has been selected in vitro by PMEA and has been observed to develop in two of 29 patients undergoing extended therapy in a phase I/II clinical trial of adefovir dipivoxil (Mulato et al., 1998). HIV-1 recombinant viruses containing this mutation have also been shown to result in diminished sensitivity to PMEA and 3TC in vitro (Cherrington et al., 1996b). The enzymatic analyses using recombinant K70E RT presented here showed minor decreases in affinity for PMEApp and 3TCTP, with K_i values increasing by approximately 2.5-fold using a DNA or RNA template. The magnitudes of these increased K_i values are in agreement with the moderate decreases in drug sensitivities observed with the K70E recombinant virus in cell culture. The specific activity of K70E RT was also reduced compared with wild-type. Finally, in single-cycle processivity assays, the K70E RT demonstrated moderately reduced processivity compared with wild-type RT. These results support the published observation of the reduced in vitro replication capacity of HIV-1 expressing the K70E RT mutation (Cherrington et al., 1996b). Interestingly, an AZT-associated mutation at the same amino acid, K70R, is reported to not alter the replication capacity of HIV expressing K70R (Sharma and Crumpacker, 1997), suggesting an amino acid specificity for this phenotype.

A correlation between processivity and replication capacity has also been observed for the 3TC-associated M184V RT mutation (Boyer and Hughes, 1995; Back et al., 1996). In our experiments, the M184V mutant again demonstrated significant impairment in processivity. However, despite the demonstrated in vitro replication deficiency (Back et al., 1996), the M184V RT mutant HIV-1 is readily selected for in patients after 3TC treatment begins and is associated with a gradual return toward baseline viral loads (Schuurman et al., 1995; Wainberg et al., 1995). Thus, under the in vivo selective pressure of 3TC treatment, the replication deficient M184V virus is more fit than wild-type HIV-1. The replication impairment, however, may contribute to the gradual character of the increase in viral load. This slow return toward baseline viral load contrasts with resistance to the non-nucleoside RTI nevirapine, which is marked by a sharp return toward baseline viral loads (de Jong et al., 1997). In the case of the K70E mutation, both the replication and processivity impairment, as well as the reduction in susceptibility to PMEA, are less notable than with the M184V mutation and 3TC. These results suggest that the K70E mutant of HIV-1 might establish itself quite slowly in response to adefovir dipivoxil treatment, consistent with our limited clinical observations.

The observation presented here of slightly increased processivity for the K65R mutant of HIV-1 RT confirms an earlier observation by Arion et al. (1996). These authors suggested that the increased processivity of the K65R mutant may be caused by reduced template/primer dissociation, hence better elongation. Our observation of the reduced specific activity of K65R in conjunction with its increased processivity also suggests altered association/dissociation characteristics. Thus, although the K65R mutant seems less catalytically active, enhanced template/primer binding characteristics may serve to promote single-cycle processivity. Interestingly, although the K65R mutant can develop in vivo in response to ddC or ddI therapy in a minority of the patients (Gu et al., 1994b; Zhang et al., 1994; Winters et al., 1997), the K65R mutation has not developed in any patient treated with adefovir dipivoxil to date. This is curious because the in vitro susceptibility of the K65R mutant to PMEA is more notably reduced than that of the K70E mutant to PMEA.

Of the two patients who developed the K70E mutation during extended adefovir dipivoxil therapy, one of the patients was undergoing monotherapy and both showed continued viral load suppression during treatment (Mulato et al., 1998). Although anecdotal, this is noteworthy, because viral load often rebounds once resistance mutations can be defined in plasma-derived virus (Schuurman et al., 1995; Zazzi et al., 1996). Possibly, the reduced RT processivity and specific activity, as well as the reduced in vitro replication capacity of the K70E mutant may, in vivo, balance the minor changes in K_i and IC₅₀ values for PMEA, resulting in continued drug effectiveness as measured by HIV-1 RNA levels in plasma. The larger ongoing clinical trials should more clearly establish the role of the K70E mutation, if any, in clinical resistance to adefovir dipivoxil therapy.