|Home | About | Journals | Submit | Contact Us | Français|
We report key mechanistic differences between the reverse transcriptases (RT) of human immunodeficiency virus type-1 (HIV-1) and of xenotropic murine leukemia virus-related virus (XMRV), a gammaretrovirus that can infect human cells. Steady and pre-steady state kinetics demonstrated that XMRV RT is significantly less efficient in DNA synthesis and in unblocking chain-terminated primers. Surface plasmon resonance experiments showed that the gammaretroviral enzyme has a remarkably higher dissociation rate (koff) from DNA, which also results in lower processivity than HIV-1 RT. Transient kinetics of mismatch incorporation revealed that XMRV RT has higher fidelity than HIV-1 RT. We identified RNA aptamers that potently inhibit XMRV, but not HIV-1 RT. XMRV RT is highly susceptible to some nucleoside RT inhibitors, including Translocation Deficient RT inhibitors, but not to non-nucleoside RT inhibitors. We demonstrated that XMRV RT mutants K103R and Q190M, which are equivalent to HIV-1 mutants that are resistant to tenofovir (K65R) and AZT (Q151M), are also resistant to the respective drugs, suggesting that XMRV can acquire resistance to these compounds through the decreased incorporation mechanism reported in HIV-1.
Xenotropic murine leukemia virus-related virus (XMRV) is a gammaretrovirus that was first identified in some prostate cancer tissues (1,2) While some subsequent reports confirmed the presence of XMRV in prostate cancer samples (3–6), several others found little or no evidence of the virus in patient samples (7–9). XMRV DNA was also reported in 67% of patients with chronic fatigue syndrome (CFS) (10), but several subsequent studies in Europe and the USA failed to identify XMRV DNA in CFS patients or healthy controls (11–15). Hence, the relevance of XMRV to human disease remains unclear (16) and have been challenged (17). Most recently, it has been reported that XMRV has been generated through recombination of two separate proviruses suggesting that the association of XMRV with human disease is due to contamination of human samples with virus originating from this recombination event (18). Nonetheless, as a retrovirus that can infect human cells, XMRV can be very helpful in advancing our understanding of the mechanisms of retroviral reverse transcription, inhibition and drug resistance. XMRV RT is similar to the Moloney murine leukemia virus (MoMLV) RT, which has been the subject of structural and biochemical studies (19–24). Most of the differences between these gammaretroviral enzymes are at the RNase H domain (Supplementary Figure S1). Comparisons of human immunodeficiency virus type-1 (HIV) RT with MoMLV RT have revealed structural and sequence differences (21). For example, HIV-1 RT is a heterodimer composed of two related subunits (25,26) [reviewed in (27,28)]. Its larger p66 subunit (~66kDa) contains both the polymerase and RNase H domains; the smaller p51 subunit, (~51kDa), is derived from the p66 subunit by proteolytic cleavage and its role is to provide structural support and optimize RT’s biochemical functions (29). In contrast, structural studies have demonstrated that MoMLV RT is a monomer of about 74kDa, although one study reported that it may form a homodimer during DNA synthesis (30). So far, there are no published biochemical or structural studies on XMRV RT. Hence, the present study on this enzyme and its comparison to related enzymes provides an excellent opportunity to advance our biochemical understanding of the mechanism of reverse transcription, its inhibition and drug resistance.
The plasmid pBSK-XMRV containing the coding sequence of XMRV RT from the VP62 clone (GenBank: DQ399707.1) was chemically synthesized and optimized for bacterial expression by Epoch Biolabs Inc (Missouri City, Texas, USA). The 2013bp XMRV RT sequence was amplified from pBSK-XMRVRT by PCR, using the forward and reverse primers 1 (all primer sequences are shown in Supplementary Table S1), resulting in NdeI and HindIII restriction sites. Drug resistant XMRV RT mutants Q190M and K103R (equivalent to HIV-1 Q151M RT and K65R) were generated by site-directed mutagenesis using forward and reverse primers 2 and 3. The digested amplicons were ligated into pET-28a (Novagen), resulting into a construct that expresses an N-terminal hexa-histidine tag. pET-28a-MRT encoding full-length wild-type MoMLV RT was provided by Dr M. Modak (New Jersey Medical School, Newark NJ, USA).
Expression and purification of MoMLV and XMRV RTs were carried out similarly to our previously published protocols (23,24). Briefly, RTs were expressed in BL21-pLysS Escherichia coli (Invitrogen) grown at 37°C and induced with 150µM IPTG at OD600 0.8, followed by 16h growth at 17°C. A cell pellet from a 3l culture was incubated with 40ml lysis buffer (50mM Tris–HCl, pH 7.8, 500mM NaCl, 1mM PMSF, 0.1% NP-40, 1% sucrose and 2mg/ml lysozyme), then sonicated and centrifuged at 15,000g for 30min. The supernatant was diluted 2-fold in Buffer A (50mM Tri–HCl pH 7.8, 1mM PMSF, 4% streptomycin sulfate and 10% sucrose), stirred on ice for 30min and centrifuged. The supernatant was loaded on a Ni-NTA column and bound proteins were washed with 20ml Buffer B (20mM Tris–HCl pH 7.5, 500mM NaCl) and 5mM imidazole, followed by 20ml Buffer B with 75mM imidazole. RT was eluted in 2ml fractions with 20ml buffer B containing 300mM imidazole. Fractions with RT were pooled and further purified by size exclusion chromatography (Superdex 75; GE Healthcare). RTs (>95% pure) were stored in 50mM Tris–HCl pH 7.0, 100mM NaCl, 1mM DTT, 0.1% NP-40 and 30% glycerol in 10μl aliquots at −20°C. Protein concentrations were determined by measuring UV280 (molar extinction coefficients of 106 and 103M−1cm−1 for XMRV and MoMLV RT).
HIV-1 RT was cloned in a pETduo vector and purified as described previously (29,31,32). Oligonucleotide sequences (IDT-Coralville, IA, USA) of DNA/RNA substrates are shown in Supplementary Table S1. Nucleotides were purchased from Fermentas (Glen Burnie, MD, USA). They were treated with inorganic pyrophosphatase (Roche Diagnostics, Mannheim, Germany) as described previously (33) to remove PPi that might interfere with excision assays.
Steady state parameters Km and kcat for dATP incorporation were determined using single nucleotide incorporation gel-based assays. XMRV RT and MoMLV RT reactions were carried out in 50mM Tris–HCl pH 7.8, 60mM KCl, 0.1mM DTT, 0.01% NP-40 and 0.01% bovine serum albumin (BSA) (Reaction Buffer) with 6mM MgCl2 or 1.5mM MnCl2, 0.5mM EDTA, 200nM or 100nM Td26/5′-Cy3-Pd18b, 20nM or 5nM RT for XMRV and MoMLV RTs, respectively and varying concentrations of dNTP in a final volume of 10µl. The reactions for HIV-1 RT were carried out in Reaction Buffer with 100nM Td26/5′-Cy3-Pd18b, 10nM HIV-1 RT and 6mM MgCl2 in a 20µl reaction. All the concentrations mentioned here and in subsequent assays reflect final concentration of reactants otherwise mentioned reactions were stopped after 15min for XMRV, 4min for MoMLV RT, and 2.5min for HIV-1. The products were resolved on 15% polyacrylamide–7M urea gels. The gels were scanned with a Fuji Fla-5000 PhosphorImager (Stamford, CT, USA) and the bands were quantified using MultiGauge. Results were plotted using GraphPad Prism 4. Km and kcat were determined graphically using Michaelis–Menten equation.
Formation of RT-DNA binary complex: 20nM Td31/5′-Cy3-Pd18a (Supplementary Table S1) was incubated for 10minutes with increasing amounts of MoMLV or XMRV RT in 50mM Tris–HCl pH 7.8, 0.01% BSA, 5mM MgCl2 and 10% (v/v) sucrose. The complexes were resolved on native 6% polyacrylamide 50mM Tris–borate gel and visualized as described above.
Active site concentrations and kinetic constants of DNA binding for XMRV, HIV-1 and MoMLV RTs were determined using pre-steady state experiments. Reactions with XMRV and MoMLV RTs were carried out in the reaction buffers listed above. For XMRV RT 100nM protein was pre-incubated with increasing concentrations of Td31/5′-Cy3-Pd18a, followed by rapid mixing with a reaction mixture containing 5mM MgCl2 and 100µM next incoming nucleotide (dATP). The reactions were quenched at various times (5ms to 4s) by adding EDTA to a final concentration of 50mM. The amounts of 19-mer product were quantified and plotted against time. The data were fit to the following burst equation:
where A is the amplitude of the burst phase that represents the RT–DNA complex at the start of the reaction, kobs is the observed burst rate constant for dNTP incorporation, kss is the steady state rate constant and t is the reaction time. The rate constant of the linear phase (kcat) was estimated by dividing the slope of the linear phase by the enzyme concentration. The active site concentration and T/P binding affinity (KD.DNA) were determined by plotting the amplitude (A) against the concentration of T/P. Data were fit to the quadratic equation (Equation 2) using non-linear regression:
where KD is the dissociation constant for the RT–DNA complex, and [RT] is the concentration of active polymerase. HIV-1 RT’s DNA binding affinity was determined as previously described (29).
We used surface plasmon resonance (SPR) to measure the binding constants of XMRV and HIV-1 RTs to double-stranded DNA. Experiments were carried out using a Biacore T100 (GE Healthcare). To prepare the sensor chip surface we used the 5′-biotin-Td37/Pd25 oligonucleotide (Supplementary Table S1). One hundred and twenty RUs of this DNA duplex were bound in channel 2 of a streptavidin-coated sensor chip [Series S Sensor Chip SA (certified)] by flowing a solution of 0.1µM DNA at a flow rate of 10µl/min in a buffer containing 50mM Tris pH 7.8, 50mM NaCl. The binding constants were determined as follows: RT binding was observed by flowing solutions containing increasing concentrations of the enzyme (0.2, 0.5, 1, 2, 5, 10, 20, 50, 100 and 200nM) in 50mM Tris pH 7.8, 60mM KCl, 1mM DTT, 0.01% NP40 and 10mM MgCl2 in channels 1 (background) and 2 (test sample) at 30µl/min. The trace obtained in channel 1 was subtracted from the trace in channel 2 to obtain the binding signal of RT. This signal was analyzed using the Biacore T100 Evaluation software to determine KD.DNA, kon and koff.
The optimal nucleotide incorporation rates (kpol) were obtained by pre-steady state kinetics analysis using single nucleotide incorporation assays. A solution containing XMRV RT (150nM final concentration) and Td31/5′-Cy3-Pd18a (40nM) was rapidly mixed with a solution of MgCl2 (5mM) and varying dATP (5–200μM) for 0.1 to 6s before quenching with EDTA (50mM) (all concentrations in parentheses are final, unless otherwise stated). Products were resolved and quantified as described above. Burst phase incorporation rates and substrate affinities were obtained from fitting the data to Equation 1. Turnover rates (kpol), dNTP binding to the RT-DNA complex (Kd.dATP), and observed burst rates (kobs) were fit to the hyperbolic equation:
HIV-1 RT’s DNA binding affinity was determined as previously described (29).
The fidelity (error-proneness) of XMRV RT was determined and compared with that of MoMLV RT and HIV-1 RT by primer extension assays using 10nM heteropolymeric Td100/5′-Cy3-Pd18a. Reactions (10µl) were carried out in Reaction Buffer containing all four dNTPs (100μM each) or only three dNTPs (missing either dATP, dGTP or dTTP) at 100μM each. Incubations of the XMRV and MoMLV (50nM) reactions were at 37°C for 45min and 30min for HIV-1 RT (20nM). Reactions were initiated by adding dNTPs, stopped with equal volume of formamide-bromophenol blue, and an aliquot was run on a 16% polyacrylamide–7M urea gel.
For these experiments, instead of including the next correct nucleotide (dATP) in the polymerase reactions, we used dTTP as the mismatched incoming nucleotide. Hence, 50nM XMRV RT was pre-incubated with 35nM Td31/5′-Cy3-Pd18a in reaction mixture. Reactions were initiated by adding dTTP (5–750µM) and 5mM MgCl2, followed by incubation (37°C) for 5min, due to the decreased mismatch incorporation rate of XMRV. For MoMLV RT, 30nM RT and 20nM DNA used and the reactions were carried out for 2.5minutes. For HIV-1, 30nM RT, 20nM DNA and 0–200µM nucleotide were used and the reactions were carried out for 2.5min. The amount of extended primer was quantified and plotted against the concentration of dTTP. The data were used to derive the Kd.dNTP of incorrect nucleotide binding, the rate kpol (using Equations 1 and 3) and the efficiency of the misincorporation reaction (kpol/Kd.dTTP).
ANGIE P cells, which contain a retroviral vector (GA-1) that encodes a bacterial β-galactosidase gene (lacZ) and a neomycin phosphotransferase gene, were plated (5×106 cells/100mm dish) and after 24h were transfected using the calcium phosphate precipitation method with a plasmid expressing either XMRV or amphotropic MLV (AM-MLV) (three independent transfections per vector). After 48h, the culture medium with XMRV or (AM-MLV) was harvested, serially diluted and used to infect D17 target cells (2×105 cells/60mm dish) in the presence of polybrene. The infected D17 cells were selected for resistance to G418 (400µg/ml) in the presence of 1µM AZT to suppress reinfection, and characterized by staining with 5-bromo-4-chloro-3-indoyl-β-d-galacto-pyranoside (X-Gal) ~2 weeks after G418 selection. The frequencies of inactivating mutations in lacZ quantified as described before (blue versus white colonies) (34).
Processivity reactions were carried out in Reaction Buffer containing 20nM Td100/Pd18, 100μMof each dNTP, 30nM HIV-1 RT, 50nM MoMLV RT or 100nM XMRV RT and 1μg/μl unlabeled calf thymus DNA trap in 50μL. Enzymes were pre-incubated with Td100/Pd18 for 1min before adding dNTPs (100µM each) together with the calf thymus DNA trap. Reactions were incubated at 37°C, and 10μl aliquots were taken out at 3, 7.5 and 15min for HIV-1 RT or at 7.5, 15 and 30min for XMRV RT and MoMLV RT, and mixed with equal volume of loading dye. The effectiveness of the trap was determined by pre-incubating the enzyme with the trap before adding Td100/Pd18. Control DNA synthesis was measured in absence of trap under the same conditions. Reaction products were resolved as above.
Thirty nanomolar Td31/5′-Cy3-Pd18a was pre-incubated for 10min with 100nM XMRV or 50nM MoMLV RT in Reaction Buffer, then rapidly mixed with 100µM dNTPs, 5mM MgCl2 for varying times (0.1–45s) before quenching with EDTA (50mM final). Single turnover processivity of HIV-1 RT was assayed with 40nM enzyme, 20nM DNA and 50μM of each nucleotide were used. The reaction products were resolved and quantified as described above. The data were fit to a one-phase exponential decay equation for the elongation of the 18-mer primer. The rates of appearance and extension of products from subsequent nucleotide incorporations (19- and 27-mer) were obtained by fitting the intensities of corresponding bands to double exponential (Equation 4):
DNA synthesis by 50nM XMRV RT or MoMLV RT was carried out in Reaction Buffer using 20nM Td100/5′-Cy3-Pd18a, 2.5µM dNTP, 5mM MgCl2 and varying amounts of NRTI (0–100µM). Reactions were quenched with 95% formamide after 1h incubation at 37°C (38). In experiments with aptamers 10nM XMRV RT, 20nM Td31/5′-Cy3-Pd18a and 50µM dNTPs were used in the presence of varying amounts of aptamer for 30min (0–500nM for m.1.3; 0–25nM for m.1.4 and m.1.1FL). The inhibition of DNA polymerization was monitored by resolving the products on 15% polyacrylamide–7M urea gels and visualized as described above. Bands corresponding to full extension products were quantified using MultiGauge Software and IC50s were obtained from dose–response curves using GraphPad Prism.
The ability of enzymes to use PPi or ATP to unblock template-primers that had AZT-MP (T/PAZT-MP) or EFdA-MP (T/PEFdA-MP) at their 3′ primer ends was measured as follows: 20nM of T/PAZT-MP or T/PEFdA-MP were prepared as described before (32). They were incubated at 37°C with either 60nM HIV-1 RT or 200nM XMRV RT in the presence of 0.15mM PPi or 3.5mM ATP for PPi- or ATP-dependent rescue reactions, respectively. Reactions were initiated by the addition of MgCl2 (6mM). Aliquots were removed at different times (0–90min) and analyzed as above. Rescue assays were performed in the presence of 100µM dATP to prevent EFdA-MP reincorporation, 0.5µM dTTP, 10µM ddGTP and 10mM MgCl2.
The sequence of XMRV RT from the VP62 clone was aligned with that of MoMLV RT (PDB: 1RW3) (21,22) using ClustalW. To generate the homology model of XMRV RT, we used the Prime protocol of the Schrödinger software suite (Schrödinger Inc. NY). The resulting molecular model was further energy minimized by OPLS2005 force field using the Impact option of Schrödinger. The final model was validated with PROCHECK v.3.5.4.
The XMRV and MoMLV enzymes are closely related (~95% sequence identity) with most of the differences between them being in the RNase H domain (Supplementary Figure S1). While XMRV and MoMLV differ significantly from HIV-1 RT, the known polymerase motifs (A–F) are well conserved in all three enzymes (Supplementary Figure S1). Specifically, the active site aspartates in Motifs A and C (Figure 9) (D150, D224, D225 in XMRV RT; D150, D224, D225 in MoMLV RT; D110, D185, D186 in HIV-1 RT) are conserved in all three RTs. Also, the three enzymes are similar in Motif B, which is involved in dNTP binding and multidrug resistance (AZT and dideoxy-nucleoside drugs) through the decreased incorporation mechanism (27,39–41). Specifically, all three enzymes have a glutamine at the start of this motif (Q151 in HIV-1 RT, Q190 in XMRV RT and Q190 in MoMLV RT). Motif D includes HIV-1 RT residues L210 and T215, which when mutated they enhance excision of AZT from the AZT-terminated primer terminus. This motif is mostly different in XMRV and MoMLV RTs, where the corresponding residues are N226 and A231 (Supplementary Figure S1). K219 of HIV-1 RT Motif D is proximal to the dNTP-binding pocket and is also conserved in the other enzymes (K235). The DNA primer grip (Motif E) (36,42) in HIV-1 RT (M230G231Y232) is slightly different in the gammaretroviral enzymes (L245G246Y247). Motif F at the fingers subdomain of all enzymes has two conserved lysines that bind the triphosphate of the dNTP (K65 and K72 in HIV-1 RT; K103 and K110 in XMRV and MoMLV RTs).
Several HIV-1 residues involved in NRTI resistance have the resistance mutations in XMRV and MoMLV RTs (Table 1). Hence, XMRV and MoMLV RTs have a Val as the X residue (codon 223) of the conserved YXDD sequence of Motif C. An M184V mutation at this position in HIV-1 RT causes strong, steric hindrance-based, resistance to 3TC and FTC (43–45), and to a lesser extent to ddI, ABC [reviewed in (46)], and translocation defective RT inhibitors (TDRTIs) (43) (Table 1). Similarly, the M41L mutation, which causes excision-based AZT resistance in HIV is already present in XMRV and MoMLV RT (L81, Table 1). The gammaretroviral enzymes differ from HIV-1 RT in several other HIV drug resistance sites (HIV residues 62, 67, 69, 70, 75, 77, 115, 210, 215) (Table 1). Finally, there are also differences in residues that are essential for NNRTI binding in HIV-1 RT: W229 changes to Y268 in XMRV RT, Y181 to L220, Y188 to L227 and G190 to A229 (Table 1) (27,28,47–49).
The sequence coding for full-length XMRV RT from the VP-62 clone (NCBI RefSeq: NC_007815) (1) was optimized for expression in bacteria, synthesized by Epoch Biolabs and cloned as described in ‘Materials and Methods’ section. Both XMRV RT and MoMLV RT were tagged with a hexahistidine sequence at the N-terminus and expressed with a yield of ~2mg/l of culture. Purified enzymes (>95% pure, Supplementary Figure S2) were stored at −20°C. The presence of NP-40 or glycerol was critical for enzyme stability.
Initial polymerase activity assays using Td31/5′-Cy3-Pd18a displayed overall slower polymerase activity of XMRV RT compared to HIV-1 and MoMLV RTs. This observation led us to investigate the steady state nucleotide incorporation properties of XMRV RT using single nucleotide incorporation assays. The estimated values for kcat (19.9min−1 for HIV-1 RT (32), 3.3min−1 for MoMLV RT, 0.6min−1 for XMRV RT) and Km.dNTP (0.07µM for HIV-1 RT (32), 3.3μM for MoMLV RT, 3.0µM for XMRV RT) show that XMRV RT has a drastically reduced efficacy (kcat/Km.dNTP) at nucleotide incorporation, compared to both MoMLV and HIV-1 RTs.
To assess if the efficiency of XMRV RT was also affected by a lower DNA binding affinity we measured the DNA binding affinity of the enzymes using three methods: gel-mobility shift assays, pre-steady state kinetics and SPR. Gel-mobility shift assays showed that the KD.DNA for XMRV RT was marginally higher than that for HIV-1 RT and MoMLV RT (data not shown) (50) suggesting weaker binding to DNA.
Pre-steady state kinetics allows estimation of the fraction of active polymerase sites as well as the KD.DNA value for the enzyme. The amplitudes of DNA extensions using XMRV RT and/or MoMLV RT at varying DNA concentrations were plotted against the DNA concentration and the data were fit to the quadratic equation (Equation 2), yielding a KD.DNA of 33nM for XMRV RT, 19nM for MoMLV RT (Table 2) and 12.5nM for HIV-1 RT (32). These values did not change significantly when tested with DNA of different lengths (data not shown). Hence, the transient kinetic experiments confirmed the findings of the gel-mobility shift assays showing XMRV RT to have lower DNA binding affinity than HIV-1 RT.
Measurements of KD.DNA using gel-mobility shift assays and pre-steady state kinetic methods do not offer insights regarding the kinetics of binding and release of nucleic acid from the viral polymerases. Hence, we used SPR to measure directly DNA binding and the DNA dissociation components of the KD.DNA. We attached on the SPR chip a nucleic acid biotinylated at the 5′ template end and immobilized it on a streptavidin sensor chip. Various concentrations of either XMRV or HIV-1 RT were flowed over the chip to measure the association (kon) and dissociation (koff) rates of the enzymes in real time (Figure 1). HIV-1 RT had considerably slower dissociation rates than XMRV RT, and longer dissociation phases were needed to obtain reliable values.
Several methods were tested to best fit our data. The ‘heterogeneous ligand’ method gave the best fit for both XMRV and HIV-1 RT. In this model the x2 values for DNA binding to XMRV and HIV-1 RT were 9.3 RU2 and 48.1 RU2, respectively, compared to 15.1 RU2 and 152 RU2 when we tried fitting the data in a ‘homogeneous ligand’ model. The former model assumes that RT binds DNA in two different modes and provides two association (kon) and two dissociation constants (koff).
Our data show that XMRV RT has a slightly faster rate of association (kon) than HIV-1 RT. We measured two kon values of 7.3×106M−1s−1 and 8.2×104M−1s−1 for XMRV RT versus 7.6×105M−1s−1 and 1.2×106M−1s−1 for HIV-1 RT. Interestingly, the dissociation rate of XMRV RT was significantly faster than that of HIV-1 RT (0.28s−1 and 0.0045s−1 for XMRV RT and 7.8×10−4 s−1and 0.0076s−1 for HIV-1 RT) (Table 3). This difference in dissociation rate resulted in a KD.DNA at least 1 order of magnitude higher for XMRV RT compared to HIV-1 RT (38 and 54nM versus 1.0 and 6.1nM for XMRV and HIV-1 RT, respectively) (Table 3).
A transient-state kinetics approach was used to estimate the dNTP binding affinity (Kd.dNTP) and maximum nucleotide incorporation rate (kpol) (55). The rates at varying concentrations of next incoming nucleotide (dATP) were determined by plotting the amount of extended primer as a function of time. The rates were then plotted against dATP concentration. The data were fit to a hyperbola (Equation 3). The Kd.dNTP for XMRV RT is 26.6μM and the kpol is 8.9s−1 (Figure 2) (Table 2). Under similar conditions the Kd.dNTP and kpol were 1.3µM and 24.4s−1 for HIV-1 RT (29) and 25µM and 14.1s−1 for MoMLV RT.
To assess whether XMRV RT displays high nucleotide incorporation fidelity we monitored the incorporation of three dNTPs by XMRV RT and compared with HIV-1 RT (52). The results of fidelity assay are shown in Figure 3. The lanes marked ‘4dNTPs’ for all enzymes represent the DNA synthesis using a Td100/5′-Cy3-Pd18a template-primer in the presence of all four dNTPs. The subsequent lanes, marked ‘-dNTP’, correspond to the synthesis of DNA in the absence of that specific deoxynucleotide triphosphate. The comparison of the DNA synthesis in the absence of one nucleotide by HIV-1 RT, MoMLV RT and XMRV RT shows that HIV-1 and MoMLV RTs were able to misincorporate and extend the primer beyond the missing nucleotide more efficiently than XMRV RT, suggesting that the latter is a less error prone DNA polymerase. It should be noted that the higher fidelity of XMRV is not the result of measuring a smaller number of errors because of the decreased replication rate, as the assay conditions were optimized to allow production of the same amount of full length product in the presence of all four dNTPs for and MoMLV RTs. To further investigate the fidelity of DNA synthesis by XMRV RT, the kinetics of mismatch nucleotide incorporation were carried out in a quantitative manner by monitoring the incorporation of single mismatched nucleotide under pre-steady state conditions. The estimated KD.dTTP (mismatch) and kpol values show that XMRV RT has a lower affinity for a mismatched nucleotide but comparable turnover number than MoMLV RT, suggesting that the observed higher fidelity over MoMLV RT is due to differences during the nucleotide-binding step (Table 4). However, compared to HIV-1 RT, XMRV RT has decreased both affinity and incorporation rate, suggesting that its higher fidelity is the result of both decreased binding of mismatched nucleotides and slow rate of incorporation.
The ANGIE P cells used for this assay are a D17-based encapsidating cell line and contain an MLV-based retroviral vector (GA-1), which encodes a bacterial β-galactosidase gene (lacZ) and a neomycin phosphotransferase gene (neo). Replication fidelity is a measure of the frequency of lacZ inactivation and was determined by measuring lacZ non-expressing white colonies. The results show that the number of white colonies was not statistically different in the case of XMRV as compared to AM-MLV, suggesting that under these conditions the fidelity of XMRV is not significantly different than that of AM-MLV (Figure 4).
Processivity is the probability of translocation of a polymerase along a template and predicts the number of cycles of nucleotide incorporation during one productive enzyme–DNA binding event. We assessed XMRV RT’s processivity of DNA synthesis in comparison to HIV and MoMLV RTs using both a gel-based trap assay and a quantitative pre-steady state assay. In the gel-based assay, the enzymes were pre-incubated with template-primer, then the reaction was initiated by the addition of all four nucleotides together with calf thymus DNA, which was used as a trap to bind free enzyme dissociated from the substrate during the course of the reaction (38). The length of the DNA product is an inverse measure of termination probability, as previously described. As a control, we used lanes where no trap was present; establishing that the same amount of total polymerase activity (processive and non-processive) is provided in all cases. The results indicate that XMRV RT is less processive than HIV-1 and MoMLV RTs with shorter DNA product after 30min of reaction in the presence of trap (Figure 5).
To measure processivity quantitatively we applied a single turnover processivity assay developed by Patel et al. (35) (Figure 6). In this assay, the rates of consecutive nucleotide incorporations under single turnover conditions are monitored. The rate of elongation incorporation (k1) and the rate of processive DNA synthesis (k2) (Equation 4) were calculated at several template positions for each enzyme. The ratio of the rate of processive DNA synthesis to the rate of nucleotide incorporation (k2/k1) is referred to as the processivity index (35). The absolute values of these constants for HIV-1 RT, XMRV and MoMLV RT and their ratios are collected in Table 5. XMRV RT is clearly the least processive for each extension product. The difference in processivity varies significantly depending on sequence or sequence context (decrease in processivity from 3-fold up to 10-fold). While the current data do not allow generalization of rules for pausing at specific sites, this clearly shows consistently that XMRV is not as efficient as MoMLV RT in polymerizing processively through ‘difficult spots’.
Previous studies have shown that XMRV is inhibited by some antivirals (53–56). However, the susceptibility of XMRV RT has not been tested against a wide variety of nucleoside RT inhibitors (NRTIs) that block replication by chain-terminating the primer, or by preventing translocation after their incorporation into the nascent DNA chain (TDRTIs) (32,57,58). In addition, the susceptibility of XMRV RT to non-nucleoside RT inhibitors (NNRTIs) or RNA aptamers that can be selected to block reverse transcriptases (59–63) has not been established.
Hence, we performed gel-based primer extension assays in the presence of various inhibitors. As shown in Table 6, most of the HIV-1 RT inhibitors also block XMRV RT with significantly varying IC50s. The most potent inhibitors tested were ENdA (4′-ethynyl-2-amino-2′-deoxyadenosine) followed by EFdA. EFdA was also potent at inhibiting wild-type XMRV replication in cell culture with an EC50 of 40nM from three independent experiments (standard error was 10nM).
Unlike HIV-1 RT, XMRV RT and MoMLV RT lack the two tyrosine residues (Y181 andY188 in HIV-1 RT) (Supplementary Figure S1) that are known to contribute to NNRTI binding. Hence, the gammaretroviral enzymes were not inhibited by the NNRTIs tested (TMC-125 and efavirenz) (Supplementary Figure S3).
We also tested XMRV RT’s susceptibility to three independent RNA aptamers that had been previously selected against MoMLV RT (60). The aptamers inhibited XMRV RT to varying extents with IC50s ranging from 2 to 52nM (Figure 7). Most notable was the m.1.1FL aptamer which gave IC50s of 2 and 4nM for XMRV RT (Figure 7) and MoMLV RT respectively, without inhibiting HIV-1 RT (data not shown). These inhibition assays utilized truncated forms of aptamers m.1.3 and m.1.4 lacking the original primer-binding segments of the aptamers, demonstrating that these 5′ and 3′ segments are not required.
A key mechanism of NRTI resistance in HIV-1 RT is based on inhibitor excision from the primer end, using a pyrophospholytic reaction (64,65). The pyrophosphate donor in vivo is likely to be ATP, although PPi can efficiently unblock NRTI-terminated primers. This excision activity is present in wild-type HIV-1 RT, and is enhanced in the presence of AZT-resistance mutations. We measured the ability of wild-type XMRV to unblock primers terminated with AZT or EFdA in the presence of PPi. We found that unlike HIV-1 RT that excised AZT-MP efficiently under these conditions, XMRV RT had considerably lower excision activity (Figure 8). Similar excision experiments where ATP was used instead of PPi showed that XMRV is very inefficient in ATP-based excision as compared to HIV-1 RT (data not shown).
The HIV-1 RT mutation Q151M confers resistance to AZT by enhancing discrimination of the nucleotide analog leading to its reduced incorporation (37,66–68). Another HIV-1 RT mutation, K65R, decreases susceptibility to tenofovir (69,70). Since AZT and tenofovir are potent inhibitors of XMRV (Table 6) (54–56), we wanted to investigate whether the XMRV RT mutant equivalents of HIV Q151M and K65R (XMRV Q190M and K103R) would confer XMRV RT resistance to AZT and tenofovir. We constructed these mutant clones and tested their susceptibility to AZT and tenofovir in the same manner as wild-type XMRV RT. Interestingly, Q190M XMRV RT has a decreased susceptibility to AZT (approximately 5-fold increase in the IC50). Similarly, the K103R XMRV RT mutant enzyme was less susceptible to tenofovir, increasing the IC50 by at least 2-fold.
Given the significant sequence similarity between XMRV and MoMLV RTs, the resulting homology model of XMRV RT is highly similar to MoMLV RT (>1.5Å rms) and of excellent quality. Since the input structure of MoMLV RT did not contain the RNase H domain of the enzyme, the XMRV RT model is also missing this domain. The molecular model of the polymerase domain of XMRV RT is shown in Figure 9. An alignment of the MoMLV RT crystal structure (22) with the XMRV RT homology model highlights the few changes in the polymerase domain of XMRV RT. These are L29 (P in MoMLV), Q234 (L in MoMLV), R238 (Q in MoMLV) and N422 (D in MoMLV). From these, residue 422 is located in the nucleic acid binding cleft and may contribute to differences in the interactions with nucleic acid substrate. However, most of the differences between the gammaretroviral enzymes are in their RNase H domains and also in the first 30N-terminal residues of the polymerase domain, for which we do not have structural information since they were not included in the original crystal structure of MoMLV RT. The differences between XMRV RT and HIV-1 RT are very significant. Unlike the HIV enzyme, XMRV RT appears to be a monomer in solution. Moreover, alignment of the HIV-1 RT–DNA complex with XMRV RT based on their active sites at the palm subdomains shows that the thumb subdomain of XMRV RT would have to be repositioned to be able to accommodate nucleic acid.
Early studies reported the presence of XMRV in stromal cells from prostate cancer patient samples and also in CFS clinical samples. Some of the subsequent studies confirmed these findings whereas several others failed to identify XMRV in prostate cancer or in CFS patients, even when same samples were used (71). It was recently reported that human sample contamination with mouse DNA can occur frequently (17,72–74). Moreover, two coauthors from this study have recently demonstrated that XMRV is the product of recombination events between two MLV proviruses, suggesting that XMRV may not be relevant to human disease (18). Nonetheless, XMRV is still an important human retrovirus and comparisons with HIV can provide valuable insights into the fundamental mechanisms of DNA polymerization, RT inhibition and drug resistance. (75).
There is high degree of sequence similarity between the XMRV and MoMLV RTs (95% amino acid identity), and much less so with HIV-1 RT. Based on gel filtration experiments we conclude that unlike HIV-1 RT, but similar to MoMLV RT, XMRV RT exists in solution primarily as a monomer. We also included comparisons with HIV-1 RT in this study as it has been extensively studied and provides an excellent frame of reference.
We report here that there are significant differences in the DNA polymerization efficiency of the three enzymes. Although the polymerase active sites of the XMRV and MoMLV enzymes are almost identical, there is a considerable decrease in the efficiency of nucleotide incorporation by XMRV RT. Most differences in sequence are at the RNase H domain and are likely to affect polymerization by changing the positioning of DNA at the polymerase active site.
We have recently solved the crystal structure of the XMRV RNase H at high resolution (1.5Å) (pdb 3P1G) (Kirby, K.A. et al., submitted for publication). We observed major differences in affinity for nucleic acid that we determined with gel-mobility shift assays and with pre-steady state kinetics. SPR experiments dissected in more detail the specific defect of XMRV RT in binding DNA. Surprisingly, XMRV RT can associate very rapidly with DNA, even more so than HIV-1 RT (Figure 1 and Table 3). However, it dissociates from DNA much faster than the HIV enzyme, resulting in an overall reduced binding affinity. A possible reason for the fast association and dissociation rates of XMRV RT may be the apparent monomeric state, which might offer facile access to the nucleic acid binding cleft, although with less contacts and lower affinity than HIV-1 RT, which is a heterodimer (76,77). This high rate of XMRV RT dissociation from DNA likely contributes to the decreased processivity observed in our study, and may have consequences in the recombination rates of this virus.
Previous sequences of XMRV from prostate cancer tumors showed low variability, suggesting that the virus may have a high fidelity of replication (1,10). Our study demonstrated that HIV-1 RT and MoMLV RT incorporated mismatched nucleotides and extended past the mismatches more efficiently than XMRV RT. Pre-steady state kinetics established that the higher overall fidelity of XMRV RT over MoMLV RT is due to a lower affinity for mismatched nucleotides. When compared to HIV-1 RT, however, XMRV RT has differs in both the nucleotide binding and incorporation steps. Nonetheless, XMRV did not have higher fidelity than a related amphotropic MLV virus or HIV-1 in a cell-based assay. It is possible that the high dNTP concentration in dividing cells can suppress mismatching events. We have previously shown (39) that as nucleotide concentrations vary in different cell lines, this can affect viral susceptibility to NRTIs, and possibly in this case also incorporation of mismatched nucleotides. Additional cell-based studies using multiple cell lines and a large panel of viruses should provide a better understanding of the relation between in vivo and in vitro fidelity.
Early studies have reported susceptibility of XMRV to some antiretrovirals that have been used in the treatment of HIV infection (53–56). In those studies the compounds were tested at the virus level. To better understand the interactions of inhibitors at their RT target level we tested here the ability of these and several more compounds to block the polymerase activity of XMRV RT. We found that two TDRTIs, EFdA-TP and ENdA-TP were very potent RT inhibitors (IC50s: 0.43µM and 0.14µM, respectively). Unlike other NRTIs, these compounds have a 3′ OH group and are known to efficiently inhibit HIV replication by blocking translocation (32,58,78). Preliminary experiments demonstrated that they also block XMRV RT by the same mechanism (data not shown).
In HIV, moderate resistance to EFdA is conferred by the emergence of the M184V mutation at the conserved X position of the conserved YXDD motif of the polymerase active site. Interestingly, XMRV and MoMLV RTs already have a valine (V223) at this position. This difference is likely to contribute to the better potency of EFdA against HIV-1 RT than XMRV RT or MoMLV RT (57,58). It may also contribute to the decreased ability of XMRV RT to unblock chain-terminated primers, as was also reported for M184V HIV-1 RT (79) and to the enhanced fidelity reported here for XMRV RT, which is also reminiscent of the previously reported high fidelity of M184V HIV-1 RT (80,81). Nonetheless, despite the presence of a Val in the YMDD motif of XMRV RT we found EFdA to inhibit very efficiently replication-competent or pseudotyped XMRV, with submicromolar EC50s (40 and 110nM, respectively).
Previously, highly potent aptamers were selected to inhibit MoMLV RT (60). We demonstrate here that the three aptamers we tested have varying potency against XMRV RT. Aptamer m.1.1FL was the most potent inhibitor of XMRV RT and MoMLV RT in in vitro assays (IC50=2 and 4nM, respectively). The fact that XMRV and MoMLV RTs are inhibited by the same aptamers at comparable efficiencies suggests that the RT residues that are different in the two enzymes are not critical to the binding of the aptamer. In contrast, heterodimeric HIV-1 RT has a very different binding cleft and is not inhibited by these aptamers.
Tenofovir is an essential component of HIV therapies and is also a potent inhibitor of XMRV RT. HIV resistance to tenofovir is conferred by a single codon mutation (K65R). HIV-1 RT residue 65 is known to interact with the incoming dNTP or the activated tenofovir analog (tenofovir diphosphate) (82). K65R causes resistance to tenofovir by lowering the kpol for the incorporation of the inhibitor into the nascent DNA. We prepared XMRV RT with the equivalent mutation, K103R, and determined that it has decreased susceptibility to tenofovir. Hence, it is possible for XMRV to develop tenofovir resistance through the same mechanism as HIV-1 RT. HIV resistance to AZT can occur by either decreased binding/incorporation or increased excision of the chain-terminating NRTI (33,83). HIV-1 RTs containing the M41L, D67N, K70R, T215Y/F, K219E/Q mutations show enhanced removal of AZT. Our experiments show that unlike wild-type HIV-1 RT, XMRV RT is not able to excise NRTI-terminated primers. Similarly, it was previously shown that MoMLV RT is not capable of unblocking chain-terminated primers (33).
In HIV, decreased binding of AZT is conferred initially in the presence of the primary Q151M mutation, followed by secondary mutations F77L, A62V, V75I and F116Y (27,47,84). XMRV RT already differs from wild-type HIV-1 RT in the first three of these residues (P104, Q113 and L115 versus A62, V75 and F77) (Table 1). We demonstrated that introducing the primary Q→M mutation at the equivalent XMRV RT site (Q190M) resulted in an enzyme with decreased susceptibility to AZT. Hence, it appears that these residues can confer AZT resistance to XMRV by reduced incorporation of nucleotide analogs, as is the case in HIV-2 (41). At this point we do not know if introduction of as yet unknown mutations could endow XMRV RT with the ability to unblock chain-terminated nucleic acids. The details of the molecular mechanism of XMRV resistance to tenofovir and AZT are under investigation.
In conclusion, our study provides detailed biochemical analysis of the mechanisms of polymerization, inhibition, fidelity, processivity and drug resistance of XMRV RT and how it compares with the closely related enzyme MoMLV RT and the more distantly related HIV-1 RT. The findings enhance our understanding of the basic mechanisms of reverse transcription.
Supplementary Data are available at NAR Online.
NIH grants (AI076119, AI079801, and AI094715, to S.G.S.), (AI074389, to D.H.B.), (AI079801 to M.A.P.); NIH Bench-to-Bedside Award and the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research (to V.K.P); Ministry of Knowledge and Economy, Bilateral International Collaborative R&D Program, Republic of Korea; Canadian Institutes of Health Research (CIHR) and University of Missouri (to S-L.L.); amfAR Mathilde Krim Fellowship and a CIHR Fellowship (to B.M.). Funding for open access charge: NIH grants (AI076119, AI094715, AI074389, AI079801).
Conflict of interest statement. None declared.
The content of this publication does not necessarily reflect the views or policies of the Department of Health Human Services, nor does mention of trade names, commercial products or organizations imply endorsement by the U.S. Government.