Susceptibility of the Human Retrovirus XMRV to Antiretroviral Inhibitors

Robert A Smith; Geoffrey S Gottlieb; A Dusty Miller


Retrovirology. 2010;7(70) 

In This Article


In this study, we used a novel marker transfer assay to directly compare the susceptibility of XMRV and HIV-1 to a panel of antiretroviral drugs in the same host cell type. Our experimental approach and findings differ from previous studies of XMRV in several important ways. With regard to NRTIs, the initial report by Sakuma et al.[21] suggested that XMRV is sensitive to AZT but resistant to 3TC, d4T and tenofovir. Importantly, the single dose of tenofovir used in their experiments (30 nM) was substantially lower than the EC50 observed in our assays (~5 μM; Table 1), leading the authors to conclude that XMRV was resistant to the drug. Our analysis shows that tenofovir is equally potent against XMRV and HIV-1 in culture (Table 1). A subsequent study by Singh et al.[23] used differing cell types to compare XMRV and HIV-1, and as a result, differences in the intrinsic NRTI susceptibilities of the two viruses could not be resolved from host cell-specific differences in NRTI activity. In fact, careful inspection of their data suggests that XMRV is relatively resistant to AZT, tenofovir and TDF (a prodrug of tenofovir), as the EC50 values for these analogs were 15-94-fold higher for XMRV compared to HIV-1. Our data are more congruent with the findings of Paprotka et al.,[22] who showed that XMRV and HIV-1 are comparably sensitive to AZT and TDF in prostate cancer cells. We extend these observations by demonstrating that, in addition to AZT and tenofovir, the NRTIs AZddA, AZddG and adefovir are equally active against XMRV and HIV-1 (Table 1). Taken together, our analysis resolves disparities among earlier reports of XMRV drug susceptibility and illustrates that XMRV is sensitive to a broader range of NRTIs than was previously appreciated.

Overall, the patterns of drug susceptibility observed in our analysis of XMRV are similar to those seen in previous studies of Moloney MLV (MoMLV). MoMLV is sensitive to AZT, adefovir and tenofovir, but is relatively resistant to ddI, D4T, 3TC, abacavir and PFA.[33–36] In addition, purified MoMLV protease is highly resistant to PIs,[37] whereas both raltegravir and elvitegravir have been shown to inhibit MoMLV replication in culture.[38,39] In agreement with our findings for XMRV (Table 2), MoMLV is moderately resistant to elvitegravir, as evidenced by a 7-fold greater EC50 for the drug relative to HIV-1.[39] These concurrent drug sensitivity patterns are consistent with the high degree of amino acid sequence similarity shared between XMRV and MoMLV, which are 99% identical in the protease and RT polymerase domain and 90% identical in the integrase catalytic core domain (CCD).

To gain further insights into the molecular basis of antiretroviral drug resistance in XMRV, we constructed amino acid alignments of the inferred XMRVVP62 and HIV-1NL4-3 sequences for the entire protease enzyme, the portion of RT spanning the conserved polymerase motifs, and the integrase CCD (Figure 3). Within these three regions, XMRV and HIV-1 share 27-31% amino acid identity and 18-21% amino acid similarity. Importantly, the XMRV and HIV-1 sequences differ at several sites that are critical for antiretroviral drug resistance. XMRV protease contains three residues (V54, S81, and L92) that correspond to PI resistance-conferring replacements in HIV-1 (I47V, T74 S, and I84L, respectively) (Figure 3A).[40] XMRV also contains several amino acid residues in the RT polymerase domain that, in HIV-1, result in NNRTI resistance (K101P, K103 H, Y181L, Y188L, and G190A) and dideoxynucleoside analog resistance (T69N, L74V, Y115F) (Figure 3B).[40,41] These sites likely contribute to intrinsic drug resistance in XMRV. In addition, XMRV integrase contains a serine at the position corresponding to Q148 in HIV-1 (Figure 3C), which is known to be critical for integrase inhibitor resistance in HIV-1.[42] This amino acid difference may contribute to moderate elvitegravir resistance in XMRV (Table 2).

Figure 3.

Alignment of Pro and Pol amino acid sequences for XMRV and HIV-1. Alignments are shown for the protease (panel A), amino-terminal RT (panel B) and integrase catalytic core domain (CCD) sequences (panel C) of XMRVVP62 and HIV-1NL4-3 ([GenBank: NC_007815.1] and [GenBank: M19921, respectively). Numbering for XMRVVP62 is based on assigned amino acid numbers for the corresponding MoMLV peptides [GenBank: AF033811]. Alignments were generated using EMBOSS[62] with the following settings: gap-open = 10, gap extend = 0.5, algorithm = needle (global), scoring matrix = BLOSUM62. Amino acid identities between XMRVVP62 and HIV-1NL4-3 are shown with yellow boxes, conserved amino acid residues (BLOSUM62 score ≥1) are shown with grey boxes, and alignment gaps with are indicated with a dash (-). Catalytic active site residues are indicated with an asterisk (*). For RT, the initial EMBOSS alignment was manually adjusted to conform to a recent structural alignment of MoMLV and HIV-1 RTs.[63] Boundary boxes for conserved polymerase motifs A-D are shown as previously assigned.[64] Boundaries for motif F are shown as identified in alignments of viral RNA-dependent RNA polymerases.[65] The X at position five of XMRV protease indicates the location of a termination codon that, in MLV, is suppressed during translation of Gag-Pol-encoding RNA. Sites involved in antiretroviral drug resistance in HIV-1, as tabulated by the International AIDS Society-USA (for protease and RT)[40] or in the Stanford University HIV Drug Resistance Database (for integrase)[66] are indicated in bold, colored letters. The locations of primary PI, NRTI, and NNRTI resistance mutations, as well as changes associated with resistance to the integrase inhibitors raltegravir and elvitegravir, are shown in red. Sites involved in NNRTI resistance are shown in blue. Pound signs (#) indicate amino acid residues believed to be important for the positioning of strand transfer inhibitors, based on a recent structural analysis of prototype foamy virus integrase.[51]

As observed in previous studies of MoMLV RT,[43,44] XMRV was highly resistant to the L-pseudosugar nucleoside analogs 3TC and FTC (Table 1). Both MoMLV and XMRV RT encode a valine at the second position of the conserved YXDD sequence of polymerase motif C, whereas the corresponding residue in HIV-1 RT is methionine 184 (Figure 3B). Although the M184V replacement confers high-level resistance to 3TC and FTC in HIV-1,[45] mutants of MoMLV that harbor the reciprocal change in the YXDD sequence (V223M) remain highly resistant to 3TC.[43,44] It is therefore likely that amino acid sites outside the YXDD sequence of RT contribute to intrinsic 3TC/FTC resistance in XMRV.

In HIV-1 RT, specific substitutions at positions 41, 67, 70, 210, 215 and 219 (commonly known as thymidine analog mutations or TAMs) confer AZT resistance by enhancing RT-catalyzed excision of AZT-5'-monophosphate from the nascent DNA strand.[46] Although the sequences of XMRV and HIV-1 differ at five of the six TAM sites in RT (Figure 3B), these residues are unlikely to influence AZT susceptibility in XMRV, as the excision activity of MoMLV RT is orders of magnitude lower than that of the HIV-1 enzyme.[47] Indeed, we observed that XMRV and HIV-1 were comparably sensitive to AZT as well as two other NRTIs containing a 3'-azido modification (AZddA and AZddG; Table 1). Based on previous studies of HIV-1 and MoMLV,[29,48,49] we expect that XMRV RT can utilize the 5'-triphosphate forms of these analogs as alternative nucleotide substrates, resulting in chain termination of DNA synthesis. Additional biochemical analyses are required to characterize the nucleotide selectivity and excision activity of XMRV RT.

Two recently-published reports have shown that the integrase inhibitor raltegravir inhibits XMRV replication in culture at nanomolar concentrations of the drug.[22,23] Our results confirm these findings and demonstrate that elvitegravir is also active against XMRV, although the concentrations of elvitegravir needed to inhibit XMRV infection were higher than those required for raltegravir (Table 2). A third integrase inhibitor, L-870812, has also been reported to exert moderate antiviral activity against XMRV in culture, with an EC50 32-fold greater than that of raltegravir.[23] Although raltegravir, elvitegravir and L-870812 are structurally divergent, these three inhibitors share a common pharmacophore that binds the active site metal ions essential for integrase strand transfer catalysis.[50] Recent crystallographic studies have identified three amino acid residues that are believed to influence the positioning of strand transfer inhibitors in the integrase active site,[51] and based on our alignment of the CCD, these residues are conserved in the XMRV and HIV-1 integrase sequences (Figure 3C). Taken together, these data suggest that the strand transfer inhibitor-binding sites of XMRV and HIV-1 integrase share a similar overall topology despite numerous amino acid differences in the CCD.

We used two independent sources of XMRV for our studies: one derived from the infectious molecular clone VP62[11] and the other from 22Rv1 prostate carcinoma cells.[7] Our rationale for this choice was that the VP62 clone might encode alterations that influence drug susceptibility, whereas 22Rv1 cells harbor at least 10 proviral copies of XMRV, presumably providing a more diverse sample of the virus. However, a recent analysis of XMRV sequences from 22Rv1 cells revealed that the proviruses are nearly identical to each other and to the VP62 molecular clone.[22] There are only two nucleotide differences between the consensus XMRV22Rv1 and XMRVVP62 sequences ([GenBank: FN6900043] and [GenBank: EF185282], respectively); these result in single amino acid changes in Gag and Env, whereas the Pro and Pol proteins are identical. Thus, the key proteins targeted by the antiretroviral drugs tested in our study are identical in XMRV22Rv1 and XMRVVP62. This identity is reflected in the similar EC50 values obtained for these two viruses (Tables 1 and 2). Strikingly, all six of the full-length XMRV sequences currently available in GenBank show a high degree of nucleotide identity (Figure 4). Although the lack of variation reported in XMRV is difficult to reconcile with the known mutation rates of MoMLV and other retroviruses, collectively, these sequencing results suggest that the drugs that are active against XMRV22Rv1 and XMRVVP62 should be similarly active against other XMRV strains.

Figure 4.

Phylogenetic analysis of XMRV. All full-length XMRV sequences available in GenBank (accessed on April 28, 2010) were aligned using ClustalW. Unrooted (panel A) and rooted (panel B) phylogenetic trees were generated using the neighbor-joining algorithm of MEGA 4.0[67] with default settings. Scale bars indicate evolutionary distance in base substitutions per site (i.e., the distance shown in panel A equals 2 substitutions per 10,000 bases). Note that after the original sequencing of XMRV strains VP62, VP42 and VP35,[3] strain VP62 was resequenced ("VP62 corrected"[11]). The resulting sequence reveals a closer similarity between VP62 and other XMRV strains and suggests that the branch lengths of VP35 and VP42 are also likely overestimated due to PCR or sequencing errors. mChrom13 indicates an endogenous MLV sequence located on Mus musculus chromosome 13 [GenBank: CT030655.7], and is the most closely related non-XMRV sequence found by BLAST search of GenBank using the XMRV22Rv1 sequence. DG-75 indicates DG-75 MLV [GenBank: AF221065].


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.