Ribavirin Can Be Mutagenic for Arenaviruses ^▿

Cells, virus, and drugs.

Lymphocytic choriomeningitis virus (LCMV) strain Armstrong (Arm) 53b is a clone from LCMV Arm CA that was plaque purified three times and passaged four times in BHK-21 cells. This virus was used for all cell culture infections described in the present study. Procedures to grow BHK-21 and Vero cells and to infect cells with LCMV have been previously described (39–41, 64). Briefly, for single-step infections, BHK-21 cells were infected with LCMV at a multiplicity of infection (MOI) of 10 PFU/cell in the presence or absence of ribavirin, mycophenolic acid (MPA), or guanosine or combinations of these drugs at the concentrations indicated for each experiment. Progeny virus was collected at 48 h postinfection (p.i.) from the cell culture medium. For serial passages, infections of BHK-21 cells were carried out at an MOI of 0.01 PFU/cell in the presence or absence of drugs as indicated for each experiment. In each passage, the cells were infected with the progeny virus from the preceding infection and collected at 48 h postinfection. For LCMV infections in the presence of ribavirin, BHK-21 cells were preincubated with the desired amount of ribavirin (Sigma) (0, 10, 20, or 100 μM) for 7 h, and the same ribavirin concentration was maintained throughout the infection. The toxicity of ribavirin for BHK-21 cells under the culture conditions used in the present study has been previously reported (3, 77). BHK-21 cells maintained at least 90% viability after 48 h of treatment with 100 μM ribavirin compared to control BHK-21 cells maintained in parallel in the absence of the drug. For LCMV infections in the presence of 0, 5, and 30 μM MPA, no toxicity for the cells was detected (100% survival after 48 h in 30 μM MPA). Treatment of BHK-21 cells with guanosine (100 μM), which did not result in detectable toxicity for the cells, was carried out as previously described (5). LCMV was titrated by plaque assay on Vero cell monolayers (4). Mock-infected BHK-21 and Vero cells were maintained in parallel, and the corresponding cell culture supernatants were titrated to control possible viral contamination. No infectivity was detected in the mock-infected cultures at any time.

RNA extraction, RT-PCR amplification, and LCMV RNA quantification.

RNA was extracted from the supernatants of LCMV-infected cultures using Trizol (Invitrogen), following the manufacturer's instructions. LCMV RNA was amplified by reverse transcriptase PCR (RT-PCR) using RT transcriptor (Roche) and Pfu DNA polymerase (Promega) and the following pair of primers, primers L3654F (F stands for forward) (5′-AGTTTAAGAACCCTTCCCGC-3′; residues 3654 to 4268) and L4260R (R stands for reverse) (5′-CGAGACACCTTGGGAGTTGTGC-3′; residues 4239 to 4260). Nucleotide positions are given in the viral (genomic) sense and refer to the consensus genomic sequence determined previously for the L segment (GenBank accession number AY847351-L) (39). To ensure that in the RT-PCR amplification, excess LCMV RNA template was present for quasispecies analysis, 1:10 and 1:100 dilutions of the initial template preparations were amplified in parallel. Only preparations that yielded a positive amplification band at the two dilutions were subjected to molecular cloning and sequencing of individual clones (5). The amplified cDNAs were either purified with a Wizard PCR purification kit (Promega) or subjected to agarose (Pronadisa) gel electrophoresis; the cDNA band was extracted from the gel using a QIAEX II gel extraction kit (Qiagen). Purified DNA was sequenced by Macrogen, Inc., to obtain the consensus sequence of the corresponding population.

Genomic large (L) RNA was quantified by Light Cycler DNA Master SYBR green I kit (Roche), according to the manufacturer's instructions. The polymerase-coding region was amplified with primers L4183F (5′-ATCGAGGCCACACTGATCTT-3′; residues 4183 to 4202) and L4260R (5′-CGAGACACCTTGGAGTTGTGC-3′; residues 4239 to 4260). An LCMV RNA fragment spanning nucleotides 3662 to 4268 was used as the standard. This was obtained as a runoff transcript from a molecular DNA clone of the polymerase-coding region in the genomic sense, cloned into pGEM-T Easy vector (Promega). The denaturation curve of the amplified DNAs was determined to monitor the specificity of the reaction. Negative controls (without template RNA) were run in parallel with each amplification reaction mixture. Each value is the average of at least three determinations. The specific infectivity of LCMV was calculated by dividing the number of progeny infectivity (PFU) by the amount of LCMV RNA in the same volume of culture medium.

Molecular cloning and calculation of mutant spectrum complexity.

Molecular clones were prepared from cDNA (the band corresponding to the RT-PCR amplification obtained with undiluted template) using primers L3654F and L4260R (described above in “RNA extraction, RT-PCR amplification, and LCMV RNA quantification”). cDNA was ligated to the pGEM-T Easy vector (Promega) and transformed into Escherichia coli DH5α. cDNA from individual positive E. coli colonies was amplified with Templiphi (GE Healthcare) and sequenced (Macrogen, Inc.).

The average mutation frequency among components of the mutant spectrum of an LCMV population was calculated by dividing the number of different mutations found by the total number of nucleotides sequenced. The Shannon entropy (S) (which measures the proportion of different sequences in the region analyzed) was calculated using the formula S = −[∑i (p_i × ln p_i)]/ln N where p_i is the proportion of each sequence in the mutant spectrum and N is the total number of sequences compared (94). An S value of 0 means that all sequences are identical, while a value of 1 means that the sequences are different from each other. Statistical significance values were calculated using Prism software program version 5.0 or higher. The mutation frequency calculated for LCMV passaged in the absence of ribavirin was at least 2.8-fold larger than can be attributed to the error incorporation during the RT-PCR procedure used (82).

Assessment of the inhibitory and mutagenic activity of ribavirin during LCMV replication in cultured cells.

We first compared the inhibitory effect of ribavirin on LCMV multiplication in BHK-21 cells following infection at a low MOI and a high MOI (Fig. 1). The concentrations of ribavirin that produced a decrease of 99% in the yield of infectious progeny (99% inhibitory concentrations [IC₉₉]) were 4.28 ± 0.24 μM for the infections carried out with an MOI of 0.01 PFU/cell and 12.83 ± 0.61 μM for the infections carried out with an MOI of 10 PFU/cell. Thus, the inhibitory effect of ribavirin on LCMV was more pronounced in infections carried out at a low MOI.

Ribavirin has been recognized as a mutagen for several RNA viruses (reviewed in reference 37). To investigate whether ribavirin could exert a dual inhibitory and mutagenic activity during LCMV replication, single-step infections were carried out at an MOI of 10 PFU/cell in the presence and absence of 20 μM or 100 μM ribavirin. The progeny infectivity, progeny RNA, mutant spectrum complexity, and types of mutations in the progeny populations were analyzed. The results (Table 1) confirm a strong inhibitory activity of ribavirin, with at least a 10⁶-fold reduction of virus titer and 10²-fold reduction of viral RNA levels in the presence of 100 μM ribavirin. It must be noted that the specific infectivity of progeny LCMV decreased 28.5-fold in the infection carried out in the presence of 20 μM ribavirin and at least 2,170-fold in the presence of 100 μM ribavirin (considering the limit of detection of infectivity [Table 1]). The significant decrease of specific infectivity may be relevant to the mechanism of inhibition of LCMV replication by ribavirin (see Discussion). The mutation frequency among components of the LCMV population passaged in the presence of 20 μM ribavirin was significantly higher (2.3-fold; P = 0.0385 by χ² test) than in the population passaged in the absence of ribavirin. In contrast, no increase in mutation frequency was detected in the mutant spectrum of LCMV replicated in the presence of 100 μM ribavirin compared to LCMV replicated in the absence of ribavirin. The difference of population complexity was also reflected in the Shannon entropy (Table 1). Furthermore, examination of the mutation types shows that the infection in the presence of 20 μM ribavirin, but not in the presence of 100 μM ribavirin, led to a specific increase of G→A and C→U transitions (P < 0.05 by χ² test) (Table 1). G→A and C→U are the types of mutations previously associated with ribavirin mutagenesis (3, 5, 18). In the population passaged in the absence of ribavirin, A→G and U→C transitions were more frequent than G→A and C→U transitions, and the same bias was present in the progeny of the infection carried out in the presence of 100 μM ribavirin (Table 1). In all cases, transitions were 3- to 5-fold more frequent than transversions. The results strongly suggest that ribavirin can exert a dual mutagenic and inhibitory activity on LCMV.

Effect of guanosine on the inhibitory and mutagenic activity of ribavirin exerted on LCMV.

In the cell, ribavirin is phosphorylated into ribavirin mono-, di-, and triphosphate (RMP, RDP, and RTP, respectively) by cellular enzymes (91, 98). RMP acts as a competitive inhibitor of inosine monophosphate dehydrogenase (IMPDH) and reduces the intracellular concentration of GTP (33, 90, 91). RTP is used as a substrate by the RNA-dependent RNA polymerases (RdRps) of picornaviruses in place of GTP or ATP during RNA elongation resulting in mutagenesis of viral RNA (3, 18, 30, 37, 69, 87). The addition of guanosine to the cell culture medium can restore intracellular GTP levels despite the presence of RMP (5, 50, 54, 98). To study whether GTP levels were involved in the inhibition and mutagenesis of LCMV by ribavirin, single-step infections were carried out in the presence or absence of ribavirin and in the presence of 100 μM guanosine, a concentration that compensates for the decrease of GTP produced by ribavirin treatment of BHK-21 cells (5). The results (Fig. 2 and Table 1) show that the presence of guanosine in the culture medium greatly reduced the inhibition of infectious LCMV progeny production by 100 μM ribavirin (16.6-fold decrease of infectious progeny production in the presence of guanosine versus at least a 112,121-fold reduction in the absence of guanosine [Table 1]). Guanosine also prevented the increase of mutation frequency and Shannon entropy associated with the presence of 20 μM ribavirin and abolished the differences in the frequency of transition types observed between the mutant spectra of LCMV passaged in the presence and absence of 20 μM ribavirin (Table 1). Thus, the inhibitory and mutagenic activities of ribavirin exerted on LCMV during virus replication were largely overcome by the presence of guanosine in the culture medium, suggesting that inhibition and mutagenesis of LCMV by ribavirin were mediated or enhanced by low intracellular GTP levels.

Effect of treatment with mycophenolic acid on LCMV multiplication in cultured cells.

Low intracellular GTP caused by ribavirin treatment may favor misincorporation of AMP (if RTP is not a substrate for LCMV polymerase) or both AMP and RMP (if RTP is a substrate for the polymerase) in place of GMP. To quantitate the mutagenic activity evoked by low intracellular GTP levels in the absence of ribavirin, we conducted LCMV infections in the presence or absence of MPA, an inhibitor of IMPDH that cannot be incorporated into nucleic acids (33). We conducted single-step LCMV infections (MOI of 10 PFU/cell) in the presence or absence of 5 and 30 μM MPA, as 30 μM MPA results in a decrease of intracellular GTP levels at least equal to the decrease produced by 500 μM ribavirin (5). The infections were done in the presence or absence of 100 μM guanosine. The results (Table 2) show that MPA reduced LCMV infectious progeny production by 2.1- to 2.3-fold, an inhibition which is at least 48,000-fold lower than the inhibition produced by ribavirin (compare Tables 1 and 2). Guanosine totally compensated for the inhibition exerted by MPA on LCMV progeny production (Table 2). The presence of MPA did not lead to any increase in mutant spectrum complexity, a result which contrasts with that obtained with 20 μM ribavirin (Tables 1 and 2). In fact, the mutation frequency and Shannon entropy of the LCMV that replicated in the presence of 30 μM MPA decreased 4.3- and 3.6-fold, respectively, relative to the populations replicated in the absence of MPA (Table 2). Thus, in contrast to ribavirin, MPA is not mutagenic for LCMV. The decrease of mutant spectrum complexity in the presence of MPA was associated mainly with a lower frequency of A→G and U→ C transitions, the transition types that LCMV tends to produce during replication. This lower frequency is expected if this mutational tendency is favored by high intracellular GTP concentrations during LCMV replication. Surprisingly, the amount of LCMV progeny RNA decreased significantly in the virus produced in the presence of MPA (P < 0.05 by Student's t test) (Table 2). As a consequence, the specific infectivity of the LCMV produced in these infections was 1.2- to 3.2-fold higher than that of the population generated in the absence of drugs. Since mutagenesis of RNA viruses is associated with decreases in specific infectivity (2, 3, 34, 35, 39, 40, 76, 77), the increase in specific infectivity that the virus undergoes in the presence of MPA and guanosine reinforces the conclusion of the absence of mutagenic activity of MPA on LCMV. How alterations of intracellular GTP levels can modify the specific infectivity of LCMV is unknown (see Discussion). Thus, the mutagenic activity of ribavirin cannot be attributed to the inhibition of IMPDH exerted by RMP.

Assessment of ribavirin mutagenic activity and its prevention by guanosine during serial LCMV infections in cultured cells.

The mutagenic activity of ribavirin on LCMV and its compensation by guanosine (Fig. 2 and Table 1) were studied in single-step infections carried out at a high MOI. The IC₉₉ for ribavirin was 3-fold higher in the infections carried out at an MOI of 10 PFU/cell than at 0.01 PFU/cell (Fig. 1). Infection at a high MOI promotes accumulation of defective interfering (DI) particles of LCMV (80, 95, 96), whereas infection at a low MOI acts as a filter to eliminate DI particles whose replication is dependent on standard, infectious LCMV. Previous studies showed that a subset of defective but RNA replication-competent LCMVs termed defectors could interfere with replication of standard LCMV and contribute to virus extinction (40, 42, 63). Since the proportion of defector LCMV genomes increases with moderate intensities of mutagenesis (40), it was important to ascertain that the mutagenic activity of ribavirin observed during LCMV infection at a high MOI was not dependent on DI particles and defector genomes. To achieve this aim, we conducted parallel infections of BHK-21 cells at a low (0.01 PFU/cell) and high (10 PFU/cell) MOI in the presence of increasing concentrations of ribavirin (Fig. 3). Quantification of infectivity and viral RNA after the first passage shows that the inhibition of infectious progeny production by ribavirin measured at 24 h and 48 h postinfection (p.i.) was 100- to 300-fold higher at a low MOI than at a high MOI (Fig. 3A and B). In contrast, the decrease in viral RNA progeny production by ribavirin was less pronounced at 48 h p.i. and almost absent at 24 h p.i. at a low MOI (Fig. 3C and D).

The different effects of ribavirin on the yield of infectious units and viral genomes resulted in a significant decrease of specific infectivity due to ribavirin activity at an MOI of 0.01 PFU/cell at high concentrations of ribavirin (100 μM), not at low concentrations of ribavirin (5 and 20 μM) (Fig. 3E). The decrease in specific infectivity in the infection carried out in the presence of 100 μM ribavirin was diminished at a high MOI (Fig. 3F). The results with an MOI of 10 PFU/cell are in agreement with those reported in the previous series of independent infections described in Table 1.

A total of nine passages were performed at a low MOI in the presence of 10 μM ribavirin to ensure continued virus infectivity (absence of extinction) over nine passages. Evolution of the infectivity values in the presence or absence of ribavirin and guanosine in the course of the nine passages indicated an inhibitory activity of ribavirin that decreased with passage number and that was largely compensated for by the presence of guanosine at all passages (Fig. 4 and Table 3). The yields of infectivity and viral RNA at passage 1 were significantly reduced by 10 μM ribavirin, resulting in a 15-fold increase in specific infectivity. At passage 9, the inhibition by ribavirin diminished and no difference in specific infectivity was detected (Fig. 4 and Table 3). The mutation frequency and Shannon entropy at passage 9 were significantly higher in the virus passaged in the presence of ribavirin than in the virus passaged in the absence of ribavirin (P < 0.005 by χ² test), and the increase was associated with G→A and C→U transitions, the transition types expected from ribavirin mutagenesis. Guanosine compensated for the decrease of infectious progeny production and progeny RNA levels associated with the presence of ribavirin but compensated only partially for the increase of mutant spectrum complexity evoked by ribavirin (Table 3). Thus, several passage regimes indicate that ribavirin displays a dual inhibitory and mutagenic activity for LCMV during virus replication in cell culture, and both activities can be either totally or partially prevented by the presence of guanosine in the culture medium. The mutagenic activity of ribavirin on LCMV cannot be explained as a result of the inhibition of IMPDH by RMP.