The present article describes the steps required to isolate and characterize RNA polymerase fidelity variants of RNA viruses and how to use mutation frequency data to confirm fidelity changes in tissue culture.
RNA viruses use RNA dependent RNA polymerases to replicate their genomes. The intrinsically high error rate of these enzymes is a large contributor to the generation of extreme population diversity that facilitates virus adaptation and evolution. Increasing evidence shows that the intrinsic error rates, and the resulting mutation frequencies, of RNA viruses can be modulated by subtle amino acid changes to the viral polymerase. Although biochemical assays exist for some viral RNA polymerases that permit quantitative measure of incorporation fidelity, here we describe a simple method of measuring mutation frequencies of RNA viruses that has proven to be as accurate as biochemical approaches in identifying fidelity altering mutations. The approach uses conventional virological and sequencing techniques that can be performed in most biology laboratories. Based on our experience with a number of different viruses, we have identified the key steps that must be optimized to increase the likelihood of isolating fidelity variants and generating data of statistical significance. The isolation and characterization of fidelity altering mutations can provide new insights into polymerase structure and function1-3. Furthermore, these fidelity variants can be useful tools in characterizing mechanisms of virus adaptation and evolution4-7.
1. Determine the range of mutagen concentrations that is minimally toxic to cells
The purpose of this exercise is to determine what range of mutagen concentrations can be used during an infection without excess cell toxicity. Essentially, you will want to reproduce the conditions that will be required for virus infection. For most viruses, infections last between 2 and 7 days. Prepare enough plates to sample cells at each day. If non-adherent cells are used, modify the protocol accordingly.
2. Determine the optimal non toxic mutagen concentration that moderately reduces virus titers (approximately 0.5-2 log reduction)
This exercise serves to determine the concentration of mutagen that will exert a strong selective pressure, without over-mutagenizing the population. For RNA mutagens, we find that this corresponds to 10.5-2 logs reductions in virus titer. At these concentrations, every genome is mutated in at least one or two positions. The mutagen may aid in the generation of the resistance mutation, which will then be selected over passage. If too many mutations are introduced (at very high mutagen concentrations), the mutants bearing the resistance mutation will themselves be lethally mutagenized, impeding their isolation.
3. Isolation and identification of mutagen resistant variants
Perform large population size passages in the optimal mutagen concentration defined above and check virus titers across the passage series. As a control, passage virus in growth medium without any mutagen. As another control to monitor the potential emergence of defective interfering particles (DI), perform fresh infections in absence of mutagen at each passage step (unpassaged control).
4. Once a mutation has been identified, isolate or generate the variant and confirm the resistance phenotype to several RNA mutagens
Next, the variant presenting the identified mutation is isolated to confirm its link to the resistance phenotype. It is essential that the mutation suspected of changing fidelity is studied in a genetically clean background (that is, not presenting additional mutations elsewhere in the genome). In the best situation, an infectious cDNA clone exists that would permit the generation of a stock of the mutagen-resistant variant by site directed mutagenesis on a clean genetic background. In this case, section 4 is not necessary. However, if a cDNA clone is not available, isolation can be done by plaque purification of virus, described below. More than one round of plaque purification may be required to isolate the variant on a clean, genetic background.
5. Check replication rates
Since fidelity- altering mutations most often map to the polymerase, it is possible that the same polymerase mutation will significantly alter replication kinetics and it is important to determine similarities and differences in replication that will permit a better comparison of differences in mutation frequencies performed below. To do so, examine replication by at least two complimentary approaches – one that examines virus production and another that examines RNA synthesis.
6. Measure mutation frequencies
This is a critical step in confirming that the identified polymerase mutation conferring resistance to mutagen alters replication fidelity. It is important to note that the mutation frequencies measured here are not mutation rates. To determine rates, a very careful measurement of replication kinetics (amount of RNA synthesized and length of replication cycle) must be factored in. Measuring mutation frequencies however, as long as passage history and replication kinetics are monitored, provides reproducible, quantitative measures of replication fidelity. Mutation frequencies can be determined either in the viable virus population (plaque clones or limiting dilution) or in the total virus population (virus stock or supernatant). To determine mutation frequencies, prepare virus stocks from a later passage (e.g. passage 2 or beyond). It is important that the virus population has had time to expand its genetic diversity closer to a mutation-selection equilibrium.
7. Sequence analysis
Perform sequence analyses using a reference or consensus sequence for each population and appropriate alignment software. We recommend Lasergene or Sequencher that can readily identify SNPs with respect to consensus.
8. Representative results:
The dose dependent effect of mutagen concentration on cell viability and virus viability is shown in Figure 1. In this example, we found that virus passage in 100 μM AZC was reduced in virus titer by the targeted 10.5-2 log, but HeLa cell viability was not negatively impacted during the 2 days required for virus infection. This pilot experiment led to the choice of 100 μM AZC concentration for serial passage of virus, to select for mutagen resistance. Figure 2 illustrates the initial reduction in titer, followed by emergence of a mutagen resistance phenotype. During the first few passages in mutagen, as lethal mutations accumulate, a significant drop in virus titers occurs. Gradually, a mutagen-resistant variant emerges an its emergence coincides with a return to virus titers no different from untreated controls. At this stage, a large percentage of the virus population presents the resistance mutation. Sequencing of this virus population reveals the amino acid change(s) responsible. Once identified, and isolated or newly generated, the mutagen-resistant virus may be less sensitive than wild type to different RNA mutagens (base analogs of different structure, for example). Figure 3 shows a RNA mutagen resistant Coxsackie virus B3 that titers higher than wild type in the presence of ribavirin, 5-fluorouracil, and 5-azacytidine, and high MgCl2 and MnCl2. Broad resistance to RNA mutagens is a strong indicator of increased replication fidelity. Verifying that replication kinetics of the fidelity variant is similar to wild type virus will aid in comparison of mutation frequencies. Figure 4 depicts the one step-growth kinetics of a high fidelity variant compared to wild type. If replication rates and final titers are not similar, then steps should be taken to compare virus populations of similar size, that have undergone the same number of rounds of replication. The link between RNA synthesis rate and replication fidelity is not well characterized, particularly in vivo. A slower replication rate may result in a decreased mutation frequency (higher fidelity), although this is not an absolute rule, as shown in Figure 4. With the above parameters established, the mutation frequencies of the fidelity variant and wild type populations can be compared to obtain genetic confirmation of altered replication fidelity. Figure 5 shows a sequence alignment of a wild type and high fidelity variant, with point mutations identified. The mutations are counted, ranked according to number of mutations per clone (Table 1), and represented as an average mutation frequency per population, per 10,000 nucleotides sequenced, Figure 6.
Figure 1. Determining the optimal conditions to select for RNA mutagen resistance: retention of high cell viability with a moderate (1-2 log drop in virus titer). HeLa cells were treated with indicated concentrations of ribavirin and infected with wild type Coxsackie virus B3 at a MOI of 0.01. 48 hours post infection, the progeny virus was harvested and titers were determined by TCID50. The percentage of cells surviving treatment at 48 hours, determined by Trypan blue staining, is indicated below the x-axis. The results show that concentrations of 100 and 200 μM reduce virus titers by 1-2 log, without affecting cell viability.
Figure 2. Serial passage in the presence of moderate concentrations of RNA mutagens selects for mutagen resistant populations. In this figure, Chikungunya virus was passaged in HeLa cells in the presence of 50 μM ribavirin (grey bars). Control passages were performed in absence of ribavirin (black bars). After each passage, virus progeny was quantified by classic plaque assay on BHK cells. The mutagenic effect is evident during the first passages (p1 and p2 compared to p0 starting population) where the treated virus titers drop by 2 log. Gradually, titers return to normal (untreated) levels. No significant differences are observed in passage 5 mutagen treated populations compared to untreated, suggesting that resistant variants have been selected. Indeed, consensus sequencing of the population identified unique mutations in the virus population undergoing ribavirin treatment.
Figure 3. Confirmation of broad resistance to RNA mutagens of different structure. Shown here, the high fidelity A372V variant of Coxsackie virus B3 that was initially isolated in the screen described in Section 3 was generated from an infectious clone and tested for its relative sensitivity to different concentrations of different RNA mutagens (ribavirin, 5-fluorouracil, 5-azacytidine). HeLa cells were treated with indicated concentrations of ribavirin and infected with wild type Coxsackie virus B3 at a MOI of 0.01. 48 hours post infection, the progeny virus was harvested and titers were determined by TCID50. Shown here are the titers of wild type (solid lines) and A372V variant (dashed lines) as a function of mutagen concentration. A372V consistently titers higher than wild type under all conditions tested.
Figure 4. Replication rates and fidelity variants. To determine the one-step growth kinetics of virus production, HeLa cells were infected at MOI = 10 with either wild type (solid line), high fidelity variant A372V (long dashes) or replication deficient variant Cx64 (short dashes) of Coxsackie virus B3. At time points indicated, virus progeny was harvested from cells and supernatants by freeze-thaw and titered by TCID50. The fidelity increase of A372V does not coincide with an observable replication defect in tissue culture. The variant Cx64 presents a significant delay in replication kinetics and reaches maximum titers that are 1000-fold lower than wild type virus.
Figure 5. Alignment of TopoTA cloned sequences from each virus population. Using the approach described in Section 7, each sequence obtained from cloned RT-PCR product presumably originates from a single, unique genome within the total virus population and would thus, carry unique mutations. The figure shows a typical alignment, following clean up of poor quality sequences and visualization of SNPs. The total SNPs (10 in this figure) within a population are counted, and the number of SNPs appearing on each clone are noted. For example, the clone underlined by a bar, contains 2 unique mutations whereas 8 other clones contain a single, unique mutation. This data is used to compile Table 1. To view a larger version of this figure please click here.
Figure 6. Graphic representation of mutation frequencies of virus populations. For easier interpretation, the numerical data obtained from sequence and statistical analyses can be represented as either a chart, or histogram (shown here). A372V virus generates fewer mutations than wild type and presents a significantly lower mutation frequency (*, p<0.01). The Cx64 variant, that replicates to titers 1000-fold lower than wild type, presents the same mutation frequency (ns, not significant) indicating that replication speed and fidelity are not necessarily linked. The same Chikungunya virus (CHIKV) population gives similar mutation frequencies whether the virus stock, or a 105-fold dilution, is used for RNA extraction.
Mutation distribution summary for statistical analysis.
Note: For each clone, it is essential that the same genomic region (and length of sequence) is covered. In this case, 859 nucleotides per clone. This is critical for statistical analyses. On the other hand, rank sum tests used for statistical analysis do not require the sample sizes to be the same, the researcher is free to compare population of differing sample size. Hence, the 142 clones of wild type can be compared to the 84 clones of A372V.
# clones with n mutations | wild type | A372V |
7 mutations | 0 | 0 |
6 mutations | 0 | 0 |
5 mutations | 0 | 0 |
4 mutations | 0 | 0 |
3 mutations | 1 | 0 |
2 mutations | 6 | 2 |
1 mutations | 40 | 14 |
0 mutations | 95 | 68 |
Total mutations | 55 | 18 |
Total clones sequenced | 142 | 84 |
Total nucleotides sequenced | 121,978 | 72,156 |
Mutations/104 nt | 4.51 | 2.49 |
Table 1. Mutation distribution summary for statistical analysis. Note: For each clone, it is essential that the same genomic region (and length of sequence) is covered. In this case, 859 nucleotides per clone. This is critical for statistical analyses. On the other hand, rank sum tests used for statistical analysis do not require the sample sizes to be the same, the researcher is free to compare population of differing sample size. Hence, the 142 clones of wild type can be compared to the 84 clones of A372V.
Choice of cell line. The efficacy of base analogs as RNA mutagens correlates with their relative uptake by different cell types 11. If the cell line that is normally used for virus passage proves to be refractory to mutagen uptake or too sensitive (high cell toxicity), it may be necessary to use another cell line that meets these requirements and is still permissive to viral replication. Once the mutagen resistance variant is isolated, the remainder of the characterization can be performed in the original, preferred cell line. In our experience, HeLa cells readily take up mutagen; BHK cells require up to 10-fold higher concentrations and Vero cells are refractory to mutagen uptake.
Choice of mutagen. In trying to isolate fidelity variants by mutagen treatment, the likelihood of success is increased if more than one type of mutagen is used. Base analog mutagens of different structure, that are erroneously incorporated into genomes during replication will predominantly induce result in a specific subset of mutations in subsequent replication cycles: ribavirin treatment favors GtoA and CtoU transition mutations 12; 5-azacytidine has a similar bias, with the addition of CtoG and GtoC transversions 13; 5-fluorouracil preferentially induces AtoG and UtoC transitions 14. Alternatively, higher concentrations of Mg2+ or Mn2+ can be supplemented to the medium to increase the overall mutation frequency of RNA viruses without the bias described above 12. Depending on the virus’ codon sequences, and the codon changes required to generate a fidelity variant, some of these conditions will favor the emergence of this variant over others. For the higher fidelity poliovirus G64S and Coxsackie virus A372V, ribavirin treatment most readily selected for the variants because the required AtoG transition at the codon site corresponded to the mutations predominantly generated by this ribavirin.
MOI vs. population size. In virology, protocols for tissue culture infection pay particular attention to the multiplicity of infection (MOI), to avoid the accumulation of defective interfering particles (low MOI) or to promote recombination between viruses (high MOI), for example. To select for emergence events over serial passaging, it is also important to consider virus population size. Since the resistant mutant initially exists at low frequency, it is best to transfer as large a population size as possible from one passage to the next (105-106 viruses, e.g.) to avoid losing these emerging variants at each passage. Scaling up the size of well or flask (number of cells infected) may help to minimize the increase in MOI if this is of concern. On the other hand, for experiments in which the sensitivity of a virus to mutagen is being tested, low MOI infection is performed in order to increase the number of replication cycles occurring in the experiment and to avoid rescue of mutagenized genomes by higher fitness genomes through complementation in co-infected cells. This is important since the mutations generated on progeny genomes during the first round of replication will not be immediately detected. Most of these mutagenized RNAs will still be packaged into virion. It is in the next round of infection that lethal mutations present in these genomes will result in an aborted replication cycle, and reduction in virus titer. It may be necessary to allow for several rounds of accumulation of mutations before a significant effect of lethal mutagenesis is observed. Finally, if over the passage series in the presence of mutagen, the virus titers continue to drop until extinction, then the researcher should try passaging virus in gradually increasing amounts of mutagen (starting from a very low concentration).
Isolation and generation of the RNA mutagen resistant clone from the RNA mutagen-resistant population. RNA mutagens introduce multiple random mutations to each genome, but selection for resistance will only enrich (and fix to consensus sequence) the resistance mutation. To identify this mutation, we sequence the mutagen resistant population (consensus of the population) and not individual viruses. Hence, the single, random mutations created by the mutagen are not detected in the sequence; only the mutations that result in consensus changes following selection, are found. In our experience we only identify one or two such consensus sequence changes. Once the mutagen resistant population is obtained and the resistance mutation is identified, it is necessary to generate a more pure stock of this variant. Above, we described a plaque purification procedure. Alternatively, if the virus of interest does not produce easily identifiable plaques, the desired variant may be purified by limiting dilution. This approach is essentially a TCID50 in 96-well format, where the virus stock is diluted such that less than 50% of wells are infected. Using this dilution, the same approach as above is taken, in isolating up to 10 individual variants and confirming their sequences. As mentioned, in the best cases, an infectious cDNA clone of the virus strain is available. Isolation of the variant would thus not be necessary. In our experience, fidelity variants are the result of single amino acid substitutions and can thus be generated using simple, commercialized mutagenesis kits such as Quikchange (Agilent). A secondary option is to use a cDNA clone of a closely related strain. However, if a related strain is used, we strongly suggest using both this approach and virus isolation (e.g. plaque purification) because we have found that the same fidelity altering mutation on two closely related viruses will not necessarily have the same effect.
Fidelity and replication. Selection of RNA mutagen resistant variants have resulted in the isolation of both higher and lower fidelity variants with growth characteristics that are similar to their wild type counterparts 4,12,15. Currently, the link between polymerase activity rates and fidelity is not fully understood. In vitro biochemical studies using purified RNA polymerase have shown that higher fidelity variants have slower processing rates, while lower fidelity variants tend to have faster processing 1-3,12. In tissue culture, these differences are not usually evident, suggesting that availability of resources, rather than intrinsic polymerase activity kinetics, is the rate-limiting step. If the fidelity variant replicates with kinetics that are not significantly different from wild type, then a comparison of their mutation frequencies can be directly made. If a very significant change in replication kinetics exists, then the data should be normalized to account for kinetic differences, for example by comparing viruses that have undergone the same number of replication cycles. In our experience, although no significant differences in one-step growth kinetics were observed between wild type and high fidelity variants, we observed that higher fidelity variants consistently titer higher (within 1 log) compared to wild type but they make slightly less RNA (within the same order of magnitude), further suggesting that the genomes they produce contain fewer mutations and are thus, more infectious.
Sample preparation and sequencing. For all steps in these protocols, it is imperative that high-fidelity, proof-reading enzymes are used for PCR and RT-PCR to limit introducing additional mutations since they cannot be distinguished from biologically relevant mutations. It is critical that the virus populations to be compared have been prepared in the same conditions (passage history, tissue culture medium, temperature, RNA extraction method, RT-PCR protocols, etc.) It is also important to ensure that enough starting material was obtained from the RNA extraction such that a strong band is generated by RT-PCR. A 1/100 dilution of the RNA sample should also give a detectable RT-PCR band, indicating that the sample contains sufficient numbers of RNA molecules to avoid representation bias (amplifying the same genome repeatedly). Since mutation frequency is a distribution, one would expect that similar values will be obtained regardless of population size, provided the aforementioned bias is not occurring. As Figure 6 shows, a 105-fold dilution of a virus stock gives a mutation frequency that is not significantly different from the parental stock.
Until the optimal conditions for TopoTA cloning are found, confirm the presence of inserts after blue/white screening, by colony PCR before sequencing. As a control for mutational noise (mutations introduced by RT-PCR and sequencing), clone a PCR product from a plasmid bearing the same viral sequence and/or clone and sequence RT-PCR products of in vitro transcribed RNA corresponding to virus genome (be aware that different in vitro transcription enzymes have different error rates and may not give useful information as to the real background error in your procedure). Some virus sequences may be toxic to bacteria, so it is important to verify this before deciding on the region of viral genome to be sequenced for mutation frequency. In analyzing sequences obtained by TopoTA, note that each clone should contain only one insert/sequence. If a double peak is observed, suggesting a mixed population, it is possible that two neighboring bacterial colonies were selected. It is also possible, although highly unlikely given the low mutation frequencies in bacterial replication, that the mutation was introduced during amplification of plasmid in the bacterial culture. In plaque purified populations, a double peak may represent overlapping plaques, or a virus that is acquiring a new mutation or reverting a mutation during plaque development. Be consistent and decide on whether to count or not count these mutations.
Finally, keep in mind that the mutation frequencies used here are relative values. They are valid only in comparing virus populations grown under the same condition, and sequenced over the same region! They should not be taken as absolute values of mutation rate, or the mutation frequency of the genome as a whole. However, when conditions are controlled, they do permit reproducible, quantitative comparisons of differences in mutation distribution and frequency.
The authors have nothing to disclose.
This work was supported by funding from the Medical and Health Research grant from the City of Paris, the French National grant ANR-09-JCJC-0118-1, and the ERC Starting Grant RNAvirusPopDivNVax Project no. 242719.
Name of the reagent | Company | Catalogue number | Comments |
---|---|---|---|
ribavirin | Sigma | R9644-10MG | |
5-fluorouracil | Sigma | F6627-1G | |
5-azacytidine | Sigma | A2385-100MG | |
MgCl2 | Sigma | M1028-100ML | |
MnCl2 | Sigma | M1787 | |
Trypan blue | Sigma | T8154-20ML | |
TopoTA cloning kit | Invitrogen | 10351021 | |
Quikchange mutagenesis kit | Agilent | 200516 | If a cDNA infectious clone is available |
96-well miniprep kit | Macherey-nagel | 740625 | |
Lasergene, Sequencher | DNAstar, Gene Codes Corporation | www.dnastar.com www.genecodes.com |
Or other alignment software |