Norovirus Polymerase Fidelity Contributes to Viral Transmission In Vivo

Virus replication fidelity and hence the intrahost genetic diversity of viral populations are known to be intricately linked to viral pathogenesis and tropism as well as to immune and antiviral escape during infection. In this study, we investigated whether changes in replication fidelity can impact the ability of a virus to transmit between susceptible hosts by the use of a mouse model for norovirus. We show that a variant encoding a high-fidelity polymerase is transmitted less efficiently between mice than the wild-type strain. This constitutes the first experimental demonstration that the polymerase fidelity of viruses can impact transmission of infection in their natural hosts. These results provide further insight into potential reasons for the global emergence of pandemic human noroviruses that display alterations in the replication fidelity of their polymerases compared to nonpandemic strains.

IMPORTANCE Virus replication fidelity and hence the intrahost genetic diversity of viral populations are known to be intricately linked to viral pathogenesis and tropism as well as to immune and antiviral escape during infection. In this study, we investigated whether changes in replication fidelity can impact the ability of a virus to transmit between susceptible hosts by the use of a mouse model for norovirus. We show that a variant encoding a high-fidelity polymerase is transmitted less efficiently between mice than the wild-type strain. This constitutes the first experimental demonstration that the polymerase fidelity of viruses can impact transmission of infection in their natural hosts. These results provide further insight into potential reasons for the global emergence of pandemic human noroviruses that display alterations in the replication fidelity of their polymerases compared to nonpandemic strains.
KEYWORDS: RNA polymerases, noroviruses, polymerase fidelity, quasispecies, virus transmission R eplication fidelity is a major trait influencing the pathogenesis of RNA viruses. The high error rates of viral RNA polymerases during genome replication result in genetically diverse virus populations, known as quasispecies, which are extremely flexible and adaptable to dynamic host environments and selection pressures (reviewed in reference 1). Alterations in replication fidelity typically lead to reduced virus infectivity in the host (2)(3)(4)(5). This evidence has led to studies on the development of antiviral strategies based on the manipulation of virus replication fidelity. Such strate-gies have involved either the use of mutagenic compounds that can drive viral extinction through increased replication error rates (lethal mutagenesis; reviewed in references 6 and 7) or the design of mutant viruses with altered fidelity causing an attenuated phenotype in vivo, which can be used for the development of live vaccine candidates (4,8,9).
We recently demonstrated the antiviral activity of a novel mutagenic nucleoside, namely, T-705, or favipiravir, against persistent norovirus infections in vivo (10), providing the first link between norovirus polymerase fidelity and norovirus pathogenesis. Human noroviruses (HuNoVs) are considered the main cause of diarrheal episodes and foodborne gastroenteritis globally (11)(12)(13), with an estimated Ͼ500 M infections annually and Ͼ200,000 associated fatalities (13). Mortality rates are especially high in children under the age of 5 living in low-income countries, although other cohorts of vulnerable patients include elderly and immunocompromised populations (11,14). HuNoV infections have also been linked to effects on an increasing number of other severe disorders such as exacerbation of inflammatory bowel disease, ulcerative colitis, and life-threatening chronic diarrhea (15,16). Despite the significant impact on global health and the elevated economic losses associated with HuNoVs (11), there are no licensed vaccines or antiviral drugs available for the treatment of disease and control of outbreaks.
Since 2002, a sharp increase in the global incidence of norovirus outbreaks that is associated with the emergence of genogroup II.4 (GII.4) pandemic strains has been reported (17,18). Evidence suggests that emerging HuNoV GII.4 pandemic strains transmit from person to person more efficiently than the previously dominant genogroup I strains (19,20). Recent studies have demonstrated that the viral RNAdependent RNA polymerases from pandemic HuNoV GII.4 strains display reduced replication fidelity and increased intrahost diversity in their resulting viral populations in comparison to nonpandemic strains (21). A newly emerging HuNoV GII.17 isolate, first identified in China, seems to be outcompeting the HuNoV GII.4 in some parts of Asia, resulting in increased outbreaks since the winter season of 2014 to 2015 (22). Recent studies suggest that this novel variant may be spreading globally, which could lead to the replacement of the pandemic genotype GII.4 by GII.17 (22)(23)(24)(25)(26). Fitting with the hypothesis that fidelity is linked to rapid emergence of norovirus isolates, the GII.17 isolate displays evolutionary rates at least 1 order of magnitude higher than those seen with GII.4 (25). Despite this evidence, a possible relationship between the reported reduced fidelity (and greater genetic diversity) in pandemic HuNoV GII.4 and the emerging GII.17 isolates and increased transmissibility has yet to be examined.
Here, we have assessed the influence of polymerase fidelity and genetic diversity on norovirus infectivity and transmission in vivo using a persistent strain of murine norovirus (MNV) as a model. MNV provides a robust surrogate system for studying HuNoV fidelity due to its high replication rates in cultured cells and to the availability of efficient reverse genetics systems and small-animal models (10,27,28). Here we have identified a high-fidelity MNV polymerase mutant (I391L) which shows delayed replication kinetics during the establishment of a persistent infection in vivo. Despite being able to successfully establish an infection and persist, the high-fidelity I391L mutant showed reduced mouse-to-mouse transmission compared to wild-type (WT) virus. Significantly, the transmission of a mutant with in vivo replication kinetics similar to those of the I391L mutant, but with WT-like polymerase fidelity (S313T), was unaffected, suggesting that the reduced transmission of I391L was linked to fidelity. These results support the role of genetic diversity as a significant trait influencing norovirus infectivity and transmission in vivo and provide new insight into the potential role of polymerase fidelity in the emergence of norovirus strains.

RESULTS AND DISCUSSION
Generation of MNV polymerase fidelity mutant candidates. With the aim of identifying possible MNV polymerase (NS7) fidelity mutants, we carried out site-directed mutagenesis of residues previously identified as fidelity determinants in the ortholo-gous positions of other viral RNA-dependent RNA polymerases. Given the substantial amount of data available on picornavirus polymerases and their relatively close genetic relationship with caliciviruses, we designed mutations in the MNV polymerase based on previous studies with poliovirus (PV), foot-and-mouth disease virus (FMDV), and coxsackievirus (CV) (3)(4)(5)(29)(30)(31)(32). To date, the most paradigmatic example of viral attenuation as a result of increased fidelity is the PV polymerase (3D pol ) G64S mutation, which resulted in a loss of pathogenicity with restricted virus tissue tropism in a lethal mouse model of PV infection (3,4). While the norovirus polymerases do not share a great degree of structural homology with the PV 3D pol in the domain surrounding the G64S position, MNV NS7 residue R77 (R74 in HuNoV NS7) lies within this domain and similarly establishes multiple hydrogen bond interactions with residues in the active site (30,33,34). The G64S change in PV 3D pol is thought to alter fidelity by changing the hydrogen bond network, indirectly leading to subtle conformational changes of catalytic site residues (8,30). Therefore, we examined the impact of substitutions at amino acid R77 on virus replication by reverse genetics (Table 1). All the mutations studied were lethal in the context of the MNV infectious clone, suggesting that this position is critical for (

3) No
a Substitutions generated in this study were based on fidelity changes found in other viruses (4,5,29,31,32): G64S in PV; P169S in FMDV; and S164P, S298T, and A372V in CV. b Effect on virus polymerase fidelity caused by substitutions shown in the first column. c MNV NS7 polymerase mutants generated based on the fidelity of the residues shown in the first column. Mutants V330I and T35I were constructed after isolation of these mutations in MNV populations treated with increasing concentrations of ribavirin (see Fig. S1 in the supplemental material). Other variants of position 330 (V330A or -S) were also prepared based on the isolation of V330I during ribavirin treatment. These mutants were not viable. d MNV strain. NS7 mutations were introduced in plasmids containing full MNV-1 or MNV-3 genome sequences. Hence, "(1)" denotes MNV-1, "(3)" denotes MNV-3, and "(1, 3)" denotes those mutations tested in both strains. For the experiments represented in the figures, only MNV-3 variants were used. e The viability of NS7 mutants was assessed by reverse genetics after recovery in BHK-21 (or BHK-21-derived BSR-T7) cells followed by 3 serial passages in RAW264.7 cells as described in reference 27. f Phenotype in cell culture. "Modest increase or decrease" (in virus replication) indicates statistically significant changes in virus titers of Յ1 log 10 (two-way analysis of variance [ANOVA] test) at any given time point during replication kinetics. "Significantly decreased" in virus replication (K174S) refers to statistically significant changes in virus titers of~2 log 10 at an early replication time point. In this column, "P72S" and "E75S" refer to the corresponding NS7 mutants in the MNV-3 genome context. g Phenotype in vivo. "Modest increase" (in virus replication) in vivo indicates a highly significant change in the titer of virus shed in feces (Ͼ1 log 10 ; P Ͻ 0.001; twoway ANOVA). "Significantly decreased" indicates highly significant changes in levels of both viral RNA and infectious virus shed in feces (P Ͻ 0.001; ANOVA test). "Modest decrease (viral RNA)" indicates significantly lower RNA levels but not virus titers shed in mouse feces. In this column, "P72S" and "E75S" refer to the corresponding NS7 mutants in the MNV-3 genome context. polymerase function (Table 1). In order to identify tolerated MNV variants in this region of NS7, we mutated several neighboring amino acids of R77 that were not predicted to establish as many direct hydrogen bonds with amino acids involved in the active site and that would be predicted to result in more-subtle rearrangements of motif A. A series of positions, including V65, L70, P72, E75, and G78, were identified as possible tolerant sites within NS7 (Table 1). We also identified homologous positions in MNV NS7 based on other fidelity variants isolated in related picornaviruses, such as CV and FMDV, namely, K174, P175, S313, and I391 (Table 1). We successfully recovered mutants encoding NS7 substitutions P72S, E75S, K174R, K174S, S313T, I391L, and I391V, whereas the remaining mutants were nonviable ( Table 1). The recovery of K174S and S313T yielded lower virus titers than the MNV WT, suggesting reduced replication fitness associated with these polymerase changes ( Table 1). The stability of the viable mutants was confirmed by sequencing virus populations obtained after 4 serial passages in cell culture, with no additional changes identified within the NS7-coding region for any mutant (data not shown).
Since the propagation of different RNA viruses in cell culture in the presence of ribavirin generally leads to the selection of fidelity mutants (e.g., PV G64S, FMDV P169S, and CV A372V, among others) (29,31,32), we similarly investigated whether the passage of MNV in the presence of ribavirin also selected for changes in NS7 conferring higher polymerase fidelity. We identified substitutions T35I and V330I in NS7, which emerged in two lineages of MNV that had been repeatedly passaged in the presence of ribavirin (see Fig. S1 in the supplemental material). No substitutions were found in the NS7 RNA polymerase of MNV passaged in parallel in the absence of the drug. These mutations were introduced in the MNV infectious cDNA clone, and viable mutants were recovered (Table 1).
Substitution I391L results in increased polymerase fidelity. As an indirect approach to identify any fidelity alterations in the viable MNV mutants, we tested the sensitivity of the viable mutants to the following three different nucleoside analogues with mutagenic activity against a variety of RNA viruses, including MNV: favipiravir, 5-fluoruracil, and ribavirin (3,5,10,35). High-fidelity mutants are typically less sensitive to mutagenic treatment, while mutator (lower fidelity) polymerase variants are typically more sensitive (5,29,32). The I391L mutant had reduced sensitivity to all three mutagenic compounds, while the S313T mutant was more sensitive to 5-fluorouracil and favipiravir (Fig. 1). I391V sensitivity varied depending on the mutagenic compound, and the I391V mutant was more sensitive to 5-fluorouracil and favipiravir and less sensitive to ribavirin (see Fig. S2 in the supplemental material). The remaining mutant viruses showed no significant differences in their sensitivities to any of these mutagenic compounds compared to the WT (see Fig. S3).
Given this pattern of sensitivity to different mutagenic nucleosides, substitutions I391L, I391V, and S313T were further examined for any impact on replication fidelity. Several studies have demonstrated that alterations in polymerase fidelity typically lead to changes in the genetic diversity of the resulting virus population (4,5). With this in mind, we measured the genetic diversity in MNV NS7 mutant populations after their serial passage in cell culture in the absence of any mutagenic treatment. We found that the genetic diversity in the I391L NS7 mutant virus population was~3-fold lower than in the WT NS7 population, suggesting that the I391L substitution results in a higherfidelity NS7 ( Fig. 1D; see also Table S1 in the supplemental material). The NS7 mutants S313T and I391V did not show any significant difference in their mutation frequencies compared to the WT, suggesting that their increased sensitivity to mutagens was not related to any impact on polymerase fidelity ( Fig. 1D; see also Fig. S2D and Table S1). Hence, the S313T and I391V mutants display WT-like fidelity phenotypes.
The high-fidelity I391L NS7 mutant shows delayed replication in vivo but not in cell culture. We investigated whether the increase in fidelity conferred by I391L resulted in any phenotypic alteration of norovirus replication in vivo, as previously documented for other viruses (4,9). Prior to this, a more detailed characterization of the replication kinetics of the I391L mutant and the remaining NS7 mutants was performed in cell culture. The I391L mutant showed no significant difference from WT MNV in terms of the virus yields obtained following viral recovery by reverse genetics from a full-length cDNA construct ( Fig. 2A). Similarly, no replication defect was observed at either a low multiplicity of infection (MOI) (0.01 50% tissue culture infective dose [TCID 50 ]/cell; Fig. 2B) or a high MOI (5 to 10 TCID 50 /cell; Fig. 2C and D) for immortalized macrophage cells. However, I391L showed significantly delayed kinetics of replication in vivo as reflected in decreased viral RNA levels and lower titers of infectious virus secreted in the feces of the mice (Fig. 3A and B). This observation suggests that the higher replication fidelity of I391L results in reduced fitness during the establishment of a persistent infection in vivo. The early defect in I391L replication was recovered by day 3 postinfection (Fig. 3A and B) and then remained comparable to the WT level for the duration of the experiment (28 days; data not shown). Animals infected with a 10-fold-higher viral titer also showed delayed replication kinetics for I391L, although viral loads similar to the WT levels were again observed by day 2 (Fig. 3C). The observed lower virus titers and RNA levels in feces of mice infected with I391L were associated with a trend toward lower viral RNA levels in different tissues harvested at 48 h postinoculation, although the differences observed did not reach statistical significance (Fig. 3D).Taking the data together, the reduced replication kinetics of the MNV I391L NS7 mutant were apparent only in vivo and not during replication in cell culture. This is in line with observations of other reported fidelity mutants (4,32).
Of the remaining NS7 mutants, S313T and K174S also displayed decreases in replication in vivo to levels similar to those seen with I391L ( Fig. 3A and B; see also Fig. S4 in the supplemental material). However, in contrast to I391L, both S313T and K174S also displayed small but significant replication defects in cell culture. Both the S313T and K174S mutants yielded reduced virus titers in virus recoveries by reverse genetics ( Fig. 2A; see also Fig. S4A) that were typical of viruses with a replication defect. K174S also showed significantly delayed kinetics during infection in RAW264.7 cells (see Fig. S4B). In contrast, the S313T mutant displayed minimally but reproducibly lower virus titers at 16 h postinfection during kinetics assays performed with both high and low MOI's ( Fig. 2B to D). The mutants I391V and T35I exhibited modest replication defects in vivo, although to a lesser extent than I391L (see Fig. S4 and S5). Surprisingly, significantly higher numbers of genome copies were secreted in mice infected with the P72S mutant, with slightly increased replication also being observed in cell cultures (see Fig. S6), suggesting that this substitution confers a replication advantage to MNV. Given that P72S did not differ in its sensitivities to mutagenic nucleosides, we hypothesized that the observed phenotype was not due to any impact on fidelity and it was therefore not characterized in more detail.
We subsequently investigated whether the reduced replication of I391L in mice could be rescued by a chemical expansion of the quasispecies diversity by favipiravir treatment. We have previously demonstrated that favipiravir elicits antiviral mutagenic activity associated with increased genetic diversity of the viral quasispecies in vivo (10). Treatment of mice with favipiravir resulted in decreased virus titers shed by both WT MNV and I391L MNV-infected animals (Fig. 4), suggesting that the dose of favipiravir used elicits antiviral activity against both viruses. Despite this, I391L displayed greater tolerance than the WT to favipiravir during infection in vivo, in agreement with observations in cell culture (Fig. 1B).

The high-fidelity MNV NS7 I391L mutant is transmitted less efficiently than the WT and the S313T mutant in vivo.
Several studies have found that viral genetic diversity contributes to organ and tissue tropism expansion during viral infections by facilitating avoidance of or escape from different physiological barriers (3,4,36). However, there is limited information on whether genetic diversity contributes to virus transmission between susceptible hosts. Recent studies suggest that reduced polymerase replication fidelity may be a significant determinant linked to the pandemic expansion of recent HuNoV GII.4 isolates (21) and that person-to-person transmission is more efficient for GII.4 pandemic strains than for nonpandemic isolates (19,20). To determine whether the high-fidelity I391L NS7 mutant had altered transmission in vivo, we established an MNV transmission model whereby one mouse per cage was orally inoculated and transmission to five sentinel littermates was monitored by quantifica-  (39). The presence of the desired mutations in the viral genome was confirmed by sequencing. Data were analyzed by two-way ANOVA (*, P < 0.05; **, P < 0.01; ***, P < 0.001). gRNA eq, genomic RNA equivalents. (C) Animals were infected as described for panel A, but the mice were inoculated with a dose of 3 ؋ 10 3 TCID 50 MNV per mouse (n ‫؍‬ 5). Data represent results from a two-way ANOVA test (***, P < 0.001). (D) Viral RNA levels detected in various tissues collected from mice inoculated with 3 ؋ 10 2 TCID 50 of WT MNV or the NS7 polymerase mutants at 48 h postinfection. MLN, mesenteric lymph node (two-way ANOVA test results were nonsignificant). In all panels, the dashed line represents the detection limit. The apparent absence of SEM bars for some values in panels B and C indicates that the bar is smaller than the respective symbol.
tion of viral RNA in the feces. Since a defect in transmission could be attributed to delayed replication in I391L-infected mice, we also included the S313T mutant as it exhibits WT-like polymerase fidelity (Fig. 1D) but slower replication in vivo, to an extent similar to that seen with the I391L mutant ( Fig. 3A and B). We found that transmission rates for the I391L mutant were significantly lower than those found for the WT and S313T (Fig. 5), suggesting that genetic diversity is linked to efficient virus transmission between hosts. Reduced transmission in the I391L-infected group was associated with lower viral RNA levels shed in the feces of sentinel animals (see Fig. S7 in the supplemental material). Hence, changes in polymerase fidelity may impact the rate of transmission of noroviruses within a susceptible population of naive hosts, providing additional support to recent studies indicating that the pandemic HuNoV GII.4 isolates possess polymerases with lower fidelity (21,28).
In this study, we demonstrated that a high-fidelity norovirus mutant has reduced replication in vivo accompanied by reduced rates of transmission between hosts. This  report builds on our previous investigations demonstrating that norovirus replication fidelity and intrahost genetic diversity influence viral pathogenesis (10). Together, these results are highly significant in light of understanding the influence of replication fidelity on factors that affect the emergence of new norovirus variants with pandemic potential. Given the recent development of potential cell culture propagation methods for HuNoVs (37,38), the identification of fidelity determinants in NS7 could also contribute to the design of genetically stable attenuated vaccines, as has been proposed for other viruses (8). Animal infections and antiviral treatment. Male C57BL/6 mice (4 to 5 weeks old) were orally infected with known amounts of MNV-3 as previously described (10) and as stated in the text. Fecal samples were collected at different time points, and the presence of infectious particles and viral RNA was determined as described in reference 10. For the determination of viral RNA levels in tissues, animals were euthanized at 48 h postinfection and the cecum, colon, and mesenteric lymph node harvested.

Ethics
Mice infected with MNV-3 were treated with favipiravir following protocols previously described in reference 10. Briefly, 1 week after inoculation, persistently infected animals were treated twice a day by oral gavage with favipiravir (600 mg/kg of animal body weight/day) for 4 days. Fecal samples were collected at different times during the course of treatment and virus titers and RNA levels quantified.

Viral transmission experiments.
To determine MNV transmission in vivo, we scored the proportion of infected mice within the population of noninoculated animals in contact with mice infected by oral-gavage inoculation. One mouse was inoculated per cage and was housed with five contact uninfected sentinel mice. We determined the presence of MNV RNA in the feces of naive mice at different time points postinfection (days 1, 2, 3, 4, and 6).

Cells, infections, and reverse genetics recovery of viruses.
Procedures for the cultivation of cells and MNV infections have been previously described (10). RAW264.7 murine leukemia macrophage cells were used for the propagation and titration (TCID 50 assay) of the MNV-1 and MNV-3 mutants used in this study. All the different cell lines were cultured in Dulbecco's modified Eagle medium (DMEM) with 10% FCS, 100 U/ml penicillin, and 100 mg/ml streptomycin (complete DMEM) and were maintained at 37°C with 5% CO 2 .
The wild-type strain and NS7 mutants MNV-1 and MNV-3 used in this study were obtained after reverse genetics recovery of infectious virus as previously described (10,27). Briefly, BHK-21 cells were transfected with plasmids containing a T7 promoter followed by full-length genomic MNV cDNA encoding different NS7 mutants. Then, transfected cells were infected with helper fowlpox virus expressing recombinant T7 polymerase (FPV-T7) as previously described in reference 27. The resulting population was titrated and used as a passage 0 stock. All the experiments carried out in this study involved NS7 mutants in the context of an MNV-3 genome, with the exception of a limited number of NS7 mutant virus recoveries described in Table 1 that were generated in the context of an MNV-1 genome.
Cell culture infections in the presence of mutagenic compounds. Experiments performed to determine levels of sensitivity to different mutagenic compounds were carried out using RAW264.7 cells. A total of 4 ϫ 10 5 cells per well were seeded on 24-well plates, incubated for 3 h at 37°C and 5% CO 2 to allow attachment to the plate, and then infected with different MNV mutants at an MOI of 0.01 TCID 50 /cell. Cells were incubated at 37°C and 5% CO 2 for 1 h; the supernatants were then removed and cells washed once with complete DMEM, before addition of 1 ml of complete DMEM containing mutagenic compounds at the indicated concentrations. Infections were harvested at 42 h and virus progeny released through 2 consecutive cycles of freeze-thawing. Viral RNA extraction, reverse transcription-PCR (RT-PCR) amplification, quantitative PCR (qPCR), and mutation frequency analysis of virus populations. Viral RNA was extracted from 100 l of viral samples (either supernatant from lysed infected cultures or phosphate-buffered-saline [PBS]resuspended feces from animals) using a GenElute RNA miniprep purification system following protocols provided by the manufacturer. Viral RNA was quantified using a two-step qPCR approach following protocols described previously (27).
For the calculation of mutation frequency in any virus population, we followed protocols previously described (10). We analyzed the mutation frequency of mutant and WT populations obtained after eight passages in RAW264.7 cells to allow sufficient rounds of replication and accumulation of mutations to facilitate the quantitative analysis. Briefly, 4 l of purified RNA was subjected to reverse transcription in a 20-l final volume using SuperScript III (Roche) as indicated by the manufacturer. cDNA (3 l) was then PCR amplified using high-fidelity Hot Start KOD DNA polymerase (Toyobo) and primers spanning MNV-3 genomic positions 951 to 3395. Positive PCR band products were excised from an agarose gel and purified with EconoSpin columns (Epoch) and then directly ligated to plasmid PCR Blunt using a Zero Blunt PCR cloning kit (Life Technologies). Positive Escherichia coli colonies were identified by PCR screening using primers flanking the vector-cloning site and GoTaq polymerase (Promega). The resultant PCR products corresponding to individual MNV cDNA clones were sequenced, and the mutation frequency in each population was calculated.
Statistical analysis. Statistical significance was examined using GraphPad Prism as described in the figure legends. For statistical analysis of mutation frequencies, a Mann-Whitney test was used that compares the ranked scores of the numbers of mutations found in individual clones grouped by population (Mann-Whitney test; 0.01 Ͻ P Ͻ 0.05, **), as previously described (5).

ACKNOWLEDGMENTS
We particularly thank Larissa Thackray (Washington University) for helpful discussions and for performing pilot animal experiments during the early stages of the project. This work was supported by a Wellcome Trust Senior Fellowship to I.G. (grant reference 097997/Z/11/Z).

FUNDING INFORMATION
This work, including the efforts of Ian Goodfellow, was funded by Wellcome Trust (097997/Z/11/Z).