Prof. Yukinobu Tohya, Nihon University, Japan, provided the MNV S7 PP3 strain. MNV S7 PP3 was propagated, enumerated, and purified according to the published protocols. RAW 264.7 cells [American Type Culture Collection (ATCC) TIB-71] were used as host cells and were cultured in Dulbecco’s Modified Eagle Medium (Nissui) containing 10% (v/v) fetal bovine serum (Gibco), 0.1% (w/v) NaHCO₃, 2 mM L-glutamine, 10 mM non-essential amino acids, penicillin (100 mg/ml), and streptomycin (100 U/ml).

A virus population was taken from a laboratory strain of MNV S7 PP3 as an ancestor of serial passages with and without lime treatment (Figure 1). MNV populations derived from the nth passage without lime treatment (−lime) and with lime treatment (+lime) were designated as pn + lime and pn−lime virus, respectively (n = 1, 2, 3, 4, and 5).

The lime solution was prepared by dissolving 15 mg of calcium hydroxide (Fujifilm-Wako, Osaka, Japan) in 30 ml of sterilized pure water so that the pH value increased to approximately 12. A 100-μl pn + lime MNV suspension in the culture medium [10⁶–10⁷ plaque forming unit (PFU)/ml] was exposed to 900 μl of lime solution at room temperature (22°C) and inactivated to 10³ PFU/ml. The treatment time for less sensitive populations was prolonged so that the infectious titer would be 10³ PFU/ml after treatment. A 100-μl pn−lime MNV suspension in the culture medium was diluted with the culture medium to create 10³ PFU/ml, in parallel. Each of the virus suspension was inoculated onto the confluent RAW cells cultured in T75 flask containing 25 ml of the culture medium [the multiplicity of infection (MOI) was approximately 10^–4]. The MNV was incubated at 37^°C with 5% CO₂ for 3 days. The cell suspension was subjected to three rounds of freeze-thawing and then transferred to a 50-ml falcon tube. The falcon tube was centrifuged at 10,000 g for 10 min at 4^°C, and the supernatant was passed through a PolyVinylidene DiFluoride (PVDF) membrane filter (0.2-μm pore). The collected MNV suspension, pn+1+lime and pn+1−lime, was exposed to the lime solution or diluted with the culture medium and then inoculated onto the confluent RAW cells (Figure 1). These processes were repeated five times for each passage type. The entire selection process was repeated twice. The first and second selections were referred to as the first and second rounds, respectively. The obtained MNV suspensions were stored at −80^°C until the disinfection experiment and next generation sequencing (NGS) library preparation.

The lime sensitivity of viruses was evaluated on the basis of the reduction of infectious titers by exposure to lime solutions. The lime sensitivity assay was conducted in triplicate for each sample. MNV suspension (100 μl) in the cultured medium was added to a 1.5-ml tube containing 900 μl of the lime solution. Experiments were carried out at 22°C. In the sensitivity assay for the populations in −lime and +lime lineages, samples were periodically taken during the 5-min exposure to the lime solution and immediately diluted with the culture medium. In the sensitivity assay for the plaque-purified isolates and the infectious clones, samples were taken at 4 min because MNV was not always detected after the longer exposure time.

Murine norovirus infectivity was determined by plaque assays in the sensitivity assay for the populations in −lime and +lime lineages. One milliliter of the sample was inoculated onto the confluent RAW cell cultured in the six-well plate and incubated at 37°C with 5% CO₂ for 90 min with periodical swaying. The virus suspension was removed after the 90-min incubation, and 2 ml of the agar medium (1.25% agar) were added. The six-well plates were incubated for 2 days at 37°C with 5% CO₂ and then dyed with 0.03% neutral red solution and incubated for 2–3 h. The neutral red solution was removed and the six-well plates were incubated for 1 h, and then, the plaques were counted.

Murine norovirus infectivity was determined by Median Tissue Culture Infectious Dose (TCID₅₀) assays in the sensitivity assay for the plaque-purified isolates and the infectious clones produced using the plasmid-based reverse genetics system. We performed the TCID₅₀ assays because these clones and isolates did not form clear plaques with unknown reasons. We did not compare the data obtained using the two different methods for titration, and we evaluated the lime sensitivity within the data obtained from identical method. Three-fold serial dilutions of virus were prepared in 96-well plates containing culture medium. Fifty microliters of the RAW cell suspension were inoculated in each well. The 96-well plates were incubated at 37°C with 5% CO₂. After 2 days, a cytopathic effect was observed with light microscopy. TCID₅₀ values were calculated according to the Spearman and Kärber algorithm (Hierholzer and Killington, 1996).

The populations obtained from the +lime and −lime lineages were inoculated onto the confluent RAW cells cultured in six-well plates (MOI of 10^–4). Plaques were formed using the method described above. Well-separated plaque from any other plaques was taken using 1,000 μl of barrier tips cut by a sterilized cutter, placed into a 1.5-ml tube containing 500 μl of culture medium, and vortexed. The isolate suspensions were incubated overnight at 4°C and then passed through a membrane filter (0.2-μm pore). The first plaque-purified virus suspension was subjected to a second round of plaque purification. After the third round of plaque purification, virus suspensions were stored at −80°C until the subsequent analysis. Finally, we obtained six isolates originated from the p5 + lime populations, six isolates originated from the p5−lime populations, and three isolates originated from the ancestor population in the experimental adaptation (Supplementary Figure 1). Each of 15 MNV isolates was amplified in a single passage at MOI of 10^–4 on the confluent RAW cells to increase the viral titer for sensitivity assays and NGS library preparation.

The construct pMNV_A6089G plasmid was generated from pMNV S7 (Katayama et al., 2014) by introducing mutations detected in the ancestor virus and the less sensitive viruses. The In-Fusion system (Takara) was used for fragment insertion into plasmid DNA. PrimeSTAR max DNA Polymerase (Takara) was used for mutagenesis of pMNV S7. Primers used for the In-Fusion reaction and mutagenesis are listed in a table in Supplementary Figure 2.

Infectious clones (referred to as clone MNV) were produced using a plasmid-based reverse genetics system as described previously (Katayama et al., 2014). HEK293T cells cultured in a six-well plate with 2 ml of culture media were transfected with pMNV_A6089G using Lipofectamine 2000 (Invitrogen). Supernatants from transfected cultures were collected at 48 h after transfection and inoculated onto Huh7.5.1 CD300lf cells cultured in six-well plates. The virus was passaged three times in Huh7.5.1 CD300lf cells to obtain a sufficient titer for disinfection experiments. The whole genome sequence of propagated virus was confirmed using next-generation sequencing.

Viral RNA was extracted from the MNV suspension using the QIAamp Viral RNA Mini kit (Qiagen, Germany) according to the manufacturer’s instructions. Ribosomal RNA (rRNA) in RNA suspension was removed using the NEBNext rRNA Depletion Kit (Human/Mouse/Rat) (New England, Ipswich, MA, United States). A 300-bp fragment library was constructed for each sample using the NEBNext Ultra II RNA Library Prep Kit for Illumina according to the manufacturer’s instructions. Samples were added bar codes for multiplexing using the NEBNext Multiplex Oligos for Illumina (Dual Index Primers Set 1). Library purification was done using Agencourt AMPure XP magnetic beads (Beckman Coulter, Pasadena, CA, United States) according to the NEBNext Protocol. The quality of the purified libraries was assessed on an Agilent 2100 Bioanalyzer using a High Sensitivity DNA Kit (Agilent Technologies, Santa Clara, CA, United States). The concentration of the purified library was determined on a Qubit 2.0 fluorometer using the Qubit HS DNA Assay (Invitrogen, Carlsbad, CA, United States) and the NEBNext Library Quant Kit for Illumina. Libraries generated from the populations obtained in serial passages, plaque purified isolates, and infectious MNV were sequenced with paired-end 150-bp reads on an Illumina NovaSeq 6000, paired-end 250-bp reads on an Illumina MiSeq, and paired-end 150-bp reads on an Illumina iSeq (Illumina, San Diego, CA, United States), respectively.

The FASTQ formatted sequence data were analyzed using CLC Genomics Workbench Software version 10.1.1 (CLC Bio, Aarhus, Denmark). The Trim Sequences function trimmed index primer sequences used for library amplification and sequencing. The MNV S7 PP3 genome sequence obtained from National Center for Biotechnology Information (NCBI) (GenBank: AB435515.1) was used as the reference sequence. The trimmed sequence reads were aligned to the reference sequence using the Map Reads to Reference function, and then realigned using the Local Realignment function. Sequence coverages were obtained by the locally realigned read files to SAMtools (Danecek et al., 2021). The consensus sequence of the MNV laboratory strain (the ancestor population in the serial passages) was determined using the Extract Consensus Sequence function, so that it was used as a reference sequence for the variant detection for the populations amplified through the serial passages.

For detection of the single-nucleotide polymorphisms (SNPs) in the MNV isolates, the MNV S7 PP3 genome sequence (GenBank: AB435515.1) was used as a reference sequence. The frequency of SNPs was determined using the Low Frequency Variant Detection function. The threshold value of the error rate was set to 1% to avoid the false positive of variant detection. SNPs with a coverage depth of at least 10 were considered according to the default setting of the CLC Genomics Workbench. Genome-wide and site-specific nucleotide diversities (π_S and π_N) were calculated with SNPGenie (Nelson et al., 2015).

We performed a regularized regression modeling to evaluate the significance of SNPs for the viral lime sensitivity using the NGS data and sensitivity data of the MNV isolates. We considered the following multiple linear regression model:

where y is the logarithm of the inactivation ratio [the log reduction values (LRVs)] of viruses at 4 min after exposure to lime solutions, β is a regression coefficient, and ε is the residuals. The vectors y and ε consist of n elements that correspond to the number of samples with a set of the LRVs and SNPs. The vector β consists of p + 1 elements that are allocated to p regression coefficients and one intercept. X is a n × (p + 1) design matrix that includes the frequency of SNPs at p polymorphic sites in n samples. We assumed that the value of β_i indicated the contribution of the SNPs at site i in the viral tolerance to lime treatment. The Elastic Net (EN) method (Zou and Hastie, 2005) was employed to estimate β in Eq. 1 as follows:

where λ(λ > 0) is a regularizing parameter and α(0 < α < 1) is a penalty weight for regularization. The EN regression was performed by the glmnet package (Hastie and Qian, 2016) on the R software (version 3.6.1) (R Development Core Team, 2019). The value of λ was determined by k-fold cross-validation to minimize the mean square value plus standard error (minMSE+1SE).

Inactivation of the +lime and −lime populations is shown in Figure 2 as a function of treatment time. The LRVs at 5 min of the p1 + lime and p1−lime were 3.0–4.2 and 2.5–3.5, respectively (Figure 3). The LRVs of the −lime populations did not change across the five passages (p = 0.55 and p = 0.15 in the first round and the second round, by linear regression non-zero slope T-test; Figure 3). Meanwhile, the LRVs of the +lime populations got smaller with each passage (p = 0.0048 and p = 0.031 in the first and second round, by linear regression non-zero slope T-test; Figure 3). The mean LRVs were significantly different between the +lime and −lime populations (p = 2.5 × 10^–5 and p = 5.1 × 10^–6 in the first and second rounds, respectively, using the Mann–Whitney U test).

The depth of coverage was greater than 300 reads across the entire coding region for 99.9% of the sequencing libraries. We estimated synonymous nucleotide diversity (π_S) and non-synonymous nucleotide diversity (π_N) for the three protein coding genes (ORFs 1, 2, and 3) separately (Figure 4).

For ORF1 in the −lime populations, the π_S and π_N were within the 10^–4 order, and the π_N /π_S values were close to 1, whereas, in the +lime populations, π_S increased and π_N decreased over the course of serial passages (Figure 4). For ORF2 in the +lime and −lime populations, the π_N /π_S values were less than 1, whereas π_N got larger with each passage in the −lime populations (Figure 4). In the +lime populations, in contrast, π_N got close to zero after passage 2, whereas the unique non-synonymous nucleotide change (A6089G) accounted for over the 90% of the virus populations (Figure 5). ORF3 in the +lime populations had no synonymous mutations, and π_S values were zero after passages 2 and 3, in the first and second rounds, respectively. Meanwhile, non-synonymous mutations were detected in all populations, and accordingly the π_N /π_S values were greater than 1 (Figure 4).

Single-nucleotide polymorphisms called at greater than 1% frequency, and at least 300 reads of coverage were used for the analysis. Four non-synonymous mutations were detected in the laboratory strain with > 99% frequency: G1564A (ORF1), C6289G (ORF2), T6605C (ORF2), and T6751C (ORF3). The consensus sequence of the ancestor population was used as the reference sequence for the SNPs detection from populations obtained in the serial passages.

T561C and T7279C accounted for over 35% of the −lime populations (Figure 5). T561C and T7279C were recurrences of the MNV S7 PP3; the origin of our laboratory strain and, thus, the two SNPs could be common variants in the viral population of the MNV S7 PP3. They were fixed as a consequence of several bottleneck events. The trajectories of the C7215T in the first round and four SNPs (C829T, A733G, A78C, and C329T) in the second round were parallel with T561C, whereas those of the two SNPs (C7358T and T6202C) in the first round and TC7385T in the second round were parallel with T7279C (Figure 5). The parallel trajectories with the common variants indicated that those SNPs were the consequence of genetic hitchhiking with T561C and T7279C.

In the +lime lineages, one synonymous SNP (A2519G) and one non-synonymous SNP (A6089G) arose in the populations with high frequency (Figure 5). In the first round, the frequency of A6089G increased from less than 0.5% to greater than 90% by passage 2, and increased to 98.8% by passage 5 in the first round. In the second round, the frequency of A6089G was less than 35% by passage 2 and increased to 96.6% by passage 5. The mutation A6089G associated with the amino acid substitution of lysine (K) to arginine (R) at site 345 on VP1. We confirmed the location of K345R by assigning the nucleotide changes to the sequence of MNV S7 PP3 reported by Nelson et al. (2018) and visualized using PyMOL (Schrödinger LLC, 2015). The K345R is located on the P2 subdomain, specifically on the opposite side of the contact site to the receptor CD300lf (Supplementary Figure 3).

A2519G and A6089G did not demonstrate parallel trajectories indicative of genetic hitchhiking. In the first round, the frequency of A2519G increased to 66.4% by passage 3, and, similarly, in the second round, the frequency of A2519G increased to 54.7% by passage 4 (Figure 5). The trajectories of A2519G were reproductive in the populations from the +lime lineages, which indicated that A2519G frequency was increased due to the lime treatment.

The frequency of A6089G and A2519G was associated with the greater tolerance of virus populations to the lime treatment, and thus, we hypothesized that those two SNPs contributed to the lime sensitivity of MNV. To verify this hypothesis, we increased the sample size by plaque isolation from the sensitive and less sensitive MNV populations and evaluated the association between the SNPs frequency and the LRVs of the isolates.

The LRVs were smaller in the isolate originated from the less sensitive populations (J, K, L, M, N, and O) compared to the other isolates (Supplementary Figure 4). The less sensitive six isolates shared A6089G, and the LRVs were significantly smaller in the isolates with the A6089G mutation (p = 0.027 by Mann–Whitney U test; Supplementary Figure 4). More SNPs were identified in the 15 isolates compared to the founder virus populations (Supplementary Figure 5). Although the 15 isolates shared the five major SNPs (G1564A, C6289G, T6605C, T6751C, and C7112T), the appearances of polymorphic sites were not consistent across the isolates (Supplementary Figure 5). Less sensitive populations were purified in the plaque-to-plaque transfer, whereas the genetical heterogeneity increased during the replication in the cultivation conducted before the NGS library preparation because mutations could be induced in each replicate. The A6089G remained in the isolates from the less sensitive populations, but A2519G was not detected among all the less sensitive isolates (Supplementary Figure 5). In this study, the coefficient values in the EN regression indicate the significance of SNPs for the viral tolerance to lime treatment. The coefficient value that corresponded to A6089G was not zero, but that for other SNPs was estimated to be zero (Supplementary Figure 6), which indicated that A6089G was important for the viral tolerance in an alkali environment, whereas A2519G was not.

The sole effect of A6089G on the lime sensitivity was verified by comparing the LRVs of infectious clones produced using a plasmid-based reverse genetics system (clone MNV) and ancestor virus used for the serial passage with lime treatment. Those two populations shared non-synonymous mutations on ORF2 (C6289G and T6605C) and ORF3 (T6751C), whereas clone MNV had a specific mutation on ORF2 (A6089G). The LRVs were significantly smaller in clone MNV than ancestor virus (p = 0.004 by Mann–Whitney U test; Figure 6), which indicated that A6089G contributed solely to the viral tolerance to lime treatment.