Multiple levels of triggered factors and positive effect of cell-to-cell bottleneck in mutation repair of RNA viruses

Debilitating mutation in RNA viruses can be repaired via different mechanisms, while triggering factors of mutation repair were poorly understood. In this study, multiple levels of triggering factors of mutation repair was identified based on genetic damage of tRNA-like structure (TLS) in cucumber mosaic virus (CMV). TLS mutation in different RNAs of CMV distinctively impacted the pathogenicity and mutation repair. Relative quantity defect of RNA2 or quality defect of RNA3 resulting from TLS mutation was correlated with high rate of mutation repair, and TLS mutation of RNA1 failed to be repaired. However, TLS mutation of RNA1 can be repaired in the mixed inoculation with RNA2 having pre-termination mutation of 2b or at the low dose of original inoculation, especially around dilution end-point. Taken together, TLS mutation resulting into quality or quantity defect of viral genome or TLS mutation at low dose around dilution end-point was inclined to be repaired. In addition, different levels of mutation repair of TLS necessarily required the cell-to-cell movement, which implied the positive effect of cell-to-cell bottleneck on evolution of low-fitness virus, a phenomenon opposite to the Muller ratchet. This study provided important revelations on virus evolution and application of viral mild vaccine. Author summary Due to the low-fidelity of replicase, debilitating RNA viruses can be repaired through different mechanisms, which implied the resilience of RNA viruses. In this study, we identified multiple levels of triggered factors and occurrence occasion of mutation repair using the divided genome of CMV, which contained conserved cis-element tRNA-like structure (TLS) at the 3’end. TLS mutation of different RNA in CMV presented different rate of mutation repair from 0-80%. TLS mutation resulting into genomic quality or quantity defect or at low dose around dilution end-point was inclined to be repaired. However, all above types of mutation repair necessarily required cell-to-cell movement, which presented the positive effect of cell-to-cell bottleneck on virus evolution and increased fitness of low-fitness RNA viruses. It is an opposite phenomenon to the Muller ratchet, in which bottleneck always decreased the fitness of viruses. Except to identify the triggering factors of mutation repair in RNA viruses, this study also provided important revelations on creation and application of mild vaccine based on RNA viruses.


10
Many factors caused different type of mutations in RNA viruses, which have evolved 11 several tools to maintain their genome integrity due to the infidelity of the replicative 12 machinery (Agol and Gmyl, 2018). Negative selection can eliminate less-fit or lethal 13 variants, while some debilitating mutations in RNA virus can be repaired or 14 remodeled by reversion or compensatory mutation through RNA recombination even 15 in trans complementation from other coinfecting viruses (Barr and Fearns, 2010; Agol 16 and Gmyl, 2018). RNA Recombination played essential role on the genetic evolution 17 as well as genetic stability of all living organism, which is responsible for the 18 rearrangement of viral genes, the repair of debilitating mutations, and the acquisition 19 of nonself sequences (Sztuba-Solińska et al., 2011). Based on the infidelity of the 20 replicative machinery, genome of well-adapted RNA viruses can maintain their 21 identity, whereas genome of weak-adapted or debilitated RNA viruses are unstable 22 and can be repaired even become higher-fitness variant (Agol and Gmyl, 2018). RNA 23 recombination is one of the strongest forces shaping robustness and resilience of RNA 24 viruses, which are two connected aspects of viral evolution, although robustness and 25 resilience of RNA viruses seems to be conflicative (Sztuba-Solińska et al., 2011; Agol 26 and Gmyl, 2018). 27 Two models have been proposed to interpret RNA recombination: the potential  Agol, 2005). The latter is the most widely accepted model for 32 3 recombination. Replicative-related RNA recombination is mediated by copy-choice 1 mechanism, which occurs at specific crossover sites between different RNA templates 2 possibly involving in related viruses, distantly related viruses and even host RNAs 3 (Figlerowicz and Nagy, 1997; Sztuba-Solińska et al., 2011). In copy-choice 4 mechanism of RNA recombination, the specific crossover site between two RNA 5 templates and error-prone replicase are substrate basis, which ensures the potential 6 occurrence of RNA recombination (Sztuba-Solińska et al., 2011). Each RNA virus 7 seems to be capable of recombining since RNA recombination does not require 8 complicated machinery. RNA recombination on different RNA viruses will promote 9 viral evolution, while RNA recombination on debilitating viral genome may cause 10 mutation repair. However, factors triggering mutation repair or determining the 11 frequency of mutation repair via RNA recombination was unclear. 12 In order to evaluate the factor triggering occurrence and frequency of mutation 13 repair mediated by RNA recombination, deleterious mutation and the specific 14 crossover site among different RNAs are required simultaneously. In this study, the 15 divided genome of cucumber mosaic virus (CMV) was used to evaluate the 16 relationship between debilitating genome and mutation repair mediated by RNA 17 recombination, because RNA1, RNA2 and RNA3 of CMV have identical 3'UTR 18 containing conserved tRNA-like structure (TLS), which was essential for replication 19 and had some core cis-elements such as caacg loop (Sivakumaran et al., 2000).

20
Identical mutations on TLS in RNA1, RNA2 or RNA3 presented quite different 21 frequency of mutation repair via RNA recombination. Further assay suggested that 22 mutation repair can be triggered by not only quality or quantity defect of genome 23 caused by mutation but also debilitating genome at low original dose. In addition, all 24 these types of mutation repair on TLS mutation required cell-to-cell movement. These 25 results provided new revelation on viral evolution and mild vaccine application. 26

27
TLS mutation in different RNAs of CMV presented remarkable difference of 28 repair frequency mediated by copy-choice type of RNA recombination 29 In order to identify the relationship between mutation repair and debilitating genome, 30 debilitating genome was required. In this study, the tRNA-like structure (TLS) was 31 mutated to cause debilitating genome because TLS was previously reported to be 32 4 essential for in vitro replication of CMV (Sivakumaran et al., 2000). TLS of RNA1, 1 RNA2 or RNA3 was respectively mutated in two ways, which included replacement R3-Tm1 and R3-Tm2 was 50%, 37.5%, 80% and 62.5%, respectively (Fig.1B). Same 10 type of TLS mutation in RNA3 had higher mutation repair frequency than that in 11 RNA2. In addition, replacement type mutation on caacg loop (m1) had higher 12 mutation repair than insertion type mutation on caacg loop (m2). Taken together, 13 identical TLS mutation in RNA1, RNA2 or RNA3 of CMV presented different 14 characteristic of mutation repair, including the sureness of occurrence and the 15 frequency of occurrence. 16 Due to the conservation of TLS region among RNA1, RNA2 and RNA3 in CMV, 17 RNA recombination may be the candidate mechanism to mediate mutation repair on 18 TLS mutation. To identify whether RNA recombination accomplished the mutation 19 repair on TLS mutation in CMV, identical TLS mutant of RNA1, RNA2 and RNA3 20 were mixed to simultaneously infect plants followed by analysis of mutation repair  R1-Tm1/R2-Tm1/R3-Tm1 or R1-Tm2/R2-Tm2/R3-Tm2 failed to be repaired, which was 27 due to the absence of copy-choice template through RNA recombination.

28
Quality or quantity defect of genomic RNAs resulting from TLS mutation in 29 RNA2 or RNA3 of CMV triggered mutation repair 30 Based on above data, single TLS mutation of RNA2 or RNA3 can be repaired, while 31 identical TLS mutation of RNA1 failed to be repaired ( Fig.1A & 1B). For TLS 32 mutation of RNA1, RNA2 or RNA3, the prerequisite of RNA recombination was 33 5 possessed, but occurrence frequency was remarkable difference. It is implied that 1 mutation repair of TLS mutation mediated by RNA recombination was triggered by 2 not only own characteristic of TLS mutation, but also other factors. To identify why 3 TLS mutation of different RNAs in CMV had different rate of mutation repair, the 4 effect of different TLS mutation on CMV pathogenicity and RNA accumulation was 5 analyzed. To exclude the obstruction of repaired mutation, we selected the special 6 samples, in which TLS mutation was not repaired (Fig.1C). Mutant of R1-Tm1 or 7 R1-Tm2 had similar pathogenicity with wt, which caused severe stunting (Fig.1C). 8 Mutant of R2-Tm1 or R2-Tm2 had weak pathogenicity causing slight stunting (Fig.1C).

9
Mutant of R3-Tm1 or R3-Tm2 had weaker pathogenicity than wt and presented weaker CMV pathogenicity and synthesis of corresponding RNAs (Fig.1). Quality defect of 29 RNA3 size caused by TLS mutation (Tm1 or Tm2) in RNA3 triggered high frequency 30 of mutation repair at 80% or 62.5%. Quantity defect of RNA2 synthesis caused by 31 TLS mutation (Tm1 or Tm2) in RNA2 also triggered high frequency of mutation repair 32 at 50% or 37.5%. TLS mutation in RNA1 did not cause quantity or quality defect, 33 which is correlated with the nonoccurrence of mutation repair. It is suggested that 34 6 quality or quantity defect of genome segment of CMV caused by TLS mutation was 1 the important triggered factor for mutation repair. Although TLS mutation of RNA1 failed to be repaired (Fig.1), after all it caused the 5 debilitating TLS in RNA1. Whether can the debilitating TLS in RNA1 be repaired at 6 other special condition such as mixed infection with other type of mutant causing 7 genome defect? Pre-termination mutant (R2-2bPT) of gene-silencing suppressor 2b 8 presented mild pathogenicity without stunting and low RNA accumulation (Fig.2). 9 When TLS mutation in RNA1 (R1-Tm1 or R1-Tm2) and R2-2bPT co-infected plants, 10 TLS mutant in RNA1 was repaired at the rate of 100% in progeny RNAs (Fig.2B), 11 while single TLS mutation in RNA1 failed to be repaired (Fig.1B). Similarly, when 12 TLS mutation in RNA3 (R3-Tm1 or R3-Tm2) and R2-2bPT co-infected plants, TLS 13 mutation was repaired at the rate of 100% in progeny RNAs (Fig.2B), which was 14 higher than that of single TLS mutation (R3-Tm1 or R3-Tm2) (Fig.1B). Presence of 15 R2-2bPT can remarkably improve the repair frequency of TLS mutation in RNA 1 or 16 RNA3, which may be due to relative low dose of RNA accumulation resulting from 17 the defect of gene-silencing suppressor 2b (Fig.2C). To further confirm the 18 relationship of the low dose of CMV genome and mutation repair of TLS mutation, 19 dilution inoculation of R1-Tm1 /R2/R3 or R1-Tm2 /R2/R3 was performed (Fig.3).  Fig.1 & Fig.3).

29
Cell-to-cell movement was necessarily required for mutation repair of TLS 30 mutation, which presented the positive effect of bottleneck on mutation repair 31 Based on above data, mutation repair can be triggered by quality or quantity defect of 32 genome resulted from TLS mutation or TLS mutation at the special low dose of 33 7 original inoculation, which was related with low-fitness of viral genome 1 (Fig.1&Fig.2&Fig.3). In addition to these multiple levels of trigger factors, 2 occurrence of mutation repair may require special occasion. Previous study had 3 reported that some mutation of essential element was not repaired in protoplast system 4 but repaired directly or indirectly in plants (Yuan et al., 2010). It is implied that the 5 occurrence of mutation repair may require the cell-to-cell movement of virus. In order 6 to analyze the role of cell-to-cell movement on mutation repair, cell-to-cell 7 movement-defected mutant R3-m5 was constructed via two Ala replacement 8 corresponding to position 549-554 in 3a ORF (Ding et al., 1995; Fig.4A&4B). When 9 R2-Tm1 or R2-Tm2 and R3-m5 were co-inoculated into plants, mutation repair failed to 10 be occurred (Fig.4C), while single R2-Tm1 or R2-Tm2 had high rate of mutation repair 11 at 50% or 37.5% (Fig. 1B). When R1-Tm1 or R1-Tm2 and R3-m5 were co-inoculated 12 into plants, mutation repair failed to be occurred at any dose (Fig.5), while single 13 R1-Tm1 or R1-Tm2 appeared mutation repair around dilution end-point (Fig. 3). It is 14 suggested that mutation repair of TLS mutation required the cell-to-cell movement of 15 virus. 16 Taken together, mutation repair of TLS mutation in CMV presented multiple levels    16 Error-prone performance of the viral RdRp was inclined to cause genetic mutation  frequency of mutation repair from 0% to 100% such as the case of TLS mutation in 7 CMV RNA1 (Fig. 1& Fig.3). There may appear the low dose around dilution 8 end-point when mild vaccine entered a new plant during the transmission due to the 9 low acquisition and secondary infection of mild vaccine by aphid. This special low 10 dose may be inclined to trigger mutation repair, which will produce wild type of CMV 11 destroying the safety of mild vaccine. Therefore, transmission characteristic of mild 12 vaccine should be removed to avoid the potential mutation repair.

13
Materials and methods 14 Plasmid construction 15 All mutant plasmids were derived from pCB301-Fny1, pCB301-Fny2 and cell-to-cell movement defect. 26 All above mutants were constructed from the plasmid pCB301-Fny1,  Mutation repair assay 13 To evaluate the stability and mutation repair of mutation sites in progeny RNAs, total 14 RNA was extracted from inoculated and systematic leaves at 14dpi after 15 agroinfiltration followed by RT-PCR. Specific primers located at upstream and 16 downstream of corresponding mutation sites were used in RT-PCR. RT-PCR products 17 were purified and directly sequenced using specific primers. Mutation repair assay for 18 each mutant was performed at least 10 individual repeats. Detailed information of 19 primers used to amplify mutation regions were shown in supplementary table2. 20 Northern blot 21 Total RNAs were extracted from leaves at 14 dpi after agroinfiltration and Northern 22 blot was performed as previously described (Yuan et al., 2006). cDNA probes were 23 labelled using random primers. Four types of cDNA probes were used to detect 24 different genomic and corresponding subgenomic RNAs. cDNA probe 1 (cP1) 25 corresponding to 2951-3150 of RNA1 was designed to exclusively detect RNA1.