Mutations in virus-derived small RNAs

Abstract

RNA viruses exist as populations of genome variants. Virus-infected plants accumulate 21-24 nucleotide small interfering RNAs (siRNAs) derived from viral RNA (virus-derived siRNAs) through gene silencing. This paper describes the profile of mutations in virus-derived siRNAs for three members of the family Potyviridae: Turnip mosaic virus (TuMV), Papaya ringspot virus (PRSV) and Wheat streak mosaic virus (WSMV). For TuMV in Arabidopsis thaliana, profiles were obtained for mechanically inoculated rosette leaves and systemically infected cauline leaves and inflorescence. Results are consistent with selection pressure on the viral genome imposed by local and systemic movement. By genetically removing gene silencing in the plant and silencing suppression in the virus, our results showed that antiviral gene silencing imposes selection in viral populations. Mutations in siRNAs derived from a PRSV coat protein transgene in the absence of virus replication showed the contribution of cellular RNA-dependent RNA polymerases to the generation of mutations in virus-derived siRNAs. Collectively, results are consistent with two sources of mutations in virus-derived siRNAs: viral RNA-dependent RNA polymerases responsible for virus replication and cellular RNA-dependent RNA polymerases responsible for gene silencing amplification.

Figure 1

Profile of TuMV-derived siRNAs in wild type Arabidopsis thaliana. Values are average and standard error from two biological replicates. Abundance was normalized to reads per million. Mechanically inoculated rosette leaves were sampled at 7 days post inoculation (dpi). Samples from systemically infected inflorescence were collected at 10 dpi. (a) Size distribution and abundance of 18–30 nt unique sequences with no mutations, and sequences with one or two mismatches. (b) Polarity and abundance by tissue and number of mismatches. Only 21- to 24-nt siRNAs were included. Te ratio of siRNAs with mismatchest to siRNAs with no mutations is indicated for each tissue.

Figure 2

Genome-wide distribution of TuMV-derived siRNAs in wild type Arabidopsis thaliana and de-novo assembly of the TuMV genome. 21- to 24-nt virus-derived siRNAs were pooled and correspond to samples in Fig. 1. Abundance was normalized to reads per million. (a) Genome-wide distribution of TuMV-derived siRNAs by match class in inoculated rosette leaves and in systemically infected inflorescence. (b) Abundance of siRNAs with one or two mismatches relative to siRNAs with no mutations. (c) Relationship between siRNAs of positive and negative polarity and containing mutations. (d) De-novo assembly of the TuMV genome using virus-derived siRNAs without mutations. Lines represent contigs and red arrowheads point to gaps in the assembly. In this TuMV clone, HC-Pro is marked with 6xHIS at the N terminus.

Figure 3

Comparison of TuMV-derived siRNAs in inoculated rosette leaves and systemically infected inflorescence. 21- to 24-nt siRNAs were pooled and correspond to samples in Fig. 1. Unique sequences were counted per tissue and their abundance normalized to reads per million. Values are the average of two biological replicates. Virus-derived siRNAs with (a) no mutations and (b) with one or two mismatches. (c) Accumulation of single nucleotide polymorphisms (SNPs) in virus-derived siRNAs in inoculated rosette leaves, in systemically infected inflorescence, and their correlation. The asterisk (*) indicates significant difference (p-value < 0.001). (d) Genome-wide distribution of single nucleotide polymorphisms in virus-derived siRNAs estimated in a 50-nt window.

Figure 4

Accumulation of 21- to 24-nt virus-derived siRNAs per cistron in TuMV in inoculated Arabidopsis rosette leaves and systemically infected inflorescence. Values are the average and standard error of two biological replicates as in Fig. 1. Abundance was normalized to reads per million and to the length of the cistron. (a) Accumulation of siRNAs with no mutations. (b) Accumulation of unique sequences with one or two mismatches. (c) Relative abundance of siRNAs expressed as the ratio of siRNAs with one or two mismatches to siRNAs with no mutations.

Figure 5

Genome-wide distribution of siRNAs derived from suppressor-deficient TuMV-AS9 and de-novo genome assembly. siRNAs 21- to 24-nt were pooled and classified based on the number of mismatches. Numbers represent average and standard error from two biological replicates from inoculated rosette leaves (7 dpi) and systemically infected cauline leaves (15 dpi) of ago2-1 single mutant Arabidopsis. Number of reads were normalized to reads per million. (a) Unique sequences and their abundance by polarity. The ratio of siRNAs with mismatches to siRNAs with no mutations is indicated for each tissue. (b) Genome-wide distribution of virus-derived siRNAs by match class, polarity, and tissue. (c) Correlation between siRNAs of positive and negative polarity, by match class and tissue. (d) De-novo assembly of the TuMV-AS9 genome using siRNAs with no mutations. Coverage is expressed as percent of the genome. Red arrow heads point to gaps in the assembly.

Figure 6

Comparison of suppressor-deficient TuMV-AS9-derived siRNAs in inoculated rosette leaves and systemically infected cauline leaves of ago2-1 single mutant Arabidopsis. 21- to 24-nt siRNAs were pooled and correspond to samples in Fig. 5. Values are the average of two biological replicates. Unique sequences were counted per tissue and their abundance normalized to reads per million. Virus-derived siRNAs with (a) no mutations and (b) with one or two mismatches. (c) Accumulation of single nucleotide polymorphisms (SNPs) in virus-derived siRNAs in inoculated rosette leaves, in systemically infected inflorescence, and their correlation. The asterisk (*) indicates significant difference (p-value < 0.001). (d) Genome-wide distribution of single nucleotide polymorphisms in virus-derived siRNAs estimated in a 50-nt window.

Figure 7

Genome-wide distribution of Wheat streak mosaic virus (WSMV)-derived siRNAs and de-novo genome assembly. SiRNAs (21- to 24-nt) were pooled and correspond to two contrasting cultivars, Arapahoe (susceptible) at 18 °C and 27 °C and Mace (susceptible at high temperature) at 27 °C. Abundance was normalized to reads per million. (a) Genome-wide distribution of WSMV-derived siRNAs in cultivar Arapahoe, and (b) cultivar Mace. (c) De-novo assembly of the WSMV genome using siRNAs with no mutations. Genome coverage is indicated in percent. Red arrow heads point to gaps (12 nt) in assembly, which map to the coat protein from nt 8331 to 8343.

Figure 8

Genome-wide distribution of Papaya ringspot virus (PRSV)-derived siRNAs. Abundance of siRNAs 21 to 24-nt was normalized to reads per million. (a) PRSV-derived siRNAs in non-transgenic cultivar AU9. (b) Transgene-derived siRNAs in leaves from transgenic plants cultivar SunUP, expressing a coat protein transgene.

Figure 9

Comparison between the accumulation of mutations in virus-derived small RNAs and in the genome. Using full genome sequences or small RNAs, single nucleotide polymorphisms were estimated and normalized to a 50-nt window. The average and a 99% confidence interval are represented by a horizontal line. (a) Turnip mosaic virus. SiRNA variation in inflorescence is as in Fig. 2a. (b) Wheat streak mosaic virus. SiRNA variation in cultivar Arapahoe at 18 °C is as in Fig. 7a. (c) Correlation between genomic and siRNAs variation.

Figure 10

Model for the introduction of mutations in genomic viral RNA and in virus-derived siRNAs. Viral RNA-dependent RNA polymerases introduce mutations in viral RNA during replication and transcription. Cell-to-cell movement, systemic virus movement, and antiviral gene silencing impose purifying selection. During antiviral gene silencing amplification, cellular RNA-dependent RNA polymerases introduce mutations.