Analysis of codon usage and allele frequencies reveal the double-edged nature of cross-kingdom RNAi

Abstract

BackgroundIn recent years, a new class of small 21- to 24-nt-(s)RNAs has been discovered from microbial pathogens that interfere with their hosts gene expression during infection, reducing the hosts defence in a process called cross-kingdom RNA interference (ckRNAi). According to this model, microbial sRNAs should exert selection pressure on plants so that gene sequences that reduce complementarity to sRNAs are preferred. In this paper, we test this consequence of the ckRNA model by analyzing changes to target sequences considering codon usage and allele frequencies in the model system Arabidopsis thaliana (At) - Hyaloperonospora arabidopsidis (Ha) and Hordeum vulgare (Hv) - Fusarium graminearum (Fg). In both pathosystems, some selected sRNA and their corresponding target have been described and experimentally validated, while the lengthy methodology prevents the analysis of all discovered sRNAs. To expand the understanding of ckRNAi, we apply a new in silico approach that integrates the majority of sRNAs. ResultsWe calculated the probability (PCHS) that synonymous host plant codons in a predicted sRNA target region would show the same or stronger complementarity as actually observed and compared this probability to sets of virtual analogous sRNAs. For the sets of Ha and Fg sRNAs, there was a significant difference in codon usage in their plant gene target regions (for Ha: PCHS 24.9% lower than in the virtual sets; for Fg: PCHS 19.3% lower than in the virtual sets), but unexpectedly for both sets of microbial sRNA we found a tendency towards codons with an unexpectedly high complementarity. To distinguish between complementarity caused by balancing sRNA-gene coevolution and directional selection we estimated Wrights F-statistic (FST), a measurement of population structure, in which positive deviations from the background indicate directional and negative deviations balancing selection at the respective loci. We found a negative correlation between PCHS and FST (p=0.03) in the At-Ha system indicating deviations from codon usage favoring complementarity are generally directionally selected. ConclusionThe directional selection of complementary codons in host plants suggests an evolutionary pressure to facilitate silencing by exogenous microbial sRNAs, which is not consistent with the anticipated biological role of pathogen sRNAs as exclusively effectors in cross-kingdom RNAi. To resolve this conflict, we propose an extended model in which microbial sRNAs are perceived by plants via RNA interference and, via coevolution, primarily help to fine-tune plant gene expression.