Role of methylation and siRNA on differential allelic expression in a hybrid of distantly related citrus species

Abstract DNA methylation plays a central role in the regulation of gene expression. In plants, methylation occurs in the CG, CHG and CHH contexts, via distinct DNA methyltransferases including MET1, CMT3 and the RNA-directed DNA Methylation (RdDM) pathway via DRM2. In interspecific hybrids, these epigenetic mechanisms are confronted to a mixed small RNA population and two subgenomes harbouring specific methylation patterns, therefore generating unique expression profiles. The aim of this work was to understand these regulations by analysing gene expression, DNA methylation and small RNAs in a Citrus hybrid resulting from the cross between C. reticulata (mandarin) and C. australasica (finger lime). Haplotype-resolved subgenomes assembly identified hundreds of allele-specifically expressed genes. Asymmetric reprogramming of methylation was observed, in particular an increase in CHH in C. australasica haplotype. Surprisingly, CHH methylation, usually associated with gene silencing, was correlated here with increased expression, but also 24nt small RNA populations at their promoter regions. Similar analyses of the parental lines and other citrus species suggest the correlation between CHH methylation-enriched promoter and high expression level is not due to the hybridization, but seem to be generally true for all citrus. These observations suggest that, in citrus fruit, RdDM could activate transcription. This work also provides a full pipeline to analyse the expression profiles and DNA methylation in complex hybrids, which could be crucial for anticipating varieties resistant to diseases and the current threats affecting citriculture such as the Huanglongbing disease. Competing Interest Statement The authors have declared no competing interest. Footnotes Substantial modifications have been made : First, to clarify whether the bias identified between methylation and expression was related to hybridisation or rather to parental contexts, we generated or used new transcriptomic and methylome data from additional citrus species, namely Citrus lemon, Citrus Reshni, Citrus sinensis and Poncirus trifoliata. Regardless of the species tested, all exhibited this bias, which appears to be shared among all citrus species, although we remain factual on this point to not over-interpret our data. In addition, these new analyses also include analyses of roots, particularly of the commonly used rootstock P. trifoliata. Therefore, we demonstrate that this association between high expression and de novo methylation is not tissue-specific. Secondly, we validated some of the RNA-seq data via RT-qPCR, as well as WGBS methylation data via BS-PCR, followed by Sanger sequencing. Finally, a thorough review and modifications to the discussion section have made this revised article clearer, with conclusions and hypotheses that are better supported by the new data sets.