Bisulphite sequencing reveals dynamic DNA methylation under desiccation and salinity stresses in rice cultivars
Stress treated seedlings were harvested, snap frozen in liquid nitrogen and stored at -80 C till further use.
we analyzed role of DNA methylation under desiccation and salinity stresses in three (IR64, stress-sensitive; Nagina 22, drought-tolerant and Pokkali, salinity-tolerant) rice cultivars via bisulphite sequencing.
Methylation in CG context within gene body and methylation in CHH context in distal promoter regions were positively correlated with gene expression. Hypomethylation in Nagina 22 and hypermethylation in Pokkali in response to desiccation and salinity stresses, respectively, were correlated with higher expression of few abiotic stress response related genes.
DMR-DEGs harboring differentially methylated cytosines due to DNA polymorphisms between the sensitive and tolerant cultivars in their promoter regions and/or coding regions were identified, suggesting the role of epialleles in abiotic stress responses.
Correlation between DNA methylation and gene expression was determined by plotting methylation density of genes expressed at different levels in each sample. Based on FPKM values, genes expressed at very low (<1 FPKM), low (>1-5 FPKM), moderate (>5-25 FPKM) and high (>25 FPKM) levels were categorized. Methylation density in gene body, gene ends and flanking regions was estimated for these sets of genes expressed at varying levels. The correlation between differential methylation and differential gene expression (>2 fold-change with <0.05 q-value) under desiccation/salinity stress as compared to control condition within same cultivar was analyzed by estimating methylation level differences in different sequence contexts within gene body and 2 kb flanking regions.
We generated about (76-86 million) high-quality 90-bp long paired-end reads with >30x sequencing depth for each sample. A total of about 46-55 million uniquely mapped reads covered ~88% of the rice genome and 81.5-82.8% of total cytosine residues in each sample analyzed (Table S1).
Average methylation level of mCs in CG context (85.45-90.37%) was much higher than CHG (66.41-70.15%) and CHH (39.76-45.69%) contexts (Fig. 1B; Fig. S1).
we compared methylation levels between stress treated and control samples for each cultivar. Interestingly, methylation levels showed most variations in CHH context followed by CHG context in response to desiccation and salinity stresses in Nagina 22 and Pokkali, respectively (Fig. 1B).
increased methylation level in Pokkali under salinity stress were observed in both the sequence contexts.
methylation levels in CG context were marginally lower under desiccation stress in Nagina 22 and under salinity stress in Pokkali.
The control and stress treated sample(s) of the same cultivar clustered together, suggesting that methylome divergence among the cultivars is much higher than the methylation changes in response to abiotic stress within a cultivar (Fig. 1C).
pericentromeric and centromeric regions harboring high density of transposable elements (TEs) were found to be extensively methylated in CG and CHG contexts in control and stress treated samples (Fig. 1D). In contrast, higher fraction of mCs in CHH context was detected in gene rich regions under control and stress conditions in all the cultivars (Fig. 1D), suggesting an important role of CHH context DNA methylation in abiotic stress response.
The methylation density at gene ends representing transcription start site (TSS) and transcription termination site (TTS) was lower as compared to their body and flanking regions in all the sequence contexts.
Likewise, methylation density difference between control and salinity stress conditions in IR64 and Pokkali cultivars was analyzed. Higher methylation density in IR64 and lower methylation density in Pokkali were observed in all the sequence contexts under salinity stress in both protein coding genes and TEs (Fig. 2C, D; Fig. S3B).
Interestingly, we observed a positive correlation between DNA methylation in CHH context in distal flanking regions (-500 to -2000 bp) and gene expression in control (Fig. S4) and stress treated samples (Fig. 3). In contrast, an antagonistic correlation between CG and CHG contexts DNA methylation in all other genic regions and gene expression was observed.
Likewise, we analyzed DMRs under salinity stress in IR64 and Pokkali. A total of 2511 and 3580 DMRs representing 2002 and 2834 genes were detected in IR64 and Pokkali, respectively, under salinity stress. Majority of DMRs were detected in CHH context followed by CG and CHG contexts in IR64 and Pokkali cultivars (Fig. 4E, F).
This suggests that CHH context hypermethylation may be involved in salinity stress response in rice.
In Pokkali, GO terms related to abiotic stress response, including response to salt stress in the hypermethylated genes (Fig. S5).These results suggest that hypomethylation under desiccation stress in Nagina 22 and hypermethylation under salinity stress in Pokkali may be important to elicit stress tolerance in these cultivars.
This suggests that methylation dynamics in CHH context guided by de novo methylation and demethylation may be important during stress response within a cultivar. However, methylation differences detected in CG context between cultivars may be due to diversification of DNA methylomes during selection/domestication in different rice cultivars.
To understand the role of DNA methylation in response to salinity stress, we analyzed differential expression of DMR-associated genes in different sequence contexts and gene regions in IR64 and Pokkali. A total of 34 and 74 DMR associated DEGs under salinity stress were detected in IR64 and Pokkali, respectively (Fig. 6A, B). Most (86.1-88.57%) of the DMR-DEGs under salinity stress showed differential methylation in CHH context in both the cultivars (Fig. 6C, D). About (77.94-87.1%) of the DMR-associated genes in CHH context showed higher expression under salinity stress in both the cultivars and most of these genes harbored DMRs in their flanking regions (82.3-83.87%).
In IR64, CHH context hypermethylation in promoter (55.5%), gene body (80%) and downstream regions (69.2%) was correlated with higher gene expression under salinity stress (Fig. 6C). However, correlation of CHH context hypermethylation in promoter (94.7%), gene body (80%) and downstream (91.7%) regions with higher gene expression was found to be more significant in Pokkali (Fig. 6D). In addition, the number of genes showing correlation was much higher (~3 times) in Pokkali as compared to IR64. These results suggest that hypermethylation in CHH context may be associated with salinity stress response in Pokkali.
Further, we analyzed correlation of differential methylation with differential gene expression in the sets of genes involved in abiotic stress response. In IR64, four genes exhibited CHH context hypermethylation and higher expression under salinity stress (Fig. 6E). In Pokkali, a total of five genes involved in abiotic stress response showed hypermethylation in CHH context and higher gene expression under salinity stress (Fig. 6F).
Next, we analyzed overlap of DMR-associated DEGs in the sensitive and tolerant rice cultivars underdesiccation and salinity stresses. Likewise, about 74-78% of the DMR-associated DEGs were specific under salinity stress in IR64 and Pokkali (Fig. 7).
Uncovering Differentially Methylated Regions (DMRs) in a Salt-Tolerant Rice Variety under Stress: One Step towards New Regulatory Regions for Enhanced Salt Tolerance
We analyzed genome wide DNA methylation of a salt-tolerant rice variety under salinity and identified a set of differentially methylated regions (DMRs) between control and stress samples using high-throughput sequencing of DNA immunoprecipitated with the 5-methylcytosine antibody (MeDIP-Seq). The examination of DNA methylation pattern at DMRs regions revealed a general tendency for demethylation events in stress samples as compared to control. In addition, DMRs appear to influence the expression of genes located in their vicinity.
Following our previous studies [13] we have furthered investigated the patterns of DNA methylation in 'Pokkali', a rice variety that simultaneously shows a great capacity to tolerate salinity stress and to rapidly shape DNA methylation levels when exposed to salinity.
The percentage of uniquely mapped reads was considerably higher on the non-immunoprecipitated sample (approximately 50%) contrasting with approximately 25% for the immunoprecipitated samples (Table 1). Regarding the genome coverage, approximately 7.5% of cytosines were covered by at least one uniquely mapped read (Table 1).
The analysis of differential methylation based on MeDIP-Seq data, using the MEDIPS program, as mentioned in the methods section, enabled the identification of 53 DMRs between control and salt stress samples. The DMRs that were close to each other (less than 500 bp) were merged originating 22 DMRs (ranging from 100 to 1000 bp) (Table 2). Although all methylation contexts were present (CG, CHG, and CHH), in DMR2 the methylation was mainly in the CHH context, while in DMR15 the CHG context was predominant.
The DMRs were analyzed according to their position relative to the nearest gene, and more than 70% of the DMRs were in close proximity to genes (less than 2 kbp away) (Figure 2A). Furthermore, over 75% of the DMRs identified were associated with transposable elements and repetitive sequences (Figure 2B).
The functional annotation of genes flanking or overlapping salt-induced DMRs was performed using Blast2GO [35] and multilevel pie charts were generated for the three main classes: cellular component, biological process, and molecular function (Figure S4).