Salinity tolerance associated Meta-QTLs in rice
A total of 265 QTLs related to 32 traits were collected in this study using the Simple Sequence Repeats (SSR) markers (Table S1, S2) among which, 126 and 139 QTLs were selected for further analysis in normal and salinity conditions (Table S3). Most of the QTLs belonged to the salinity tolerance score (STS) (27 QTLs), shoot potassium concentration (KS) (26 QTLs), shoot sodium concentration (NS) (21 QTLs), chlorophyll content (CHL)(19 QTLs) and shoot dry weight (DSW) traits (19 QTLs) (Fig.S1). In contrast, the rare QTLs belonged to the number of sterile spikelets (NSS) [20], dead seedling rate (DSR), leaf potassium concentration (KLV), reduction of seedling height (RSH) and reduction of leaf area (RLA) traits (Fig.S1). The highest number of QTLs were observed on chromosome 1 (37 QTLs) and 2 (36 QTLs) followed by chromosome 7 (29 QTLs), while chromosome 8 (12 QTLs) and 11 (12 QTLs) had the lowest number of QTLs (Fig.S2). The phenotypic variance described by the original QTLs was different from 0.7% to 33.25% and the confidence interval (CI) of markers was different from 0.99 to 84.36 cM (Table S3). After the integration of all the collected QTLs on the consensus map, 46 meta-QTLs were identified in 12 chromosome of rice (Fig.1). There were meta-QTLs with a CI of 95% based on the lowest Akaike information criterion (AIC) values. Remarkably, second meta-QTLs on Chr7: M-QTL2, Chr2: M-QTL2, and Chr1: M-QTL2 included the highest number of initial QTLs (17 ,16 and 12, respectively), which covered a relatively narrow CI (4.78, 1.82 and 2.84 cM, respectively) (Table S4). These meta-QTLs support the important traits; for example, ratio of the shoot sodium and potassium concentration (NKS), number of fertile spikelets (NFS), root length (RTL), and chlorophyll content (Table S4). Chr12: M-QTL4, Chr 9: M-QTL3 and Chr3: M-QTL2 had the highest mean percentage of phenotypic variation (R2), which can be considered as the main effective QTLs for the involved traits (Table S4). A total of 9366 genes were detected in 46 meta-QTL positions, among which, Chr8: M-QTL2 contained the highest number of genes (868 genes); while, Chr12: M-QTL2 contained the lowest number of genes (14 genes) (Table S4). Moreover, the proportion of functionally characterized annotated genes (27%) is actually limited compared to the about 73% of unannotated genes with allocated putative functions. It is intersting to note that, 81 genes were identified on Chr1: M-QTL2 which were located in Saltol region.
Expression profiling analyses in the salinity tolerant genotypes of rice
The DEGs were identified under salinity stress compared to control conditions in the salinity tolerant genotypes. A total of 1714 DEGs were observed in the roots of FL478 as a salinity tolerant genotype, among which, 927 and 787 were up- and down-regulated in the salinity conditions [31]. DEGs from multiple RNA-seq datasets were combined and the DEGs were classified into root, shoot, seedling, and leaves to have a deeper understanding about the salt responsive genes in the salinity tolerant rice genotypes. A total of 3030, 396, 703 and 723 DEGs were merely identified in root, shoot, seedling and leaves, respectively (Fig.S3). Also, raw microarray data from nine independent experiments were downloaded (Table S5) and analyzed uniformly. Microarray meta-analysis suggested 11694 DEGs, among which, 4121, 13, 6247 and 1199 DEGs were exclusively expressed in root, shoot, seedling and leaves, respectively (Fig.S4). In addition, a total of 4763 and 5862 DEGs were merely up- and down-regulated, respectively, in the salinity tolerant genotypes.
Integration of DEGs from two Meta-Analysis approaches
Identified DEGs in both RNA-Seq and microarray meta-analysis were combined to confirm the consistency of the obtained results. A list of overlapping DEGs were detected in four tissues, separately after removing all the duplicate genes.
Comparative transcriptome analysis indicated that 227, 2, 311, and 84 DEGs were commonly detected by the RNA-Seq and microarray respectively in root, shoot, seedling, and leaves tissues (Fig.2). A total of 4255 and 10980 DEGs were merley identified by the RNA-Seq and microarray meta-analysis, while only 156 DEGs were previously reported in the literature (Fig.2).
Detection of the DEGs in the meta-QTL positions
There were a total of 1345, 86, 1729, and 552 DEGs in the meta-QTL positions in root, shoot, seedling and leaves, respectively (Fig.3). Among the identified DEGs in the meta-QTL positions, 664 and 2359 DEGs were identified by the RNA-Seq and microarray meta-analysis, respectively while, only 82 DEGs located in the meta-QTL positions were previously reported in the literature (Fig.3).
Functional annotation of DEGs located in the meta-QTL positions
Gene ontology enrichment analysis was performed to determine the biological roles of the DEGs located in the meta-QTL positions. Carbohydrate metabolic process, regulation of cellular process, regulation of transcription, response to stress and regulation of nitrogen compound metabolic process were indicated as dominant terms in the biological processes (BP) (Fig.S5). Moreover, some BP terms including regulation of transcription, inorganic anion transport, anion transport, ion transport as well as regulation of gene expression, cell wall organization and modification were significantly enriched (Fig.S5). The most significant over-represented molecular function (MF) terms were nucleotide binding, ATP binding, anion transmembrane transporter activity, inorganic anion transmembrane transporter activity, transcription factor activity and oxidoreductase activity (Fig.S5). In terms of cellular component (CC) ontology, the most significant enriched terms were intrinsic to membrane and integral to membrane (Fig.S5).
Mining the potential candidate genes in the meta-QTL positions
Exploring the meta-QTL regions for the common genes were resulted in finding 60 potential candidate genes in the root (Table S6), among which, only four genes were previously reported associated to the salinity response. Remarkably, LOC_Os01g20980.1 (coding Pectinesterase) was found in Chr1: M-QTL2 located in Saltol region (Table S6). Ion homeostasis related QTLs were also found in Chr1: M-QTL2 which controling the KLV, NS, NKS, KS and RN traits (Table S4). Overall, identified potential candidate genes were classified into several terms in the root tissue, for example, transcription factor (e.g., TIFY, GRAS, HOX, WRKY and MYB family), signaling (e.g., OsWAK125, pectinesterase ,OsMKK1, and CHIT15), transporter (e.g., OsHKT1 and some genes coding transmembrane transport and anion transporter) and some other functions (e.g., NUDIX family, genes coding the aspartic protease ) (Table S6).
Four genes in meta-regions on Chr2, 3, and 8 were identified as potential candidate genes in the shoot, as discussed in the literature; for instance, TIP2-1 (LOC_Os02g44080.1) in Chr2: M-QTL4 (Table S6). Chr2: M-QTL4 was integrated with seven initial QTLs controlling RTL and some other related traits (e.g. S, KS, NKS, SIS, and NS) (Table S4). Moreover, two transcription factors (LOC_Os03g08310.1 and LOC_Os08g15050.1) were identified respectively as possible candidate genes in Chr3: M-QTL1 and Chr8: M-QTL2 (Table S6) supporting the root length and photosynthesis related traits, respectively (Table S4). It is interesting to note that, LOC_Os03g08310.1 (coding TIFY11A) was identified as common candidate gene in the root and shoot (Table S6).
Our results indicated 98 potential candidate genes in the seedling including 84 DEGs located in the M-QTLs that were not reported yet. However, 14 genes have been already considered in the literature (Table S6). Functional classification of these potential candidate genes further suggested that they were related to the transcription regulation (e.g., AP2, WRKY, HOX, and GRAM family), signal transduction (e.g., CIPK24, GDSL) and there were some genes with another functions including kinase, phosphatase, and transporter terms under salinity stress in seedling tissue (Table S6). Remarkably, LOC_Os01g20830.1 (coding a transporter protein) and LOC_Os01g21144.1 (with unknown function) were found in Saltol region on Chr1: M-QTL2 (Table S6). As well, there were some potential candidate genes in hotspot-regions; for example, WRKY70 (LOC_Os05g39720.1) in Chr5: M-QTL4 and PP2C (LOC_Os06g48300.1) in Chr6: M-QTL4 (R2= 10.31%) (Table S4, S6). Moreover, some genes were identified as potential candidate genes in Chr2: M-QTL1, Chr8: M-QTL1, Chr10: M-QTL3, and Chr11: M-QTL1; these meta-regions were integrated the importance of the initial QTLs for photosynthesis, straw dry weight, yield components (e.g. QGW, DF and NFS) and RTL traits (Table S4, S6).
Totally, 28 potential candidate genes were identified in the leaves among which, 14 genes were found in the literature. The LOC_Os01g22249.1 (coding the peroxidase) located in Saltol region in Chr1: M-QTL2 was identified as another leading candidate gene. Notably, OsHKT1 (LOC_Os06g48810.1) and PP2C (LOC_Os06g48300.1) were found in the hotspot-regions in Chr6: M-QTL4 (Table S4, S6).
The obtained results indicated that, 20 genes were located on the hotspot-regions containing original QTLs for both yield components and ion homeostasis traits which could be suggested as promising candidate genes (Fig.4, Table 1). The promising genes were related to the following functions: pectinesterase, peroxidase, transcription regulation, high-affinity potassium transporter, protein serine/threonine phosphatase, cell wall organization and a CBS domain containing gene, among which, there were 2 genes in Saltol region (Table 1).
Validation of differential gene expression using qRT-PCR
To further validate the potential candidate genes, 15 genes were selected for qRT-PCR in FL478 as a salt tolerant genotype (Fig.5). The qRT-PCR results were confirmed the outcome of the meta-analysis (Fig.S6).