Rice Gene And Srna Expression Quantitative Trait Loci Mapping And Regulatory Network Construction

Expression of genes or small RNAs (sRNAs) is coordinately regulated in space and time in an organism to deliver the gene products when needed, according to the developmental programs and in response to environmental cue. The ability to reveal the regulatory architecture of the genes/sRNA at the whole genome scale by constructing the regulatory network is critical for understanding the biological processes and genome function of the organism.Analysis of expression QTLs (eQTLs) that treats the expression values of the genes or sRNA as a quantitative trait (e-trait) to perform QTL analysis using a genetically segregating population appears to be a useful strategy for suggesting regulatory relationships between genetical factors. Similar to QTL analysis, this analysis can detect genetic elements that regulate the expression variation of the e-trait acting in cis if the QTL is located in the vicinity of e-trait, or trans if it is located in a distant position. Such analysis could suggest the existence of a potential regulator in a genomic region for a gene (e-trait). Moreover, one regulator could be the target of another regulator. Thus connecting such regulator-target relations revealed by multiple e-traits and eQTLs of different layers would lead to a net-like structure, providing basic elements for the construction of regulatory network. More recently, an approach combining eQTL with co-expression analysis was proposed for identifying regulator candidates underlying eQTLs, which may greatly refine the resolution of the analysis.1. eQTL mapping in rice shoots at72h after germinationIn this study, we used an Affymetrix GeneChip Rice GenomeArray to analyze eQTLs in rice shoots at72h after germination from110recombinant inbred lines (RILs) derived from a cross between Zhenshan97and Minghui63using601recombinant bins. We obtained16,372expression traits (e-traits) each with at least one eQTL, resulting in26,051eQTLs in total, including4,464cis-eQTLs. The cis-eQTLs tended to have higher LOD scores than trans-eQTLs. The distribution of the eQTLs deviated significantly (χ2=6924.86, P<2.2e-16) from random occurrence on the12chromosomes.We investigated the distribution of eQTLs along the genome by dividing the whole genome into1cM partitions.171potential eQTL hotspots were identified in the rice genome, each of which controls transcript variations of many e-traits. Gene ontology analysis revealed an enrichment of certain functional categories of genes in some of the eQTL hot spots. In particular, eQTLs for e-traits involving the DNA metabolic process was significantly enriched in several eQTL hot spots (Chr03:401-402cM; Chr05:717-718cM; Chr10:1205-1206cM,1243-1244cM,1244-1245cM).We calculated correlations of the transcript levels between4,464e-traits with cis-eQTLs and the e-traits with trans-eQTLs located in the1.5LOD-drop support intervals of corresponding cis-eQTLs.99genes showed significant correlations with30or more e-traits that had trans-eQTLs located in these gene regions, including24genes showing co-expression with100-425e-traits. These data indicated that genes other than transcript factors also play important roles as master regulators for gene expression at the whole genome level.We also detected correlations between QTLs for shoot dry weight and eQTLs, revealing possible candidate genes for the trait. These results provided clues for the identification and characterization of the regulatory network in the whole genome at the transcriptional level.2. eQTL mapping in rice flag leaf at heading stageThe ability to reveal the regulatory architecture of the genes at the whole genome level by constructing the regulatory network is critical for understanding the biological processes and developmental programs of the organism. Here we conducted an eQTL guided function-related co-expression analysis for identifying the putative regulators and constructing gene regulatory network.We performed an eQTL analysis of210RILs derived from a cross between two indica rice lines, Zhenshan97and Minghui63, the parents of an elite hybrid, using data obtained by hybridizing RNA samples of flag leaves at heading stage with Affymetrix whole genome arrays. Making use of an ultra-high density SNP bin map constructed by population sequencing,13,647eQTLs for10,725e-traits were detected, including5079cis-eQTLs (37.2%) and8568trans-eQTLs (62.8%). Four chromosomes (Chr01,Chr03, Chr06, Chr09) had greater numbers of eQTLs than expected based on both physical and genetic maps. The analysis revealed138trans-eQTLs hotspots which were both significantly more than the expectations based on the sizes of the bins (P<0.01) and the numbers of e-traits with cis-eQTLs (P<0.01), each of which apparently regulates the expression variations of many genes. Co-expression analysis of functionally related genes within the framework of regulator-target relationships outlined by the eQTLs led to the identification of putative regulators in the system. The usefulness of the strategy was demonstrated with the genes known to be involved in flowering. We presented a simple, yet effective method for gene regulatory network reconstruction using an eQTL guided function-related co-expression analysis. We combined data of gene expression from a rice RIL population, gene functional annotation of a prior knowledge and iGA with the co-expression level to identify candidate regulators. Eight obvious regulatory groups were identified. Ghd7, Hdl and RFT1were identified as master regulators in this analysis, which were the trans-eQTLs for several flowering related genes. To confirm the network, we checked the transcript abundance of the candidate downstream targets for modulator of Ghd7using flag leaves at heading date from Zhenshan97and NIL(mh7) which have identical genetic background except the introgressed segment.80%of the candidate targets were confirmed by experiment result. These discrepancies may be due to differences of the genetic materials (RILs vs NILs) and/or environmental factors, since the cis-eQTL for Ghd7could not fully explain the expression variation of the gene in the population (LOD=67.6,R2=68.5%).We also applied this strategy to the analysis of QTLs for yield traits, which also suggested the likely candidate genes. The eQTL-guided co-expression analysis may provide a promising solution for outlining a framework for the complex regulatory network of an organism.3. eQTL mapping in rice flag leaf at pre-booting stageThirdly, huge efforts for the analysis of expression quantitative trait loci (eQTLs) have been exploited to discover genetic function loci that explain variation in gene expression levels and identify specific cis or trans regulatory regions at genome-wide scale. The utilization of eQTL mapping could be broadened to small RNA expression variation. In order to detect the regulatory regions that have controlled the expression divergence of sRNAs in population, small RNA-sequencing technologies could be applied to collect the sRNA expression level in flag leaf at pre-booting stage from Zhenshan97, Minghui63and their98IMF2population. These elucidations provided a sight of the genetic composition of sRNA expression variations, especially for dominance effect, and understood the expression regulatory patterns of sRNA expression diversity. We constructed SNPs replaced parents’ reference genome. Three categories reads from heterozygous materials should be considered,(1) the sRNAs conserved between parents,(2) allele-specific expression sRNAs,(3) specifically aligned to any parent.0.71%SNPs contained sRNAs added17%parents specifically expressed sRNAs might associated with the genetic variation between Zhenshan97and Minghui63.81,096sQTLs were obtained for53,904sRNA e-traits, which produced27,375cis-sQTLs in the diagonal and other53,721trans-sQTLs. We identified significant hotspots for sRNA e-traits whether all e-traits and e-traits with sQTLs mapped. The most significant hotspots were Bin385and Bin969.Totally,45,438sQTLs (56.03%) displayed significant dominance effect, in which, the vast majority of sQTLs (92.26%) showed negative dominance effect.36.89%sQTLs exhibited negative overdominance effect of all significant negative dominance sQTLs where the performance of sRNAs present weak prospect in heterozygous genotype than parents genotypes.96%sQTLs with significant overdominance effect were trans-sQTLs either positive or negative overdominance. There were several bin abundant with the trans-sQTLs displaying positive dominance effect, but no peaks for cis-sQTLs. We also identified significant hotspots for cis-and trans-sQTLs. Most of hotspots for cis and trans-sQTL were distinct, the most significant hotspots for cis-sQTLs were on Chr02, Chr03and Chr07. However, Chr04, Chr05and Chr09for trans-sQTLs.We compared the eQTLs for sRNA biogenesis genes about Dicer-like (OsDCLs), Argonaute (OsAGOs) and OsRDRs with the trans-sQTL hotspots. The results showed that few eQTLs for small RNA biogenesis genes were located in the most significant trans-sQTL hotspots regions. Although these genes played vital roles in the sRNA biogenesis, but no remarkable evidences have detected that were the trans-eQTLs (regulators) for sRNA expression variations in the population.The expression variations for the well-known miRNAs for the different mature sequences were analyzed, and produced20cis-miQTLs and390trans-miQTLs for153miRNA with different mature sequences. Our miQTL mapping data showed that different isoforms of the miRNAs were always controlled by different genetic factors in intricate transcript processes. miRNA did not have a significantly detectable effect on corresponding mRNA targets in genetic population variations.All data displayed a more dynamic perspective of small RNA regulatory activities.