Genetic Dissection On The Basis Of Heterosis In An Elite Cotton Hybrid XZM2

As a major source of fibers, cotton is an important economic crop and plays animportant role in the global economy. With the development of textile technology andsocial demand, it is urgent to breed and plant cotton varieties with high yield productionand super fiber quality. Hybrid cotton could greatly increase lint yield, modestly improvefiber quality and raise tolerance to stresses. Accordingly, hybrid cotton could createremarkable economic and social benefits. However, heterosis of cotton has not been used aswidely as that of rice and corn. Deficiency of combinations with high heterosis and theexpense of producing F₁ seeds are two main reasons leading to this present situation.Encouragingly, the second problem can be resolved to some degree nowadays, sinceprevious studies showed that F₂ heterosis existed in cotton. Therefore, it is urgent to studythe genetic mechanism of cotton heterosis to promote selection of elite hybrid cotton.In this research, a population of 180 recombinant inbred lines (RILs) was developedby single seed descended from the cross of high yield Upland cotton (Gossypium hirsutumL.) varieties Zhongmiansuo12 (ZMS12) and 8891, the two parents of Xiangzamian2(XZM2). Field trials in two locations and two years were performed and QTL for importanttraits were tagged using data of the RIL population. An immortalized F₂ population wasconstructed by intercrossing of the RILs and was planted in two years to map QTL andreveal the genetic mechanism of heterosis. DDRT-PCR and DNA methylation wereperformed using leaves in different growth stages of XZM2 and its two parents so as todetect the molecular mechanism of heterosis. The results were listed as the following:1. QTL mapping of yield and yield-component, fiber quality and plant architecturetraitsData of 149 polymorphic loci were used to construct a genetic linkage map. Finally,132 loci were mapped to 26 chromosomes/linkage groups covering 865.20 cM,approximately 18.57% of the total recombinational length of the cotton genome, and theaverage genetic distance between two loci was 6.55 cM.The RIL population was planted at Jiangpu and Guanyun in 2002 and 2003. Jiangpu islocated in Yangtze River Valley cotton-growing region, while Guanyun is located inYellow River Valley cotton-growing region. Plant architecture, yield and yield-component and fiber quality traits were investigated.QTL for all traits were analyzed by composite interval mapping method usingWindows QTL Cartographer 2.0. Data sets from single environment (separate analysis) anda set of data from the means of multiple environments (joint analysis) were used to performQTL mapping.For yield and yield-components, altogether 15 QTL were identified in the jointanalysis. In addition, 34 QTL for yield and yield components were identified in the threeenvironments in separate analysis and 10 of them were detected in joint analysis. In aseparate analysis, 48 QTL for fiber quality traits were identified in the four environments,and 16 of them were detected by joint analysis. For plant architecture traits, a total of 16QTL were mapped.Notably, a stable lint percentage QTL qLP-A10-1 was detected both in joint analysisand in two environments of separate analysis. A fiber length QTL qFL-D2-1 explaining16.1% of phenotypic variance on average was simultaneously detected in joint analysis andin four environments of separate analysis, showing high stability across differentenvironments.QTLMapper1.6 based on a mixed linear model was used to determine QTL withadditive effects at the single-locus level, digenetic interactions between two different lociand interactions between QTL and the environments were analyzed.Epistatic QTL were detected for all of yield and yield-component traits. Altogether 16pairs of QTL interactions (AA) were identified. However, only 4 of the loci involved hadsinge-locus effects, indicating the complexity of epistasis and their important contributionsto yield and yield-component traits.The epistatic effects of fiber length explained 21.8% of PV, which is a little higherthan additive main effects (20.2%). QE interactions contributed less when compared withadditive and epistatic effects. The epistatic effects were smaller for other fiber quality traits.Many digenetic gene loci contributing notably to all the plant architecture traits weredetected, most of which did not have the main effects themselves. The total epistatic effectsexplained phenotypic variance from 15.2% to 51.1%. In fact, more epistatic QTL thanmain-effect QTL were detected and contributed significantly to plant architecture traits.This would explain the phenomenon that though there were large differences of plantarchitecture between ZMS12 and 8891, few polymorphic molecular marker loci could bedetected. From the results we presumed that main effects as well as epistatic effects of the QTLplayed an important role in XZM2-derived RILs.Lint yield of cotton is determined by its component traits, such as bolls/plant, bollweight and lint percentage. Integrated results of regression analysis, correlation analysisand path analysis showed that bolls/plant, boll weight and lint percentage had positiveeffects to seed yield and lint yield; their contribution order to lint yield was bolls/plant＞boll weight＞lint percentage, which is consistent with their mid-parent heterosis value in F₁.Consequently, in cotton breeding, bolls/plant can be considered in priority and otheryield-component traits could be measured as a whole to implement variety enhancementand hybrid selection of cotton. Additionally, due to influence of AE and AAE, the predictedsuperior lines in different environment might be different. In cotton breeding, it is apractical strategy to carry out breeding program according to special environment factors.2. QTL mapping in the immortalized F₂ population and genetic mechanism ofheterosisA set of immortalized F₂ population was constructed by intercrossing of RILs. Theimmortalized F₂ population was planted at Jiangpu in 2004 and 2005, and QTL of yield andyield-component, fiber quality and plant architecture traits were tagged.Of the QTL detected by composite interval mapping method, two QTL for yield andyield-components including the lint percentage QTL qLP-A10-1 and 8 QTL for fiberquality traits including the fiber length QTL qLP-A10-1 could also be detected in the RILpopulation, which showed high stability and might be of special value for marker-assistedselection (MAS).Most of the QTL detected by composite interval mapping method showed partialdominance. Comparison of trait performance among different genotypes in flankingmarkers of these QTL showed that heterozygosity did not always show higher performancethan corresponding homozygosity; only 22.9% of the markers showed overdominance, andthe percentage of overdominance was even lower when considering the existence ofpseudo-overminance. The results indicated that dominance contributed more thanoverdominance for heterosis in XZM2. Correlations between general genotypicheterozygosity and trait performance were not significant except for minor traits in theimmortalized F₂ Population, neither were correlations between genotypic heterozygosityand mid-parent heterosis, and some RILs had even higher performance than F₁ on all traits,which further confirmed the above conclusions. Epistatic QTL were detected for most of the traits, and they explained comparativelylarge part of phenotypic variance. When combined with conclusions of the RIL population,the results showed that dominance and epistasis are the major genetic basis of heterosis inXZM2.3. Differential gene expression patterns in leaves between XZM2 and its two parentsAll the genes in hybrid F₁ are derived from its two parents, however, heterosisemerged in hybrid and hybrid performance is different from that of two parents. It can bededuced that differential gene expression between hybrids and parents is responsible forheterosis.In this study, Differential Display Reverse Transcription PCR (DDRT-PCR) was usedto analyze differences of gene expression in XZM2 and its two parents ZMS12 and 8891.Leaves of three growing stages, namely seedling, budding and flowing stages werecollected and analyzed. Six patterns of differential gene expression were detected, whichinclude: fragments observed in either parent but not in F₁ (UNP₁/UNP₂); fragments presentin F₁ but in neither of the parents (UNF₁); fragments observed in both parents but not in F₁(ABF₁); fragments that occurred in either of the parent and F₁ (DMP₁/DMP₂).The distribution order of six patterns is UNF₁＞DMP₂＞UNP₂＞DMP₁＞UNP₁＞ABF₁ (F₁ silenced) when the data of three stages were compared, which is consistent withthe fact that F₁＞P₂＞P₁ in yield level. The result indicated that yield production is highlycorrelated with differential gene expression, and the high gene expression level in F₁ mightbe responsible for heterosis.The differential fragment were cloned and sequenced. The BLASTn results showedthat the genes showing differential expression between XZM2 and its two parents wereinvolved in photosynthesis, respiration, signal transduction and fiber development. Part ofthe differential expression was confirmed by RT-PCR.4. DNA methylation of XZM2 and its two parentsDNA methylation of XZM2 and its two parents in three growing stages was analyzedbased on methylation-sensitive AFLP (MSAP). A total of 64 primer combinations wereused and 139 DNA methylation loci were detected. Four patterns of DNA methylation weredetected, which include: methylation of the internal cytosine in two strands; methylation ofthe external cytosine in one strand; methylation at the external cytosine or both of thecytosine residues in two strands; methylation level decreased, or demethylation.The pattern of methylation at the internal cytosine in two strands constitutes majority of the four patterns; while the pattern of methylation at the external cytosine in one strandranked secondly. The methylation level of F₁ was lower than that of parents, which isconsistent with the highest gene expression level in F₁ detected by DDRT-PCR, indicatingthat low level of DNA methylation and high level of gene expression in F₁ might beresponsible for heterosis. Growing stage specific DNA methylation was observed. DNAmethylation level increased or decreased with the progress of growth, or showingundulation, which can partly explain differential gene expression in different growingstages.