【作者】 李博；

【导师】 张志毅；

【作者基本信息】 北京林业大学 ， 林木遗传育种， 2009， 博士

【摘要】 长期以来,复杂数量性状遗传学研究一直是遗传学研究的重点和难点,这是因为:一方面,多数的经济性状都属于数量性状,深入揭示这些复杂性状的遗传学机制是对其进行遗传改良的前提;另一方面,复杂数量性状通常由多基因协同调控,其遗传机制相当复杂。QTL作图和分离QTL内主效基因的成功,为开展复杂性状的研究奠定了重要的理论和实践基础。木本植物的遗传连锁作图和QTL定位研究发展较快,许多树种都构建了较高密度的基因组DNA遗传连锁图谱,并且定位了许多重要性状的QTL位点。然而,多年生木本植物的生殖周期较长、遗传背景复杂等特点严重制约了通过图位克隆（positional cloning）分离QTL的发展,长期以来使得木本植物的QTL研究仅限于理论水平。杨树是木本植物遗传学研究的模式物种,也是第一个全基因组测序的木本植物,其遗传学资源十分广泛。本研究通过利用毛白杨（Populus tomentosaCarr.）与毛新杨（P.tomentosa×P.bolleana）的F₁代回交群体,构建了两张特异组织材料的转录组图谱（transcriptome maps）,并利用该转录组图谱对包括生长、生理和木材品质在内的28个性状进行了QTL分析。同时借助杨树基因组学和转录组学手段,对QTL区间内的候选基因进行了分析和筛选,显著地缩小了候选基因的筛选范围,为将来分离林木QTL内主效基因奠定了重要基础。主要研究结果如下:（1）利用毛新杨与毛白杨回交子代7年生群体,构建了木本植物首张未成熟木质部转录组图谱,其中毛白杨的转录组图谱包括18个较大的连锁群,全长为1536cM,毛新杨转录组图谱由13个较大的连锁群构成,总长度为1030cM,两者分别覆盖预测基因组长度的72.3%和67.1%。为了进一步改善图谱质量,成功的构建了毛白杨和毛新杨的根萌苗幼化群体,并利用该群体构建了根萌苗转录组图谱。随后,对两组特异组织转录组图谱进行了比较分析和整合研究,整合后形成的图谱在图谱长度、标记密度以及标记平均距离上均得到明显改善。（2）在构建三组转录组图谱（木质部图谱、根萌苗图谱及整合图谱）基础上,对群体的生长、光合生理以及木材品质等28个性状进行了QTL定位分析,并对各图谱的QTL定位结果进行了比较。利用单标记作图法,共获得402个相关联的标记,其中7个生长性状检测到127个相关联标记,10个光合生理性状检测到166个相关标记,同时检测到109个与11个木材品质性状相关联的标记。利用整合图谱,采用区间作图策略,发现了大量的QTL位点,当LOD＞1时,共检测到118个QTL位点,其中LOD＞3的位点有10个,单个QTL对表型变异的贡献量一般小于10%,但是也检测到了若干个贡献量在30%以上的主效QTL位点。（3）利用QTLs两侧的标记序列,将14个QTLs位点从遗传连锁图谱上转移到杨树基因组上,对QTLs区间内的基因数目、基因频率、基因功能进行了分析。发现所得QTL位点在基因组上的物理距离在500Kb到3500Kb之间,QTLs内基因数目由34到284个不等,表现出在该杂交群体的生殖过程中,基因组不同染色体区域内迥异的交换频率分布。（4）利用杨树芯片数据库（UPSC-BASE）中668张杨树表达芯片数据,构建了杨树基因共表达模型,进而利用该模型对木材形成过程中,可能参与纤维素合成和木质素合成的基因进行了预测。在预测出的可能参与木材形成的基因中,有部分基因已经证明的确参与了木材形成过程,表明该模型具有很好的预测能力。同时利用该模型检测出大量的未知功能基因,这些基因的功能将在后续工作中加以证明。（5）应用上述基因共表达模型,对2个控制纤维素含量和1个控制木质素含量的QTLs内基因进行候选基因筛选,缩小控制QTL内主效基因的筛择范围。利用两个纤维素合成酶PCESA4和PTCESA7,对控制纤维素含量的QTLαCC4内基因进行共表达分析,预测出该QTL内estExt_Genewisel_v1.C_LG_X12954（PU07734）、fgenesh4_pg.C_LG_XI000690（PU05329）和grail3.0062007201（PU01733）最有可能是该QTL的主效基因。根据同样的策略,对QTLαCC5和QTLLC1两个位点内的主效基因进行了预测。更多还原

【Abstract】 For years,genetics of complex traits is always situated in the most difficult center of genetics because most economic traits are quantitative nature and are the most significant targets for improvement which needs more understanding of the genetic mechanisms underlying the complex traits.However,the facts that hundreds of gene networks contribute to the formation of phenotype lead to a worse situation.The success of QTL mapping and major QTL cloning built a theoretical and practical foundation for research on genetics of complex traits.Recently,linkage mapping and QTL analysis involving with woody plants have been developed rapidly,and plenty of tree species established dense genomic DNA linkage maps and abundant QTLs controlling the important traits had been located on the genetic maps.However,the nature of long reproductive cycle and high heterozygous genomic background for perennial plants blocks the development of positional-cloning the QTLs in practical phase.Populus L.which had been full-sequenced and accumulated abundant informative resources is a model species for genetic research on woody plants.In this study,a F₁ backcross population of（Populus tomentosa×Populus bolleana）×Populus tomentosa was used for constructing the first transcriptome maps for woody plants and QTLs analysis was carried out for 28 traits related to growth, photosynthetic traits and wood quality.Furthermore,gene contents within QTL genomic intervals had been deeply analyzed and some most potential QTL candidate genes had been screened out by introducing the genomic and transcriptomic tools.This study is first try to identify QTL in woody plant genome by focusing on the candidate gene selection with gene co-expression prediction approach which can dramatically shrink the primary candidate gene pool and efficiently picking up the most potential genes for further research.In detail,as follow:1) Two separate transcriptome maps were constructed with developing xylem for P. tomentosa and its hybrid P.tomentosa×p.bolleana.For P.tomentosa（male parent）,the map comprised of 18 linkage groups which is a total of 1536cM in length,for P.tomentosa×p.bolleana（female parent）,13 linkage groups in length of 1030cM made of the map.The coverage of deduced genome is 72.3%and 67.1%for both parents,respectively.To improve the quality of the transcriptome maps,another transcriptome maps for parents had been constructed with the whole plant of sucker seedling（without root） collecting from the juvenilized population.Afterward,two set of transeriptome maps had been compared and integrated into combined maps which had an obvious improvement in map length,marker density and average marker intervals.2) QTL analysis was carried out for 28 traits with these 3 sets of transcriptome maps （combined maps included） and QTL results had been drawn a comparison.With single marker mapping method,402 associated markers had been detected for all traits,including 127 markers for 7 growth traits,166 for 10 photosynthetic traits and 109 for wood quality traits.By using interval mapping,118 QTLs were detected at LOD>1.0,only 10 QTLs existed when LOD>3.0.For each QTL,contribution to the phenotype variation was limited under 10 percents,but also some major QTLs with over 30 percents contribution for some phenotypes had been discovered.3) Fourteen QTL intervals had been transferred from genetic maps into the Populus genome by anchoring the boundary markers covering the QTLs into the whole sequence of Populus.Analysis had been done on the gene contents within each QTL including gene number,gene frequency and gene function.What had been found is every index varied dramatically depending on different QTLs,for example,the gene number varied from 34 to 284 and the physical length of QTLs ranged from 500Kb to 3500Kb.It can be drawn a conclusion that recombination frequency within hotspot and cold spot changed according to the different chromosome segments.4) A gene co-expression model had been constructed with 668 public Populus cDNA chips data which had been used to predict the potential novel genes required in the process of wood formation.Of the top potential genes required in cellulose synthesis and lignin synthesis,some had been testified for their function in this biological process which also was a guarantee for the prediction with this model.5) With this co-expression model in use,the candidate genes within two QTLs controlling cellulose content and one QTL controlling lignin content had been searched in order to deduce the gene mumber for selection.The results showed that 3 mostly highly co-expressed with two cellulose synthesis enzymes had been discovered including gene model estExt_Genewisel_vl.C_LG_XI2954（PU07734）,fgenesh4_pg.C LG_XI000690 （PU05329） and grail3.0062007201（PU01733）.Using the same methods,corresponding candidate genes had been selected for another two QTLs.更多还原