醋栗番茄全基因體分子標誌之開發與探究控制番茄雄蕊長度之候選基因

Development of Genome-Wide High-Density SNP Markers in Solanum pimpinellifolium and Investigation of Candidate Loci of Stamen Length in Tomato

林亞平(Ya-Ping Lin)

指導教授 ： 陳凱儀

國立臺灣大學/生物資源暨農學院/農藝學研究所/博士(2019年)

https://doi.org/10.6342/NTU201900689

摘要

自達爾文提出演化論後，異型花的遺傳機制一直是植物學家感興趣的議題之一。古典的研究認為控制自交不親和性與異花型的基因緊密連鎖，稱之為S基因座，這些調控相關性狀且緊密連鎖的基因稱為超級基因。一般認為植物從異交演化成自交的過程中會先失去自交不親和性，而後在超級基因內發生重組，使得異型花變成同型花，以確保自交成功的機率。然而伴隨著分子技術的進步，現今的分子證據卻顯示同型花可能是由半合子造成，而非傳統上認為的罕見重組。在農業上，研究此議題可以了解作物在馴化過程中，受到強烈選拔壓力後對基因體造成的改變。此外，也可控制作物的自交不親和性與花型，如此不僅能有效地生產雜交種子，也能藉由提高授粉率而增加產量。
醋栗番茄為野生番茄的一種，原生於秘魯與厄瓜多沿岸，是栽培番茄的近親。因為醋栗番茄具有許多抗病性狀，且可與栽培種番茄相互雜交，故為重要的番茄種原之一。目前醋栗番茄已提供番茄育種工作上一些抗病基因座，也應用於農藝性狀相關的全基因體關聯性定位中。前人研究發現醋栗番茄可分成異交、自交與中間型三種交配系統，異交的醋栗番茄不僅具有較高的遺傳歧異度，且具有較突出的雌蕊。由於醋栗番茄在交配系統與花的形態上都具有多型性，故很適合利用於自交不親和性與異型花的研究。
現今分子標誌已廣泛地應用於作物育種上，伴隨著次世代定序的成本下降，開發重要種原的全基因體分子標誌已成為基礎的育種工作。本研究針對99個醋栗番茄收集系進行PstI限制酶關聯性定序，定序範圍涵蓋12,790個基因。我們一共得到24,330個單一核苷酸多型性分子標誌，其中16,365個分子標誌座落於7,383個基因上。我們觀察到定序範圍與基因的分佈類似，顯示使用PstI限制酶來篩選基因體片段的策略適合應用於尋找候選基因的研究上。此外，該族群可以分成三個先祖次族群及四個混合基因體的次族群。主成份分析、成對Fst、AMOVA皆支持這樣的分群，顯示這組高密度分子標誌可穩定估計族群結構。接著，我們估計該族群整體的連鎖衰變在18千鹼基對，意味著此族群可以在全基因體關聯性分析得到精密的解析度，甚至可以定位到單一基因。然而要滿足這樣的解析度，至少需要50,000個分子標誌。
在雄蕊長度的全基因體關聯性定位中，我們利用98個醋栗番茄收集系進行混合線性模式分析，定位到三個候選基因座，但這三個基因座皆為高度錯誤發現率。由於全基因體關聯性定位的檢定力及錯誤發現率皆與研究樣本的族群大小有關，我們建議兩種增加樣本的方式，一是在各個次族群中均勻地增加取樣數目，這個方法也可能使罕見對偶基因變成一般對偶基因，故也可能增加分子標誌的數目。另一個方法是在秘魯北部增加取樣數目，因為此處是醋栗番茄的發源地，遺傳歧異度大，也可能增加對偶基因數。
另一方面，前人研究顯示style2.1下游附近有兩個控制雄蕊長度的基因 stamen2.2及stamen2.3，我們利用轉錄體組定序來挑選這兩個候選基因，使用的材料是栽培種番茄品系M82及其滲透系TA3178，TA3178在style2.1附近是野生番茄（潘那利番茄）的染色體片段。我們藉由單一核苷酸多型性的數量差異來界定滲透片段的範圍。接著，依據本研究室之前的結果，我們篩選從標誌cLED19A24到CT9該區間的18個候選基因，比較這些候選基因的表現量及多型性後，發現Solyc02g087960.2、Solyc02g087970.1與Solyc02g088070.2可能是stamen2.2及stamen2.3的候選基因。

關鍵字

番茄 ； 雄蕊長度 ； 族群基因體學 ； 關連性分析

並列摘要

Botanists have been fascinated by the genetic mechanism of heterostyly since Darwin’s theory of evolution. It was believed that the genes controlling self-incompatibility and floral morphology were linked tightly, so-called S-locus. According to the classical evolutionary studies, when a plant evolved from outcrossing to selfing, it was necessary to lose self-incompatibility and then adjusted the positions of male and female floral organs through the rare recombination within the S-locus. However, new evidence suggested that homostyly resulted from hemizygote rather than the rare recombination. In agriculture, studying the genetic mechanism of self-incompatibility and heterostyly can understand the changes of crop genomes under the selection forces during domestication processes. Additionally, it can accelerate the production of hybrid seeds or ensure the pollination to increase yield.
Solanum pimpinellifolium is a wild tomato originated from the coastal region of Peru and Ecuador. It serves as an important germplasm in tomato breeding programs because it displays many resistant traits and can freely cross to cultivated tomatoes. Previous studies classified this species as complete or near complete allogamy, complete autogamy and intermediate type based on its mating system. In addition, allogamous accessions displayed higher genetic diversity and more exsertion of stigma than autogamous ones. Because S. pimpinellifolium contains the variations of outcrossing rate and floral morphology within its own species, it could be an ideal material to study the genetic mechanism of self-incompatibility and heterostyly.
Nowadays, molecular markers have been applied to crop breeding extensively. Accompanying by the cost down of next generation sequencing, the development of genome-wide high-density markers for germplasm becomes essential in breeding programs. In this research, we performed the PstI-digested associated DNA sequencing for 99 accessions of S. pimpinellifolium, resulting in 24,330 SNPs. The coverage extended to 12,790 genes, and a total of 7,383 genes were targeted directly by 16,365 SNPs. Besides, the sequencing regions and the annotated genes presented similar distributions through each chromosome. This suggested that PstI-digested associated DNA sequencing was an appropriate strategy to investigate candidate genes. This collection was divided into three subpopulations of single-ancestral genome and four subpopulations of mix-ancestral genome by ADMIXTURE. Principle component analysis, pairwise Fst and AMOVA all supported the subpopulations, implying this set of high-density markers was capable to estimate the subpopulations stably. Moreover, the overall LD decay was within 18 Kb, suggesting a fine resolution in genome-wide association study even to a single-gene level. However, to achieve such fine resolution, at least 50,000 markers were required.
Three candidate loci controlling stamen length were identified via the mixed linear model in genome-wide association study of 98 S. pimpinellifolium accessions, but all three loci presented high false discovery rate. Since the power and false positive rate of genome-wide association study depend on the sample size of a studying population, we suggest two approaches to increase sample size. One is to increasing samples in each subpopulation evenly. This approach can potentially make rare alleles to common alleles by increasing the allele frequency. The other is to sampling more individuals in the northern Peru because the accessions in the northern Peru present more genetic diversity. This approach can also increase both rare alleles and common alleles.
On the other hand, following the previous studies, stamen2.2 and stamen2.3 were located in the downstream interval next to style2.1. We performed a RNA sequencing experiment of M82 and TA3178. TA3178 is an introgression line of M82 and contains a segment of Solanum pennellii near style2.1. We identified this introgression region by comparing the difference of SNPs between these two lines. Afterwards, following the previous work in our team, we screened 18 candidate genes from marker cLED19A24 to CT9 by comparing the fold change and cDNA polymorphism between M82 and TA3178. This result suggested that Solyc02g087960.2, Solyc02g087970.1 and Solyc02g088070.2 should be the candidates of stamen2.2 and stamen2.3.

並列關鍵字

Solanum pimpinellifolium ； stamen length ； population genomics ； GWAS

參考文獻

Chapter 1

Anderson, A. D., & Weir, B. S. (2007). A maximum-likelihood method for the estimation of pairwise relatedness in structured populations. Genetics. https://doi.org/10.1534/genetics.106.063149

Astle, W., & Balding, D. J. (2010). Population Structure and Cryptic Relatedness in Genetic Association Studies. Statistical Science. https://doi.org/10.1214/09-sts307

Auer, P. L., & Doerge, R. W. (2010). Statistical design and analysis of RNA sequencing data. Genetics. https://doi.org/10.1534/genetics.110.114983

Bedinger, P. A., Chetelat, R. T., McClure, B., Moyle, L. C., Rose, J. K. C., Stack, S. M., … Royer, S. (2011). Interspecific reproductive barriers in the tomato clade: Opportunities to decipher mechanisms of reproductive isolation. Sexual Plant Reproduction. https://doi.org/10.1007/s00497-010-0155-7