Gene expression modularity reveals footprints of polygenic adaptation in Theobroma cacao

First author: Tuomas Hämälä; Affiliations: University of Minnesota (明尼苏达大学): St Paul, USA

Corresponding author: Peter Tiffin

Separating footprints of adaptation from demography is challenging. When selection has acted on a single locus with major effect, this issue can be alleviated through signatures left by selective sweeps. However, as adaptation is often driven by small allele frequency shifts at many loci, studies focusing on single genes are able to identify only a small portion of genomic variants responsible for adaptation. In face of this challenge, we utilize co-expression information to search for signals of polygenetic adaptation in Theobroma cacao, a tropical tree species that is the source of chocolate. Using transcriptomics and a weighted correlation network analysis, we group genes with similar expression patterns into functional modules. We then ask whether modules enriched for specific biological processes exhibit cumulative effects of differential selection in the form of high FST and dXY between populations. Indeed, modules putatively involved in protein modification, flowering, and water transport show signs of polygenic adaptation even though individual genes that are members of those groups do not bear strong signatures of selection. Modelling of demography, background selection, and the effects of genomic features reveal that these patterns are unlikely to arise by chance. We also found that specific modules were enriched for signals of strong or relaxed purifying selection, with one module bearing signs of adaptive differentiation and an excess of deleterious mutations. Our results provide insight into polygenic adaptation, and contribute to understanding of population structure, demographic history, and genome evolution in T. cacao.

从群体结构中分离出适应性足迹是十分具有挑战性的工作。当选择作用于起主效应的单个位点上时，通过选择性清除留下的信号我们可以降低解决该问题的难度。然而，由于适应性通常由多位点上等位基因频率小的改变共同驱动，对于单个基因的研究只能鉴定作用于适应性的很小一部分的基因组变异。面对这个挑战，作者利用共表达的信息搜索热带植物可可树中多基因适应性信号。利用转录组学与加权相关网络分析WGCNA，作者将具有类似表达模式的基因聚成功能模块。接着，作者检测那些富集特定生物学进程的模块是否会存在不同群体之间差异选择的累积效应，这种群体间的差异选择主要通过高的FST和dXY来体现。确实，作者发现参与蛋白修饰、开花以及水分运输的模块存在多基因适应的信号，尽管这些模块中的单个基因并不存在强烈的选择效应。通过对群体结构、背景选择及基因组特性影响的建模，作者确认这些结果不太可能是偶然出现的。同时，作者还发现有的模块会富集强烈的纯化选择信号，而另外一些模块则富集宽松的纯化选择信号，仅只有一个模块存在适应分化的信号和过剩的有害突变。本文的研究结果为多基因适应的研究提供了新的视野，同时有助于我们进一步理解可可树的群体结构、族群演化历史以及基因组演化。

通讯：Peter Tiffin (https://cbs.umn.edu/contacts/peter-tiffin)

研究方向：植物免疫系统基因的分子演化；植物对于CO2浓度增加以及其他环境变化的演化响应；物种区域扩张的生态和演化限制。

doi: https://doi.org/10.1093/molbev/msz206

Journal: Molecular Biology and Evolution

Published online: September 10, 2019