Sela Itamar, Wolf Yuri I, Koonin Eugene V
National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, MD, 20894, USA.
BMC Biol. 2021 Feb 10;19(1):27. doi: 10.1186/s12915-021-00960-2.
The genomes of bacteria and archaea evolve by extensive loss and gain of genes which, for any group of related prokaryotic genomes, result in the formation of a pangenome with the universal, asymmetrical U-shaped distribution of gene commonality. However, the evolutionary factors that define the specific shape of this distribution are not thoroughly understood.
We investigate the fit of simple models of genome evolution to the empirically observed gene commonality distributions and genome intersections for 33 groups of closely related bacterial genomes. A model with an infinite external gene pool available for gene acquisition and constant genome size (IGP-CGS model), and two gene turnover rates, one for slow- and the other one for fast-evolving genes, allows two approaches to estimate the parameters for gene content dynamics. One is by fitting the model prediction to the distribution of the number of genes shared by precisely k genomes (gene commonality distribution) and another by analyzing the distribution of the number of genes common for k genome sets (k-cores). Both approaches produce a comparable overall quality of fit, although the former significantly overestimates the number of the universally conserved genes, while the latter overestimates the number of singletons. We further explore the effect of dropping each of the assumptions of the IGP-CGS model on the fit to the gene commonality distributions and show that models with either a finite gene pool or unequal rates of gene loss and gain (greater gene loss rate) eliminate the overestimate of the number of singletons or the core genome size.
We examine the assumptions that are usually adopted for modeling the evolution of the U-shaped gene commonality distributions in prokaryote genomes, namely, those of infinitely many genes and constant genome size. The combined analysis of genome intersections and gene commonality suggests that at least one of these assumptions is invalid. The violation of both these assumptions reflects the limited ability of prokaryotes to gain new genes. This limitation seems to stem, at least partly, from the horizontal gene transfer barrier, i.e., the cost of accommodation of foreign genes by prokaryotes. Further development of models taking into account the complexity of microbial evolution is necessary for an improved understanding of the evolution of prokaryotes.
细菌和古菌的基因组通过基因的大量丢失和获得而进化，对于任何一组相关的原核生物基因组来说，这都会导致形成一个泛基因组，其基因共性呈现普遍的、不对称的U形分布。然而，决定这种分布具体形状的进化因素尚未被完全理解。
我们研究了基因组进化的简单模型与33组密切相关细菌基因组的经验观察到的基因共性分布和基因组交集的拟合情况。一个具有无限外部基因库可供基因获取且基因组大小恒定的模型（IGP-CGS模型），以及两个基因周转率，一个用于缓慢进化的基因，另一个用于快速进化的基因，允许两种方法来估计基因内容动态的参数。一种方法是将模型预测与恰好k个基因组共享的基因数量分布（基因共性分布）进行拟合，另一种方法是分析k个基因组集合共有的基因数量分布（k核）。尽管前者显著高估了普遍保守基因的数量，而后者高估了单拷贝基因的数量，但两种方法产生的总体拟合质量相当。我们进一步探讨了去掉IGP-CGS模型的每个假设对基因共性分布拟合的影响，结果表明，具有有限基因库或基因丢失和获得速率不相等（更大的基因丢失率）的模型消除了对单拷贝基因数量或核心基因组大小的高估。
我们研究了通常用于模拟原核生物基因组中U形基因共性分布进化的假设，即无限多个基因和恒定基因组大小的假设。基因组交集和基因共性的综合分析表明，这些假设中至少有一个是无效的。这两个假设的违背反映了原核生物获得新基因的能力有限。这种限制似乎至少部分源于水平基因转移障碍，即原核生物容纳外源基因的成本。考虑微生物进化复杂性的模型的进一步发展对于更好地理解原核生物的进化是必要的。
