Polyploidy—whole‐genome duplication—is a key process in plant evolution, with one or more duplication events evident in the genomes of nearly all land plant lineages (Leebens‐Mack et al., 2019). Although polyploidization often also involves hybridization (allopolyploidy), it is clear that doubling the genome, in and of itself (autopolyploidy), is a macromutation: a single event with many phenotypic consequences. The mechanisms by which this macromutation alters phenotypes remain mysterious, particularly at the cellular level (Doyle and Coate, 2019), but at least some of them fall under the broad heading of “epigenetics”, which Cavalli and Heard (2019, p. 489) define as “molecules and mechanisms that can perpetuate alternative gene activity states in the context of the same DNA sequence”. Doubling a diploid genome creates an autotetraploid that typically will differ in gene expression and phenotype from its isogenic diploid progenitor (e.g., Robinson et al., 2018; Corneillie et al., 2019) and can also differ in classic epigenetic features such as methylation pattern (Zhang et al., 2015). Such an autopolyploid thus has qualitatively “the same DNA sequence” as its diploid progenitor yet has “alternate gene activity states” that it can “perpetuate” over many generations. In short, autopolyploidy is, in and of itself, an epigenetic macromutation. Accordingly, techniques for studying the epigenome have begun to shed new light on what autopolyploidy “does”, affording insights into the direct effects of genome doubling minus the dramatic impact of hybridity that characterizes allopolyploids (e.g., Wendel et al., 2018), and which also do not involve millions of years of adaptation to changes in genome size (e.g., Roddy et al., 2020).

An autopolyploid that is isogenic with its diploid progenitor differs from it quantitatively, in doubled dosage of every gene and in doubled DNA content. Both effects lead to phenotypic changes not only through altered gene expression but also by directly or indirectly increasing the sizes of cells and organelles, leading to quantitative changes such as decreased surface to volume ratios (e.g., of the nucleus or plastid), increased distances molecules must travel (e.g., mRNAs from a larger nucleus to the cytoplasm), and altered concentrations (e.g., of transcription factors in the nucleus). Transcriptional dosage responses to experimentally doubling the genome of Arabidopsis thaliana are nonlinear and variable among genes (Fig. 1A). There is often a global reduction in expression on a per‐genome basis, combined with differential effects on the regulation of particular classes of genes, notably those that are dosage sensitive (Hou et al., 2018; Coate et al., 2020; Song et al., 2020). Neither overall patterns of methylation nor proximity of transposable elements to genes appears to explain these genome‐wide or gene‐class‐specific phenomena, leaving chromatin folding in space and time (the 4D nucleome; Dekker et al., 2017) as a possible mechanism for coordinated responses of numerous unlinked genes.

Hou et al. (2018) suggested that deviation from doubling of the overall mRNA transcriptome in autotetraploid A. thaliana could be a result of increased (but not doubled) cell size. Implication of cell size in global control of transcription brings up the “nucleotype” hypothesis: that cell size and associated phenotypes (e.g., nuclear volume, cell cycle duration) are strongly influenced by bulk DNA content rather than being determined entirely by genotype, perhaps as a consequence of biophysical laws (Bennett, 1971; Doyle and Coate, 2019). What is the relationship between the nucleotype and the epigenome? Altered metabolism is one possibility, not only due to its effects on all aspects of cell biology, including chromatin dynamics (Sharma and Rando, 2017), but also given the interdependence of cell size and metabolic activity, and the impact of gene dosage on metabolic flux (Bekaert et al., 2011). Altered dimensions also affect crowding and compaction, which in the case of chromatin is known to affect gene expression.

New insights into the epigenetics of genome doubling have been provided by a high‐throughput chromosome conformation capture (Hi‐C) study comparing chromatin organization of a synthetic A. thaliana autopolyploid with its Col‐0 diploid progenitor (Zhang et al., 2019; Fig. 1B–F). In the doubled nucleus, individual duplicated chromosomes appear to maintain separate chromosome territories. Nuclear volume increases with genome size (Simova and Herben, 2012), but genome doubling produces less than a doubling in nuclear volume (Sas‐Nowosielska and Bernas, 2016; Robinson et al., 2018), which could affect chromatin compaction and alter contacts among chromosomes. Indeed, Zhang et al. (2019) found increased inter‐chromosomal interactions and reduced intra‐chromosome arm interactions in the autotetraploid. Moreover, about 12% of the genome, including over 2600 genes, changed position between loose vs. condensed structural domains (LSD vs. CSD), which correspond, respectively, to the largely euchromatic, transcriptionally active “A” and heterochromatic, transcriptionally repressed “B” compartments found in animals and many plants. Previously, such alteration of partitioning in a polyploid had been described in a natural allopolyploid (Wang et al., 2018), but there the effect of genome doubling is complicated by hybridity and evolutionary divergence. In the synthetic A. thaliana autopolyploid, there was evidence of increased compaction and greater repressive methylation in CSDs, and looser structure and increased activating methylation in LSDs, but overall histone methylation patterns were not altered dramatically.

Zhang et al. (2019) reported that nearly 750 genes were differentially expressed between the diploid and autotetraploid, many of them involved in stress responses as in other autopolyploids (Coate and Doyle, 2019). Differential expression did not correlate with changed methylation pattern, but over 70% of genes with altered expression were located in regions of the genome that showed changes in their levels of cis‐ and trans‐interactions; in particular, promoter–promoter interaction frequency was higher in the autotetraploid. In contrast to this strong effect of interaction, fewer than 30 of the 2600 genes in genomic regions that shifted between CSDs and LSDs were differentially expressed, suggesting that chromatin interaction could be a more important determinant of gene expression shifts than overall chromatin compaction when the genome doubles. Among the differentially expressed genes was the flowering time repressor, FLC, whose increased expression correlated with altered cis‐chromatin interactions in the later‐flowering autopolyploid. The involvement of a gene known to be affected by cell size (Ietswaart et al., 2017) suggests general links among altered chromatin, gene expression, and novel functional traits produced by genome multiplication. Importantly, this example concerns a phenotype that could certainly be acted upon by natural selection, thus emphasizing the evolutionary relevance of the epigenome in autopolyploidy.

Additional studies that integrate epigenomics, nucleome architecture, gene expression, and phenomics will ultimately be necessary to understand fully the effects and consequences of genome multiplication and whether there are “rules” that dictate responses to polyploidization. For one thing, the Zhang et al. (2019) study used tissue from aerial parts of whole seedlings, and as interesting as their results are, they represent the average effect of genome doubling on many different cell types. But not all cell types respond to genome duplication the same way—or at all. Robinson et al. (2018) found that although both epidermal pavement cells and stomatal guard cells increased in area with increased ploidy, the two cell types had different response curves. Even more dramatically, Katagiri et al. (2016) found that although leaf epidermal pavement cells were larger in synthetic A. thaliana autopolyploids, palisade mesophyll cells were not. Moreover, when Katagiri et al. (2016) induced palisade cells to express ATML1, a transcription factor whose ectopic expression generates epidermal features, the transformed palisade cells behaved like epidermal cells and enlarged in response to polyploidization. Consequently, any rules governing cell size and other phenotypic responses to polyploidy are likely specific to a particular cell type. It is now possible to study plant cell biology at single‐cell resolution, not only transcriptomically, but with Hi‐C (Zhou et al., 2019), and we look forward to the application of these approaches to synthetic autopolyploids.

Identifying the connections between phenotypes and chromatin‐level variation is a key step toward mechanistic understanding of any epigenetic phenomenon, but it is only a first step. Separating cause from effect is hindered by the observation that the same phenotype can be produced either genetically, generally by altering gene function, or epigenetically, by mis‐regulation due to alteration of chromatin context (Zoghbi and Beaudet, 2016). Moreover, mutations in genes directly involved in epigenetic processes (e.g., chromatin remodelers) can cause phenotypes by altering the epigenetic states at one or more downstream loci. Zoghbi and Beaudet (2016, p. 11) conclude that all of these effects “are not caused by epigenetic mutations, but the mutated genes secondarily alter chromatin states that are critical components of the epigenotype”. We suggest that whole organism autopolyploids could be a useful model for dissecting the effects of epigenetic modifications of phenotypes because in autopolyploids the alteration of chromatin is not the direct result of a specific genetic mutation. Rather, even though gene action could certainly be involved through feedbacks involving metabolic changes caused by nucleotypic and/or gene dosage effects (e.g., increased cell size, organelle number; Fig. 1) it is clear that, in an autopolyploid, genome duplication has a direct effect on chromatin state. In the Zhang et al. (2019) study, each gene whose transcription level or chromatin compartment is altered in the autopolyploid thus represents an opportunity to explore the interrelations between chromatin changes and gene expression in a genetic background unaltered by conventional genetic mutations.

In conclusion, whole‐genome duplication causes sweeping alterations to the 4D nucleome, which likely drive phenotypic changes independent of classic genetic mutation, making autopolyploidy an epigenomic macromutation. Emerging techniques to quantify chromatin‐level changes will yield key insights into the effects of “pure polyploidy” (Spoelhof et al., 2017) and position autopolyploids as key models for understanding epigenetic interactions and their effects on evolutionarily relevant phenotypes.

多倍体-全基因组复制-是植物进化的关键过程，几乎所有陆地植物谱系的基因组中都存在一个或多个复制事件（Leebens-Mack等，2019）。尽管多倍体化通常也涉及杂交（alalpolyploidy），但很明显，将基因组本身加倍（自身多倍体）是一种巨大的突变：单个事件具有许多表型后果。这种宏观突变改变表型的机制仍然是未知的，特别是在细胞水平上（Doyle和Coate，2019年），但是至少其中一些属于“表观遗传学”的广泛标题，Cavalli和Heard（2019年），第 489）定义为“可以在同一DNA序列的背景下延续替代性基因活性状态的分子和机制”。将二倍体基因组加倍会生成一个自四倍体，其基因表达和表型通常会不同于其同基因二倍体祖先（例如Robinson等，2018； Corneillie等，2019），并且在经典表观遗传特征（例如甲基化模式）方面也可能有所不同（Zhang等人，2015）。因此，这种同倍体在质量上与其二倍体祖先“相同的DNA序列”，却具有可以“延续”许多代的“替代基因活性状态”。简而言之，自身多倍体本身就是表观遗传的宏变异。因此，用于研究表观基因组的技术已开始为自多倍体“做什么”提供了新的思路，从而洞悉了基因组倍增的直接作用，减去了表征异源多倍体的杂交的巨大影响（例如，Wendel等人，2018年），以及这也不涉及对基因组大小变化进行数百万年的适应（例如，Roddy等人，2020年）。

与其二倍体祖细胞同基因的异源多倍体在数量上与它不同，不同之处在于每个基因的剂量加倍，DNA含量加倍。这两种作用不仅通过改变基因表达，而且通过直接或间接增加细胞和细胞器的大小，导致表型变化，从而导致定量变化，例如表面积/体积比（例如核或质体）降低，分子距离增加必须传播（例如，从较大核到细胞质的mRNA），并改变浓度（例如，核中转录因子的浓度）。转录剂量对拟南芥基因组实验倍增的反应是非线性的，并且在基因之间是可变的（图1A）。在每个基因组的基础上，全球表达通常会下降，再加上对特定基因类别（尤其是对剂量敏感的基因）的调控差异（Hou等人，2018 ; Coate等人，2020 ; Song等人，2020）。甲基化的整体模式或转座因子与基因的接近都似乎无法解释这些基因组范围内或基因类特异的现象，从而使染色质在空间和时间上折叠（4D核仁; Dekker等人，2017）用于众多未连锁基因的协调反应。

侯等人。（2018）提出在拟四倍体拟南芥中总mRNA转录组倍增的偏离可能是细胞大小增加（但不是增加）的结果。细胞大小在全局转录控制中的含义提出了“核型”假说：细胞大小和相关的表型（例如核体积，细胞周期持续时间）受大量DNA含量的强烈影响，而不是完全由基因型决定，生物物理定律的结果（本尼特，1971年;多伊尔和科特，2019年）。核型和表观基因组之间有什么关系？改变代谢是一种可能性，不仅因为它对细胞生物学各个方面的影响，包括染色质动力学（Sharma and Rando，2017），而且考虑到细胞大小和代谢活性的相互依赖性以及基因剂量对代谢的影响通量（Bekaert et al。，2011）。尺寸改变也会影响拥挤和压实，就染色质而言，这会影响基因表达。

高通量染色体构象捕获（Hi-C）研究提供了对基因组倍增表观遗传学的新见解，该研究比较了拟南芥自多倍体与其Col-0二倍体祖细胞的染色质组织（Zhang等，2019 ;图1B–F）。在双核中，单个重复的染色体似乎保持着独立的染色体区域。核体积随基因组大小的增加而增加（Simova和Herben，2012年），但基因组的倍增所产生的核量却不到一倍（Sas-Nowosielska和Bernas，2016年； Robinson等人，2018年），这可能会影响染色质的紧缩并改变接触在染色体之间。确实，张等人。（2019）发现同源四倍体中染色体间相互作用增加而染色体内臂相互作用减少。此外，约12％的基因组（包括2600多个基因）在松散结构域与压缩结构域之间（LSD与CSD）改变了位置，分别对应于大部分为常色的，转录活跃的“ A”和经转录抑制的异色的在动物和许多植物中发现的“ B”格。以前，多倍体中的这种分配变化已在天然同质多倍体中进行了描述（Wang等，2018），但是由于杂交和进化差异，基因组倍增的作用变得复杂。在人工拟南芥中 同源多倍体，有证据表明CSD中压实度增加和抑制性甲基化程度更高，而LSD中松散的结构和活化甲基化程度增加，但总体组蛋白甲基化模式没有明显改变。

张等。（2019）报道了近750个基因在二倍体和同源四倍体之间差异表达，其中许多与其他同源多倍体一样参与了应激反应（Coate和Doyle，2019年）。差异表达与甲基化模式改变无关，但是超过70％的具有改变表达的基因位于基因组区域，显示出顺式和反式水平的变化互动 特别是，同源四倍体中启动子-启动子的相互作用频率更高。与这种强大的相互作用相反，在CSD和LSD之间转移的基因组区域中的2600个基因中，只有不到30个被差异表达，这表明当基因组整体染色质比整体染色紧密时，染色质相互作用可能是更重要的基因表达变化的决定因素。加倍。在差异表达的基因中有开花时间阻遏物FLC，其表达增加与后来开花的多倍体中顺式染色质相互作用的改变有关。已知受细胞大小影响的基因的参与（Ietswaart等人，2017）表明染色质改变，基因表达和基因组繁殖产生的新功能性状之间的一般联系。重要的是，这个例子涉及的表型肯定会受到自然选择的影响，从而强调了表观基因组在多倍体中的进化相关性。

最终需要进行整合表观基因组学，核仁结构，基因表达和表观学研究的其他研究，以充分了解基因组倍增的影响和后果以及是否存在“规则”决定对多倍体化的反应。一方面，张等人。（2019）研究使用了整个幼苗空中部分的组织，结果令人感兴趣，它们代表了基因组倍增对许多不同细胞类型的平均作用。但是，并非所有的细胞类型都对基因组重复产生相同或完全相同的反应。罗宾逊等。（2018年）发现，尽管表皮路面细胞和气孔保卫细胞的面积均随着倍性的增加而增加，但两种细胞的响应曲线却有所不同。更令人震惊的是，Katagiri等人。（2016）发现，尽管人工合成拟南芥自多倍体中的叶表皮路面细胞较大，但栅栏叶肉细胞却没有。此外，当Katagiri等人。（2016）诱导栅栏细胞表达ATML1，一种异位表达产生表皮特征的转录因子，转化后的木栅细胞表现得像表皮细胞，并在多倍体化反应中扩大。因此，任何控制细胞大小和对多倍性的其他表型反应的规则都可能特定于特定细胞类型。现在有可能以单细胞分辨率研究植物细胞生物学，不仅是转录组学，而且还可以使用Hi-C（Zhou et al。，2019），并且我们期待将这些方法应用于合成同源多倍体。

识别表型和染色质水平变异之间的联系是对任何表观遗传现象进行机械理解的关键一步，但这只是第一步。由于观察到相同的表型可以通过遗传方式产生，通常是通过改变基因功能，也可以是表观遗传的，即由于染色质环境的改变而引起的调节失调，因而阻碍了将结果从结果中分离出来（Zoghbi and Beaudet，2016）。此外，直接参与表观遗传过程的基因中的突变（例如染色质重塑剂）可通过改变一个或多个下游基因座的表观遗传状态而引起表型。Zoghbi和Beaudet（2016，第 11）得出结论，所有这些影响“不是由表观遗传突变引起的，而是突变的基因继而改变了染色质状态，而染色质状态是表观基因型的关键组成部分”。我们建议整个生物体多倍体可能是剖析表型表观遗传修饰作用的有用模型，因为在多倍体中染色质的改变不是特定基因突变的直接结果。确切地说，尽管基因作用肯定可以通过涉及由核型和/或基因剂量效应（例如，细胞大小增加，细胞器数目增加；图1）引起的代谢变化的反馈来参与，但很明显，在同倍体中，基因组重复具有直接影响染色质状态。在张等人。（2019）研究中，每个基因在其多倍体中的转录水平或染色质区室都发生了变化，因此代表了探索在传统遗传突变未改变的遗传背景中染色质变化与基因表达之间相互关系的机会。

总之，全基因组复制会导致4D核仁发生彻底改变，这可能会导致表型变化，而与经典的基因突变无关，从而使多倍体成为表观基因组的大突变。量化染色质水平变化的新兴技术将产生对“纯多倍体”效应的关键见解（Spoelhof等人，2017年），并将同倍体定位为理解表观遗传相互作用及其对进化相关表型的影响的关键模型。