As a notable illustration of totipotency, somatic embryogenesis (SE) is the developmental reprogramming of somatic cells towards the embryogenesis pathway (Yang and Zhang, 2010). Investigations examining the totipotency process are of great fundamental and practical importance in crop biotechnology. Moreover, high‐frequency regeneration of SE has been limited due to the genotype‐dependent response. To date, the epigenetic molecular basis underlying embryogenic redifferentiation during SE remains largely unexplored.

Plant embryogenesis is accompanied by changes at chromatin level and reprogramming of gene expression, highlighting the central role of epigenetic regulation (Miguel and Marum, 2011). During SE, DNA methylation is continually changing to satisfy cell requirements (Nic‐Can and De‐la‐Peña, 2014). The methylation of DNA is essential to SE (De‐la‐Peña et al. , 2015; Kumar and Van Staden, 2017). Recently, Ji et al. (2019) and Li et al. (2019) also reported DNA methylation variations during plant SE.

SE is the concerted process involving multiple cellular pathways controlled by epigenetic and genetic variability (De‐la‐Peña et al. , 2015; Miguel and Marum, 2011). Genome‐wide dissection of dynamic methylation modification features is conducive to explaining the complex underlying genotype‐dependent SE transdifferentiation at overall level. In this study, a single‐base resolution of genome‐wide bisulfite sequencing (BS‐seq) and transcriptome sequencing was performed to comprehensively analyse the DNA methylation and gene regulatory patterns involved in SE transdifferentiation in two cotton genotypes with distinct embryogenic abilities. Three typical stages of early SE: hypocotyls (HY), nonembryogenic calli (NEC) and primary embryogenic calli (PEC), extending from callus dedifferentiation (NEC‐VS‐HY) to embryogenic redifferentiation (PEC‐VS‐NEC) were examined for BS‐seq (Figure 1a–c). Two genotypes, Yuzao 1 (YZ) with a high embryogenic ability (>80%) and Lumian 1 (LM) with a very low ability (<10%) (Jin et al. , 2006), were selected.

Total methylcytosines (^mCs) were identified at dedifferentiation and embryogenic redifferentiation during early SE in the two genotypes. The percentages of genomic methylation dynamic of ^mCG and ^mCHG had similar patterns among the samples with the opposite of ^mCHH methylation (Figure 1d). The overall ^mCG levels accounted to the highest extent followed by ^mCHG and then ^mCHH (Figure 1e). Notably, ^mCs levels presented different patterns during embryogenic redifferentiation in the two genotypes, continuing to rise in LM but decreased at the PEC stage in YZ (Figure 1e).

The methylcytosine levels of three sequence contexts (^mCG, ^mCHG and ^mCHH) were further overviewed in different genic regions, gene body (exon and intron), 2 kb upstream and downstream of transcription start sites as well. (Figure 1f–h). Results showed that DNA methylation in the three sequence contexts was not evenly distributed among genomic transcriptional elements. Upstream and downstream regions were most highly methylated, particularly for ^mCG. Moreover, to assess DNA methylation between developmental specific stages and between genotypes, hierarchical clustering of methylcytosine levels was performed. The results showed that the global pattern of ^mCG was more distinguishable between the two genotypes than between the developmental stages (Figure 1f), whereas it was more discernible between the developmental stages at the ^mCHH site (Figure 1h). These observations remarkably indicated that methylation levels at the CG site were genotype‐specific, whereas differentiation stage‐specific at the CHH site during early SE process.

The methylcytosine levels of ^mCHH in genome‐wide transcriptional elements were further investigated during early SE in LM and YZ (Figure 1i,j). During embryonic redifferentiation, patterns of CHH methylation showed significant differences in the two genotypes. A lower (significantly declined) level of CHH methylation was observed at YZ_PEC (Figure 1j). This result could, to some extent, explain the highly embryogenic redifferentiation ability in YZ, which suggested that CHH hypomethylation marked and distinguished embryonic redifferentiation.

To further investigate SE initiation promoting methylated genes, the differentially methylated genes (DMGs) were identified and significantly enriched in lipid biosynthetic and metabolic processes in YZ embryonic redifferentiation (Figure 1k). Differentially methylated key genes involved in lipid pathway were confirmed to be transcriptionally affected during embryogenic redifferentiation. The results in the highly embryogenic genotype were consistent with and extended our recent report (Guo et al. , 2019).

Simultaneously, for association analysis of DNA methylation and expression levels at embryonic redifferentiation during SE transdifferentiation in the two genotypes, a cross‐analysis identified 1569 and 1977 genes in two genotypes respectively showing significant variations in both methylation and gene expression (termed codifferential genes) (Figure 1l). Among these genes, 1263/306 and 1606/371 codifferential genes were modified by methylation in their upstream/gene‐body regions, respectively. Furthermore, we quantitatively examined the correlations between variations in DNA methylation and variations in gene expression during SE initiation. The results showed that compared with YZ, there was a higher negative correlation of variations in LM in both upstream and gene‐body regions (Figure 1m,n), which suggested that transcription variations were more negatively modulated by DNA methylation in LM, the SE recalcitrant genotype in cotton.

For successful achievement of plant SE, genotype‐dependent DNA methylation remains crucial. In this study, we reported that CHH demethylation could serve as the critical epigenetic marker and associated with embryonic redifferentiation in the highly embryogenic genotype, while CHH hypermethylation in the recalcitrant genotype, which suggested the negative effect on SE‐associated genes during embryonic redifferentiation. However, future research is necessary to explain how DNA methylation is established and to elucidate the molecular mechanisms regulating SE transdifferentiation.

The systematic epigenetic molecular basis underlying cell totipotency and SE transdifferentiation are poorly understood in plants. Especially, the genotype‐dependent critical methylation features associated with embryogenic redifferentiation remains largely unexplored. In our study, integrated maps of genome‐wide DNA methylomes at single‐base resolution and transcriptomes were generated during cotton SE, spanning cell dedifferentiation to embryogenic redifferentiation, in two genotypes with distinct embryogenic abilities. Dynamic DNA methylation variations and their relationships with transcriptional divergence between different genotypes and developmental stages were globally surveyed. Our data revealed that total methylcytosine (^mC) levels presented a hypomethylation pattern during embryogenic redifferentiation in the highly embryogenic genotype. DNA methylation (^mCG, ^mCHG and ^mCHH) were significantly distributed on genomic up and downstream transcriptional elements. Significantly, the global pattern of ^mCG displayed genotype‐specific, and the ^mCHH pattern was particularly determined to be differentiation stage‐specific during SE process. The hypomethylated ^mCHH notably marked and distinguished embryonic redifferentiation. And differentially methylated genes (DMGs) were significantly enriched in the lipid pathway in embryogenic redifferentiation. Furthermore, systematic association analysis of DNA methylome and transcriptome indicated that gene expression variations were more strongly modulated by DNA methylation in the recalcitrant genotype. Compared with previous significant report of the genome‐wide increase in CHH methylation during SE, using one genotype (Ji et al. , 2019; Li et al. , 2019), our current study characterized CHH hypermethylation in LM with low SE ability, but CHH hypomethylation in YZ with high SE ability during embryogenic redifferentiation process. These results suggested the importance of genotype‐dependent methylation modes. The results in this study revealed a comprehensive overview of genotype‐dependent dynamic DNA methylation associated with regulated gene expression during cotton SE. Our study provides new insights into the underlying epigenetic molecular basis and critical methylation modes associated with embryogenic competence acquisition during SE transdifferentiation, thereby holding great promise for its advancement in recalcitrant plant species.

作为全能性的一个显着例证，体细胞胚发生（SE）是指体细胞向胚发生途径的发育重编程（Yang和Zhang，2010年）。检验全能过程的调查在作物生物技术中具有重要的基础和实践意义。此外，由于基因型依赖性反应，SE的高频再生受到限制。迄今为止，在SE期间胚胎发生再分化的表观遗传分子基础仍未开发。

植物胚胎发生伴随着染色质水平的变化和基因表达的重编程，突出了表观遗传调控的核心作用（Miguel和Marum，2011年）。在SE期间，DNA甲基化不断变化以满足细胞需求（Nic-Can和De-la-Peña，2014年）。DNA的甲基化对SE是必不可少的（De-la-Peña等人，2015 ; Kumar和Van Staden，2017）。最近，Ji等。（2019）和Li等。（2019）还报道了植物SE期间DNA甲基化的变化。

SE是涉及表观遗传和遗传变异性控制的多个细胞途径的协调过程（De-la-Peña等人，2015 ; Miguel和Marum，2011）。全基因组范围内的动态甲基化修饰特征剖析有助于从整体上解释复杂的潜在基因型依赖性SE转分化。在这项研究中，进行了全基因组亚硫酸氢盐测序（BS-seq）和转录组测序的单碱基解析，以全面分析涉及两种具有不同胚发生能力的棉花基因型中SE转分化涉及的DNA甲基化和基因调控模式。检查了BS的早期SE的三个典型阶段：胚轴（HY），非胚性愈伤组织（NEC）和原胚性愈伤组织（PEC），从愈伤组织去分化（NEC‐VS‐HY）延伸至胚性再分化（PEC‐VS‐NEC） -seq（图1a–c）。两种基因型，具有高胚发生能力的玉藻1（YZ）（> 80％）和具有非常低的胚芽发育能力的Lumian 1（LM）（<等。（2006年）。

在两种基因型的早期SE期间，在去分化和胚发生再分化中鉴定出总甲基胞嘧啶（^m Cs）。在样品中，^m CG和^m CHG的基因组甲基化动力学百分比与^m CHH甲基化相反（图1d）。总体上，^m CG含量最高，其次是^m CHG，然后是^m CHH（图1e）。值得注意的是，在两种基因型的胚胎发生再分化过程中，^m Cs的水平呈现出不同的模式，LM继续升高，但在YZ的PEC阶段降低（图1e）。

在不同的基因区域，基因体（外显子和内含子），转录起始位点上游和下游2 kb的不同基因区域，进一步概述了三种序列环境（^m CG，^m CHG和^m CHH）的甲基胞嘧啶水平。（图1f–h）。结果表明，在三个序列范围内，DNA甲基化在基因组转录元件之间分布不均。上游和下游区域甲基化程度最高，特别是对于^m CG。此外，为了评估发育特定阶段之间和基因型之间的DNA甲基化，对甲基胞嘧啶水平进行了分级聚类。结果表明，在全球格局^米CG在两种基因型之间比在发育阶段之间更可区分（图1f），而在^m CHH位点的发育阶段之间更可辨别（图1h）。这些观察结果显着表明，CG位点的甲基化水平是基因型特异性的，而CHH位点在早期SE过程中则是分化阶段特异性的。

在LM和YZ早期的SE早期，进一步研究了全基因组转录元件中^m CHH的甲基胞嘧啶水平（图1i，j）。在胚胎再分化过程中，CHH甲基化的模式在两种基因型中显示出显着差异。在YZ_PEC处观察到CHH甲基化水平较低（显着下降）（图1j）。这个结果可以在一定程度上解释YZ的高度胚胎发生再分化能力，这表明CHH的低甲基化标志着并区分了胚胎重新分化。

为了进一步研究促进SE启动的甲基化基因，鉴定了差异甲基化基因（DMG），并在YZ胚胎再分化过程中显着丰富了脂质生物合成和代谢过程（图1k）。证实参与脂质途径的差异甲基化关键基因在胚胎发生性再分化过程中受到转录影响。高度胚胎发生基因型的结果与我们最近的报道相符并扩展了我们的报道（Guo et al。，2019）。

同时，为了对两种基因型的SE转分化过程中胚胎再分化的DNA甲基化和表达水平进行关联分析，一项交叉分析确定了两种基因型的1569和1977个基因，分别显示了甲基化和基因表达的显着差异（称为共差异基因）（图1l）。在这些基因中，分别在其上游/基因体区域中通过甲基化修饰了1263/306和1606/371共差异基因。此外，我们定量检查了SE启动过程中DNA甲基化变异与基因表达变异之间的相关性。结果表明，与YZ相比，上游和基因体区域的LM变异具有更高的负相关性（图1m，n），

为了成功实现植物SE，依赖基因型的DNA甲基化仍然至关重要。在这项研究中，我们报道了CHH去甲基化可以作为重要的表观遗传标记，并在高度胚胎发生的基因型中与胚胎再分化有关，而CHH过度甲基化的顽固性基因型则表明对SE相关基因在胚胎再分化过程中具有负面影响。但是，未来的研究是必要的，以解释如何建立DNA甲基化并阐明调节SE转分化的分子机制。

植物对细胞全能和SE转分化的系统表观遗传学基础知之甚少。特别是，与胚发生再分化有关的基因型依赖性关键甲基化特征仍未得到充分探索。在我们的研究中，在棉花SE期间，生成了两种具有独特胚胎发生能力的基因型，涵盖了从细胞去分化到胚胎发生再分化的全基因组DNA单基因组分辨率和转录组整合图。全局调查了动态DNA甲基化变异及其与不同基因型和发育阶段之间转录差异的关系。我们的数据显示，总甲基胞嘧啶（^mC）水平在高度胚发生基因型的胚发生再分化过程中呈现低甲基化模式。DNA甲基化（^m CG，^m CHG和^m CHH）显着分布在基因组上下游转录元件上。值得注意的是，^m CG的整体模式显示出基因型特异性，并且特别确定^m CHH模式在SE过程中是分化阶段特异性的。低甲基化的^mCHH显着地标志着和杰出的胚胎再分化。并且在胚胎发生性再分化中，脂类途径中的差异甲基化基因（DMGs）明显富集。此外，对DNA甲基化组和转录组的系统关联分析表明，顽固基因型中的DNA甲基化对基因表达变化的影响更大。与以前的一项重要报道相比，使用一种基因型在SE中全基因组CHH甲基化增加（Ji等人，2019 ; Li等人，2019），我们目前的研究的特征是在胚胎发生再分化过程中，具有低SE能力的LM中CHH甲基化较高，而具有高SE能力的YZ中的CHH低甲基化。这些结果表明基因型依赖性甲基化模式的重要性。这项研究的结果揭示了棉花SE期间基因型依赖性动态DNA甲基化与调控基因表达相关的全面概述。我们的研究为潜在的表观遗传分子基础和与SE转分化过程中的胚胎发生能力获得相关的关键甲基化模式提供了新的见识，从而为其在顽plant植物中的发展提供了广阔的前景。