Expression profiles of somatic embryogenesis--specific genes as a signature of cell fate
This paper reports on one of the first global analyses of SE gene expression on 12 key developmental stages covering the whole regeneration process from leaf explant dedifferentiation to embryo formation. Two pre-requisites were crucial for this study: (i) the availability of large-scale protocols, offering high reliability and cell synchronicity in each developmental stage as well as good biological efficiency, (ii) the availability of recent omics technologies that allowed to rapidly provide masses of information. This global analysis of coffee SE could serve as a reference for a wide range of plant species because the intensive sampling of successive key developmental stages that are conserved among species provided an overview of the SE process and enabled us to open some real black boxes [24]. The statistical approach we used to analyze transcriptomic data allowed us to cluster the whole SE process in seven main developmental phases and six key developmental phase switches that are the basis of SE. Many authors recently reported the need for a better understanding of the SE process to remove existing bottlenecks [8, 22, 24, 26]. Many developmental stages are undistinguishable when conventional morphological and histological approaches (e.g. early stages of dedifferentiation and redifferentiation) are used. A number of studies in different species assumed that clear correlations exist among the different transcriptome profiles and certain SE stages [28, 29, 31]. Our global analysis is a proof of concept that transcripts are good markers of all cell fate transitions and, in the near future, could be used to understand and better pilot the optimization of the SE culture conditions by using them as a milestone of successful developmental stages. This goes beyond morphological and histological descriptions, which until now, were the most common way to support empirical protocol optimization. We believe that this global scale transcriptome study, combined with a metabolic approach, will lead to a much clearer understanding of the molecular mechanisms underlying cell reprogramming.
Genes encoding regulatory, metabolic, hormonal and stress-related pathways are the most differentially expressed during coffee somatic embryogenesis
A number of genes were strongly up or downregulated during the six developmental phase switches identified. The transition of a leaf cell to an embryogenic cell is a long process of cell division and organization that occurs in the dark, hence upregulated genes related to mitosis and downregulated genes related to photosynthesis were expected [63]. SE is driven by exogenously supplied plant growth regulators [64]. Although most plants require similar physical conditions (temperature, light regime) for the induction of SE, only a specific composition of the medium can trigger and subsequently support the process. According to Sugimoto et al. [65], the prime characteristic of plant regeneration is cell fate reprogramming induced by wounding, stress, and hormones, in agreement with our results. Auxins and cytokinins are widely known to play essential roles in the induction of embryogenic culture [66, 22, 55, 56]. Additionally, many authors have reported the involvement of certain transcription factors (TFs) in the induction of somatic embryos in different species. However, the time points at which the genes encoding these TFs are highly active were previously unknown. In this study, we showed the kinetics of these genes during the successive SE steps for the first time. It has been proposed that together with auxin and cytokinin, TFs play a crucial role in the maintenance of the stem cell niche in the shoot apical meristem in totipotent cells in Arabidopsis [67] and in the cell pattern specification during the transition from totipotent-to-embryonic cell at the onset of redifferentiation [68]. Our results confirmed that an exogenous supply of auxin and cytokinin enabled enrichment in the transcripts of genes related to meristem development in coffee embryogenic cells, while enrichment in transcripts of genes related to embryonic cell pattern specification and embryonic formation was evidenced at the onset of redifferentiation. Concerning metabolic pathways, we previously showed that carbohydrates, starch, amino acids and secondary metabolites are differentially accumulated during the different SE steps [56]. Here, we confirmed that DEGs involved in metabolic pathways are tightly linked to the accumulation of their respective metabolites.
Genes regulating cell fate are highly modulated by environmental drivers during somatic embryogenesis
This study provides solid proof that environmental drivers are the main regulators of cell fate as they are tightly linked to the regulation of genes involved in crucial biological processes. This study allowed us to measure the direct effects of the environmental drivers usually used to control SE, particularly light, growth regulators, and cell density [8, 12] as these drivers were clearly perceived by the cells at molecular level. Environmental drivers, conventionally named ‘culture conditions’, are usually optimized in an empirical way to guarantee the appropriate nutritional and physico-chemical environment for a particular genotype for SE induction [22, 24]. Since these drivers can be perceived at a molecular level, the genes modulated by the drivers are of huge interest since they can be used to pilot SE optimization in a rational way. This is the case for genes involved in photosynthesis and circadian rhythm that are tightly linked to light intensity and photoperiod, genes involved in the response to auxin and cytokinin stimulus, and genes involved in embryo pattern specification that are tightly linked to cell density. Many authors have focused on the complex gene networks involved in the response to growth regulators, mainly auxin and cytokinin, to understand the expression of cell totipotency in Arabidopsis and in cotton [8, 68, 69]. The genes identified in these model plants were also found in coffee and showed similar patterns during the developmental stages we studied, indicating conserved pathways of cell totipotency between species. We believe that coffee SE could be used as a reference system to better understand fundamental mechanisms behind the response of woody plants to growth regulators.
Hormone-related genes play a major role in the expression of totipotency
SE is driven by exogenously supplied plant growth regulators. At the transcriptomic level, the upregulation of a total of 13 AUX/IAA genes, which are repressor genes [68] as well as efflux carrier genes (PIN), showed that the deprogramming process leading to the formation of undifferentiated cells in C. arabica was highly dependent on exogenously supplied auxin. Endogenous IAA concentrations are known to be tightly linked to expression of the YUCCA gene in Arabidopsis [70]. In coffee, once auxin was removed from the medium to allow cell redifferentiation, all AUX/IAA genes were switched off and YUCCA genes were switched on, enabling synthesis of endogenous IAA. Our results confirmed the concomitant increase in YUCCA gene expression and in endogenous IAA during redifferentiation.
ARF5 appeared to be only upregulated in embryogenic cells and can now be considered as a potential molecular marker of this developmental stage in coffee. Wójcikowska et al. [71] showed that, in Arabidopsis, ARF5 is highly expressed in embryogenic cultures and regulates the expression of numerous genes involved in somatic embryo formation including LEC2 (LEAFY COTYLEDON2), which is an activator of the YUC1, YUC4, YUC10 (YUCCA) genes involved in auxin biosynthesis during SE. ARF5 has also been reported to be involved in the cytokinin response pathway in Arabidopsis, tightly linked to ARR10 during the patterning and cell organization of meristem cells [71]. We also confirmed upregulation of ARR10 in coffee embryogenic cells that occurred at the same time as upregulation of the ARF5 gene. This shows that SE pathways are highly conserved between species and that the upregulation of ARR10 can be considered as a molecular candidate of embryogenic potential (Table 2).
ABI3 has been reported to play a major role in the regulation of SE induction in many species [22, 35, 55], while ABI5 has been reported to inhibit seed germination and promote embryo maturation in conifers [72, 73]. This shows once again that SE pathways are highly conserved between species and that activation of ABI3 can be considered as a molecular candidate of embryogenic cells, while the activation of ABI5 is a potential marker of the embryo maturation process (Table 2).
Identifying molecular candidates of embryogenic capacity
The formation and proliferation of embryogenic cells are the most crucial stages for the success of the SE process in all plant species because the efficiency of redifferentiation (i.e. mass regeneration of embryos) depends directly on their abundance. Embryogenic cell formation is a real bottleneck in the SE process for all plant species including coffee. It requires improvement of culture conditions to reduce the long time required (7 months for coffee). Consequently, many authors have focused their research on comparing embryogenic and non-embryogenic callus on a morphological [1] or molecular level [8, 32]. Since this question is of interest to many researchers, we decided to add the transcriptomic comparison between embryogenic and non-embryogenic callus to our study. A huge majority of the obtained DEGs (355) were downregulated in embryogenic cells compared to non-embryogenic ones. This is in agreement with Yang et al. [8], who showed in cotton that the existing developmental information of somatic cells must be switched off, most probably by an epigenetic regulation, in order to express the embryogenic capacity. The identified DEGs could serve as predictors of regenerative capacity, i.e. used to rapidly select or eliminate cell lines based on their presence/absence. Since coffee embryogenic callus is a compact, rapidly proliferating structure, its constitutive embryogenic cells show functional mitotic activity, upregulation of genes encoding organelle and DNA organization, morphogenesis, cell cycle and division. In addition, upregulation of genes related to wounding (WIND genes) [74] in embryogenic cells suggests that they result from a controlled stress-related pathway, while non-embryogenic cells result from an uncontrolled stress-related pathway implicating strong upregulation of genes involved in stress and oxidation processes [75]. Histological studies demonstrated that non-embryogenic callus is a spongy and oxidated callus containing numerous vacuolated and degenerating cells [56]. Non-embryogenic cells were characterized by upregulation of genes coded to respond to metal ion, oxidation reduction, and phosphate ion transport, in agreement with a number of studies on conifers [13, 32] suggesting that the main fate of non-embryogenic cells was survival, while embryogenic cells were mainly a transient state before the cell fate transition. This implies that the markers of embryogenic state in woody plants are conserved between species. We suggest that reducing oxidative stress by improving gaseous O2/CO2 exchange and reducing ethylene would increase embryogenic capacity.
In our study, transcripts of genes encoding amino acids were also more abundant in non-embryogenic cells, suggesting that the embryogenic cell genes were involved in the synthesis of more complex structures (proteins, DNA) as reported in other species [29, 32].
Secondary metabolism-related genes are switched off during dedifferentiation and switched back on at the onset of redifferentiation
Somatic cells in the plant contain all the genetic information needed to create a new complete functional plant [8]. During cell dedifferentiation, the existing developmental information of somatic cells must be switched off or reconfigured to make the somatic cells ready for an embryogenic program [76]. Our results clearly showed that genes encoding phenolic compounds and alkaloids were sharply downregulated during this stage and completely switched off in embryogenic cells. This is in agreement with the huge re-configurations observed in cell metabolic pathways during dedifferentiation [56]. Nic-Can et al. [77] provided evidence that these compounds inhibited the embryogenic process by affecting DNA methylation in C. canephora. Magnani et al. [78] also reported that biochemical pathways in Arabidopsis were shut off in order to activate the transcriptional machinery. Conversely, our results showed upregulation of the secondary metabolism-related genes in the early days of redifferentiation followed by the resumption of phenolic compound synthesis, mainly chlorogenic acids, as they are key intermediaries for cell wall biogenesis [56]. Furthermore, these antioxidant compounds could intervene as protectants since embryo formation has been widely reported to be a stress-related phenomenon [76, 79]. Therefore, genes involved in chlorogenic acid synthesis, such as HQT or lignin synthesis, such as CAD [80], can be considered as potential molecular markers of the redifferentiation pattern of coffee SE. Transcripts of genes encoding precursors of caffeine (XMT1) [81] also accumulated during redifferentiation, probably meaning that caffeine is produced in later embryo developmental stages.
Molecular candidates of cell fate to pilot optimization of somatic embryogenesis
We identified a set of potential molecular markers of cell fate transition during coffee SE (summarized in Table 2). A molecular marker is first defined by a clear expression pattern, i.e. a gene is switched on (or sharply upregulated) then switched off (or sharply downregulated) at one or several developmental stages. Secondly, a molecular marker is chosen based on its gene expression level, i.e. a gene with a high expression level is preferred to a gene with a low expression level. Finally, a molecular marker can be validated by comparing it between non-optimal conditions or developmental stages. To be able to undertake detailed sampling of all developmental stages of the coffee SE process, we limited this study to only one genotype. Further RT-qPCR analyses of a set of genotypes that are more or less recalcitrant to the induction of somatic embryos, will be crucial to study the candidate molecular markers. Molecular markers can be used to efficiently pilot the SE process optimization more rapidly, reliably, and more cost-effectively by testing a number of contrasted culture conditions in order to select the optimal ones, particularly in the case of recalcitrant genotypes or species.