Transcriptome Dynamics of Epidermal Reprogramming during Direct Shoot Regeneration in Torenia fournieri.

Shoot regeneration involves reprogramming of somatic cells and de novo organization of shoot apical meristems (SAMs). In the best-studied model system of shoot regeneration using Arabidopsis, regeneration is mediated by the auxin-responsive pluripotent callus formation from pericycle or pericycle-like tissues according to the lateral root development pathway. In contrast, shoot regeneration can be induced directly from fully differentiated epidermal cells of stem explants of Torenia fournieri (Torenia), without intervening the callus mass formation in culture with cytokinin; yet, its molecular mechanisms remain unaddressed. Here, we characterized this direct shoot regeneration by cytological observation and transcriptome analyses. The results showed that the gene expression profile rapidly changes upon culture to acquire a mixed signature of multiple organs/tissues, possibly associated with epidermal reprogramming. Comparison of transcriptomes between three different callus-inducing cultures (callus induction by auxin, callus induction by wounding and protoplast culture) of Arabidopsis and the Torenia stem culture identified genes upregulated in all the four culture systems as candidates of common factors of cell reprogramming. These initial changes proceeded independently of cytokinin, followed by cytokinin-dependent, transcriptional activations of nucleolar development and cell cycle. Later, SAM regulatory genes became highly expressed, leading to SAM organization in the foci of proliferating cells in the epidermal layer. Our findings revealed three distinct phases with different transcriptomic and regulatory features during direct shoot regeneration from the epidermis in Torenia, which provides a basis for further investigation of shoot regeneration in this unique culture system.

Introduction

Plant development is often featured by its high plasticity in contrast to the limited plasticity of animal development. The plastic nature of plant development can be seen in regeneration phenomena such as organ regeneration and somatic embryogenesis, through which many plants are able to recreate most parts of or even the entire plant body. The regeneration processes generally involve some kind of cell reprogramming and de novo organization of meristems that contain stem cells. For a representative example, shoot regeneration from mature tissues relies on reprogramming from the original, differentiated state and the subsequent generation of the shoot apical meristem (SAM) of an adventitious bud.

Since Skoog and Miller (1957) discovered that callus, shoots and roots can be artificially induced and manipulated by the application of the phytohormones auxin and cytokinin in tissue culture, tissue culture has become one of the main tools to study plant organogenesis including shoot regeneration. For the efficient induction of shoot regeneration, two-step culture systems, consisting of callus induction by auxin-rich culture in the first step and adventitious shoot induction by cytokinin-rich culture in the second step, have been developed for various plant species and widely used (e.g. Nishi et al. 1968, Christianson and Warnick 1983, Koornneef et al. 1987, Coleman and Ernst 1990).

A similar two-step culture system was also established in the model plant Arabidopsis thaliana (Arabidopsis) (Valvekens et al. 1988, Akama et al. 1992), which has facilitated the molecular biological analysis of shoot regeneration. In this culture system, if skipping the first step for callus induction, explants can form no or few adventitious buds, which implies that explant cells undergo reprogramming to acquire competence for shoot regeneration in the first step culture.

With the two-step culture system of Arabidopsis, transcriptome analysis was performed for the gene expression profiling of callus formation and adventitious SAM formation (Che et al. 2002, 2006, Xu et al. 2012), and spatial and temporal expression patterns of genes encoding major SAM regulatory transcription factors and phytohormone signaling factors were investigated in the process leading to SAM formation (Gordon et al. 2007, Cheng et al. 2013). The two-step shoot induction culture of Arabidopsis was used for the forward genetics of shoot regeneration as well (Yasutani et al. 1994, Tamaki et al. 2009). Recently, studies of the two-step shoot regeneration have been expanded, incorporating various new lines of research such as functional analysis of epigenetic regulation (He et al. 2012, Lee and Seo 2018, Ishihara et al. 2019) and genome-wide association analysis of natural variations (Lardon et al. 2020), which has accumulated increasing pieces of information.

One of the most important outcomes of research concerning the two-step culture of Arabidopsis over the last decade is the understanding that the callus formed in the first step is not a fully undifferentiated cell mass but a disorganized root meristem–like tissue originating from pericycle or pericycle-like tissues via the pathway of lateral root formation (Atta et al. 2009, Sugimoto et al. 2010). Furthermore, it was also shown that transcription factors regulating lateral root formation, such as LATERAL ORGAN BOUNDARIES DOMAINs (LBDs) and PLETHORAs (PLTs), participate not only in the callus formation but also in the acquisition of shoot regeneration competence (Fan et al. 2012, Kareem et al. 2015). Moreover, it was reported that lateral root primordia at certain developmental stages can be converted directly into adventitious shoot buds by the exposure to high concentrations of cytokinin (Atta et al. 2009, Rosspopoff et al. 2017). These findings suggest that auxin-induced callus shares many features with a lateral root primordium and that the root development pathway offers a mechanism of cell reprogramming during this type of callus formation. It is noted here that this reprogramming starts not from fully differentiated cells but from pericycle (or pericycle-like tissue) cells that are generally considered to remain partially meristematic (De Smet et al. 2006, Atta et al. 2009).

Arabidopsis plants can occasionally form a regenerative callus at wounded sites without hormone application. The molecular basis of this wound-induced callus formation has been also studied well for cell reprogramming, resulting in the identification of several key regulatory factors including the AP2/ERF transcription factor WOUND-INDUCED DEDIFFERENTIATION 1 (WIND1) (Iwase et al. 2011, 2017). Another good material for studying cell reprogramming is mesophyll protoplasts, which are reactivated from the quiescent state to enter the cell cycle and form regenerative callus in culture (Zelcer and Galun 1976). Transcriptome analysis of these processes revealed dynamic transcriptional changes possibly associated with cell reprogramming during wound-induced callus formation and protoplast culture (Chupeau et al. 2013, Ikeuchi et al. 2017).

In other plants, there are more diverse paths of shoot regeneration. In some cases, a preparatory callus formation step is not necessary for shoot regeneration (Hicks 1980), and the origin of regenerated shoots is not restricted to pericycle or pericycle-like cells (Chlyah and Van 1975, Creemers-Molenaar et al. 1994). In 1973, Chlyah reported a notable example of such shoot regeneration with the tissue culture of Torenia fournieri (Torenia). In this culture, adventitious bud SAMs formed directly on the surface of stem segments without intervening callus mass growth phase (Chlyah 1973, 1974a, 1974b). Importantly, histological analyses demonstrated that these adventitious bud SAMs originated exclusively from epidermal cells (Chlyah 1974a). From the 1970s to 1990s, various physiological studies were carried out with the Torenia culture system, which revealed a promotive role of cytokinin in shoot regeneration (Kamada and Harada 1979, Tanimoto and Harada 1982, 1984) and suggested wound stress as another promoting factor (Takeuchi et al. 1985). However, this culture system has never been used for molecular biological studies of shoot regeneration.

The Torenia stem culture system has three distinct features of shoot regeneration in contrast to the Arabidopsis two-step culture system: first, the entire process of shoot regeneration is triggered simply by one-step culture; second, drastic reprogramming should occur during the transformation of fully differentiated epidermal cells into meristem cells; and third, the initial process of de novo organization of SAMs takes place in a two-dimensional field of the epidermis. Because of these features, the Torenia stem culture can serve as a unique alternative experimental system for studying regeneration, particularly advantageous for cell reprogramming and SAM organization. For this reason, in the present study, we chose the Torenia stem culture to reveal the hidden aspects of shoot regeneration, which are difficult to uncover only with one model system. We performed the cytological and transcriptomic characterization of shoot regeneration with the Torenia culture system. The results obtained depict global and temporal changes in the gene expression profile that are likely to associate with each elementary process of the shoot regeneration in the Torenia stem culture, which provides a basis for further investigation of the relevant molecular mechanisms. We also compared the transcriptome data of Torenia with those reported for Arabidopsis callus-inducing cultures to gain information of core reprogramming mechanisms common to various types of regeneration-related events and identified possible candidates for factors universally involved in the molecular network of cell reprogramming.

Results

Effects of culture conditions on adventitious bud formation

Based on previous reports (Chlyah 1974b, Tanimoto and Harada 1982) and our preliminary experiments, we set up the culture system of Torenia stem segments for effective induction of shoot regeneration, in which, the first internodes were excised from young plants at the age of 4 weeks (), half-sectioned longitudinally, cut into 1.5-mm-long segments () and cultured on half-strength Murashige and Skoog (MS) medium containing 1 mg/l N6-benzyladenine (BA) as cytokinin at 22°C under continuous light.

Fig. 1

Tissue culture system of Torenia for direct shoot regeneration from stem explants. (A) A 4-week-old plant of Torenia. Bar = 5 mm. (B) Preparation of stem segments. Bars = 1 mm. (C–E) Stem explants cultured for 14 d with 1 mg/l BA under continuous light (BA+, light), without BA under continuous light (BA-free, light) or with BA in the dark (BA+, dark). Arrowheads in the magnified image of (C) indicate adventitious buds. Scale bars = 1 mm. (F) Number of adventitious buds formed on the epidermal surface per explants. Error bars = SE (n = 50).

We first tested the effect of cytokinin and light on the frequency of adventitious bud formation. While the culture of stem explants in the presence of BA under continuous light induced as many as 10 adventitious buds on the epidermis of each explant on average, the culture without BA in the light and the culture with BA in the dark induced much less intensive bud formation (). Particularly in the culture without BA, very few buds were formed (). These results showed that both exogenously supplied cytokinin and light are important, with the former being more critical, for adventitious bud induction in this culture. We employed culture with BA in the light as the standard bud-inductive culture and culture without BA in the light and culture with BA in the dark as controls in the subsequent experiments.

Early cytological events and cell division activation

In an early report on the Torenia stem culture, it is mentioned that nuclei and nucleoli of the epidermal cells increased in the acetocarmine stainability and volume before the commencement of cell division, although without quantitative data (Chlyah 1974b). Similar changes in the nuclear and nucleolar appearance were reported in several types of dedifferentiating plant cells, implicating nuclear and nucleolar development in the preparation of cell division activation (Feldman and Torrey 1977, Williams and Jordan 1980, Paul et al. 1989, Williams et al. 2003). To verify Chlyah’s observation quantitatively and evaluate nuclear and nucleolar development in the Torenia stem culture, we stained epidermal cells of stem explants before and after culture for 2 d in three conditions with 4ʹ,6-diamidino-2-phenylindole (DAPI) and RNAselect to visualize nuclei and nucleoli, respectively, and measured their size. In the standard bud-inductive culture, both nuclei and nucleoli were enlarged markedly in the first 2 d of culture (). In the dark control culture, the nuclear size increased to the same extent as in the bud-inductive culture. Enlargement of nucleoli also occurred in the dark, although they were somewhat smaller than those in the bud-inductive culture (). In contrast to these cultures, the BA-free control culture induced only a limited increase in the size of nucleoli (). These observations confirmed nuclear and nucleolar development in epidermal cells ready to divide and revealed that it occurs mostly depending on cytokinin and under the relatively minor positive influence of light.

Fig. 2

Changes in the size of nuclei and nucleoli in the early stage of culture as influenced by BA and light. (A) DAPI-stained nuclei and RNAselect-stained nucleoli in epidermal cells of stem explants immediately after excision (day 0) and cultured for 2 d in the standard bud-inductive condition (day 2). Bars = 5 µm. (B) Boxplots of nuclear and nucleolar sizes of epidermal cells of stem explants immediately after excision (day 0) and cultured for 2 d (day 2) under the standard bud-inductive, BA-free control or dark control conditions. n = 37–45. Asterisks indicate statistically significant differences by Mann–Whitney–Wilcoxon test with Benjamini–Hochberg correction (*P < 0.05; **P < 0.01; ****P < 0.0001; ns, not significant).

To investigate the activation of cell cycle progression in cultured stem explants, nuclear DNA content and cell division frequency were examined. The nuclear DNA content was assessed by flow cytometry for stem explants before and after culture for various times in the standard bud-inductive condition (). Before culture, there were a dominant 2C peak and a very small 4C peak in the histogram of nuclear DNA content, and after culture, the 4C peak increased within 2–3 d. These results indicated that most cells of the stem of Torenia do not show endoreduplication and are in the G0/G1 phase of the cell cycle and that cell cycle progression from the G1 to S phase is activated in the bud-inductive culture.

Fig. 3

Cell division activation during culture as influenced by BA and light. (A) Changes in the distribution of the flow cytometric measured nuclear DNA content in stem explants during culture under the standard bud-inductive condition. (B) Changes in the number of anillin blue–stained nascent cell walls in the epidermis of stem explants during culture under the standard bud-inductive condition. Error bars indicate standard errors (n = 7). (C) Comparison of the numbers of anillin blue–stained nascent cell walls in the epidermis of stem explants between standard bud-inductive, BA-free control and dark control conditions. Error bars indicate standard errors (n = 6–7). (D–F) CLSM observation of the PI-stained epidermis of the stem explants at the beginning of culture (D) and after 4 d (E) and 6 d (F) of culture in the standard bud-inductive condition. Scale bars = 100 µm.

Nascent cell walls in the epidermis of cultured stem explants were detected by aniline blue staining and counted as an index of cell division frequency. In the standard bud-inductive culture, the number of nascent cell walls increased after 2–4 d of culture, indicating that epidermal cells are activated to resume cell division in this period (). In the control cultures, however, cell division in the epidermis was much less active, and particularly in the BA-free culture, it was substantially limited (). These results indicated that the activation of cell division requires cytokinin application and is promoted by light.

Cell division patterns leading to adventitious bud formation

The stem explants cultured in the standard bud-inductive condition were inspected for cell division patterns in the epidermis. Confocal laser scanning microscopy (CLSM) after staining with propidium iodide (PI) clearly detected the appearance and development of foci of small cells, indicating local activation of cell division in the epidermal layer (). Serial observation of the surface of cultured stem explants with a metallographic microscope showed that locally activated cell division generated the foci of proliferating cells, from which adventitious bud SAMs eventually developed (). In all cases observed, each SAM arose from a subpopulation of dividing cells originated from more than one epidermal cells, and borderlines of these ‘presumptive SAM areas’ were inconsistent with those of original epidermal cells ().

Fig. 4

Serial observation of cell division and adventitious bud formation in the epidermis of cultured stem explants. (A) Epidermis of stem explants immediately after excision. Bar = 100 µm. (B–G) Serial observation of the epidermis of stem explants forming adventitious buds in the standard bud-inductive culture. Two adventitious buds are indicated by arrowheads. Bars = 100 µm. (H) Areas developed into adventitious SAMs (presumptive SAM areas, red dashed lines) and the boundaries of original epidermal cells (yellow lines) in the course of shoot regeneration shown in (B–G). The boundaries of presumptive SAM areas and original epidermal cells are shown on the image of day 4 in which both boundaries are clear. Bar = 100 µm. (I–M) The boundaries of presumptive SAM areas (red dashed lines) and original epidermal cells (yellow lines) in another five cases of shoot regeneration observed serially. In each case, a picture that presumptive SAM areas and the boundaries of original epidermal cells are clear is chosen from the serial pictures. Bars = 100 µm.

Global changes in gene expression during the entire course of shoot regeneration

Transcriptome data were obtained by RNA sequencing (RNA-seq) for stem explants cultured for 0, 1, 2, 4 or 8 d in the standard bud-inductive condition and stem explants cultured for 4 or 8 d in the BA-free and dark control conditions. Principal component analysis (PCA) was performed on the normalized transcription data (). As a result, the first three components explained 79.2% of the total variance among all samples. Biological replicates of each sample clustered together, confirming the reproducibility of the analysis. Among 27,137 genes of which transcripts were detected, 23,697 genes showed significant temporal changes (false discovery rate (FDR) < 0.05) in their expression level during culture under the standard bud-inductive condition. These genes were subjected to k-means clustering analysis and classified into eight clusters according to their changing patterns. Then, the Gene Ontology (GO) term enrichment analysis was performed on each of the clusters using the Blast2GO software.

The gene clusters were characterized by enrichment of different GO terms (). With respect to possible relationships with the cellular and morphogenetic events observed during shoot regeneration, notable GO enrichment was found in clusters 2, 4 and 5. In cluster 2, where expression levels increased in the first day and then decreased slightly, GO terms related to protein synthesis such as ‘translation’ and ‘ribosome biogenesis’ were very highly enriched. The high enrichment of these terms well corresponded to the nucleolar enlargement in the first 2 d. Genes in cluster 4 showed a gradual increase in the expression levels during the first 4 d and maintained a high expression throughout the later stage. This cluster was represented by GO terms related to the cell cycle. This seemed to reflect the cell division activation from 2 to 4 d of culture. In cluster 5, where expression levels started to increase after 2 d of culture and continued to increase thereafter, GO terms related to developmental processes were enriched. More specific GO terms such as ‘plant organ development’, ‘tissue development’, ‘shoot system development’ and ‘meristem development’ were also enriched in this cluster. Indeed, this cluster contained many SAM regulatory gene orthologs that are likely to participate in the formation of adventitious SAMs.

Fig. 5

Clustering of gene expression patterns during shoot regeneration in stem explants. Eight clusters of genes showing different expression patterns during the culture of stem explants in the bud-inductive condition. Expression levels of each individual gene are expressed as Z scores that were calculated from TPM values such that their mean and standard deviation over time were equal to 0 and 1, respectively. Magenta and gray lines indicate the average expression of all genes in each cluster and expressions of individual genes, respectively. Top eight GO terms enriched in each cluster are presented with a color scheme representing log10(P-value) as well as the number and fold enrichment of the relevant genes in it. BP, CC and MF indicate GO categories ‘biological process’, ‘cellular component’ and ‘molecular function’, respectively.

To look at the overall trends of how temporal changes of gene expressions are regulated by culture conditions, we also picked up genes that were significantly upregulated or downregulated after 4 and 8 d of culture in each of two control conditions as well as the bud-inductive condition (), classified them into the above clusters and examined how they were overlapped between culture conditions (). Throughout the clusters, genes significantly changed in the bud-inductive culture were overlapped more with those in the dark control than with those in the BA-free control. From the comparison among clusters, the ratio of overlap between the bud-inductive culture and the BA-free control was found to be lowest in cluster 1 and second lowest in cluster 5. In these clusters, gene expressions show a transient increase or decrease followed by a gradual reverse change, and the reverse change phase seemed to be suppressed in the BA-free control. These results showed that cytokinin has a dominant impact on the regulation of gene expressions at relatively late stages of culture while light has a minor effect.

Temporal expression patterns of SAM regulatory genes

We identified Torenia orthologs of the major SAM regulatory genes of Arabidopsis encoding transcription factors: three SHOOT MERSITEMLESS (STM) orthologs named TfSTM1, TfSTM2 and TfSTM3; two orthologs of CUP-SHAPED1 (CUC1) and CUC2 named TfCUC1/2a and TfCUC1/2b and two WUSCHEL (WUS) orthologs named TfWUS1 and TfWUS2 (). Expression levels of these genes in cultured stem explants and various parts of young plants were examined by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) analysis.

In young plants, all of the three TfSTM genes were expressed at a high level at the shoot apical region and expressed more or less in the stem as well (). Two TfCUC1/2 genes and two TfWUS genes also showed a high expression at the shoot apical region, but unlike TfSTMs, they were not or only weakly expressed in the stem ().

In stem explants cultured under the standard bud-inductive condition, where visible adventitious bud formation occurred after 8 d (), the expression level of TfSTM1 increased after 4 d of culture, while the expression of TfSTM2 showed a remarkable decrease within the first 2 d followed by a slight increase after 6 d (). TfSTM3 expression increased within 2 d and remained at a relatively high level in the bud-inductive culture. Expressions of the TfCUC1/2 and TfWUS genes increased dramatically after 4–6 d of culture in the bud-inductive condition. Of note, the expression levels of TfWUSs at day 8 were more than several times higher than those at the shoot apical region of young plants. The expression increases of TfSTM1, TfSTM2, TfCUC1/2s and TfWUS2 were smaller in the dark control and not clearly observed in the BA-free control in a good correlation with the frequency of adventitious bud formation (). At day 8 of the bud-inductive culture, all of the SAM regulator genes tested were expressed in both inner tissues and epidermal layers with CUC1/2s and WUSs showing a higher expression in the epidermal layers than in the inner tissues (), which might be associated with SAM development from the epidermis.

Fig. 6

Expression of SAM regulatory genes during culture of stem explants. Expression of TfSTMs, TfCUC1/2s and TfWUSs in cultured stem explants was quantified by RT-qPCR. The left panels show expression levels in culture under standard and control conditions (green, standard bud-inductive condition; orange, BA-free control; purple, dark control), and the right panels show expression levels in the epidermal layer (light green) and inner tissue (gray) of explants cultured in the standard condition. Error bars indicate standard errors (n = 3). Relative expression levels were calculated with reference to the expression in the shoot apical region of young plants.

Effect of cytokinin on gene expression in the early stage of culture

As our cytological observation detected clear differences in the nuclear and nucleolar enlargement between the standard bud-inductive culture and the BA-free control culture as early as day 2, we performed the second set of RNA-seq transcriptome analysis of stem explants focusing on cytokinin effects during the first 2 d of culture. Stem explants cultured for 0, 3, 6, 12, 24 or 48 hours in the standard bud-inductive condition and in the BA-free control condition were subjected to analysis together with the root, leaf and shoot apex of young plants. As a result of PCA of the transcriptome data, the first three components explained 81.3% of the total variance among all samples (). According to these components, uncultured stem explants are quite distant from stem explant cultured for 3 hours or longer, indicating that there was a drastic transcriptional change in the first 3 hours of culture.

Genes differentially expressed between the standard bud-inductive condition and the BA-free condition were limited to only a few in the initial 12 hours of culture and increased drastically in the subsequent 36 hours (; ). This result indicated that exogenous cytokinin plays a substantial regulatory role in gene expression only after 12 hours of culture, agreeing with the abovementioned trend seen in .

Table 1

Numbers of genes differentially expressed between the standard bud-inductive and BA-free control cultures

	3 h	6 h	12 h	24 h	48 h
Higher expression in standard bud-inductive culture	2	0	17	275	814
Lower expression in standard bud-inductive culture	0	0	3	362	1126

Fig. 7

Effect of cytokinin on the gene expression profile in the early stage of culture. (A–B) Expression patterns of genes that showed a significant difference in the expression level between the standard bud-inductive and BA-free control cultures at least at one timepoint. The heat maps show mean log2 values of the fold changes of expressions during culture in the standard bud-inductive and BA-free control conditions relative to the expression before culture (A) and mean log2 values of the fold changes of expression in the standard culture to that in BA-free culture at each time (B).

As compared to the BA-free control culture, 868 genes and 1181 genes showed higher and lower expression levels in the standard bud-inductive culture at least at one timepoint, respectively. These two groups of genes were subjected to GO term enrichment analysis. In the lower expression group, 554 GO terms were significantly enriched (FDR < 0.05), of which top positions were largely occupied by the terms belonging to ‘response to stimuli’ including stress response–related terms (). This result implicated exogenous cytokinin in the suppression of responses to stress and/or other kinds of stimuli at the initial stage of culture.

Table 2

Top 10 terms in three GO categories (biological process, cellular component and molecular function) enriched in genes expressed at a lower level in the standard bud-inductive culture than in the BA-free control culture

Biological process
GO term	P-value	Number of genes	Fold enrichment
Response to stimulus	1.8E-41	676	1.455
Response to external stimulus	3.9E-38	345	1.918
Response to oxygen-containing compound	9.7E-34	362	1.786
Response to chemical	2.9E-32	468	1.580
Response to stress	6.0E-30	469	1.545
Response to organic substance	5.6E-28	373	1.652
Response to biotic stimulus	1.6E-27	268	1.887
Response to external biotic stimulus	2.6E-27	267	1.884
Response to other organisms	2.5E-27	267	1.884
Interspecies interaction between organisms	3.5E-27	272	1.865
Response to stimulus	1.8E-41	676	1.455
Cellular component
Cell periphery	1.2E-22	397	1.517
Plasma membrane	8.9E-19	331	1.528
Cellular anatomical entity	5.2E-18	968	1.115
Membrane	1.6E-15	608	1.241
Anchoring junction	1.6E-08	183	1.455
Cell–cell junction	1.6E-08	183	1.455
Cell junction	2.0E-08	183	1.450
Extracellular region	2.0E-08	117	1.642
Plasmodesma	2.2E-08	182	1.450
Symplast	2.2E-08	182	1.450
Molecular function
Caffeoyl-CoA O-methyltransferase activity	2.7E-10	8	21.667
Caffeoyl CoA:S-adenosyl-L-methionine O-methyltransferase activity	2.7E-10	8	21.667
Irihydroxyferuloyl spermidine:S-adenosyl-L-methionine O-methyltransferase activity	2.7E-10	8	21.667
Tricaffeoyl spermidine:S-adenosyl-L-methionine O-methyltransferase activity	2.7E-10	8	21.667
Trihydroxyferuloyl spermidine O-methyltransferase activity	2.7E-10	8	21.667
DNA-binding transcription factor activity	3.9E-09	120	1.677
Molecular function regulator	9.6E-09	158	1.522
Molecular transducer activity	2.0E-07	36	2.538
Protein binding	2.6E-07	492	1.165
Oxidoreductase activity, acting on peroxide as acceptor	2.9E-07	21	3.592

In the higher expression group, 603 GO terms were found to be enriched, which contained many nucleolus, ribosome biogenesis or protein synthesis-related terms and their upper-order category terms (). From this result, together with the observation of nucleolar enlargement, we can reasonably speculate that nucleolar development, the increase in its activity of synthesizing ribosomes and the consequent acceleration of protein synthesis occur dependently on cytokinin in the bud-inductive culture.

Table 3

Top 10 terms in three GO categories (biological process, cellular component and molecular function) enriched in genes expressed at a higher level in the standard bud-inductive culture than in the BA-free control culture

Biological process
GO term	P-value	Number of genes	Fold enrichment
Gene expression	1.5E-36	310	1.948
Organic substance biosynthetic process	9.1E-34	385	1.694
Biosynthetic process	3.0E-33	390	1.675
Cellular biosynthetic process	6.2E-33	375	1.701
Cellular macromolecule biosynthetic process	7.0E-33	291	1.924
Macromolecule biosynthetic process	1.0E-32	295	1.907
Cellular process	3.2E-31	702	1.256
Cellular nitrogen compound biosynthetic process	1.2E-30	291	1.869
Ribonucleoprotein complex biogenesis	1.7E-29	92	3.888
Ribosome biogenesis	1.5E-26	82	3.914
Gene expression	1.5E-36	310	1.948
Cellular component
Non-membrane-bounded organelle	2.6E-42	233	2.488
Intracellular non-membrane-bounded organelle	2.6E-42	233	2.488
Cellular anatomical entity	1.8E-36	771	1.219
Nucleolus	9.2E-36	128	3.452
Nucleus	2.9E-32	399	1.641
Protein-containing complex	3.8E-32	306	1.859
Nuclear lumen	3.6E-32	161	2.691
Intracellular	2.4E-31	675	1.282
Intracellular organelle lumen	6.2E-31	169	2.544
Membrane-enclosed lumen	6.2E-31	169	2.544
Molecular function
Structural molecule activity	1.1E-24	77	3.874
Structural constituent of ribosome	3.2E-21	62	4.083
RNA binding	6.3E-21	124	2.435
mRNA binding	5.0E-16	70	2.933
Binding	8.3E-15	621	1.184
Protein binding	4.6E-14	407	1.322
Heterocyclic compound binding	3.3E-11	390	1.282
Organic cyclic compound binding	3.9E-11	390	1.281
DNA binding	4.6E-11	150	1.671
Nucleic acid binding	2.0E-10	270	1.383

Although not top-ranked, cell cycle–related GO terms such as ‘cell cycle’ (rank 21 in the category ‘biological process’, P-value = 2.34E-17, gene number = 86, fold enrichment = 2.958) and ‘cell cycle process’ (rank 77 in the category ‘biological process’, P-value = 1.36E-09, gene number = 58, fold enrichment = 2.628) were also enriched in the higher expression group, consistent with the much more active cell division in the bud-inductive culture than in the BA-free culture.

To further investigate the role of cytokinin in cell cycle progression, we compared expression levels of cell cycle phase–specific genes between the bud-inductive culture and BA-free culture. In this analysis, Torenia orthologs to the E2F target genes of Arabidopsis (Vandepoele et al. 2005) and to the Arabidospsis G2/M-specific genes (Haga et al. 2011) were used as putative indices of the G1-to-S transition phase and G2-to-M transition phase, respectively. In the standard bud-inductive condition, many of the E2F target genes showed elevation in the expression level after 24 hours of culture while the G2/M-specific genes became highly expressed after 48 hours of culture (). After 24 hours of culture, when cells were in the first round of cell cycle, expression levels of the E2F target genes were considerably higher in the bud-inductive culture than in the BA-free culture, suggesting a role of cytokinin in cell cycle progression from the G1 phase to the S phase (). Expression of the G2/M-specific genes was also higher in the bud-inductive culture ().

Fig. 8

Effect of cytokinin on the expression of cell cycle genes during culture. (A–B) Expression patterns of E2F target genes (A) and G2/M-specific genes (B) during culture of stem explants under the standard bud-inductive and BA-free control conditions. The heat map shows mean log2 values of the fold changes relative to the expression before culture.

Expression profile of genes characteristic of various plant parts in the early stage of culture

The next analysis was performed to characterize the transcriptome of cultured stem explants with reference to gene expression profiles of various plant parts. Global gene expression data of the stem, root, leaf and shoot apex of young plants were obtained by RNA-seq. A total of 17,453 genes that showed significant differences in the expression level among plant parts were subjected to k-means clustering analysis and classified into 30 clusters according to their expression patterns. As a result, clusters of genes that were preferentially expressed in one plant part were identified as genes characterizing each part (). For these genes, expression profiles in cultured stem explants were extracted from the transcriptome data of the early stage of culture (). In the first 3 hours of culture, the expression of genes characteristic of the stem declined while many genes characteristic of the root, leaf or shoot apex became actively expressed at the same time. These changes occurred in both the standard bud-inductive and BA-free control cultures. These results demonstrated that upon culture, regardless of the presence or absence of exogenous cytokinin, stem explants rapidly alter their gene expression profile to have signatures of multiple organs/tissues.

Fig. 9

Expression patterns of genes characteristic of various plant parts in the early stage of culture of stem explants. (A–D) Expression patterns of genes that showed preferential expression in the stem (A), root (B), leaf (C) or shoot apex (D) of young plants in the early stage of culture of stem explants under the standard bud-inductive and BA-free control conditions. The heat map shows mean log2 values of the fold changes relative to the expression before culture.

Comparison of transcriptome data between different culture systems of Arabidopsis and Torenia stem culture

It is of great interest how common or different the molecular networks are among various plant culture responses involving cell reprogramming. In an attempt to answer this problem, we conducted an comparative transcriptome analysis between the Torenia stem culture and three culture systems of Arabidopsis, callus induction by wounding, callus induction by auxin and mesophyll protoplast culture. First, we compared significantly upregulated genes in the cultures of Arabidopsis. Analysis of published transcriptome data (Xu et al. 2012, Ikeuchi et al. 2017) identified 4,494 genes significantly upregulated in the first 24 hours of callus induction by wounding and 351 genes significantly upregulated genes in the first 24 hours of callus induction by auxin. We also used 1,284 genes that were previously reported as being upregulated in the first 24 hours of mesophyll protoplast culture (Chupeau et al. 2013) in our analysis. Of these genes, 15 genes were found to be upregulated in all three culture systems, and 860 genes were upregulated in at least two culture systems ().

Fig. 10

Expression patterns of genes upregulated in callus induction culture systems of Arabidopsis in Torenia stem culture. (A) Venn diagram showing numbers of overlapping and non-overlapping genes among gene sets that were identified to be upregulated during callus induction by auxin, callus induction by wounding and protoplast culture of Arabidopsis. (B–H) Expression patterns in the early stage of culture of Torenia stem explants under the standard bud-inductive and BA-free control conditions for the Torenia orthologs to the Arabidopsis genes upregulated in one, two or all three of three different callus induction culture systems of Arabidopsis: callus induction by auxin, callus induction by wounding and protoplast culture. The heat maps show mean log2 values of the fold changes relative to the expression before culture. Parenthesized numbers indicate percentages of genes that were significantly upregulated at least at one timepoint in the Torenia stem culture.

Next, using our transcriptome data of the early stage of the Torenia stem culture, we examined expressions of Torenia orthologs to the Arabidopsis genes that were identified by the above analysis as being upregulated in at least one of the three culture systems (). The results clearly showed a trend that genes upregulated in common to more culture systems of Arabidopsis were also upregulated in the Torenia stem culture at a higher proportion. For nine genes out of 15 genes that were upregulated commonly in three culture systems of Arabidopsis, we found Torenia orthologs, most of which were upregulated in the Torenia stem culture under both the standard bud-inductive and BA-free control cultures (; ). The nine genes, including an ortholog of HISTONE DEACETYLASE 3 (HDA3), an epigenetic regulator gene, were expected to be involved in the initial culture response universally beyond variations in culture protocols and plant materials.

Table 4

List of the Torenia orthologs to the Arabidopsis genes commonly upregulated in all three different callus-inducing culture systems of Arabidopsis

AGI	Primary gene symbol	Torenia transcript	Significance
AT4G36670	POLYOL/MONOSACCHARIDE TRANSPORTER 6 (PMT6)	TfB096621	+
AT4G36670	POLYOL/MONOSACCHARIDE TRANSPORTER 6 (PMT6)	TfB101385	+
AT3G07390	AUXIN-INDUCED IN ROOT CULTURES 12 (AIR12)	TfA043399	+
AT1G21750	PDI-LIKE 1-1 (PDIL1-1)	TfA023511	+
AT2G15760		TfB102428	+
AT1G09970	LRR XI-23	TfB077885	+
AT1G09970	LRR XI-23	TfA029790	+
AT3G44750	HISTONE DEACETYLASE 3 (HDA3)	TfB090237	+
AT3G04120	GLYCERALDEHYDE-3-PHOSPHATE DEHYDROGENASE C SUBUNIT 1 (GAPC1)	TfB100536	+
AT3G04120	GLYCERALDEHYDE-3-PHOSPHATE DEHYDROGENASE C SUBUNIT 1 (GAPC1)	TfA002952	+
AT1G68560	ALPHA-XYLOSIDASE 1 (XYL1)	TfA011629	+
AT3G54960	PDI-LIKE 1-3 (PDIL1-3)	TfB087184	+
AT4G36670	POLYOL/MONOSACCHARIDE TRANSPORTER 6 (PMT6)	TfB089190	ns
AT3G07390	AUXIN-INDUCED IN ROOT CULTURES 12 (AIR12)	TfB097832	+
AT3G44750	HISTONE DEACETYLASE 3 (HDA3)	TfA014513	+
AT4G36670	POLYOL/MONOSACCHARIDE TRANSPORTER 6 (PMT6)	TfB077745	ns
AT3G44750	HISTONE DEACETYLASE 3 (HDA3)	TfB086922	ns

Genes are listed with Arabidopsis Genome Initiative (AGI) codes in the same order as in . The + sign denotes significant upregulation at least at one timepoint in the Torenia stem culture.

Discussion

In this study, we revisited the stem segment culture of Torenia, which was reported nearly half a century ago to successfully induce adventitious buds from the epidermis simply in one step (Chlyah 1973), and used its modified version as a unique experimental system for the investigation of direct shoot regeneration involving cell reprogramming from the fully differentiated state. As described in the previous papers (Tanimoto and Harada 1982, 1984), cytokinin application was confirmed to be critical in this culture, and light was also found to be important. With the 1 mg/l BA-supplemented culture in the light as a standard bud-inductive culture and the BA-free culture in the light and the BA-supplemented culture in the dark as controls, we performed the cytological analysis of nuclear and nucleolar development and cell cycle progression and several lines of gene-expression analysis in the course of shoot regeneration in Torenia stem explants. These analyses and further comparison with the transcriptomes of three different Arabidopsis culture systems yielded plenty of information suggestive of the processes of shoot regeneration.

Cytokinin-dependent nucleolar development precedes cell division activation

In stem explants cultured under the standard bud-inductive condition, cell division became active after 2–3 d, which was followed by adventitious bud formation after 8 d of culture (). Microscopic observation at day 2 revealed that, before the activation of cell division, nuclei and nucleoli were markedly enlarged in epidermal cells (). Similar changes in the size of nuclei and nucleoli have been reported for dedifferentiating cells provoked to divide in various plant cell/tissue cultures (Feldman and Torrey 1977, Williams and Jordan 1980, Paul et al. 1989, Williams et al. 2003). Taking these reports into consideration, the nuclear and nucleolar enlargement observed in the Torenia stem culture can be considered as a sign preceding cell division activation.

The nucleolus is a specialized subnuclear domain functioning as a site of ribosome biogenesis, and its development is generally linked with the stimulation of ribosome biogenesis and the resultant acceleration of protein synthesis. Indeed, consistent with nucleolar enlargement, transcriptome analysis with GO enrichment analysis indicated that expressions of genes related to ribosome biogenesis and protein synthesis were upregulated in the early stage of the culture of Torenia stem explants (). Upregulation of ribosome-biogenesis-related genes prior to cell proliferation was also reported from the transcriptomic characterization of Arabidopsis cultures for wound-induced callus formation (Ikeuchi et al. 2017) and mesophyll protoplast–derived callus formation (Chupeau et al. 2013). Additionally, in the mesophyll protoplast culture of tobacco, an increase in the rate of protein synthesis was observed by the [14C]Leu-feeding experiment before the start of cell division (Zelcer and Galun 1976). These previous findings together with our results suggest that nucleolar development accompanying stimulation of ribosome biogenesis and acceleration of protein synthesis is commonly involved in preparation for cell division activation in dedifferentiating plant cells.

All the above-described cytological events and gene expression changes were greatly reduced in the BA-free control culture compared to the standard bud-inductive culture. Without BA application, nuclei and nucleoli were only slightly enlarged (), the frequency of cell division was very low (), few adventitious buds were formed (), and nucleolus, ribosome biogenesis and protein synthesis–related genes were significantly less upregulated (). For a simple explanation of these results, we can hypothesize that cytokinin primarily regulates gene expression for nucleolar development accompanying stimulated ribosome biogenesis, which is a prerequisite for cell division activation and subsequent morphogenesis. In partial support of this hypothesis, previous transcriptomic studies demonstrated that ribosomal protein genes are upregulated downstream of cytokinin signaling (Brenner et al. 2005, Kiba et al. 2005).

In the dark control, cell division adventitious bud formation was reduced to about 30% and 20% of those in the standard bud-inductive culture (), while the reduction of nucleolar enlargement was not so large (). It is therefore likely that light is involved not only in the nucleolar development process but also in later processes.

Cell cycle restarts from G1 to S progression

As argued above, the role of cytokinin in cell division activation in the Torenia stem culture may be attributed to the cytokinin requirement of nucleolar development and stimulation of ribosome biogenesis in the preparation for cell division. Even if this is true, however, the possibility is not excluded that cytokinin more directly functions to activate cell division as well. Indeed, cytokinin is known to promote cell cycle progression in at least two pathways: one directs the G1-to-S transition via transcriptional upregulation of the D-type cyclin CYCD3 (Riou-Khamlichi et al. 1999), and the other directs the G2-to-M transition through the activation of A-type cyclin-dependent kinase (CDKA) (Zhang et al. 1996, Orchard et al. 2005).

Our transcriptome analysis of Torenia stem explants revealed that both the G1/S and G2/M genes showed a much higher expression in the bud-inductive culture than in the BA-free control (). As cells were mostly in the G0 or G1 phase at the beginning of culture (), this result suggests that the main point of regulation of cell cycle progression by cytokinin is at the G1-to-S phase transition in the present culture system, aside from whether it is direct or indirect.

De novo SAM organization occurs from cell division foci in association with SAM regulator gene expression

Previous anatomical studies of the Torenia stem culture showed that cell division was unevenly activated in the epidermis to form local areas of active cell proliferation and that adventitious bud SAMs formed from the centers of such areas (Chlyah 1974b, Chlyah et al. 1975). Additionally, it was also pointed out that the origin of the SAM was not always a single epidermal cell (Chlyah 1974b). In agreement with these reports, our serial observation indicated that adventitious SAMs arise in cell division foci derived from more than one epidermal cell (). Of note, the boundary of the presumptive SAM region, which later developed into the SAM, did not correspond to the borderlines of original epidermal cells. This finding suggests that the presumptive SAM regions are determined only after several rounds of cell division ().

There have been many papers reporting that shoot regeneration is closely associated with a high expression of SAM regulatory transcription factor genes such as STM, CUCs and WUS (Daimon et al. 2003, Zhang et al. 2017). Similarly, in Torenia stem explants cultured in the standard bud-inductive condition, the expression of most of the major SAM regulatory genes tested was found to increase greatly a few days before visible adventitious bud formation (). In particular, expression levels of TfCUC1/2s and TfWUS2 showed a good correlation with epidermal shoot regeneration in the following respects: lack of expression before culture, a reduced expression levels in parallel with the reduced frequencies in two control cultures and a higher expression in epidermal layers than in inner tissues. It can be inferred from these results that transcriptional activation of the SAM regulatory network is key to adventitious SAM morphogenesis.

Gene expression profile rapidly changes upon culture likely reflecting cell reprogramming

Transcriptome analysis of Torenia stem explants focusing on the early stage of culture revealed very rapid changes in the gene expression profile from the original state to a new state with multiple organ/tissue signatures characterized by simultaneous expression of genes that are normally expressed preferentially in the shoot apex, leaf or root. Previous studies with various organisms and various experimental systems have shown that pluripotent stem cells or cells undergoing reprogramming co-express genes that are typically associated with specific cell types (Hu et al. 1997, Buganim et al. 2012, Efroni et al. 2016, Mozgová et al. 2017, Omary et al. 2020). Recently, it has been proposed that such gene expression profile exhibiting a ‘mixed identity’ can be a common feature of the pluripotent state in either animals or plants (Moris et al. 2016, Efroni 2018). Our findings in the Torenia culture coincide with these arguments and may reflect that the explants gain pluripotency during cell reprogramming.

Comparison between genes upregulated during callus induction in three different culture systems of Arabidopsis and examination of the expression of their orthologs in the Torenia stem culture identified nine genes that were expressed in the early stage in all the four culture systems as potential candidates of genes that function in cell reprogramming commonly in any type of culture system. Interestingly, this gene list () included an ortholog of the histone deacetylase gene HDA3, which was reported to participate in the auxin-induced callus formation (Lee et al. 2016). Functional analysis of this and other genes in the list would give a clue for a common molecular network involved in cell reprogramming.

Our analysis indicated that cytokinin has little impact on gene expression in the first 12 hours of culture (). Accordingly, transcriptional changes in the initial phase of culture noted above must be triggered not by exogenous cytokinin but by some other stimuli during tissue culture operation. At the start of tissue culture, explants necessarily suffer wounding when excised from the mother plant, and wounding is well-known to induce a wide spectrum of cellular responses, including cell reprogramming and the resultant callus formation not only in Arabidopsis but also in various plant species (Ikeuchi et al. 2013). In Torenia stem segment culture, it was also reported that additional wounding promotes shoot regeneration (Takeuchi et al. 1985). Wounding is, therefore, one of the likely triggers of initial changes of gene expression in the Torenia culture. The wounding-triggered molecular cascade leading to cell reprogramming has been extensively studied in Arabidopsis (Ikeuchi et al. 2017, Iwase et al. 2017). However, there is also the possibility that other factors besides wounding trigger these changes. For example, explant isolation may affect the endogenous signaling molecule levels as a result of tissue disconnection, which can be a trigger of gene expression changes. Indeed, tissue disconnection by incision of the stem was shown to alter the distribution of auxin and consequently induce changes of gene expression leading to the process of tissue reunion in Arabidopsis (Asahina et al. 2011). These works would guide future investigation of the roles of wounding and tissue disconnection in shoot regeneration in the Torenia culture.

In summary of the results obtained through the present study, the whole course of shoot regeneration in the Torenia culture system is schematized in . Global changes of gene expression rapidly occur upon culture independently of cytokinin, possibly reflecting the epidermal cell reprogramming from the fully differentiated state to enter the pluripotent state. These initial gene expression changes include the upregulation of genes commonly implicated in various callus-inducing cultures of Arabidopsis. Then, cytokinin transcriptionally stimulates nucleolar development and ribosome biogenesis, which is followed by the activation of epidermal cell division. Finally, adventitious bud SAMs are de novo organized from cell division foci in the epidermis. Our findings depict the overall process of direct shoot regeneration in the Torenia stem culture consisting of three distinct phases with different transcriptomic and regulatory features, which opens the way for further investigation of hidden aspects of shoot regeneration with this unique system.

Fig. 11

Schematic sketch of processes of shoot regeneration in stem segment culture of Torenia.

Materials and Methods

Plant material and growth conditions

All experiments were carried out using a genetically homogeneous inbred line of Torenia fournieri Lind. that had been developed through 11 generations of self-pollination. Surface-sterilized and stratified seeds were sown on basal medium, which was half-strength MS medium containing 2% (w/v) of sucrose, buffered with 0.05% (w/v) of 2-morpholinoethanesulfonic acid at pH 5.7 and solidified with 0.25% (w/v) of gellan gum, and plants were aseptically grown at 22°C under continuous light (60–100 µmol/s/m2).

Tissue culture

Stems were excised from the internodes between the cotyledons and the first pair of true leaves of 4-week-old plants. Each internode stem of a quadrangular prism shape with wider and narrower lateral faces was sliced longitudinally along the midline of the narrower side into two sections and then cut into 1.5-mm-long segments. The stem segments were placed on the basal medium described above or the basal medium supplemented with 1 mg/l of BA such that the sliced surface was in contact with the medium. The subsequent culture was conducted at 22°C under continuous light (50–70 µmol/s/m2) or in the dark.

Flow cytometric analysis of nuclear DNA content

To isolate nuclei, samples were chopped in CyStain UV Precise P Nuclei Extraction Buffer (Sysmex, Kobe, Hyogo, Japan) in petri dishes placed on ice and filtered through 20-µm CellTrics filter (Sysmex) after 1-minute incubation on ice. Isolated nuclei were stained with CyStain UV Precise P Staining Buffer (Sysmex), and then, the DNA content of each nucleus was quantified with SyFlow SL (Partec, Görlitz, Saxony, Germany).

Microscopic analysis

For RNAselect and DAPI staining, stripped epidermis of stem explants was fixed in methanol at −20°C for at least 10 minutes. The fixed samples were washed in phosphate buffered saline at pH 7.2 (PBS) and then stained in PBS solution containing 25% (v/v) CyStain UV Precise P Staining Buffer (Sysmex), 1 µM CYTO RNAselect Green Fluorescent Cell Stain (Invitrogen, Waltham, MA, USA) and 0.1%(w/v) TritonX-100 for 30 minutes at room temperature while being protected from light. The stained samples were washed in PBS before observation.

For detection of nascent cell walls, the epidermis of stem explants was stained with aniline blue according to the protocol described in Schenk and Schikora (2015) with minor modifications. Stripped epidermis was fixed in a 1:3 mixture of acetic acid and ethanol for at least 24 hours at room temperature. After washing in 150 mM K2HPO4 for 30 minutes, the samples were stained in 1% (w/v) aniline blue solution containing 150 mM K2HPO4 for 2.5 hours at room temperature while being protected from light. The stained samples were washed in 150 mM K2HPO4 before observation.

The epidermis samples stained with DAPI, RNAselect or aniline blue were observed under the Olympus BX50F4 microscope.

For CLSM, explants were thinly sliced and vacuum-infiltrated with 1 µg/ml PI. Samples were then observed using a confocal laser scanning microscope (Leica, Wetzlar, Hesse, Germany; TCS SP5).

Serial observation of the surface of cultured explants was performed with a metallurgical microscope (WRAYMER, Osaka, Osaka, Japan; BM-3400TL).

Transcriptome analysis

All transcriptome analyses were carried out in three biological replicates. Collected samples were immediately frozen with liquid nitrogen and stored at −80°C until use. Total RNA was isolated from the frozen samples with Direct-zol RNA MiniPrep Kit (Zymo Research, Irvine, CA, USA).

For RNA-seq analysis of a set of samples consisting of stem explants cultured for 0, 2, 4, 6 and 8 d, libraries were prepared from total RNA with messenger RNA (mRNA)-seq Kit with KAPA mRNA Capture Beads (KAPA, Wilmington, MA, USA), NEBNext Multiplex Oligos for Illumina Index Primers Set 1–4 (NEB, Ipswich, MA, USA) and Agincourt AMPure XP (Beckman Coulter, Brea, CA, USA) according to manufacturers’ protocols. The libraries were sequenced with Nextseaq600 (Illumina, San Diego, CA, USA). Raw reads containing adapter sequences were trimmed using bcl2fastq (Illumina), and nucleotides with low-quality (QV < 25) were masked by N using the original script. Reads shorter than 50 bp were discarded, and the remaining reads were mapped to the Torenia cDNA database (http://dandelion.liveholonics.com/torenia/), which had been constructed from mRNAs of leaves and roots of young seedlings and floral organs of flowering plants, using Bowtie with the following parameters: ‘--all --best --strata’ (Langmead et al. 2009). Reads were counted by transcript models.

For RNA-seq analysis of a set of samples consisting of various parts of 4-week-old plants (shoot apices, the first and second pairs of true leaves, the first internode stems and whole roots) and stem explants at the early stage of culture (explants cultured for 0, 3, 6, 12, 24 and 48 hours), libraries were prepared with mRNA HyperPrep Kit (KAPA) and Multiplex Oligos for Illumina Index Primers Set 1–4 (NEB) according to manufacturers’ protocols. The libraries were sequenced with Novaseaq6000 (Illumina). Reads were mapped to the Torenia cDNA database (http://dandelion.liveholonics.com/torenia/) using Bowtie2 (Langmead and Salzberg 2012), and the expression level of each transcript was quantified with Salmon (Patro et al. 2017).

The log-transformed Transcripts Per Million (TPM) values of the RNA-seq transcriptome data were subjected to PCA using the R prcomp function.

Differential expression analysis was performed with edgeR (Robinson et al. 2009, McCarthy et al. 2012) and limma (Ritchie et al. 2015) packages of R. K-means clustering analysis was performed on Multiple Experiment Viewer platform (Saeed et al. 2003). Assignment of GO annotation to transcript sequences of Torenia was conducted with Blast2GO (Conesa and Götz 2008) based on the results of homology search against the Arabidopsis subset and Viridiplantae subset of the NCBI non-redundant database and also on the protein domains identified by InterPro domain search. GO enrichment analysis was also carried out on Blast2GO by Fisher’s exact test with cutoff at FDR <0.05.

For comparison of transcriptome data between Arabidopsis and Torenia, RNA-seq data and microarray data of Arabidopsis were obtained from the public resource. The Arabidopsis RNA-seq data were processed as described above. The microarray data were normalized by a variant of MAS5.0 with robust radius-minimax estimators (Kohl and Deigner 2010). Then, differentially expressed genes were identified with the rank products method with a cutoff at FDR <0.05 using the Rank Prod R package (Del Carratore et al. 2017).

Identification of Torenia orthologs to Arabidopsis genes

Orthologs of Torenia to the SAM regulator genes of Arabidopsis were identified by homology search against the amino-acid sequence database deduced from the Torenia cDNA database (http://dandelion.liveholonics.com/torenia/) with the amino-acid sequences of Arabidopsis SAM regulators as queries followed by phylogenetic tree construction. In other cases, Torenia orthologs to a set of Arabidopsis genes of interest were identified using OrthoFinder (Emms and Kelly 2015, 2019).

RT-qPCR analysis

Total RNA was isolated with Direct-zol RNA MiniPrep Kit (Zymo Research). From each RNA preparation, potentially remaining genomic DNA was eliminated and the first-strand cDNA was synthesized using PrimeScript RT reagent Kit with gDNA Eraser (Perfect Real Time) (TaKaRa, Kusatsu, Shiga, Japan). Then, qPCR was performed with gene-specific primers () using TB Green Premix Ex Taq II (Tli RNaseH Plus) (TaKaRa) on Step One Real-Time PCR System (Applied Biosystems, Waltham, MA, USA). The data were normalized with the ubiquitin gene TfUBQ10, a Torenia ortholog of Arabidopsis UBQ10, as an internal control.

Click here for additional data file.