Comparative transcriptome profiling reveals the multiple levels of crosstalk in phytohormone networks in Brassica napus

Summary

Plant hormones are the intrinsic factors that control plant development. The integration of different phytohormone pathways in a complex network of synergistic, antagonistic and additive interactions has been elucidated in model plants. However, the systemic level of transcriptional responses to hormone crosstalk in Brassica napus is largely unknown. Here, we present an in‐depth temporal‐resolution study of the transcriptomes of the seven hormones in B. napus seedlings. Differentially expressed gene analysis revealed few common target genes that co‐regulated (up‐ and down‐regulated) by seven hormones; instead, different hormones appear to regulate distinct members of protein families. We then constructed the regulatory networks between the seven hormones side by side, which allowed us to identify key genes and transcription factors that regulate the hormone crosstalk in B. napus. Using this dataset, we uncovered a novel crosstalk between gibberellin and cytokinin in which cytokinin homeostasis was mediated by RGA‐related CKXs expression. Moreover, the modulation of gibberellin metabolism by the identified key transcription factors was confirmed in B. napus. Furthermore, all data were available online from http://yanglab.hzau.edu.cn/BnTIR/hormone. Our study reveals an integrated hormone crosstalk network in Brassica napus, which also provides a versatile resource for future hormone studies in plant species.

Introduction

Plant growth and development are adjusted by many extrinsic and intrinsic factors. Phytohormones, the small signalling molecules, are one of the intrinsic factors that produced by plants and regulate a variety of fundamental biological processes and responses to environmental stimuli (Ciura and Kruk, 2018; Müller and Munné‐Bosch, 2021; Takeuchi et al., 2021). Hormones are synthesized in plant tissues, which are received by hormone receptors with different affinities, and stimulate a range of gene expressions through signal transduction via protein–protein interactions, post‐translational modifications and regulation of transcription factor (TF) activities, ultimately altering cellular function and plant behaviour (Takeuchi et al., 2021; Verma et al., 2016; Yamamuro et al., 2016).

The functions of plant hormones have been elucidated through studies of exogenous hormone application combined with analysis of the relative biosynthetic and signalling mutants (Blázquez et al., 2020; Wang et al., 2020). Numerous components of the core signalling pathways of various plant hormones have been identified through genetic screens and hypothesis‐driven approaches, leading to the elucidation of partial or entire metabolic pathways and signal transduction cascades (Belkhadir and Jaillais, 2015; Davière and Achard, 2013; Kieber and Schaller, 2018; Lavy and Estelle, 2016; Liu and Timko, 2021; Raghavendra et al., 2010; Yang et al., 2015). For example, gibberellins (GAs), together with auxins, and brassinosteroids (BR), regulate cell expansion along longitudinal axis and strongly influence plant architecture and organ size (de Lucas and Prat, 2014; Gonzalez et al., 2012; Perrot‐Rechenmann, 2010), mutation of core signalling factors leads to dwarfism (e.g. gai, axr2‐1, bri1‐3; Noguchi et al., 1999; Peng et al., 1997; Timpte et al., 1994); ethylene and cytokinins act primarily to increase cell expansion along transverse axes and strongly reduce hypocotyl elongation of dark‐grown seedlings (Cary et al., 1995; Smets et al., 2005; Yu and Huang, 2017); abscisic acid (ABA) and ethylene promote senescence and abscission of leaves and flowers (Iqbal et al., 2017; Pandey et al., 2000; Zhao et al., 2017), which have also been shown to antagonize GA and BR‐mediated cell elongation (Liu and Hou, 2018). Genetic screening and dissection of hormone‐insensitive and hypersensitive mutants provide evidence for the non‐redundant roles of hormones. For example, loss of a single pathway of GA, auxin and BRs, results in dwarfism, despite the apparent overlap in the growth‐promoting activities of these three hormones (Yu et al., 2002).

Different phytohormones have been shown to integrally influence overlapping processes in a complex network of synergistic, antagonistic and additive interactions (Depuydt and Hardtke, 2011; Durbak et al., 2012). The mechanisms underlying the crosstalk between different hormone pathways have been addressed both at the level of hormone response and biosynthesis, creating a delicate response network (Berens et al., 2017; Jaillais and Chory, 2010). For instance, DELLA proteins act as repressors of a central growth regulatory pathway that are rapidly degraded or accumulated in the presence or absence of GA respectively (Sun, 2011; Wang and Deng, 2014), and the turnover of DELLA protein levels is required for a normal GA growth response (Van De Velde et al., 2017). Exogenous application of auxin or ethylene accelerates or delays GA‐induced DELLA degradation respectively (Achard et al., 2003; O'Neill et al., 2010). Further evidence confirmed that the GA‐induced disappearance of DELLAs is dependent on the auxin and ethylene signalling pathways (He and Yamamuro, 2022; Luo et al., 2013). Furthermore, the signal integration between GA and other hormones is established by DELLA proteins through their interaction with transcription factors (TFs) of other phytohormones (Bao et al., 2020; Davière and Achard, 2016), indicating that DELLAs have been proposed as key integrators of plant growth from multiple hormonal signals.

Brassica napus (AACC, 2n = 38) originated from a spontaneous hybridization between Brassica rapa (AA, 2n = 20) and Brassica oleracea (CC, 2n = 18) and is one of the most important allotetraploid oil crops in the world (Chalhoub et al., 2014; Lu et al., 2019; Wang et al., 2018). Understanding the molecular and systemic levels of hormone interactions will help us predict the effects of disruption or over‐activation of specific parts of the network on the overall plant response. Transcriptional profiling and binary protein–protein interactions have illustrated the crosstalk and integration of plant hormone pathways in Arabidopsis (Nemhauser et al., 2006). However, little is known about the crosstalk of different phytohormones in other plant species, especially in the allotetraploid oil crops Brassica napus. In this work, we treated 14‐day‐old B. napus seedlings with seven phytohormones at five timepoints (0, 0.5, 1, 3, 6 h), and used RNA‐seq to investigate the overlapping in transcriptional effects of the different hormones. Differentially expressed genes (DEGs) analysis was used to reveal genes that co‐regulated by seven hormones, and the regulatory networks between seven hormones were constructed, revealing the novel regulatory network between gibberellin and cytokinin. In addition, a comprehensive database was made available online at http://yanglab.hzau.edu.cn/BnTIR/hormone to facilitate the search for gene expression patterns. These studies show that the long‐term effects of all hormone treatments represent a ‘domino effect’, resetting many systems within the plant and provide a versatile resource of genome‐wide gene expression for future hormone studies in the plant kingdom.

Results

The experimental design and data overview of global transcriptomes for phytohormone treatment in B. napus

Phytohormones play a central role in plant growth and development. To investigate the dynamic changes in the transcriptome with hormone treatment at the seedling stage, the ZH11 (Zhongshuang 11) seeds were first germinated for 2 days. The well‐germinated seedlings were selected and placed on the floats, which were submerged in the Hoagland nutrient solution container and grown for a further 12 days. The 14‐day‐old seedlings were treated with Hoagland solution containing different hormones (abscisic acid (ABA), gibberellic acid 3 (GA), indole‐3‐acetic acid (IAA; auxin), 1‐amino‐cyclopropane‐1‐carboxylic acid (ACC; ethylene precursor), t‐zeatin (CK; cytokinin), brassinolide (BL; brassinosteroid) and methyl jasmonate (MJ; jasmonate)), and the relative root and shoot samples were collected at different timepoints (0, 0.5, 1, 3 and 6 h; Figure 1a). Meanwhile, the 14‐day‐old ZH11 seedlings treated with Hoagland solution without any hormones were also collected at different timepoints, which were considered as mock (control) for each timepoint. Each experiment was performed in triplicate as biological replicates. The RNA‐seq was then performed on the Illumina sequencing platform (HiSeq 2500), and 198 RNA‐seq datasets were collected (Figure 1a).

Figure 1.

Schematic of experimental design, as well as the global transcriptome response of the B. napus to hormone treatment. (a) The experimental procedure and the total sequencing size. (b) The expression level of ABA, ETH, BR, GA, IAA, tZ and MeJA marker genes under ABA, ETH, BR, GA, IAA, tZ and MeJA treatments. The x‐axis represents the timepoints (0, 0.5, 1, 3 and 6 h). The y‐axis represents the expression levels of three biological replications. The TPM values were divided by 10 or 100 if the gene expression was >100. Student's t‐test was used to determine P value, the data are presented as mean ± SD (n = 3) and significant differences are indicated: *P < 0.05; **P < 0.01; ***P < 0.001. (c) The number of DEGs under different hormone treatments. The grey, orange and deep blue histograms represent specific DEGs (differentially expressed under one hormone), variable DEGs (differentially expressed under two to six hormones) and common DEGs (differentially expressed under seven hormones). (d) The upset plot shows the number of differentially expressed genes under different hormone treatments, simultaneously. The bar graph shows the number of up‐ and down‐regulated DEGs.

A total of 14.13 billion high‐quality reads were generated, and then mapped to the B. napus ZS11 reference genome (Song et al., 2020) using Hisat (Kim et al., 2015). On average, ∼98.03% of the reads were mapped to ZS11 (Table S1), which were further used to calculate the normalized gene expression level as transcripts per million mapped reads (TPM). The expression values among the three biological replicates were highly correlated (average R ² = 0.96; Figure S1; Table S2). The average TPM value of the three replicates was taken as the expression level for the sample at each timepoint, and the TPM ≥ 1 was defined as a gene as expressed.

The validation of hormone response genes in B. napus

To further validate the quality of the gene profiles, the effects of each hormone were confirmed using the Arabidopsis hormone response marker genes (Tables S3 and S4). Each marker gene was up‐ or down‐regulated according to previously reported patterns (Figures 1b, S2 and S3). For instance, BnaA1.IAA13 (BnaA03G0160600ZS) and BnaA6.ARR5 (BnaA06G0170100ZS) were significantly induced by IAA and TZ treatment respectively (D'Agostino et al., 2000; Effendi et al., 2011), at all four timepoints (Figures 1b and S2); in contrast, BnaA6.DET2 (BnaA04G0244700ZS) and BnaA4.PYL6 (BnaA04G0257000ZS) were suppressed by BR and ABA treatment respectively (Tanaka et al., 2005; Zhao et al., 2020), at all four timepoints (Figures 1b and S2), indicating that the treatments were effectively altered the expression of hormone response marker genes. As B. napus is a young allotetraploid species, the majority of genes in B. napus (AACC) are multiple‐copy genes with high sequence similarity (Chalhoub et al., 2014; Lu et al., 2019; Wang et al., 2018). Approximately 5.0%–39.3% of the homologous genes showed a divergence in the expression pattern (Figure S4), indicating that these homologous genes might have divergent functions in hormone responses. Notably, the auxin and gibberellin marker genes were strongly induced in the root rather than in the shoot (Figure S5A,B), consistent with the fact that the roots, but not the whole plants were immersed in the hormone solution. Finally, the DEGs from the root and shoot datasets were combined for further analysis.

The criteria (|log₂(fold change|) ≥ 1, and P‐value < 0.05 with false discovery rate (FDR) adjustment) were used to identify differentially expressed genes (DEGs) that were affected by hormone treatment at any of the four timepoints. The DEGs were categorized as up‐regulated (genes were up‐regulated at least at one timepoint), down‐regulated (genes were down‐regulated at least at one timepoint) and complex (genes were up‐regulated at one timepoint but down‐regulated at another timepoint; Figures 1c and S6; Tables S5–S8). Approximately 1913–19 769 (7325 on average) DEGs were identified compared to the control at each timepoint (Tables S5–S8). The number of hormone‐response genes (up‐ and down‐regulated) was shown in Tables S5–S7. Similar to Arabidopsis, ABA‐ and BR‐responsive genes formed the largest (17 094 up‐regulated, 15 336 down‐regulated) and smallest (4915 up‐regulated, 4425 down‐regulated) groups respectively (Figure 1c). In total, 980 to 6951 (2869 on average) genes were specifically up‐ or down‐regulated by a hormone at a single timepoint (Figure S6), while 758 to 10 806 (3590 on average) genes were viably expressed at different timepoints (Figure S6), indicating that the expression patterns of hormone‐responsive genes varied dynamically. Notably, very few genes were consistently up‐ or down‐regulated by IAA, tZ or ETH at four timepoints, and 2012 and 3152 genes were consistently up‐ or down‐regulated by ABA, and MeJA at four timepoints respectively (Figure S6).

To gain insight into the relationships between the phytohormones, the common hormone‐responsive genes among the hormones were identified by comparing the lists generated from the DEGs. 628–10 813 (4513 on average) DEGs were responsive to only one hormone treatment, and 8261–21 166 (13 754 on average) DEGs were responsive to 2–6 types of hormone treatments (Figure 1c,d). In particular, the number of genes affected by all seven hormones was strikingly small (451 genes with the same direction under seven hormones; 135 up‐regulated, 316 down‐regulated; Figure 1d). Remarkably, 63 genes were changed in the same direction by GA, IAA and BL treatments—40 induced and 23 repressed—none with known function. Similarly, 4332 genes (2227 induced, and 2105 repressed) were changed in the same direction by MeJA and ABA treatments (Figure 1d), indicating that MeJA and ABA might play synergistic roles in seed germination and response to water deficit (de Ollas and Dodd, 2016; Varshney and Majee, 2021). The effects of GA are antagonized by the ABA pathway (Liu and Hou, 2018), and 3964 genes were altered in opposite directions by GA and ABA treatments (Figure 1d). These common transcriptional targets of ABA and GA were as likely to be affected in the same as in the opposite direction.

The Gene Ontology (GO) terms were then used to identify trends in the responsive genes for each hormone. Genes encoding proteins involved in ‘DNA binding’, ‘intracellular signal transduction’ and ‘oxidation–reduction process’ were significantly enriched for almost all treatments, consistent with the role of hormones as triggers of signal transduction cascades (Figure S7). Significantly, 1684 genes responded to all seven hormones, and the different response patterns were generated based on the expression profiles of these genes (Figure S8A; Table S9). We identified eight significant expression clusters from all DEGs in response to all seven hormone treatments (Figure S8B). We further performed GO enrichment analysis using the DEGs. The results showed that many hormones‐ and stress‐related biological processes were enriched, including ‘response to chitin’, ‘response to bacterium’, ‘response to cytokinin’ and ‘response to abscisic acid’ (Figure S8C). These results indicate that different phytohormones interact extensively, and the DEGs presented here will be a versatile resource for future hormone interaction studies.

Hormone treatments trigger widespread effects on hormone metabolism in B. napus

Despite very few common transcriptional targets among seven hormones, other evidence suggests that one hormone‐regulating gene is involved in another hormone metabolism pathway (Nemhauser et al., 2006). We then investigated whether there was a significant pattern between each set of hormone metabolism genes (biosynthesis and catabolism) side by side. Genes assigned to hormone metabolism pathways were identified based on GO annotation (Table S10; see Methods). As a result, there were 5324 hormone‐related genes in ZS11, including 702, 257 and 4365 biosynthesis, metabolism and signalling genes respectively.

The networks of hormone effects on hormone metabolism were then generated according to the previous criteria (Goda et al., 2008; Nemhauser et al., 2006). Lines with arrowheads and blocked arrows represent up‐ and down‐regulation of hormone biosynthetic genes, respectively, and diamond arrowheads indicate changes in gene expression with ambiguous outcomes (e.g. genes affected include those associated with both increased and decreased hormone levels; Figure 2a). Similar to Arabidopsis, exogenous IAA, BR, tZ and GA treatment suppressed their own biosynthetic genes (IAA: up‐regulated/down‐regulated genes = 18/37, P = 1.74 × 10⁻²; BR: up‐regulated/down‐regulated genes = 3/19, P = 5 × 10⁻⁴; tZ: up‐regulated/down‐regulated genes = 3/13, P = 9.56 × 10⁻³; GA: up‐regulated/down‐regulated genes = 7/15, P = 7.27 × 10⁻²) and promoted the catabolic genes (IAA: up‐regulated/down‐regulated genes = 21/10, P = 6.24 × 10⁻²; BR: up‐regulated/down‐regulated genes = 9/1, P = 1.63 × 10⁻³; tZ: up‐regulated/down‐regulated genes = 15/4, P = 7.12 × 10⁻²), suggesting that these bioactive growth‐promoting hormones are governed by feedback regulation (Figures 2a and S9A; Tables S11 and S12). In contrast to the growth‐promoting hormones, the biosynthetic and catabolic genes of ABA and MeJA were up‐regulated by ABA and MeJA respectively (ABA: up‐regulated/down‐regulated biosynthetic genes = 36/9, P = 9.56 × 10⁻⁷; MeJA: up‐regulated/down‐regulated catabolic genes = 28/2, P = 3.21 × 10⁻⁷; Figures 2a and S9A; Tables S11 and S12), indicating that plants must maintain the homeostasis of these hormones through both synthetic and metabolic pathways. In Arabidopsis, the IAA‐induced ethylene production is affected by enhancing the expression of the ACC synthase (ACS) genes (Alonso and Ecker, 2001). We found that the ethylene biosynthetic genes were mainly repressed by IAA (up‐regulated/down‐regulated genes = 8/28, P = 1.38 × 10⁻³), and the majority of them were ACS genes (e.g. Bna.ACS2, Bna.ACS5 and Bna.ACS7; Figure 2a). In addition to the synergistic and additive transcriptional regulation between the hormones, the antagonistic transcriptional effects between the hormones were also observed. For example, 22 of the GA biosynthesis genes were repressed by exogenous ABA (Figure 2a). The complex web of such interactions suggests that the long‐term effects of all hormone treatments represent a ‘domino effect’, resetting many systems within the plant.

Figure 2.

Crosstalk network of hormone biosynthetic genes. (a) Crosstalk of hormone biosynthesis genes. Lines with different colours represent different hormone responses. The number of up‐ and down‐regulated genes under each hormone treatment are shown on the line. Lines with arrowheads represent up‐regulation of hormone biosynthesis genes. Blocked arrows represent down‐regulation of genes involved in hormone biosynthesis. Diamond arrowheads indicate changes in gene expression with ambiguous outcomes (e.g. genes affected include those linked to both increased and decreased hormone levels). (b) Overlapping of differentially expressed hormone biosynthesis genes under different hormone treatments. The red bar represents the 25 genes differentially expressed under seven hormone treatments simultaneously. (c) The heatmap shows the expression profile of hormone biosynthesis genes differentially expressed under at least three hormone treatments. The heatmap clustered the log₂(fold change) of each hormone biosynthetic gene, and we selected the largest fold change of the four timepoints. Blue and red represent the up‐ and down‐regulations under the hormone treatments. (d) Co‐expression network of 25 hormone biosynthesis genes. Red circles represent hub hormone biosynthesis genes. Triangle with different colours represent different hormone biosynthesis genes. (e) Subnetworks of the hub biosynthetic gene co‐expression network. (f) Co‐expression related to the subnetwork in (e). (g) The heatmap shows the expression profile of ACO2 genes under GA and tZ treatment. Log₂(FC), log₂ fold change. (h) Relative expression levels of the BnaA9.ACO2 genes which were co‐expressed with BnaRGA and BnaCXK2 selected by RNA‐seq in 14 d seedlings of Bnarga (L27 and L46), WT and BnaA9.CKX2‐OE (L1 and L2) analysed by RT‐qPCR. The gene expression level of WT was set as 1. Values are the means ± SD of three biological replicates. Student's t‐test was used to calculate significance. The * and ** indicate P < 0.05 and P < 0.01 respectively.

We then attempted to determine whether there was a significant overlap between each set of hormone metabolic genes. Among all metabolic genes, 49 genes (25 biosynthetic and 24 catabolic genes) were simultaneously altered by seven hormone treatments, of which nine genes belonged to the MeJA biosynthetic pathway and seven genes belonged to the tZ catabolic pathway (Figures 2b,c and S9B,C; Tables S13 and S14). Subsequently, the co‐expression networks were generated using 25 biosynthetic and 24 catabolic genes as the hub genes, respectively, with an incorporated connection (Figures 2d and S9D). For example, the expression of 75 genes was correlated with ACO2 (Figure 2e; Table S15). As expected, several ethylene biosynthesis‐related genes such as BnaA10.ACO4 and BnaC5.ACO4 were significantly associated with ACO2 (Pearson's correlation coefficient = 0.82, 0.72 respectively; P < 2.2 × 10⁻¹⁶, P < 2.2 × 10⁻¹⁶ respectively; Figure 2e; Table S15). In addition to these ethylene biosynthetic genes, ACO2 was also co‐expressed with GA and cytokinin‐related genes (e.g. RGA and CKX2; Figure 2f; Table S15). ACO2 was simultaneously repressed by GA and cytokinin (Figure 2g). Previously, we generated the genome‐editing quadruple mutants of BnaRGA (Bnarga; Yang et al., 2017). In Bnarga, the expression of BnaA9.ACO2 was decreased (Figure 2h), suggesting that the GA‐suppressed BnaACO2 expression might be mediated by RGA. Cytokinin levels were decreased in B. napus BnA9.CKX2 overexpressing lines (unpublished data), and the expression of BnaA9.ACO2 was strongly induced in the two BnA9.CKX2 overexpressing lines (L1 and L2; Figures 2h and S10), consistent with BnaACO2 being suppressed by cytokinin.

Identification of key transcription factors regulating hormone metabolism

Hormone homeostasis can be monitored and reprogrammed by hormone‐responsive transcription factors (Hill, 2015). We then investigated the hub transcription factors which involved in the regulation of hormone metabolism by co‐expression analysis. The RNA‐seq datasets were obtained from this study and BnTIR (Liu et al., 2021), and the relative hormone metabolism pathway genes are listed in Table S9. The co‐expression networks of each hormone were generated by correlating the expression patterns of genes with hormone biosynthesis genes (Wang et al., 2022; Tables S16–S18). Taking GA as an example, bioactive GAs are synthesized from transgeranylgeranyl diphosphate in the plastid, and the main GA metabolic pathway has been elucidated in model plants (Yamaguchi, 2008). The expression of these GA structural genes showed variable patterns in B. napus by different hormone treatments (Figure 3a), and the transcripts of these genes were highly correlated with 306 TFs (PCC > 0.8), including the B3, bHLH, Dof and MIKC_MADS families (Figure 3b; Table S16).

Figure 3.

Identification of key transcription factors regulating gibberellin biosynthetic. (a) Biosynthetic pathway of GA in Arabidopsis. Substrates and products are represented by plain text; gene‐encoding enzymes are indicated by italics. Genes expressed under different hormone treatments are indicated in red and blue respectively. CPS, ent‐copalyl diphosphate synthase; GA2ox, GA 2‐oxidase; GA3ox, GA 3‐oxidase; GA20ox, GA 20‐oxidase; KAO, ent‐kaurene acid oxidase; KO, ent‐kaurene oxidase; KS, ent‐kaurene synthase; GAMT, gibberellin methyltransferase. (b) The regulatory network of gibberellin biosynthesis. Pink circles represent gibberellin biosynthetic genes. Diamonds with different colours represent different families of transcription factors whose transcripts were correlated with the expression of gibberellin biosynthetic genes. (c) Transcriptional regulatory network of Bna.DOF5.7, Bna.AGL15, Bna.FUS3. Diamonds represent key transcription factors regulating gibberellin biosynthesis in B. napus. (d) Schematic representation of transcription factor binding sites in the promoters of different gibberellin biosynthesis genes. (e) Schematic representation of the constructs used for the dual‐luciferase assay. The reporter construct containing the firefly luciferase was driven by the BnaA6.KO and the BnaC4.GA20OX4 promoter, respectively, and the Renilla luciferase (REN) was driven by the CaMV 35S promoter. The effector constructs contained BnaA6.DOF5.7, BnaC9.AGL15 and BnaC2.FUS3 was driven by the CaMV 35S promoter. (f) The LUC/REN ratios in the dual‐luciferase assay. Data are means ± SD obtained from three biological experiments. Student's t‐test was used to calculate significance. The *** indicates P < 0.001. (g) Yeast one‐hybrid assay of interactive effects between BnaA6.KO promoter and BnaC2.FUS3. Transformed yeast cells were grown on SD/‐Leu and SD/‐Leu containing 100 ng/mL aureobasidin A respectively. (h) EMSA analysis of the recombinant BnaC2.FUS3‐His protein and the promoter sequence of BnaA6.KO. Schematic diagram (upper panel) shows DNA sequence of BnaA6.KO promoter (−1070 to −1021) designed for EMSA probe. The underline shows RY elements and the mutated probe. The image (bottom panel) shows the result of EMSA analysis. The red arrow indicates the shift bands. The asterisk indicates that the mutated probe was used as the competing probe, and the 32a‐His protein was used as control.

To investigate the potential regulatory effect of these TFs, three TFs, including BnaDOF5.7, BnaAGL15 and BnaFUS3 were selected and the transcriptional regulation of GA structural genes was examined. Regulatory network analysis revealed that BnaDOF5.7, BnaAGL15 and BnaFUS3 transcripts were highly correlated with seven, seven and eight GA structural genes respectively (Figure 3c; Table S16). The conserved DOF5.7, AGL15 and FUS3 motifs, known to be the putative binding sites of the Dof‐, MADS‐ and B3‐type TFs, were identified in the promoter sequences of GA structural genes (Figure 3d). The potential regulatory effects of BnaDOF5.7, BnaAGL15 and BnaFUS3 on GA structural genes were investigated. The promoter regions of BnaA6.KO and BnaC4.GA20OX4 containing the corresponding cis‐elements were fused to the firefly luciferase (LUC), while the Renilla luciferase (REN) reporter driven by the 35S promoter served as an internal control (Figure 3e). Co‐expression of the effectors with the proBnaA6.KO::LUC and proBnaC4.GA20OX4::LUC in protoplasts isolated from Arabidopsis leaves resulted in a significant increase or decrease in luminescence intensity (Figure 3f). There were two predicted FUS3‐binding sites in the BnaA06.KO promoter, designated P1 (−1049 to 1043 bp) and P2 (−253 to −247 bp) respectively. Y1H and EMSA results indicated that BnaC2.FUS3 could directly bind to the P1 site of the BnaA6.KO promoter (Figure 3g,h). However, the region containing the P2 site showed strong self‐activation in the Y1H assay, and EMSA results also showed that there was no direct binding between BnaC2.FUS3 and P2 site of BnaA6.KO promoter (Figure S11). These results demonstrate that BnaC2.FUS3 inhibits the expression of BnaA6.KO by specifically binding to the P1 site of the BnaA6.KO promoter.

In addition, we also constructed the regulatory network to characterize the transcription factors putatively regulating IAA and ETH biosynthesis. The 286 TFs (PCC > 0.8) including the WRKY, bHLH and MIKC_MADS families were highly correlated with IAA structural genes (Figure S12A,B; Table S17). BnaWRKY69 and BnaWRKY72 transcripts were highly correlated with five IAA structural genes (Figure S12C). In addition, 244 TFs including WRKY, G2‐like, C2H2 and NCA families were highly correlated with ETH structural genes (Figure S12D,E; Table S18). BnaHRS1 and BnaWRKY61 transcripts were highly correlated with four ETH structural genes (Figure S12F). Our results provide an important resource for studying the regulation of hormone synthesis.

Transcriptional integration of hormone signalling in B. napus

Although many interactions between hormone signalling pathways have been described, few molecular mediators of signalling crosstalk have been isolated (Depuydt and Hardtke, 2011). To integrate the hormone signalling at the transcriptional level, genes characterized as hormone signalling pathways were first identified by GO annotation (Table S10). Then, the networks of hormone effects on signalling genes were generated (Figure S13A). In contrast to the hormone metabolism regulation network, the majority of gene expressions showed ambiguous outcomes (Figure S13A; Table S19). One possibility is that the hormone biosynthesis gene expressions are feedback regulated by their own signalling cascades, which could be misinterpreted as one pathway being regulated by another at the transcriptional level (Fukazawa et al., 2017).

Then, 163 genes were simultaneously regulated by seven hormones (Figure S13B; Table S20), and 53 genes belonged to the ETH signalling pathway (Table S20). Later, the co‐expression networks were generated using 163 signalling genes as the hub genes, with an incorporated connection (Figure S13C,D). For example, the B. napus IAA26 was co‐expressed with 117 hormone‐related genes (including 9 metabolic genes, and 90 signalling genes; Table S21); 62 were auxin‐related genes (Figure S13E). IAA26 was interacted with a group of AUX/IAA and ARF proteins which are involved in auxin signalling (Figure S13F; Altmann et al., 2020; Piya et al., 2014; Trigg et al., 2017). These results indicate that IAA26 might regulate the auxin signal pathway through transcriptional and protein–protein interactions. However, the hierarchical alignment and/or ultimate direction of these hub transcription factors needs to be investigated in the future.

The homeostasis of cytokinin is mediated by RGA‐dependent GA signalling pathway

The antagonistic effects on numerous developmental processes are found between GA and cytokinin (Fleishon et al., 2011; Weiss and Ori, 2007). In Arabidopsis, GA blocks the cytokinin responses via SPY (SPINDLY; Greenboim‐Wainberg et al., 2005). Based on the effects of hormones on hormone metabolic networks, we found that four cytokinin biosynthetic genes IPTs were induced by GA, while 18 cytokinin oxidase genes CKXs were repressed by GA (Figures 4a and S9A). The results were further confirmed by RT‐qPCR, that the cytokinin synthetic genes BnaIPT1, BnaIPT5 and BnaIPT7 were induced and repressed by GA and PAC (the inhibitor of GA biosynthesis) respectively; on the contrary, the cytokinin oxidase genes BnaCKX3, BnaCKX5, BnaCKX6 and BnaCKX7 were repressed and induced by GA and PAC respectively (Figure 4b). Notably, the key cytokinin transcription factors, ARR genes, were also repressed by GA (Figure 4a; Table S22), and induced by PAC (Figure 4b). The synthetic cytokinin reporter, TCSn (two‐component signalling sensor) that harbours the concatemerized binding motifs for activated type‐B nuclear response regulator (RR) and fused with GUS, was then transformed into B. napus. The GUS activities were induced by tZ (Figure 4c), indicating that the TCSn::GUS could allow the visualization of cytokinin‐responsive tissues. Then, TCSn::GUS transgenic lines (L3 and L6) were immersed in the buffer containing different concentrations of GA and PAC, and further GUS staining was performed. Without any treatment, the GUS staining signals were found only in the root tips, and the signals were suppressed by GA (Figure 4c); in contrast, the GUS signals were induced by PAC (Figure 4c), and the expression level of TCSn‐GUS was significantly suppressed and induced by GA and PAC respectively (Figure 4c). However, there were no significant differences in TCSn‐GUS expression between different concentrations of GA or PAC treatments (Figure 4c). These results indicate that the endogenous GA level is essential for the cytokinin response, and that GA‐ or PAC‐mediated TCSn‐GUS expression was not dose‐dependent.

Figure 4.

Interactions validate between cytokinin and gibberellin. (a) The heatmap shows the expression profile of cytokinin genes. The heatmap clustered the log₂ fold change (log₂(FC)) of each gene. Biosynthesis, catabolism and signalling genes are marked on the left. The dotted box represents BnaARRs, BnaCKXs and BnaIPTs. (b) The bar graph shows the relative expression levels of cytokinin biosynthesis, catabolism and signalling genes in 14‐day‐old seedlings after GA₃, PAC or Mock treatments. The cytokinin biosynthesis, catabolism and signalling‐related genes are coloured in red, blue and green respectively. (c) GUS staining of TCSn‐GUS reporter lines (L3 and L6) in roots after GA₃, PAC, tZ or Mock treatments (left panel). Scale bar, 100 μm. The bar graph shows the relative GUS expression level of TCSn‐GUS reporter lines in roots after GA₃, PAC, tZ or Mock treatments (right panel). (d) The bar graph shows the relative expression levels of BnaCKX5 and BnaCKX7 in 14‐day‐old seedlings of Bnaa6.rga‐D, Bnarga and WT after GA₃, PAC or Mock treatments. The bar graph shows the relative expression levels of BnaCKX2s in 20‐DAP seeds of Bnaa6.rga‐D, Bnarga and WT after GA₃, PAC or Mock treatments. DAP, day after pollination. (e) The bar graph shows the relative tZ content in 14‐day‐old seedling of Bnaa6.rga‐D, Bnarga and WT after GA₃, PAC or Mock treatments. (f) Schematic representation of the constructs used for the dual‐luciferase assay (left) and LUC/REN ratios (right) in the dual‐luciferase assay. Data are means ± SD obtained from three biological experiments. Student's t‐test was used to calculate significance. The * and ** indicate P < 0.05 and P < 0.01 respectively. (g) Relative expression levels of GA biosynthesis genes in 14‐day‐old seedlings after tZ or Mock treatments. The gene expression level of Mock was set as 1. (h) The bar graph shows the relative GA₃ content in 14‐day‐old seedlings after tZ or Mock treatments. Values are the means ± SD of three biological replicates. Student's t‐test was used to calculate significance. *** indicates P < 0.001. (i) Model shows the interactions between gibberellin and cytokinin in Brassica napus seedling. In (b), (c) and (g), the gene expression level of Mock were set as 1. BnaActin7 was used as the internal control. Values are the means ± SD of three biological replicates. Student's t‐test was used to calculate significance. The * and ** indicate P < 0.05 and P < 0.01 respectively. In (d) and (e), the data are means ± SD obtained from three biological experiments, and the letters a, b, c, d and e on the bar indicate the statistical differences between different groups at P < 0.05 (Student's t‐test). Letter a indicates the statistical differences between GA‐treated versus mock‐treated (e.g. GA‐treated WT vs. mock‐treated WT); letter b indicates the statistical differences between PAC‐treated versus mock‐treated (e.g. PAC‐treated WT vs. mock‐treated WT); letter c indicates the statistical differences between mock‐treated BnaRGA mutants (Bnaa6.rga‐D and Bnarga) versus mock‐treated WT; letter d indicates the statistical differences between GA‐treated BnaRGA mutants (Bnaa6.rga‐D and Bnarga) versus GA‐treated WT; and letter e indicates the statistical differences between PAC‐treated BnaRGA mutants (Bnaa6.rga‐D and Bnarga) versus PAC‐treated WT. In (b), (c), (d), (e), (g) and (h), Mock indicates that the plants were treated with buffer containing the same amount of ethanol as the GA₃, PAC and tZ treatment solution.

DELLA proteins function as the key repressors of GA signalling. There were 10 DELLA‐encoding genes in B. napus, of which the BnaRGA was one of the most widely studied genes, which has a critical function in vegetative growth and seed development (Wu et al., 2020; Yan et al., 2021). We have previously generated the BnaRGA quadruple mutants of (Bnarga) and the gain‐of‐function mutant of BnaA6.RGA (Bnaa6.rga‐D, L4; Yang et al., 2017). We wonder whether GA‐promoting cytokinin levels are mediated by DELLA proteins. To test this hypothesis, the expression of BnaCKXs genes was compared between Bnarga, Bnaa6.rga‐D and WT by GA and PAC treatments. Compared to the WT, the GA‐suppressed BnaCKX expression was less repressed in Bnaa6.rga‐D (Figure 4d), and the PAC‐induced BnaCKX expression was more induced in Bnaa6.rga‐D and less induced in bnarga (Figure 4d). Degradation of cytokinin is controlled by cytokinin oxidase/dehydrogenase (CKX) enzymes, and the activities of CKXs are correlated with the changes in cytokinin pool (Sakakibara, 2006; Schmülling et al., 2003). One of the endogenous bioactive CKs, t‐zeatin (tZ), was then quantified in bnarga, bnaa6.rga‐D and WT after GA and PAC treatments. Compared to WT, tZ accumulated more in Bnarga and less in Bnaa6.rga‐D without any treatment (Figure 4e). After GA treatment, the tZ levels were increased in WT, whereas tZ was less increased in Bnarga (Figure 4e), probably because the other DELLA proteins (e.g. BnaRGL1, BnaRGL2 and BnaRGL3) might also be involved in GA‐mediated tZ level regulation. The GA‐induced tZ enrichment was significantly suppressed in Bnaa6.rga‐D (Figure 4e). In contrast, tZ was suppressed by PAC, and the inhibition of endogenous tZ by PAC was counteracted in Bnarga (Figure 4e). In addition, the ~1.2 kb promoter region of BnaA7.CKX2 was fused to the firefly luciferase (LUC), while the Renilla luciferase (REN) reporter driven by the 35S promoter served as an internal control (Figure 4f). Co‐expression of the BnaA6.RGA and BnaC7.RGA with the proBnaA7.CKX2::LUC in protoplasts isolated from Arabidopsis leaves resulted in a significant increase in luminescence intensity (Figure 4f). However, the luminescence intensity was suppressed by co‐expression of the BnaA9.RGA with the proBnaA7.CKX2::LUC in protoplasts (Figure 4f), suggesting that the functional differentiation occurs between different RGA homologous proteins.

In Arabidopsis, cytokinin inhibits the expression of GA20ox and GA3ox and promotes that of RGA and GAI (Brenner et al., 2005). We found that the GA biosynthesis genes, CPS, KS, KO and GA20ox were induced by TZ, and the endogenous GA₃ levels were induced by TZ treatment (Figures 4g,h and S14; Table S23). Based on the above results, we propose a network between GA and cytokinin, in which the cytokinin homeostasis is mediated by RGA‐dependent CKXs expression (Figure 4i).

An online database for exploring phytohormone transcriptomes

To provide an easy access and enable visualization of these transcriptome datasets, the ‘Hormone’ module was built on a Linux Ubuntu server (version 7.5) and included in the BnTIR browser (http://yanglab.hzau.edu.cn/BnTIR/hormone), which allows basic queries on the annotated gene names. Online visualization of spatially resolved transcriptomes was implemented using Highcharts, ECharts, Handsontable, DataTables and d3 JavaScript libraries.

To visualize temporally resolved transcriptomes, the eFP module and the heatmap module were used to display the selected gene expression pattern based on the TPM of hormone transcriptomes (Figure 5a and S15). In the eFP module, users could browse the temporal expression profile of the gene of interest in the selected sample, and it was displayed in ‘absolute’, ‘relative’ and ‘compare’ modes. The colour represents the gene expression level. The colour bar is displayed at the top of the graph. A pop‐up box will open to show detailed information (TPM and treatment type) by hovering the mouse over it. The expression patterns after hormone treatments were also displayed as a heatmap, and users could view the expression data in one or more hormone treatments (Figure 5b). The results could be viewed in full screen, print the graph, viewed the data table, and downloaded the image or data in various formats by clicking the graph context menu in the upper right corner of the graph. Users can click and hold down the left mouse button and drag a rectangle in the graph to zoom in.

Figure 5.

The function introduction of hormone transcriptome platform for B. napus. (a) Two examples of the hormone eFP module. Jasmonic acid carboxyl methyltransferase BnaC5.JMT was up‐regulated in different tissues after 10 μM JA treatment. Gibberellin biosynthesis gene BnaA6.GA3ox1 shows high expression at 1 h of shoot and 6 h of root after 10 μM GA₃ treatment. (b) DEGs module. The expression profiles of genes after hormone treatment could be retrieved by entering the gene ID, genome region or gene index. (c) Hormone pathways module. (d) Phylogenetic tree and expression profile of selected genes in hormone pathway. (e) Differential expression of selected genes under seven hormone treatments. (f) Expression comparison of selected genes under different hormone treatments. (g) Hormone crosstalk module. (h) The number of DEGs under hormone treatment.

Many genes involved in hormone metabolism and signalling pathways have been identified in model plants (Li et al., 2017). Therefore, we designed an interactive scalable vector graphics (SVGs) to explore the genes involved in the hormone pathways (Figure 5c). The three types of visualizations were generated to visualize the expression results and performed differential expression analysis and comparison of hormone‐related genes by different hormone treatments side‐by‐side (Figure 5d–f). Users could select ‘hormones’ and ‘other pathways’ from the drop‐down list, and then the annotation of homologues, heatmap, bar chat and line chat was displayed. For each graph, a specific homologue gene and sample could be selected, and its temporal expression profile will be displayed immediately after the selection. We also generated the crosstalk module to display the integration of plant hormone pathways in B. napus (Figure 5g). Taking ‘IAA’ as an example, users could select the gene lists of ‘biosynthesis’, ‘catabolic’ or ‘signalling’ from the drop‐down list, then click the ‘ABA’ button to display the differentially expressed gene numbers and fold change distribution of ABA ‘biosynthesis’, ‘catabolic’ or ‘signalling’ related genes after IAA treatment (Figure 5h). For each DEG gene, users could retrieve and download the gene information at ‘Information of differential expression genes’. Furthermore, we provided the external links in ‘Information of differential expression genes’ to explore the expression profile in 91 tissues of B. napus (Liu et al., 2021), which could explore the candidate variations of DEGs for various traits in B. napus (Yang et al., 2022). The user‐friendly web‐based tools and database presented here will be a valuable resource for future hormone studies, and provide a reference to researchers efficiently exploit hormone response patterns in B. napus.

Discussion

Plant hormones are the key intrinsic factors that control plant ontogeny throughout the whole life cycle. Homeostasis and crosstalk of hormones act in response to various of fundamental biological processes such as development, reproductive, longevity and responses to environmental stimuli (Ciura and Kruk, 2018; Müller and Munné‐Bosch, 2021; Takeuchi et al., 2021). B. napus (AACC, 2n = 38) is derived from a spontaneous hybridization between Brassica rapa (AA, 2n = 20) and Brassica oleracea (CC, 2n = 18; Chalhoub et al., 2014). Phytohormones are also essential for seedling development, organ size regulation, fatty acid biosynthesis and stress responses in B. napus (Tian et al., 2023; Wu et al., 2020; Xiong et al., 2022; Yan et al., 2021). However, few studies have been uncovered the genetic and molecular mechanisms of these hormones crosstalk in controlling B. napus seedling development. Here, we provide a comprehensive insight into how hormones shape B. napus seedling growth, and establish the complex cross‐regulatory relationships between hormone pathways at the transcriptional level. Finally, an online database for exploring the phytochrome transcriptomes has been created, allowing many queries by the annotated gene names.

The highly interconnected web of genes involved in hormone haemostasis in B. napus

Plant growth is shaped by hormonal pathways that not only coordinate the intrinsic developmental programs but also mediate the environmental inputs (Durbak et al., 2012; Verma et al., 2016). Several classes of phytohormones have been characterized, and the hormone haemostasis modulates various growth phenomena. For instance, gibberellins (GAs) strongly influence plant architecture and organ size by regulating cell expansion along the longitudinal axis, and the endogenous GA haemostasis is tightly controlled by its metabolic genes (Davière and Achard, 2013; Sprangers et al., 2020). Mutation of biosynthetic genes (e.g. GA20OX2) or overexpression of catabolic genes (e.g. GA2OX6) leads to dwarfism by reducing endogenous GA levels (Andrew‐Peter‐Leon et al., 2021; Yan et al., 2017). The expression of these GA metabolic genes is also regulated by exogenous GA, whose biosynthesis genes are repressed after GA treatment (Wang et al., 2015). These expression patterns are also observed in B. napus (Figure 2a), suggesting that the conserved mechanisms of GA‐mediated metabolic gene regulation exist in different plant species.

In plants, the metabolites are catalysed by a large number of enzymes encoded by a group of structure genes (Hart, 1987). The expression of these structural genes (e.g. flavonoid biosynthesis) is tightly regulated by hub transcriptional regulators (Ye et al., 2019). Hormone homeostasis can also be reprogrammed by hormone‐responsive transcription factors (Hill, 2015). However, fewer hub transcriptional regulators have been characterized in the regulation of hormone pathways. By constructing the co‐expression networks of relative B. napus hormone metabolic genes, a number of transcriptional regulators responsible for relative hormone metabolic pathways were identified (Figures 3b and S12B,E; Tables S15–S17), and these hub transcriptional regulators could either promote or repress the downstream gene expression (Figure 3f). Notably, a B. napus B3 transcription factor, BnaFUSCA3 (BnaFUS3), repressed the expression of BnaKO (ent‐kaurene oxidase) binding directly to the promoter sequence (Figure 3g). In Arabidopsis, AtFUS3 represses the GA‐biosynthesis gene AtGA3ox2 by physically interacting with two RY elements (CATGCATG) present in the AtGA3ox2 promoter (Curaba et al., 2004), implying that the FUS3‐mediated GA biosynthesis mechanisms are conserved in B. napus and Arabidopsis.

Different phytohormones collectively influence hormone homeostasis in a complex network of synergistic, antagonistic and additive interactions (Depuydt and Hardtke, 2011; Durbak et al., 2012). Here, we show the highly interconnected networks of genes involved in hormone metabolism (Figures 2a and S9A and S13A), which help us to investigate that the treatment with one hormone should lead to changes in the levels of multiple hormones. Auxin, GA and BR have been shown to play a central role in regulating plant growth, in particular by promoting the cell elongation (de Lucas and Prat, 2014; Gonzalez et al., 2012; Perrot‐Rechenmann, 2010). Approximately 21 and 9 BR biosynthesis genes were induced by exogenous auxin and GA respectively (Figure 2a), whereas the expression of 20 and 10 auxin and GA biosynthesis genes was induced by BR (Figure 2a). These results support the idea that auxin and GA regulate BR biosynthesis, and that auxin and GA might also rely on synthesized BRs for their additive growth‐promoting effects in B. napus. We also found that eight and five BR catabolic genes were induced by exogenous auxin and GA respectively (Figure S9A). One explanation for this phenomenon might be that the induction of catabolic genes is essential for maintaining the endogenous BR homeostasis. In addition to the synergistic and additive transcriptional regulation between the hormones, the antagonistic effects between hormones were also observed. GA and ABA mostly play antagonistic roles in plant developmental processes and environmental responses (Liu and Hou, 2018). Indeed, 22 GA biosynthesis genes were repressed by exogenous ABA (Figure 2a), and seven ABA catabolic relative genes were induced by exogenous GA (Figure S9A). Interestingly, FUS suppresses the GA‐biosynthesis genes in B. napus and Arabidopsis (Figure 3f; Curaba et al., 2004); meanwhile, FUS3 could also inhibit germination by increasing ABA levels (Chiu et al., 2012), suggesting that FUS3 might be the central TF in controlling the endogenous ABA/GA ratio. We also observed a few cases where the gene‐expression changes that predicted ambiguous outcomes (both increase and decrease) were induced by the same hormones (Figures 2a and S9A, diamond‐headed arrows). This could be due to the different effects in subsets of cells or tissues, and this could be resolved by using more sensitive techniques (e.g. single cell transcriptomes and metabolomes) to measure the gene expression associated with hormone levels.

The crosstalk and integration of hormone signalling pathways in B. napus

It is likely that the multiple interacting hormone pathways are acquired during growth control have greatly expanded the developmental spectrum of plants, including morphological variation, at both intraspecific and interspecific levels (Depuydt and Hardtke, 2011; Santner et al., 2009; Wolters and Jürgens, 2009). The crosstalk and integration of plant hormone signalling pathways have been illustrated by transcriptional profiling and binary protein–protein interactions in Arabidopsis (Nemhauser et al., 2006). Compared to the protein–protein interaction networks, an unexpectedly small number of genes have been found to be co‐regulated by multiple hormones (Nemhauser et al., 2006). Approximately 163 genes were co‐regulated by seven hormones in B. napus (Figure S13B), suggesting that each hormone acts independently at the transcriptional level by targeting different members of gene families involved in development and environmental responses. Although only a few transcriptional regulators were co‐regulated by the hormones, the co‐expression network of these genes reveals that this group of hub genes might be involved in the crosstalk of hormone signalling pathways in B. napus. Since hormone signalling cascades typically involve feedback regulation by adjusting their own biosynthesis, this could be misinterpreted as regulation of one pathway by the other at the level of transcription.

The existence of a core module that serves as an integration site for many hormone signals has been proposed (Nemhauser et al., 2006). Taking DELLA proteins as an example, the plants carrying mutations in DELLA genes show altered responses to exogenous hormone treatments (Navarro et al., 2008; Oh et al., 2014; Stewart Lilley et al., 2013), and the stability of DELLA proteins are also affected by other hormones (Achard et al., 2003; O'Neill et al., 2010). Despite the lack of a DNA binding domain, DELLA proteins serve as an integration site for other hormone signals by interacting with the transcription factors of other hormone signalling pathways (Bao et al., 2020; Davière and Achard, 2016). Based on the hormone metabolism network, we found that the cytokinin biosynthesis genes IPTs and metabolism genes CKXs were induced and repressed by GA respectively (Figures 4a and S9A). Indeed, the endogenous t‐zeatin levels were increased and suppressed by GA and PAC respectively (Figure 4d), which could be due to the RGA‐related cytokinin homeostasis (Figure 4i). However, the GA‐induced tZ was less accumulated in Bnarga mutants (Figure 4d). One possibility is that the other DELLA proteins (e.g. BnaRGL1, BnaRGL2 and BnaRGL3) might also be involved in the regulation of GA‐induced tZ levels. An antagonistic interaction between GA and cytokinin has been demonstrated in Arabidopsis, where the GA‐signalling suppressor SPY represses GA signalling and promotes cytokinin responses (Eckardt, 2005; Greenboim‐Wainberg et al., 2005). The synergistic and antagonistic interactions between GA and cytokinin in B. napus and Arabidopsis suggest that the original interpretation of crosstalk may need to be revisited in some of these cases, which is essential to investigate both the cellular conditions and the biochemical functions of the gene products in different plant species.

In conclusion, the comprehensive transcriptome atlas of hormone treatment presented in this study provides a valuable resource for uncovering the function of hormone‐responsive genes in B. napus. Coupled with detailed physiological and kinetic analyses, these studies will help us to investigate the hormone crosstalk among different plant species, and ultimately determine whether there are common transcriptional growth modules represented by common target genes. As an allotetraploid crop, B. napus contains a large number of homologous genes, and functional analysis of homologues from the A and C genomes is of considerable importance for molecular manipulation in oilseed rape (Calderwood et al., 2020; Schranz et al., 2006). As the expression patterns and function of homologous genes may have changed (Figure S4; Peng et al., 2015; Shah et al., 2018), our transcriptome dataset provides a valuable resource to study the functional divergence of homologous genes in B. napus. Furthermore, in combination with our tissue expression datasets (Liu et al., 2021), the function of hormone‐related genes in a specific important trait (e.g. seed size) of B. napus could be explored through co‐expression networks. An important next step will be to utilize this resource to guide the B. napus breeding in the future.

Methods

Plant materials and growth condition

ZS11 is a conventional semi‐winter B. napus cultivar. All plant materials, including transgenic plants, were grown in a growth chamber under a 16‐h light/ 8‐h dark photoperiod at 22 °C, with a light intensity of 100 μmol/m²/s.

Total RNA extraction and RNA‐seq library preparation

For the hormone treatments, the ZS11 seeds were placed in a Petri dish with wet filter paper and germinated for 2 days. Well‐germinated seedlings were selected and placed on the floats, which were submerged in the Hoagland nutrient solution container and grown for a further 12 days. The 2‐week‐old seedlings were then transferred to new containers containing Hoagland solution with or without hormones. The final concentration was 10 μm for each hormone. The seedlings were treated with hormones, and the samples were then collected at different timepoints (0, 0.5, 1, 3 and 6 h). The shoots and roots were separated by scissors, and immediately frozen by liquid nitrogen.

For RNA‐seq preparation, the total RNA was extracted using Trizol Reagent (15596026, Invitrogen Life Technologies, Carlsbad, CA, USA) according to the manufacturer's instructions. A total amount of 1.5 μg RNA per sample was used to generate RNA‐seq libraries by the TruSeq RNA Sample Preparation Kit (RS‐122‐2001, Illumina, San Diego, CA, USA). To select cDNA fragments of the preferred length (200 bp), the library fragments were purified using AMPure XP system (Beckman Coulter, Beverly, CA). DNA fragments with adaptor molecules ligated at both ends were selectively enriched using Illumina PCR Primer Cocktail in a 15 cycle PCR reaction. The products were then purified (AMPure XP system) and quantified by the Agilent high‐sensitivity DNA Assay on a Bioanalyzer 2100 system (Agilent, Santa Clara, CA, USA). The library preparations were sequenced on a Hiseq platform (Illumina) by GENOSEQ (Wuhan, China).

Identification and analysis of differentially expressed genes

The quality of the RNA sequencing reads was checked by FastQC (v0.11.9; https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Barcode adaptors and low‐quality reads (read quality < 80 for paired‐end reads) were removed by Trimmomatic (v0.38; Bolger et al., 2014). The filtered reads were then aligned to the B. napus reference genome (ZS11; Song et al., 2020) using Hisat2 (v2.1.0; Kim et al., 2015) with default parameters. Bam files containing aligned reads were inputted into StringTie (v1.3.3b; Pertea et al., 2016) to measure the gene expression levels. Gene‐level raw count data files were generated using featureCounts (v1.6.4; Liao et al., 2014). The raw count data were imported into the Bioconductor package DESeq2 (Love et al., 2014) in the R language to identify the differentially expressed genes. Genes with a log₂‐converted fold change ≥1 or ≤−1 with an FDR (False Discovery Rate) ≤ 0.05 were considered as DEGs. Cluster analysis for DEGs under seven hormone treatments was performed using Short Time‐series Expression Miner (STEM) software with the maximum number of model profiles set to 40 (Ernst and Bar‐Joseph, 2006). The gene ontology (GO) of ZS11 was retrieved from tair (https://arabidopsis.org/index.jsp) according to the homologous genes in BnTIR (Liu et al., 2021). Fisher's exact test was used to test whether the functional categories were overrepresented for the DEGs under seven hormone treatments. The resulting P values were adjusted to Q values using the Benjamini–Hochberg correction and the FDR of 5% was applied.

The construction of hormone crosstalk network

The genes involved in hormone biosynthesis, catabolism and signalling pathways were collected from previously reported (Li et al., 2017). In addition, the keywords of ‘abscisic acid’, ‘ethylene’, ‘brassinosteroid’, ‘gibberellin’, ‘auxin’, ‘jasmonic acid’ and ‘cytokinin’ were used to search for hormone genes from the GO file of ZS11. The highly interconnected web of genes involved in hormones predicts that treatment with one hormone should lead to changes in the levels of multiple hormones. The hormone crosstalk networks were constructed based on the changes in hormone biosynthesis, catabolism and signalling genes. We defined hormone‐responsive genes as those with the absolute of log₂‐converted fold change ≥1 in at least one timepoint or tissue. The up‐regulation effect under hormone treatment was defined as the number of up‐regulated hormone‐responsive genes >1.5 times the number of down‐regulated hormone‐responsive genes. If the difference between up‐regulated and down‐regulated hormone‐responsive genes was <1.5 times, the effect was equivocal. Statistical tests under hormone treatment were then performed using a Fisher's exact test based on the following 2*2 table (Goda et al., 2008). We used the expression changes of ABA biosynthesis genes under IAA treatment as an example. Here, the number of DEGs of seven‐hormone biosynthetic gene under IAA treatment is designated as N ₁ (up‐regulated) and N ₂ (down‐regulated); the number of DEGs of ABA biosynthetic gene under IAA treatment is designated as X ₁, and X ₂ respectively. The significant effect of ABA biosynthesis genes under IAA treatment was generated based on the following 2*2 table:

	Up‐regulated	Down‐regulated
ABA Biosynthesis genes	X 1	X 2
Not ABA Biosynthesis genes	N 1 – X 1	N 2 – X 2

Co‐expression analysis and gene network visualization

The hormone hub genes were defined as the differentially expressed genes under seven hormone treatments, simultaneously. The hormone hub genes were assembled into a ‘guide‐gene set’, which was used to construct the co‐expression network. All hormone genes were defined as a ‘candidate‐gene set’. The connectivity between guide genes and candidate genes was retrieved. The gene pairs were retained only if the absolute value of the Pearson correlation coefficient (PCC) of the gene pairs was >0.5. The co‐expression networks of hormone biosynthesis, catabolism and signalling hub genes were constructed based on the gene pairs. Cytoscape v3.6 software ‘yFiles Organic Layout’ was used to visualize the co‐expression network when the PCC was >0.5 (Shannon et al., 2003).

System architecture and software for database construction

As a module in the BnTIR, the entire web interface of the hormone transcriptome platform was built using HTML5, CSS3, JQuery (http://jquery.com) and Bootstrap (https://getbootstrap.com). The gene expression profile heatmap, and the fold change of DEGs were visualized by Highcharts (https://www.highcharts.com/), ECharts (http://echarts.baidu.com) and plotly.js (https://plotly.com/). The statistical information of the genes was managed by Handsontable (https://handsontable.com/) and DataTables (https://datatables.net). An ‘Electronic Fluorescent Pictograph’ browser was used to explore and analyse the expression or the fold change of DEGs under hormone treatment (Winter et al., 2007). An interactive Scalable Vector Graphics (SVG) was designed to explore the genes involved in the hormone pathway and the crosstalk network.

Plasmid construction and plant transformation

The full‐length CDS BnaA9.CXK2 was amplified by PCR and then cloned into pCambia1300 vector. The pCambia1300‐BnaA9.CXK2 and pMDC162‐TCSn‐GUS vectors were then transformed into B. napus using the standard Agrobacterium‐mediated transformation procedure (Dai et al., 2020). Positive transgenic plants were screened by RT‐qPCR. At least two individual transgenic lines were selected for further studies. All relative primers are listed in Table S23.

For the transient expression assay, the ~1.2 kb promoter of BnaA6.KO, BnaC4.GA20OX4 and BnaA7.CKX2 were individually assembled into pGreenII 0800‐LUC as the reporter, and the CDS of BnaA6.DOF5.7, BnaC9.AGL15, BnaC2.FUS3, BnaA6.RGA and BnaC7.RGA was individually assembled into the vector pGreenII 62‐SK as the effector respectively. For the yeast one‐hybrid assay, the 478 bp (−1070 to −1021) and 222 bp (−400 to −179 bp) promoters of BnaA6.KO were cloned into the vector pAbAi (630491, Clontech, Mountain View, CA, USA) as bait, while the CDS of BnaC2.FUS3 was assembled into the vector pGADT7 (630491, Clontech, Mountain View, CA, USA). All primers used in this study are listed in Table S23.

RNA isolation and quantitative reverse‐transcription PCR

According to the manufacturer's instructions, total RNA was isolated from leaf tissue using an Ultrapure RNA Kit (LS1040, Promega, Madison, WI, USA). Subsequently, 2 μg of total RNA was used to synthesize cDNA using a Transcript RT kit (R223‐01, Vazyme, China). RT‐qPCR was performed as previously described (Yan et al., 2021). The expression level of BnaActin7 was used as an internal standard. All primers used in the RT‐qPCR analysis are listed in Table S24.

Hormone treatment assay

GA₃ (S18001; Shyuanye, China), paclobutrazol (PAC, S18040, Shyuanye, China) and trans‐zeatin (tZ, A600748, Sangon Biotech, China) were first dissolved in ethanol as a stock solution, and then diluted with water to 10 mm. Seedlings were treated with GA₃, PAC, tZ or 1% ethanol (control) for 6 h.

Yeast one‐hybrid assay

The yeast one‐hybrid assay was performed according to the instructions of the Matchmaker Gold Yeast One‐Hybrid Library Screening System (Clontech, Mountain View, CA, USA). Briefly, the recombinant bait plasmids were first linearized by BstBI or BbsI and integrated into the yeast strain Y1HGold using the LiAc method (Wu et al., 2020). The transformed yeast strains were then incubated on the synthetic defined (SD) medium ‐Ura for 2–3 days. Subsequently, the recombinant prey plasmids were transformed into the positive yeast strain (containing the bait genome) and incubated on SD medium ‐Leu ‐Ura. Finally, 100 ng/mL of aureobasidin A (AbA, CA2332, Coolaber, China) was added to the ‐Leu ‐Ura SD medium to assess the protein‐DNA interactions. At least six clones were selected for each experiment.

Electrophoretic mobility shift assay

The BnaC2.FUS3‐His plasmids were first transferred into the Escherichia coli BL21 to express the recombinant protein in the presence of 0.05 mm IPTG at 37 °C for 3 h. The Mag‐Beads His (20515ES08, Yesen, China) was used to purify the His fusion proteins according to the instructions of manufacturer. For the EMSA assay, Cy5‐labelled probes and recombinant proteins were mixed in EMSA/Gel‐Shift binding buffer (GS005, Beyotime, China) at 25 °C for 20 min in the presence or absence of unlabelled competitor DNA. The reaction mixture was then electrophoresed on 6% non‐denaturing polyacrylamide gels under ice water conditions.

Dual‐Luciferase assay

As previously described, the recombinant pGreenII 0800‐LUC and pGreenII 62‐SK plasmids were co‐transformed into Arabidopsis mesophyll protoplasts for dual‐luciferase assay (Huang et al., 2015). The Firefly LUC and Renilla LUC (REN) activities were detected using the Dual‐Luciferase^® Reporter Assay System (E1910, Promega, Madison, WI, USA) according to the manufacturer's instructions. The ratio of LUC/REN was calculated to normalize the LUC activity. At least three independent experiments were performed for each assay.

The GUS staining assay

The GUS staining was performed using a GUS staining kit (SL7160, Coolaber, China). The T₂ generation lines of the TCSn‐GUS report lines were used for the GUS staining assay. Roots were collected from 4‐day‐old TCSn‐GUS lines and then stored in cold 90% (v/v) acetone solution for 1 h. Subsequently, GUS staining buffer was then added to replace the acetone and placed under vacuum for 30 min. After all the tissues were immersed, the samples were incubated at 37 °C for 12 h, followed by rinsing with a graded series of ethanol (40%, 60%, 80% and 100%) for 2 h until completely clear for photography. Two individual T₂ transgenic lines (L3 and L6), with at least 15 plants were used for the GUS staining assay.

Quantification of hormone levels

One hundred milligrams of lyophilized samples was extracted with 1 mL of 70% methanol (A452‐4, Fisher Scientific, Waltham, MA, USA) at 4 °C. After centrifugation at 13 800 g for 10 min, the supernatant was aspirated and filtered (0.22‐μm pore size) for LC–MS analysis. The hormone concentration was detected by an LC‐ESI‐MS/MS system, and the analytical conditions were performed as previously described (Chen et al., 2013). The standard of GA₃ and trans‐Zeatin were detected to draw the standard curve (concentration: 0.5, 1, 2, 5, 10, 20, 50, 100 ng/mL respectively) before the determination of each batch. At least three independent experiments were performed for each assay.

Statistical analyses

Statistical analysis was performed to identify significant between genotypes, using Student's t‐test, at P values <0.05 or <0. 01.

Conflict of interest

The authors declare no conflict of interests.

Authors' contributions

C.D., Q.Y. and L.G. designed the research. D.L., G.Y., S.L. and L. Y. performed the experiments. G.Y., D.L. and W.L. analysed and visualized the data. C.D., Q.Y., D.L. and G.Y. wrote the manuscript. S.W. developed the web interface for data visualization. All authors read and approved the manuscript.

Supporting information

Figure S1 Correlation between biological replications, four materials were randomly selected (R range from 0.83 to 0.99).

Figure S2 The expression level of ABA, ETH, BR, GA, IAA and tZ marker genes under ABA, ETH, BR, GA, IAA and tZ treatments.

Figure S3 The heatmap shows the expression profiles of marker genes of ABA, ETH, BR, GA, IAA, tZ and MeJA under ABA, ETH, BR, GA, IAA and tZ and MeJA treatments.

Figure S4 The percent of differentially expressed homologous gene pair.

Figure S5 The heatmap shows the expression profile of marker genes of IAA (A) and GA (B) in different tissues.

Figure S6 The number of DEGs at each timepoint under different hormone treatments.

Figure S7 GO enrichment of DEGs under different hormone treatments.

Figure S8 DEGs under seven hormone treatments, simultaneously.

Figure S9 Crosstalk network of hormone catabolism genes.

Figure S10 Relative expression levels of the BnaCKX2 in 14 days seedlings of WT and BnaA9.CKX2‐OE (L1 and L2) analysed by RT‐qPCR.

Figure S11 Yeast one‐hybrid assay and EMSA analysis of interactive effects and recombinant between BnaA6.KO promoter and BnaC2.FUS3.

Figure S12 Identification of key transcription factors regulating auxin and ethylene biosynthetic.

Figure S13 Crosstalk network of hormone signal genes.

Figure S14 The heatmap shows the expression profile of gibberellin genes.

Figure S15 Expression profile of hormones marker genes.

Table S1 Summary of accessions for RNA sequencing.

Table S2 Correlation between biological replicates.

Table S3 Marker genes of different hormones.

Table S4 Expression profiles of marker genes.

Table S5 Up‐regulated DEGs at different timepoints with different hormone treatments.

Table S6 Down‐regulated DEGs at different timepoints with different hormone treatments.

Table S7 ‘Up’ indicates genes that have significantly higher signal under treated conditions for at least one timepoint and do not have significantly lower signal at any timepoints.

Table S8 The number of DEGs at different timepoints with different hormone treatments.

Table S9 Expression profiles of DEGs under seven hormone treatments, simultaneously.

Table S10 Summary of hormone biosynthetic, catabolism and signal genes.

Table S11 The significance level of responsive type of biosynthesis genes under different hormone treatments.

Table S12 The significance level of responsive type of catabolism genes under different hormone treatments.

Table S13 Twenty‐five hormone biosynthetic genes were differentially expressed under seven hormone treatments, simultaneously.

Table S14 Twenty‐four hormone catabolism genes were differentially expressed under seven hormone treatments, simultaneously.

Table S15 Co‐expression genes with Bna.ACO2 in B. napus.

Table S16 Co‐expression transcription factors with GA biosynthesis genes in B. napus.

Table S17 Co‐expression transcription factors with IAA biosynthesis genes in B. napus.

Table S18 Co‐expression transcription factors with ETH biosynthesis genes in B. napus.

Table S19 The significance level of responsive type of signal genes under different hormone treatments.

Table S20 Hormone catabolism genes were differentially expressed under seven hormone treatments, simultaneously.

Table S21 Co‐expression genes with Bna.IAA26 in B. napus.

Table S22 Expression changes of cytokinin biosynthetic, catabolism and signal genes under GA treatment.

Table S23 Expression changes in gibberellin biosynthetic, catabolism and signal genes under tZ treatment.

Table S24 Primers used for making constructs.

Table S25 Primer used for RT‐qPCR analyses.

Data availability

All transcriptome data were available at BnTIR (http://yanglab.hzau.edu.cn/BnTIR).