Vernalization Promotes GA-Mediated Bolting Initiation via the Inhibition of ABA and JA Biosynthesis

Abstract

To understand the relationship between vernalization and endogenous phytohormones on bolting, phytohormone levels and transcriptome changes were measured by LC-MS/MS and RNA sequencing before vernalization (CK), at the end of 16 weeks of vernalization (Vel) and at 5 days after vernalization (Re). A total of 32 phytohormone compounds significantly changed after vernalization; especially, the content of abscisic acid (ABA) and jasmonic acid (JA) was dramatically decreased more than sixteen and three times in the Vel and Re samples. In addition, GA₁₉ accumulated after vernalization, while the content of GA₅₃ and GA₂₀ decreased. A total of 7471 differentially expressed genes (DEGs) were identified in response to vernalization. These DEGs were enriched with GO terms including response to stimulus (GO:0050896), response to hormone (GO:0009725) and shoot system development (GO:0048367). KEGG mapping indicated that 16, 13 and 11 DEGs were involved in gibberellic acid (GA), ABA and JA biosynthesis, suggesting a negative role of vernalization in phytohormone biosynthesis. Vernalization also repressed the expression of BvABFs and BvMYC2s, implying the inhibition of ABA and JA signalling. Additionally, vernalization plays a positive role in GA signalling, owing to the down-regulation of BvDELLAs. We also found that GA₃-induced bolting could be retarded by exogenous ABA and methyl jasmonate (MeJA). In total, our results suggest that vernalization can promote GA-induced bolting by decreasing BvDELLA repressors of GA signalling and eliminating the antagonistic effects of ABA and JA.

1. Introduction

Sugar beet (Beta vulgaris L.) belongs to the Amaranthaceae and contributes approximately 30% of the global sugar produced [1]. As a typical biennial crop, the yield and sugar content of sugar beet are highly dependent on the duration of vegetative growth. After vernalization, the translocation of sugar from the taproots to the shoots through the phloem meets the energy consumption of bolting and flowering and results in lower sugar production [2]. Thus, understanding the mechanisms of vernalization is important for early sowing in spring and achieving autumn-sown beets. In sugar beet, several genes, including BvBTC1, BvFT1, BvFT2 and BvBBX19, have been revealed to be involved in the vernalization pathway [3]. BvBTC1 was cloned from bolting locus B and identified as an essential component of the vernalization pathway owing to an insertion mutation located in the promoter region that determines whether sugar beets are biennial [4]. After vernalization, the down- and up-regulation of BvFT1 and BvFT2, respectively, are controlled by BvBTC1 for the initiation of bolting [4]. BvFT1 and BvFT2 are an antagonistic pair of genes that are homologous to AtFT, which acts as a floral integrator gene in Arabidopsis [5]. In addition, BvBBX19 was identified from bolting locus B2 and cooperates with BvBTC1 to regulate BvFT1 and BvFT2 [3].

In addition to the vernalization pathway, the presence of phytohormones is closely related to bolting and flowering initiation. Langridge (1957) first reported that applications of gibberellic acid (GA) overcome late flowering caused by short days in Arabidopsis [6]. Subsequently, the promotion of bolting and flowering by applications of GA were described in various species, such as sugar beet, radish (Raphanus sativus), lettuce (Lactuca sativa) and cabbage (Brassica oleracea var. capitata) [7,8,9,10]. By observing the flowering phenotype in mutants with a loss of function of many components of GA biosynthesis and signalling transduction, a GID1-DELLA-FT/SOC module has also been identified in the GA pathway that controls the flowering time in Arabidopsis thaliana [11,12,13]. Although abscisic acid (ABA) is well known as a stress-related phytohormone, several studies suggest a role of ABA in the regulation of flowering [14,15]. AtABI4 and AtABI5, which are key factors in ABA signalling, recognize the core element G-box on the promoter of the flowering inhibitor AtFLC and promote its transcriptional activation [14,15,16]. The abi3 and abi4 mutants show an early flowering phenotype, while the overexpression of AtABI4 results in a late flowering phenotype [14,15]. ABA also inhibits GA biosynthesis by up-regulating negative regulators of GA signalling, namely AtSPY and AtRGL2 [17]. Additionally, jasmonic acid (JA) and indole-3-acetic acid (IAA) have been found to be antagonistic with GA in the control of flowering time. The crosstalk between JA and GA primarily focuses on the interaction between DELLA and JAZ1 that facilitates the binding of MYC2 to the promoter of flowering genes [18]. Thus, the overexpression of AtJAZ1 in Arabidopsis results in early flowering compared with the wild type [19]. The GA-induced degradation of DELLA repressor proteins also decreases JA accumulation and the expression of DAD1 and increases the expression of MYB and elongation of the filament growth [20]. IAA has been hypothesized to inhibit flowering in morning glory (Pharbitis nil) by enhancing the biosynthesis of ethylene, which can be partially overcome by GA [21]. GA metabolic enzymes, such as gibberellin-20-oxidase (AtGA20ox) and gibberellin-3-oxidase (AtGA3ox), and the stability of DELLA proteins are also regulated by auxin signalling [22,23].

Previous research suggests that the crosstalk between vernalization and the GA pathway co-regulate bolting and flowering. The requirement of the vernalization time is shortened by the application of GA₃ in radish, although treatment with GA₃ did not overcome the need for a period of cooling [24]. The promotion of GA biosynthesis by vernalization has also been observed in other species. The accumulation of kaurenoic acid, which is a precursor to GA, and an increase in the activity of kaurene oxidase and kaurenoic acid hydroxylase has also been detected in vernalized field pennycress (Thlaspi arvense) [25]. An increase of two- to twenty-five-fold of several GA compounds, including GA₁, GA₃, GA₈, GA₁₉ and GA₂₀, have been studied in winter canola (Brassica napus L.) at the end of the vernalization treatment [26]. The relationship between GA and vernalization in the regulation of bolting is also supported by a statistical analysis of bolting characterization in sugar beet [27]. In Arabidopsis, the relationship between GA and vernalization involves the targeting of several genes involved in GA metabolism and signalling by AtFLC on the basis of an analysis of the genome-wide binding sites [28]. Moreover, DELLAs directly interact with AtFLC and enhance its negative regulation of downstream genes, including AtFT [29]. AtFLC is an important flowering inhibitor of the vernalization pathway that inhibits flowering by repressing the downstream flowering genes [30]. However, in sugar beet, the expression of BvFLC is not like that of AtFLC, which gradually decreases during vernalization and maintains a low level after temperature increase. In RNAi and the overexpression of BvFLC transgenic beets, there have been no significant changes observed in the requirement for vernalization and the expression of BvFT1 and BvFT2 [31]. These results suggest that BvFLC is not the primary regulator of vernalization in sugar beet. How the vernalization cooperates with the GA pathways in the regulation of bolting and flowering in sugar beet remains unclear. Some phytohormones, such as ABA and JA, are known as stress-responsive phytohormones and can crosstalk with GA and modulate the plant growth and developmental stages [32]. How does the prolonged low temperature of vernalization affect the content of phytohormones in sugar beet and do these changes affect GA-induced bolting in sugar beet? To answer these questions, changes in the levels of phytohormones and gene expression affected by vernalization in sugar beet were analyzed by liquid chromatography–mass spectrometry/mass spectrometry (LC-MS/MS) and RNA sequencing, respectively. Additionally, the exogenous applications of phytohormones were designed to test the effect of phytohormones on the occurrence of bolting.

2. Materials and Methods

2.1. Plant Materials and Vernalization Treatment

KWS9147, a bolting insensitive commercial variety of sugar beet, was selected for this study. Seeds were sown in a field at Gaomi, Shangdong, China (N: 36.2° E: 119.4°). After 3 months of growth, taproots that were 3 to 5 cm diameter were harvested and transplanted into pots (12 cm × 12 cm × 10 cm) that contained chernozem. The plants were then transferred to a culture room with 24 ± 2 °C, 200 μmol·m⁻²·s⁻¹ light intensity and 16 h:8 h light/dark. After 2 weeks of restoration, the plants were transferred into a cold cellar that was 4 ± 2 °C, 200 μmol·m⁻²·s⁻¹ light intensity and 16 h:8 h light/dark for 16 weeks of vernalization. After vernalization, the plants were transferred back to the culture room, and the rate of bolting was recorded daily. The tissues in shoot apexes without young leaves were sampled before vernalization (CK) and at the end of 16 weeks vernalization (Vel). To assess memory of vernalization, tissues at 5 days after vernalization (Re) were also sampled. All samples were stored at −80 °C to determine the contents of phytohormones and conduct ssRNA-seq and qRT-PCR. Three biological replicates were performed for all treatments.

2.2. RNA Extraction and qRT-PCR

The RNA was extracted, and real-time quantitative reverse transcription PCR (qRT-PCR) was performed as previously described [33]. Briefly, the frozen samples were fully ground into powder, and the total RNA was extracted using an RNA-easy isolation reagent (Vazyme, Nanjing, China). RNA purity and concentration was checked on an agarose gel and quantified by NanoDrop 2000c (Thermo Fisher Scientific, Waltham, MA, USA).Genomic DNA digestion and cDNA synthesis were performed using a PrimeScript RT Reagent Kit (Takara, Dalian, China) according to the manufacturer’s instructions. Three biological replicates were set up for each treatment. Floral activators, such as BvSOC1, BvBTC1 and BvFT2, were selected as markers to assess the vernalization response. Some genes involved in GA, ABA and JA biosynthesis and signalling were also selected for validation of RNA-seq. BvGAPDH was used as the reference gene to normalize the levels of expression [30]. The specific primers (Table S1) were designed by Primer-BLAST along with restrictions as previously described [33]. The requirements for primer design were as follows: (1) Tm values of primers were restricted at 60 ± 1 °C, (2) the product length was limited from 150 to 300 bp, (3) at least one primer spans two exons (3′-UTR for genes with signal exon) and (4) there was no nonspecific amplification detected in B. vulgaris. All the qRT-PCR reactions were performed using Fast qPCR Master Mixture (DiNing, Beijing, China) and a QuantStudio 1 plus instrument (Applied Biosystems, Foster City, CA, USA). A one-way analysis of variance (ANOVA) and fold change were applied and calculated to determine significant differences.

2.3. Determination of Phytohormone Contents

A total of 50 mg frozen powder, which was ground as described above, was dissolved and extracted with 1 mL of methanol/water/formic acid (15:4:1, v/v/v). In each extract, 10 μL of a 100 ng/mL internal standard mixed solution that contained 88 types of phytohormone compounds was added for quantification (Table S2). After 10 min of vortexing, the mixture was centrifuged at 12,000 g for 5 min at 4 °C. The supernatant was evaporated to dryness under a stream of nitrogen gas and then dissolved in 100 μL of 80% methanol (v/v). After membrane filtration with a pore size of 0.22 μm, 2 μL samples were uploaded onto an LC-ESI-MS/MS system (UPLC, ExionLC AD; ESI-MS/MS, Triple Quad 6500). A Waters ACQUITY UPLC HSS T3 system (Waters, Milford, MA, USA) was used for chromatography with a C18 column (1.8 µm, 100 mm × 2.1 mm) with a 12 min gradient at 0.35 mL/min. The gradient program started at 5% acetonitrile with 0.04% acetic acid (0–1 min), increased to 95% acetonitrile with 0.04% acetic acid (1–8 min), was maintained at 95% acetonitrile with 0.04% acetic acid (8–9 min) and finally ramped back to 5% acetonitrile with 0.04% acetic acid (9–12 min). ESI-MS/MS was operated in positive and negative ion modes and controlled by Analyst software version 1.6.3 (AB SCIEX, Foster City, CA, USA). The ESI source operation parameters were set as follows: source temperature 550 °C; ion spray voltage (IS) 5500 V for the positive ion model, −4500 V for the negative ion model and curtain gas (CUR) was set at 35 psi, respectively. The data of all the phytohormones were acquired and quantified by Metware Biotechnology Co., Ltd. (Wuhan, China, http://www.metware.cn/ (accessed on 31 March 2023)). The contents of phytohormones were analyzed by a one-way ANOVA followed by the Duncan test to make comparisons among the treatments (p < 0.05).

2.4. RNA-Seq and Differential Expression Analysis

A total of 1 µg RNA per sample was used to construct the library. Sequencing libraries were prepared using an NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biosystems, Ipswich, MA, USA) following the manufacturer’s recommendations, including mRNA purification by poly-T oligo attached magnetic beads, synthesis of the first and second cDNA strands, end prep of the DNA fragments, adaptor ligation, purification of the DNA fragments with AMPure XP Beads (Beckman Coulter, Brea, CA, USA), PCR enrichment and the purification of PCR products. The quality of the library was assessed using an Agilent Bioanalyzer 2100 system (Agilent Technologies, Santa Clara, CA, USA). The index-coded samples were clustered by a cBot Cluster Generation System using a TruSeq PE Cluster Kitv3-cBot-HS (Illumina, San Diego, CA, USA) according to the manufacturer’s instructions. An Illumina HiSeq platform was used to sequence the library preparations according to a 150 bp paired-end sequencing protocol. Three biological replicates were set up for each treatment.

To obtain clean reads, reads containing adapters, poly-N and low-quality reads were filtered and removed from raw reads using FASTP (version 0.19.3). The Q20, Q30, GC-content and sequence duplication level of the high-quality clean reads were calculated. HISAT (v2.1.0) was utilized to map clean reads to the B. vulgaris reference genome, which was downloaded from Phytozome v13. Transcript assembly was followed by StringTie (v1.3.4d). The fragments per kilobase of transcript per million fragments mapped (FPKM) of the assembled transcripts were calculated and normalized by featureCounts (v1.6.2) in accordance with the global normalization parameters. Subsequently, the Benjamini–Hochberg corrected p-values and fold change among treatments were calculated using DESeq2. Genes that met the corrected p-value < 0.05 and fold change > 2 were considered to be differentially expressed genes (DEGs).

2.5. Bioinformatics

The gene ontology (GO) enrichment analysis was performed by TBtools according to the hypergeometric test [34]. The GO reference of sugar beet was downloaded from Dicots PLAZA 5.0 [35]. All the genes and phytohormone compounds were mapped into the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Phytohormone metabolism and signalling pathways were referred from the KEGG pathway maps and transcriptional regulatory network as described by Bao et al. [12]. The FactoMine, pheatmap and Cor functions in the R language were applied for principal component analysis (PCA), hierarchical clustering and the calculation of correlation coefficients [36].

2.6. Exogenous Phytohormones Treated for Sugar Beet Seedlings

To assess the inhibition of ABA and JA for GA-induced bolting, the effect of different combinations of GA₃, ABA and methyl jasmonate (MeJA) on bolting was observed. KWS9147 seeds were sown in pots with vermiculite. Cultivation conditions were the same as described in Section 2.1. After germination, seedlings were transferred to containers with Hoagland nutrient solution at pH 5.8. Exogenous phytohormones were dropped onto stem tips from 5 days after transplanting. Seedlings treated with 1 mM GA₃ were set as control. GA₃ mixed with 0.1 and 1 mM ABA and MeJA were used for treatment. A total of 20 μL phytohormones mixture were dropped daily. Each treatment had 12 seedlings. The number of bolting seedlings were counted every day. The length of bolting stem was measured by ImageJ at 15 days after treatment (DAT).

3. Results

3.1. Changes in Bolting-Related Gene Expression and the Bolting Rate after Vernalization

The process of material cultivation and sampling is shown in Figure 1A. The beet roots grew to approximately 3−5 cm in diameter after 12 weeks of field cultivation. After transplanting into pots, the beets returned to normal growth within 2 weeks, and the shoot apexes were then sampled for subsequent measurements. The growth was strictly retarded by vernalization, but the survival rate was >95%. Vernalized plants restored their growth within 2 days after they were transferred back to room temperature. Visible bolting initiation was observed after 14 days of growth. The bolting rate increased gradually and reached 91.7% at 25 days after transfer (Figure 1B). qRT-PCR showed that the expression of bolting activators, including BvBTC1, BvSOC1 and BvFT2, was significantly up-regulated after 16 weeks of vernalization (Figure 1C–E). BvBTC1 and BvSOC1 continued to be up-regulated during the post-vernalization period. Although the level of expression of BvFT2 was down-regulated after the transfer back to room temperature, the level of expression was still higher than that before vernalization. These results suggest that 16 weeks of vernalization were sufficient for the induction of bolting.

3.2. The Effect of Vernalization on the Levels of Phytohormone

A total of 39 compounds that included ABA, auxin, cytokinin (CTK), ethylene (ETH), GA, JA and salicylic acid (SA) were successfully detected in sugar beets (Table S2). Among them, 32 compounds, including two types of ABA, 11 auxin, nine CTK, one ETH, four GA, five JA and one SA, changed significantly in Vel and Re (Figure 2A). Based on these changes in content, five groups were classified by hierarchical clustering (Figure 2B). Only 2-oxindole-3-acetic acid (OxIAA) was classified in group A, which was only reduced in the Re samples. Group B and C had 11 and 15 compounds, respectively, and their contents significantly decreased after vernalization and were reduced more or stayed at a continuous low level after the plants were transferred back to room temperature. 1-Aminocyclopropane-1-carboxylate (ACC) and three CTK compounds (2MeScZR, DHZROG and DZ) significantly accumulated in the Re samples and were classified into group D. Only two phytohormones, namely GA₁₉ and iP7G, accumulated after vernalization and were maintained at high levels in the Re samples. In terms of phytohormone classes, a continuous reduction in the contents of auxin, ABA, JA and SA was caused by vernalization. Different members of GA and CTK showed contrasting changes in response to vernalization. Additionally, the accumulation of ETH was induced by vernalization when the ambient temperature increased to room temperature.

Since GA-induced bolting initiation was described in previous studies, we focused on changes in the contents of GAs (Figure 2C–E) [7,8,9,10]. The content of GA₁₉ significantly increased under Vel and Re treatment, while changes in GA₅₃ were the opposite. GA₂₀ content was lower in sugar beet compared with GA₁₉ and GA₅₃. Approximately 0.51 to 0.97 ng/g FW of GA₂₀ was detected in the three replicates of CK, while the signal intensity was missing in the Vel and Re samples. This was probably because the GA₂₀ levels in the Vel and Re samples had dropped below the detection range of the instrument. Among the GAs detected, GA₁₉ (approximately 2.05 to 4.15 ng/g FW) and GA₅₃ (approximately 0.43 to 3.09 ng/g FW) comprised more than 77% of the total GA. They remained consistent with the percentage of total GA in the CK, Vel and Re samples. In addition to the GAs, the change in ABA content was worthy of note owing to the crosstalk between GA and ABA. A dramatic decrease in the ABA contents, measuring 180.59 ng/g fresh weight (FW) in the CK samples and decreasing to 11.11 and 5.27 ng/g FW in the Vel and Re samples, respectively, were detected and this led to a high rate of GA and ABA after vernalization (Figure 2F). In addition, JA-related compounds, which were antagonistic with the GAs, also significantly decreased after vernalization. Approximately 30.76 ng/g FW JA was measured in the CK samples, but none was detected in the Vel samples (Figure 2G). An increase in the JA content (approximately 9.06 ng/g FW) was found in the Re samples, but the content was still far less than that in the CK samples. Similarly, JA-Ile, which is the primary bioactive JA compound in plants, was only detected in the CK samples (Figure 2H).

3.3. Transcriptome Changes in the Shoot Apex of Sugar Beet during Vernalization

After Illumina sequencing, more than 6 Gb of clean data were obtained from each library. The Q30 of the base ratio >92% and <89% of the clean reads were mapped to the EL10 reference genome. After the assembly and annotation of the transcripts using StringTie, 20,362 annotated and 4863 novel genes were detected (Table S3). The FPKM value was calculated and used for the PCA to evaluate the reproducibility of the data. A plot of two dimensions presented high reproducibility for our sequencing data (Figure 3A). High reproducibility was also suggested by a correlation analysis, which indicated that the correlation coefficient between replicates within the treatments was >0.97 (Figure 3B). Subsequently, DEGs were identified based on the common criteria (fold change > 2 and p value < 0.05). A total of 7471 DEGs were identified among the CK, Vel and Re treatments. There were 1392 and 2923 DEGs up-regulated and down-regulated, respectively, after 16 weeks of vernalization (Figure 3C). Subsequently, the levels of expression of 2034 and 1952 DEGs were up-regulated and down-regulated at room temperature for 5 days, respectively (Figure 3C). Under the CK versus Re treatment, we identified 1464 and 2312 DEGs whose expression was up- and down-regulated, respectively (Figure 3C). A total of 2586 DEGs were significantly changed after vernalization and memorized this significant change post-vernalization, of which 1876 genes remained unchanged between Vel and Re samples (Figure 3D). The other 710 DEGs were significantly different among the three treatments. These results showed that the transcriptome changes were caused by vernalization and not restored with the increase in temperature.

To understand the biological processes associated with vernalization, GO terms were enriched based on the functional annotation of DEGs. A total of 519, 633 and 455 GO terms were enriched in the CK versus Vel, CK versus Re and Re versus Vel, respectively (Figure 4). As expected, hundreds of DEGs (262 to 1030 genes) were annotated in the stress-related GO terms, including response to stimulus (GO:0050896), external stimulus (GO:0009605), abiotic stimulus (GO:0009628), chemical (GO:0042221) and organic substance (GO:0010033). In terms of development, the enrichment of shoot system development (GO:0048367) was detected in the three comparisons. Flowering-related terms (GO:0009908, GO:0048575 and GO:0048579) were also enriched in the CK versus Vel, CK versus Re and Re versus Vel, respectively. Additionally, DEGs primarily participated in the response to hormone (GO:0009725) with an FDR that was 1.31E-7, 1.67E-14 and 8.36E-11 in the CK versus Vel, CK versus Re and Re versus Vel, respectively. We also found that several hormone-related GO terms, such as the gibberellin metabolic process (GO:0009685) and the response to abscisic acid (GO:0009737), were enriched after vernalization. These results suggested that the genes involved in phytohormone metabolism and signalling were dramatically regulated by vernalization.

3.4. The Change of DEGs Involved in Phytohormone Metabolism

The KEGG mapping indicated there were 33 and 39 genes in the sugar beet genome that were identified and involved in GA and ABA metabolism, respectively (Figure 5 and Table S4). Among them, a total of 16 and 13 DEGs were affected by vernalization, respectively. BvCrtB and BvKS catalyzed the production of phytoene and ent-kaurene, respectively, from geranylgeranyl diphosphate (GGPP), which are precursors shared by both GA and ABA biosynthesis. A significant down-regulation of BvCrtB was observed in the Re samples, although the level of expression was increased (not in excess of two-fold) during vernalization. In contrast, six of the eight BvKS genes were significantly up-regulated in the Vel or Re samples. Several genes (BvCrtZ, BvNCED and BvABA2) that participate in multiple steps of ABA biosynthesis were repressed by vernalization, and the decrease in their expression was sustained in post-vernalization conditions. It is noteworthy that the down-regulated BvCrtZ was a single-copy gene in the sugar beet genome. In GA biosynthesis, BvKO and BvKAO, which encode proteins that catalyze the transformation of ent-kaurene into GA₁₂, were repressed by vernalization. However, the levels of expression of BvKO and BvKAO increased post-vernalization and were significantly up-regulated compared with the CK samples. BvGA20ox encodes a key oxidase enzyme that catalyzes the conversion of GA₁₂ and GA₅₃ to GA₉ and GA₂₀, respectively. The down-regulation of BvGA20ox was detected in both the Vel and Re samples. This was consistent with the decrease in GA₉ and GA₂₀ and the accumulation of an intermediate product (GA₁₉). We also detected the down-regulation of BvCYP707 and BvGA2ox in both the Vel and Re samples, implying that the decrease in contents of GA and ABA was not owing to the degradation and inactivation of the pathway. In JA biosynthesis, a total of 28 genes from the sugar beet genome were mapped through the KEGG database. BvLOX2S, BvAOS, BvAOC, BvOPR and BvACCA1 that were implicated in multiple steps of the JA biosynthesis pathway were found to be down-regulated in response to vernalization. This was consistent with the reduction of intermediate metabolites, such as OPDA and OPC-4, and end products, such as JA, JA-Ile and JA-Val (Table S2). In total, vernalization presented a negative role in the biosynthesis of ABA, JA and GA at the transcriptional level. The inhibition remained for several days after the end of vernalization.

3.5. The Change of DEGs Involved in Phytohormone Signalling

Approximately 50% of the GA, ABA and JA signalling-related genes were also affected by vernalization (Figure 6 and Table S4). There were one and seven BvGID1s detected that were up-regulated in the Vel and Re samples, respectively. In contrast, most of the BvDELLAs were down-regulated in the Vel samples, suggesting the activation of GA signalling. In ABA signalling, two BvSnRK2s and three BvPP2Cs were increased and decreased in the Vel samples, respectively, which promoted the activation of BvABFs. However, some BvSnRK2s and BvPP2Cs presented an opposite trend in response to vernalization. Most expressions of BvABFs were down-regulated, disturbing ABA-regulated gene expression. It is suggested that vernalization disturbed ABA signalling by the repression of BvABFs, but not the phosphorylation-dependent switch in regulators. Similarly, vernalization played a negative role in JA signalling at the transcriptional level, owing to the down-regulation of many types of BvMYC2. In total, vernalization could promote GA signalling and inhibit ABA and JA signalling based on the transcriptional changes in key regulators, including BvDELLA, BvABF and BvMYC2.

3.6. qRT-PCR-Based Experimental Validation of Select Genes Involved in ABA, JA and GA Biosynthesis and Signaling

To verify the transcriptome profiling results, qRT-PCR was employed to determine the expression levels of a representative group of genes involved in ABA, JA and GA biosynthesis and signalling (Figure 7). As mentioned above, BvNCED and BvLOX participate in key steps of ABA and JA biosynthesis. A significant decrease in the expression of BvNCED and BvLOX was detected by both qRT-PCR and ssRNA-seq. The results of ssRNA-seq show the significant down-regulation of BvGA20ox in Vel and Re treatment, while significant down-regulation was only identified by qRT-PCR post-vernalization. In terms of the signalling pathway, both qRT-PCR and ssRNA-seq revealed that two BvDELLAs were significantly repressed by vernalization. Especially, the significant down-regulation of BvDELLA-2 was identified in the Vel and Re treatments. Vernalization also affected the expression of BvABF, which is a key regulator in ABA signaling. A continuous decrease in the expression of BvABF was detected under vernalization and post-vernalization by ssRNA-seq and qRT-PCR. In JA signaling, BvJAZ, BvMYC2-1 and BvMYC2-2 were repressed by vernalization. The expression pattern under vernalization and post-vernalization was validated by ssRNA-seq and qRT-PCR. In summary, most of the expression patterns of the selected genes detected by qRT-PCR were consistent with those revealed by ssRNA-seq, although the expression fold change was not strictly consistent for all genes.

3.7. Bolting Traits of Seedlings Treated by Combinations of GA₃, ABA and MeJA

To verify that the ABA and JA signalling pathways inhibit GA-induced bolting, GA₃ mixed with different concentrations of ABA and MeJA was dropped onto sugar beet seedlings. Sugar beet seedlings treated with only GA₃ were set as a control. Exogenous ABA and MeJA could result in late bolting, although GA-induced bolting was not completely inhibited (Figure 8A). The average bolting stem length was 3.427 cm after GA₃ treatment (Figure 8B). The length of the bolting stem length significantly decreased to about 26.50% and 52.12% after an additional 0.1 mM and 1 mM ABA treatment. Obviously short bolting stems were not detected in the 0.1 mM MeJA treatment, while a high concentration of MeJA caused about a 30.70% reduction in bolting stem length. Besides bolting stem length, ABA and MeJA delayed GA-induced bolting initiation. The number of bolts in the ABA and MeJA treatment were less than that of the control at 7 d after treatment. All sugar beet seedlings were bolting at 8 d after exogenous GA₃ treatment, while bolting in some of seedlings treated with ABA and MeJA was delayed by 1 to 2 days (Figure 8C,D).

4. Discussion

The developmental transition from a vegetative phase to reproduction is complexly regulated by external factors, such as vernalization and photoperiod, and internal factors, such as changes in phytohormones [37]. A clue to the linking of external and internal factors was that vernalization and photoperiod typically affected the level of endogenous phytohormones. In this study, LC-MS/MS and RNA sequencing were applied for understanding change in phytohormone content and the expression of genes involved in phytohormone biosynthesis and signalling. Reductions in the contents of 26 endogenous phytohormones were detected. More than 40 DEGs involved in multiple steps of phytohormones biosynthesis and signalling were changed at the transcriptional level. These results showed that vernalization suppressed the biosynthesis of phytohormones, especially ABA and JA, alleviating the inhibitory effect on GA-induced bolting. GA signalling was transcriptionally activated by vernalization, although GA biosynthesis was partially repressed. In addition, most of the changes in phytohormone content and gene expression persisted for several days after vernalization. These memories of repression of ABA and JA biosynthesis and the activation of GA signalling at a transcriptional level was consistent with the traits of vernalization.

4.1. Vernalization Suppressed Biosynthesis of ABA and JA, Alleviating the Inhibitory Effect on GA-Induced Bolting

Several studies have reported that the content of endogenous phytohormones, such as ABA and MeJA, is changed by low temperature and contributes to the regulation of bolting and flowering [38,39,40]. TaABA8’OH2 and TaLOX2S, which play central roles in ABA and JA metabolism, respectively, were also identified as a target of vernalization-related regulator (TaVRN1) by chromatin immunoprecipitation and sequencing [41]. As one of the key phytohormones involved in stress resistance, the accumulation of ABA was commonly found in plants under short-term low temperatures. However, the change of ABA content was infrequently noticed in long-term low temperatures, namely vernalization. In this study, a more than 16-fold decrease in ABA levels was detected after vernalization. The levels of expression of BvCrtZ, BvZEP, BvNCED and BvABA2, which are all nearly implicated during the entire pathway of ABA biosynthesis, were repressed by vernalization and kept at low FPKM in post-vernalization conditions. Additionally, BvCYP707A, which were mainly genes contributing to ABA degradation, was down-regulated rather than up-regulated by vernalization. Therefore, the significant decrease in ABA content might be attributed to the inhibition of biosynthesis. In ABA signalling, the effect of vernalization on the expression of BvPP2Cs and BvSnRK2s, which derived reversible phosphorylation for switching the activity of ABF transcription factors, did not coincide [42]. However, vernalization directly repressed the expression of BvABFs at a transcriptional level. Considering the decline in ABA content together, ABA response might be impaired after vernalization. Further, we hypothesized that the limitation of GA-induced bolting by ABA might be weakened by vernalization, owing to an antagonism between GA and ABA in some biological processes [43]. However, the role of crosstalk between ABA and GA in the regulation of flowering remains unclear. The overexpression of AtABI4, which positively mediates ABA signalling, results in late flowering and GA reduction [14]. In contrast, the up-regulation of AtFT and AtTSF in response to drought accelerates bolting and flowering, while the content of GA is reduced by drought [44,45]. To test the effects of crosstalk between ABA and GA, the bolting-related traits were investigated after treatment with exogenous GA and ABA. Whatever the bolting initiation and bolting stem growth, they were significantly repressed by 0.1 and 1 mM ABA. It suggested that ABA competed with GA and disrupted the bolting in sugar beet induced by GA and the vernalization-induced inhibition of ABA biosynthesis and signalling removed the limiting factor for bolting.

Similar to the expression pattern of ABA biosynthesis genes, BvLOX2S, BvAOS, BvAOC, BvOPR and BvACAA1 were repressed by vernalization. These genes contribute to several steps of metabolism from linolenic acid to JA and JA-Ile. Not only were the contents of JA and JA-Ile significantly down-regulated, but also the contents of intermediate products, such as OPC-4, were significantly decreased after vernalization. Our results reflected that vernalization is different from the short-term low temperatures that often cause an increase in ABA and JA content [46]. In the JA signalling pathway, most of the BvMYC2s was down-regulated after vernalization. This was consistent with the weakened JA response caused by lower JA levels. JAs have been reported to compete with GA to regulate bolting and flowering in many species. The negative role of JA signalling in flowering is recognized by the early flowering observed in both coi1 mutants and seedlings that overexpress AtJAZ1 [19]. In Nicotiana attenuata, high levels of JA strongly antagonize the biosynthesis of GA and suppress stem growth by inhibiting the accumulation of transcripts of GA20ox [47]. Diallo et al. (2014) found that vernalization can result in the accumulation of MeJA in hexaploid winter wheat (Triticum aestivum L.) [39]. The accumulation of MeJA is rapidly restored post-vernalization. Further study on the mvp wheat mutant, in which non-flowering phenotypes were caused by deletions, suggest a role for MeJA in modulating vernalization and flowering time in wheat [39]. In sugar beet, 1 mM MeJA could obviously inhibit GA-induced bolting, while the inhibition would be slight under 0.1 mM MeJA. It is suggested that the role of MeJA in the regulation of plant growth and development depended on concentration. Similarly, high concentrations of MeJA have been reported in the inhibition of callus growth in Allium jesdianum [48]. In contrast, callus growth was promoted by a lower concentration of MeJA.Taken together, vernalization might alleviate the inhibition of GA-induced bolting by a repression of JA biosynthesis and signalling.

4.2. Vernalization Directly Repressed the Expression of BvDELLAs for Bolting, Although GA Biosynthesis Was Partially Repressed

Low temperature (4 °C) treatment increases the content of GA and enhances the accumulation of GA during the initiation of flowering pak choi (B. rapa) [40]. The effect of vernalization on the levels of endogenous phytohormones are also demonstrated at the transcriptional level. Consistent with increase in GA content, more genes involved in GA synthesis were found to be up-regulated in pak choi after low-temperature treatment and at the floral bud differentiation stage than down-regulated [40]. In sugar beet, a previous study on the statistical analysis of bolting frequency and bolt height treated with different treatment combinations showed that the three-way interaction of vernalization × GA × genotype reached a significant level (p < 0.01), which was more than the vernalization × GA and vernalization × genotype interactions [10]. This suggests that the crosstalk between vernalization and GA responses promotes stem growth in biennial sugar beet. Sorce et al. (2002) showed that the increase in total GA content occurs in sugar beet after vernalization and during bolting [49]. Consistent with this, we found that the up-regulation of most of the BvKS genes was detected after vernalization and post-vernalization. BvKS catalyzes the production of ent-kaurene, which is a precursor for GA biosynthesis in plants [49]. A significant increase in the level of GA₁₉ in both the Vel and Re treatments was also detected in this study. However, GA₁₉ is a C₂₀-GA compound that is an intermediate product during the conversion of GA₅₃ to bioactive GA₁ and GA₃. We also detected the reduction of GA₅₃ and GA₂₀, which are up- and downstream products in GA biosynthesis. This implied that the conversion of GA₁₉ to GA₂₀ was suppressed by vernalization but not the conversion of GA₅₃ to GA₁₉. Consistent with these results, the level of expression of two BvGA20OXs, which are responsible for the production of C₁₉-GAs using C₂₀-GAs as substrates, were repressed by vernalization and reached significant down-regulation post-vernalization [50]. Thus, our results suggest that GA biosynthesis is partially repressed by vernalization. The increase in total GA might be caused by the accumulation of inactive GAs.

Typically, GA-dependent transcriptional regulation consisting of the degradation of DELLA by the 26S proteasomes is mediated by oscillation of GID1 abundance in response to GA [51]. DELLAs, as the central repressors of the GA signalling pathway, interact with dozens of transcription factors regulating many biological processes, such as bolting and flowering [10]. In Arabidopsis, DELLAs directly interact with the CCT-domain of AtCO and the DNA-recognition domain of AtPIF4, causing the down-regulation of AtSOC1 and AtFT expression and late flowering [52]. The heterologous transformation of AtGAI, which encodes DELLA proteins, into sugar beet increases the thermal time required for post vernalization. In this study, more than one quarter of the BvGID1s in the sugar beet genome were up-regulated in the Vel or Re treatment, while down-regulation of other BvGID1s were detected. This might be owing to the partial repression of GA biosynthesis. Interestingly, nine BvDELLAs were noticed to be significantly down-regulated after vernalization. It is implied that vernalization does not depend on GAs to activate downstream genes by the direct transcriptional repression of BvDELLAs for bolting.

5. Conclusions

In this study, we found that the content of dozens of compounds decreased after vernalization and remained at low levels for several days in the post-vernalization condition. In particular, the contents of ABA and JA decreased by more than sixteen- and three-fold after vernalization. At the transcriptional level, 7471 DEGs were identified in response to vernalization. Subsequently, GO enrichment also showed that vernalization had a substantial impact on phytohormone metabolism and signalling. In terms of phytohormone metabolism, the genes involved in multiple steps of the ABA and JA biosynthetic pathways were repressed by vernalization. The GA biosynthesis might be affected by vernalization, owing to the down-regulation of BvGA20ox. In terms of phytohormone signalling, the down-regulation of BvABF and BvMYC2, which are key regulators in ABA and JA signalling, was caused by vernalization. The downstream genes of GA signalling might be activated by vernalization via the direct repression of BvDELLAs. These results implied that vernalization relieves the antagonistic effect of ABA and JA on GA signalling and promotes the expression of floral integrated genes. This was also verified by the observation of bolting by the application of GA₃, ABA and MeJA in sugar beet. Our study on the profile of change of phytohormones and DEGs in response to vernalization provides new insights into the crosstalk of external and internal factors in the regulation of bolting in sugar beet.