Introduction

Nitrogen, which constitutes approximate 2% of plant dry matter, is an essential component for plants to make amino acids, proteins, and DNA (Frink et al. 1999). Plants have evolved a set of adaptive abilities to a limiting nitrogen supply to finish their life cycles and produce offspring, including the repression of photosynthesis and growth, remobilization of nitrogen from mature organs to growing ones, and accumulation of abundant anthocyanins (Diaz et al. 2006; Peng et al. 2008; Feyissa et al. 2009).

Anthocyanins, a class of plant secondary metabolites, often accumulate in leaves in response to many stress conditions, including low temperature and nutrient depletion (Gould 2004; Feyissa et al. 2009). The anthocyanin biosynthesis is catalyzed by a lot of enzymes. According to the previous reports (Dubos et al. 2008; Gonzalez et al. 2008; Petroni and Tonelli 2011), the coding genes of these metabolic enzymes are often divided into two groups, designated as early biosynthetic genes (EBGs), including chalcone synthase (CHS), chalcone isomerase (CHI), and flavonol 3-hydroxylase (F3H), and late biosynthetic genes (LBGs), including flavonol 3′-hydroxylase (F3′H), dihydroflavonol reductase (DFR), leucoanthocyanidin dioxygenase (LDOX), UDP-flavonoid 3-glucosyl transferase (UF3GT), and anthocyanidin reductase (ANR). For investigating how anthocyanin accumulation is regulated in plants, many regulators were identified. The complex composed of MYB-bHLH-WD40 (MBW) transcription factors, including Transparent Testa (TTG1, a WD40-repeat protein), Transparent Testa 8 (TT8), Glabra 3 (GL3), enhancer of GL3 (EGL3), production of anthocyanin pigmentation 1 (PAP1), and PAP2, could activate expressions of LBGs, leading to the production of anthocyanins (Baudry et al. 2006; Gonzalez et al. 2008; Feyissa et al. 2009). Ectopic expressions of PAP1 and PAP2 increase anthocyanin accumulation in a TTG1-dependent manner, but their down-regulation reduces expressions of LBGs (Gonzalez et al. 2008). In yeast two-hybrid assay, PAP1 and PAP2 interact with TTG1, EGL3, GL3, and TT8, implying that multiple MBW complexes can be formed. Moreover, in transcriptional transient expression assays using protoplasts, coexpression of PAP1 with EGL3, GL3, or TT8 could stimulate expression of a reporter gene driven by the DFR promoter (Zimmermann et al. 2004).

Phytohormones, including cytokinin, ethylene, jasmonic acids, and gibberellins (GAs), are important internal factors affecting anthocyanin biosynthesis (Loreti et al. 2008; Jeong et al. 2010; Qi et al. 2011; Zhang et al. 2011a; Das et al. 2012a, b). GAs, a family of tetracyclic diterpene plant hormones, could control diverse aspects of growth and development, and their biosynthesis and signaling pathway have been well characterized, particularly in Arabidopsis (Plackett et al. 2012; Rieu et al. 2008a, b; Yamaguchi 2008). The production of bioactive GAs from common diterpene precursor trans-geranylgeranyl diphosphate requires a set of enzymes, including terpene synthases, cytochrome P450 monooxygenases, and two types of 2-oxoglutarate-dependent dioxygenases: GA 20-oxidases (GA20ox) and GA 3-oxidases (GA3ox). Whereas GA 2-oxidases (GA2ox), another class of dioxygenases, insert a 2β-hydroxyl to inactivate the bioactive GAs, and thereby serve to reduce the concentration of bioactive GAs (Rieu et al. 2008a, b; Yamaguchi 2008; Sun 2011; Plackett et al. 2012). Endogenous GA content in plants was often controlled through direct regulation of GA metabolism via changes of GA20ox, GA3ox, and GA2ox expressions in response to environmental or developmental stimuli (Achard et al. 2007; Jiang et al. 2007; Hauvermale et al. 2012).

Many components of the GA signaling, including the GA receptor Gibberellin Insensitive Dwarf 1 (GID1), the growth inhibitor DELLA proteins (DELLAs), and the F-box proteins Sleepy 1 (SPY1), have been identified (Sun 2011; Davière and Achard 2013). It was proposed that GA signal promotes growth by overcoming DELLA-mediated growth restraint (Sun 2011; Davière and Achard 2013). GA targets DELLAs for ubiquitylation and subsequent destruction through the 26S proteasome-dependent pathway, thereby relieving their restraining effects on plant growth and development (Hauvermale et al. 2012; Davière and Achard 2013). GA signaling transduction will be reduced in gain-of-function mutants of DELLA genes but enhanced in loss-of-function mutants of DELLAs (Achard et al. 2007; Davière and Achard 2013). There are five DELLAs (GA-insensitive, GAI; repressor of ga1-3, RGA; RGA-Like 1, RGL1; RGL2 and RGL3) in Arabidopsis, which play distinct but overlapping functions in repressing GA responses (Davière et al. 2008; Sun 2011; Davière and Achard 2013).

Previously, it was found that sucrose-induced transcriptional expression of DFR and PAP1 is repressed by application of exogenous GA3, indicating that GAs act as an antagonist of sucrose-induced anthocyanin biosynthesis (Loreti et al. 2008). Furthermore, GA-insensitive mutant gai is less sensitive to GA-dependent repression of sucrose-induced DFR expression, implying that GA modulates sugar-induced anthocyanin biosynthesis through regulation of the amount or activity of DELLAs (Loreti et al. 2008; Das et al. 2012a; Li et al. 2014). Similarly, phosphate starvation-induced anthocyanin accumulation is also repressed by exogenous GA3, and moreover, the DELLA quadruple mutant gai-t6/rga-t2/rgl1-1/rgl2-1 is insensitive to phosphate starvation up-regulation of several anthocyanin biosynthetic genes, such as F3′H and LDOX, suggesting the involvement of the GA-DELLA pathway in phosphate starvation-induced anthocyanin biosynthesis (Jiang et al. 2007). In addition, our previous work also proposed that GA-DELLA participates in low-temperature-induced anthocyanin accumulation (Zhang et al. 2011a). But whether and how GA-DELLA pathway plays a role in nitrogen deficiency-induced anthocyanin accumulation still remain unclear. Here, we found that GAs negatively regulate nitrogen deficiency-induced anthocyanin accumulation and biosynthetic gene expressions, and further studies indicate that DELLAs, especially RGA and GAI, are necessary regulators during these processes. Interestingly, yeast two-hybrid assay, pull-down assay, as well as luciferase complementation assay collectively showed that RGA could interact with PAP1. Transient expression assays further indicated that RGA and GAI could transcriptionally enhance the effects of PAP1-induced promoter activities of F3′H and DFR. These data thereby establish the prime role of the GA-DELLA system in nitrogen deficiency regulation of anthocyanin biosynthesis.

Materials and methods

Plant materials and growth conditions

The seeds of mutants ga20ox1 (SALK_094207C), ga20ox2 (SALK_029533C), ga3ox1 (CS6943), ga3ox1/ga3ox2 (ga3ox1/3ox2, CS6944), ga2ox1 (SALK_095011C), and ga2ox2 (SALK_011239C) were obtained from Arabidopsis Biological Resource Center (ABRC), which are in Columbia-0 (Col-0) ecotype background. Mutant lines of ga20ox1/ga20ox2 (ga20ox1/20ox2), rga-24/gai-t6, gai, gai-t6/rga-t2/rgl1-1/rgl2-1/rgl3-1 (della), and transgenic line 35S::TAP-RGAd17 (over-expressing TAP-tagged RGA but lacking a 17 amino-acid motif within the DELLA domain), which are all in Landsberg erecta (Ler) ecotype background, were as described previously (Achard et al. 2007; Loreti et al. 2008; Feng et al. 2008; Plackett et al. 2012). All seeds were surface sterilized with 20% (V/V) bleach solution for 10 min, rinsed with sterile water for three times, and then sown on 0.8% agar-solidified 1/2 strength Murashige and Skoog (1/2MS) media containing 3% sucrose and indicated concentrations of potassium nitrate (KNO3) but lacking ammonium. Seeds on agar plates were cold-treated at 4 °C for 3 days in the dark and then transferred to a controlled-environment chamber (22 °C, 16 h light). After the germination of the seeds finished within 36 h, the germinated seeds were transferred to the same new media but applied with or without GA3 (an active form of GAs) or paclobutrazol (PAC, a specific GA biosynthesis inhibitor) for further growth. Unless specified otherwise, the concentrations of potassium nitrate were 10 mM (high nitrate, HN) or 0.1 mM (low nitrate, LN), and the concentrations of GA3 and PAC were 10 and 1 μM, respectively.

Anthocyanin quantification

Anthocyanin content of seedlings was determined as described previously (Vandenbussche et al. 2007; Liu et al. 2015).

Real-time quantitative reverse transcription PCR (qRT-PCR)

RNA extraction, RT reaction, and qPCR assay were performed as described previously (Zhang et al. 2015). Primer sequences used here can be found in Supplementary Table 1.

Yeast two-hybrid (Y2H), pull-down assay, and luciferase complementation assay

The C-terminal coding sequences of RGA were cloned into indicated restriction enzyme sites of the pGBKT7 vector. The full-length coding sequences of PAP1 and PAP2 were cloned into the indicated restriction enzyme sites of the pGADT7 vector. Yeast two-hybrid assays were performed as the Matchmarker GAL4-based two-hybrid system (Clontech). Constructs were cotransformed into the yeast AH109. The presence of the transgenes was confirmed by growth on SD media lacking tryptophan (Trp) and leucine (Leu) (SD/-T/-L) plates. To assess protein interactions, the transformed yeast was selected on SD media lacking Trp, Leu, and histidine (-His) but containing 40 μg/mL X-α-gal (SD/-T/-L/-H/X-α-gal), and SD media lacking Trp, Leu, His, and adenine (Ade) but containing 40 μg/mL X-α-gal (SD/-T/-L/-H/-A/X-α-gal). Interactions were observed after 4 days of incubation at 30 °C. All the primers used for plasmid constructions in Y2H assays were listed in the Supplementary Table 1.

The procedure of pull-down assay was performed according to Qi et al. (2014) with some modification. The full-length coding sequence of RGA was inserted into pMAL-C2x vector to produce RGA-MBP. The MBP and MBP-fused RGA were expressed in Escherichia coli BL21 and purified by amylose resin beads (NEB). Full-length coding sequence of PAP1 was introduced into pSAT6-EYFP-C1 (CD3-1103) to get 35S:PAP1-EYFP. About 2 × 106 Col-0 protoplasts were transfected with 200 μg 35S:PAP1-EYFP plasmid and cultured overnight. The protoplasts were then harvested and total proteins were extracted. A portion of the protein extracts was mixed with 10 μg purified recombinant RGA-MBP and the equal amount of protein extract was mixed with 10 μg MBP as a negative control. Then, 10 μL MBP resin (NEB) was added to the samples. After incubation for 6 h at 4 °C with constant rotation, the resin was washed five times with ice-cold washing buffer, and then used for immunoblotting analyses using anti-GFP antibody (Clontech).

The luciferase complementation assays were performed according to Chen et al. (2008). In brief, the full-length coding sequences of RGA and PAP1 were introduced into pUC19-cLUC (CD3-1702) and pUC19-nLUC (CD3-1701) to form CLuc-RGA and PAP1-NLuc, respectively. These vectors were transformed into Arabidopsis Col-0 protoplasts. After culture overnight, the firefly LUC activities were determined using Firefly Luciferase Activity Assay kit (Beyotime). Relative LUC activity is equivalent to luminescence intensity/200 protoplasts. Each data point consisted of three replicates. Primers used for construction are presented in Supplemental Table 1.

Transient expression assay

The transient expression assays were performed in Arabidopsis protoplasts as previously described (Zhang et al. 2015). To generate the effectors, full-length coding regions of PAP1, RGA, and GAI from Col-0 were, respectively, cloned into pGreenII 62-SK vectors under control of CaMV 35S promoter. To generate the reporter constructs, the ~2-kb promoter sequences of F3′H, DFR and LDOX were, respectively, amplified from genomic DNA of Arabidopsis Col-0 and cloned into pGreenII 0800-LUC (Hellens et al. 2005). All primers used for making these constructs are listed in Supplementary Table 1. Arabidopsis mesophyll protoplasts preparation from 4-week-old Ler, della, or gai seedlings grown under short photoperiod (12 h light/12 h dark photoperiod) and subsequent transfection were performed as described previously (Yoo et al. 2007). Firefly luciferase (fLUC) and renilla luciferase (REN LUC) activities were measured using the Dual-Luciferase Reporter Assay System (Promega).

Statistical analysis

The significance of differences between data sets was evaluated using paired student’s t test using the OriginPro8.0 software (OriginLab).

Accession numbers

Sequence data used in our experiments can be found from the Arabidopsis Genome Initiative database: Actin2 (At3g18780), CHS (At5g13930), CHI (At3g55120), F3′H (At5g07990), DFR (At5g42800), LDOX (At4g22880); UF3GT (At5g54060); PAP1 (At1g56650); TT8 (At4g09820); TTG1 (At5g24520); GL1 (At3g27820); GL3 (At5g41315); RGA (At2g01570); GAI (At1g14920); RGL1 (At1g66350); RGL2 (At3g03450); RGL3 (At5g17490); GA20ox1 (At4g25420); GA20ox2 (At5g51810); GA3ox1 (At1g15550); GA3ox2 (At1g80340); GA2ox1 (At1g78440); and GA2ox2 (At1g30040).

Results

GAs negatively regulate nitrogen deficiency-induced anthocyanin accumulation and biosynthesis gene expression

To test whether GAs play a role during nitrogen deficiency inducing anthocyanin biosynthesis, the anthocyanin levels were determined in 6-day-old Col-0 seedlings grown on 1/2MS agar media lacking ammonium but containing various concentrations of nitrate with or without 10 μM GA3 or 1 μM PAC. The results showed that as the nitrate concentration decreased from 30 to 0.1 mM, the anthocyanin levels increased significantly (Fig. 1a). It was worthy to note that application of GA3 impaired, but PAC strengthened such effect, respectively (Fig. 1a), which coincide with the phenotypes that the purple of hypocotyls and cotyledons of the 0.1 mM nitrate-treated seedlings faded after GA3 treatment, but was deepened by PAC treatment (Fig. 1b). These data suggest that GAs play a negative role in nitrogen deficiency-induced anthocyanin accumulation in Arabidopsis seedlings. Such effect might result from GAs down-regulation of expressions of anthocyanin biosynthetic genes. To test this hypothesis, the transcription levels of several genes involved in anthocyanin biosynthesis pathway were assayed in 5-day-old Col-0 seedlings grown on 1/2MS agar media lacking ammonium but containing 10 mM nitrate (higher nitrogen, HN) or 0.1 mM nitrate (lower nitrogen, LN) with or without 10 μM GA3 or 1 μM PAC. As shown in Fig. 2, transcripts of CHS and especially F3′H, DFR, LDOX and UF3GT were significantly enhanced by LN treatment. Furthermore, application of GA3 and PAC decreased and increased LN-induced transcript levels of these genes, respectively (Fig. 2). These results indicate that GAs negatively regulate LN-induced anthocyanin biosynthesis gene expressions.

Fig. 1
figure 1

a Effects of exogenous GA3 and PAC on anthocyanin accumulation in the presence of various concentrations of nitrate. Anthocyanin levels were determined in 6-day-old Col-0 seedlings grown on 1/2MS media lacking ammonium but containing various concentrations of potassium nitrate with or without GA3 (10 μM) and PAC (1 μM) treatment. Each value is the mean of three biological replicates ± SE. b Representative images of Col-0 seedlings grown under 1/2MS but only containing 0.1 mM nitrate with or without 10 μM GA3 and 1 μM PAC for 6 days

Full size image
Fig. 2
figure 2

a Effects of GA3 and PAC on transcript levels of several anthocyanin biosynthesis genes under nitrogen deficiency. RNA were extracted from 5-day-old Col-0 seedlings grown on 1/2MS media containing 10 mM of potassium nitrate (HN) or 0.1 mM of potassium nitrate (LN) with or without GA3 (10 μM) and PAC (1 μM) treatment, and the transcript levels of genes CHS, CHI, F3′H, DFR, LDOX, and UF3GT were assayed via qRT-PCR. Each value is the mean of three biological replicates ± SE

Full size image

These results led us to propose that decrease of endogenous GA levels would enhance, but increase of them would impair, LN-induced anthocyanin accumulation. Previously, it was found that mutations of GA20ox or GA3ox, key genes encoding enzymes involved in GA activation pathway, will decrease endogenous bioactive GAs in Arabidopsis, whereas mutations of GA2ox, key genes encoding enzymes involved in GA inactivation pathway, will enhance endogenous bioactive GAs (Rieu et al. 2008a, b). Therefore, we subsequently evaluate the negative effects of GAs on LN-induced anthocyanin accumulation and biosynthesis gene expression using several Arabidopsis mutants related to these genes. In all seedlings tested, LN obviously enhanced the anthocyanin levels compared to those grown on HN (Fig. 3a). LN exhibited similar effects on induction of anthocyanin accumulation in single mutants ga20ox1, ga20ox2, ga3ox1 to wild-type Col-0, but the effects were enhanced in double mutants ga20ox1/ox2 and ga3ox1/ox2, and, conversely, was impaired in ga2ox1 and ga2ox2 mutants when compared to Col-0 (Fig. 3a). To strengthen this model, we subsequently measured the effects of LN on transcript levels of F3′H, DFR, LDOX, and UF3GT. As shown in Fig. 3b, LN increased the transcript levels of F3′H, DFR, LDOX, and UF3GT (9.7-, 15.3-, 10.5-, and 5.7-fold, respectively) in Col-0 seedlings, but LN increased the levels more dramatically in the strains involved in GA activation pathway (13.0-, 30.2-, 13.5-, and 6.7-fold, respectively, in the ga20ox1/ox2 double mutant, and 11.8-, 28.1-, 12.7-, and 5.9-fold, respectively, in the ga3ox1/ox2 double mutant). By contrast, LN exhibited reduced effects on the expressions of these genes when the GA inactivation mutant strains were examined (only 7.0-, 8.3-, 6.0-, and 3.7-fold increase for F3′H, DFR, LDOX, and UF3GT, respectively, for the ga2ox1 mutant and 6.6-, 7.8-, 6.2-, and 4.2-fold increase, respectively, in the ga2ox2 mutant). Collectively, these results show that changes of endogenous GAs could affect LN-induced anthocyanin biosynthesis.

Fig. 3
figure 3

a Effects of LN on anthocyanin accumulation in several mutants involved in GA metabolism. Anthocyanin levels were determined in 6-day-old seedlings of Col-0, ga20ox1, ga20ox2, ga20ox1/ga20ox2 (ga20ox1/20ox2), ga3ox1, ga3ox1/ga3ox2 (ga3ox1/3ox2), ga2ox1, and ga2ox2 grown on 1/2MS media containing HN or LN. b Effects of LN on transcript levels of several key genes involved in anthocyanin biosynthesis in various mutants involved in GA metabolism. RNA were extracted from 5-day-old seedlings of Col-0, ga20ox1/ox2, ga3ox1/ox2, ga2ox1, and ga2ox2, and the transcript levels of genes F3′H, DFR, LDOX, and UF3GT were assayed via qRT-PCR. Each value is the mean of three biological replicates ± SE. The numbers over bars indicate the fold changes of LN relative to HN, and the different lowercase letters below each bar represent significance of differences at the level of P < 0.01 according to the Student’s t test

Full size image

DELLAs are necessary for nitrogen deficiency fully induction of anthocyanin accumulation and biosynthetic gene expression

Because DELLAs are important factors involved in GA signaling, we next asked if the GA negatively regulation of LN-induced anthocyanin biosynthesis is via DELLAs. The effects of LN on anthocyanin accumulation were determined in loss-of-function mutants gai-t6/rga-24 and della, GA-insensitive mutant gai, and the transgenic line 35S::TAP-RGAd17, as well as their wild-type Ler. As shown in Fig. 4a, LN increased anthocyanin levels by 6.0-fold in Ler, but by 4.8- and 4.7-fold in gai-t6/rag-24 and della, respectively, and with 7.8- and 8.9-fold in 35S::TAP-RGAd17 and gai, respectively. These data indicate that DELLAs are positive regulators for LN fully induction of anthocyanin accumulation. In addition, it was found that high sucrose as well as low phosphorus-induced anthocyanin accumulation were impaired in della mutant (Fig. S2). Consistently, the requirement for DELLAs on LN-induced enhancements of transcript levels of genes F3′H, DFR, LDOX, and UF3GT was observed. For instance, LN increased F3′H expressions about 4.5-fold in Ler, but 3.5- and 6.8-fold in della and gai, respectively. Similar results were also found for the expression of DFR, LDOX and UF3GT (Fig. 4b), implying that DELLAs positively involve in LN-induced anthocyanin biosynthetic gene expression.

Fig. 4
figure 4

a Effects of LN on anthocyanin accumulation in DELLA mutants and transgenic line. Anthocyanin levels were determined in 6-day-old seedlings of wild-type Ler, loss-of-mutants gai-t6/rga-24, gai-t6/rga-t2/rgl1-1/rgl2-1/rgl3-1 (della), GA-insensitive mutant gai, and the transgenic line 35S::TAP-RGAd17 grown on 1/2MS media containing HN or LN. The numbers over bars indicate the fold changes of LN relative to HN. b Effects of LN on transcript levels of several key genes involved in anthocyanin biosynthesis in seedlings of Ler, della, and gai. RNA were extracted from 5-day-old seedlings of Ler, della, and gai, and the transcript levels of genes F3′H, DFR, LDOX, and UF3GT were assayed via qRT-PCR, after that, the transcript changes of each gene in seedlings treated LN relative to HN were calculated. Each value is the mean of three biological replicates ± SE. Different lowercase letters below each bar in a represent significance of differences at the level of P < 0.01 according to the Student’s t test. * and ** in b represent significance of differences at the levels of P < 0.05 and P < 0.01, respectively, based on the Student’s t test

Full size image

Next, we asked if LN has some effects on DELLAs. To this end, we measured the transcript levels of RGA, GAI, RGL1, RGL2, and RGL3 using qRT-PCR assays in wild-type seedlings grown under LN and HN conditions for 5 days. Regrettably, LN exhibited little effects on transcript levels of RGL1, RGL2, and RGL3 and just slight effects on RGA and GAI (Fig. 5a), suggesting that LN has a little effect on DELLA gene expression. Instead, LN exerts its effects on transcriptions of several key genes involved in GA metabolism. As shown in Fig. 5b, after Col-0 seedlings exposure to LN for 5 days, the transcript levels of GA20ox1 and GA3ox1, encoding both enzymes involved in the GA activation pathway, were significantly decreased, whereas the transcript levels of GA2ox1 and GA2ox2, encoding both enzymes involved in the GA inactivation pathway, were significantly enhanced, implying that LN might decrease the endogenous GAs through both down-regulation of GA activation genes and up-regulation of GA inactivation genes.

Fig. 5
figure 5

Effects of LN on transcript levels of a DELLA genes and b GA metabolic genes. RNA were extracted from 5-day-old seedlings Col-0 seedlings grown on 1/2MS media containing HN or LN, and the transcript levels of genes RGA, GAI, RGL1, RGL2, RGL3, GA20ox1, GA3ox1, GA2ox1, and GA2ox2 were assayed via qRT-PCR. Each value is the mean of three biological replicates ± SE. * , **Significance of differences at the levels of P < 0.05 and P < 0.01, respectively, when compared to the levels of HN treatment based on the Student’s t test

Full size image

RGA could interact with PAP1 and affect its transcriptional activities on anthocyanin biosynthetic genes

Subsequently, we wonder how DELLAs regulate anthocyanin accumulation. It was found that LN significantly increased transcript levels of PAP1, TT8, TTG1, and GL3, encoding several key components of the MBW complex, in wild-type Ler as well in mutant della, to comparable extents (Fig. S1), implying that DELLAs exhibit little roles on MBW at the transcript level. Previously, it was shown that DELLAs are involved in regulation of many progresses often through directly interaction with other transcriptional factors, such as Brassinazole Resistant 1 (BZR1) and JASMONATE ZIM-DOMAIN (JAZ) (Li et al. 2012; Qi et al. 2014). Therefore, whether some relationships exist between DELLAs and MBW was analyzed. To do this, Y2H assays were first performed. The full-length DELLAs all displayed strong autoactivation when fused with the GAL4 DNA-binding domain; therefore, the truncated C-terminal part of RGA (cRGA) was fused with BD domain to test its interaction with GL1 (Glabra 1), PAP1, PAP2, and TTG1. Interestingly, compared to the positive interaction between cRGA and GL1, which has been found by Qi et al. (2014), cRGA could strongly interact with PAP1 and weakly interact with PAP2 (Fig. 6a). To further confirm the interaction of RGA with PAP1, a MBP pull-down assay was performed. As shown in Fig. 6b, the EYFP-tagged PAP1 protein could be pulled down by RGA-MBP, but not by MBP alone, indicating that RGA interacts with PAP1 in vitro. Consistent with the results of the pull-down assay, an in vivo luciferase complementation assay also showed that cLUC-RGA fusion proteins could interact with PAP1-nLUC when both were transiently coexpressed in Arabidopsis protoplasts (Fig. 6c). These results indicate that RGA may regulate anthocyanin biosynthesis by forming heterodimers with PAP1.

Fig. 6
figure 6

RGA interacts with PAP1. a Yeast two-hybrid assays to detect the interactions of C-terminal of RGA (cRGA) with PAP1 and PAP2 in yeast AH109, and the relationships between cRGA with GL1 and AD were used as positive control and negative control, respectively. BD, pGBKT7; AD, pGADT7. b Pull-down assay of MBP-fused RGA expressed in Escherichia coli BL21 with PAP1-EYFP expression Arabidopsis protoplasts. Total proteins were pulled down by amylose resin and detected using an anti-GFP antibody. c Luciferase complementation assay of CLuc-RGA with PAP1-NLuc in Arabidopsis protoplasts. After vectors were cotransformed and cultured overnight, the relative firefly luciferase activities were determined. Each value is the mean of three biological replicates ± SE. **Significance of differences at the level P < 0.01 based on the Student’s t test

Full size image

The interaction between RGA and PAP1 raised the question of whether RGA affects PAP1’s transcriptional activities on its target genes, such as F3′H and DFR. To this end, transient transcriptional expression assays were performed in protoplasts. We introduced PAP1, RGA as well as GAI coding sequences into pGreenII 62-SK vector to get the effectors, and introduced the promoters of F3′H and DFR into the pGreenII 0800-LUC vector to get the reporters (Fig. 7a). After transformation of PAP1 and the reporters into protoplasts of Ler, della, and gai, it was found that the fLUC activities notably increased when compared to those transformed with the empty vector (Fig. 7b), indicating that PAP1 positively regulates the promoter activities of F3′H and DFR. It was worthy to note that these effects were significantly impaired in della but enhanced in gai when compared to those in Ler (Fig. 7b). More importantly, application of RGA or GAI could significantly or slightly increase the effects of PAP1 on LUC activities in Ler as well as in della (Fig. 7b). Taken together, these results indicate that RGA and GAI could enhance the transcriptional activities of PAP1 on F3′H and DFR.

Fig. 7
figure 7

a Schematic maps of the constructs used in the transient expression assay. Coding sequences of RGA, GAI, and PAP1 were separately cloned into pGreenII 62-SK vector to get the effectors. Empty vector (EV) pGreenII 62-SK was used as the negative control. Promoters of F3′H and DFR were separately inserted into pGreenII 0800-LUC vector to get the reporters. 35S; CaMV35S promoter; fLUC, firefly luciferase; REN LUC, Renilla luciferase; pF3′H, promoter of F3′H; pDFR, promoter of DFR. b The effects of RGA and GAI on PAP1-induced transcriptional activities of F3′H and DFR promoter reporters in Ler, della, and gai protoplasts. Each value is the mean of three biological replicates ± SE. Different lowercase letters represent significant differences at the level of P < 0.05 based on the Student’s t test

Full size image

Discussion

Nitrogen deficiency is a key environmental factor inducing anthocyanin accumulation in plants, observed many years ago (Martin et al. 2002). In our experiment, it was found that with nitrate concentration decrease in media, the anthocyanin levels increase significantly (Fig. 1). Interestingly, such effect was significantly impaired by exogenous GA3, but improved by PAC (Fig. 1), as we found that GAs negatively regulate nitrogen deficiency-induced anthocyanin accumulation through down-regulation of the biosynthesis genes, including CHS, F3′H, DFR, and LDOX (Fig. 2). The results obtained from GA biosynthesis mutants further support this notion (Fig. 3). Consistently, Loreti et al. (2008) found that sucrose-induced anthocyanins and transcript levels of DFR and PAP1 were decreased by GAs. Similarly, Jiang et al. (2007) reported that phosphate limitation induction of anthocyanin accumulation results from up-regulation of F3′H and LDOX, whereas GAs play a negative role during this process. GAs can also negatively regulate anthocyanin accumulation triggered by other factors, such as low temperature (Zhang et al. 2011a) and jasmonate (Xie et al. 2016). Hence, our results as well as previous reports lead us to conclude that GAs is one of the universal elements participating in environmental factors induction of anthocyanin biosynthesis, even though the molecular mechanisms underlying these effects remain unclear.

DELLAs repress seed germination, growth, and almost all known GA-dependent processes, whereas GA relieves their repressive activity (Achard et al. 2007; Yamaguchi 2008; Davière and Achard 2013). LN-induced anthocyanin accumulation and biosynthesis gene expressions were impaired in loss-of-mutant della but enhanced in GA-insensitive mutant gai and transgenic line 35S::TAP-RGAd17 (Fig. 4), indicating that such effect of GAs is mediated, at least partially, through DELLAs. Affecting transcription of genes encoding MBW complex might be a mechanism for DELLAs regulation of anthocyanin accumulation. Regretfully, we did not find that DELLA mutation exhibited any significant effects on gene expressions of PAP1, TT8, TTG1, and GL3 (Fig. S1). This is consistent with the results obtained by Jiang et al. (2007), in which it was found that DELLA mutation significantly impaired low phosphorus-induced up-regulation of genes F3′H, LDOX, and UF3GT, but only exhibited slight effect on PAP1 expression. However, Li et al. (2014) found that high sucrose up-regulation of PAP1 was largely attenuated in della mutant when compared to Ler. Such inconsistency may indicate that DELLAs regulation of anthocyanin accumulation is via different pathways under various environmental stresses.

In general, DELLAs are not able to directly bind to gene promoters, but function in plants indirectly through interaction with other transcription factors (TFs) (Davière and Achard 2013). For instance, DELLAs regulate hypocotyl elongation and plant defense by interacting with BZR1 and JAZs, respectively (Li et al. 2012; Qi et al. 2014). For these interactions, DELLAs generally block the DNA-binding capacity of the transcription factors or otherwise inhibit the transcriptional activity of them. Just recently, Xie et al. (2016) found that DELLAs could sequester MYBL2 and JAZ, two known repressor of anthocyanin biosynthesis, leading to the release of bHLH/MYB subunits and subsequently to the formation of active MBW complex, which then activates the anthocyanin biosynthesis. We here also found that RGA could directly interact with PAP1 (Fig. 6). Furthermore, RGA and GAI could positively enhance the activities of PAP1 on its downstream genes F3′H and DFR, which may be inconsistent with the notion that DELLAs generally function as transcriptional repressors. In point of fact, a few reports also found that RGA is able to function as transactivation factors via association with DNA (Zhang et al. 2011b; Davière and Achard 2013). Hence, the roles of DELLAs on transcription activities of TFs are very likely to be complicated. DELLAs are involved in anthocyanin accumulation may be either through interaction MYBL2/JAZ to sequester their transcriptional repression or through interaction PAP1 to strengthen its transcriptional promotion on anthocyanin biosynthetic gene expression. To clearly address the relationships among DELLAs, MBW complex, and MYBL2/JAZ, more works should be done with the help of biochemistry and genetics.

Sucrose is clearly important for inducing anthocyanin accumulation in agar-grown Arabidopsis seedlings (Teng et al. 2005; Loreti et al. 2008). The GA biosynthesis mutant ga1-5 accumulates more anthocyanins, and the GA-insensitive mutant gai is less sensitive to GA-dependent repression of sucrose-induced DFR expression (Loreti et al. 2008), implying that GA-DELLA pathway is involved in sucrose-inducing anthocyanin biosynthesis. However, sucrose alone was not sufficient for the accumulation of anthocyanins, but was triggered by additional lack of nitrogen (Lea et al. 2007; Zhou et al. 2012). In this study, 3% of sucrose was present in the media, and we additionally found that as the concentration of sucrose decreased, the effect of LN on anthocyanin accumulation was impaired (Fig. S2). Consistently, the ratio between carbohydrates and nitrogen (C/N) has been suggested to be an important factor for regulation of anthocyanin accumulation and biosynthesis gene expression (Lea et al. 2007). We here found that DELLAs play positive roles during both nitrogen deficiency and high sucrose induction of anthocyanin accumulation (Figs. 4, S2), but more detailed mechanisms how C/N regulates anthocyanin biosynthesis through GA-DELLA pathway need to be addressed.

It seems unlikely that LN affects DELLAs at transcriptional level, as LN exhibited no significant effects on transcript levels of DELLA genes (RGA, GAI, RGL1, RGL2, and RGL3; Fig. 4a). However, we found that LN reduced transcript levels of GA3ox1 and GA20ox1, whereas increased transcript levels of GA2ox1 and GA2ox2 (Fig. 5b), which may cause decreased GA levels and DELLA accumulation in LN-treatment seedlings. Similarly, light exhibits a little effect on RGA expression at transcript level, but DELLAs are expected to be altered through regulations of GA20ox1, GA3ox1, as well as GA2ox1 (Achard et al. 2007). In addition, phosphate starvation was found to cause changes in the transcripts levels of GA metabolism genes and a subsequent decrease in the level of bioactive GA4, which, in turn, causes DELLA accumulation and modulation of anthocyanin biosynthesis (Jiang et al. 2007). Therefore, nitrogen deficiency inducing anthocyanin biosynthesis may be through decrease of endogenous bioactive GAs, although their contents still remain to be assayed in the treated plants.

Finally, we also found that nitrogen deficiency-induced anthocyanin accumulation is largely light-dependent (Fig. S3), as can be explained by the finding that PAP1 is degraded in darkness in a COP1-dependent manner (Maier et al. 2013). Meanwhile, the della mutant exhibited the decreased sensitivity in response to light periods under nitrogen deficiency condition (Fig. S3), which may result from DELLAs’ positive roles for PAP1’s transcriptional activity on anthocyanin biosynthetic genes, as stated above. However, it should be noted that in the absence of DELLAs, LN could still significantly stimulate anthocyanin accumulation (Fig. S3), so not only DELLAs, but also other proteins must be involved in this process. The light signaling component HY5 is a candidate, as it has been shown to be a key positive regulator for light-inducing anthocyanin biosynthesis (Lee et al. 2007; Vandenbussche et al. 2007; Zhang et al. 2011a). Moreover, it was recently shown that HY5 could transcriptionally activate PAP1 via directly binding to its promoter (Shin et al. 2013). In addition, we found that nitrogen deficiency-induced anthocyanin accumulation is impaired in the loss-of-function mutant hy5-215 (data not shown). Hence, the complicated relationships among DELLAs, HY5, and MBW components in nitrogen deficiency regulation of anthocyanin biosynthesis may be an interesting research area.

In summary, this work shows that GAs are negative regulators of nitrogen deficiency induction of anthocyanin accumulation in Arabidopsis seedlings, whereas DELLAs play a positive role in this process. RGA interact with PAP1, resulting in PAP1’s enhanced transcriptional activity on its target genes encoding anthocyanin biosynthesis enzymes, including F3′H and DFR.

Author contribution statement

The experiments were conceived and designed by Yongqiang Zhang, Zhongjuan Liu, and Weifeng Xu. The experiments were performed by Yongqiang Zhang, Zhongjuan Liu, Jianping Liu, Sheng Lin, and Jianfeng Wang. The data were analyzed by Yongqiang Zhang, Zhongjuan Liu, and Weifeng Xu. Yongqiang Zhang, Wenxiong Lin, and Weifeng Xu wrote the paper.