https://orcid.org/0000-0003-0826-1101
Abstract
Deciphering the genetic basis of organoleptic traits is critical for improving the quality of fruits, which greatly shapes their appeal to consumers. Here, we characterize the citrus R3-MYB transcription factor TRIPTYCHON-LIKE (CitTRL), which is closely associated with the levels of citric acid, proanthocyanidins (PAs), and anthocyanins. Overexpression of CitTRL lowered acidity levels and PA contents in citrus calli as well as anthocyanin and PA contents in Arabidopsis leaves and seeds. CitTRL interacts with the two basic helix–loop–helix (bHLH) proteins CitbHLH1 and ANTHOCYANIN 1 (CitAN1) to regulate fruit quality. We show that CitTRL competes with the R2R3-MYB CitRuby1 for binding to CitbHLH1 or CitAN1, thereby repressing their activation of anthocyanin structural genes. CitTRL also competes with a second R2R3-MYB, CitPH4, for binding to CitAN1, thus altering the expression of the vacuolar proton-pump gene PH5 and Leucoanthocyanidin reductase, responsible for vacuolar acidification and proanthocyanidins biosynthesis, respectively. Moreover, CitPH4 activates CitTRL transcription, thus forming an activator–repressor loop to prevent the overaccumulation of citric acid and PAs. Overall, this study demonstrates that CitTRL acts as a repressor of the accumulation of citric acid, PAs, and anthocyanins by a cross-regulation mechanism. Our results provide an opportunity to simultaneously manipulate these key traits as a means to produce citrus fruits that are both visually and organoleptically appealing.
Introduction
Fleshy fruits contain abundant primary metabolites (sugars, organic acids) and secondary metabolites (phenolics, terpenoids) that are pivotal components in determining fruit palatability (Tucker, 1993). Organic acids greatly affect sensory traits of fruits, with fruits that have high acidic contents being generally less acceptable to consumers (Baldwin et al., 2014). Proanthocyanidins (PAs), also known as condensed tannins, are polyphenolic metabolites that protect plants from various stresses, facilitate seed dissemination, and benefit human health, but they also confer an astringent taste to fruits and juices, thus reducing their acceptance by consumers (Dixon et al., 2005). Another important sensory trait of fruits is color, which is attributable to the accumulation of pigments such as carotenoids and anthocyanins (Huang et al., 2018; Zhu et al., 2021). Developing delicious horticultural fruits with conspicuous color and high nutrient contents is an ongoing goal of both researchers and breeders.
Citrus fruits are among the most important fruits worldwide, and citric acid is the predominant organic acid they accumulate (Hussain et al., 2017; Wang et al., 2018; Tadeo et al., 2020). The genes involved in citrate biosynthesis and degradation have been well studied (Sadka et al., 2001; Terol et al., 2010; Hussain et al., 2017), and there is increasing focus on the genes that affect the accumulation of citrate in vacuoles (Shimada et al., 2006; Strazzer et al., 2019). V-type ATPases and V-type pyrophosphatases (V-PPases) are the most abundant proteins at the vacuolar membrane and help maintain the vacuole less acidic than the cytoplasm in most species (Shiratake and Martinoia, 2007). In addition, a vacuolar proton-pumping complex, formed by the two P-type vacuolar ATPases PH1 and PH5, is responsible for the hyper-acidification of modern citrus fruits. PH5 can directly transport protons (H+) across the tonoplast (Faraco et al., 2014; Strazzer et al., 2019). The expression of the genes encoding P-type H+ transporters is transactivated by the CitPH4–CitAN1 (ANTHOCYANIN 1) complex (Strazzer et al., 2019). Transient overexpression of CitPH4, also named CsMYB73, induced the accumulation of citric acid in tobacco (Nicotiana tabacum) leaves (Li et al., 2015). Similarly, mutations in CitAN1, encoding a basic helix–loop–helix (bHLH) transcription factor, resulted in a loss of acidity in citrus fruits (Butelli et al., 2019).
The homologues of citrus CitbHLH1 and CitAN1 in Arabidopsis are GLABRA 3 (GL3) and TRANSPARENT TESTA 8 (TT8), respectively, which are regulators of flavonoid biosynthesis (Montefiori et al., 2015; Huang et al., 2018; Zhang et al., 2020). Numerous studies have reported that bHLH transcription factors generally interact with R2R3-MYB and WD40 proteins to form a MYB–bHLH–WD40 (MBW) hierarchical complex participating in the biosynthesis of anthocyanins and proanthocyanins (Kleindt et al., 2010; Dixon et al., 2013; Xu et al., 2017). R2R3-MYB binds directly to genes promoters and ultimately determines the accumulation of flavonoid (Li et al., 2009; Xu et al., 2014; Yan et al., 2015; Naing and Kim, 2018; Jia et al., 2020). In citrus, the R2R3-MYB CitRuby1 cooperates with two bHLH proteins, CitbHLH1 and CitAN1, to activate the transcription of downstream genes such as CHALCONE SYNTHASE (CHS), ANTHOCYANIDIN SYNTHASE (ANS), FLAVANONE 3-HYDROXYLASE (F3H), and DIHYDROFLAVONOL 4-REDUCTASE (DFR), to promote anthocyanin accumulation (Huang et al., 2018; Strazzer et al., 2019). Another R2R3-MYB, CitPH4, the activator of citric acid accumulation mentioned above, also positively regulates PA biosynthesis in the seed coat of sweet orange (Citrus sinensis) together with CitAN1 by forming a protein complex (Zhang et al., 2020).
Besides MYB activators, several R2R3-MYB repressors and R3-MYBs also participate in the regulation of anthocyanin and PA biosynthesis (Zhu et al., 2009; Albert et al., 2014; Chen et al., 2019; Huo et al., 2020). Among them, CAPRICE (CPC)-like R3-MYBs are small proteins (about 75–112 amino acids) that passively repress the expression of structural genes because they lack the R2 domain and a repressor motif. Arabidopsis CPC was initially identified as a modulator of trichome and root hair formation, but was later shown to also negatively regulate anthocyanin biosynthesis by competing with PAP1/2 to bind to the bHLH partner (Wada et al., 2002; Zhu et al., 2009). In chrysanthemum (Chrysanthemum × morifolium), CmMYB#7 negatively modulated anthocyanin accumulation by preventing the activator CmMYB6 from binding to its CmbHLH2 partner (Xiang et al., 2019). Notably, the hierarchical interactions between MYB activators and repressors involved in flavonoid accumulation appear to be conserved among plants. Indeed, in petunia, the anthocyanin activator DEEP PURPLE (PhDPL) induces the transcription of the repressor gene PhMYB27 to form a fine-tuned loop that balances anthocyanin biosynthesis (Albert et al., 2014). Similar regulatory loops have also been described in tomato (Solanum lycopersicum) between SlAN2-like and SlMYBATV, in peach (Prunus persica) between PpMYB10.1 and PpMYB18, and in citrus between CitRuby1 and CitMYB3 (Zhou et al., 2019; Huang et al., 2020; Yan et al., 2020).
In this study, we identified a repressor R3-MYB transcription factor that simultaneously modulates citric acid, PA, and anthocyanin contents. We named this protein CitTRL for citrus TRIPTYCHON-like. Furthermore, we showed that CitTRL transcription is activated by CitPH4 to form a fine-tuned regulatory loop balancing citric acid and PA accumulation. CiTRL thus acts as an important node, as it connects the regulation of fruit flavor and color by competing with different MYB activators. The spatiotemporal manipulation of CitTRL transcription will provide an efficient way to produce visually pleasing and delicious citrus fruits.
Materials and methods
Plant materials
Citrus species including ‘HAL’ (Hong Anliu) sweet orange, ‘AL’ (Anliu) sweet orange, ‘WSY’ (Wusuan) pummelo, ‘HB’ pummelo, sweet lemon, ‘YLK’ (Youlike) sour lemon, and wild mandarin ‘JYYJ’ (Jiangyong) were cultivated in the germplasm resource nursery at Huazhong Agricultural University (Wuhan, China). The fruits of cultivated mandarin ‘BTJ’ (Bingtang) were collected from Yunnan province. The developmental blood orange and purple pummelo were collected previously (Huang et al., 2018, 2020). The pulps were sampled from three to six fruits of these varieties, and the seed coats were collected from at least 20 seeds of ‘HongAnliu’ and ‘Anliu’ sweet orange for one biological repeat. All the samples were divided and frozen in liquid nitrogen upon harvest, and then were stored at −80 °C for further analysis.
Analysis of sequence homology and phylogenetic relationship
To reveal the sequence homology among CitTRL, CitTRY, and other identified R3-MYB proteins in plants, we performed sequence alignment using translated amino acid sequences of these proteins by ClustalX. The alignment result was used to construct a phylogenetic tree using the neighbor-joining method with 1000 bootstrap replicates.
RNA extraction and real-time quantitative PCR
Total RNA was extracted from the pulp and pericarp of citrus, stems and young siliques of Arabidopsis, and tobacco leaves using Trizol iso plus (Takara). Total RNA (1 mg) was treated with 4× gDNA Wiper and reverse-transcribed to cDNA using HiScriptII Q RT SuperMix Kit (Vazyme Biotech). Quantitative RT-PCR was performed with SYBR-Green PCR Mastermix (YEASEN) on a 384-well plate and the fluorescence was real-time detected by Light Cycler 480 (Roche) (Zhang et al., 2022). The data were normalized to the expression of the actin gene (GenBank accession: GU911361, CAP ID: Cs1g05000). The equally mixed cDNA from three biological samples was used for the experiment with three repeats. We adopted the classic method to analyse RT-qPCR data. The quantitative primers are given in Supplementary Table S1.
Subcellular protein localization
The full-length coding sequence without stop codon of CitTRL from sweet orange was inserted into a modified pMDC43 vector (Nelson et al., 2007). The p35S: CitTRL-GFP was co-expressed with p35S: OsGhd7-RFP, a nuclear-localized protein, in Nicotiana benthamiana leaves mediated by Agrobacterium. After 48 h of infiltration, the fluorescence was detected using a confocal laser scanning microscope (TCIT SP2, Leica). Primer information is provided in Supplementary Table S1.
Construction of the expression vectors and genetic transformation
To generate CitTRL overexpression vectors, the full-length cDNA from sweet orange was amplified and recombined to pK7WG2D and pH7WG2D gateway vector using BP and LR enzymes. The overexpression plasmid pK7WG2D carrying CitTRL was introduced into Arabidopsis ecotype Columbia mediated by Agrobacterium strain GV3101 following the floral dip method (Clough and Bent, 1998). The positive transformants were screened out by resistance to kanamycin in the medium and were transplanted to soil after 10–15 d. Homozygous transgenic lines were used for further analysis.
The transformation of citrus calli ‘Guoqing 1’ (Citrus reticulata, ‘G1’) was performed as described previously (Duan et al., 2007). We first introduced the pH7WG2D carrying CitTRL into citrus calli to obtain CitTRL-OE transformants. The transgenic citrus calli carrying CitPH4 and CitTRL were generated by overexpressing CitTRL in the background of overexpression of CitPH4 and the transformants were screened out by double resistance to kanamycin and hygromycin. The transgenic calli sub-cultured at least six times were used for further studies. Primer information is represented in Supplementary Table S1.
Transient expression assays
The full-length coding sequence of CitbHLH1, CitAN1, CitRuby1, and CitPH4 was inserted into the pK7WG2D vector to generate overexpression vectors. All plasmids were individually genetically introduced into GV3101. Different combinations (R2R3-MYB+bHLH TF+CitTRL and R2R3-MYB+bHLH TF+empty vector) were transiently expressed in N. benthamiana leaves and citrus fruits designated kumquat (Fortunella margarita) (Gong et al., 2021). After 3 d (for N. benthamiana) or 5 d (for citrus fruits) of infiltration, the samples were photographed to observe phenotype and then were frozen immediately in liquid nitrogen at −80 °C until use. For citrus fruits, at least six fruits of uniform size and maturity were regarded as a biological replicate, and the data were obtained from three biological replicates.
Anthocyanin extraction and measurement
Anthocyanin content was measured following a previously reported method with some modifications (Huang et al., 2018). The stems and leaves of transgenic lines and the control from Arabidopsis and N. benthamiana were weighed and ground into powder in liquid nitrogen, and then extracted with 1 ml of 100% methanol+0.1% (v/v) HCl at 4 °C for 2 h. Subsequent centrifugation for 10 min at 15 000 g was employed to obtain the supernatant. Measured absorbance values of a total of 200 μl supernatant at 530 nm and 657 nm were recorded and the ratio of (A530−0.25×A657)/fresh weight was counted as anthocyanin content.
Qualitative and quantitative measurement of proanthocyanidins
To qualitatively evaluated PA content, the samples were stained with 1% (w/v) 4-dimethylaminocinnamaldehyde (DMACA) in methanol: 6 M HCl at room temperature for 48 h (Huang et al., 2020); 70% ethanol was used to wash the treated samples at least three times. The phenotypes of stained tissues were photographed using a stereomicroscope (Leica MZFL III). Three independent lines of each genotype were used in the assay.
To evaluate accurately the PA content, approximate amounts of transgenic line and the control (0.1 g T2 seeds and 1 g citrus calli) were used to determine the soluble and insoluble PA content. The samples were ultrasonically extracted in a 5 ml extraction solution with 70% acetone and 0.5% acetic acid at room temperature for 1 h, followed by centrifugation. Chloroform and hexane were used individually to treat transferred supernatants followed by centrifugation. For analysis of soluble PAs, the supernatant was reacted with 0.2% w/v DMACA in methanol–3 M HCl reagent for 30 min and then the absorbance at 640 nm was measured to calculate content, with (+)-catechin (MedChemExpress, USA) was used as a standard sample. To extract insoluble PAs, a mixture of the obtained pellet and 2 ml butanol–HCl was sonicated for 1 h followed by centrifugation. Then 1 ml supernatant was transferred to a new centrifuge tube with 6 ml of butanol–HCl reagent and 200 µl of 0.2% NH4Fe(SO4)2 solution followed by boiling for 40 min and cooling. The absorbance values of procyanidin B1 (MedChemExpress) with concentration gradients were recorded to draw a standard curve, which was used to convert the A550 values of supernatant into PA equivalents.
Dual-luciferase reporter assay
The promoter regions of CitCHS, CitDFR, CitANS, CitLAR, and CitPH5 were inserted into the pGreen0800 vectors as reporters, and the overexpression vectors harboring the open reading frame (ORF) of CitbHLH1, CitAN1, CitRuby1, CitPH4, and CitTRL as well as empty vector (EV) were used as effectors. All the vectors were introduced separately into the Agrobacterium GV3101 strain. Cultures were resuspended to OD600=0.8. Resuspensions of regulators and reporter were mixed in a proportion of 5:1 (v/v), of which R2R3-MYB, bHLH TF together with CitTRL or EV constituted a regulator mixture at a ratio of 1:1:1, and then the resuspensions were injected into both sides of the same leaf. Infiltrated N. benthamiana leaves were sampled to examine enzyme activities of firefly luciferase (LUC) and Renilla luciferase (REN) following the instructions of the Dual-Luciferase Reporter Assay System (Promega, USA). The primer sequences are listed in Supplementary Table S1.
Yeast two-hybrid assay
The investigation of protein interaction was conducted via a yeast two-hybrid assay. The ORF of CitTRL and bHLH proteins (CitbHLH1 and CitAN1) was cloned separately into PGBKT7 and PGADT7 vectors as bait and prey, respectively. Different combinations of bait and prey vectors were co-expressed in the yeast (Saccharomyces cerevisiae) strain AH109 with Clontech Takara Kit, of which co-transformed yeast with bait and PGADT7 was used as a negative control. Transformed positive yeasts were cultured with 600 µl liquid SD/–Leu–Trp medium at 200 rpm for 24–36 h. Thereafter, cultures were dotted on SD/–Trp–Leu–His–Ade plates to observe growth after 3–5 d incubation at 30 °C. The primers used in the assay are given in Supplementary Table S1.
Bimolecular fluorescence complementation
Expression vectors JW771 and JW772 individually fused with the N- and C-terminal halves of firefly luciferase (LUC) were used in bimolecular fluorescence complementation (BiFC) assays. The full-length coding sequence of bHLH TFs (CitbHLH1 and CitAN1) and MYB TFs (CitTRL, CitRuby1 and CitPH4) was cloned into JW771 and JW772, respectively. Four different mixtures, CitTRL–nLUC+cLUC–bHLH, CitTRL–nLUC+cLUC, nLUC+cLUC–bHLH, and nLUC+cLUC, were infiltrated into different parts of N. benthamiana leaves; 0.2 mM luciferase substrate was smeared on the transformed leaves’ surface in the dark for 5 min. Images were captured using a NightShade imaging apparatus (LC985).
In addition, CitTRL-OE was added to the mixture of R2R3-MYB–nLUC (CitRuby1 and CitPH4) and cLUC–bHLH to verify whether CitTRL interfered with the formation of the R2R3-MYB–bHLH complex. The assay was performed as described above. Primer information is presented in Supplementary Table S1.
Electrophoretic mobility shift assay
An electrophoretic mobility shift assay (EMSA) assay was performed using purified CitPH4–maltose binding protein (MBP) fusion protein. The synthesized oligonucleotides containing a potential MYB-binding site and the adjacent 10-bp sequence were labeled with 5ʹ 6-FAM (Sangon). The annealed probes were inoculated with CitPH4–MBP protein in the dark for 40 min, and then 100 V electrophoresis was conducted in the dark dark for 1 h. The primers used in the assay are given in Supplementary Table S1.
Results
Identification of a potential regulator for citric acid, proanthocyanidins, and anthocyanins in citrus
The sweet orange cultivar ‘Hong Anliu’ (HAL, Citrus × sinensis cv. Hong Anliu) is a mutant of the cultivar ‘Anliu’ (AL) with lower levels of citric acid and PAs in fruit pulp and is characterized by white seeds due to the loss of PA accumulation (Guo et al., 2016; Zhang et al., 2020) (Fig. 1A, B). Comparative transcriptome analysis of the HAL and AL cultivars demonstrated that several genes encoding transcription factors (CitTRY, CitWRKY44, and CitMYB179) are co-expressed with the key regulators PH4 (encoding a R2R3-MYB) and AN1 (encoding a bHLH transcription factor) (Zhang et al., 2020), which promote the accumulation of both citric acid and PAs (Butelli et al., 2019; Strazzer et al., 2019; Zhang et al., 2020). Of those, CitTRY (Cs5g_pb002540) and its homolog Cs9g_pb020880, referred to as CitTRL (TRY-like) hereafter, shared 68% and 66% sequence identity, respectively, with Arabidopsis TRY, an R3-MYB transcription factor that regulates trichome and root hair development, along with anthocyanin biosynthesis (Supplementary Fig. S1A) (Schellmann et al., 2002; Nukumizu et al., 2013). Phylogenetic analysis showed that both CitTRY and CitTRL cluster together with AtTRY (Fig. 1C). Multiple sequence alignment identified an intact R3 domain and a WxM motif responsible for movement between cells in the C-terminal domain of the two citrus homologs (Supplementary Fig. S1B).
Analysis of the phylogenetic relationship and expression level of CitTRL and CitTRY. (A) The fruit and seed of HAL and AL sweet orange. HAL is a mutant of AL for which the fruit tastes acidless and the seeds are white with a low level of proanthocyanins. (B) Citric acid content in the fruit juice of HAL and AL. (C) Phylogenetic analysis of CitTRY, CitTRL, and other characterized R2R3-MYBs and R3-MYBs. The phylogenetic tree was constructed using the neighbor-joining method with 1000 bootstrap replicates by MEGA (v7.0) software. Scale bar represents 0.050 substitutions per site. (D, G) Relative expression of CitTRY (D) and CitTRL (G) in fruit pulp and seed coat of sweet oranges. (E, H) Relative expression of CitTRY (E) and CitTRL (H) in the developmental peels of blood orange. (F, I) Relative expression of CitTRY (F) and CitTRL (I) in the developmental peels of purple pummelo. The data were obtained from three biological replicates. The peels present pink to purple due to the biosynthesis of anthocyanins as the fruits mature. Scale bar represents 1 cm.
We validated the expression levels of CitTRY (Fig. 1D) and CitTRL (Fig. 1G) in the fruit pulp and seed coat of HAL and AL cultivars by RT-qPCR. The expression of both genes was positively correlated with the levels of citric acid in pulp (Fig. 1B) and PAs in the seed coat (Zhang et al., 2020), with CitTRL exhibiting a greater rise in its expression in AL relative to HAL than CiTRY. As a subset of R3-MYBs negatively regulates anthocyanin biosynthesis (Chen et al., 2019), we also determined the relative transcript levels of CitTRY and CitTRL in a collection of citrus tissues collected previously with different levels of anthocyanins (Huang et al., 2018, 2019). The fruit peels of blood orange (Citrus × sinensis ‘Blood orange’) and purple pummelo (Citrus maxima) take on a pink to purple color in late developmental stages caused by the accumulation of anthocyanins. RT-qPCR analysis indicated that, in both blood orange (Fig. 1H) and purple pummelo (Fig. 1I), CitTRL is most highly expressed at the early green fruit stage, and its expression declines as the fruit expands before weakly increasing during late developmental stages. By contrast, CitTRY expression levels differed modestly over the sampled developmental stages of blood orange (Fig. 1E) and did not display consistent expression patterns between blood orange and purple pummelo (Fig. 1F). Considering these results together, we selected CitTRL for further analysis in this study.
We generated a construct for subcellular localization analysis, in which we cloned the CitTRL coding sequence in-frame and upstream of the green fluorescent protein (GFP) sequence. The recombinant plasmid was transiently expressed in N. benthamiana leaves. We detected fluorescence signals from CitTRL–GFP in the nucleus that co-localize with those of the nuclear marker OsGhd–RFP (a fusion between rice protein Grain number, plant height, and heading date7 (Ghd7) and red fluorescent protein (RFP)), indicating that CitTRL is a nucleus-localized protein (Supplementary Fig. S1C).
CitTRL is a repressor of citric acid and PA accumulation in citrus
To investigate the function of CitTRL in citrus, we overexpressed CitTRL in wild-type citrus calli, whose levels of citric acid and PAs are relatively low (Supplementary Fig. S2A, B). Although the expression of the H+-pumping gene CitPH5 and several structural genes involved in PA biosynthesis (CitCHS, CitF3H, and CitANS) decreased in CitTRL-OE transformants (Supplementary Fig. S2F, G), we observed no significant changes in the contents of citric acid or PAs between wild type (WT) and CitTRL-OE transformants (Supplementary Fig. S2C–E).
We thus tested the overexpression of CitTRL (Fig. 2A, B) in transgenic citrus calli already overexpressing CitPH4 (CitPH4-OE), encoding a R2R3-MYB that promotes citric acid accumulation (Li et al., 2015) and soluble PAs (Zhang et al., 2020). We selected three independent lines each for CitPH4-OE and CitPH4-OE+CitTRL-OE for further analysis. The introduction of CitTRL in the CitPH4-OE background resulted in lower contents of citric acid (Fig. 2C, D) and soluble PAs (Fig. 2E, F). RT-qPCR analysis showed that the expression levels of CitAN1, CitPH5, CitCHS, CitF3H, and CitLAR are higher in CitPH4-OE lines, but lower in the transformants co-expressing CitPH4 and CitTRL, as expected (Fig. 2G). Collectively, these results indicated that CitTRL acts as a negative regulator of citric acid and PA accumulation.
Overexpression of CitTRL in the transgenic citrus calli overexpressing CitPH4. (A) The phenotype of WT in citrus calli, and the transgenic lines overexpressing CitPH4 with or without CitTRL. (B) Relative expression of CitPH4 and CitTRL in WT, CitPH4-OE and CitPH4-OE+CitTRL-OE transformants. (C, D) The pH (C) and citric acid content (D) in WT and transgenic citrus calli. (E) Extraction of soluble PAs from citrus calli. (F) Quantification of soluble PA content in citrus calli. (G) RT-qPCR analysis of the expression of acidity-associated and PA-associated genes in WT, CitPH4-OE, and CitPH4-OE+CitTRL-OE transformants. The data were obtained from three independent transgenic lines. Asterisks indicate statistically significant differences relative to WT. Statistical significance: n.s., P>0.01; *0.01<P<0.001; **P<0.001.
Overexpression of CitTRL represses anthocyanin biosynthesis
To investigate the function of CitTRL in regulating anthocyanin accumulation, we generated stable transgenic lines in Arabidopsis overexpressing full-length CitTRL driven by the cauliflower mosaic virus (CaMV) 35S promoter. We selected three independent transgenic lines with high CitTRL expression relative to WT for further analysis (Fig. 3A).
CitTRL represses anthocyanins biosynthesis in Arabidopsis. (A) Relative expression level of CitTRL in wild type (WT) and CitTRL-OE transformants. (B) The seedlings of WT and transgenic lines that overpressed CitTRL. The enlarged seedlings in the white dotted frame are shown at the top right, and extraction solution of anthocyanin is shown at the bottom right. (C) The anthocyanin content in the stems of WT and the transgenic lines. FW, fresh weight. (D) Relative expression of flavonoids-related structural genes in the stems of WT and the CitTRL-OE transformants in Arabidopsis. The data were obtained from three biological replicates. Asterisks indicate statistically significant differences relative to WT. Statistical significance: n.s., P>0.01; **P<0.001.
We noted a purple color along the petiole of rosette leaves and the base of stems from WT, whereas the CitTRL-OE transformants remained green (Fig. 3B). Indeed, the CitTRL-OE transformants accumulated very small amounts of anthocyanins compared with WT (Fig. 3C). In agreement with this, Arabidopsis DFR and ANS expression was much lower in the basal part of stems of CitTRL-OE transformants relative to WT (Fig. 3D). These results indicated that CitTRL represses the expression of the downstream genes, and thus anthocyanin biosynthesis.
The CitTRL-OE transformants also affected PA biosynthesis. Mature Arabidopsis seeds generally have a brown seed coat, which turns black when stained with DMACA due to the presence of abundant PAs. The CitTRL-OE transformants only produced pale yellow seeds (Supplementary Fig. S3A). The contents of PAs were indeed lower in CitTRL-OE transformants relative to control seeds (Supplementary Fig. S3B). In addition to DFR and ANS, ANTHOCYANIDIN REDUCTASE (ANR) was also less expressed in young siliques of transgenic CitTRL-OE lines, as determined by RT-qPCR (Supplementary Fig. S3C). These results indicated that CitTRL represses the expression of flavonoid structural genes, and thus anthocyanin and PA biosynthesis.
CitTRL interacts with two bHLH transcription factors
The bHLH transcription factors CitbHLH1 and CitAN1 are both activators of anthocyanin biosynthesis. CitAN1 has additional roles in promoting the accumulation of PAs and citric acid in citrus (Huang et al., 2018; Butelli et al., 2019). We speculated that CitTRL may exert its function by competitively binding to a bHLH protein and weakening the activation mediated by R2R3-MYBs. We explored this possibility by conducting a yeast two-hybrid assay to test for interaction between CitTRL and CitbHLH1 or CitAN1. Yeast colonies containing BD-CitTRL and AD-CitbHLH1/AD-CitAN1 grew on selective medium, whereas the negative control carrying AD empty vector and BD-CitTRL did not, indicating that CiTRL interacts with CitbHLH1 and CitAN1 in yeast (Fig. 4A). We also turned to a luciferase (LUC) complementation imaging assay to test the interaction between these proteins in planta. In transient infiltration assays, only N. benthamiana leaves co-infiltrated with the CitTRL-nLUC and cLUC-CitbHLH1 or cLUC-CitAN1 constructs emitted luminescence, indicative of the reconstitution of full-length LUC (Fig. 4B), while other combinations of constructs, including CitTRL-nLUC+cLUC, nLUC+cLUC-CitbHLH1/CitAN1, and nLUC+cLUC, resulted in only weak LUC signals (Fig. 4B). Taken together, these data indicated that CitTRL can interact with CitbHLH1 and CitAN1 in yeast and in plants.
CitTRL physically interacts with CitbHLH1 and CitAN1. (A) CitTRL interacts with two bHLH transcription factors, CitbHLH1 and CitAN1, verified by yeast-two hybrid. The transformed yeasts were 10-fold diluted and dotted on SD/−L/−T and SD/−L/−T/−A/−H media to observe growth. Co-transformation of AD and BD-CitTRL into yeasts serves as a negative control. (B) The interactions between CitTRL and bHLH transcription factors in luciferase complementation imaging assays. CitTRL–nLUC+cLUC–bHLH together with the negative controls, including CitTRL–nLUC+cLUC, nLUC+cLUC–bHLH, and nLUC+cLUC, were injected into different parts of the same N. benthamiana leaves. Luminescence was detected 3 d after infiltration with NightShade imaging apparatus.
CitTRL interferes with the activation of anthocyanin, proanthocyanidin and acidity genes mediated by R2R3-MYB–bHLH complexes
To investigate the repression mechanism of CitTRL, we performed transient expression assays in N. benthamiana leaves with reporter constructs in which the promoter of several downstream genes drove the transcription of the LUC reporter gene, using Renilla luciferase (REN) driven by the CaMV 35S promoter as internal control. Co-infiltration of N. benthamiana leaves with the anthocyanin activators CitRuby1 (encoding a R2R3-MYB) and CitbHLH1 (Fig. 5A) or CitRuby1-CitAN1 (Fig. 5B) led to a rise in transcription from the CitCHS, CitDFR, and CitANS promoters. Co-infiltrating CitPH4 and CitAN1 effector constructs also induced the transcription of LEUCOANTHOCYANIDIN REDUCTASE (CitLAR) and CitPH5, which are critical for PA biosynthesis and vacuolar acidification, respectively (Fig. 5C, D). The co-infiltration of CitTRL decreased the transcription rate from the promoters of these downstream genes relative to that obtained with the R2R3-MYB and bHLH complex (Fig. 5A–D).
CitTRL represses downstream gene expression by interfering with the activation capacity of R2R3-MYB–bHLH complexes. (A, B) CitTRL impaired the expression of genes induced by anthocyanin activator complex CitRuby1–CitbHLH1 (A) or CitRuby1–CitAN1 (B). (C, D) CitTRL inhibited the target genes activated by CitPH4–CitAN1 complex. The promoter of proanthocyanidins gene (LAR) and acidity gene (PH5) was used as reporter vector in (C) and (D), respectively. (E) Schematic illustration of the vectors used in modified BiFC assays. (F) CitTRL reduced the affinities of CitRuby1–CitbHLH1/CitAN1 and CitPH4–CitAN1 complexes. The luminescence of N. benthamiana leaves was captured 3 d after infiltration. Error bars represent the mean ±SD of six biological replicates. Different letters indicate a significant difference using Duncan’s test: P<0.001.
To test the influence of CitTRL on the interaction between R2R3-MYB and bHLH transcription factors, we performed a modified BiFC assay using the CitTRL-OE construct as a competitor (Fig. 5E). Co-infiltration of different R2R3-MYB and bHLH pairs fused to nLUC or cLUC (CitRuby1–nLUC+cLUC–CitbHLH1/CitAN1, CitPH4–nLUC+cLUC–CitAN1) generated strong luminescence whose signal intensity decreased when CitTRL-OE was included (Fig. 5F). Collectively, these results indicated that CitTRL can compete for binding with bHLH TF and attenuate its interaction with R2R3-MYB, thus interfering with the activation by CitRuby1 (anthocyanins) or CitPH4 (citric acid and PAs) transcription factors of their downstream structural genes.
CitTRL represses two sets of R2R3-MYB–bHLH complexes
To further verify the function of CitTRL, we performed transient expression assays in N. benthamiana leaves and citrus fruits. Transient co-infiltration of CitRuby1+CitbHLH1 in N. benthamiana leaves resulted in an accumulation of red pigments (anthocyanins), while leaves co-infiltrated with CitRuby1+CitbHLH1 and CitTRL showed no visible color change, although anthocyanins did accumulate relative to the negative control (EV), albeit not to the same extent as with CitRuby1+CitbHLH1 (Fig. 6A–C). RT-qPCR analysis showed that the expression of N. benthamiana anthocyanin structural genes was up-regulated upon co-infiltration with CitRuby1+CitbHLH1, but largely repressed when CitTRL was included (Fig. 6D). Moreover, we observed a similar repression of anthocyanin biosynthesis when co-infiltrating CitTRL with CitRuby1+CitAN1 (Supplementary Fig. S4A–D).
Transient overexpression of CitTRL leads to a reduction in the level of metabolites induced by R2R3-MYB–bHLH complex. (A) Phenotype of N. benthamiana leaves infiltrated with empty vector (EV), CitRuby1-CitbHLH1, and CitRuby1-CitbHLH1-CitTRL. (B) Relative expression of CitRuby1, CitbHLH1, and CitTRL in N. benthamiana leaves shown in (A). (C, D) Analysis of the anthocyanin content (C) and the expression of endogenous structural genes of flavonoid pathway (D) in N. benthamiana leaves. (E) The infiltrated citrus fruits (F. margarita) under bright-field (top) and fluorescence channels (bottom). (F) RT-qPCR analysis of the transcripts of CitPH4, CitAN1, and CitTRL in citrus fruits when infiltrated with EV, CitPH4-CitAN1, and CitPH4-CitAN1-CitTRL. (G, H) Measurement of pH (G) and citric acid (H) in citrus fruits when infiltrated with EV, CitPH4-CitAN1, and CitPH4-CitAN1-CitTRL. (I) The soluble PA content in citrus fruits when infiltrated with EV, CitPH4-CitAN1, and CitPH4-CitAN1-CitTRL. (J) Relative expression of H+-pumping genes and PA-associated genes. The data were obtained from at least three biological replicates. Statistical significance: n.s., P>0.001; **P<0.001.
To validate these results in the citrus genus, we performed transient expression assays in citrus fruits from kumquat (Fig. 6E, F). Co-expressing CitPH4 and CitAN1 lowered the acidity level and slightly increased PA contents relative to EV controls, while the concomitant transient expression of CitTRL returned both values to EV controls (Fig. 6G–I). The CitPH4–CitAN1 complex activated the expression of H+-pumping genes (CitPH1 and CitPH5) and PA biosynthetic genes (CitCHS, CitANS, and CitLAR), whereas the added overexpression of CiTRL mitigated their expression levels in kumquat fruits overexpressing CitPH4, CitAN1, and CitTRL (Fig. 6J). In addition, infiltration of CitPH4+CitAN1 raised the PA contents of N. benthamiana leaves and activated the expression of PA-related genes over EV controls, and PA contents and gene expression levels were significantly reduced in when CitPH4, CitAN1, and CitTRL were co-infiltrated (Supplementary Fig. S4E–G).
Taken together, our data showed that CitTRL substantially suppresses the metabolic processes induced by the CitRuby1–CitbHLH1/CitAN1 or CitPH4–CitAN1 complexes.
CitTRL and CitPH4 form an activator–repressor loop
CitTRL and CitPH4 showed similar expression patterns in fruit pulp (Fig. 7A) and seed coat (Fig. 7B) of sweet orange, and relative CitTRL transcript levels increased in CitPH4-OE transgenic calli and kumquat fruits compared with controls (Figs 2B, 6F). To determine whether CitTRL and CitPH4 are similarly co-expressed in other citrus species, we measured their transcript levels in three other pairs of citrus varieties with different acidity levels: wild mandarin (‘JYYJ’) and low acid mandarin (‘BTJ’) (Wang et al., 2018); high acid pummelo (‘HB’) and acidless pummelo (‘WSY’); and sweet lemon and sour lemon (‘YLK’). Indeed, CitTRL was consistently co-expressed with CitPH4, and its expression levels positively correlated with acidity levels (Fig. 7C). Moreover, CitTRL transcript levels were inversely correlated to those of CitRuby1 during green fruit stages but positively correlated after the breaker stage (Supplementary Fig. S5A, B).
Transcriptional activation of CitTRL by CitPH4. (A, B) Expression pattern of CitTRL and CitPH4 in the pulp (A) and seed coat (B) of sweet orange. The pulp of AL is moderately acidic and the seeds of AL accumulate abundant PAs, while the seeds of acidless HAL are white with a low level of PAs. The expression data of CitTRL were extracted from Fig. 1 and used here for comparison. (C) Expression pattern of CitTRL and CitPH4 in the pulp of three pairs of citrus varieties with different levels of acidity. The low-acid group contains sugar mandarin BTJ, acidless pummelo WSY, and sweet lemon, while wild mandarin JYYJ, HB pummelo, and sour lemon YLK produce high-acid fruits. The expression of CitPH4 is normalized using the transcript of CitTRL and is plotted on the right y-axis. The grey columns indicate the expression of CitTRL and black dotted lines indicate the expression of CitPH4. (D) Transient activation of CitTRL by dual-luciferase assays. Different effectors (indicated on the figure) together with TRLpro: LUC were injected into different parts of the same N. benthamiana leaf. EV serves as a control. The data were collected from at least six biological replicates. (E) Direct binding of CitPH4 to the MBS of CitTRL promoter in EMSAs. The unlabeled and mutated probes were added as competitors. The upper bands refer to the protein-labeled probe complex and the lower bands indicate the free probe.
We next asked whether the two R2R3-MYB transcription factors contribute to the regulation of CitTRL expression by conducting a dual-luciferase reporter assay. CitPH4, but not CitRuby1, transactivated the transcription of CitTRL, as evidenced by the luminescence signal detected from the CiTRLpro:LUC reporter (Fig. 7D). In addition, expressing CitAN1 alone failed to induce CitTRL transcription over EV controls. Notably, co-infiltration of CitPH4+CitAN1 increased CitTRL promoter activity over EV controls, but this effect was mitigated by the co-overexpression of CitTRL with CitPH4 and CitAN1 (Fig. 7D). Furthermore, to confirm that CitPH4 binds directly to the CitTRL promoter, we performed EMSAs using the predicted MYB binding site (MBS) labeled with 6-FAM. We detected a shift in the mobility of the labeled probe corresponding to the protein–DNA complex when recombinant CitPH4–MBP was incubated with the labeled probe (P1, P2, P3), but the intensity of the band decreased with an increasing amount of unlabeled probe as competitor (Fig. 7E). Together, our results indicated that CitTRL is transcriptionally induced by CitPH4, and the induced CitTRL, in turn, represses the activation capacity of CitPH4, thus forming an activator-and-repressor regulatory loop to balance citric acid and PA metabolism.
Discussion
The pleiotropic effect of CitTRL on two R2R3-MYB–bHLH complexes in citrus
Citric acid is the predominant organic acid in citrus that confers on the fruits their characteristic pleasant flavor (Hussain et al., 2017; Tadeo et al., 2020). Citrus fruits contain abundant flavonoids, of which anthocyanins and PAs play important roles in determining the organoleptic quality and nutritional values of the fruit (Huang et al., 2020). In this study, we established that CitTRL is a negative regulator that interacts with two bHLH transcription factors, thus interfering with the activation capacity of downstream genes by their respective MYB–bHLH complexes, ultimately leading to lower levels of citric acid, PAs, and anthocyanins (Fig. 8). CitTRL acts as a key node in connecting the regulation of fruit flavor and color. The identification of CitTRL provides new strategies, such as manipulation of CitTRL tissue-specific expression, to improve multiple traits efficiently.
Working model of CitTRL in cross-regulation of vacuolar acidification and biosynthesis of anthocyanin and proanthocyanidin. CitTRL is transcriptionally activated by activator CitPH4. CitTRL suppresses vacuolar acidification and proanthocyanidin biosynthesis by competing with CitPH4 for binding to CitAN1. In addition, CitTRL inhibits anthocyanin accumulation by competing with CitRuby1 for binding to CitAN1 or CitbHLH1 transcription factor. Yellow, brown, and purple boxes represent the critical genes involved in the accumulation of vacuolar acidification, proanthocyanidin, and anthocyanin. Black bold arrows indicate strong activation of target genes, and black thin arrow indicate weak activation. Italic letters indicate genes and regular letters indicate proteins.
Both CitTRL and its homolog CitTRY clustered with CPC-like R3-MYBs with an intact R3 domain and conserved WxM motif (Fig. 1), raising the possibility that CitTRY may function redundantly or cooperatively with CitTRL, which will need further investigation. CPC-like R3-MYBs are proposed to modulate trichome initiation, root hair differentiation, stomatal formation, and flowering (Serna, 2008; Zhu et al., 2009). Recent studies have reported R3-MYBs as negative regulators of flavonoid biosynthesis. Overexpression of poplar (Populus trichocarpa) RML1 or Arabidopsis CPC blocked anthocyanin accumulation in Arabidopsis (Zhu et al., 2009; Hu et al., 2016). High expression of the R3-MYB gene MYBL1 was associated with white flower color in nightshade (Iochroma loxense) (Gates et al., 2018). We demonstrated here by multiple functional assays that CitTRL not only represses the biosynthesis of anthocyanins and PAs, but also has a significant negative effect on the accumulation of citric acid.
The absence of the R2 domain or an active repressor motif suggested that CitTRL is a passive negative regulator. CitTRL interacted with the two bHLH transcription factors CitbHLH1 and CitAN1 (Fig. 4). CitTRL repressed citric acid and PA accumulation by competing with CitPH4 for binding to CitAN1 (Figs 2, 5, 6), and suppressed anthocyanin biosynthesis by competing with CitRuby1 for binding to CitAN1 or CitbHLH1 (Figs 5, 6). Both CitPH4 and CitRuby1 are R2R3-MYB transcription factors. However, the repressive effect of CitTRL was limited, as it failed to fully abolish the accumulation of these metabolites such as in high acid lemon. In addition, the regulatory effects of CitTRL relied on the presence of the R2R3-MYB–bHLH complexes. For instance, overexpression of CitTRL in Arabidopsis resulted in lower AtANR expression in young siliques but not in stems (Fig. 3D; Supplementary Fig. S3C). We attribute this observation to the expression pattern of AtTT8, an Arabidopsis ortholog of CitAN1 required for AtANR expression, being predominantly in young siliques (Nesi et al., 2000). Therefore, we speculate that CitTRL represses AtANR expression by impairing AtTT8 function, and it represses AtDFR and AtANS expression through GLABRA3 (AtGL3) and ENHANCER OF GL3 (EGL3). Moreover, the unaltered metabolite content when overexpressing CitTRL in WT citrus calli further supported the notion that CitTRL functions in the background of R2R3-MYB–bHLH complexes.
The CitPH4–CitTRL regulatory loop
Regulatory loops have rarely been reported in the regulation of citric acid accumulation. CitTRL was specifically transcriptionally activated by CitPH4 but not by CitRuby1. CitTRL was highly expressed in high acid citrus fruits, in which the activator gene, CitPH4, was highly expressed (Fig. 7). Biochemical experiments and the elevated expression levels of CitTRL in CitPH4-overexpressing calli and citrus fruits indicated that CitTRL is transcriptionally activated by CitPH4 to form a feedback loop that prevents vacuolar hyper-acidification and over-accumulation of citric acid in fruit pulp (Fig. 7). This regulatory model also existed in the seed coat for the deposition of PA promoted by CitPH4. Regulatory loops between MYB activators and repressors have been reported for the regulation of flavonoid biosynthesis (Zhou et al., 2019; Yan et al., 2020). The petunia anthocyanin activator DPL induces the expression of the repressor MYBx to prevent excessive anthocyanin accumulation in vegetative tissues (Albert et al., 2014). In peach, PpMYB18 negatively modulates anthocyanin and PA accumulation, and the expression of the encoding gene is also activated by the activator PpMYB10.1 (Zhou et al., 2019). In citrus, a hierarchical regulatory loop has been characterized for the modulation of anthocyanin biosynthesis, which consist of the activator CitRuby1 and the repressor CitMYB3 (Huang et al., 2020). The identification here of the CitPH4–CitTRL regulatory loop expands understanding of the mechanisms underlying citric acid and flavonoid accumulation and provides a new module for manipulating citrus fruit flavor.
The complex regulation mechanism of CitTRL expression for the accumulation of anthocyanins
In this study, the expression of CitTRL positively correlates with that of CitRuby1 and anthocyanin contents during late developmental stages. However, transient expression assays showed that CitTRL is not induced, but rather repressed, by CitRuby1. This result was supported by our observation of the highest expression of CitTRL in the peel of early green fruits, where CitRuby1 expression levels are very low. In Arabidopsis, MYB96 not only activates abscisic acid (ABA)-responsive genes, but also interacts with the histone deacetylase HDA15 to repress ABA negative regulators to further enhance ABA sensitivity (Lee and Seo, 2019). Thus, other partners may be recruited to the CitTRL promoter to turn CitRuby1 into a transcriptional repressor. Furthermore, light signaling is one of the most pivotal environmental factors that influence anthocyanin biosynthesis. High light intensities are reported to induce the expression of anthocyanin activators such as Arabidopsis PAP1/2 and citrus CitRuby1, while repressing the transcription of the Arabidopsis anthocyanin repressor MYB-LIKE 2 (MYBL2) (Dubos et al., 2008; Huang et al., 2019). We found no difference in CitTRL expression in the blood orange or purple pummelo fruits grown in the shade compared with fruits exposed to normal light conditions (Supplementary Fig. S6), indicating that CitTRL expression is not light responsive, and neither is it regulated by CitRuby1 in the light. Thus, the high expression of CitTRL and the low expression of CitRuby1 restricted the biosynthesis of anthocyanins during the green fruit stage, and overaccumulation of anthocyanins during late stages stimulated the expression of CitTRL in a CitRuby1-independent manner. The complex regulatory mechanism of CitTRL expression will need further dissection.
Collectively, this study identified the novel repressor CitTRL of citric acid, anthocyanin and PA accumulation in citrus, and revealed the activator–repressor loop CitPH4–CitTRL as a means to balance citric acid and PA metabolisms. Manipulation of the tissue-specific expression of CitTRL may help produce good-looking fruits with moderate levels of acidity and astringency. The new regulator module described here thus opens the door to improving multiple fruit quality-related metabolic traits efficiently.
Supplementary data
The following supplementary data are available at JXB online.
Fig. S1. Analysis of protein sequence and subcellular localization of CitTRL.
Fig. S2. Overexpression of CitTRL in citrus calli.
Fig. S3. Overexpression of CitTRL represses PA accumulation in the seeds of Arabidopsis.
Fig. S4. Co-expression of CitTRL and CitRuby1-CitAN1 significantly suppressed anthocyanin biosynthesis.
Fig. S5. Expression pattern of CitTRL and CitRuby1 in citrus.
Fig. S6. Evaluation of the effect of light signal on the expression of CitTRL in citrus.
Table S1. List of primer sequences used in this study.
Author contributions
QX and XD designed the research. JH, YX, JF, ZL, LW, and RX performed the experiments. JH, DH, and YZ were involved in the collection of samples. JH analysed the data. JH and QX wrote the article with contributions from LL.
Conflict of interest
The authors declare no competing interests.
Funding
This work was supported by the National Key Research and Development Program of China (No. 2018YFD1000101), National Natural Science Foundation of China (Nos 31925034 and 31872052), and the Science and Technology Major Project of Guangxi (Gui Ke AA18118046).
Data availability
The data supporting the findings of this study are available from the corresponding author (QX), upon request.
Comments