Hyperoside, a flavonoid, induces lnc187 and lnc999 to enhance pigeon pea pollen tube growth and seed production

Abstract

Flavonoid hyperoside boosts okra reproduction; its role in pigeon pea, with poor seed set, remains unclear. We found that hyperoside increases pigeon pea seed set by promoting pollen tube growth, a process that benefits from long noncoding RNAs (lncRNAs). Two lncRNAs, lnc187 and lnc999, are regulated by hyperoside and synergize to regulate target genes. Among them, lnc187 is the key effector; its loss may abolish downstream function. These lncRNAs up-regulate CcGPP (GRAS phosphorylase) kinase expression by binding to its promoter and acting as a scaffold to connect MRP23 and RNA polymerase while simultaneously inhibiting CcGDP (GRAS dephosphorylase) phosphatase activity by binding to key protein sites. Genetic evidence also confirms the relationship between lnc187/lnc999, MRP23, and the downstream CcGPP/CcGDP. This study clarifies the flavonoid hyperoside triggered regulatory axis where lnc187/lnc999 promotes hyperphosphorylated CcGRAS to regulate pollen tube growth and seed set in pigeon pea, offering insights into reproductive development research for high-value woody species.

Hyperoside induces lnc187/lnc999 to recruit MRP23 to CcGPP promoter and inhibit CcGDP, promoting seed setting.

INTRODUCTION

Flowers, as the primary reproductive structures in plants, facilitate the production of progeny. The emergence of flowering plants (1) during the Cretaceous period significantly enhanced food availability for insects, thereby facilitating their pollination of these flowers and consequently increasing seed production (2, 3). This evolutionary development has resulted in a greater diversity of ecological niches, enhanced biodiversity, and increased complexity within ecosystems. The finite life span of flowers plays a crucial role in determining the effective pollination period, which is recognized as a key factor influencing yield (4). Traits associated with the effective pollination period include the duration until full bloom and the growth rate of pollen tubes (5, 6). Consequently, there exists a substantial scientific rationale for investigating flower development with respect to bloom timing and pollen tube growth. The pigeon pea (Cajanus cajan L.), a globally recognized woody perennial and widely cultivated staple crop, provides substantial calories and nutrients to humanity. It holds notable economic importance, particularly in the subtropics, due to its role in staple food production, the abundance of secondary metabolites it contains, and its medicinal value. It not only helps humans with carbohydrates but also plays a key role in maintaining global food security and alleviating poverty. However, adverse effects on pigeon pea pollen germination and development lead to a reduced seed set rate, consequently diminishing yield (7, 8). Therefore, flower development is crucial for maintaining the desirable flowering characteristics and high seed setting rates in pigeon pea. Research suggests that flavonoids play an important role in flower development, influencing not only flower color formation (9) but also fragrance of flowers (10) and pollinator attraction (11), thereby promoting flower reproduction and species perpetuation (12). Hyperoside, a flavonol glycoside, attenuates d-galactose–induced renal aging and injury in humans by inhibiting AMPK-ULK1 signaling–mediated autophagy (13), and prevents cyclophosphamide-induced ovarian damage and fertility decline by inhibiting HIF-1α (hypoxia inducible factor-1α)/BNIP3 (BCL2 interacting protein 3) –mediated autophagy (14). At the same time, it is also integral to several critical developmental processes throughout the plant life cycle, including reproductive development (15), flower blooming (16), and pollen tube growth (17). In our previous study, we found that hyperoside was able to affect the expression of ADF2, ADF2-like, which in turn affected the pollen tube growth and seed setting rate (15–17). Plants primarily respond to flavonoid small-molecule compounds by activating physiological response mechanisms, which encompass perception, signal transduction, regulation of gene expression, and posttranslational modifications of proteins, ultimately lead to notable epigenetic changes (18–20). However, most research has focused on the role of protein-coding genes (21), often overlooking the important role of long noncoding RNAs (lncRNAs). Consequently, there is a scarcity of reports detailing how lncRNAs affect reproductive development in plants.

LncRNAs are defined as being longer than 200 nucleotides and do not encode proteins. At the initial stage when lncRNAs were discovered, it was also called the “dark matter” between genes due to reasons such as its function and biological significance being unclear. At this stage, with the gradual deepening of the study of lncRNAs, based on their genomic coordinates relative to neighboring protein-coding genes, their positional categories can be pictorially described as antisense, intronic, sense-overlapping, intergenic, and bidirectional (22). LncRNAs with different genomic locations usually perform different biological functions. In animals, lncRNAs are highly expressed in a cell- and tissue-specific manner in reproductive organs. They are involved in the molecular processes of sexual reproduction by precisely coordinating the response to intrinsic and extrinsic cellular signals (23). In plants, the genomic location of an lncRNA is often indicative of its potential role or importance in a particular biological process (24–26). In addition, lncRNAs exhibit several relatively stable regulatory patterns. These patterns involve the regulation of gene expression through interactions with DNA or other RNA molecules (27–29). Beyond regulating gene expression, lncRNAs can also function as signaling molecules in the regulation of cell signaling pathways (30, 31). Certain lncRNAs have the capacity to interact with protein signaling molecules, thereby influencing their activity and distribution. For example, lncRNA NEAT1 (nuclear enriched abundant transcript 1) interacts with the protein kinase Pumilio, enhancing kinase activity and thereby regulating cell cycle progression and proliferation (32). Recent investigations into noncoding RNA (ncRNA) have validated that lncRNAs play a role in plant flower development. In roses, the lncRNA LncWD83 promotes flowering by modulating the ubiquitination of the floral repressor RcMYC2L (33). Current studies delineate multiple regulatory modalities of lncRNAs across diverse biological processes and underscore their pivotal roles in floral development; however, the precise molecular mechanisms orchestrating their function remain insufficiently elucidated, particularly in perennial woody species. Therefore, the specific regulatory modes and molecular functions of lncRNAs during flower development in woody plants warrant further exploration.

LncRNAs require interaction with one or more proteins to function. Understanding the protein-lncRNA interaction network may provide important clues for a deeper understanding of intracellular signaling mechanisms (22). Protein modifications serve as a crucial mechanism for regulating protein signaling functions. Among the various covalent modifications in protein posttranslational modifications, phosphorylation is the most important and specific type. It is widely acknowledged as the most notable regulatory modification in both prokaryotic and eukaryotic organisms (34). Kinases can initiate intracellular cascade regulation upon receiving external signals (35). Initial investigations into kinases primarily concentrated on stress responses (36–38), with CIPK24 binding to CBL4 to regulate salt tolerance in Arabidopsis (39) and the CBL1/9-CIPK1 complex modulating drought tolerance in the same species (40). The mitogen-activated protein kinase (MAPK) module plays a crucial role in plant stress tolerance (41). Yeast Dbf2 kinase enhances salt, drought, cold, and heat tolerance when overexpressed in yeast and transgenic plant cells (42). As research on the kinases has progressed, it has become evident that kinases not only are implicated in plant stress resistance but also play notable roles in growth and developmental processes, such as seed germination (43) and pollen tube growth (44). In addition, increasing numbers of interactions between phosphorylation and other protein modifications have been explored. For example, phosphorylation and ubiquitination interact in response to the regulation of protein degradation (45), and balance between phosphorylation and sumoylation to regulate immunogenic iron death (46). As the reverse modification of phosphorylation, dephosphorylation is equally important in plant growth and development (47, 48). A kinase (49, 50) or transcription factor (TF) (51, 52) in its phosphorylated state interacts with a phosphatase to undergo dephosphorylation, thereby altering its function. This mechanism enhances the plant’s resistance to drought and shade, or promotes the accumulation of anthocyanins. However, studies on the balance between phosphorylation and dephosphorylation, particularly regarding the effects of flavonoids and lncRNAs on this balance, have not been reported.

Here, we found that under the influence of the flavonoid hyperoside, lnc187 acts as a critical regulator affecting floral developmental fate. However, when lnc187 was present alone, it merely exhibited a normal rate of seed set. In contrast, when both lnc187 and lnc999 were present, the seed set rate of pigeon pea was greatly promoted. But when lnc999 was present alone, it did not affect seed set, regardless of its expression level. These findings validate that the two lncRNAs function as co-regulators, each exhibiting dual regulatory roles at distinct phases of gene expression: transcription and translation. Specifically, lnc999 and lnc187 interact with the promoter region of the kinase CcGPP (GRAS phosphorylase), thereby significantly enhancing its transcriptional activity. Concurrently, the two lncRNAs interact with the CcGDP (GRAS dephosphorylase) protein, influencing the phosphatase CcGDP’s modifying activity at the translational level. The CcGPP kinase competes with the CcGDP phosphatase for binding to the downstream substrate CcGRAS. Both phosphorylated and nonphosphorylated forms of CcGRAS can regulate the transcript levels of downstream genes, such as CcADF2 and CcADF2-like. These genes, which are directly related to pollen tube growth, have undergone alterations when high levels of phosphorylated CcGRAS are present. Expression of the key cytoskeletal genes that ensure polar growth of pollen tubes CcADF2 and CcADF2-like increased, ultimately leading to an increase in seed setting. These findings highlight the notable role of lncRNA-mediated phosphorylation in plant growth and development, and also provide unique insights for research into woody plant seed setting.

RESULTS

Flavonoid hyperoside promotes plant reproduction

Hyperoside, a flavonoid compound, was found to regulate the expression of key flowering-related genes, specifically, up-regulating genes involved in flowering promotion such as ADF2 and ADF2-like (17). In okra—a high-hyperoside species—our previous work showed that hyperoside orchestrates the entire flowering-to-seed setting transition through coordinated expression of ROP2, ADF1, MYB30, and CDPK6. Here, the question arises whether hyperoside can universally regulate flowering and seed setting across plant species, irrespective of their endogenous hyperoside levels. To address this, we systematically compiled data from species previously reported to either contain or lack hyperoside. Hyperoside content was also measured in the flowers of the screened species. Under favorable growth conditions, hyperoside rather than other flavonoid (fig. S1A) accumulation was positively correlated with seed setting rate (Fig. 1A), indicating that hyperoside functions as a development-responsive signal that enhances reproductive growth when resources are plentiful. On the basis of this extensive dataset, we choose the woody plant pigeon pea MT-Y1 (C. cajan L.), characterized by its low hyperoside levels, as a model system for validating hyperoside’s functional role. Pigeon pea were classified into two groups, WT and MT-Y1, based on their differing hyperoside content. Furthermore, we dissected the complete pigeon pea flower into four parts: petals, stamens, pistil, and pollen. Analysis of the hyperoside content in different tissue parts of the flowers of the two pigeon pea plants validated that the hyperoside content in whole flowers, petals, and pollen of WT was significantly higher than that of MT-Y1 (Student’s t test), whereas there was no significant difference in the content in the pistils of WT and MT-Y1 (Fig. 1B). Although other flavonoids (rutin, isoquercetin, myricetin, quercetin-3-O-glucoside, and quercetin) also exhibit different expression patterns in different floral tissues (petals, stamen, pistil, pollen), there is no obvious correlation between them and seed set rate (fig. S1B). Moreover, the genetic mechanisms underlying high or low seed set differ markedly between genotype and are tightly linked to their intrinsic hyperoside levels (Fig. 1, B to D). The results suggest that the high content of hyperoside in pollen may have a positive effect on the seed setting rate of pigeon pea.

Fig. 1. Interspecific differences in reproductive success were significantly correlated with flavonoid composition and metabolic function.

(A) Evolutionary relationships among 10 species including pigeon pea and their content of hyperoside in the whole floral organ and seed setting rate. Tree scale represents the genetic distance. High stands for 80 to 100% seed setting rate, relatively high for 60 to 80% seed setting rate, medium for 40 to 60% seed setting rate, relatively low for 20 to 40% seed setting rate, and low for 0 to 20% seed setting rate. FW denotes direct determination of content in fresh plant samples. Seed setting rate is defined as the number of effective seed set as a percentage of the total number of flowers, reflecting the reproductive efficiency of the plant from flowering to seed set. To measure this, three branches were randomly labeled at full bloom, the total number of flowers per branch was recorded, and the number of normally developing seeds was counted after seed stabilization. Seed setting rate was defined as the percentage of total seeds to total flowers. At least three biological replicates were set up for each treatment. Values are means ± SD (n = 3). (B) Phenotypes and hyperoside content of different tissues (flowers, petals, stamen, pistil, pollen) of pigeon pea WT and MT-Y1 flowers. DW stands for determination of content after freeze-drying of plant samples. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (C) Phenogram of pigeon pea WT flowers, pod morphology, and seed set. Scale bar, 1 cm. (D) Phenogram of pigeon pea MT-Y1 flowers, pod morphology, and seed set. Scale bar, 1 cm.

Hyperoside boosts pigeon pea pollen and seed

Previous studies from our group showed that the flavonoid hyperoside can promote extended flowering periods, enhanced pollen tube growth, and increased seed setting in okra (Abelmoschus esculentus) (15–17). To investigate its role, we first designed CcUF3GaT1-RNAi (RNA interference) plants using wild-type (WT) plants as a background to silence the UF3GaT gene in pigeon pea, thereby reducing hyperoside synthesis. The results demonstrated that decreasing hyperoside synthesis led to a significant reduction in the seed setting rate (fig. S1C). Inversely, hyperoside treatment ultimately resulted in a higher seed setting rate at MT-Y1 plant (fig. S1D). To clarify changes in hyperoside content in pigeon peas during hyperoside treatment, we measured hyperoside content in pigeon peas treated for different numbers of days (0 to 11 days). The results showed that hyperoside content reached its highest level on the fifth day of treatment and then declined (fig. S1E). In our study, the MT-Y1 pigeon pea was used as a model for the hyperoside backfill experiment. Previous studies have identified the UF3GaT gene crucial for hyperoside biosynthesis (table S1). On the basis of its expression in pigeon pea flowers and its response to hyperoside, CcUF3GaT1 (LOC109816145) was selected as a key gene regulating hyperoside content (fig. S1F). This gene significantly reduces hyperoside content without affecting the expression of other key flavonoids in flower tissues (fig. S1, G to I).

The application of hyperoside significantly increased the number of pollen grains adhering to the pistil of pigeon pea flowers, as observed through scanning electron microscopy (SEM) and fluorescence electron microscopy. Furthermore, both the quantity of pollen grains and the pollen germination rate in hyperoside-treated flowers were found to be 2 to 3 times greater than those in untreated flowers (Fig. 2A). This observation was further corroborated by aniline blue staining. Notably, in vivo, the length of pollen tubes of hyperoside-treated pistils increased by 1000 to 2000 μm compared to untreated pistils (Fig. 2, B and C). In vitro, the length of pollen tubes of hyperoside-treated pistils increased by 50 to 100 μm compared to those of untreated pollen (Fig. 2, D and E). Collectively, these factors led to a significant enhancement in seed quality following hyperoside treatment, evidenced by improved seed fullness, increased diameter, higher thousand-grain weight, and better seed viability (Fig. 2, F and G). However, SEM analysis validated no significant alterations in the seed epidermis or glandular hairs after hyperoside treatment (fig. S2, A and B). There was also no significant change in the key flavonoid content in the seeds (fig. S2C). In contrast, notable differences were observed in seed diameter, thousand-grain weight, and overall seed health (Fig. 2, F and G). Specifically, hyperoside-treated pigeon pea seeds displayed greater fullness, increased diameter, higher thousand-grain weight, and improved seed viability. On the basis of this, we generated a schematic representation of the morphological state of pistil and pollen growth before and after treatment (Fig. 2H). This process contributed to the increase in seed set in pigeon pea with high hyperoside content.

Fig. 2. Hyperoside promotes seed setting in pigeon pea by affecting the pollination efficiency of the pistil.

(A) Flower pistil phenotypes under different treatments under scanning electron microscope and electron microscope. “-Hyperoside” represents UF3GaT-RNAi. “+Buffer” represents buffer without hyperoside. “+Hyperoside” represents in vitro spraying of hyperoside. The percentages in the graph below are the percentage of pollen tubes entering the pistil. Scale bars, 10 μm. (B) Flower pistil stained with aniline blue under different treatments. Scale bars, 100 μm. (C) Pollen tube length in vivo. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (D) Pollen germination under different treatments. Scale bars, 50 μm. (E) Mean pollen tube length in vitro. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (F) Pod status of WT and MT-Y1 after different treatments. Statistics on seed length, width, and thickness are shown above the pods. Data are expressed as mean ± SD (n = 3). Scale bars, 1 cm. (G) Mean pod length and thousand-grain weight. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (H) Diagram of the pattern of pollen tubes from growth to pollination for WT and MT-Y1. There was significant high in the number of pollen grains on the pistil, pollen tube length, and the seed setting rate in the WT pistil.

Hyperoside-induced lnc187/999 modulate reproduction via CcGPP/CcGDP

LncRNAs play a crucial role in the phenotypic regulation of flower development. To elucidate the genetic regulation of seed setting in pigeon pea, we conducted transcriptome sequencing and lncRNA analysis of flowers with high and low levels of hyperoside at different stages of pollen development. Initially, we identified a total of 3,875 lncRNAs from the transcriptome data using the coding-noncoding index (CNCI), coding potential calculator 2 (CPC2), and predictor of lncRNAs and messenger RNAs based on an improved k-mer scheme (PLEK) algorithms (53). Subsequently, we analyzed the fragments per kilobase of transcript per million mapped reads (FPKM) values and the fold change of all lncRNAs to identify those responding to hyperoside treatment and pollen development (fig. S3). On the basis of this analysis, the two most notable responding lncRNAs, lnc187 and lnc999, were selected. Ultimately, we identified the target genes CcGPP and CcGDP as being associated with the specific lncRNAs through genomic location and correlation analysis. Notably, CcGPP and CcGDP exhibited a very strong correlation [Pearson correlation coefficient (Pcc) = 0.99] with both lncRNAs lnc187 and lnc999 (Fig. 3A). Consistent with this, the relative expression levels of lnc187, lnc999, and CcGPP were strongly correlated with reproductive development stages and the hyperoside treatment. Conversely, CcGDP expression showed an inverse correlation with CcGPP. In addition, lnc187 was significantly down-regulated in pollen and pistil of MT-Y1 flowers compared to WT (Fig. 3, B and C). This is consistent with our hypothesis that lnc187 and lnc999 primarily operate in pollen and pistil, thereby affecting pigeon pea seed set. To further explore the relationship between lnc187 and lnc999, we demonstrated by in vitro experiments that the two act independently and do not form a reciprocal complex (fig. S4A). Together, these findings suggest that lnc187 and lnc999 synergistically regulate CcGDP and CcGPP expression in pigeon pea mediated by hyperoside.

Fig. 3. Expression profiling of key lncRNAs and corresponding target genes in different tissues.

(A) Positional relationship between lnc187 and lnc999 and their two corresponding target genes, CcGPP and CcGDP, on the genome. Pcc (Pearson correlation coefficient) is a statistical index that measures the strength and direction of the linear relationship between two variables. In lncRNA research, Pcc is used to quantify the covariance between lncRNA and the expression level of target gene, and its value ranges from −1 ≤ Pcc ≤ 1. When Pcc > 0, it indicates that there is a positive correlation between the lncRNA and the target gene, and that the lncRNA changes in the same direction as the expression of the gene (e.g., the lncRNA promotes the gene transcription, and there may be a direct activation effect). A larger Pcc value indicates a stronger correlation between the lncRNA and the target gene. The Pcc value between each of the two lncRNAs and the target gene was 0.99. (B) Relative expression of lnc187 and lnc999 and their corresponding target genes CcGPP and CcGDP in pollen under different treatments. Values are means ± SD (n = 3). Different lowercase letters indicate significant differences at P < 0.05 (one-way ANOVA). (C) Relative expression of lnc187 and lnc999 and their corresponding target genes CcGPP and CcGDP in pistil under different treatments. Values are means ± SD (n = 3). Different lowercase letters indicate significant differences at P < 0.05 (one-way ANOVA).

Hyperoside induces lnc187/999 to recruit MRP23 and promotes CcGPP transcription

Given that lncRNAs regulate both transcription and posttranslational modifications that occur typically in distinct locations in cells, we investigated how they overcome spatial limitations to exert their functions. Therefore, we assessed the subcellular localization of lnc187 and lnc999. Nucleoplasmic separation experiments showed that both lncRNAs were localized to the nuclear membrane in the absence of treatment. However, after treatment with hyperoside, lnc187 and lnc999 were predominantly found in the nucleus (fig. S4B). To elucidate the regulatory mechanisms by which these two lncRNAs regulate CcGPP, we conducted dual-luciferase reporter gene assays. The results validated that hyperoside treatment increased the interaction strength between the lncRNAs and the CcGPP promoter. Furthermore, this binding affinity was significantly greater when both lncRNAs acted in concert (Fig. 4A). To identify the specific regions of lncRNA that interact with the CcGPP promoter, we divided the promoter region into three segments based on fragment length, designated as segments a, b, and c (Fig. 4B). Similarly, the lncRNA was divided into two segments based on its secondary structure, referred to as segment a and segment b (Fig. 4C). The results demonstrated that the b segment of the lncRNA was capable of binding to the c segment of the CcGPP promoter, and this binding was further enhanced in the presence of both lncRNA b fragments. Additionally, hyperoside treatment further enhanced this binding interaction (Fig. 4, B and C, and fig. S4, C and D). Subsequently, we reconfirmed the result that the simultaneous presence of lncRNA homologs increases the capacity for interactions by ChIRP-qPCR (chromatin isolation by RNA purification–quantitative polymerase chain reaction) experiments (fig. S4E).

Fig. 4. Hyperoside promotes the co-binding of lnc187 and lnc999 to the downstream target gene CcGPP promoter and facilitates its transcription.

(A) Method of vector construction of lnc187, lnc999, and the promoter regions of CcGPP. LUC/REN relative activity of lnc187, lnc999, and the promoter regions of CcGPP. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (B) Interaction of lnc187 and lnc999 with three promoter regions (a, b, and c) of CcGPP. (C) Interaction of lnc187 (a and b) and lnc999 (a and b) with the CcGPP promoter region (c). Lnc187 was divided into two segments, lnc187-a (1 to 835 bp) and lnc187-b (836 to 1949 bp), with no duplication between the two segments. Lnc999 was also divided into two segments, lnc999-a (1 to 751 bp) and lnc999-b (752 to 1295 bp), with no duplication between the two segments. Values are means ± SD (n = 3). Significant differences using Student’s t test: at *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (D) Mass spectrometric detection of proteins interacting with lnc187 and lnc999 under hyperoside treatment. The pie chart represents the proportion of several proteins with a size of about 55 kDa. Among them, the large area in yellow is the RNA transcriptional regulation–related proteins, and there are three kinds of them, namely, ATP synthase subunit β, MRP23, and eukaryotic initiation factor 4A-8. The specific proportions of the three proteins in the RNA transcriptional regulation–associated proteins are separated by black lines. The plant material was pigeon pea flower tissue. (E) Protein level of CcGPP under hyperoside treatment. The plant material was tobacco. (F) The pattern map of lnc187 and lnc999 regulates the promoter region of CcGPP.

Here, we suggest that lncRNAs themselves do not directly influence gene transcription. This raises the question: Do lncRNAs act as transcriptional activators by recruiting some proteins? To investigate whether lncRNA binding to the promoter regulates the transcriptional level of CcGPP by recruiting specific transcription-related regulatory factors, we conducted immunoprecipitation mass spectrometry (IP-MS) experiments. The findings indicated that lnc187 and lnc999 recruit transcription-related regulatory factors, including adenosine triphosphate (ATP) synthase subunit β, mediator of RNA polymerase II transcription subunit 23 (MRP23), and eukaryotic initiation factor 4A-8, thereby enhancing the transcriptional level of CcGPP (Fig. 4D).

Notably, experimental validation confirmed that the most potent of these transcription-related regulators was MRP23, which directly promotes CcGPP transcription by recruiting RNA polymerase, whereas lncRNA acts as a scaffold in this process (fig. S4F). Conversely, inhibiting CcMRP23 expression does not affect the expression of lnc187 or lnc999, but it significantly impairs CcGPP expression (fig. S4G). This further indicates the important role of CcMRP23 in the transcriptional regulation of CcGPP. Moreover, we observed that elevated CcGPP transcript levels are accompanied by a rise in its protein abundance (Fig. 4E). Our results conclusively demonstrate that hyperoside promotes the binding of lncRNAs to the CcGPP promoter region. Additionally, lncRNAs here act as scaffolds to regulate the transcriptional activity of CcGPP by recruiting transcriptional regulators such as RNA polymerase, and hyperoside enhances the interactions between lncRNAs, promoters, and transcription-related regulators (Fig. 4F). These findings emphasize the key role of hyperoside in facilitating these multi-component interactions among lncRNAs, promoters, and transcription-related regulators.

CcGDP binds lnc187/999, enabling lncRNA regulation

To further investigate the function of CcGDP, a target gene of lncRNAs, we performed the following validations. Firstly, Gene Ontology (GO) analysis validated that CcGDP protein belongs to the class of RNA binding proteins. Additionally, predictions using AlphaFold indicated that the RNA binding domain of this protein can interact with lnc187/999 (fig. S4H). Next, we confirmed in vitro that lnc187 and lnc999 interact with CcGDP protein. We purified CcGDP protein with glutathione S-transferase (GST) tags and used the respective antisense strands of the lncRNAs as controls (Fig. 5A). Similarly, in vivo, we enriched the green fluorescent protein (GFP)–tagged CcGDP protein by RNA pull-down experiments and found that it also interacted with both lncRNAs (Fig. 5B). Ultimately, to verify the specific binding sites of CcGDP with the two lncRNAs, we divided each lncRNA into two segments, and the RNA pull-down experiments showed that both segments (lncRNA-a and lncRNA-b) interacted with CcGDP (Fig. 5C). These findings suggest that lnc187 and lnc999 use distinct regulatory mechanisms for their respective target genes, CcGPP and CcGDP. Specifically, lnc187 and lnc999 enhance the transcriptional activity of CcGPP while also directly binding to the RNA binding protein CcGDP, potentially affecting its posttranslational modification (Figs. 4F and 5D).

Fig. 5. Hyperoside promotes the binding of lnc187 and lnc999 to the downstream target gene CcGDP protein.

(A) RNA pull down of lnc187, lnc999, and CcGDP in vitro and percentage of bound CcGDP to input CcGDP. CcGDP proteins were immunoprecipitated using a GST antibody. Antisense strands of antisense RNA for lnc187 and lnc999 were used as control. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (B) RNA pull down of lnc187, lnc999, and CcGDP in vivo and percentage of bound CcGDP to input CcGDP. The CcGDP protein uses GFP tag. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (C) RNA pull down of lnc187-a, lnc187-b, lnc999-a, lnc999-b, and CcGDP in vivo and percentage of bound CcGDP to input CcGDP. The CcGDP protein uses GFP tag. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (D) Pattern map of lnc187 and lnc999 regulation of the CcGDP. Lnc187 and lnc999 were able to bind to CcGDP proteins when induced by hyperoside.

Four TFs identified as downstream substrates of the CcGPP

Previous studies found that hyperoside influences the expression of ADF, MYB, and other genes in okra (15–17). Consistent with these findings, a high expression level of ADF was also observed in the current investigation (fig. S5A). To validate and extend pathways from the previous studies, we tested interactions between ADF2, MYB30, CIPK5, and CIPK14 (selected from our previous work) and CcGPP/CcGDP. We found no interaction (fig. S5, B and C), prompting us to screen for other potentially significant downstream genes (fig. S6).

We conducted a Venn diagram analysis of differentially expressed genes across various developmental stages, validating that approximately 10,218 genes exhibited changes in each pairwise combination (fig. S6A). As pigeon pea pollen progresses in development, an increase in up-regulated genes is observed. Specifically, the pollen maturity exhibit a significantly higher number of differentially expressed genes compared to the pollen immaturity period. Furthermore, the analysis confirms that the pollen maturity period has a greater number of up-regulated genes than the pollen immaturity period (fig. S6B). Finally, 157 genes were screened for up-regulated expression in response to hyperoside (fig. S6, C and D). Additionally, based on GO enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses on these up-regulated genes, we identified nine genes that met the selection criteria (fig. S6E). We conducted phosphorylation site predictions for the nine identified TFs and found that each TF contained phosphorylation sites. From this group, we selected four TFs with high phosphorylation scores and multiple phosphorylation sites for functional validation (fig. S6F).

CcGPP promotes CcGRAS phosphorylation, and lncRNAs curb CcGDP phosphatase activity

Ultimately, we confirmed that CcGRAS is a crucial downstream TF co-regulated by CcGPP and CcGDP, as evidenced by yeast two-hybrid (Y2H), bimolecular fluorescence complementation (BiFC), and dual-luciferase assays (Fig. 6, A to C). Furthermore, CcGPP, CcGDP, and CcGRAS were all localized to the nuclear membrane (Fig. 6D), which reinforces the notion of their interaction. We also validated the interaction between CcGPP and CcGDP with CcGRAS proteins through pull-down and Co-IP experiments (Fig. 6, E and F). We also investigated the functional roles of CcGRAS. The results showed that the CcGRAS gene plays an important role in hyperoside-mediated pollen tube growth, and when the CcGRAS gene was suppressed, pollen tube growth was not affected by hyperoside, and the expression of the downstream ADF gene did not change before and after hyperoside treatment (Fig. 6G). We analyzed the phosphorylation sites of CcGRAS and identified two linked phosphorylation sites that are highly likely to be phosphorylated (Fig. 7A). Notably, the nonphosphorylated form of CcGRAS was significantly diminished under hyperoside treatment (Fig. 7B), suggesting that the hyperoside-induced up-regulation of CcGPP expression influences the modification state of CcGRAS. As shown in Fig. 7B, the ratio of phosphorylated to nonphosphorylated CcGRAS (P+:P−) increased under 50 mM hyperoside treatment. In addition, hyperoside treatment was able to promote seed setting in pigeon pea flowers, so the phosphorylated state of CcGRAS was able to promote seed setting in pigeon pea flowers. In contrast, 0 mM hyperoside treatment showed significantly more nonphosphorylated CcGRAS, suggesting that the nonphosphorylated state of CcGRAS does not promote seed setting in pigeon pea flowers. The CcGDP protein exhibits a dual identity, acting not only as an RNA binding protein but also as a phosphatase. Therefore, we propose that hyperoside treatment also inhibits the phosphatase activity of CcGDP. This, in turn, further leads to an increase in the phosphorylation state of CcGRAS, which in turn leads to increased seed formation.

Fig. 6. Screening that CcGRAS acts as a substrate for the co-action of phosphatases CcGDP and kinases CcGPP.

(A) Y2H assay on the binding of kinases, phosphatases, and substrates. The red font represents two proteins with binding. (B) BiFC assay on the binding of kinases, phosphatases, and substrates. The red font represents two proteins with binding. Scale bars, 50 μm. (C) LUC complementation imaging assays of the interactions of kinases, phosphatases, and substrates. The red font represents two proteins with binding. (D) Subcellular location of CcGPP, CcGDP, and CcGRAS. Scale bars, 50 μm. (E) Pull-down experiments for CcGPP and CcGRAS, and CcGDP and CcGRAS. (F) Co-IP experiments for CcGPP and CcGRAS, and CcGDP and CcGRAS. (G) Pollen tube growth after oligonucleotide transfection of CcGRAS. Numbers in the upper right corner represent average length for pollen tubes. Scale bars, 20 μm. Relative expression of CcGRAS, CcADF2, and CcADF2-like after oligonucleotide transfection. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05.

Fig. 7. Involvement of CcGPP and CcGDP in phosphorylation of CcGRAS.

(A) Map of phosphorylation sites and mutant patterns in CcGRAS. The blue represents phosphorylation score. The number of orbs represents the type of kinase that can phosphorylate the site. (B) Phosphorylation and dephosphorylation status of CcGRAS in vivo under hyperoside treatment. (C) Dephosphorylation of CcGRAS by the phosphatase CcGDP was assayed in vitro and statistics on the percentage of phosphorylated to unphosphorylated amounts of CcGRAS for the different treatments. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (D) Phosphorylation of CcGRAS by the kinase CcGPP was assayed in vitro and statistics on the percentage of phosphorylated to unphosphorylated amounts of CcGRAS for the different treatments. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (E) Phosphorylation of CcGRAS^{S148A S149A} by kinase CcGPP was experimented in vitro. (F) In vivo phosphorylation of CcGRAS in the presence of CcGPP and CcGDP together, and CcGPP or CcGDP only. Lnc187 and lnc999 were also added to obtain the dephosphorylation of CcGRAS by the phosphatase CcGDP when both lncRNAs were present. Statistics on the percentage of phosphorylated to unphosphorylated amounts of CcGRAS for the different treatments. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05.

To further elucidate whether the CcGDP phosphatase can directly dephosphorylate CcGRAS, we purified phosphorylated CcGRAS-GFP from hyperoside-treated tobacco leaves using anti-GFP magnetic beads. Concurrently, CcGDP-GST was also purified. The incubation of CcGRAS-GFP with CcGDP-GST followed by Phos-tag analysis validated a significant increase in the nonphosphorylated form of CcGRAS in the presence of CcGDP. These findings indicate that CcGDP can dephosphorylate CcGRAS in vitro, as shown in Fig. 7C. Through both in vivo and in vitro phosphorylation experiments, we established that CcGPP can phosphorylate CcGRAS (Fig. 7, D and F). By mutating the threonine residues at positions 148 and 149 of CcGRAS to alanine, we inhibited phosphorylation at these loci, demonstrating that CcGRAS could no longer be phosphorylated by CcGPP when these sites were mutated (Fig. 7E). Moreover, comparable outcomes were observed in vivo, which further supports the idea that CcGDP inhibits the CcGPP-mediated phosphorylation of CcGRAS. Notably, the addition of lnc187 and lnc999 resulted in a reduction of CcGDP’s dephosphorylation activity. This observation suggests that lnc187 and lnc999 not only enhance the transcription of CcGPP but also inhibit the phosphatase activity of CcGDP, thereby indirectly promoting the downstream regulatory effects mediated by CcGPP (Fig. 7F). Collectively, these results imply that CcGDP inhibits the phosphorylation of CcGRAS by CcGPP through the dephosphorylation of CcGRAS, ultimately inhibiting CcGRAS function. However, this process is simultaneously inhibited by lnc187 and lnc999.

Inhibiting lnc187/999/CcGPP blocks hyperoside-driven seed set

We investigated the functional roles of two lncRNAs and their target genes CcGPP and CcGDP (figs. S7, A and B, and S8). The inhibition of CcGPP prevented hyperoside, enhancing the pollen germination of pigeon pea (fig. S7, A and B). Similarly, suppressing the lncRNAs lnc187 and lnc999 also abolished hyperoside’s stimulatory effect on pollen germination. Conversely, relieving the inhibitions resulted in a significant hyperoside-induced enhancement of both pollen germination rate and tube length (fig. S7A), consistent with the observed gene expression patterns. Notably, inhibition of lnc999 up-regulates lnc187, while silencing lnc187 does not affect lnc999 expression. This suggests that lnc999 may be functionally compensated by lnc187, i.e., lnc187 can partially substitute for lnc999 to fulfill its function. Moreover, the presence of both lnc187 and lnc999 appears to amplify their downstream facilitatory effect. Conversely, suppressing CcGPP expression had no impact on the levels of either lncRNA and did not alter their response to hyperoside (fig. S7A). Significantly, when both lncRNAs were inhibited simultaneously, the suppression of CcGPP was further enhanced (fig. S8), providing strong evidence for their synergistic function. Consistent with the above results, the overexpression of lnc187, lnc999, and CcGPP in plants significantly improved seed setting rates compared to the control group (Fig. 8, A and C). Furthermore, the expression levels of the two lncRNAs significantly influenced the transcript abundance of their target gene, CcGPP (Fig. 8, A and B). Moreover, the expression of the TF CcGRAS, as well as CcADF2 and CcADF2-like, which were previously implicated in pollen tube growth (17), was also examined (figs. S7 and S8). These findings indicate that the overexpression of lnc187 and lnc999 more effectively promotes the transcription of CcADF2 and CcADF2-like, ultimately leading to an improved seed setting rate (Fig. 8C). Similarly, the overexpression of CcGPP exhibited comparable effects (Fig. 8 and figs. S7 and S8). These results further support our hypothesis that the “hyperoside-lnc187/999-CcGPP-CcGRAS” molecular network regulates pigeon pea seed set.

Fig. 8. Seed setting status and relative expression of key genes after transgenesis.

(A) Seed setting status of transgenic plants and relative expression of relevant genes. The picture shown above shows the average seed pattern per single branch in each plant. The numbers in the lower left corner indicate the percentage in each plant showing this pattern of seed setting. In the chart below, the criterion for seed setting statistics was the average total number of seeds per plant for the same treatment. Scale bar, 1 cm. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (B) Seed setting status of plants and relative expression of related genes after simultaneous overexpression or suppression of both lncRNAs. The numbers in the lower left corner indicate the percentage in each plant showing this pattern of seed setting. In the chart below, the criterion for seed setting statistics was the average total number of seeds per plant for the same treatment. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05. (C) Seed set rate in all transgenic plants. Values are means ± SD (n = 3). Significant differences using Student’s t test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05.

DISCUSSION

Flavonoids, due to their hormone-regulating properties, have been shown to influence reproductive system functions by maintaining hormone levels (54, 55). On the basis of this theory, our study further validated that hyperoside, a specific flavonoid, also exhibits multiple regulatory effects during pigeon pea reproduction. First, hyperoside levels in total flowers, petals, and pollen of WT plants were significantly higher than those of MT-Y1, suggesting that high levels of hyperoside may positively affect seed setting rate by enhancing pollen quality. However, as an important organ of the flower, the pistils may also affect seed set (4). But there was no significant difference in hyperoside content between the WT and MT-Y1 pistil in this experiment. Therefore, we believe that the difference in seed set in pigeon pea may be mainly due to pollen rather than the pistils. Second, the significant enrichment of hyperoside in WT petals suggests that pollination and seed setting may be indirectly affected by regulating the flowering process in pigeon pea plants. In addition, the sensitivity of pigeon pea to hyperoside is associated with two key lncRNAs, lnc187 and lnc999. Through tissue-specific analysis, we found that both lncRNAs were predominantly expressed in pollen and pistils. This suggests that both lncRNAs primarily control seed production by regulating pollen quantity within the pistil and ovary, as well as the growth of pollen tubes. Here, we found that lnc187 plays a primary role. When lnc999 is present alone, there is a modest enhancement of CcGPP transcription, although this effect is not particularly strong. Optimal regulation, however, is achieved only when lnc187 is also present, working in concert with lnc999. The central scientific question that arises from this is: Is the sensitivity of plant lncRNAs to hyperoside universal at the cross-species or cross-tissue level, or is it only an episodic phenomenon within a particular system? To investigate this, we conducted homology comparisons among the genomes of several species responsive to hyperoside, including C. cajan L., Glycine max, A. esculentus, Malus domestica, and Vitis vinifera. Notably, the same lncRNAs were identified in G. max and M. domestica, with the gene IDs LOC121172993 and LOC114824109 (fig. S9). In addition, gene expression analysis of lncRNAs and target genes validated that the expression of these key factors in soybean and apple was also positively correlated with the accumulation of hyperoside, which is consistent with the findings in pigeon pea. Consequently, the lncRNAs identified in this study may serve as common regulators in hyperoside-sensitive plants. This evolutionary conservation highlights flavonoids as universal regulators of reproductive success across different kingdoms. Nevertheless, the specific molecular regulatory mechanisms governing these processes involved remain to be elucidated. Collectively, our data provide unique insights into the mechanisms underlying flower development in woody plants. LncRNAs, as ubiquitous regulatory elements in eukaryotic cells, orchestrate diverse protein-centric processes through multifaceted molecular mechanisms (56). In plants, well-characterized examples include the rice lncRNA MISSEN, which directly regulates endosperm development by interacting with protein complexes (25), and Arabidopsis COOLAIR, which epigenetically silences the flowering repressor FLC to control flowering time (57). The two lncRNAs we identified, lnc187 and lnc999, regulate transcription and posttranslational modification, respectively. These findings align with established paradigms in both animal and plant systems, where lncRNAs act as molecular decoys, guides, or scaffolds (58–60). Notably, we uncovered an unprecedented collaborative mechanism: lnc187 and lnc999 function as synergistic partners, spatially segregated yet functionally interconnected, coordinating transcriptional and posttranslational regulatory layers. This discovery expands the mechanistic repertoire of lncRNAs, validating their capacity for multi-tiered regulation within biological networks. Here, we primarily investigated the effects of two lncRNAs on the flowers on woody plant pigeon pea.

Lnc187 and lnc999 recruit ATP synthase subunit β, MRP23, and eukaryotic initiation factor 4A-8. Notably, MRP23 appears to be the most critical among these factors, as it promotes CcGPP transcription by recruiting RNA polymerase. Lnc187 and lnc999 play a scaffolding role here. The recruitment of these transcriptional regulators strengthens the positive transcriptional regulatory effect of lncRNAs on the CcGPP promoter region. Meanwhile, the energy produced by ATP synthase and other factors may also positively influence downstream CcGPP phosphorylation. Notably, the binding of small amounts of lncRNAs lnc187 and lnc999 to CcGDP in the cytoplasm effectively suppressed the dephosphorylation of downstream substrates by CcGDP. This study provides unique insights into the functional development of lncRNAs. Beyond this study, existing research in plant and animal systems has identified diverse lncRNA action modes, including circRNA-miRNA-lncRNA regulatory networks (61), lncRNA-guided phase separation during vesicle formation (62), and lncRNA-dependent chromatin remodeling mechanisms (63). Collectively, these findings highlight the challenge of rigorously validating the species-specific functional roles of lncRNAs under distinct physiological contexts.

Our study identifies two ncRNAs that function as partners in regulating kinase and phosphatase homeostasis. We suggest that hyperoside promotes the massive nuclear entry of lnc187 and lnc999, thereby facilitating the transcription of CcGPP. Following hyperoside treatment, CcGPP expression demonstrated a pronounced increase, correlating with the phenotypes of increased seed set. During this period, the phosphorylation and dephosphorylation of downstream substrates, mediated by CcGPP and CcGDP, respectively, operate in a competitive manner. In addition, regarding the interactions between CcGPP and CcGDP, based on the studies of phosphatases by others (49, 50), we suggest that CcGDP may exert an inhibitory effect on the phosphorylation of CcGPP, although more specific results require verification by subsequent experiments. Upon analyzing the expression patterns of CcGPP and CcGDP during the development of pigeon pea flowers, we propose a working model. At the initial stages of normal flowering in pigeon pea, CcGPP exhibited a trend of low expression. Subsequently, during the later stages of flower development, the expression of CcGPP gradually increased, enabling the interaction with the downstream specific TF CcGRAS. This interaction culminated in enhanced phosphorylation of CcGRAS by CcGPP and diminished dephosphorylation of CcGRAS by CcGDP under hyperoside treatment, thereby promoting the expression of genes associated with flower development and indirectly affecting seed setting in pigeon pea (Fig. 9). Furthermore, CIPKs (calcineurin B-like interacting protein kinases) are known to interact with a majority of CBLs (calcineurin B-like proteins) (36–38), suggesting that they may function collaboratively as a complex to phosphorylate TFs during reproductive development (fig. S10). A key finding of our study is the identification of downstream substrates for CcGPP and CcGDP, including the TF CcGRAS, which fulfills the criteria for interaction. Nevertheless, the precise mechanisms underlying CcGRAS’s interaction with flowering-related genes and its role in regulating flowering in pigeon pea require further elucidation. Collectively, our results provide unique insights into the mechanisms by which lncRNAs, together with kinases and phosphatases, regulate plant development at the transcriptional and translational levels, respectively.

Fig. 9. Hyperoside affects pigeon pea reproductive development by regulating the lncRNA-CcGPP-CcGRAS module.

In MT-Y1 plants, i.e., under low hyperoside levels, the transcription of lnc187 and lnc999 is reduced, resulting in less binding to the promoter region of CcGPP. CcGPP is able to phosphorylate CcGRAS, while CcGDP is able to dephosphorylate the phosphorylated CcGRAS. In WT plants, i.e., under high hyperoside levels, the transcription of lnc187 and lnc999 increased dramatically, which resulted in more binding to the promoter region of CcGPP. Lnc187 and lnc999 formed a scaffold in the promoter region of CcGPP and recruited mediator of RNA polymerase that in turn recruited RNA polymerase, which ultimately activated and facilitated the transcription of CcGPP. The gray element bound to the CcGPP promoter region with MRP23 is RNA polymerase. The marked increase in CcGPP transcript expression promoted a significant increase in CcGPP protein content. CcGPP was capable of phosphorylating CcGRAS, while CcGDP was capable of dephosphorylating the phosphorylated CcGRAS. However, the higher expression levels of lnc187 and lnc999 were able to bind to CcGDP and reduce the phosphatase activity of CcGDP. Resulting in far more CcGPP-mediated phosphorylation than CcGDP-mediated dephosphorylation. High phosphorylated CcGRAS ultimately promotes pigeon pea seed setting.

MATERIALS AND METHODS

Plant material and growth conditions

Pigeon pea seeds used in this experiment were obtained from Northeast Forestry University in 2018. New seeds are harvested annually, and the seeds used in this study were specifically harvested in 2022. The cultivation method was adapted from a previously established protocol for pigeon pea in our laboratory, with minor modifications (64). Initially, the seeds were soaked in deionized water for 24 hours at room temperature, followed by sterilization with sodium hypochlorite. Subsequently, the seeds were placed on sterile gauze that was moistened with a one-fifth concentration of Hoagland’s solution and incubated at 37°C until germination occurred. Once germinated, the seeds were planted in soil, which had been sterilized at 121°C for 2 hours before use. The plants were then cultivated in a greenhouse under a temperature regime of 30°C/25°C (day/night) for a duration of 2 to 3 months. Upon reaching a height of 30 to 50 cm, the seedlings were transferred to field conditions located in Sanqing Yuan, Haidian District, Beijing, China, for further cultivation. Following the onset of flowering, blooms and other plant tissues were collected and stored at −80°C for subsequent analysis. Pollen was collected, dried, and stored at −20°C for future use.

The two pigeon pea genotypes used in this experiment were WT and MT-Y1. These two plants were screened by assaying the content of hyperoside. Hyperoside content in the floral organs was determined by high-performance liquid chromatography (HPLC). Compared with WT, MT-Y1 showed a significant decrease in seed setting rate, as well as a significant decrease in hyperoside content in its total flowers, petals, and pollen, while the rest of the cultivation conditions remained the same.

Plasmid construction and plant transformation

The complete sequences of lnc999, lnc187, CcGPP, CcGRAS, CcADF2, CcADF2-like, and others were cloned using the corresponding primers as detailed in table S2, following the methodology established (16). All gene accession numbers are listed in table S3. The genes lnc999, lnc187, CcGPP, CcGDP, and CcGRAS were cloned into the vectors pFGC5941, pROK2, and pCAMBIA1390 through homologous recombination, using the Gateway Cloning Protocols (Thermo Fisher Scientific, Waltham, MA, USA) and the BP Clonase II enzyme mixture. These vectors are regulated by the 35S/UBQ promoter. These plasmids were subsequently transformed into Escherichia coli DH5α (Shanghai Weidi Biotechnology Co. Ltd., catalog no. DL1001S) for conservation. The relevant genes contained in each plasmid vector were transformed into Agrobacterium tumefaciens GV3101 (Shanghai Weidi Biotechnology Co. Ltd., catalog no. AC1001) using the heat shock method.

This investigation builds upon our foundational research in Agrobacterium-mediated hairy root transformation using the model legume pigeon pea. Through iterative methodological refinements of in planta injection-based transformation protocols over the past decade, our laboratory has now achieved protocol standardization for generating high-efficiency, heritable transgenic events in recalcitrant woody species, with transformation success rates exceeding 85% across three generations (65–67). Initially, Agrobacterium strains overexpressing CcGPP and lncRNA were centrifuged and subsequently resuspended in a resuspension solution consisting of 5% sucrose and 0.02% Silwet L-77. The resuspended solution was then injected into pigeon pea stem segment sites and bud sites by syringe aspiration, and transgenic shoots and roots appeared after 2 to 3 weeks. The nontransgenic portion was cut off, only the transgenic portion was cultured until the young shoots grew to a large size and flowered, the seed setting rate was counted, and samples of the flowers of different pigeon pea transgenic plants were stored at −80°C for further experiments (fig. S11). Additionally, we found that infiltrating the transgenic points of plants with trace elements and exogenous small peptides such as CLE20 can promote the regeneration and transformation efficiency of plants to a certain extent.

Detection of key floral flavonoids including hyperoside

The content of hyperoside and other key flavonoids (rutin, isoquercetin, myricetin, quercetin-3-O-glucoside, and quercetin) in flowers was determined in different treated samples and at different stages of pigeon pea flowers. For the screening of these five flavonoids, we referred to previous studies on flavonoid compounds in okra and pigeon pea and selected five flavonoid compounds that may be present in flowers for verification. The samples were analyzed using an Agilent 1260 liquid chromatography system equipped with a Luna C18 column (250 by 4.6 mm, 5 μm Inner Diameter; Phenomenex) at a wavelength of 254 nm. The elution solvent consisted of mobile phase A (0.1% formic acid solution) and mobile phase B (acetonitrile). Gradient elution conditions were as follows: 0 to 10 min, 20% to 30% B; 10 to 25 min, 30% to 95% B. The total run time was 25 min with a constant flow rate of 1 ml/min. The injection volume of all samples was 20 μl, and the column temperature was maintained at 35°C. The flavonoid content was quantified by using standards (Sichuan Weikeqi Biological Technology Co. Ltd.).

Hyperoside supplementation and inhibition methods

In accordance with our previous research findings (16), we prepared a hyperoside solution for external application. A volume of 100 ml of the solution was administered per plant. The average height of the treated plants was recorded at 1.5 m. Spraying was conducted every 3 days, while the control group received only a buffer solution. The flowers that were treated with the spray were subsequently collected and stored at −80°C for further experimentation.

On the basis of previous studies on okra (16) and loquat (68), we screened all UF3GaT1 genes in pigeon pea according to their structural domains. In addition, we performed a fine screening based on the E value and percent identity of this gene with the UF3GaT1 gene in okra and loquat, and screened five compliant genes. Subsequently, we analyzed the expression of all five screened pigeon pea UF3GaT1 genes in transcriptome data, and screened UF3GaT1 (LOC109816145), which has the highest expression in floral tissues and responds to hyperoside and floral developmental processes, as a major functional gene for RNAi experiments, and subsequently performed UF3GaT1 expression assay and hyperoside content measurements to clarify the regulatory role of UF3GaT1 on hyperoside.

SEM assay

We used an S-3400N II scanning electron microscope (Hitachi, Japan) to analyze the samples. Initially, the pistils were immersed in phosphate-buffered saline (PBS) and rinsed two to three times. Subsequently, the samples were treated with pentylene glycol fixative and subjected to vacuum for a duration of 1 hour. Following this, the samples underwent a gradient dehydration process using ethanol concentrations of 50%, 70%, 90%, 95%, and 100% before being placed in tert-butanol. The drying of the samples was conducted using Freeze Dryer ES-2030. Subsequently, the desiccated plant material underwent sputter-coating for a duration of 5 min to achieve uniform gold coverage. The samples were then affixed to a sample holder using double-sided conductive tape. The analysis was performed under the following conditions: an accelerating voltage of 3.0 kV and an operating temperature of 25°C.

Aniline blue staining assay

In the aniline blue staining assay, pistils from flowers that had been open for more than 24 hours were selected and subjected to vacuum fixation in a solution of acetic acid and ethanol (1:3) for a duration exceeding 2 hours. Subsequently, the flowers were rehydrated through a series of ethanol solutions (70%, 50%, and 30%) followed by deionized distilled water (ddH₂O), with each step lasting 15 min. After an overnight treatment with 8 M sodium hydroxide (NaOH), the samples were rinsed twice with ddH₂O and stained with an aniline blue solution (0.3% decolorized aniline blue, 108 mM K₃PO₄) for a period exceeding 2 hours. The stained samples were then examined under a fluorescence microscope (Leica MDi8) equipped with an ultraviolet filter, where pollen tubes that enter the pistil were deemed to have successfully penetrated the pistil. This analysis was conducted in triplicate at a minimum.

Pollen germination rate and pollen tube length analysis

Fresh pollen from pigeon pea was collected and subsequently immersed in a pollen germination solution composed of 40% sucrose, 0.025% boric acid, and 0.03% calcium nitrate. The pollen was allowed to germinate for 1 hour at a temperature of 25°C in a dark environment. Following this initial incubation period, a hyperoside standard was introduced at a concentration of 50 mg/liter, and the incubation continued for an additional 3 hours. Pollen germination and pollen tube length were then assessed using a microscope, with a minimum of three biological replicates conducted for each treatment.

Antisense oligonucleotide transfection

The design of phosphorothioate antisense oligodeoxynucleotides (as-ODNs) and sense oligodeoxynucleotides (s-ODNs) targeting lncRNA and other related genes was conducted (17, 69). Pigeon pea pollen was first evenly sown in a pollen germination solution and incubated for 1 hour at 25°C. Subsequently, transfection reagents at a concentration of 10 mM as-ODN and s-ODN were added to the pigeon pea pollen germination solution containing the germinated pollen, and incubation was continued for 3 to 4 hours at 25°C. Finally, the pollen was observed by fluorescence electron microscopy. A subset of the pollen was used to assess the pollen germination rate and the length of the pollen tubes. The pollen from the different treatments was collected by centrifugation and stored at −80°C for further experiments.

Nuclear plasmid separation experiment

Pigeon pea flowers were collected and subjected to lysis using 1 ml of Plant Cell Lysis Buffer (Beyotime Biotechnology, Shanghai, China). The methodology has been modified as previously described (70). Following a 10-min incubation on ice, the samples were subjected to centrifugation for 3 min at 4°C at a speed of 1000 rpm. The resulting supernatant enriched the cytoplasmic fraction, while the nuclear fraction was contained within the precipitate. The precipitate, representing the nuclear fraction, was collected and resuspended in lysis buffer. To isolate the nucleolus, the precipitate was dissolved in Resuspension Buffer I, which consists of 340 mM sucrose and 5 mM MgCl₂, and sonicated until complete dissolution was achieved. Subsequently, an equal volume of Resuspension Buffer II, containing 880 mM sucrose and 5 mM MgCl₂, was added, followed by centrifugation at 2000 rpm for 20 min at 4°C. The supernatant enriched the nucleoplasmic fraction, while the nucleolus fraction was found in the pellet. Finally, RNA was extracted from the cytoplasm, nucleoplasm, and nucleolus using TRIzol.

RNA-sequencing and lncRNA analysis

RNA sequencing was conducted at three distinct time points—half-open, full-open, and wilting—and across two treatment conditions: hyperoside treatment and control. Since the half-open samples served as controls, only one set of samples without any treatment was sent. Each sample had three biological replicates. Therefore, 15 sequencing samples were finally generated. The sequencing was executed by Wuhan Benagen Tech Solution Co. Ltd. (Wuhan, China; http://www.benagen.com) using an Illumina Novaseq 6000 sequencer. Subsequent to sequencing, data were filtered to yield clean reads. The filtered transcriptome sequences were aligned with established reference genes of pigeon pea to facilitate differential gene expression analysis, GO enrichment analysis, and KEGG pathway classification. Newly identified transcripts underwent coding potential prediction using CNCI (version 2.0; default parameters) (71), CPC2 (version standalone_python3 v1.0.1) (72), and PLEK software (Beijing Yung Biotechnology Co., Ltd., Beijing, China), leading to the identification of eligible lncRNAs (53).

RNA isolation and quantitative reverse transcription PCR

The objective of this study was to quantify the expression levels of lncRNAs and genes across various treatments and tissues. RNA was extracted using the Promega RNA extraction kit. The concentration and quality of the extracted RNA were assessed using a Thermo NanoDrop spectrophotometer (Thermo Scientific, USA). Subsequently, the RNA was reverse-transcribed using a reverse transcriptase kit from TaKaRa (Japan). Gene transcription levels were determined using gene-specific primers (refer to table S4) in conjunction with SuperStar Universal SYBR Master Mix (CoWin Biosciences, Jiangsu, China). Actin expression served as an internal control, and finally, the data were analyzed using the 2^−ΔΔCt method (64).

AlphaFold molecular interaction analysis

AlphaFold is an artificial intelligence program developed by DeepMind. The most recent iteration, AlphaFold3, enhances its capabilities to predict the structures and interactions of a diverse array of biological molecules, encompassing proteins, nucleic acids, small molecules, and ions (73). Here, we used the AlphaFold Server to predict interaction potential across various combinations, including protein-protein, protein-DNA, RNA-DNA, and RNA-protein interactions.

Subcellular localization

The Agrobacterium solution was subjected to centrifugation to eliminate the supernatant and subsequently resuspended in a buffer containing 0.2 mM acetosyringone, 10 mM MgCl₂, and 10 mM MES. This resuspension was introduced into tobacco leaves using a 1-ml syringe and incubated in a greenhouse at a temperature range of 22° to 24°C for a duration of 48 hours. The fluorescence of the treated tobacco leaves was examined using a Leica confocal microscope (Leica SP8).

BiFC assays

CcGPP, CcCBLs, and other genes were cloned into the pSPYNE and pSPYCE vectors, which contain the coding sequences for yellow fluorescent protein (YFP), to create N-terminal or C-terminal fusion proteins, respectively. BiFC assays were conducted as previously described (16).

Y2H assays

Genes such as CcGPP were cloned into the pGADT7 vector using Eco RI and Bam HI restriction sites. Similarly, the gene CcGRAS was cloned into the pGBKT7 vector using Nde I and Bam HI sites, with the corresponding primers detailed in table S2. A minimum of three independent Y2H assays were conducted, yielding consistent representative results. The activation domain (AD) and binding domain (BD) vectors were transformed into the yeast strain Y2HGold. The pGBKT7-SV40 and pGADT7-p53 constructs served as positive controls, while the empty pGADT7 and pGBKT7 vectors were used as negative controls. Following a growth period of 4 to 6 days on SD/-Leu-Trp medium at 30°C, the clones were subsequently transferred to selective medium (SD/-Leu-Trp-His-Ade) at the same temperature for an additional 3 to 4 days. Finally, positive clones (cotransformed) were then spotted at diluted concentrations of 1:10, 1:100, and 1:1000.

Luciferase complementation assay

In the transient luciferase (LUC) activity assay, 2000–base pair (bp) promoter sequences of CcGPP were inserted into the pGreen II-0800-LUC vector to generate the proCcGPP: LUC plasmid. To investigate gene expression regulated by lncRNAs or proteins, the lncRNA sequence driven by the 35S promoter was cloned into the pGreen II-62-SK vector. Both vectors were subsequently transfected into A. tumefaciens strain GV3101 (pSoup-p19) (Shanghai Weidi Biotechnology Co. Ltd., catalog no. AC1003). Finally, the Agrobacterium was co-injected into tobacco plants. Fluorescence intensity was measured using a Proteoluminescence Imaging System (Tanon 5200Multi). The dual-luciferase reporter system was used to quantify the intensity of activation of the CcGPP promoter by lncRNA, reflecting the ability of lncRNA to bind to the promoter region (33). With appropriate modifications to better meet the requirements of this experiment, the CcGPP promoter was inserted into pGreen II-0800-LUC carrying the Firefly-LUC reporter gene, driven by the promoter to be tested (RcFT promoter), and, at the same time, the sea kidney luciferase (Renilla-LUC) as an internal reference to normalize differences in transfection efficiency. LncRNA inserts pGreen II-62-SK for overexpression of lncRNA, whose 35S promoter drives lncRNA transcription. The LUC/REN ratio reflects the promoter transcriptional activity, and an increase in the ratio is indicative of binding and activation of the promoter.

Protein purification

E. coli BL21 (DE3) was induced overnight at a temperature of 16°C using 250 mM isopropyl-β-d-thiogalactopyranoside (IPTG). The target proteins were subsequently eluted and purified from the cell lysate using a 10-ml Ni–nitrilotriacetic acid (NTA) agarose or GST column (Qiagen). Finally, the purified target proteins were snap-frozen in liquid nitrogen and stored at −80°C for future experimental applications.

Pull down and RNA pull-down assays

CcGPP protein and CcGDP protein were subjected to pull-down experiments with CcGRAS protein. An appropriate amount of CcGRAS-GST protein solution was mixed with glutathione agarose beads, and the mixture was incubated at 4°C for 2 hours to allow the proteins to bind fully to the beads. CcGPP-His protein and CcGDP-His protein were added separately to the affinity chromatography resin bound with CcGRAS-GST protein. The mixture was incubated at 4°C overnight with constant agitation to allow the two proteins to interact fully. After incubation, the reaction system was washed multiple times with washing buffer to remove unbound or nonspecifically bound proteins. Finally, elution buffer is added to elute the CcGPP-His protein and CcGDP-His protein specifically bound to the CcGRAS-GST protein from the affinity chromatography resin. The eluate is collected, and the eluted proteins are separated and detected by SDS–polyacrylamide gel electrophoresis (PAGE) electrophoresis, followed by Western blotting analysis.

The DNA sequences of lnc999 and lnc187 were used as templates for RNA transcription, using the T7 High Yield RNA Transcription Kit (Nanjing Novozymes Bioscience and Technology Co. Ltd.). The resulting lncRNA was biotinylated and subsequently isolated using the Pierce RNA 3′ End Desthiobiotinylation Kit and the Pierce Magnetic RNA-Protein Pull-Down Kit (Thermo Fisher Scientific, Waltham, MA, USA). Following this, the biotinylated lncRNA was incubated with either purified protein or crude plant protein at 4°C for a duration of 30 to 40 min, after which elution was performed at 37°C for 15 to 30 min. The RNA binding proteins were identified through gel electrophoresis and IP-MS.

Co-IP assays

Co-immunoprecipitation (Co-IP) experiments were performed with CcGPP, CcGDP, and CcGRAS proteins. CcGPP-HA and CcGDP-HA were mixed with CcGRAS-flag and expressed in tobacco cells. Tobacco leaves were then harvested, ground in liquid nitrogen, and mixed with an appropriate amount of plant tissue lysis buffer. The mixture was lysed on ice for approximately 30 min. After lysis, the lysate was collected into prechilled centrifuge tubes and centrifuged at 4°C at 12,000 rpm for 10 min. The supernatant was then used for subsequent immunoprecipitation experiments. An appropriate amount of lysate was added to a centrifuge tube containing Protein A + G Agarose beads (Beyotime, P2055), along with an appropriate amount of the corresponding antibody (Flag antibody, 2 to 5 μg). The mixture was incubated overnight at 4°C with gentle rotation to allow the antibody to bind fully to the Protein A + G Agarose beads. After incubation, the immunoprecipitation complex was washed multiple times with prechilled washing buffer. Finally, an appropriate amount of elution buffer was added to elute the target protein. The eluted protein complex was separated by SDS-PAGE electrophoresis, and Western blotting analysis was performed.

ChIRP-qPCR

ChIRP was conducted as previously described, with certain modifications (26, 74). Antisense lncRNA probes were designed to target the full lengths of lnc187 and lnc999, and were biotinylated at the 3′ end (Thermo Fisher Scientific, Waltham, MA, USA). The lncRNA justice probe served as a negative control. DNA and RNA were extracted from pigeon pea flower samples treated with CK/buffer/hyperoside. Initially, biotin-labeled RNA probes were combined with 30 μl of streptavidin affinity magnetic beads and incubated at 4°C overnight. Subsequently, the magnetic bead complex was mixed with the extracted DNA or RNA from the various treatments and incubated for 1 hour at room temperature. Finally, the DNA or RNA bound to the RNA probe was eluted using an appropriate elution solution. The resulting samples were then subjected to qPCR experiments.

In vitro kinase activity assay and CcGRAS phosphorylation assay

The prediction of phosphorylation sites in proteins was conducted using the NetPhos 3.1 tool (https://services.healthtech.dtu.dk/services/NetPhos-3.1/). For the in vitro phosphorylation assay, 0.5 to 2.0 mg of purified protein were introduced into a kinase reaction buffer composed of 20 mM tris (pH 7.2), 6 mM MgCl₂, 1 mM CaCl₂, 1 mM dithiothreitol (DTT), 2 mM ATP, and 1× Cocktail I. The reaction was incubated for 2 hours at 30°C and subsequently terminated by the addition of 5× SDS-PAGE loading buffer. Finally, the samples were analyzed using 12% SDS-PAGE gels and subjected to immunoblotting (75).

Protein dephosphorylation assays

In vitro dephosphorylation assays, GFP-tagged CcGRAS, and mCherry-tagged CcGPP were encoded in tobacco leaves and treated with hyperoside to clarify the effect of hyperoside on CcGRAS phosphorylation. Hyperoside enables CcGPP and CcGRAS to complete phosphorylation reactions in vivo. Total proteins extracted from UBQ: CcGRAS-GFP transgenic tobacco leaves were enriched using BeyoMag Anti-GFP (TransGen Biotech, HT801-01) Magnetic Beads. The extracted CcGRAS that has been phosphorylated was co-incubated with CcGDP-GST purified protein in a shaker (1200g, 30°C) for 2 hours. Protein loading buffer and sample are treated at 100°C for 10 min. Finally, the mixtures were analyzed by immunoblotting.

In in vivo dephosphorylation assay, CcGRAS-GFP was infiltrated into Nicotiana benthamiana leaves by injection. Subsequently, hyperoside-treated and untreated tobacco leaves were lysed and the CcGRAS-GFP protein was purified. Finally, the protein upload buffer and samples were incubated at 100°C for 10 min before the mixtures were directly analyzed by immunoblotting. In a semi–in vivo dephosphorylation assay, CcGPP-GFP and CcGDP-mCherry were mixed in equal proportions and co-infiltrated into N. benthamiana leaf blades by injection. Total proteins were extracted from N. benthamiana leaf blades after 48 to 72 h. Equal volumes of purified CcCIPK-GFP, CcGDP-mCherry, and CcGRAS-GST were mixed separately and then incubated in a shaker (1200g, 30°C) for 2 hours. Protein loading buffer and sample are treated at 100°C for 10 min. Finally, the mixtures were analyzed by immunoblotting.

Mn²⁺–Phos-tag SDS-PAGE

Protein samples were subjected to trichloroacetic acid (TCA) precipitation and separated on 10% conventional SDS-PAGE gels. A 5 μM Phos binding reagent (Phosbind) acrylamide (APExBIO, USA) was incorporated into the gel. The electrophoresis was conducted for a duration of 1.5 hours (49). The gel was then washed three times for 10 min each with gentle agitation in Normal Transfer Buffer supplemented with 10 mM EDTA, followed by an additional 10 min of washing in Normal Transfer Buffer. The gel was used for immunoblot analysis. Finally, the fusion protein was detected using an anti-GFP antibody (TransGen Biotech, HT801-01). The upper band detected in the gel corresponds to the phosphorylated form (P+), which migrates more slowly than the nonphosphorylated form of CcGRAS-GFP (P−).

Immunoblotting

Proteins were extracted from the leaves of N. benthamiana using Plant Cell Lysis Buffer for Western blotting and immunoprecipitation (Beyotime Biotechnology, Shanghai, China). The proteins were separated via SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane. Following transfer, the membranes were blocked with a solution of 5% (w/v) semi-skimmed milk and bovine serum albumin (BSA) powder at room temperature. The extracted proteins were subsequently immunoblotted (76) by primary antibody incubation with either Sigma Silk Threonine Pan Antibody (Sigma-Aldrich, SAB5701877), anti-GFP (TransGen Biotech, HT801-01), anti-GST (TransGen Biotech, HT601), anti-His (TransGen Biotech, HT501), anti-mCherry (Beyotime, AG9L4, RRID:AB_2891288), anti–hemagglutinin (HA) (TransGen Biotech, HT301-01), anti-FLAG (Beyotime, AF519), anti–c-Myc (TransGen Biotech, HT101-01), or anti–glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (TransGen Biotech, HC301-01). The secondary antibody was then incubated with the corresponding Goat Anti-Rabbit IgG (H+L) (HS001-01) or Goat Anti-Mouse IgG (H+L) (HS002-01). Finally, immunoblotting profiles were generated using a ChemiDoc Touch chemiluminescence imaging system (Bio-Rad, Hercules, USA).

Statistical analysis

All data in this study were obtained from at least three independent replicated experiments. The repeat represents a biological repeat. Data were analyzed using Prism software (GraphPad, San Diego, CA, USA). Quantitative data for expression analysis and physiological measurements were examined for statistical significance using Student’s t test and one-way analysis of variance (ANOVA) followed by Duncan’s multiple range test to assess significant differences for pairwise comparisons [*P < 0.05, **P < 0.01, ***P < 0.001, not significant (ns) P > 0.05].

Supplementary Materials

This PDF file includes:

Figs. S1 to S11

Tables S1 to S4