Fruits were harvested from H. undatus (Vietnam No. 1) 35 days post artificial pollination (DAAP) at Miaoshui Base, located on Luoge Road in Ruyang County, Henan Province, China. These collected fruits served as samples. A total of eighty H. undatus fruits exhibiting consistent shape, size, and color were chosen for the study. The fruits were kept in a controlled environment at 25°C with a relative humidity of 85%. Throughout the storage duration, the exocarp’s flavonoid levels were assessed every three days. Trypsin (bovine source, 5 × 10⁵ units/g) was purchased from Amersco, USA.

Trypsin was used as a preservative, and the fruits were divided into a control group and a trypsin group. Five H. undatus fruits are selected in each group. The treatment conditions for trypsin were the same as those previously reported, with a final concentration of 2.41 × 10⁻⁶ mol/L [19]. The control group was soaked in distilled water following the same procedure. Samples were taken for transcriptome sequencing after one week of H. undatus fruits storage. The sequencing process is carried out according to our previous research. Differential expression analysis of the transcriptome was performed using the DESeq2 software (|log₂FC| = 1.05, p-value <= 0.05) [20].

The molecular weight and theoretical isoelectric point of the protein were predicted using the ExPASy online website (https://web.expasy.org/protparam/) (accessed on 18 February 2025) [21]. Similar sequences were searched in the NCBI database. Multiple amino acid sequences were aligned using DNAMAN software, and a phylogenetic tree was constructed using MEGA software [22].

The total RNA of H. undatus peel was extracted using the RNAprep Pure Micro Kit (DP432, TransGen Biotech Co., Ltd., Beijing, China), and cDNA first strand synthesis was performed using the TransScript^® II One-Step gDNA Removal and cDNA Synthesis SuperMix (AH311-02, TransGen Biotech Co., Ltd., Beijing, China). Amplification was conducted using specific primers for HuLBD1. The primers used for VIGS were 5^′CCGGAATTCATGGTATATGAAGCGAATGCTC-3^′(sense) and 5^′CGCGGATCCTCAAGTCCAAAGAGGCTCCC-3^′(antisense), which were synthesized by Genscript Biotech Corporation. The PCR product was subjected to agarose gel electrophoresis, and the target band was recovered and ligated with the pEASY^®-Blunt Cloning vector (CB101-01, TransGen Biotech Co., Ltd., China). The ligation product was then transformed into Escherichia coli TOP10 competent cells by heat shock. Single colonies that were confirmed as positive by PCR were sent to Sangon Biotech Co., Ltd. (Shanghai, China) for sequencing. After confirming the correctness of the sequencing, the product was digested with EcoR I and BamH I restriction enzymes and ligated with T4 DNA Ligase (FL101-01, TransGen Biotech Co., Ltd., China) to the pTRV2 vector [23], constructing the pTRV2-HuLBD1 recombinant plasmid. After PCR verification, the recombinant plasmid was transformed into Agrobacterium GV3101 to induce gene silencing in H. undatus.

In accordance with the approach described by Zhang et al. [24], a solution known as the infection buffer was created, consisting of 10 mM MES, 10 mM MgCl₂, and 200 μM AS. For the injection treatment, the peel of H. undatus was immersed in a bacterial suspension with an OD600 of 0.6 using a 1 mL syringe [3,25]. The fruits infected with a 1:1 Agrobacterium infection solution carrying pTRV1 and pTRV2-HuLBD1 were used as the experimental group (LBD-i), while the fruits infected with a 1:1 Agrobacterium infection solution carrying pTRV1 and empty pTRV2 were used as the control group (CK). Thirty-five H. undatus fruits are selected in each group. Both groups of H. undatus were placed in a 25°C incubator. The color, shape, and dehydration-induced wrinkles or lesions of the fruit were observed every 24 h, photographed, and recorded. The water loss rate was measured [23]. After one week of storage, the treated H. undatus peel was cut with a scalpel, rapidly frozen in liquid nitrogen, and then preserved at −80°C for subsequent transcriptomic sample preparation.

The total flavonoid content was assessed using the aluminum chloride colorimetric method [26]. The soluble protein and soluble sugar content were determined using methods previously studied [27].

GSEA was used as a statistical method to estimate the enrichment between a query gene set and reference gene features [28]. KEGG enrichment analysis was performed according to the method of Candar-Cakir et al. [29].

The regulatory network between proteins was constructed using Cytoscape software. The CytoHubba plugin was used to score and rank the nodes, while MCODE was used to calculate and collect clusters in the protein-protein interaction network [19].

Reverse RT-qPCR was conducted according to the method of Li et al. with β-Actin used as an internal reference for gene expression normalization [30]. The relative copy number of genes was calculated using the method [6].

Statistical analysis was performed using the SPSS (version 12.0.1) statistical software package. The Wilcoxon test, a non-parametric test, was utilized to analyze differences between the entire sample groups. OriginPro 2024 (version 10.1.0.178) analyzed data and drew graphs [31].

After trypsin treatment, the transcriptome data showed enrichment for a total of 70 LBD-related genes, including 31 upregulated genes and 39 downregulated genes(Table S1). Among these, TRINITY_DN52324_c0_g1 exhibited the largest change in expression among the downregulated genes. The protein sequence was obtained from NCBI (https://www.ncbi.nlm.nih.gov/) (accessed on 18 February 2025) and analyzed using ExPASy. Its open reading frame (ORF) is 546 bp in length, encoding a polypeptide of 182 amino acids. The calculated molecular weight was 20.06 kDa, with a theoretical isoelectric point of 5.16. Using the Blast online alignment tool in NCBI (Fig. 1A), six proteins with high homology to TRINITY_DN52324_c0_g1 were retrieved, including those from Arabidopsis (NP_172268.1), Oryza sativa (XP_015638880.1), Solanum lycopersicum (XP_004251265.1), Nicotiana tabacum (XP_016487519.1), Amaranthus tricolor (XP_057521184.1), and Chenopodium quinoa (XP_021715798.1). A multiple sequence alignment was performed, and a phylogenetic tree was constructed using MEGA 11.0 software (Fig. 1B). The findings revealed that this protein shares the highest homology with LBD1 from Arabidopsis thaliana. It possesses a relatively conserved Lateral Organ Boundary (LOB) domain at the N-terminus, containing a zinc finger motif responsible for DNA binding (CX2CX6CX3C), a GAS-block motif (Gly-Ala-Ser), and a leucine zipper motif (LX6LX3LX6L) responsible for protein dimerization, which are characteristic features of LBD TFs. Therefore, it was named HuLBD1.

To confirm the function of this gene in preserving fruits, the VIGS technique was utilized to inhibit the expression of HuLBD1 mRNA in H. undatus. Employing cDNA extracted from newly harvested H. undatus as a template, a 312 bp segment of the HuLBD1 gene was successfully obtained through PCR using specific primers. The target fragment was then inserted into the pTRV2 vector (Fig. 2A). The results obtained through RT-qPCR indicated that, compared to the CK group, the expression level of the HuLBD1 gene in the LBD-i group decreased. This finding demonstrated that the silencing of the HuLBD1 gene in H. undatus was effective (Fig. 2B).

By observing the phenotypic changes of the fruit, in the early stages of storage, both the CK group and LBD-i group showed high-quality fruit, with bright color (Fig. 2C). After 6 days of storage, the scales of the CK group began to shrivel, and the color became dull, while the LBD-i group maintained a relatively bright color with no obvious fungal spots on the surface. After 12 days, the scales of the CK group were completely dry, with numerous disease spots, while the LBD-i group maintained a favorable fruit color, with only a few disease spots (Fig. 2C).

With an increase in storage duration, an upward trend was observed in the weight loss rate of the H. undatus. Notably, the weight loss rate in the LBD-i group remained consistently lower compared to the CK group. After 12 days of storage (p < 0.01), the weight loss rate of the LBD-i group (12.67%) was markedly lower than that of the CK group (15.44%), indicating that VIGS treatment slowed down the water loss rate of the H. undatus (Fig. 2D).

The amount of flavonoid in the H. undatus peel exhibited a pattern of initially rising and then falling as the storage duration extended (Fig. 2E). On the third day of storage, both groups exhibited a “peak,” but the LBD-i group (0.163 mg/mL) was significantly higher than the CK group (0.136 mg/mL) (p < 0.01). On the ninth and twelfth day of storage, the LBD-i group (0.089 mg/mL) also had a higher flavonoid content than the CK group (0.076 mg/mL) (p < 0.01). Throughout the storage period, the trends of both groups were almost identical, but the LBD-i group consistently had higher levels, indicating that the silencing of HuLBD1 effectively enhanced the flavonoid content in the H. undatus peel.

To clarify the changes in fruit quality during storage, we measured the content of reducing sugars, soluble proteins, and pH in the fruit juice. During the early stage of storage, no significant differences were observed between the two groups for these indicators (p > 0.05). After 3 days of storage, the reducing sugar content and soluble protein content in the juice of both groups showed a decreasing trend (Fig. 2F,G). On the 12th day, the reducing sugar content in the LBD-i group (35.27 mg/mL) was significantly higher than that in the CK group (30.48 mg/mL) (p < 0.01). Similarly, the soluble protein content in the LBD-i group (0.14 mg/mL) was higher than that in the CK group (0.13 mg/mL) (p < 0.05). The pH of the fruit juice in both groups showed an increasing trend, with the CK group showing a significant acceleration in pH increase after day 3, while the LBD-i group remained relatively stable (Fig. 2H).

HuLBD1, as a plant-specific TF, identification of downstream target genes is particularly crucial for studying the function of LBD genes. Therefore, comparing the transcriptomic data of the CK group with the LBD-i group, and analyzing differentially expressed genes, was essential. The results of sample correlation analysis indicated a strong correlation between biological replicates (Fig. 3A). From the analysis of expression level differences, 5075 significantly differentially expressed genes were identified, comprising 2757 upregulated genes and 2318 downregulated genes (Fig. 3B).

The differentially expressed genes that were identified had been subjected to GSEA to annotate them to five metabolic pathways. As shown in Table S2, these pathways included Phenylpropanoid biosynthesis (MAP00940), Flavonoid biosynthesis (MAP00941), Stilbenoid, diarylheptanoid, and gingerol biosynthesis (MAP00945), Isoquinoline alkaloid biosynthesis (MAP00950), and Tropane, piperidine, and pyridine alkaloid biosynthesis (MAP00960). The biosynthesis pathways of both Phenylpropanoid and Flavonoid originate from phenylalanine and involve a sequence of enzyme-catalyzed reactions, indicating a close relationship between them. MAP00940 and MAP00941 pathways had the most enriched genes. Among them, 26 genes had been clustered in the leading edge of the MAP00940 pathway, while 10 genes had been clustered in the leading edge of the MAP00941 pathway (Fig. 3C,D). As the key genes had already been identified in the GSEA results, they played significant roles in Phenylpropanoid and Flavonoid biosynthesis pathways.

In our transcriptomic data, we identified 435 genes related to flavonoid and other phenylpropanoid synthesis, 5075 differentially expressed genes after HuLBD1 silencing, and 36 leading subsets obtained from GSEA analysis. Functional annotation analysis was performed on the three gene sets mentioned above (Fig. 3E), and a total of 16 genes were annotated to be present in all three gene sets. Hierarchical clustering analysis was conducted on these 16 differentially expressed genes (Fig. 3F), revealing that 4 genes showed significantly increased expression while 12 genes exhibited decreased expression, indicating significant differences between the two groups. Enrichment string analysis indicated that the four upregulated genes (Fig. 3G), HuCHS1-2, HuFLS1, HuCYP75B2 and HuCHS5-2, mainly participated in phenylpropanoid and flavonoid biosynthesis. These results suggest that HuLBD1 may have promoted the accumulation of flavonoid compounds and influenced the senescence process of H. undatus by regulating the expression of genes related to phenylpropanoid and flavonoid biosynthesis, such as HuCHS1-2, HuFLS1, HuCYP75B2, and HuCHS5-2.

To further validate the core genes, 69 genes related to flavonoid and other phenylpropanoid synthesis, screened from transcriptomic data, were used to construct a PPI network (Fig. 4) and performed co-expression network analysis (Fig. 5) using Cytoscape. The PPI network consisted of 36 nodes and 100 edges, with 31 upregulated nodes and 38 downregulated nodes represented by triangles and squares, respectively. Based on the CytoHubba sorting method “Maximal Clique Centrality (MCC),” the Top10 nodes in the overall protein network were ranked (Fig. 4B). The results showed that HuCHS5-2, HuFLS1, and HuCYP75B2 ranked in the top ten, ranking first, third, and ninth, respectively (Fig. 4B), indicating their core roles in the PPI network. MCODE was used to divide the expression-related network into two clusters (Fig. 5) and analyzed the correlation between genes. Cluster 1 comprised genes related to phenylpropanoid synthesis (Fig. 5A), such as phenylalanine ammonia-lyase (PAL), which catalyzed the non-oxidative deamination of phenylalanine to trans-cinnamic acid, a key enzyme in the phenylpropanoid metabolism pathway. β-glucosidase (BGLU) catalyzed the conversion of cinnamic acid to coumarins, while catechol-O-methyltransferase (COMT) catalyzed the conversion of caffeic acid to ferulic acid. Cluster 2 comprised genes related to flavonoid synthesis (Fig. 5B), such as chalcone synthase (CHS), which catalyzed the formation of naringenin chalcone from 4-coumaroyl-CoA and malonyl-CoA, initiating the biosynthesis of flavonoids. Flavonol synthase (FLS), a key enzyme downstream in the flavonoid biosynthesis pathway, promoting the accumulation of flavonol. The clustering results indicated that there was an expression correlation between HuCHS1-2 and HuFLS1, jointly promoting the biosynthesis of phenylpropanoids and flavonoids. The results of PPI network and co-expression network analysis indicated that HuLBD1 primarily regulated the expression of genes related to phenylpropanoid and flavonoid biosynthesis pathways, such as HuCHS1-2, HuFLS1, HuCYP75B2, and HuCHS5-2.

RT-qPCR was performed to detect the expression levels of four upregulated genes, HuCHS1-2, HuCHS5-2, HuFLS1, and HuCYP75B2 (Table 1), using RNA-seq data (Fig. 6). The R-squared value (R² = 0.9541) between the RT-qPCR results and RNA-seq data indicates a high level of reliability of the transcriptomic data.

The LBD transcription factor family members are crucial in the genetic evolution of various species, including lateral root development, secondary metabolism, stress response, fruit ripening, and senescence [32]. H. undatus has a high water and sugar content, with tender tissue that is prone to dehydration, wilting, rotting, and rapid senescence, resulting in a short shelf life [33]. However, little is known about the functions of LBDs in H. undatus. In this study, 70 LBD-related genes were identified in H. undatus, of which 17 belong to the LBD transcription factor family. Through analysis of the expression changes of LBD-related genes, the key gene HuLBD1 was screened, and the HuLBD1 TF in H. undatus was identified through bioinformatics analysis.

During long-term storage of fruits, water loss is a key factor affecting quality. Research indicates that maintaining ideal water levels in fruits may decrease cellular metabolism and prolong senescence while in storage [34]. In tomato fruits, inhibiting the expression of the SlLOB1 transcription factor significantly reduces water loss during storage and delays fruit ripening [35]. The results of the weight loss measurement in H. undatus in this study were similar to those in tomato fruits. As the storage time increased, the weight loss rate of H. undatus continuously increased, but after silencing the HuLBD1 gene, the weight loss rate increased more slowly. This suggests that HuLBD1 may play an important role in water retention and delaying senescence in H. undatus fruits.

In addition, sugars and organic acids are also key factors in determining the flavor and health value of fruits [36]. Post-harvest, H. undatus undergoes intense metabolic activity, leading to the continuous consumption and breakdown of nutrients such as proteins, sugars, and acids in the flesh. In the current study, it was observed that the content of reducing sugars and soluble proteins initially increased and then decreased, particularly showing a decline on the third day of storage. This could be due to the gradual hydrolysis of starch into soluble sugars during the early stages of storage, leading to an increase in reducing sugar content. Similarly, the soluble protein content also increased due to metabolic activity during the early stages of storage. However, with prolonged storage, fruit respiration and metabolism intensified, resulting in the extensive consumption of these substances, which led to a decline in their levels. A similar trend was also observed in pears, where reducing sugar content increased during the early stages of storage, but started to decrease after four days [37]. After silencing the HuLBD1 gene, the rate of decrease in reducing sugar and soluble protein content was delayed, suggesting that inhibiting the expression of HuLBD1 can slow down the decline in fruit quality. The pH level gradually increased with the extension of storage time. Compared to the control group, the fruit with silenced HuLBD1 showed a smaller change in pH level. A similar finding was reported by Rashid et al. [38], where apples showed smaller pH level changes after preservation treatment. This indicates that the silencing of HuLBD1 affects the physiological metabolic responses in the fruit.

It is noteworthy that flavonoids play an important role in postharvest fruit senescence processes, such as in stress resistance and influencing the quality, flavor, and nutritional value of fruits [39]. It has been reported that in species such as citrus [40,41], flavonoid-related metabolites in the fruit peel change during the ripening process and play a key role in fruit maturation and senescence. This study’s findings distinctly illustrate a bimodal trend in the accumulation of flavonoids as senescence progresses, with the highest flavonoid levels observed on the third day of storage. Compared to the control group, the flavonoid content in the peel of HuLBD1-silenced fruits was significantly increased and maintained at a higher level. This suggests that HuLBD1 plays an important role in the flavonoid biosynthesis process in H. undatus. HuLBD1 may regulate the expression of key genes in the flavonoid biosynthesis pathway, affecting the synthesis and accumulation of flavonoids, thereby delaying the senescence process of the fruit. Therefore, further research is needed to analyze the regulatory mechanism of HuLBD1 in fruit senescence.

Previous research has shown that trypsin can upregulate the expression of CsMYC2 in cucumber (Cucumis sativus), thereby promoting flavonoid biosynthesis to achieve preservation effects [42]. Qi et al. [43] investigated the ripening and softening processes of sweet cherry fruit (Prunus avium L.) and revealed that the expression of PaPG1, which is positively regulated by PaMADS7, facilitates the ripening and softening of sweet cherry fruit. The silencing of PaMADS7 notably hinders the ripening of sweet cherry fruit. As for the LBD transcription factor, the tomato fruit (Solanum lycopersicum) SlLOB1 is upregulated during ripening and activates cell wall-related genes. Inhibiting SlLOB1 in fruit hinders softening, while overexpression of SlLOB1 promotes premature induction of cell wall genes and fruit softening [35]. In other species such as Arabidopsis, Banana, and Citrus, members of the LBD family have been reported to activate downstream cell wall gene expression to participate in the regulation of fruit senescence [44–46]. Based on these existing reports, we speculate that HuLBD1 may participate in the senescence process of H. undatus by regulating the expression of certain genes.

However, unlike the previous understanding of LBD regulation in the maturation and senescence processes of fruit and vegetable cell walls, our results indicate that HuLBD1 acts through negative regulation of the expression of genes involved in flavonoid biosynthesis, promoting the accumulation of flavonoid compounds in H. undatus peel, enhancing fruit resistance, and delaying fruit senescence. In Camellia sinensis, CsLBD can directly bind to cis-elements in the promoters of CsC4H, CsDFR, and CsUGT84A, positively regulating flavonoid synthesis, thus affecting the growth, development, and resistance to abiotic stress in Camellia sinensis [47]. PtrLBD41 promotes the expression of chalcone synthase (CHS) under salt stress conditions, affecting the biosynthesis of flavonoids and enhancing tolerance to salt stress [48]. In this research, inhibiting LBD expression could promote the expression of genes such as HuCHS1-2 (Log₂FC = 7.1111), HuCHS5-2 (Log₂FC = 2.1918), HuFLS1 (Log₂FC = 4.5660), HuCYP75B2 (Log₂FC = 4.0781), thereby increasing the flavonoid content in the fruit peel, enhancing the fruit’s disease resistance and stress tolerance, and delaying fruit senescence.

It is worth noting that among the aforementioned genes, CHS showed the most significant expression changes. CHS is the first enzyme in the plant flavonoid biosynthesis pathway and is also one of the key enzymes in plant secondary metabolism, holding significant physiological significance for plants [49]. High expression of GmCHS8 in soybeans can promote the production of isoflavones [50]. Overexpression of the CitCHS gene in citrus leads to a significant increase in flavonoid content [51]. This is consistent with our results. Combining the expression pattern of the HuCHS gene in fruits with the transcriptome data, we observed induced expression of HuCHS in HuLBD1-silenced plants. We consider HuCHS and other genes as a downstream target gene of LBD, capable of promoting flavonoid compound synthesis. By measuring the flavonoid content in the fruit peel, we have demonstrated that the increased expression of chalcone synthase in H. undatus significantly enhances flavonoid content.

Previous studies have shown that trypsin can eliminate superoxide anions (O₂⁻) and protect cells, exhibiting anti-senescence effects [52]. H₂O₂ is a stable element of ROS when compared to its precursor. H₂O₂ can be transported into the cytoplasm by particular members of the aquaporin (AQP) super family [53] and the accumulation of H₂O₂ can cause oxidative damage to intracellular macromolecules. Treatment with trypsin significantly reduced the excessive accumulation of endogenous ROS, such as O₂⁻ and H₂O₂, in H. undatus, hindered cell damage, and improved the storage quality of H. undatus [3,19,20]. In this study, the expression of HuLBD1 was inhibited by trypsin. Transcriptome data showed that HuLBD1 promotes the accumulation of flavonoid compounds by negatively regulating the expression of genes such as HuCHS, thereby delaying the postharvest senescence in H. undatus (Fig. 7).