Integrated analysis of structure, physiology, transcriptome and metabolome reveals key metabolic pathway responses of cold-tolerant pear cultivar germplasms under low-temperature stress

Abstract

Freezing injury during winter is a critical abiotic stress that severely impacts the growth, development and, fruit quality of deciduous fruit trees. Cold tolerance can be induced through seasonal cold acclimation, which involves coordinated adjustments in tissue structure, physiology, and biochemistry driven by natural low-temperature, with distinct strategies across plant species. However, the cold-tolerant mechanisms of pear trees during cold acclimation remain poorly understood. Here, one-year-old branches of 10 pear cultivar germplasms were evaluated for cold tolerance based on the semi-lethal low temperature (LT₅₀), with three biological replicates across two consecutive experimental years (2021–2022). LT₅₀ values varied significantly among these materials, ranging from -42.43 ℃ to -32.59 ℃ and exhibited a highly significant negative correlation with field freezing injury indices (r = 0.86091, p < 0.0001). Subsequently, integrating anatomical structure observation, physiological index determination, metabolomics, and transcriptomics (with three biological replicates, each with three technical replicates for all omics and molecular experiments), we compared the low-temperature stress responses of cold-resistant ‘Shanli’ and cold-sensitive ‘Hanhong’, with statistical validation via one-way ANOVA, Duncan’s multiple range test, Pearson’s correlation analysis, and gray relational analysis. The results demonstrated that with decreasing temperature, the xylem ratio and lignin content in ‘Shanli’ branches increased markedly compared to ‘Hanhong’, and overwintering capability was correlated with branch lignin synthesis (r = -0.7783, p < 0.01). Cold acclimation enhanced lignin accumulation in ‘Shanli’ branches by increasing guaiacyl (G) and syringyl (S) units, associated with increased activities of key enzymes (shikimate hydroxycinnamoyl transferase (HCT), caffeoyl shikimate esterase (CSE), ferulate 5-hydroxylase (F5H), cinnamyl alcohol dehydrogenase (CAD), peroxidase (POD)) and critical intermediate metabolites (phenylalanine, ferulic acid, sinapic acid) in the phenylpropanoid pathway. Transcriptomic analysis identified 75 differentially expressed genes (DEGs) (|log2FoldChange|≥1 and FDR < 0.05) mapped to the phenylpropanoid pathway. Five potential key genes from the HCT, CSE, F5H, CAD and POD gene families, together with their co-expressed genes (such as ERF105-like) were identified as potentially associated with lignin content and composition modulationce in pear branches, providing novel insights into the regulatory network of lignin synthesis under natur under cold stress. Our findings reveal relationships between lignin synthesis pathways and cold toleranal low-temperature stress and candidate genes for cold-resistant pear breeding.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12870-026-08987-9.

Introduction

Abiotic stress factors (e.g., drought, salinity, heat, and low temperature) are among the most important constraints on plant growth and development, causing substantial yield and quality losses in many crop species [1–3]. For pear trees, a major deciduous fruit crop in temperate and boreal zones, prolonged winter low temperature is a predominant abiotic stress that frequently induces freezing damage, leading to reduced orchard productivity and significant economic losses [4]. To counteract this stress, pear trees acquire freezing tolerance through cold acclimation, a sophisticated adaptive process involving extensive physiological and biochemical reprogramming, activation of cold-responsive genes, and synthesis of protective proteins and metabolites to re-establish cellular homeostasis [5–7]. Elucidating the species-specific physiological and molecular mechanisms underlying pear cold tolerance is therefore essential for the targeted genetic improvement of cold-resistant cultivars.

A well-documented adaptive response during cold acclimation is cell wall thickening, enhancing cellular stability and preventing collapse under freeze-induced dehydration stress [8–10]. As a core structural component of the cell wall, lignin plays a pivotal role in this stress resistance process [11]. Lignin is biosynthesized via the phenylpropanoid pathway, and its accumulation has been preliminarily linked to cold tolerance in woody plants [12]. Under cold stress, excessive production of reactive oxygen species (ROS) in branches modulates lignin metabolism to reinforce cell wall rigidity [13]. Key enzymes in the phenylpropanoid pathway, including phenylalanine ammonia-lyase (PAL), cinnamate 4-hydroxylase (C4H), shikimate hydroxycinnamoyl transferase (HCT), ferulate 5-hydroxylase (F5H), cinnamyl alcohol dehydrogenase (CAD), peroxidase (POD) and laccase (LAC), regulate lignin content, monomer composition (p-hydroxyphenyl [H], guaiacyl [G], and syringyl [S] units) and overall deposition [14–18]. Transcription factors further fine-tune this biosynthetic pathway under cold stress, directly linking lignin regulation to cold stress resilience in woody plants [19, 20].

With the advancement of technology, transcriptomic and metabolomic analyses have identified as powerful tools for dissecting the molecular and metabolic networks of cold acclimation in woody plants [21–24]; however, their integrated use to unravel lignin biosynthesis regulation in pear under cold stress remains unexplored. The genus Pyrus harbors numerous wild relatives of cultivated pear with varying levels of abiotic stress tolerance. Among these, Pyrus ussuriensis is a highly valued pear rootstock characterized by strong cold tolerance, serving as a crucial genetic resource for mining and utilizing elite genes associated with cold tolerance [4, 25].

Despite its agronomic importance, critical knowledge gaps persist in pear cold acclimation research. First, most previous studies on pear focus solely on single-level changes (such as gene expression or enzyme activity) under cold stress, with the integrated regulatory mechanisms (involving anatomical structure, physiological indices, metabolomics and transcriptomics) of lignin biosynthesis in pear branches under natural low-temperature stress remaining unclear. Second, few studies have systematically quantified the associations between pear branch structural parameters, lignin-related metabolites or genes, and cold tolerance using rigorous statistical methods, hindering the identification of key regulatory factors. While lignin biosynthesis is known to be involved in cold resistance of other woody plants, the species-specific regulatory network of lignin-mediated cold tolerance in pear, including the key genes and metabolites driving lignin accumulation during natural cold acclimation, has not been elucidated.

To decipher the role of lignin in pear cold tolerance, we systematically compared and analyzed the differences in cold hardiness, tissue structural parameters and physiological responses of one-year-old branches from different pear cultivar germplasms under low-temperature stress by adopting multiple analytical methods, and identified the key response factors responsible for the variations in cold tolerance among them. We then integrated transcriptomic and metabolomic profiling to identify key metabolic pathways, with a particular focus on lignin biosynthesis-related mechanisms activated during pear cold acclimation. The novelty of this study lies in its integrated, multi-omics approach to unravel the coordinated regulation of lignin biosynthesis at structural, physiological, metabolic and transcriptional levels in pear, providing a theoretical foundation for breeding cold-resistant pear cultivars.

Materials and methods

Plant material and experimental design

Branches of experimental materials, comprising 10 pear cultivar germplasms (Table 1), were sampled from the pear orchard of the Pomology Institute, Jilin Academy of Agricultural Sciences, located in Gongzhuling City, Jilin Province, China (43°11′N, 124°02′E; 200 m above sea level). In this study, experimental materials did not include endangered species of wild flora. The experimental research on plants complies with institutional, national, and international guidelines. The soil type of pear orchard is brown coniferous forest soil. Each germplasm was grafted onto sorb (P. ussuriensis) as rootstock. The orchard was planted at a row spacing of 3.0 m × 4.0 m, and uniform nutrient (drip fertilization), water (regular irrigation), and pest control regimes were implemented to minimize environmental heterogeneity.

Table 1.

The list of 10 pear cultivar germplasms and their origin

Number	Germplasm	Country of origin	Pedigree
G1	Shanli	China	P. ussuriensis
G2	Jianba	China	P. ussuriensis
G3	Nanguo	China	P. ussuriensis
G4	Pingguoli	China	P. bretschneideri
G5	Hanxiang	China	‘Yanbian daxiangshui’ × ‘Pingxiang’
G6	Fuxiang	China	‘Longxiang’ × ‘115’
G7	Hanhong	China	‘Nanguo’ × ‘Jinsu’
G8	Hanlu	China	‘Yanbian daxiangshui’ × ‘Hangqing’
G9	Pingxiang	China	Natural seedling progeny of ‘Pingguoli’
G10	Hansu	China	‘Dali’ × ‘Jinsu’

The test materials consisted of 10 pear cultivar germplasms derived from different Pyrus species. For cold resistance evaluation, biological replicates were defined as three independent 10-year-old pear trees per cultivar, selected for uniform and moderate growth (consistent tree height, crown size, and no pest or disease symptoms) to ensure genetic and phenotypic homogeneity. One-year-old, healthy mature branches (0.8–1.0 cm in diameter) on the sun-exposed side of the experimental trees were sampled in December 2021 and 2022 and stored at 4 °C prior to experiments. Subsequently, anatomical structure observations and physiological index measurements were conducted with three biological replicates across all cultivar germplasms.

To elucidate the mechanisms underlying the response of pear trees to low temperature in winter, one-year-old branches of cold-resistant ‘Shanli’ (SL) and cold-sensitive ‘Hanhong’ (HH) were collected in August and December 2023 (three biological replicates per cultivar germplasm per sampling time), designated as ‘SL-8’, ‘HH-8’, ‘SL-12’, and ‘HH-12’, respectively. During these sampling periods, branch structural characteristics and lignin deposition were examined. LT₅₀ values, physiological indices, as well as metabolomic and transcriptomic profiles were also determined. All molecular experiments included three technical replicates.

Freeze-thaw treatment

Branches from 10 pear cultivar germplasms were subjected to freeze-thaw treatment for cold resistance evaluation following the method of Zhao et al. [26]. A programmable high-low temperature alternating test chamber (model: XC-GDJB-100D, China) with uniform temperature distribution (± 0.5 ℃ variation across the chamber) was used. Samples were randomly arranged on three separate shelves to ensure equal temperature exposure. Eight freezing temperatures were applied: 4 ℃ as control, -18, -22, -26, -30, -34, -38, -42, and − 46 ℃. Three branches per biological replicate for each cultivar germplasm were used per temperature. The freezing protocol was validated using calibrated thermocouples, cooling and heating rates were maintained at 4 ℃·h^{− 1} (± 0.2 ℃·h^{− 1}), with a 12 h hold at the target temperature and a 4 h room temperature (25 ± 1 ℃) recovery period for complete thawing.

Evaluation of cold resistance

2,3,5-Triphenyltetrazolium chloride (TTC) staining was performed as described by Zhao et al. [26], with minor modifications. After freeze-thaw treatment, 10 cross-sectional slices (5 mm thick) were cut from branch internodes and immersed in 10 mL of 1.0% TTC solution in 0.1 mol/L phosphate buffer (pH 7.4). The samples were incubated in the dark at 37 °C for 48 h. Subsequently, the stained tissues were dissected to obtain longitudinal sections, which were scanned using a LA2400 scanner equipped with image analysis software (WinRHIZO™; Regent Instruments Inc., Quebec City, QC, Canada). The TTC–stained areas on the longitudinal sections were quantified based on digital images. Cold injury severity for each branch was classified into one of five levels according to the percentage of stained area relative to the entire longitudinal section:

Level 1 = 0%–10.0%; Level 2 = 10.1%–40.0%; Level 3 = 40.1%–70.0%; Level 4 = 70.1%–90.0%; Level 5 = 90.1%–100% staining area (Fig. 1).

Fig. 1.

Standards for 2,3,5-triphenyltetrazolium chloride staining in pear branch

The staining index was calculated at related levels as follows:

Hardiness zone classification

The semi-lethal low temperature (LT₅₀) value was determined by fitting the staining indices to a logistic equation: y = K/(1 + ae^{− bx}), Where y represents staining index, x denotes treatment temperature, K is the saturated value of staining index, and a and b are the equation parameters. When the goodness of fit exceeded 0.95, the LT₅₀ value was subsequently calculated based on the fitted regression equation. Experiments were repeated three times with three biological replicates. Based on the LT₅₀ values, cold resistance was classified into distinct zones according to Chen [27]. Additionally, field freezing injury was surveyed in 2021 and 2022 to validate LT₅₀ results.

Observation of branch cross section and lignin staining

Cross sections of one-year-old branches from 10 pear cultivar germplasms were observed using scanning electron microscopy (SEM; HITACHI Regulus 8100, Japan). Lignin staining was performed with phloroglucinol-HCl, following the method of Yang et al. [28].

Measurement of physiological indices

Lignin content

Branch tissues were dried at 70 °C for 48 h and then ground to powder. Cell wall residues (CWRs) were obtained by washing with chloroform/methanol (1:1), 100% methanol, 50% methanol, and water three times for each solution [28]. The acetyl bromide method was employed to determine total lignin content in the branch powder using a spectrophotometer at 280 nm, following the protocol of Chang et al. [29]. Lignin monomers were determined by thioacidolysis followed by GC–MS with CWR [28].

Malondialdehyde and hydrogen peroxide contents

The same extraction solution of 0.1% trichloroacetic acid (TCA) was used for the determination of malondialdehyde (MDA) and hydrogen peroxide (H₂O₂) contents. MDA content was determined using the thiobarbituric acid method by measuring absorbance at 532 nm and 600 nm, then calculated based on the molar extinction coefficient of 155 mM^{− 1}·cm^{− 1} [30]. The calculation formula is as follows:

Where A₅₃₂ and A ₆₀₀ represent the absorbance at 532 nm and 600 nm, respectively; Vt, Vr and Vs denote the volume of sample extract, reaction solution and sample used for determination, respectively; w is cuvette path length (0.5 cm); FW is fresh weight of sample; ε represents the molar extinction coefficient (155 mM^− 1·cm^− 1). The detection limit of MDA is 0.1 nmol·g^− 1 FW.

H₂O₂ content was quantified spectrophotometrically at 390 nm according to Alexieva et al. [31]. A standard curve was constructed with H₂O₂ (5, 10, 15, 20, 25, 30 µmol·L^-1) as the standard, and the H₂O₂ content in samples was derived from this curve. The calculation formula is as follows:

Where C represents H₂O₂ concentration in the sample; Vt is the total extraction solution volume; FW is fresh weight of the sample. The detection limit of H₂O₂ is 0.01 µmol·g^− 1 FW.

Enzyme activities

Branches (1 g) were homogenized with 100 mM potassium phosphate (pH 7.8) containing 0.1 mM EDTA and 50 mM 1% (w/v) polyvinylpyrrolidone (PVPP) at 4 °C. The homogenate was centrifuged at 15,000×g for 20 min, and the supernatant was stored at − 80 °C for the analysis of enzyme activity levels [30].

The activities of antioxidant enzymes, including superoxide dismutase (SOD), catalase (CAT), and peroxidase (POD) were measured spectrophotometrically according to the manufacturer’s instructions for the enzyme activity detection kit (Jiancheng, Nanjing, China). The detection limit of SOD, CAT and POD is 5 U·g^-1 FW, 2 U·g^-1 FW and 0.5 U·g^-1 FW, respectively.

The activities of lignin biosynthesis-related enzymes including phenylalanine ammonia-lyase (PAL), cinnamate 4-hydroxylase (C4H), 4-coumarate-CoA ligase (4CL), shikimate hydroxycinnamoyl transferase (HCT), caffeoyl shikimate esterase (CSE), caffeic acid 3-O-methyltransferase (COMT), ferulate 5-hydroxylase (F5H), caffeoyl-CoA 3-O-methyltransferase (CCoAOMT), cinnamoyl-CoA reductase (CCR), cinnamyl alcohol dehydrogenase (CAD), laccase (LAC), uridine diphosphate glycosyltransferases (UGT) and peroxidase (POD) were performed through enzyme-linked immunosorbent assay (ELISA) kits (Nuominkeda, Wuhan, China) employing a double antibody one-step approach [32]. The principle is to utilize the antigen-antibody reaction for detection, using commercialized enzymes to create a standard curve (U·mL of crude extract). The concentrations of lignin biosynthesis-related enzymes in the samples were then determined by comparing the OD value of the samples to the standard curve. Enzyme activities were subsequently converted to U·g⁻¹FW based on the extraction volume and sample fresh weight. The detection limit by this assay is 1.0 U·g⁻¹FW.

Metabolite analysis of lignin pathway

Sample extraction and LC-MS/MS analysis

The freeze-dried samples were crushed using a mixer mill (MM 400, Retsch) with a zirconia bead for 1.5 min at 30 Hz. 100 mg powder was weighted and extracted overnight at 4 °C with 0.6 mL 70% aqueous methanol according to the method of Yu et al. [33].

The detection of lignin metabolites was conducted by Wuhan Metware Biotechnology Co., Ltd (Wuhan, China). The analysis of the sample extracts was carried out using an ultra-performance liquid chromatography-electrospray ionization-tandem mass spectrometry (UPLC-ESI-MS/MS) system. The UPLC system utilized was the ExionLC™ AD, and the tandem mass spectrometry system was provided by SCIEX, as referenced on their official website (https://sciex.com.cn/). The analytical conditions for the UPLC were as follows: the column used was an Agilent SB-C18 (1.8 μm particle size, 2.1 mm internal diameter, and 100 mm length). The mobile phase was composed of solvent A, which was pure water containing 0.1% formic acid, and solvent B, which was acetonitrile containing 0.1% formic acid. The sample measurements were conducted using a gradient elution program that began with an initial condition of 95% A and 5% B. Over a period of 9 min, a linear gradient was programmed to reach 5% A and 95% B, which was then maintained for 1 min. Following this, the system was readjusted to 95% A and 5.0% B within 1.1 min and held for 2.9 min. The flow rate was set at 0.35 mL·min^− 1, the column oven temperature was maintained at 40 ℃, and the injection volume was 2 µL. The effluent from the UPLC system was then directed to an ESI-triple quadrupole-linear ion trap (QTRAP)-MS for further analysis.

The operation parameters for the ESI source were meticulously set as follows: the source temperature was maintained at 500 ℃; the ion spray voltage (IS) was calibrated to 5500 V for positive ion mode and − 4500 V for negative ion mode; the ion source gases, specifically gas I (GSI) and gas II (GSII), along with the curtain gas (CUR), were adjusted to 50, 60, and 30 psi, respectively. The collision-activated dissociation (CAD) was set to a high level.

Instrument tuning and mass calibration were performed with 10 and 100 µmol·L^− 1 polypropylene glycol solutions in QQQ and LIT modes, respectively. QQQ scans were acquired as MRM experiments with collision gas (nitrogen) set to 5 psi. The declustering potential (DP) and collision energy (CE) for individual MRM transitions were further optimized. A specific set of MRM transitions was monitored for each period base on the metabolites eluting within this period.

Identification of lignin metabolites

Using the Metware Database, we conducted a comprehensive qualitative and quantitative analysis of the sample metabolites via mass spectrometry (MS). The characteristic ions for each compound were identified using a triple quadrupole mass spectrometer, and the corresponding signal intensity, measured in counts per second (cps), was captured by the detector. To process the mass spectrometry data, we employed MultiQuant software to open the samples’ mass spectrometry file. This software facilitated the integration and correction of the chromatographic peaks. The area under each chromatographic peak, denoted as ‘Area’, serves as an indicator of relative content of the respective metabolite. Ultimately, the integrated data including the areas of all chromatographic peaks were exported and saved for further analysis.

The criteria for identifying differential lignin metabolites in this study were based on two key metrics: the variable importance in projection (VIP) score and the absolute value of the log2FoldChange. Specifically, differential lignin metabolites were identified when the VIP score exceeded 1.0 and the value of |log2FoldChange| was greater than or equal to 2.0. The VIP values were derived from the results of the Orthogonal Projections to Latent Structures-Discriminant Analysis (OPLS-DA), which included score plots and permutation plots. These analyses were performed using the MetaboAnalystR package in R. Prior to conducting the OPLS-DA, the data underwent logarithmic transformation (base 2) and mean centering to stabilize the variance and improve the model’s performance. To ensure the robustness of our findings and to prevent overfitting, a permutation test with 200 iterations was conducted to validate the statistical significance of the OPLS-DA model.

RNA sequencing and data analysis

According to the methods previously described by Wu et al. [34], total RNA was purified from branch tissues of ‘Shanli’ and ‘Hanhong’ using the ethanol precipitation protocol and CTAB-PBIOZOL reagent, following the manufacturer’s instructions. The RNA sequencing was conducted by Wuhan Metware Biotechnology Co., Ltd (Wuhan, China). RNA integrity was assessed using the RNA Nano 6000 Assay Kit on the Bioanalyzer 2100 system (Agilent Technologies, Santa Clara, CA, USA). Double-stranded complementary DNA (cDNA) was synthesized using reverse transcriptase and random hexamer primers, followed by PCR amplification. The resulting PCR products were purified with AMPure XP beads to construct the final sequencing library, which was then sequenced on the BGIseq500 platform (BGI-Shenzhen, China).

All raw sequencing data were deposited in the NCBI Sequence Read Archive (SRA) under BioProject accession number PRJNA1394719. The bioinformatics pipeline for RNA-seq analysis was as follows:

Quality control of raw sequencing data

Raw sequencing data were first subjected to quality control using Fastp (v0.23.2) with the following key parameters: adapter sequences were removed; reads containing poly-N sequences (N base number > 15; parameter: --n_base_limit 15) were discarded; low-quality reads with > 50% bases having quality value Q ≤ 20 (--qualified_quality_phred 20) were filtered out. After quality control, clean reads accounted for 91.88% of total raw reads, with Q20 > 92% and Q30 > 85%.

Alignment to reference genome

The index of the reference genome was built using HISAT2 (v2.2.1) and paired-end clean reads were aligned to the Pyrus bretschneideri ‘Dangshansuli’ reference genome (v1.1) (https://doi.org/10.6084/m9.figshare.19291445.v3) using HISAT2 (v2.2.1; parameters: --dta-cufflinks). The alignment was performed using a two-pass strategy: first pass to discover splice sites, second pass to align reads using collected splice sites. The total mapped reads reached 87–91%, with uniquely mapped reads of 83–86% (exceeding the 70% threshold).

Transcript assembly and quantification

The mapped reads of each sample were assembled by StringTie (v1.3.3b) in a reference-based approach with super-reads technology enabled. StringTie builds splice graphs and flow networks, computes maximum flow to estimate transcript abundance, and iteratively updates assembly until convergence. Gene expression levels were quantified as FPKM (Fragments Per Kilobase of transcript per Million fragments mapped) using featureCounts (v1.6.1).

Differential expression analysis

Differential expression analysis was performed using DESeq2 (v1.22.1) and edgeR (v3.24.3). The screening criteria for DEGs were set as |log2FoldChange|≥1 and FDR < 0.05, with multiple testing correction using the Benjamini-Hochberg method. A total of 8426, 8468, and 7301 DEGs were identified in the comparisons SL-12 vs. SL-8, HH-12 vs. HH-8, and SL-12 vs. HH-12, respectively.

Functional annotation and enrichment analysis

Functional annotation was performed based on NCBI non-redundant protein sequences (Nr), a manually annotated and reviewed protein sequence database (Swiss-Prot), euKaryotic Ortholog Groups (KOG), Protein family database (Pfam), Gene Ontology (GO), and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases using DIAMOND and Blast (v2.14.0) with E-value thresholds of 1e−5 (Nr/Swiss-Prot), 0.01 (Pfam), 1e−3 (KOG), 1e−10 (KEGG), and 1e−6 (GO). GO and KEGG enrichment analyses were implemented by clusterProfiler (v3.10.1) with corrected p-values < 0.05 considered significant. Transcription factors were predicted using iTAK.

Data visualization and archiving

Data visualization was performed using IGV for genome browser visualization, and R packages (ggplot2, pheatmap, VennDiagram) for PCA, heatmaps and Venn diagrams. Interaction networks were constructed using STRING and visualized with Cytoscape.

Gene expression analysis using quantitative real-time PCR

The expression levels of genes associated with responses to natural low-temperature stress were determined by quantitative real-time PCR (qRT-PCR). The qRT-PCR assays were conducted on an ABI 7500 Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) using qPCR SYBR Green Master Mix. The gene expression levels were normalized using TUB-b2 as the reference gene, and the relative expression levels were calculated using the 2^−∆∆CT method [35]. Detailed information on the target genes and their specific primers is provided in Table S1 (in Supplementary Material 1). For qRT-PCR experiments in this study, both three technical replicates and biological replicates were included to ensure the rigor and reliability of gene expression detection results. Specifically, three technical replicates were set for each biological sample to reduce experimental errors caused by operational differences, and the average value of the three technical replicates was used for subsequent gene expression analysis. Additionally, biological replicates (at least three independent samples per group) were set to reflect the true biological effect of low-temperature stress on gene expression, avoiding the contingency of single-sample results. All qRT-PCR reactions were performed in strict accordance with the manufacturer’s instructions, which was consistent with the overall statistical rigor requirements of this study. Finally, all sequencing data generated in this study were deposited in the NCBI database.

Statistical analyses

Each treatment was performed in triplicate. All statistical analyses were conducted using the SAS 9.0 software system (SAS Institute Inc., Cary, NC, USA). Significant differences were determined by one-way analysis of variance (ANOVA) followed by Duncan’s multiple range test. Pearson’s correlation coefficients and gray relation degree were calculated to evaluate the associations .

The gray relation degrees, which could accurately reflect the key parameters affecting branch cold tolerance, were adopted using gray relation analysis (GRA). The GRA model was established based on the method of Qiao et al. [36]. First, original data were normalized by the mean method to eliminate dimensional interference; the reference group (LT₅₀ value) and comparison group (tissue structural parameters and physiological indices) were set. Second, the distinguishing coefficient was fixed at 0.5 (a widely accepted optimal value), and gray relational coefficients were calculated from the absolute differences between reference and comparison groups. Third, the gray relational degree (0–1) was derived by averaging coefficients to determine the correlation degree between indices.

The critical metabolites influencing the lignin biosynthesis pathway were analyzed using partial least squares regression (PLSR). Graphical figures were generated using SigmaPlot 10.0 (Systat Software Inc., San Jose, CA, USA).

Results

Evaluation of cold resistance in pear cultivar germplasms

Cold resistance of 10 pear cultivar germplasms was evaluated using LT₅₀ values determined by fitting TTC staining indices to a logistic equation over two consecutive years. LT₅₀ values varied significantly across germplasms, ranging from − 42.43 ℃ (most cold-resistant P. ussuriensis ‘Shanli’) to − 32.59 ℃ (most cold-sensitive ‘Hanhong’) (Table 2). Notably, LT₅₀ values exhibited a highly significant negative correlation with field freezing injury indices (Fig. S1), confirming the reliability of TTC–based LT₅₀ for reflecting actual field cold hardiness in pear. Based on LT₅₀ values, the 10 cultivar germplasms were classified into three cold hardiness zones (Zone 1 > Zone 2 > Zone 3), ‘Shanli’ was categorized in Zone1, three hybrid cultivars (‘Hanhong’, ‘Hanlu’ and ‘Hansu’) were assigned to Zone 3 as cold-sensitive; and the remaining six cultivars were grouped in Zone 2 as moderate cold resistance.

Table 2.

Evaluation of cold resistance in 10 pear cultivar germplasms based on mean LT₅₀ values over two consecutive years

Number	Staining index in treatment temperatures								Logistic equation	Fitting degree	LT50 / ℃	Hardiness
	-18 ℃	-22 ℃	-26 ℃	-30 ℃	-34 ℃	-38 ℃	-42 ℃	-46 ℃
G1	0.92	0.93	0.91	0.92	0.87	0.75	0.67	0.01	y = 0.9006/(1 + e(−35.0851−0.8216x))	0.9631	-42.43	Zone 1
G2	0.95	0.91	0.95	0.88	0.77	0.67	0.04	0.04	y = 0.8957/(1 + e(−44.8610−1.1381x))	0.9848	-39.21	Zone 2
G3	0.81	0.83	0.80	0.75	0.71	0.23	0.00	0.00	y = 0.7985/(1 + e(−28.0034−0.7612x))	0.9964	-36.11	Zone 2
G4	0.95	0.89	0.96	0.77	0.71	0.32	0.00	0.00	y = 0.9003/(1 + e(−20.0548−0.5464x))	0.9784	-36.29	Zone 2
G5	0.92	0.89	0.96	0.95	0.71	0.48	0.03	0.00	y = 0.9066/(1 + e(−28.0977−0.7392x))	0.9868	-37.73	Zone 2
G6	0.93	0.95	0.91	0.91	0.84	0.69	0.03	0.01	y = 0.9073/(1 + e(−25.5193−0.6431x))	0.9943	-39.37	Zone 2
G7	0.97	0.76	0.95	0.81	0.48	0.13	0.03	0.00	y = 0.8959/(1 + e(−10.3917−0.3117 x))	0.9660	-32.59	Zone 3
G8	0.85	0.79	0.93	0.77	0.60	0.03	0.00	0.00	y = 0.7891/(1 + e(−41.5238−1.1719x))	0.9515	-34.97	Zone 3
G9	1.00	0.89	0.93	0.83	0.85	0.46	0.03	0.01	y = 0.9367/(1 + e(−14.5117−0.3989x))	0.9853	-36.04	Zone 2
G10	0.85	0.85	0.79	0.73	0.60	0.21	0.00	0.00	y = 0.7983/(1 + e(−18.0122−0.5012x))	0.9866	-34.91	Zone 3

Analysis of branch cross-sectional structure and physiological indices

Significant differences were observed in cross-sectional structural parameters among the 10 pear cultivar germplasms (Fig. S2). P. ussuriensis ‘Shanli’ exhibited the lowest pith ratio (41.37%) and vessel diameter (8.46 μm), concomitant with the highest xylem ratio (28.44%) and vessel density (0.0025 unit), while cold-sensitive ‘Hanhong’ had the largest vessel diameter (16.05 μm). Phloem and cortex ratios showed moderate variation across these cultivar germplasms, ranging from 13.28% to 19.19% and from 5.77% to 8.05%, respectively (Table 3).

Table 3.

Comparison of cross-sectional structural parameters in one-year-old branches of 10 pear cultivar germplasms over two consecutive years

Number	Pith ratio (%)	Xylem ratio (%)	Phloem ratio (%)	Cortex ratio (%)	Vessel diameter (µm)	Vessel density (unit)
G1	41.37	28.44	16.43	6.77	8.46	0.0025
G2	42.36	28.35	16.77	7.50	8.98	0.0013
G3	43.57	27.59	16.27	7.30	10.18	0.0016
G4	43.19	27.41	17.42	7.19	11.94	0.0020
G5	45.72	26.78	14.07	7.67	13.01	0.0012
G6	42.68	28.33	19.19	5.77	9.70	0.0023
G7	45.39	25.92	16.28	6.88	16.05	0.0017
G8	45.62	26.11	16.68	6.66	14.12	0.0020
G9	44.23	27.06	16.15	7.76	12.93	0.0020
G10	45.92	26.47	13.28	8.05	15.01	0.0017

Physiologically, ‘Shanli’ showed the highest lignin content, SOD and POD activities, alongside the lowest H₂O₂ content. Conversely, ‘Hanhong’ displayed the lowest lignin content and POD activity, coupled with the highest MDA and H₂O₂ contents (Table 4). Gray relation analysis revealed that POD activity and xylem ratio exhibited the strongest associations with LT₅₀, with gray relation degrees of 0.9859 and 0.9767, respectively.

Table 4.

Comparison of physiological indices related to cold resistance in 10 pear cultivar germplasms over two consecutive years

Number	Lignin (mg·g−1CWR)	MDA (nmol·g−1FW)	H2O2 (µmol·g−1FW)	SOD (U·g−1FW)	CAT (U·g−1FW)	POD (U·g−1FW)
G1	26.72 ± 1.93	8.43 ± 0.53	0.40 ± 0.79	184.09 ± 28.10	158.67 ± 13.54	784.09 ± 82.40
G2	25.13 ± 7.33	4.64 ± 0.46	0.53 ± 0.22	81.68 ± 9.16	133.33 ± 27.09	702.37 ± 77.03
G3	23.70 ± 4.93	10.15 ± 0.27	0.81 ± 0.25	119.37 ± 14.26	165.33 ± 20.34	685.13 ± 69.94
G4	23.22 ± 9.63	4.99 ± 0.29	0.93 ± 0.07	102.41 ± 21.65	154.67 ± 4.13	692.41 ± 96.59
G5	26.57 ± 4.60	4.90 ± 0.25	0.90 ± 0.09	88.00 ± 5.32	325.33 ± 19.70	681.68 ± 49.61
G6	25.64 ± 2.39	8.08 ± 0.27	0.68 ± 0.35	85.13 ± 17.49	106.67 ± 21.86	730.16 ± 36.65
G7	10.50 ± 2.29	14.62 ± 0.27	2.14 ± 0.04	92.17 ± 7.78	188.00 ± 12.90	620.23 ± 46.07
G8	18.28 ± 4.54	5.68 ± 0.01	0.55 ± 0.07	85.31 ± 10.68	148.00 ± 9.47	685.31 ± 42.71
G9	26.20 ± 2.72	4.98 ± 0.70	0.59 ± 0.16	70.23 ± 11.52	166.67 ± 16.52	688.78 ± 58.85
G10	13.91 ± 2.14	9.98 ± 0.26	1.20 ± 0.17	170.16 ± 12.40	125.33 ± 7.45	649.00 ± 21.27

Correlation analysis further substantiated that xylem ratio, lignin content, and POD activity were significantly negatively correlated with LT₅₀ (p < 0.01), whereas LT₅₀ was positively correlated with vessel diameter and H₂O₂ content (p < 0.05) (Table 5). Additionally, lignin content also displayed a positive correlation with xylem ratio and POD activity, and a negative correlation with vessel diameter (p < 0.01) (Fig. S3).

Table 5.

Gray relation analysis of LT₅₀ values with tissue structure parameters and physiological indices of branches in 10 pear cultivar germplasms

Index	Pith ratio	Xylem ratio	Phloem ratio	Cortex ratio	Vessel diameter	Vessel density
Gray relation degree	0.9532	0.9767	0.9465	0.9435	0.8775	0.9073
Rank	3	2	4	5	9	7
Correlation coefficient	0.7978**	-0.8772**	-0.2898	0.2763	0.8799**	-0.3416
Index	Lignin	MDA	H2O2	SOD	CAT	POD
Gray relation degree	0.9080	0.8318	0.8181	0.8644	0.8818	0.9859
Rank	6	11	12	10	8	1
Correlation coefficient	-0.7783**	0.3986	0.7158*	-0.3080	0.0681	-0.9371**

* and ** indicate the significant difference at the 0.05 and 0.01 level, respectively. The same applies below

Comparative analysis of cold acclimation between ‘Shanli’ and ‘Hanhong’ under natural low temperature

To dissect the mechanistic basis of differential cold tolerance, we compared the most cold-resistant ‘Shanli’ and most cold-sensitive ‘Hanhong’ during natural cold acclimation from August to December. LT₅₀ values of both materials decreased progressively with declining temperature, with reductions of 22.2 ℃ and 19.0 ℃, respectively, indicating enhanced freezing tolerance during cold acclimation (Fig. S4, Fig. 2A). ‘Shanli’ consistently maintained significantly lower LT₅₀ values and MDA contents than ‘Hanhong’ at both sampling time points (p < 0.05) (Fig. 2A, B). Consistent with these physiological differences, the ‘Hanhong’ xylem frequently exhibited severe freezing damage, whereas ‘Shanli’ branches remained largely free from freezing injury after overwintering (Fig. 2C). Histochemical staining with phloroglucinol-HCl revealed more intense staining and abundant lignin deposition in ‘Shanli’ vessel elements and wood fiber cell walls compared to ‘Hanhong’ in December (Fig. 2D), confirming the higher lignin content quantified in ‘Shanli’.

Fig. 2.

Comparison of LT₅₀ values (A), MDA content (B), freezing damage (C) and histochemical staining (D) in one-year-old branches between ‘Shanli’ and ‘Hanhong’ under natural low-temperature stress

Lignin biosynthesis-related enzyme activities exhibited marked divergence between the two materials during cold acclimation. In ‘Shanli’, lignin content and activities of 4CL, HCT, F5H, COMT, CAD, UGT, and POD increased significantly in December compared with August. In contrast, ‘Hanhong’ showed no significant change in lignin content or in the CSE, F5H, COMT activities, whereas C4H and 4CL activities decreased. By December, ‘Shanli’ displayed significantly higher lignin content and activities of 4CL, HCT, COMT, F5H, CAD, UGT and POD than ‘Hanhong’ (p < 0.05). Notably, HCT, CSE, F5H, CAD, and POD activities showed significant positive correlations with lignin content (Fig. 3).

Fig. 3.

Enzyme activities and correlation analysis involved in lignin biosynthesis and lignin content in branches of ‘Shanli’ and ‘Hanhong’ under natural low-temperature stress

Integrated metabolomic and transcriptomic analysis of lignin biosynthesis under natural low-temperature stress

To elucidate the molecular mechanisms of lignin synthesis in pear branches under natural low-temperature stress, one-year-old branches of cold-resistant ‘Shanli’ and cold-sensitive ‘Hanhong’ collected in August and December were designated as SL-8, SL-12, HH-8 and HH-12.

Metabolomic profiling of the phenylpropanoid pathway

Metabolites were subjected to OPLS-DA analysis, with model validation parameters (Q², R²Y) confirming the stability and reliability of the model, indicating its suitability for sample discrimination (Fig.S5). The OPLS-DA score plot demonstrated that the four sample groups clustered into distinct clusters, reflecting significant differences in lignin-related metabolites (Fig. 4A). A total of 13 different accumulation metabolites (DAMs) involved in lignin biosynthesis within the phenylpropanoid pathway were detected (Fig. 6). Metabolites with higher VIP values are regarded as the key contributors to sample clustering, and five such metabolites (sinapaldehyde, cinnamic acid, coniferaldehyde, sinapyl alcohol and coniferyl alcohol) were identified as key discriminators for distinguishing the four sample groups (Fig. 4B). These metabolites are crucial precursors for the formation of G and S lignin monomers.

Fig. 4.

Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA) of lignin metobalism in ‘Shanli’ and ‘Hanhong’ branches under natural low temperature. A OPLS-DA score plot; B Key lignin metabolites with variable importance in projection (VIP) values greater than 1.0

Fig. 6.

Heatmap of intermediate metabolites contents and DEGs involved in lignin biosynthesis via the phenylpropanoid pathway in branches of ‘Shanli’ and ‘Hanhong’ under natural low-temperature stress

Transcriptomic profiling and differential gene expression

Principal component analysis (PCA) of RNA-seq data revealed that the experimental samples clustered into four distinct groups, reflecting differential gene expression patterns across sampling time points. The tight clustering of three biological replicates for each sample demonstrated the high reproducibility and reliability of the RNA-seq data (Fig. 5A).

Fig. 5.

Overview of transcriptome data from ‘Shanli’ and ‘Hanhong’ under natural low-temperature stress. A PCA of different comparison groups. B Numbers of total, upregulated, and downregulated DEGs in each comparison group. C KEGG enrichment analysis scatter plot of DEGs in SL-12 vs. HH-12

Comparative analysis of gene expression between months identified 8426, 8468, and 7301 DEGs in the comparisons SL-12 vs. SL-8, HH-12 vs. HH-8, and SL-12 vs. HH-12, respectively (Fig. 5B, Fig S6). KEGG pathway enrichment analysis of these DEGs revealed significant enrichment in pathways related to biosynthesis of secondary metabolites, plant-pathogen interaction, and metabolic pathways under natural low-temperature stress (Fig. 5C, Fig S6), all of which are closely associated with lignin synthesis and cold stress response.

Key DAMs and DEGs in the phenylpropanoid pathway

Notably, three critical pathway intermediates—phenylalanine, ferulic acid, and sinapic acid exhibited significantly elevated levels in cold-resistant ‘Shanli’ branches as temperature decreased from August to December (p < 0.05). These metabolites serve as precursors for cinnamic acid, sinapaldehyde, and coniferaldehyde, respectively. Furthermore, under identical low-temperature conditions in December, the contents of these three precursors were significantly higher in the cold-resistant ‘Shanli’ than in the cold-sensitive ‘Hanhong’ (p < 0.05) (Fig. 6). These results identify phenylalanine, ferulic acid and sinapic acid as key metabolic determinants of differential lignin biosynthesis in pear branches under low-temperature stress, thereby contributing to the observed variation in cold tolerance among pear cultivar germplasms.

75 DEGs mapped to the phenylpropanoid pathway—the core pathway of lignin biosynthesis—were identified and classified into 12 gene families (Fig. 6). In cold-resistant ‘Shanli’, all differentially expressed POD genes were upregulated following low-temperature stress. Within the HCT and CAD gene families, 2 of 15 HCT genes and 2 of 11 CAD genes were upregulated under natural low-temperature stress. For the CSE gene family, two CSE genes were upregulated in ‘Shanli’, while another two were downregulated in December compared to August. In the F5H family, all DEGs in ‘Shanli’ were downregulated except for novel 2906, which was uniquely upregulated in December (Fig. 6). These findings implicate phenylpropanoid pathway genes as crucial regulators of lignin biosynthesis in pear under natural low-temperature stress, with their differential expression potentially cooperates with key metabolites to modulate lignin accumulation and cold tolerance.

Collectively, the integrated metabolomic and transcriptomic analyses demonstrate that both key metabolites (phenylalanine, ferulic acid, sinapic acid) and differentially expressed structural genes (POD, HCT, CAD, CSE and F5H family members) in the phenylpropanoid pathway are closely involved in the regulation of lignin biosynthesis under natural low-temperature stress. The coordinated action of these metabolic and transcriptional factors contributes substantially to the divergence in cold tolerance between ‘Shanli’ and ‘Hanhong’.

Associated analysis of metabolomes and transcriptomes

Correlation analysis between metabolite levels and gene expression identified significant positive correlations between the contents of lignin and three key precursors (phenylalanine, ferulic acid, sinapic acid) and the expression of five hub genes: LOC125475722 (HCT), LOC125469219 (CSE), novel.2906 (F5H), LOC125479347 (CAD) and LOC125473073 (POD) (Fig. 7).

Fig. 7.

Integrative correlation analysis between metabolites and gene expression in the lignin biosynthesis pathway

Network analysis further revealed that these five hub genes were co-expressed with 17 associated TFs, predominantly belonging to the AP2/ERF, NAC, and GRF families (Fig. S7). Among these, LOC103959790 (ERF105-like), LOC103964649 (ZC3H48), and LOC125475748 (NAC17-like) identified as potential key regulatory nodes in the lignin-mediated cold response network (Fig. 8).

Fig. 8.

Co-expression gene network associated with lignin biosynthesis and its transcriptional regulation under natural low-temperature stress

To experimentally validate the expression patterns of these lignin-related genes, qRT-PCR was conducted for the five hub genes and selected co-expressed TFs (Fig. 9). Primer sequences for all validated genes are provided in Table S1. The qRT-PCR results confirmed the reliability of RNA-seq data, showing a significant positive correlation between the two datasets (r = 0.6936, p < 0.01, Fig. S8).

Fig. 9.

Relative expression levels of 10 DEGs as determined by RNA sequencing and qRT-PCR across comparison groups. Note: Different lowercase letters indicate significant differences at the 0.05 level. Black font represents the transcriptome results; red font represents qRT-PCR results

Discussion

LT₅₀ was a pivotal indicator of plant cold hardiness [37, 38], yet reliable LT₅₀-based cold resistance evaluations for pear germplasm remain scarce, with most prior studies relying on field freezing injury assessments [39]. Our standardized LT₅₀ determination via TTC staining combined with logistic equation addresses this gap, as the significant correlation between LT₅₀ values and field freezing injury indices confirms the validity of LT₅₀ in reflecting actual cold resistance performance in pear trees. This aligns with broader research emphasizing LT₅₀ as a direct and accurate cold tolerance metric, while the cold hardiness classification (e.g., ‘Shanli’ as highly cold-resistant and ‘Hanhong’ as cold-sensitive) provides actionable insights for germplasm utilization (Table 2), complementing existing frameworks for fruit tree cold resistance grading.

Plant cold tolerance is shaped by complex physiological processes, with branch structural traits emerging as key regulators [40]. Our finding that xylem ratio serves as a critical parameter for pear cold resistance (Table 5) echoes observations in raspberry stems [41], where higher xylem ratios correlate with enhanced cold hardiness (Table 3). This association likely stems from the xylem’s dual role in water transport and mechanical support—traits essential for withstanding low-temperature-induced stress [42, 43]. The contrasting freezing injury patterns between cold-resistant (‘Shanli’) and cold-sensitive (‘Hanhong’) materials (Fig. 2B, C) further underscore the xylem’s importance in mediating cold resistance, highlighting structural adaptations as a core component of pear cold acclimation.

Beyond structural adaptations, physiological and biochemical adjustments, particularly lignin metabolism, are central to pear cold tolerance [44, 45]. Among the suite of physiological parameters assessed, POD activity and lignin content exhibited the strongest associations with LT₅₀ (Table 5), consistent with lignin’s role as a first line of defense against external stresses by enhancing cell wall rigidity [46]. Notably, lignin accumulation in ‘Shanli’ was predominantly localized in the xylem under natural low temperatures (Fig. 2D), a pattern that likely reinforces branch integrity. While lignin’s role in plant cold tolerance is well-documented, its biosynthetic mechanism in pear remains poorly understood, making our focus on the phenylpropanoid pathway particularly relevant.

Our investigation into the phenylpropanoid pathway reveals key regulatory nodes underlying lignin-mediated cold tolerance. The significant correlation between HCT, CSE, F5H, CAD, and POD activities and lignin biosynthesis aligns with previous studies showing that abiotic stress upregulates these enzymes to promote G and S lignin monomer accumulation and improve the S/G ratio [29, 46]. POD, in particular, plays a dual role in lignin polymerization and ROS scavenging, both of which enhance branch cold hardiness [42, 47]. Metabolomics identified ferulic acid and sinapic acid as key metabolites driving lignin content differences between ‘Shanli’ and ‘Hanhong’ (Fig. 6), with their accumulation linked to the expression of five hub genes encoding the aforementioned key enzymes (Fig. 3)—among which LOC125473073 (POD) may be critical, echoing POD’s central role in lignin synthesis and stress tolerance across plant species.

Transcriptional regulation further modulates lignin biosynthesis under low-temperature stress [10, 15, 47]. While MYB and NAC transcription factors are well-established regulators of lignin biosynthetic genes in plants [15, 23, 48] (including pear fruit stone cell formation), our identification of ERF105-like (LOC103959790), ZC3H48 (LOC103964649), and NAC17-like (LOC125475748) as co-expressed with hub genes expands the known regulatory network (Fig. 8). ERF105-like genes have been linked to drought resistance in mulberry [49], but their role in pear abiotic stress tolerance is novel, presenting a promising target for further investigation.

In conclusion, our findings establish xylem architectural traits, lignin metabolism via the phenylpropanoid pathway, and a network of candidate regulatory genes (particularly LOC125473073 (POD) and LOC103959790 (ERF105-like)) as core determinants of pear branch cold tolerance. These insights address significant gaps in pear cold acclimation research by providing the first integrated, multi-omics characterization of lignin-mediated freezing tolerance in this economically important fruit crop. The identified candidate genes and metabolic markers offer promising targets for marker-assisted selection and molecular breeding of cold-resistant pear cultivars. Future research should prioritize: (i) functional characterization of hub structural genes and co-expressed transcription factors using stable transformation or transient expression systems; (ii) chromatin immunoprecipitation and promoter analysis to delineate direct regulatory interactions; and (iii) field validation of the identified molecular markers across diverse genetic backgrounds and environmental conditions. Such efforts will advance our mechanistic understanding of lignin-mediated cold tolerance and accelerate the development of cold-hardy pear cultivars adapted to temperate and boreal production zones.

Conclusion

Based on the LT₅₀ values obtained in this study, significant differences in cold resistance among pear cultivar germplasms were observed. Lignin biosynthesis via the phenylpropanoid pathway is a core adaptive mechanism for pear cold acclimation, where differential lignin accumulation in pear branches is driven by coordinated changes in anatomical structure, key potential phenylpropanoid enzyme activities, intermediate metabolite levels, and pathway gene expression—ultimately leading to divergent cold tolerance among pear cultivars.

Abbreviations

LT₅₀

Semi-lethal low temperature

ROS

Reactive oxygen species

TTC

2,3,5-triphenyltetrazolium chloride

MDA

Malondialdehyde

H₂O₂

Hydrogen peroxide

SOD

Superoxide dismutase

CAT

Catalase

POD

Peroxidase

PAL

Phenylalanine ammonia-lyase

C4H

Cinnamate 4-hydroxylase

4CL

4-Coumarate-CoA ligase

HCT

Shikimate hydroxycinnamoyl transferase

CSE

Caffeoyl shikimate esterase

COMT

Caffeic acid O-methyltransferase

CCoAOMT

Caffeoyl-CoA 3-O-methyltransferase

F5H

Ferulate 5-hydroxylase

CAD

Cinnamyl alcohol dehydrogenase

CCR

Cinnamoyl-CoA reductase

LAC

Laccase

UGT

Uridine diphosphate glycosyltransferases

ELISA

Enzyme-linked immunosorbent assay

UPLC

Ultra-performance liquid chromatography

LC-MS/MS

Liquid chromatography–tandem mass spectrometry

OPLS-DA

Orthogonal partial least squares-discriminant Analysis

PCA

Principal component analysis

VIP value

Variable importance in the projection value

DAMs

Different accumulation metabolites

cDNA

Complementary DNA

FPKM

Fragments per kilobase of transcript per million fragments mapped

DEGs

Differentially expressed genes

KEGG

Kyoto Encyclopedia of Genes and Genomes

qRT-PCR

Quantitative real-time PCR

Authors’ contributions

Ying Zhao, Maojun Zhang and Qiang Wang designed research and analyzed data; Ying Zhao, Xingkai Yan, Mingyan Lu, Chunhao Wu performed experiments; Ying Zhao wrote the paper.

Funding

This work was supported by the China Agriculture (Pear) Research System ( CARS-28); Jilin Provincial Science and Technology Development Program Project (20250402002ZP); Innovation Engineering Project of Jilin Province (CXGC2022RCG002).

Data availability

The raw RNA sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under the BioProject accession number PRJNA1394719 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1394719).

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.