The susceptibility of tea plants to C. camelliae was evaluated across different grafting combinations, revealing markedly distinct outcomes. The ungrafted susceptible scion cultivar ‘818’ developed the largest lesion diameter, averaging 11.67 ± 0.74 mm. In contrast, the resistant rootstock ‘419’ demonstrated strong resistance, with lesion diameters limited to 1.13 ± 0.07 mm. Self-grafted ‘818/818’ plants exhibited a significant 50.9% reduction in lesion size compared to ungrafted ‘818’. Remarkably, hetero-grafted ‘818/419’ plants showed an even greater decrease, with lesion diameters reduced by 98.7% relative to ungrafted ‘818’ and by 80.3% compared to ‘818/818’ (all differences statistically significant; p < 0.05) (Figures 1A, B). These findings robustly confirm that grafting substantially enhances resistance to C. camelliae in the susceptible cultivar, with hetero-grafting onto resistant ‘419’ rootstock conferring superior protective effects. This underscores the potential of resistant rootstocks to modulate scion disease resistance through grafting.

To elucidate the physiological and biochemical bases underpinning grafting-enhanced disease resistance, we systematically analyzed key components in the scion leaves of ‘818’, ‘419’, and grafted plants (‘818/818’ and ‘818/419’).

To explore molecular mechanisms of grafting-enhanced disease resistance, transcriptome sequencing was performed on scions from ungrafted ‘818’, self-grafted ‘818/818’, and hetero-grafted ‘818/419’. High-quality data totaled 56.77 Gb, with individual samples yielding ≥5.96 Gb clean reads, Q30>96.6%, and genome mapping rates >85%. Differential expression analysis identified 229 differentially expressed genes (DEGs; 75 upregulated, 154 downregulated) in ‘818/818’ vs ‘818’; 565 DEGs (155 upregulated, 410 downregulated) in ‘818/419’ vs ‘818/818’; and 1205 DEGs (299 upregulated, 906 downregulated) in ‘818/419’ vs ‘818’ (Supplementary Figure S1A), indicating stronger regulatory effects in plants grafted on resistant rootstocks.

Gene ontology (GO) enrichment revealed that grafting primarily modulated genes involved in metabolic and cellular processes, cellular components, catalytic activity, and binding, whereas resistant rootstocks broadened activation to include metabolic processes, signal transduction, and transcriptional regulation (Figures 3A, B). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis showed DEGs in ‘818 vs 818/818’ enriched pathways such as starch/sucrose metabolism, galactose metabolism, unsaturated fatty acid biosynthesis, phenylpropanoid biosynthesis, and peroxisome function (Figures 3C, D). In contrast, ‘818/818 vs 818/419’ DEGs were enriched in glutathione, nitrogen, sulfur metabolism, linoleic acid metabolism, pentose phosphate pathway, carbon metabolism, photosynthesis, and stress-response pathways including chloroplast light-harvesting complex assembly and cysteine/methionine metabolism. Thus, grafting mainly induced metabolic remodeling and cellular adaptation facilitating grafting stress tolerance and transport, while resistant rootstocks potentiated defense-related metabolic and signaling pathways in the scion.

Small RNA sequencing of scions from ‘818’, ‘818/818’, and ‘818/419’ yielded 109.68 million clean reads, with Q30 > 85%, and sequence lengths predominantly 21–24 nt, characteristic of plant miRNAs. A total of 615 miRNAs were identified (88 known, 527 novel). Differential expression analysis revealed 126 differentially expressed miRNA (DEMs;48 upregulated, 78 downregulated) in ‘818 vs 818/818’ and 31 DEMs (15 upregulated, 16 downregulated) in ‘818/818 vs 818/419’ (Supplementary Figure S1B). KEGG enrichment of DEM targets indicated involvement in plant hormone signaling, pathogen interaction, homologous recombination, and mRNA surveillance in ‘818 vs 818/818’ (Figure 4A); and additionally sulfur metabolism, steroid biosynthesis, diarylheptanoid/gingerol biosynthesis, endocytosis, and photosynthesis regulation in ‘818/818 vs 818/419’ (Figure 4B). These data suggest that grafting mediates physiological adaptation and defense enhancement via miRNA-regulated networks governing defense, tissue repair, and metabolism.

Widely targeted metabolomics analysis detected 1870 metabolites in scion leaves. Differential analysis identified 430 differential metabolites (261 upregulated, 169 downregulated) in ‘818 vs 818/818’, and 361 (171 upregulated, 190 downregulated) in ‘818/818 vs 818/419’ (Supplementary Figure S1C). KEGG enrichment showed upregulated metabolites in ‘818 vs 818/818’ enriched in monoterpenoid biosynthesis, phenylalanine metabolism, and ubiquinone/terpenoid-quinone biosynthesis; downregulated ones were enriched in carboxylic acid metabolism, the glyoxylate cycle, glycerophospholipid metabolism, and beta-alanine metabolism (Figure 5A). In the ‘818/818 vs 818/419’ comparison, upregulated metabolites were enriched in glycerophospholipid metabolism, caffeine metabolism, and phosphatidylinositol signaling; downregulated metabolites were involved in acetyl-CoA and lipid pathways (Figure 5B). These results indicate that grafting modulated primary energy metabolism and enhanced accumulation of secondary metabolites related to quality and disease resistance. Resistant rootstock ‘419’ further amplified disease-relevant metabolic pathways, promoting biosynthesis and accumulation of defense compounds that improve resistance and quality.

Integrative analysis of transcriptomic and metabolomic data revealed that co-expressed DEGs and metabolites in the comparison ‘818 vs 818/818’ were enriched in 34 KEGG pathways, predominantly starch and sucrose metabolism, phenylalanine metabolism, carbon metabolism, and isoflavonoid biosynthesis (Figures 5C, D). In ‘818/818 vs 818/419’, 49 enriched pathways were identified, including plant hormone signal transduction, carbon metabolism, glutathione metabolism, amino acid biosynthesis, and photosynthesis regulation (Figures 5E, F). These results suggest that grafting modulates basal metabolism and secondary metabolite synthesis, while resistant rootstocks activate comprehensive defense-related gene expression and metabolite accumulation, enhancing the plant’s immune response. Complementary miRNA sequencing data corroborated the observed enrichment of target genes in hormone signaling and glutathione metabolism pathways, underscoring the miRNA-mediated post-transcriptional regulation coordinating these responses.

To verify the sequencing data, we selected seven differentially expressed miRNAs and ten associated target genes for qRT-PCR analysis. The expression patterns were largely consistent with the high-throughput sequencing results, revealing predominant inverse correlations between the miRNAs and their putative targets (Figure 6; Supplementary Tables S2, S3). All seven validated miRNAs showed significant downregulation in the resistant hetero-graft ‘818/419’ compared to the susceptible ungrafted ‘818’ (p < 0.05). Among them, csi-miR395b-3p, novel-miR397, nta-miR160a, mdm-miR171h, novel-miR170, and vvi-miR535c were also downregulated in self-grafted ‘818/818’ and the resistant rootstock ‘419’, whereas vvi-miR159c was specifically suppressed in ‘818/419’.

Conversely, the expression of most target genes was significantly upregulated. For example, downregulation of csi-miR395b-3p was associated with strong induction of its target involved in sulfur assimilation, CsAPS1, in both ‘419’ and ‘818/419’. Suppression of novel-miR397 correlated with marked upregulation of three lignin biosynthesis-related genes—CsCCoAOMT, CsCCR2, and CsDLO2—in the resistant and hetero-grafted plants. Downregulation of nta-miR160a coincided with elevated levels of CsLAX2 (an auxin transporter) and CsCYP78A9, indicating enhanced auxin signaling. Similarly, downregulation of vvi-miR159c and mdm-miR171h aligned with upregulation of their respective putative targets: the transcription factor CsMYB30 and the signaling gene CsSCL6.

Notably, the self-grafted ‘818/818’ typically displayed intermediate expression changes, whereas hetero-grafting with ‘419’ elicited the most substantial transcriptional reprogramming. These validated miRNA-mRNA pairs constitute a core post-transcriptional regulatory network that underlies the grafting-enhanced defense response.

Grafting markedly altered the scion’s hormonal profile, especially in the heterograft ‘818/419’, where IAA and GA₃ contents increased significantly. These growth-promoting hormones likely support wound healing and vigor, thereby contributing to enhanced resistance (Sharma and Zheng, 2019; Zhang et al., 2022). In parallel, MeSA, a mobile derivative of SA associated with defense priming (Lan et al., 2025), also accumulated strongly in both self- and hetero-grafted plants, suggesting that grafting reprograms SA metabolism toward enhanced defensive readiness. At the enzymatic level, heterografting with the resistant rootstock elevated PAL activity, a key enzyme for lignin and flavonoid biosynthesis (Dixon et al., 2002), while reducing SOD and POD activities, indicating a shift from broad antioxidant responses toward specialized defense pathways. Taken together, these findings suggest that grafting, particularly onto resistant rootstocks, enhances tea plant resistance to C. camelliae by promoting IAA- and GA3-mediated growth, strengthening SA-derived MeSA defense signaling, and activating PAL-driven secondary metabolism.

The miRNA-mRNA expression patterns described above provide a mechanistic basis for understanding how resistant rootstocks influence scion defense responses. Building on these results, small RNA sequencing and qPCR validation further revealed that grafting, particularly hetero-grafting with resistant rootstock ‘419’, induces pronounced shifts in scion miRNA expression profiles, accompanied by reciprocal changes in their target or putative target genes. For example, csi-miR395b-3p was significantly downregulated in ‘419’ and ‘818/419’, while its target CsAPS1 and related sulfur metabolism gene CsLCD were upregulated, a result consistent with the known role of miR395’s in sulfur homeostasis (Kawashima et al., 2009; Tao et al., 2012). Enhanced sulfur metabolism is known to improve antioxidant capacity via glutathione synthesis and production of sulfur-containing defense compounds (Bednarek et al., 2009; Hacham et al., 2025).

Our study resulted in the identification of a novel miRNA, that we propose to name ‘novel-miR397’ (Supplementary Table S4). The downregulation of this novel-miR397 correlated with increased expression of three genes, CsCCoAOMT, CsCCR2, and CsDLO2, that are known to be linked with reinforcing lignin biosynthesis and structural barriers (Huang et al., 2021; Xue et al., 2019; Wang et al., 2019). Reduced nta-miR160a was linked to elevated CsLAX2 and CsCYP78A9, indicating enhanced auxin transport and organ development (Hao et al., 2022; Moreno-Piovano et al., 2017; Sotelo-Silveira et al., 2013). Downregulation of vvi-miR159c corresponded with higher expression of CsMYB30, a gene known to code for a stress-responsive transcription factor (Millar et al., 2019; Fichman et al., 2020), while reduced mdm-miR171h matched increased expression of CsSCL6, a gene involved in hormone signaling (Wang et al., 2020b; Yang et al., 2010). These coordinated miRNA–mRNA shifts, with intermediate changes in self-grafted ‘818/818’, support a systemic, miRNA-driven regulatory mechanism in graft-mediated resistance, a finding consistent with the role proposed by Melnyk et al. (2011b) for long-distance miRNA transport.

Integrated metabolomic and transcriptomic analyses demonstrated that grafting, especially hetero-grafting with resistant rootstock ‘419’, reprograms scion metabolism and gene expression. In ‘818/419’, upregulated metabolites were enriched in glycerophospholipid metabolism, phosphatidylinositol signaling, and caffeine biosynthesis, processes linked in previous studies to membrane stability, immune signal transduction, and pathogen inhibition (Guo et al., 2023; Abd-El-Haliem and Joosten, 2017; Zhang et al., 2010). Integrated metabolomic and transcriptomic analyses revealed that in ‘818/419’ scion leaves, differential metabolites and genes in pathways such as plant hormone signal transduction, glutathione metabolism, and phenylpropanoid biosynthesis exhibited synergistic expression, demonstrating a tight transcriptional-metabolic coupling. This further demonstrates that the increased expression of key genes such as CsAPS1, CsCCoAOMT, CsCCR2, CsDLO2, CsLAX2 and CsLCD promotes plant hormone and lignin synthesis, as well as signal transduction. These changes parallel the downregulation of their regulatory miRNAs, reinforcing the link between post-transcriptional control and metabolic adaptation.

Notably, the defense-related metabolic reprogramming induced by the resistant rootstock did not compromise the key quality-determining metabolites of the scion. ‘818/419’ was shown in our study to hold important traits associated with tea quality. It retained the characteristic high EGCG content of scion ‘818’, while showing increased EC, EGC, and L-theanine. Furthermore, it preserved two desirable traits of rootstock ‘419’: high ECG and caffeine contents. These findings demonstrate that grafting onto a resistant rootstock triggers extensive miRNA-directed regulatory cascades, broad transcriptional reprogramming, and differential enrichment of defense-associated pathways, thereby conferring robust phylloxera resistance while maintaining the scion’ s canonical levels of key quality-associated metabolites and permitting the accumulation of additional compounds contributed by the rootstock.

This study integrates multi-omics and phenotypic data to propose a hierarchical, systemic resistance network orchestrated by grafting-responsive miRNAs. The enhanced resistance in ‘818/419’ scions is a coherent cascade initiated by rootstock-derived signals (Figure 7). Specifically, the differential expression of key miRNAs (e.g., csi-miR395b-3p, novel-miR397) in the scion fine-tunes the transcriptome, leading to the upregulation of target genes governing sulfur assimilation (CsAPS1), lignin biosynthesis (CsCCoAOMT, CsCCR2), and auxin transport (CsLAX2). This transcriptional reprogramming subsequently drives a metabolic shift toward defense-ready states, characterized by the accumulation of MeSA, IAA for signaling, enhanced lignin deposition for structural reinforcement, and elevated glutathione for redox homeostasis. Consequently, the tea plant’s strategy pivots from a broad antioxidant response (reflected in altered SOD/POD activities) to a targeted deployment of specialized physical and biochemical barriers, which directly translates into the dramatic 98.71% reduction in lesion diameter.

Grafting-induced resistance conferred by resistant rootstocks is well-documented in various crops, including grapevine (Cookson et al., 2013), apple (Jensen et al., 2012), watermelon (Zhang et al., 2025), and citrus (Wang et al., 2025). The enhanced resistance in these grafted systems is primarily attributed to long-distance signaling mechanisms. In grafting systems, mobile signals such as miRNAs, hormones, and metabolites are likely transported from rootstock to scion through the phloem via source-sink dynamics, thereby reprogramming scion responses. In this study, we postulate that similar long-distance signaling enhances scion defense against C. camelliae in grafted tea plants. Numerous studies have established that 21–24 nt small RNAs, including many of the miRNAs identified here, can function as mobile silencing signals that traverse graft junctions in a selective, Argonaute-dependent manner (Melnyk et al., 2011a; Li et al., 2021). Once in the recipient scion tissue, these mobile miRNAs-transported as single-or double-stranded forms bound to Argonaute proteins-trigger target mRNA cleavage, translational repression, or epigenetic modifications such as DNA methylation. This process is frequently amplified by scion-localized RNA-dependent RNA polymerases and mirrors endogenous long-distance signaling events, such as the well-characterized shoot-to-root movement of miR399 during phosphate homeostasis (Lin et al., 2008; Pant et al., 2008). The observed changes in miRNA expression levels in the scions of grafted tea plants may therefore be associated with the mobility of miRNAs between rootstock and scion. Additionally, hormones such as MeSA, which accumulated in our hetero-grafts, represent another class of mobile signals, as volatile or phloem-transported SA derivatives, they prime systemic acquired resistance across distances (Park et al., 2007). Similarly, defense-related metabolites (e.g., from phenylpropanoid and glutathione pathways) may be conveyed with photoassimilates via the phloem, thereby enhancing scion defense compounds. Currently, our data provide indirect evidence for the long-distance transport of these substances in grafted tea plants. Future studies employing tracer assays or direct phloem profiling are needed to validate these transport mechanisms. Notably, while our multi-omics approach provides robust correlative support for miRNA-mRNA networks, functional validations such as miRNA or gene overexpression and silencing are crucial to establish causality and delineate the mechanistic details of grafting-conferred immunity.

Two-year-old tea plants (Camellia sinensis) cultivated by the Guizhou Provincial Tea Research Institute were used in this study. The C. camelliae susceptible cultivar ‘Qianmei 818’ (hereafter ‘818’) served as the scion, while the C. camelliae resistant cultivar ‘Qianmei 419’ (hereafter ‘419’) was employed as the rootstock for grafting. Grafting was conducted in spring using the cleft grafting method described by Liu et al. (2017). Two graft combinations were established: self-grafted (‘818/818’; i.e., ‘818’ scion onto ‘818’ rootstock) and hetero-grafted (‘818/419’; i.e., ‘818’ scion onto ‘419’ rootstock). Following grafting, seedlings underwent standard field management and were cultivated for one year to ensure stable growth prior to downstream experiments. This one-year period was chosen to ensure complete graft union formation, stable systemic signal transmission between rootstock and scion, and consistent physiological maturity prior to inoculation experiments, as preliminary field trials indicated that shorter periods resulted in incomplete healing and variable resistance phenotypes.

The resistance of grafted tea plants to C. camelliae was assessed using the detached leaf inoculation assay of Yoshida and Takeda (2006). The first fully expanded leaves (third leaf from the apex) were excised from healthy plants belonging to four groups: ungrafted ‘818’, ungrafted ‘419’, self-grafted ‘818/818’, and hetero-grafted ‘818/419’. The leaves were surface-sterilized by wiping with 75% ethanol and rinsed with sterile distilled water. They were blotted dry, then placed in 90-mm-diameter Petri dishes on water-soaked sterile filter paper, with their adaxial side up. The leaves were then immediately inoculated with C. camelliae stain GC-2. This strain was originally isolated from anthracnose-infected tea leaves in Guizhou Province, morphologically and phylogenetically identified as C. camelliae based on multi-gene (ITS, GAPDH, ACT, CHS) analysis combined with pathogenicity tests (as detailed in Chen, 2023, Master’s thesis, Guizhou University), and preserved in the Plant Protection Laboratory, Guizhou institute of tea science. Long-term stock cultures are maintained in 20% glycerol at -80°C, with working cultures stored on potato dextrose agar (PDA) slants at 4°C. Mycelial plugs, 5 mm in diameter, were excised from the growing margin of a 7-day old colony incubated on potato dextrose agar (PDA) at 25°C in the dark. On each leaf, a mycelial plug was deposited on the left side of the midrib, with the mycelium in direct contact with the plant tissue. A plug of non-inoculated PDA was placed on the right side of the midrib as a control. The Petri dishes were then covered with moist cotton, and incubated at 28°C in darkness. The moisture of the cotton was monitored during incubation and water was added as needed to maintain saturated humidity around the leaves. Disease development was assessed by measuring the diameter of the lesions 3 days after inoculation. To facilitate the observation of lesions with low development (diameters smaller than 5 mm) the mycelial plugs were removed from the leaves prior to their examination. Each treatment included 10 replicate leaves, and the whole experiment was carried out three times independently.

Healthy shoots comprising one bud and two leaves were excised from plants with consistent growth status in all four treatment groups (‘818’, ‘419’, ‘818/818’, and ‘818/419’). The shoots were immediately flash-frozen in liquid nitrogen and stored at -80°C until analyses were carried out. The activity of defense-related enzymes peroxidase (POD), superoxide dismutase (SOD) and phenylalanine ammonia-lyase (PAL) was quantified using commercial colorimetric assay kits (Suzhou Kaiming, China) following the manufacturer’s instructions.

The levels of plant hormones including indole-3-acetic acid (IAA), gibberellic acid (GA₃), salicylic acid (SA), and jasmonic acid (JA), alongside major catechins (epigallocatechin gallate [EGCG], epicatechin [EC], epigallocatechin [EGC], and epicatechin gallate [ECG]), L-theanine and caffeine, were determined via high-performance liquid chromatography (HPLC). Metabolites were extracted using methanol, and quantification was performed based on calibration curves generated from purified standards, with relative standard deviations maintained below 15%. Concentrations (µg/g) were calculated as follows:

Concentration = (Instrument Reading x Final Extract Volume [mL] x Dilution Factor)/Sample Mass [g].

For each of the four plant treatment groups, we analyzed three biological replicate samples, each sample being composed of shoots from five plants.

To validate RNA-Seq and sRNA-Seq findings, representative differentially expressed miRNAs and their target mRNAs were selected for quantitative real-time PCR (qRT-PCR). miRNA cDNA synthesis employed stem-loop primers, whereas mRNA was reverse transcribed via a universal reverse transcription kit (TAKARA Bio Inc., Japan). qRT-PCR was performed using SYBR Green on an ABI 7500 Real-Time PCR System. Pc-222-3p and CsGAPDH served as endogenous controls for miRNA and mRNA quantification, respectively. Relative expression was calculated using the 2^–ΔΔCt method. Each reaction included three biological replicates and three technical replicates. Primer sequences are listed in Supplementary Table S1.

Data were analyzed by one-way analysis of variance (ANOVA) using SPSS 22.0. Significant differences between groups were determined by Duncan’s multiple range test at a significance threshold of p < 0.05.