Fungal MicroRNA‐Like RNAs Facilitate Convergent Attack on a Host Immune Signalling Network

ABSTRACT

Cross‐kingdom RNA interference has emerged as an important mechanism in plant–pathogen interactions, yet how fungal small RNAs coordinate host immune suppression remains unclear. Here, we demonstrate that Alternaria alternata f. sp. mali, the causal agent of apple leaf spot, suggests microRNA‐like RNAs (milRNAs) into host tissues in association with extracellular vesicles (EVs). Small RNA sequencing of fungal EVs identified multiple milRNAs, among which AamilR251, AamilR292 and AamilR004 were the three most abundant species. These AamilRNAs target three apple (Malus domestica) immune‐related kinases (MdMAPKKK1, MdRLK2 and MdMAPKK6). Molecular and genetic analyses revealed that these proteins form an interconnected signalling module and a linear phosphorylation cascade (MdRLK2‐MdCRK10‐MdMAPKKK1‐MdMAPKK6) required for effective defence activation. Deletion of individual AamilRNAs significantly attenuated fungal virulence and restored expression of their respective host targets. Conversely, overexpression of MdMAPKKK1, MdRLK2 or MdMAPKK6 enhanced resistance to infection. Pull‐down assays identified MdCRK10 as an interacting partner of MdRLK2, forming a complex with MAPKs that activates defence signalling. MdRLK2 phosphorylates MdCRK10 at S144 and S363, and mutation of these sites disrupts signal transduction and compromises disease resistance. These findings reveal a central immune signalling network in apple and identify key components for improving resistance against fungal pathogens.

Summary statement

This study reveals a cross‐kingdom mechanism in apple (Malus domestica) where Alternaria alternata f. sp. mail milRNAs delivered via extracellular vesicles dismantle host immunity by targeting a central MdRLK2‐MdCRK10‐MAPK signalling module.

1. Introduction

Apple (Malus × domestica Borkh.) is a key global fruit crop, producing over 86 million metric tons annually (FAOSTAT, https://www.fao.org/faostat, 2023). However, fungal diseases significantly threaten the industry, impacting both yield and fruit quality. Among these, apple leaf spot caused by Alternaria alternata f. sp. mali is particularly severe, leading to defoliation, reduced photosynthesis and decreased yield (Y. Li et al. ²⁰¹³; Prechsl et al. ²⁰²³). At present, chemical control remains the primary means of reducing the incidence of apple early defoliation disease. However, the use of pesticides not only increases production costs but also pollutes the environment and endangers human health. This necessitates an in‐depth analysis of the disease resistance mechanisms of apple highly resistant varieties in response to fungal infection, as well as the pathogenic mechanisms of fungi when infecting apple highly susceptible varieties.

Understanding plant resistance mechanisms is crucial for developing these strategies (Liang et al. 2022). Plants employ two major layers of innate immunity: pattern‐triggered immunity (PTI) and effector‐triggered immunity (ETI). PTI is initiated by the recognition of pathogen‐associated molecular patterns through Receptor‐Like Kinases (RLKs), which activate immune responses (Kourelis and van der Hoorn ²⁰¹⁸; Tian et al. ²⁰²¹). In contrast, ETI is triggered when specific plant resistance (R) proteins detect pathogen effectors, resulting in a stronger, localised defence response (Kourelis and van der Hoorn ²⁰¹⁸). Both PTI and ETI involve mitogen‐activated protein kinase (MAPK) cascades, including MAPK Kinase (MAPKK) and MAPK Kinase Kinase (MAPKKK) genes, which are essential for signalling and amplifying defence responses (Feng et al. 2024). Recent studies have underscored the roles of MAPKK and MAPKKK genes in enhancing resistance to various fungal pathogens (J.‐B. Zhang et al. ²⁰²⁰). Specifically, MAPK family genes have been shown to function in responses to certain fungal diseases and abiotic stresses in apple, however, whether this critical signalling pathway is also mobilised for defence against the major pathogen ALT7 has not yet been investigated (Cheng et al. 2020; N. Wang, Liu, et al. ²⁰²²; M. F. Li et al. ²⁰²⁵). Although research on apple resistance breeding and related mechanisms can enhance the disease resistance of cultivated apples in the long run, it fails to address the practical issue of weak resistance in currently predominant apple cultivars. Therefore, while conducting studies on apple disease resistance mechanisms, it is also necessary to adopt a perspective centred on pathogenic fungi to systematically investigate their virulence mechanisms.

Current research on the pathogenic mechanisms of apple leaf spot remains insufficiently in‐depth, focusing primarily on the isolation and toxicity evaluation of pathogenic fungi, identification of toxin types, morphological characterisation of fungal cells and investigations into changes in key physiological indicators of apples during pathogen infection (Hématy et al. 2009; Underwood ²⁰¹²; Kubicek et al. ²⁰¹⁴; De Sain and Rep ²⁰¹⁵). However, the mechanisms by which pathogenic fungi break through the innate immune barrier of apples to cause large‐scale infection remain unclear. Studies have shown that some pathogens secrete effector proteins into host cells during plant infection, which suppress the host's primary defence responses to facilitate pathogen colonisation. These effectors are typically proteins, but recent research has revealed that cross‐kingdom small RNAs (sRNAs) produced by fungi can also act as effectors, entering plant cells and reducing plant resistance (J. Hu et al. ²⁰¹²; Gong et al. ²⁰²⁴; Tang et al. ²⁰²⁵). Notably, in filamentous fungi, milRNAs have emerged as key regulators of fungal physiology and virulence, acting as effector molecules that manipulate host defences (Tamburini et al. 2020). A milRNA in Fusarium oxysporum can damage the host immune response, thereby conferring resistance to tomato wilt disease (Ji et al. 2021). Similarly, an milRNA in Magnaporthe oryzae can mediate epigenetic suppression of a virulence gene (Jin et al. 2019). MilRNAs in Valsa Mali can regulate virulence genes and thus regulate the ability of fungi to infect hosts (Xu et al. 2020, ²⁰²²). These results indicate that sRNAs secreted by fungi may also serve as a means for them to reduce the resistance of host plants. However, such research has so far been limited to the infection processes of fungi on a few herbaceous plants, and there have been no relevant reports on whether the immune system of apple trees is also regulated by sRNAs released by the pathogen during infection by the causal agent of apple early defoliation disease. Extracellular vesicles (EVs) are evolutionarily conserved vehicles for intercellular communication (Mulcahy et al. 2014; Mukherjee et al. ²⁰¹⁶; Costa Verdera et al. ²⁰¹⁷; Chow et al. ²⁰¹⁹; Leidal et al. ²⁰²⁰). EVs have recently been characterised in several plant pathogenic fungi, such as Ustilago maydis (maize smut), Zymoseptoria tritici (wheat pathogen) and Fusarium oxysporum f. sp. vasinfectum (cotton pathogen) (Lee et al. 2010; Weiberg et al. ²⁰¹³; Villalobos‐Escobedo et al. ²⁰¹⁶; M. Wang et al. ²⁰¹⁶; M. Wang et al. ²⁰¹⁷; Derbyshire et al. ²⁰¹⁹; Jian and Liang ²⁰¹⁹; Yin et al. ²⁰²⁰; Dunker et al. ²⁰²⁰; Xu et al. ²⁰²⁰, ²⁰²²; Ji et al. ²⁰²¹; C. Hu et al. ²⁰²²; Kusch et al. ²⁰²³). Despite their critical role in cross‐kingdom communication, the mechanisms by which fungal EVs mediate the transfer of molecules across host–pathogen boundaries remain poorly understood. In view of this, our research group aims to identify candidate sRNAs from the apple early defoliation pathogen that target and regulate apple disease‐resistant genes, and to analyse their mechanism of action during the pathogen's infection of apple trees.

Apple leaf spot, primarily caused by ALT7, poses a significant threat to apple cultivation worldwide. Yet, the molecular mechanisms by which ALT7 employs milRNAs to modulate apple defences have not been systematically investigated. We performed high‐throughput sRNA sequencing on three EV sample types: A. alternata f. sp. mali strain ALT7, infected leaf tissues of in vitro‐grown apple susceptible variety GL‐3 plantlets at 24 h post‐inoculation (hpi) with ALT7, and non‐inoculated GL‐3 in vitro plantlets. By integrating and aligning with genomic information of apple and A. alternata f. sp. mali, we identified 28 candidate Alternaria sRNAs that met the screening criteria, designated as AamiRNA‐like RNAs (AamilRNAs). Among these, AamilR251, AamilR292 and AamilR004 were highly expressed in ALT7‐EVs and ALT7‐GL‐3‐EVs, and they targeted three apple resistant genes, namely MdMAPKKK1, MdRLK2 and MdMAPKK6. Then, we establish that this defence is orchestrated by an MdRLK2‐MdCRK10‐MAPK phosphorylation cascade, providing a direct link between a pathogen's trans‐kingdom RNAi strategy and the subversion of a specific host signalling hub. These findings offer novel molecular targets for engineering durable disease resistance in apple.

2. Results

2.1. Detection of EV‐Associated Fungal MilRNAs in Apple Tissues During Infection

To visualise the infection interface between A. alternata f. sp. mali (ALT7) and apple leaves (Gala‐3, GL‐3), transmission electron microscopy (TEM) was performed at multiple time points following inoculation. Vesicle‐like structures were observed accumulating in the apoplastic space at 12 and 24 hpi, suggesting active EV secretion during infection (Figure 1A). EVs isolated from ALT7 culture supernatants displayed the characteristic cup‐shaped morphology under TEM (Figure 1B). Nanoparticle tracking analysis further revealed a size distribution centred at approximately 158 nm, consistent with previously reported fungal EVs (Figure 1C). Together, these observations confirm that ALT7 produces EVs both in vitro and during host colonisation.

Figure 1.

Identification of microRNA‐like RNAs in extracellular vesicles (EVs). (A) Transmission electron microscopy (TEM) of the interface between Malus domestica (GL‐3) leaf cells and Alternaria alternata (ALT7) at 0, 12 and 24 h post‐inoculation (hpi). EV‐like structures (arrows) accumulate in the apoplastic space between the plant cell wall and fungal hyphae in a time‐dependent manner. Scale bars, 2 µm (overview) and 500 nm (magnified views). (B) Morphological analysis of purified ALT7 EVs by TEM. The micrograph confirms the presence of vesicles with the typical spherical or cup‐shaped morphology and a clear lipid bilayer membrane. EVs were purified from ALT7 culture supernatant and visualised by TEM after negative staining with 2% uranyl acetate. Scale bar, 200 nm (left) and 100 nm (right). (C) Characterisation of ALT7 EV size and concentration by nanoparticle tracking analysis (NTA). A representative size distribution profile from purified ALT7 EVs is shown, plotting particle concentration (particles/mL) against particle diameter (nm). The analysis revealed a mean particle size of 158.4 nm. (D) Small RNA sequencing (sRNA‐seq) of EVs isolated from GL‐3 leaves (GL‐3‐EVs), ALT7 cultures (ALT7‐EVs) and ALT7‐inoculated GL‐3 leaves (ALT7‐GL‐3‐EVs, 24 hpi). Heatmap shows log10‐transformed normalised abundance (log10(count+1)) of representative milRNAs; milRNAs highlighted in red were selected for downstream analyses. (E) Northern blot indicates that a series of AamilRNAs (AamilR251, AamilR292 and AamilR004) are stably present in EVs isolated directly from the ALT7 (ALT7‐EVs), as well as in EVs isolated from GL‐3 plant leaves 48 h post‐inoculation with ALT7 (ALT7‐GL‐3‐EVs). In contrast, these AamilRNAs were undetectable in EVs from uninoculated control plants (GL‐3‐EVs). To verify that these AamilRNAs are protected by encapsulation within the EVs, the samples were subjected to enzymatic treatments. TR: Samples were treated with 1% Triton X‐100 followed by RNase A. RPT: Sequential treatment with Proteinase K, RNase A and 1% Triton X‐100. T: Treatment with 1% Triton X‐100 only. 18S rRNA and U6 were used as loading controls for the ALT7 and GL‐3 samples. (F). In situ hybridisation revealed the accumulation of AamilR251, AamilR292 and AamilR004 in GL‐3 at 48 h post‐inoculation. No signal is detected in the control (GL‐3). 18S rRNA and Negative serve as positive and negative controls, respectively. No probe as mock control. Scale bar, 100 μm.

To investigate whether fungal sRNAs are transported via EVs, sRNA sequencing was conducted on EVs derived from ALT7 cultures, ALT7‐infected GL‐3 leaves (24 hpi) and mock‐treated GL‐3 leaves. BLAST analysis against the miRbase database (http://www.mirbase.org/) and A. alternata and apple genome sequences identified 28 microRNA‐like RNAs (milRNAs) in EVs from both ALT7 and ALT7‐GL‐3, but absent in those from mock‐treated GL‐3 leaves (Figure 1D). Of these, we selected three highly expressed milRNAs (AamilR251, AamilR292 and AamilR004) whose predicted target genes are disease resistance‐related for experimental validation (Figure 1D). Using UNAFLOD (https://www.unafold.org/), we predicted the structures of precursor sequences AaMILR251, AaMILR292 and AaMILR004 (Figure S1A). All three precursors formed typical hairpin structures, with mature AamilR251, AamilR292 and AamilR004 generated from the 3'end of these precursors (Figure S1). To confirm the presence of three milRNAs in ALT7‐Evs and ALT7‐GL‐3‐EVs, we performed northern blot analysis. This result showed that AamilR251, AamilR292 and AamilR004 were detected in ALT7‐GL‐3‐EVs and ALT7‐EVs, but not in GL‐3‐EVs (Figure S1B). To confirm the physical encapsulation of AamilR251, AamilR292 and AamilR004 within EVs, we performed RNase protection assays on EVs. Following incubation with RNase A and proteinase K, the characteristic bands for all three AamilRNAs remained protected from enzymatic degradation, whereas they were no longer detectable upon disruption of the vesicular membrane by detergent (Figure 1E). 18S rRNA and U6 were used as loading controls for the ALT7 and GL‐3 samples, respectively (Figure 1E). These results further confirm that these AamilRNAs are effectively sequestered within the EVs. Furthermore, to verify whether these milRNAs could be transferred from the fungus to apple cells, we used GL‐3 leaves inoculated with ALT7 for 48 h and untreated (Control) leaves. In situ hybridisation detected purple digoxigenin signals for AamilR251, AamilR292, AamilR004 and 18S rRNA (designated by ALT7) in inoculated GL‐3 leaves (ALT7‐GL‐3), whereas these signals were absent in wild‐type (WA) GL‐3 leaves (Figure 1F). These results indicate that we identified three fungal‐derived milRNAs‐AamilR251, AamilR292 and AamilR004‐that can translocate from the fungus to apple leaf cells as trans‐kingdom milRNAs and are highly expressed in leaves of apple susceptible varieties inoculated with ALT7.

2.2. AamilRNAs Associate With MdAGO1 and Mediate Cleavage of Immune Kinase Transcripts

To further verify whether fungal AamilRNAs bind to apple AGO proteins and cleave their target genes in apple cells, thereby exerting their biological functions, we first performed a phylogenetic analysis of MdAGO proteins with AtAGO. The results revealed that MdAGO1 has the highest homology with AtAGO1 (Figure S2). Subsequently, we purified the PAZ and PIWI domains of MdAGO1 and performed electrophoretic mobility shift assays (EMSA) to assess physical interaction. AamilR251, AamilR292 and AamilR004 specifically bound to the PAZ domain but not to the PIWI domain, consistent with canonical sRNA loading into AGO complexes (Figure 2A and Figure S3).

Figure 2.

Functional validation of a convergent targeting strategy by fungal milRNAs. (A) Electrophoretic mobility shift assay confirms direct binding of AamilR251, AamilR292 and AamilR004 to MdAGO1‐PAZ. (B) Chromosomal locations of the target mRNAs MdMAPKK6, MdRLK2 and MdMAPKKK1. (C) AamilR251 cleavage site in the PKc_MAPKKK domains of target gene were identified by 5’ RACE. AamilR292 cleavage site in the LRR domains of the target gene identified by 5’ RACE. AamilR004 cleavage site in the STKc_MAPKKK domains of target gene identified by 5’ RACE. The arrows indicate the positions of inferred cleavage sites, and the numbers above the sequences indicate the detected cleavage sites of independent clones. (D) AamilR251, AamilR292 and AamilR004 cleavage MdMAPKKK1, MdRLK2 and MdMAPKK6 identified by Nicotiana tabacum ectopic expression. Fluorescence signals were detectable upon individual expression of GFP‐MdMAPKKK1, GFP‐MdRLK2 and GFP‐MdMAPKK6. However, co‐expression of OE‐AaMILR251+GFP‐MdMAPKKK1, OE‐AaMILR292+GFP‐MdRLK2 and OE‐AaMILR004+GFP‐MdMAPKK6 resulted in the absence of fluorescence signals. pSuper1300‐cGFP was used as a control. Fluorescence signals at 488 nm. Scale bars = 20 μm. (E–G) qRT‐PCR quantification of target gene silencing corresponding to the assays in N. tabacum. Transcript levels of MdMAPKKK1 (E), MdRLK2 (F) and MdMAPKK6 (G) were significantly reduced upon co‐expression with their specific milRNAs compared to the target‐only controls. Data are presented as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences (*p < 0.05, **p < 0.01, one‐way ANOVA with Dunnett's test). [Color figure can be viewed at wileyonlinelibrary.com]

Then, we predicted the target genes of these milRNAs using psRNATarget (https://www.zhaolab.org/psRNATarget/), and they targeted three apple resistant genes, namely MdMAPKKK1, MdRLK2 and MdMAPKK6 (Figure 2B). MdMAPKKK1 is located on chromosome 15, MdRLK2 on chromosome 13 and MdMAPKKK6 on chromosome 9 (Figure 2B). To further confirm the effective cleavage of apple disease resistance genes by fungal AamilRNAs, we conducted 5ʹ RACE assays. We found that all target mRNAs were cleaved in apple leaves 48 h post‐ALT7 inoculation (Figure 2C). In addition, in tobacco plants ectopically expressing these milRNAs and their target mRNAs, fluorescence signals were detected when the target mRNAs were overexpressed, but not when the milRNAs and target mRNAs were co‐expressed (Figure 2D). RT‐qPCR further showed that, compared to overexpression of the target mRNAs alone, co‐expression of milRNAs and target mRNAs resulted in reduced expression levels of MdMAPKKK1, MdRLK2 and MdMAPKK6 (Figure 2E–G). Collectively, these results demonstrate that EV‐delivered AamilRNAs hijack the host MdAGO1 machinery to mediate sequence‐specific cleavage of immune kinase transcripts.

2.3. AamilRNAs Mediate Apple Pathogenicity by Targeting Disease Resistance Genes

To determine whether AamilRNAs contribute to fungal pathogenicity, we first generated individual overexpression strains for AamilR251, AamilR292 and AamilR004 in ALT7. Elevated accumulation of each AamilRNA was confirmed in independent transformants (Figures S4 and S5A–C). Upon inoculation onto leaves of the resistant cultivar HF (Malus domestica, ‘HanFu’), overexpression strains exhibited significantly reduced transcript levels of MdMAPKKK1, MdRLK2 and MdMAPKK6 at 48 hpi compared with WA and vector controls (Figure S5D,G,J). These results indicate that elevated accumulation of individual AamilRNAs enhances repression of immune kinase transcripts in planta. We next assessed the impact of this molecular repression on disease development. Overexpression strains produced significantly larger lesions than controls, demonstrating enhanced virulence (Figure S5E,F,H,I,K,L).

To further substantiate the functional importance of these AamilRNAs, individual knockout mutants were generated. Deletion of each AamilRNA restored the expression of its corresponding host target gene during infection (Figure 3A–C; Figure S6). Subsequently, spore suspensions of the knockout strains were inoculated onto the susceptible apple cultivar GL‐3. We analysed the expression of the target mRNA MdMAPKKK1, MdRLK2 and MdMAPKKK6 in samples collected 48 h after fungal inoculation and found that the expression levels of MdMAPKKK1, MdRLK2 and MdMAPKKK6 in GL‐3 were significantly increased in the knockout strains compared to the ALT7 and pCX62 (Figure 3D,G,J). Additionally, disease symptoms and lesion areas of GL‐3 were measured 48 hpi, we found that knockout of AamilR251, AamilR292 and AamilR004, significantly increased the resistance of susceptible cultivar GL‐3 compared to ALT7 and empty vector‐inoculation (Figure 3E,F,H,I,K,L). These results further demonstrate that deletion of AamilR251, AamilR292 and AamilR004 in ALT7 attenuates its pathogenicity to apples during infection and improves resistance in susceptible cultivars.

Figure 3.

AamilRNA knockout reduces ALT7 virulence, and host target‐gene overexpression enhances resistance. (A–C) Northern blot validation of milRNA knockout. Abundance of AamilR251 (A), AamilR292 (B) and AamilR004 (C) was significantly reduced in the respective knockout strains (KO#1‐#3) compared to the wild‐type (ALT7) and empty vector (pCX62) controls. 18S rRNA served as the loading control. (D–F) Functional analysis of AamilR251 deletion. (D) RT‐qPCR showing significantly elevated MdMAPKKK1 transcript levels in susceptible ‘Gala‐3’ (GL‐3) leaves inoculated with KO‐AamilR251 strains at 48 hpi. (E) Representative disease symptoms. (F) Quantification of lesion areas, showing attenuated pathogenicity in KO‐AamilR251 infected leaves. (G–I) Functional analysis of AamilR292 deletion. (G) RT‐qPCR showing elevated MdRLK2 levels in GL‐3 leaves inoculated with KO‐AamilR292 strains. (H) Disease symptoms. (I) Lesion area quantification. (J–L) Functional analysis of AamilR004 deletion. (J) RT‐qPCR showing elevated MdMAPKK6 levels in GL‐3 leaves inoculated with KO‐AamilR004 strains. (K) Disease symptoms. (L) Lesion area quantification. (M–N) Overexpression of host target genes enhances resistance to ALT7. (M) Representative symptoms on GL‐3 leaves overexpressing MdRLK2, MdMAPKKK1 or MdMAPKK6 (agroinfiltration; fungal inoculation at 72 h after infiltration; symptoms shown at 48 hpi). (N) Lesion area quantification. Data represent mean ± SD (n = 3 biological replicates for qPCR; n = 10 for lesion area). Asterisks indicate significant differences compared to ALT7 controls (*p < 0.05, **p < 0.01, one‐way ANOVA with Dunnett's test). [Color figure can be viewed at wileyonlinelibrary.com]

To determine whether these target mRNAs are involved in the defence pathway, we used Agrobacterium‐mediated transient overexpression of these target mRNAs in GL‐3 plants. Four days after inoculation, the expression levels of the target mRNAs (MdMAPKKK1, MdRLK2 and MdMAPKKK6) were confirmed to be upregulated (Figure S7A–C). When ALT7 was subsequently inoculated, the lesion areas were measured after 48 h, and it was observed that overexpression of these target mRNAs enhanced the resistance of GL‐3 to ALT7 (Figure 3M,N). We also used RNAi to silence these target mRNAs by generating siRNAs in HF. Four days after Agrobacterium inoculation, the expression levels of these target mRNAs were confirmed to be downregulated (Figure S7D–F). Following ALT7 inoculation, lesion areas were measured 48 hpi, and it was found that silencing of these target mRNAs decreased the resistance of HF to ALT7 (Figure S7G,H). These results indicate that the fungal milRNAs AamilR251, AamilR292 and AamilR004 regulate ALT7's pathogenicity by targeting and cleaving the apple resistance genes MdMAPKKK1, MdRLK2 and MdMAPKK6, which are involved in apple's defence pathways against ALT7.

2.4. MdCRK10, MdRLK2 and the MAPK Cascade Form a Cohesive Signalling Module

To further verify the interaction among the three target genes MdMAPKKK1, MdRLK2 and MdMAPKK6, as well as their functional co‐localisation in the same apple disease resistance pathway, we first performed protein‐protein interaction assays to clarify the interaction relationships among the three target proteins. We conducted a bimolecular fluorescence complementation assay in N. benthamiana leaves by using Agrobacterium tumefaciens infiltration. As expected, leaf epidermal cells expressing MdMAPKKK1‐YFP^N/MdMAPKK6‐YFP^C showed strong fluorescent signals in N. benthamiana, demonstrating that MdMAPKKK1 was found to interact with MdMAPKK6 (Figure 4A). To further verify the interaction relationships among MdMAPKKK1 and MdMAPKK6, we performed an in vitro protein pull‐down assay. We then performed a pull‐down assay by incubating GST‐MdRLK2‐inside (the intracellular domain of MdRLK2) with His‐MdMAPKKK1 and GST‐MdMAPKK6 with His‐MdMAPKKK1 purified from Escherichia coli and used a His‐tag antibody to detect the pulled‐down proteins. Interactions were confirmed between MdMAPKKK1 and MdMAPKK6, consistent with the results of the fluorescence assay (Figure 4B,C). These findings suggest that the identified target genes operate cooperatively rather than independently, likely interacting to mediate the signal transduction cascades that drive the plant's defence response.

Figure 4.

Interaction mapping reveals a signalling hierarchy among MdRLK2, MdCRK10, MdMAPKKK1 and MdMAPKK6. (A) Bimolecular fluorescence complementation (BiFC) assays in N. benthamiana leaves. Strong YFP fluorescence signals were observed for the pairs MdMAPKKK18/MdMAPKK6, MdMAPKKK1/MdMAPKKK18 and MdMAPKKK1/MdMAPKKK6, indicating direct interactions in vivo. No signals were detected for MdMAPKKK18/MdRLK2 or MdMAPKKK1/MdRLK2 negative controls. Scale bars = 50 μm. (B, C) GST pull‐down assays confirming direct physical associations. Recombinant His‐tagged protein (MdMAPKKK1) was incubated with GST‐tagged baits (MdRLK2‐inside, MdMAPKK6) or GST controls. Immunoblots probed with anti‐His and anti‐GST antibodies show specific interactions between: (B) MdMAPKK6 and MdMAPKKK1. Consistent with BiFC results, no direct interaction was detected between MdRLK2‐inside and MdMAPKKK1 (C). (D, E) GST pull‐down assays. Purified His‐tagged MdCRK10 was incubated with GST‐tagged MdRLK2‐inside (D) or GST‐MdMAPKKK1 (E). Immunoblots probed with anti‐His and anti‐GST antibodies confirm that MdCRK10 specifically interacts with MdRLK2 and MdMAPKKK1. (F) BiFC assays visualising protein interactions in N. benthamiana leaf epidermal cells. Reconstituted YFP fluorescence indicates direct interaction for the MdRLK2/MdCRK10 and MdCRK10/MdMAPKKK1 pairs. No fluorescence was detected for the empty vector negative controls. Scale bars = 50 μm. [Color figure can be viewed at wileyonlinelibrary.com]

To elucidate the regulatory network mediating apple defence against A. alternata, we investigated the physical interactions between identified resistance candidates. Using the intracellular domain of MdRLK2 (MdRLK2‐inside) as bait for pull‐down assays combined with LC‐MS/MS analysis, we identified MdCRK10 (a cysteine‐rich RLK) as a primary interacting partner of MdRLK2 (Figure S8). Subsequent protein‐protein interaction assays revealed that MdCRK10 also interacts with MdMAPKKK1 (Figure 4D,E). Further BiFC assays in Nicotiana benthamiana leaves confirmed that MdCRK10 physically associates with MdRLK2‐inside at the cell periphery (Figure 4F). This suggests that MdCRK10 may serve as a critical bridge between the RLK MdRLK2 and the downstream MAPK signalling components. Collectively, these findings demonstrate that MdRLK2, MdCRK10, MdMAPKKK1 and MdMAPKK6 form a robust protein complex, suggesting they operate cooperatively within a signalling module to mediate apple defence responses against ALT7.

2.5. MdRLK2 Phosphorylates MdCRK10 to Activate MAPK Signalling

Given that MdRLK2, MdCRK10, MdMAPKKK1 and MdMAPKK6 are all putative kinases, we hypothesised that their functional coordination is governed by phosphorylation‐dependent interactions. To test whether MdCRK10 is a direct substrate of MdRLK2, we performed in vitro kinase assays coupled with Phos‐tag SDS‐PAGE. Co‐incubation of recombinant His‐MdCRK10 with the kinase domain of MdRLK2 (GST‐MdRLK2‐inside) resulted in a distinct, slower‐migrating band, confirming that MdRLK2 directly phosphorylates MdCRK10 in vitro (Figure 5A). These results demonstrate that MdRLK2 directly phosphorylates MdCRK10. Furthermore, to identify the potential phosphorylation sites targeted by MdRLK2, we performed an in‐silico prediction using the NetPhos 3.1 server (https://services.healthtech.dtu.dk/services/NetPhos-3.1/) (Figure S9A). A total of seven Serine/Threonine residues in MdCRK10 were predicted as high‐confidence phosphorylation sites with a score > 0.90 (Figure S9A; Table S1). Based on the known preference of similar kinases for acidic motifs, we specifically screened these candidates for the presence of an acidic residue (Aspartate, D, or Glutamate, E) located downstream of the phospho‐acceptor site. This analysis highlighted several candidates that perfectly matched the S/T‐D/E or S/T‐x‐D/E motif (Figure S9A). Notably, Serine 144 and Serine 363 emerged as the top candidates for experimental validation due to their high prediction scores and ideal sequence contexts (Table S1). To validate these predictions, we generated a mutant version of MdCRK10^S144A, MdCRK10^S363A and MdCRK10^S144A,S363A (Figure S9B–D). The subsequent in vitro kinase assay revealed that the phosphorylation‐induced mobility shift was completely abolished in the MdCRK10^S144A, MdCRK10^S363A and MdCRK10^S144A,S363A (Figure 5B–D). This result indicates that S144 and S363 are essential sites on MdCRK10 that are phosphorylated by MdRLK2.

Figure 5.

MdRLK2 interacts with and phosphorylates MdCRK10 to activate a downstream MAPK cascade. (A–D) In vitro kinase assays determining the phosphorylation of MdCRK10 by MdRLK2. Recombinant GST‐MdRLK2‐inside was incubated with His‐tagged MdCRK10 variants. Immunoblotting with anti‐His antibody reveals phosphorylation (mobility shift) of wild‐type MdCRK10 (A). This phosphorylation is compromised in the single mutants S144A (B), S363A (C) and double mutants (D). (E, F) Reconstruction of the downstream phosphorylation cascade. In vitro assays confirm that phosphorylated MdCRK10 phosphorylates GST‐MdMAPKKK1 (E); activated MdMAPKKK1 phosphorylates His‐MdMAPKK6 (F).

Given that MdRLK2 activates MdCRK10, we next investigated the downstream signalling components targeted by MdCRK10. As many CRKs function upstream of MAPK cascades, we hypothesised that MdCRK10 might directly phosphorylate a MAPKKK. We tested this by performing an in vitro kinase assay with MdCRK10 and a candidate, MdMAPKKK1. The results showed that active MdCRK10 induced a significant mobility shift of MdMAPKKK1 on a Phos‐tag gel (Figure 5E). This result suggests that MdCRK10 directly phosphorylates MdMAPKKK1, identifying MdCRK10 as a direct link between the receptor complex and the MAPK module. To determine if MdMAPKKK1 activation leads to a full cascade, we examined its ability to phosphorylate its canonical substrate, a MAPKK. An in vitro assay using active MdMAPKKK1 and MdMAPKK6 was conducted. Indeed, MdMAPKKK1 robustly phosphorylated MdMAPKK6, resulting in a clear band shift (Figure 5F). Integrated with our previous findings, these results propose a potential phosphorylation axis that mediates apple defence responses. Furthermore, we explored the broader network by testing if other MAPKKKs could also target MdMAPKK6. Collectively, all the above results provide convincing evidence that MdRLK2 phosphorylates MdCRK10, which subsequently activates the MAPK signalling pathway.

2.6. The MdRLK2‐MdCRK10 Complex Mediates Apple Defence Responses by Activating the MAPK Signalling Pathway

To evaluate the biological function of the MdRLK2‐MdCRK10 complex in apple defence, we performed transient transformation assays in the susceptible cultivar GL‐3. Overexpression of MdRLK2, MdCRK10 or their combination (OE‐MdRLK2/MdCRK10) significantly enhanced, correlated with the significant upregulation of downstream MAPK components (MdMAPKKK1 and MdMAPKK6) and defence‐related genes (MdPR10‐2 and MdWRKY53) (Figure 6A,B and Figure S10). This enhanced resistance to ALT7, characterised by restricted lesion expansion and reduced disease incidence compared to WA (Control) and empty vector (pFGC5941) controls (Figure 6C,D). These results clearly indicate that MdRLK2 and MdCRK10 function as positive regulators of the apple defence response against ALT7, and that their functional interaction within the MdRLK2‐MdCRK10 complex is critical for mediating effective resistance in this species.

Figure 6.

MdRLK2 and MdCRK10 positively regulate the activation of the downstream MAPK cascade and resistance to ALT7. (A, B) Transcriptional analysis of signalling pathway components. Leaves were infiltrated with control (uninfiltrated), pFGC5941 (empty vector), OE‐MdRLK2, OE‐MdCRK10 or OE‐MdRLK2+OE‐MdCRK10. qRT‐PCR analysis the levels of the downstream kinases MdMAPKKK1 (A) and MdMAPKK6 (B) were significantly enhanced in the overexpression lines after 48 h post‐inoculation with ALT7. (C, D) Functional assessment of disease resistance in ‘Gala‐3’ (GL‐3) leaves. (C) Representative disease symptoms on pre‐infiltrated GL‐3 leaves at 48 h post‐inoculation (hpi) with ALT7. (D) Quantification of lesion areas. Enhanced disease resistance was observed in OE‐MdRLK2, OE‐MdCRK10 and OE‐MdRLK2+OE‐MdCRK10 leaves compared to controls. (E, F) Transcriptional analysis of signalling pathway components. Leaves were treated with Control, pFGC5941, RNAi‐MdRLK2, RNAi‐MdCRK10 or combinations of OE and RNAi vectors as indicated. qRT‐PCR confirmed that the transcript levels of the downstream kinases MdMAPKKK1 (E) and MdMAPKK6 (F) were significantly reduced in the RNAi‐treated leaves after 48 h post‐inoculation with ALT7. (G, H) Functional assessment of disease resistance in ‘Hanfu’ (HF) leaves. (G) Representative disease symptoms on pre‐infiltrated HF leaves at 48 h post‐inoculation with ALT7. (H) Quantification of lesion areas. Compromised disease resistance was observed in leaves treated with RNAi‐MdRLK2, RNAi‐MdCRK10, OE‐MdRLK2+RNAi‐MdCRK10, RNAi‐MdRLK2+OE‐MdCRK10 or RNAi‐MdRLK2+RNAi‐MdCRK10. Experimental conditions: Control: uninfiltrated; pFGC5941: empty vector. Statistics: Data represent mean ± SD (n = 3 biological replicates for qPCR; n = 10 for lesion area). Asterisks indicate significant differences compared to controls (*p < 0.05, **p < 0.01, ANOVA with Dunnett's test). [Color figure can be viewed at wileyonlinelibrary.com]

Conversely, we employed RNA interference (RNAi) and co‐transformation in the resistant cultivar HF to further dissect their roles. qRT‐PCR analysis confirmed the expression levels of MdRLK2 and MdCRK10 in the treatment lines, revealing significant downregulation of both genes in the RNAi lines; conversely, in the combination lines, the overexpressed gene showed significant upregulation while the RNAi‐targeted gene showed significant downregulation, relative to the control (Figure S11A,B). In addition, we found that the expression of downstream signalling and defence genes was markedly suppressed in all RNAi‐treated lines (Figure 6E,F and Figure S11C,D). And silencing of MdRLK2, MdCRK10, or both significantly compromised the inherent resistance of HF to ALT7 and the mutant strain KO‐AamilR292#1, leading to severe infection symptoms (Figure 6G,H and Figure S12). Notably, overexpression of one component could not rescue the susceptibility caused by the silencing of the other, indicating that both partners are indispensable (Figure 6G,H and Figure S12). These results demonstrate that the MdRLK2‐MdCRK10 complex is essential for orchestrating effective immune responses against ALT7.

To further confirm that the phosphorylation sites act as essential functional determinants for the MdRLK2‐MdCRK10 complex to initiate downstream defence signalling, we introduced site‐directed mutations at each of the three conserved phosphorylation residues in MdCRK10 individually, and subsequently examined the activation status of the MAPK pathway in these mutant lines. We performed complementation assays in HF leaves by co‐transforming RNAi‐MdCRK10 with overexpression constructs of phosphorylation‐deficient mutants (OE‐MdCRK10^S144A, OE‐MdCRK10^S363A and the double mutant OE‐MdCRK10^S144A,S363A). First, we verified the expression levels of MdCRK10 across the different treatments. MdCRK10 transcripts were significantly reduced in the RNAi‐MdCRK10 line, whereas they were significantly elevated in the various overexpression combinations (Figure S13A). Interestingly, we observed a divergent expression pattern for the upstream receptor MdRLK2: it was significantly upregulated following KO‐AamilR292#1 inoculation but downregulated after ALT7 challenge (Figure 7A). Despite these variations in MdRLK2 levels, the activation of the downstream MAPK cascade (MdMAPKKK1 and MdMAPKK6) and resistance marker genes (MdPR10‐2 and MdWRKY53) was significantly abolished in all mutant lines regardless of the inoculum used (Figure 7B,C; Figure S13B,C). Subsequently, inoculation assays further revealed that leaves harbouring any of the three mutant constructs displayed a statistically significant increase in disease incidence and developed severe pathological symptoms characteristic of ALT7 infection (Figure 7D,E). In contrast, the Controls and pFGC5941 control groups maintained robust resistance to ALT7 or KO‐AamilR292#1, whereas the RNAi‐MdCRK10 group (with pFGC5941) showed clear susceptibility (Figure 7D,E). These findings provide compelling evidence that the phosphorylation of MdCRK10 at the conserved S144 and S363 residues is a critical prerequisite for the MdRLK2‐MdCRK10 complex to activate downstream defence responses and confer resistance against ALT7 in HF (Figure 7D,E). Collectively, these data unequivocally demonstrate that the phosphorylation of MdCRK10 at Ser144 and Ser363 is essential for the initiation and transduction of its downstream signalling. These findings provide compelling evidence that the MdRLK2‐MdCRK10 complex cascade constitutes a critical regulatory checkpoint that modulates the activation of apple immune defences in response to pathogen invasion. In summary, our findings provide evidence that A. alternata utilises EVs as a primary vehicle to deliver microRNA‐like RNAs (AamilR251/292/004) into apple cells, where they hijack host MdAGO1 to cleave transcripts of MdRLK2, MdMAPKKK1 and MdMAPKK6. These targets form a key signalling axis in which MdRLK2 phosphorylates MdCRK10 to activate a downstream MAPK cascade, and disruption of this module via phosphorylation‐deficient mutations or silencing renders apples highly susceptible. These findings reveal a cross‐kingdom mechanism where fungal milRNAs dismantle a core host immune network to facilitate infection.

Figure 7.

Phosphorylation of MdCRK10 at Ser144 and Ser363 is essential for activating the downstream MAPK cascade and conferring disease resistance. (A–C) Transcriptional analysis of signalling pathway components. Complementation assays were performed by infiltrating ‘Hanfu’ (HF) leaves with control (uninfiltrated), pFGC5941 (empty vector), or RNAi‐MdCRK10 combined with either pFGC5941 (negative control), OE‐MdCRK10S144A, OE‐MdCRK10S363A or OE‐MdCRK10S144A/S363A. qRT‐PCR was used to analyse gene expression in the treated leaves. Transcript levels of MdRLK2 (A) and the downstream kinases MdMAPKKK1 (B) and MdMAPKK6 (C) were significantly induced in the Control and pFGC5941 groups but were markedly suppressed in RNAi‐MdCRK10 leaves co‐expressing the phospho‐deficient MdCRK10 variants after 48 h post‐inoculation with A. alternata strain ALT7 or KO‐AamilR292#1. (D, E) Functional assessment of disease resistance. (D) Representative disease symptoms on pre‐infiltrated ‘Hanfu’ (HF) leaves at 48 h post‐inoculation with ALT7 or KO‐AamilR292#1. (E) Quantification of lesion areas. Failure to restore disease resistance was observed in RNAi‐MdCRK10 leaves co‐expressing the phospho‐deficient MdCRK10 variants compared to controls. Experimental conditions: Control: uninfiltrated and pFGC5941: empty vector. Statistics: Data represent mean ± SD (n = 3 biological replicates for qPCR; n = 10 for lesion area). Asterisks indicate significant differences compared to controls (*p < 0.05, **p < 0.01, ANOVA with Dunnett's test). [Color figure can be viewed at wileyonlinelibrary.com]

3. Discussion

The co‐evolutionary conflict between plants and pathogens is driven by the deployment of pathogen effectors and the evolution of host immune receptors (Cai et al. 2018; He et al. ²⁰²³). In this study, we uncover a potent offensive strategy where the fungal pathogen A. alternata utilises EVs as specialised delivery vehicles to transport AamilRNAs into apple cells. This EV‐mediated translocation allows these trans‐kingdom effectors to systematically dismantle host immunity from within. To achieve this translocation, the physical integrity of these milRNAs must be preserved within the hostile environment of the host apoplast, which is rich in nucleases (Cai et al. 2019; Davila and He ²⁰²⁵). Our RNase protection assay provides evidence that AamilR251, AamilR292 and AamilR004 are encapsulated within the Evs, which shield these fungal effectors from degradation during crosskingdom transformation. This previously unknown immune cascade, which we identified and characterised, is initiated by the receptor kinase MdRLK2 and transduced through a MdCRK10‐MAPK phosphorylation module. In the resistant cultivar HF, this cascade is activated to establish resistance against ALT7, where the phosphorylation of MdCRK10 serves as a pivotal node. Critically, when this phosphorylation is abolished (S144A, S363A mutation), the signal cannot be relayed downstream, leading to a compromised defence response and a loss of resistance in HF. By mechanistically linking a pathogen's RNAi‐based offensive to the silencing of a specific host kinase cascade, our work reveals a new paradigm in fungal pathogenesis (Figure 8). This discovery not only exposes a critical vulnerability in the plant immune system but also highlights the evolution of RNA itself as a potent cross‐kingdom weapon.

Figure 8.

A proposed working model illustrating the interplay between fungal milRNA interference and host phosphorylation signalling determines apple resistance. The diagram presents the fungal pathogen as a unified source of milRNAs (top layer) targeting apple cells with contrasting genetic backgrounds (bottom panels). Top layer (fungal source): The pathogen provides a specific cocktail of milRNAs (AamilR251, AamilR292 and AamilR004) that enter the plant cell to interfere with the host immune network. Bottom Panels (Inside apple cell): (A) Wild‐type scenario (resistance): In wild‐type plants, intact phosphorylation triggers a robust signalling relay along the core axis (MdRLK2‐MdCRK10‐MdMAPKKK1‐MdMAPKK6). This high signal flux overcomes the interference from fungal milRNAs, inducing high expression of defence genes (MdPR10‐2 and MdWRKY53) and resulting in resistance. (B) MdCRK10 mutation scenario (susceptibility): Mutation of the critical phosphorylation site in MdCRK10 blocks the initial signal transmission to downstream. This primary blockade, combined with the multi‐point inhibition by fungal milRNAs on downstream nodes, results in a systemic collapse of the signalling network. Consequently, defence gene expression remains low, leading to susceptibility. Symbols: Solid arrows with (P) indicate activation via phosphorylation; T‐bars indicate inhibition by milRNAs; the bold red X indicates a blocked signalling step due to mutation. [Color figure can be viewed at wileyonlinelibrary.com]

Eukaryotic sRNAs play a key role in regulating diverse biological processes, including growth, development, metabolism, genome stability and immunity (Huang et al. 2019; Summanwar et al. ²⁰²⁰). Studies have shown that sRNAs can move between a host and interacting microbes through a mechanism known as trans‐kingdom RNAi to silence target genes in different species (Huang et al. 2019; Cai et al. ²⁰¹⁹). Fungal milRNAs represent a novel class of effectors capable of suppressing host immunity (Ji et al. 2021; Xu et al. ²⁰²²). Such milRNAs have been identified in several plant pathogenic fungi, including Botrytis cinerea, Verticillium dahliae and Fusarium graminearum, which inhibit plant immune responses by silencing host resistance genes (Weiberg et al. 2013; M. Wang et al. ²⁰¹⁷; Jin et al. ²⁰¹⁹; Ji et al. ²⁰²¹). In this study, we identified three AamilRNAs (AamilR251, AamilR292 and AamilR004) as potent virulence effectors (Figure 1). They promote disease susceptibility in apple by specifically targeting and cleaving MdMAPKKK1, MdRLK2 and MdMAPKK6, which are involved in the disease resistance process (Figure 2). Notably, these targets are key components of a MAPK signalling cascade, a central hub in plant immunity that is frequently targeted by pathogen effectors (Irieda et al. 2019; Lv et al. ²⁰²²). Maintaining miRNA homoeostasis through the tight regulation of their biosynthesis and degradation is a cornerstone of gene expression control in diverse developmental contexts (Borges and Martienssen 2015; Martinez and Köhler ²⁰¹⁷; Yu et al. ²⁰¹⁹).

As carriers of cross‐kingdom material exchange, EVs play a pivotal role in the trans‐kingdom transport of fungal virulence factors (Cui et al. 2019; Z. Wang et al. ²⁰²⁵). Previous research has established that EVs provide a membrane‐enclosed compartment that shields fungal milRNAs from degradation by nucleases within the host apoplastic space, thereby ensuring the successful delivery of functional effectors into host cells (He et al. 2023; Chen and Phillips ²⁰²⁴; Davila and He ²⁰²⁵). Once inside the host, these milRNAs hijack host AGO proteins to form the RISC, which mediates the cleavage of host target mRNAs and leads to disease susceptibility (Hua et al. 2018; Huang et al. ²⁰¹⁹; Han and Mendell ²⁰²³). Our findings align with these models, suggesting that A. alternata specifically packages a unique cocktail of milRNAs into EVs to prevent degradation, subsequently mediating the cleavage of corresponding target mRNAs and facilitating infection in apple. Actually, it has been demonstrated in mammals that certain RNA‐binding proteins (RBPs) are involved in the loading of sRNAs into EVs (Shurtleff et al. 2016; Santangelo et al. ²⁰¹⁶). Similarly, in the context of cross‐kingdom sRNA transport from Arabidopsis to B. cinerea via EVs, studies have established that the entry of sRNAs into EVs is not a stochastic process but rather involves selective loading mediated by RBPs (He et al. 2021). Therefore, we speculate that during the cross‐kingdom transport of milRNAs from fungi to plants via EVs, the loading of milRNAs into EVs may also be a selective rather than random process.

The inverse expression relationship between the AamilRNAs (AamilR251, AamilR292 and AamilR004) and their defence‐related target genes (MdMAPKKK1, MdRLK2 and MdMAPKK6) strongly suggests a negative regulatory role for these sRNAs in apple disease resistance. Specifically, the significant AamilRNAs accumulation in the susceptible cultivar GL‐3 coincided with the downregulation of target genes critical for defence signalling, whereas the resistant cultivar HF maintained low AamilRNAs levels and upregulated its defence targets during infection (Figure S14A–E). We acknowledge that the differential AamilRNAs abundance could be influenced by the pathogen load. Our quantification of the fungal ITS region showed that ALT7 colonisation in GL‐3 was approximately three times higher than that in HF (Figure S14F,G). This enhanced fungal growth in GL‐3 likely partially contributes to the overall accumulation of host sRNAs—a phenomenon previously observed in other pathosystems where higher pathogen loads can trigger greater host transcriptional responses. Despite the quantified difference in pathogen biomass, the magnitude and specificity of the target gene downregulation in GL‐3 strongly suggest an additional, active host‐mediated mechanism governing the stability or turnover rate of these AamilRNAs. A mere threefold difference in fungal biomass may be insufficient to fully account for the observed drastic and opposite expression patterns of the AamilRNAs and their targets. Therefore, we hypothesise that the susceptibility of susceptible cultivar GL‐3 involves a mechanism leading to the disproportionate accumulation of these key AamilRNAs during infection compared to the resistant cultivars HF. While the fungal biomass in the susceptible cultivar GL‐3 was approximately threefold higher than in the resistant cultivar HF, this difference alone appears insufficient to fully account for the drastic contrast in AamilRNAs levels observed. This suggests that beyond pathogen load, host‐specific factors likely regulate the effective concentration of these trans‐kingdom effectors, either by restricting their uptake or promoting their clearance. The precise molecular mechanisms underlying this differential accumulation remain an intriguing subject for future investigation.

The suppression of PTI is a central requirement for successful pathogen infection (Van Der Biezen and Jones ¹⁹⁹⁸; Dangl and Jones ²⁰⁰¹; Kourelis and van der Hoorn ²⁰¹⁸; Lolle et al. ²⁰²⁰; Y. Wang, Pruitt, et al. ²⁰²²; Du et al. ²⁰²⁵). Our results suggest that kinase‐mediated signalling modules constitute an important vulnerability within apple PTI, as multiple host kinases involved in receptor‐to‐MAPK signalling are selectively targeted by fungal AamilRNAs. Rather than disrupting individual signalling components in isolation, these AamilRNAs appear to interfere with the overall architecture of PTI signalling networks. Targeting a single node might allow for signal leakage or compensatory flux through alternative pathways. By simultaneously hitting the receptor (MdRLK2), the intermediate transducer (MdMAPKKK1) and the convergence hub (MdMAPKK6), the pathogen imposes a ‘systemic collapse’ of the defence circuit. Interestingly, the central transducer MdCRK10 is not targeted. This implies that by severing the connections upstream and downstream, the pathogen effectively isolates MdCRK10, rendering its targeting redundant. Together, these findings support the idea that cross‐kingdom RNAi can function as a means to modulate host immune network topology. Whether similar convergent targeting strategies operate in other plant–pathogen systems remains an important question for future studies.

In conclusion, our work provides evidence for an intricate molecular interaction between apple and A. alternata, suggesting a refined co‐evolutionary dynamic at the host‐pathogen interface. We have characterised a linear MdRLK2‐MdCRK10‐MAPK signalling axis that appears to play a significant role in apple's defence response. Apart from the post‐transcriptional silencing effects, our data further support a model in which the pathogen employs a ‘convergent targeting’ strategy, utilising a suite of milRNAs that are enriched in EVs to potentially modulate multiple nodes within this specific defensive circuit. These findings contribute to our understanding of cross‐kingdom RNAi, indicating a shift from general immune suppression toward a more targeted interference with host signalling architectures. On a practical level, this study identifies several candidate genes (MdRLK2, MdCRK10, MdMAPKKK1 and MdMAPKK6) that may serve as valuable targets for the genetic improvement of disease resistance in apple. Collectively, these results reveal a mechanism by which cross‐kingdom RNA signalling facilitates fungal pathogenesis by disrupting the coordination of host immune components.

4. Methods

4.1. Biological Materials and Growth Conditions

Tissue culture plants of apple cultivars (Malus domestica cv ‘Gala‐3’ [GL‐3], Malus domestica cv ‘Hanfu’ [HF]) were grown on Murashige and Skoog (MS) medium containing 0.3 mg L⁻¹ 6‐benzylaminopurine, 0.2 mg L⁻¹ indoleacetic acid and 0.1 mg L⁻¹ gibberellin₃ (GA₃). The plants were incubated in a climate‐controlled chamber at a temperature of 26°C ± 1°C with 16 h of light and 8 h of darkness. The photon flux density was kept at approximately 100 μmol m⁻² s⁻¹ using Philips Lighting TL5 28 W/865. Every 4 weeks, the plants were supplied with fresh medium (Bai et al. 2011; Q. Zhang et al. ²⁰¹⁸).

The A. alternata f. sp mali WA strain ALT7 was stored at the Laboratory of Fruit Cell and Molecular Breeding, China Agricultural University, Beijing, China. To facilitate transgenic experiments, A. alternata f. sp mali strain served as the host strain and as a WT control in all experiments. Both WT and mutant strains were cultured on complete media, Potato Dextrose Broth with 1.5% agar (PDA) medium at 28°C. E. coli strain DH5α and Rosetta (DE3) were grown on Lysogeny Broth medium at 37°C. The A. tumefaciens strains GV3101 and EHA105 were grown on Yeast Extract Peptone (YEP) medium at 28°C, respectively.

4.2. EV Isolation

The isolation of EVs from ALT7 was conducted following established protocols (Albuquerque et al. 2008; Reis et al. ²⁰¹⁹). Fungal cells were initially centrifuged at 15 000 rpm for 30 min to remove cell debris. The supernatant was then filtered through a 0.45 µm polyvinylidene difluoride membrane (Millipore, Billerica, MA). Concentration of the filtrate was achieved using a 100 kDa cutoff filter in an Amicon ultrafiltration system (Millipore), resulting in approximately 50‐fold concentration. The concentrated solution was subjected to ultracentrifugation at 60 000 rpm for 1 h at 4°C. This step was followed by two washes with phosphate‐buffered saline (PBS) and subsequent storage at ‐80°C.

EV isolation from infected plants was performed as previously described (He et al. 2021, ²⁰²³). Plant EVs were isolated from the GL‐3 apoplastic wash. Plant leaves (for GL‐3, 4‐week‐old plants) were harvested and vacuumed with infiltration buffer (20 mM MES hydrate, 2 mM CaCl2, 0.1 M NaCl, pH 6.0) and centrifuged for 10 min at 900 g to collect the apoplastic fluids. The cellular debris in apoplastic fluids was removed by centrifugation at 2000 g for 30 min, followed by filtration through a 0.45‐μm filter. Next, the apoplastic wash was further purified by centrifugation for 30 min at 10 000 g, the supernatants were then transferred to new ultracentrifuge tubes and centrifuged for 1 h at 100 000 g to obtain the P100 pellet or 40 000 g to obtain the P40 pellet. The supernatants of P40 were centrifuged for 1 h at 100 000 g to obtain the P100‐40 pellet. The pelleted material was washed with filtered infiltration buffer at 100 000 g for 1 h to collect the pellet.

4.3. TEM

GL‐3 apple leaves were collected at 12, 24, 48 and 72hpi with ALT7. Leaf segments (1 mm³) were excised and immediately immersed in 2.5% glutaraldehyde (room temperature) for 1 h, followed by fixation at 4°C for 12–24 h. Samples were then washed three times with 0.1 M PBS (pH 7.2), 15 min each time. Post‐fixation was carried out using 1% osmium tetroxide (Ted Pella Inc., 18456) for 1–2 h, followed by three washes with 0.1 M PBS and three washes with distilled water, each for 10 min. Samples were dehydrated in a graded ethanol series (30%, 50%, 70%, 80%, 95% and 100%), with each step lasting 15 min. This was followed by three 15‐min incubations in 100% propylene oxide. Infiltration was performed using mixtures of propylene oxide and embedding resin (Epoxy) at ratios of 1:1 for 2 h at room temperature, then 1:2 overnight. Finally, samples were incubated in 100% embedding resin for 4 h, repeated twice. After embedding in moulds, polymerisation was carried out sequentially at 37°C for 48 h, 45°C for 12 h and 60°C for 48 h. Ultrathin sections were prepared using an ultramicrotome (Leica UC7), stained with uranyl acetate for 25 min and lead citrate for 7 min. After washing and drying, sections were examined using a transmission electron microscope (Hitachi HT‐7800).

ALT7 EVs were visualised using negative staining TEM. The EVs were initially washed twice with PBS and fixed for 1 h in 2.5% glutaraldehyde in 0.1 M phosphate buffer at room temperature. Subsequently, 20 μL of EVs were placed on a 150‐mesh copper grid with carbon film (WFHM‐150, Servicebio, Wuhan, China) for 3–5 min. Excess sample was removed, and the grid was stained with 2% phosphotungstic acid for 1–2 min. After air‐drying at room temperature, the samples were examined using a HITACHI HT7800 transmission electron microscope operated at 120 kV.

4.4. Nanoparticle Tracking Analysis

A ZetaView fitted with 520 nm with ZetaView software version 8.05.14 SP7 (Particle Metrix) was used to determine the size distribution and concentration of the ALT7 EVs. EV samples were diluted 50 times using 0.22 μm filtered PBS and injected into the flow cell at a flow rate of 50 units. 60 s videos were taken for each sample and used to determine the size distribution and concentration of the EVs.

4.5. sRNA Library Construction and Next‐Generation Sequencing

sRNAs were extracted from EVs isolated from 4‐week‐old GL‐3 leaves, 5‐day‐old ALT7 cultures and GL‐3 leaves inoculated with ALT7 for 24 h. The sRNA libraries for Next Generation sequencing were constructed by the 5′‐phosphate dependent method as described, and the libraries were sequenced using the Illumina Novaseq. 6000 (Q. Zhang et al. ²⁰¹⁸). The bioinformatics pipeline began with mapping all sRNA (ALT7 and ALT7‐GL‐3) reads to the ALT7 reference genome (GenBank: GCF_001642055.1) via SOAP (http://soap.genomics.org.cn/soap1/). The bioinformatics pipeline began with mapping all sRNA (ALT7‐GL‐3 and GL‐3) reads to the apple reference genome (GenBank: GCF_042453785.1) via SOAP (http://soap.genomics.org.cn/soap1/). To enrich for non‐coding sRNAs, the resulting alignments were subsequently filtered against EST and protein databases to remove any sequences corresponding to protein‐coding regions. The potential of the filtered, non‐coding sRNAs to form stable hairpin secondary structures‐a hallmark of miRNA precursors‐was then assessed using Unafold (http://unafold.rna.albany.edu/). Candidate miRNAs were selected based on the following criteria: (i) a maximum of two nucleotide mismatches when compared to known miRNAs; (ii) confirmation that the sequence does not encode a protein; (iii) localisation of the mature miRNA on one arm of a hairpin precursor, with the miRNA* on the opposing arm; (iv) a thermodynamically stable precursor secondary structure, indicated by a minimum free energy (MFE) of less than ‐20 kcal/mol; and (v) an (A + U) content ranging from 30% to 70%. Finally, sequences from the GL‐3‐EVs, ALT7‐EVs and ALT7‐GL‐3‐EVs datasets that met these criteria were aligned to the miRBase (https://www.mirbase.org/, Release 22.1, 2019) database to identify conserved miRNAs. Novel miRNAs were concurrently identified from the unannotated sRNA pool using MIREAP (Version 0.20), which predicts new candidates based on their potential to fold into canonical miRNA precursor hairpin structures.

4.6. Total RNA Extraction and Northern Blot Analysis of sRNAs

Total RNA was isolated from apple leaves and EVs using a modified cetyltrimethylammonium bromide (CTAB) method (Li, Li, Han, Shu, and Li 2009). To eliminate DNA contamination, the isolated RNA was treated with DNase I (Aidlab, Beijing, China). RNA integrity was assessed via electrophoresis on a 1.2% agar gel, and its concentration was quantified using an ND‐1000 NanoDrop spectrophotometer (Thermo Fisher Scientific) (Q. Zhang et al. ²⁰¹⁸). To determine the presence of milRNAs within EVs, the extracted EVs were first incubated with Proteinase K (200 μg/mL) (Solarbio, Beijing, China) at 55°C for 30 min. Subsequently, one portion of the EVs was treated with 1% Triton X‐100 (Solarbio, Beijing, China) on ice for 30 min (He et al. 2023). The RNA extracted from EVs lysed with 1% Triton X‐100 was then digested with RNase A (100 μg/mL) at 37°C for 30 min. For the other portion, EVs were first digested with RNase A (100 μg/mL, 37°C, 30 min), and then incubated with 1% Triton X‐100. Following digestion, Northern gel blot analysis was performed to assess the integrity of the RNA.

4.7. EMSA

The 5′ biotin‐labelled AamilR251, AamilR292 and AamilR004 DNA sequences were annealed in 10× annealing buffer (100 mM Tris‐HCl [pH 7.5], 10 mM EDTA and 1 M NaCl) at 75°C for 30 min and stored at −20°C. The EMSAs were performed using a LightShift chemiluminescent EMSA kit (Thermo Fisher Scientific; 20148) according to the manufacturer's protocol (Q. Zhang et al. ²⁰¹⁸). Primers are listed in Table S2.

4.8. Protein Purification

MdRLK2‐inside, MdMAPKK6 and MdMAPKKK1 were inserted into the pGEX‐4T‐1 vector, MdCRK10 and MdMAPKKK1 were inserted into the pET28a vector and then transformed into Transetta (DE3) chemically competent E. coli cells, whose growth was then induced at 16°C. After 16 h, the cells were suspended in suspension buffer. Samples of the E. coli suspension culture were pulverised with ultrasound for 20 min and then pelleted by centrifugation (12 000 g, 1 h, 4°C). The collected supernatant was purified with GST or His resin (Life Technologies, G2879) at 4°C for 1 h. The supernatant was then discarded, and the remaining beads were rinsed with glutathione buffer three times before being used for further experiments. Primers are listed in Table S3.

4.9. GST‐Pull Down Assay

The GST pull‐down assay was performed as described previously (Si et al. 2021; N. Wang, Liu, et al. ²⁰²²). Briefly, GST‐MdRLK2‐inside+MdMAPKKK1, GST‐MdMAPKK6+His‐MdMAPKKK1, GST‐MdRLK2‐inside+His‐MdCRK10 and GST‐MdMAPKKK1+ His‐MdCRK10 fusion proteins were used for the pull‐down assay. Recombinant proteins were expressed in Rosetta (DE3) and purified using GST‐Sefinose (TM) Resin 4FF (Settled Resin) (Sangon Biotech, Shanghai, China) or His Ni Sepharose 6 Fast Flow resin (Cytiva). GST, GST‐MdRLK2‐inside+MdMAPKKK1, GST‐MdMAPKK6+His‐MdMAPKKK1, GST‐MdRLK2‐inside+His‐MdCRK10 and GST‐MdMAPKKK1+His‐MdCRK10, were mixed and incubated at 4°C for 4 h. Eluted proteins were subjected to immunoblot analyses using anti‐GST and anti‐His antibodies (Yeasen, Shanghai, China).

4.10. Phos‐Tag SDS‐PAGE

In vitro phosphorylation assays were performed by incubating pairs of recombinant proteins (0.4 mg/mL each), including MdRLK2‐inside+MdCRK10, MdRLK2‐inside+MdCRK10^S144A, MdRLK2‐inside+MdCRK10^S363A, MdRLK2‐inside+MdCRK10^S144A,S363A, MdCRK10+MdMAPKKK1 and MdMAPKKK1+MdMAPKK6. The reactions were conducted at 25°C for 30 min in a reaction buffer containing 0.30 M Tris‐HCl (pH 8.0), 50 mM KCl, 10 mM MgCl₂ and 1–40 mM ATP. Reactions were terminated by adding 0.5 volumes of 3× SDS‐PAGE loading buffer (195 mM Tris‐HCl (pH 6.8), 3.0% (w/v) SDS, 30% (v/v) glycerol, 15% (v/v) 2‐mercaptoethanol). Notably, samples were not boiled prior to electrophoresis (Kinoshita et al. 2006). The resulting protein mixtures were then separated by Phos‐tag SDS‐PAGE. Gels were prepared using Phos‐tag Acrylamide (AAL‐107, FUJIFILM Wako Pure Chemical Corporation, Japan) according to the manufacturer's instructions.

4.11. Total RNA Extraction and Northern Blot Analysis of sRNAs

Total RNA was isolated from apple leaves using a modified CTAB method (M. F. Li et al. ²⁰⁰⁹) and treated with DNase I (Aidlab, Beijing, China) to remove DNA contamination. RNA integrity was verified by electrophoresis on a 1.2% agar gel, and the RNA concentration was measured using an ND‐1000 NanoDrop spectrophotometer (Thermo Fisher Scientific) (Q. Zhang et al. ²⁰¹⁸). To identify whether milRNAs located in vesicles, the EVs underwent treatment with 1% TritonX‐100 (Solarbio, Beijing, China) on ice for 30 min, and then RNA gel blot analysis was performed.

Northern blot analysis was performed as previously reported with a Digoxin Hybridisation Detection Kit following the manufacturer's instructions (Mylab; DIGD‐120). Approximately 60 μg of RNA was separated in a 15% polyacrylamide gel containing 7 M urea and electrically transferred to Hybond‐N+ membranes (GE Healthcare) (Rio 2014; Q. Zhang et al. ²⁰¹⁸). The 5′ end‐modified digoxin‐labelled AamilR292, AamilR004, AamilR164, 18S rRNA and U6 probes were synthesised by Sangon Biotech, Shanghai, Beijing. Primers are listed in Table S2.

4.12. In Situ Hybridisation Analyses

Following inoculation, leaves were sliced into 1 cm² pieces and placed in 3.7% FAA (a solution containing 3.7% formaldehyde, 5% acetic acid and 50% ethanol) at 4°C to undergo overnight vacuum infiltration fixation. After fixation, embedding, hybridisation and assays were carried out using the methods described by Nie et al. (2021) and S. Wang et al. (2023). The embedded samples were then transversely sectioned into 8 μm sections using a paraffin slicer (Leica RM2125 RTS) (S. Wang et al. ²⁰²³). The probes AamilR251, AamilR292, AamilR004, 18S rRNA and negative) were synthesised using digoxigenin labelling by the Sangon Biotech, Shanghai. Detection of digoxin‐labelled RNA was achieved through an alkaline phosphatase luminescent substrate combination (Roche anti‐digoxin‐AP, NBT/BCIP). Images were captured using an Olympus CX23 microscope (Olympus, Japan). Primer sequences can be found in Table S2.

4.13. Fungal Transformation

The ALT7 gene was invalidated using a homologous recombination‐based technique, in accordance with previously described methods. To construct the knockout plasmids, the 5’ and 3’ regions of AamilR251, AamilR292 and AamilR004 were amplified and inserted into pCX62 plasmids (using HindIII/EcoRI and BamhI/XbaI). The entire length sequences of AamilR251, AamilR292 and AamilR004 were inserted into the pCX62 plasmid (HindIII/EcoRI) to construct the overexpression plasmids. Primer sequences can be found in Table S3.

A. tumefaciens‐mediated transformation was used as previously described with some modifications (Mullins et al. 2001; Gao et al. ²⁰¹⁰). EHA105 (containing appropriate vectors) was grown at 28°C for 2 days in YEP Medium supplemented with Ampicillin (100 mg/mL) (Sangon Biotech, Shanghai, China). For more information on subsequent co‐culture screening methods, please refer to Gao et al. (2010). Individual transformants were transferred into PDA medium containing 75 mg/mL hygromycin B (Sangon Biotech, Shanghai, China) and incubated until conidiogenesis. Conidia from each individual transformant were inoculated onto PDA plate medium and incubated at room temperature for 5 days to allow for further identification. Concurrently, these singular conidial cultures were also inoculated onto PDA medium, stored in cryopreservation tubes and refrigerated at 4°C.

4.14. Pathogen Assay

The fungus was cultured on PDA plates for 5 days at room temperature before spores were collected and diluted into sterilised water for a final concentration of 10⁵ spores per mL. Four‐week‐old plants were then inoculated. Photos were taken 2 days after inoculation and analysed by ImageJ software to measure the lesion area (%: lesion area/whole area, per leaf).

4.15. RT‐qPCR Assay

Total RNA was isolated from apple leaves using a modified CTAB method (M. F. Li et al. ²⁰⁰⁹). cDNA was synthesised using the HiScript III 1st Strand cDNA Synthesis Kit (+gDNA wiper) (Vazyme, Jiangsu, China) with OligdT, according to the manufacturer's instructions. RT‐qPCR was performed using ChamQ SYBR qPCR Master Mix (High ROX Premixed) (Vazyme, Jiangsu, China) under the following cycling conditions: 40 cycles of 95°C for 10 s and 60°C for 30 s (Applied Biosystems 7500). Relative RNA levels were calculated using the 2^−ΔΔCt method and normalised using Actin as a reference (Livak and Schmittgen 2001). Three technical replicates were performed for each cDNA sample, and three biological repeats (separate biological material) were performed for each treatment. RT‐qPCR primer sequences can be found in Table S3.

4.16. Identification and Cloning of milRNAs Targeted Genes

A search for milRNA‐targeted genes was conducted by sequence alignment with the apple genome from http://plantgrn.noble.org/psRNATarget. Detailed annotation information about the targeted genes, including nucleotide sequences, chromosome locations and predicted protein domains, was obtained from the apple genome sequence https://iris.angers.inra.fr/gddh13/the-apple-genome-downloads.html and https://www.rosaceae.org databases (Daccord et al. 2017). Primer sequences can be found in Table S3.

4.17. 5′ RNA Ligase‐Mediated Rapid Amplification of cDNA Ends

Total RNA was isolated from the same amount of ALT7 infected apple leaves and directly ligated to the 5’ RNA adaptor without any modification. cDNA was synthesised using the HiScript III 1st Strand cDNA Synthesis Kit (+gDNA wiper) (Vazyme, Jiangsu, China) with milRNAs targeted genes specific reverse transcription primers, according to the manufacturer's instructions. The first round of PCR amplification was performed using the 5’ RACE outer primer and a gene‐specific primer. The product was used as a template for nested PCR with the 5’ RACE inner primer and the nested gene‐specific primer. The PCR products were inserted into pCE2 TA/Blunt‐Zero vector (5 min TA/Blunt‐Zero Cloning Kit) (Vazyme, Jiangsu, China), and 20 individual clones were selected for Sanger sequencing. Primer sequences can be found in Table S3.

4.18. Agrobacterium‐Mediated Transient Expression Assays

Agrobacterium‐mediated transient expression was performed using the pFGC5941 construct driven by the 35S promoter. In overexpression experiments, full‐length MdMAPKKK1, MdRLK2 and MdMAPKK6 sequences were cloned into the vector pFGC5941 (NcoI/BamHI) (GenBank, AY310901). The empty pFGC5941 vector was used as the control. Loss‐of‐function MdMAPKKK1, MdRLK2 and MdMAPKK6 constructs were produced by cloning their specific sequences and their partially specific reverse sequences into pFGC5941 (NcoI/SawI and XbaI/BamHI) for MdMAPKKK1, MdRLK2 and MdMAPKK6, to produce small interfering RNAs (siRNAs). The above vectors were transformed into GV3101 by the heat shock transformation method. The cloning primers are listed in Table S3. Leaves from 4‐week‐old HF seedlings (resistant variety) and 4‐week‐old GL‐3 seedlings (susceptible variety) were infiltrated by A. tumefaciens with a silencing or overexpression construct. After agroinfiltration, the infiltrated seedlings were transferred to a fresh MS culture medium for 4 days to avoid wilting of the apple plantlets during A. tumefaciens infiltration. After 4 days, it was inoculated with ALT7.

4.19. Bimolecular Fluorescence (BiFC) Assays

The coding sequences of MdCRK10, MdMAPKKK1, MdRLK2 and MdMAPKK6 were amplified and fused with the coding sequence of the N‐terminus of YFP (YFPn) or the C‐terminus of YFP (YFPc) and inserted into vectors. Various combinations of YFPn and YFPc were transiently expressed in apple leaves by A. tumefaciens infiltration (Bai et al. 2011). Four days later, YFP fluorescence in leaves was observed and imaged at wavelengths of 500–542 nm using a Zeiss LSM900 orthogonal confocal microscope. The cloning primers are listed in Table S3.

4.20. Statistical Analysis

Statistical analyses were performed using analysis of variance followed by Dunnett's multiple‐comparison test in GraphPad Prism (v8.4.3). The statistical tests and numbers, including sample sizes or biological replications, are described in the figure legends.

Supporting information

Supporting File:

Supporting Table S1:

Supporting Table S2:

Supporting Table S3:

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.