The compound infection strain QY21 of A. alternata f. sp. mali by AaCV1 and AaMV1, as well as the AaMV1 single infection strain QY21-C1 and virus-free strain QY21-C2 obtained through detoxification of QY21, are preserved in Research Laboratory of Plant Disease Epidemiology and Integrated Pest Management of Hebei Agricultural University.

Three strains, QY21, QY21-C1, and QY21-C2, were inoculated separately onto potato dextrose agar (PDA) medium lined with cellophane at the bottom. After 72 hours of culture at 25°C, mycelia were scraped using a sterile disposable mycelium scraper, placed in RNA- and DNA-free centrifuge tubes, and rapidly frozen in liquid nitrogen. milRNA was extracted using the miRcute milRNA isolation kit (Tiangen, Beijing, China). Three biological replicates were set up for each strain, and milRNA sequencing was entrusted to Shanghai Majorbio Pharmaceutical Technology Co., Ltd (Shanghai, China). Reference species is A. alternata SRC1lrK2f. Reference genome version is Altal1 (Acc No: GCA_001642055). Specific steps were conducted, including extracting total RNA from the samples, measuring the concentration and purity of the extracted RNA using Nanodrop2000 (Thermo, MA, USA), assessing RNA integrity via agarose gel electrophoresis, and determining the RQN value using Agilent5300. Small RNA libraries were constructed using the QIAseq miRNA Library Kit (96) (Qiagen, Germany) according to the manufacturer’s protocol. The requirements for cDNA library construction are: total RNA amount of 1 μg, concentration ≥ 50 ng/μL, RQN > 7, and OD₂₆₀/OD₂₈₀ between 1.8 and 2.2. Adapter sequences were ligated to the 3’ and 5’ ends, respectively. Under the action of reverse transcriptase, a single strand of cDNA was reverse-synthesized using the RNA with ligated adapters as a template and random primers. This was followed by second-strand synthesis to form a stable double-strand structure. PCR amplification (11-12 cycles) was performed using sequencing primers to enrich the library. Target fragments were recovered by gel electrophoresis (6% Novex TBE PAGE gel, 1.0 mm, 10 well). The enriched 18-32 nt small RNA fragments were sequenced on the NovaSeq X Plus platform (Illumina Inc., San Diego, CA, USA).

The sequencing data was aligned with the reference genome sequence Altal1 to identify contamination or confusion in the samples. Low-quality data was removed before data analysis (Jiang et al., 2011). Fastp (https://github.com/OpenGene/fastp) was used for quality filtering and adaptor trimming. For the raw reads obtained from small RNA sequencing, small RNA fragments <18 nts and >30 nts were discarded, reads with an undetermined base ratio >10% were removed, reads with 3’ and 5’ adapter contamination, no inserted sequences, and single-copy sequences were removed, and reads containing polyA were discarded to obtain clean reads. After data quality control, the quality-controlled data was again statistically analyzed and quality assessed, including base quality, error rate, and content distribution statistics. This resulted in high-quality quality-controlled data (clean data) to ensure the accuracy of subsequent analysis results.

The quality-controlled raw data was aligned with the reference genome to obtain valid data (mapped reads) for subsequent analysis. TopHat2 (http://tophat.cbcb.umd.edu/) (Grabherr et al., 2011) and HISAT2 (http://ccb.jhu.edu/software/hisat2/index.shtml) (Smith-Unna et al., 2016) software were used to align the quality-controlled raw data with the reference genome. The distribution of raw data with a mapping rate higher than 65% (Total Mapped Reads) on the reference sequence was analyzed. Based on the existing reference genome, software Cufflinks (http://cole-trapnelllab.github.io/cufflinks/) (Li and Godzik, 2006) and StringTie (http://ccb.jhu.edu/software/stringtie/) (Bray et al., 2016) were used to assemble and splice the mapped reads and compare them with known transcripts. The reads aligned to the reference genome Altal1 were compared with the miRbase (Version 22.1) and Rfam (Version 14.6) databases to obtain known milRNA annotation information. MiReap software was used to predict new milRNAs for reads without annotation information using the aforementioned reference genome. DESeq2 (Zhou et al., 2014) was used to perform differential expression analysis on all small RNA loci in the nine strains, normalizing the raw read counts of each small RNA in each sample to the total read counts in that sample. The fold change (FC) between virus-infected and uninfected samples was calculated. The Benjamini and Hochberg method was used to correct the P-values obtained from the Wald test multiple times to obtain adjusted P-values (Padjust). Statistically supported differentially expressed milRNAs were retrieved (padjust ≤0.05, up-regulated when log₂(FC) ≥0.6 or down-regulated when log₂(FC) ≤-0.6). The RNAfold web server (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi) was used to predict the secondary structure of milRNA precursors, with base pairing probabilities used for coloring. The online website predictprotein (https://predictprotein.org/) was used to predict the secondary structure of proteins. Protein tertiary structure analysis was performed according to SWISS-MODEL (https://swissmodel.expasy.org/interactive) was used for target gene prediction.

The fungal genomic DNA rapid extraction kit (Sangon, Shanghai, China) was used to extract DNA from A. alternata f. sp. mali strains. Total RNA was extracted from the three strains, QY21, QY21-C1, and QY21-C2, using Trizol reagent (TransGen, Beijing, China). The integrity of the total RNA was detected using 1.0% agarose gel electrophoresis, and the concentration and purity of the nucleic acids were determined using the Eppendorf Biospectrometer^® basic UV-Vis fluorospectrophotometer (Eppendorf, HH, Germany).

The RT-qPCR detection of key milRNAs was performed using 2×FastFire qPCR PreMix (Tiangen, Beijing, China) on a 7500 quantitative PCR system (ABI, CA, USA). The primers used were: pmilR13q-F: GCGCGTGTTTATCTTCCAGA, and pmilR13q-R: AGTGCAGGGTCCGAGGTATT. U6 (U6 small nuclear RNA) was used as the housekeeping gene for milRNAs, while the Tubulin gene served as the housekeeping gene for the target genes. The relative expression levels were calculated using the 2^-ΔΔCt method. The quantitative primers for the potential target genes were: pAaSLPq-F: ATGGTGACGGTGTGGAGGATG, pAaSLPq-R: CGCAATGTCGGGCTGGAAAC; pAaDUF431q-F: AGTTCCTGCTGTCTTCCGTACC, pAaDUF431q-R: TCTGTCCGCATAGATTTCTTCAATACC; pAaNrtBq-F: GTGCTACCGTCGTGTCCAATG, pAaNrtBq-R: TCGCAGGCTTGTCGTCCAG.

To further validate that Aa-MIL1R13 serves as the functional precursor encoding mature Aa-milR13 and confirm its stability across fungal generations, we conducted a detailed precursor verification workflow. First, based on the miRNA sequencing-derived precursor sequences of Aa-milR13, we designed strain-specific forward (pmilR13-F: 5’-TGTTTATCTTCAGACTGTTGTTGT-3’) and reverse (pmilR13-R: 5’-AGTGTCAATCCACCAACACTGA-3’) primers using Premier 5.0 software. These primers were synthesized by Sangon Biotech (Shanghai) to ensure high specificity. Total RNA was extracted from mycelia of three A. alternata strains (QY2, QY21-C1, QY21-C2) subcultured for 3 generations using a standardized protocol (method 2.5). cDNA synthesis was performed with HiScript II Enzyme Mix (Vazyme) following the kit’s instructions. PCR amplification was conducted using a 25 μL reaction system with thermal cycling conditions: pre-denaturation at 95°C for 3 min, 35 cycles of 94°C (30 s), 58°C (30 s), and 72°C (30 s), followed by a final extension at 72°C for 5 min. PCR products were analyzed via 1.5% agarose gel electrophoresis and Sanger sequencing.

To evaluate the phenotypic and pathogenic characteristics of Aa-milR13 knockout and overexpression strains, two sets of experiments were conducted: colony morphology observation and detached plant tissue inoculation assays.

To elucidate the impact of virus infection on differential milRNA expression in A. alternata, we constructed milRNA libraries for the AaCV1 and AaMV1 co-infected strain QY21, the AaMV1 singly infected strain QY21-C1, and the virus-free strain QY21-C2, and performed deep sequencing. The results are shown in Table 1 . The average raw reads for the three strain libraries were 10,341,053, 12,478,073, and 11,684,185, respectively. After removing contaminants and low-quality sequences, the average clean reads obtained for the libraries were 7,899,450, 8,910,066, and 8,049,814, respectively. The average error rates of the sequenced bases were all below 0.1%, and both Q20 and Q30 were above 95%, indicating good library construction quality and sequencing quality, and reliable data.

The reads of the reference genome were aligned with the miRBase and Rfam databases to obtain annotation information for known milRNAs. The miReap software was used to predict new milRNAs for reads without annotation information. Venn analysis of all milRNAs among samples was performed, and the results are shown in Figure 1A . Twenty-five milRNAs were expressed in all three strains, accounting for 11.25%. There were 36 specific milRNAs in QY21, accounting for 16.67%, 31 specific milRNAs in QY21-C1, accounting for 14.35%, and 104 specific milRNAs in QY21-C2, accounting for 48.15%. This suggests that virus infection can cause a decrease in the number of milRNAs. In addition, we analyzed the base preference of milRNAs in the three strains, and the results are shown in Figures 1B, C . Except for the first base of 23 nt milRNAs preferring G, milRNAs of other lengths preferred U. We also found certain preferences at other sites, such as a general lack of U at positions 2 to 4, and a preference for A at position 25.

Using DESeq2, we analyzed the differential expression of milRNA in strains with different virus types. The obtained P-values were corrected for multiple testing using the Benjamini and Hochberg method. The results, shown in Figure 2 and Table 2 , revealed a total of 14 significantly differentially expressed milRNA. The largest number of significantly differentially expressed milRNA was found between the virus-free strain QY21-C2 and the co-infected strain QY21. Among the two virus-taking strains QY21 and QY21-C1, and the virus-free strain QY21-C2, there were two milRNA, Aa-milR87 (KV441478_174387) and Aa-milR13 (KV441481_212713), that showed significant differences in expression between the virus-free strain and the two virus-taking strains, but not between the two virus-taking strains themselves. Quantitative analysis of these two milRNA, shown in Figure 3 , confirmed the sequencing results, with Aa-milR13 exhibiting a more significant difference in expression. Therefore, Aa-milR13 was selected for further investigation.

The stem-loop structure of Aa-milR13 was predicted using the RNAfold web server. The result, shown in Figure 4 , indicated that the mature Aa-milR13 is located on the 5’ arm of the stem-loop structure, suggesting a highly conserved position of Aa-milR13 on its precursor gene. The thermodynamic predictions showed that the free energy of the thermodynamic system was -23.32 kcal/mol, and the minimum free energy was -22.90 kcal/mol, with a small difference between the two and both in a relatively low energy state, indicating that the structure is thermodynamically stable. In addition, the frequency of the Minimum Free Energy (MFE) structure was 50.99%, and the ensemble dispersion was 1.01, indicating a relatively high frequency and a relatively low dispersion, which suggests that this structure dominates in the thermodynamic system and is relatively concentrated, further demonstrating the stability of the Aa-milR13 stem-loop structure.

To further validate the stable presence of the Aa-milR13 precursor in A. alternata trains, three independent isolates (QY2, QY21-C1, and QY21-C2) were subcultured for 3 generations. RT-PCR amplification using strain-specific primers targeting the predicted Aa-milR13 precursor sequence produced a single prominent band of approximately 57 bp in all three subcultured strains ( Figure 5 ), consistent with the predicted length of the Aa-milR13 precursor (57 bp). Sanger sequencing of the PCR products confirmed 100% sequence identity with the genomic sequence of the predicted Aa-milR13 precursor, with no detectable mutations or sequence divergence across generations. These results demonstrate that the Aa-milR13 precursor is stably maintained in A. alternata strains, confirming its role as the functional precursor encoding mature Aa-milR13.

To elucidate the effect of Aa-milR13 on A. alternata strains, an Aa-milR13 knockout vector and overexpression vector were constructed (as shown in Figures 6A–C , 7A–C ). Knockout and overexpression transformants were obtained via PEG-mediated protoplast transformation. Stable genetically knocked-out mutant ΔQYC2-milR13 and overexpression mutant OE-milR13 were selected through hygromycin resistance screening. PCR and RT-qPCR identification results are shown in Figures 6D, E , 7D, E .

For phenotypic analysis, QY21-C2, QY21, ΔQYC2-milR13, and OE-milR13 were placed at the center of PDA plates and cultured for 1–7 days. Colony diameters were measured daily, and the results are shown in Figure 8 . The mycelial growth rate of ΔQYC2-milR13 was slower, with a significantly smaller colony diameter compared to QY21-C2. Its colony color and diameter were closer to those of QY21. In contrast, the mycelial growth rate of OE-milR13 was accelerated, and its colony diameter showed no significant difference from that of QY21-C2.

To investigate the effect of Aa-milR13 on pathogenicity, QY21-C2, QY21, OE-milR13, and ΔQYC2-milR13 were cultivated under equivalent conditions (28°C in the dark for 5 days). Five-millimeter-diameter mycelial plugs were collected from each strain and inoculated onto detached crabapple leaves and apple fruits (cv. ‘Tianhong 2’). Lesion diameters were measured 14 days post-inoculation, and the results are shown in Figures 9 , 10 . OE-milR13 exhibited the strongest pathogenicity, with leaf and fruit lesions significantly larger than those of the virus-free strain QY21-0C2 and other strains (QY21-C2 > ΔQYC2-milR13> QY21). Lesions of ΔQYC2-milR13 were significantly smaller than those of QY21-C2, indicating a marked reduction in pathogenicity. The QY21 strain showed the smallest lesions. These results suggest that deletion of Aa-milR13 significantly reduces the pathogenicity of A. alternata strains, whereas overexpression of Aa-milR13 significantly enhances their pathogenicity.

To further investigate the role of Aa-milR13 in the pathogenicity of A. alternata, we predicted the target genes of Aa-milR13. The results are shown in Table 3 . Three potential intracellular target genes, AaSLP, DUF431, and NrtB, were predicted using the RNAhybrid online website. These genes correspond to a subtilisin-like protease (CC77DRAFT_1100266), a DUF431 domain-containing protein (CC77DRAFT_1022347), and a high-affinity nitrate transporter NrtB (CC77DRAFT_1056077), respectively. 0

To clarify the synergistic expression changes between these three genes and Aa-milR13 in the strain, RT-qPCR analysis was conducted. The results, shown in Figure 11 , indicate that the expression levels of the three genes significantly increased following Aa-milR13 knockout, whereas they were significantly downregulated in the overexpression strain compared to QY21-C2. This pattern suggests a significant negative correlation between Aa-milR13 and the expression levels of these target genes, implying that Aa-milR13 may directly target AaSLP, DUF431, and NrtB. Notably, the highest expression levels of the target genes were observed in the virus-infected strain, indicating that viral infection-induced downregulation of Aa-milR13 might be the primary driver of the upregulated expression of these three genes.

Upon reviewing the literature, the functions of the aforementioned potential target genes in Alternaria species have not been reported. We analyzed the cleavage mode and target regions of Aa-milR13 to its target genes. The results, shown in Figure 12 , indicate that the target regions of Aa-milR13 are all located in the CDS regions of the target genes and act through cleavage. The pairing between Aa-milR13 and the target genes involves Watson-Crick pairing of AU and CG, wobble pairing of GU, and various other pairing modes such as AA, AC, AG, GG, UU, and UC.

The primary structures of AaSLP, DUF431, and NrtB were analyzed using the Portparam online software. The results show that AaSLP has a molecular mass of 50 kD, a theoretical isoelectric point (pI) of 9.25, and a molecular formula of C2198H3482N642O656S20. It contains multiple amino acids, with a high content of alanine (Ala) at 8.7%. DUF431 has a molecular mass of 23 kD, a theoretical pI of 4.88, and a molecular formula of C₂₅₆₇H₃₉₂₀N₆₄₆O₆₇₆S₂₃. It has the highest leucine (Leu) content at 9.7%, an instability index of 38.9, and is classified as a stable protein. NrtB has a molecular mass of 55 kD, a theoretical pI of 8.89, and a molecular formula of C2567H3920N646O676S23. It has a high glycine (Gly) content at 10.4%, an instability index of 25.66, and is considered a stable protein.

The secondary structures of AaSLP, DUF431, and NrtB proteins were predicted using the predictprotein software. The results, shown in Figure 13 , indicate that all three are proteins with heterogeneous secondary structure compositions. AaSLP and DUF431 are dominated by random coils, accounting for more than 50%, while NrtB is dominated by α-helices, accounting for 56.19% ( Figure 13A ). Homologous three-dimensional structural modeling using the SWISS-MODEL online tool revealed that the three genes are mainly composed of α-helices and random coils, with a small amount of β-sheets, which is consistent with the secondary structure prediction results ( Figure 13B ).

The RNA silencing system serves as a primary signaling pathway for virus-host interactions. RNA interference (RNAi) is a conserved mechanism in eukaryotes that regulates gene expression through mRNA cleavage, transcriptional or translational repression (Chang et al., 2012). MicroRNAs (miRNAs) are one of the key regulatory factors of RNAi in eukaryotes (Ghildiyal and Zamore, 2009). When a virus infects a host, it can alter the levels of host miRNAs, thereby regulating the expression of numerous target genes and achieving widespread gene silencing effects. This, in turn, affects the physiological metabolism and various critical functions of the host, such as growth and reproduction, substance transport and metabolism, stress response and resistance, pathogenicity, and parasitism ability. The discovery of milRNA in fungi lags behind that in plants, with a relatively small number and low conservation. With the advancement of sequencing technology, an increasing number of fungal milRNA have been discovered and identified. Using sequencing technology, two milRNA and 42 novel milRNA have been found in Sclerotinia sclerotiorum (Padmanabhan et al., 2009), 19 milRNA have been identified in F. oxysporum (Ruiz-Ferrer and Voinnet, 2009), and 57 milRNA have been identified in Valsa mali (Park and Shin, 2015). In this study, 44 putative milRNAs (homologous to miRBase-annotated miRNAs) and 172 novel milRNAs were identified in A. alternata f. sp. mali strains, representing the first comprehensive characterization of milRNAs in this species. We analyzed the differentially expressed milRNA in strains with mycovirus and without mycovirus. The expression levels of Aa-milR87 and Aa-milR13 were significantly lower in the virus-taking strains QY21 and QY21-C1 than in the virus-free strain QY21-C2, with no significant difference between the virus-taking strains. This was verified by RT-qPCR, and Aa-milR13 exhibited a more significant difference. When Aa-milR13 was knocked out in the virus-free strain, the biological traits of the mutant were closer to those of the virus-taking strain, and its pathogenicity was also significantly reduced. This confirms that Aa-milR13 is likely involved in the process of virus-regulated hypovirulence in the host.

Fungal virus infection induces differential expression of functional milRNA in the host, which forms the miRISC silencing complex by binding to AGO proteins and targets key virulence genes, leading to host virulence decline (Kobayashi and Tomari, 2016; Iwakawa and Tomari, 2022). In plants and animals, different AGOs determine the type of bound sRNA based on their 5’ terminal nucleotide preference (Eggleston, 2009). Arabidopsis AGO2 and AGO4 mainly bind sRNA starting with 5’ adenine (A), while AGO1 primarily binds 5’ uracil (U) sRNA (Mi et al., 2008). In this study, Except for 23 nt milRNA with a preference for G as the first base, milRNA of other lengths preferred U. This suggests that among the AGOs, at least one prefers binding to 5’ uracil (U) and one prefers binding to 5’ guanine (G), but further research is needed to determine which AGO proteins play a major role.

In addition, unlike plant and animal miRNAs, Aa-milR13 employs a multi-compatible base-pairing mechanism with its potential fungal target genes (AaSLP, DUF431, and NrtB). Aa-milR13 primarily interacts with the coding sequences (CDS) of its targets, diverging from animal miRNAs (3’UTR-focused) or plant miRNAs (CDS-start-site-focused) (Bartel, 2004). This CDS-centric targeting allows direct disruption of fungal protein synthesis. Aa-milR13 accommodates Watson-Crick (AU/CG), G-U wobble, and non-canonical (AA/AC/AG/GG/UU/UC) pairs, relaxing the “seed-region rigidity” of canonical miRNAs (Axtell et al., 2011). Aa-milR13 drives efficient mRNA cleavage in a pairing-pattern-dependent manner, contrasting with animal miRNAs (rare cleavage) or plant miRNAs (cleavage restricted to near-perfect complementarity). This cleavage dominance reflects fungal evolutionary adaptations to rapidly silence critical genes (Dang et al., 2011). The findings of this study provide clues for a deeper understanding of the regulatory mechanisms of gene expression.

We further predicted the function of the potential target genes of Aa-milR13. We speculated that virus infection triggers the downregulation of Aa-milR13 in the host A. alternata, which in turn upregulates the expression of target genes, thereby affecting the host’s biological traits and virulence.

The predicted Aa-milR13 potential target genes in the host include Subtilisin-like protein, high affinity nitrate transporter NrtB, and DUF431-domain-containing protein. The functions of these genes in A. alternata have not been reported. Firstly, as a type of protease, Subtilisin-like protein involved in protein hydrolysis and degradation processes. It has been reported that Subtilisin-like proteins in microorganisms are associated with inducing programmed cell death in host cells, which can affect the expression of defense enzymes and defense-related genes in the host, and trigger host immune responses (Jiang et al., 2011; Wang et al., 2022; Xiong et al., 2024). High Affinity Nitrate Transporter (NrtB) is usually found in plants, responsible for transmembrane transport of nitrate in environments with low nitrate concentrations, and is a key protein for plant nitrogen nutrition absorption. Akhtar et al. (2015) reported disruption of the Aspergillus nidulans high-affinity nitrate transporter genes NrtB prevents growth on nitrate (Akhtar et al., 2015). DUF431 domain-containing protein is predicted to be an SAM-dependent RNA methyltransferase in NCBI, catalyzing RNA methylation and RNA modification enzymes, which has not been reported in A. alternata. however, in F. verticillioides, it was found that RNA m5C methylation modification was negatively correlated with fumonisin biosynthesis (Hou et al., 2024). In our study, when A. alternata is infected by virus, the expression of Aa-milR13 is downregulated. This downregulation may relieve the inhibitory effect of Aa-milR13 on its target genes. Consequently, these alterations might indirectly influence the overall physiology and metabolism of the fungus, including its pathogenicity. Further in-depth research is required to determine the precise interactions.

In addition, we also analyzed the target genes of Aa-milR13 in plants ( Table 4 ) and found that the number of target genes in apples was much higher than that in A. alternata. Most of the target genes in apples are related to growth and development regulation and apple host defense immunity. To further clarify the cross-kingdom regulatory potential of Aa-milR13, we could perform dual-luciferase reporter assays in Nicotiana benthamiana to validate its direct post-transcriptional regulation of high-confidence apple defense genes (Feng et al., 2023), which would help reveal how Aa-milR13 interacts with host gene expression during infection. Second, to investigate the molecular mechanisms underlying its cross-kingdom activity, we could track Aa-milR13 subcellular localization in N. benthamiana using fluorescent labeling and live-cell imaging, as previously observed for fungal small RNAs like those from B. cinerea (He et al., 2023). Additionally, testing the dependence of Aa-milR13-mediated regulation on host RNAi components via knockout assays could clarify its reliance on plant machinery (Feng et al., 2017). Collectively, these studies will clarify Aa-milR13’s role as a cross-kingdom regulator.

The study of Aa-milR13 in A. alternata has important theoretical significance and practical application value. From a theoretical perspective, studying Aa-milR13 helps to gain a deeper understanding of the life activity mechanisms of fungi, especially the role of small RNAs in fungal growth, development, and pathogenicity. This will provide new clues and ideas for further revealing the biological characteristics of fungi. From a practical application perspective, research on Aa-milR13 is expected to provide new strategies and methods for the prevention and control of apple diseases. For example, inhibitors or interference technologies targeting Aa-milR13 can be developed to reduce fungal pathogenicity and enhance apple defense capabilities, thereby reducing the occurrence of apple diseases. In addition, Aa-milR13 can be utilized as a biomarker for early detection and diagnosis of apple diseases, providing a basis for timely adoption of prevention and control measures.