A collection of S. lycopersicum miRNAs (sly-miRNAs) was obtained by merging the entries found in public databases and literature. The miRbase repository (Kozomara et al., 2019) included 147 sly-miRNAs, of which 137 were unique. The sly-miRNA list reported by Kaur et al. (2017) was added, including 136 sequences (56 miRNAs classified as conserved or variants, and 60 as novel miRNAs), which account for 70 unique miRNAs after duplicate removal. Thus, a total of 207 sly-miRNAs (obtained from the two merged lists), with 185 unique sequences, represent the final collection used for tomato miRNAs. The S. lycopersicum transcript dataset (ITAG 2.4.v) was retrieved from psRNATarget (Dai et al., 2018) and included 34,725 transcripts, covering a large part of the tomato annotated genome (37,872 genes) available on NCBI.

Since M. incognita miRNAs (min-miRNAs) are not included in public databases, only literature works were used to define this collection. The list of Zhang Y. et al. (2016) included 144 min-miRNAs (38 classified as conserved and 106 as novel miRNAs), which accounts for 63 unique miRNAs after duplicate removal. The list of Wang et al. (2015) included 102 min-miRNAs, corresponding to 70 unique sequences, of which about 40% is composed of conserved miRNAs. From the 133 min-miRNAs obtained by merging the two lists, 126 were unique and represented the final collection of RKN miRNAs. The M. incognita transcript and CDS datasets were retrieved from the V2 genome assembly (Blanc-Mathieu et al., 2017), available at the INRAE Meloidogyne Genomic Resources website (https://meloidogyne.inrae.fr), which includes 43,718 transcript and CDS sequences. The M. incognita 3’UTRome was obtained by processing the transcripts and CDS lists above with a custom Python (v3.8) script, resulting in 20,201 sequences because not all the full transcripts have an annotated 3’UTR.

The cross-kingdom search of miRNA targets in S. lycopersicum and M. incognita transcriptome was performed using the psRNATarget (https://www.zhaolab.org/psRNATarget/) and RNAhybrid (https://bibiserv.cebitec.uni-bielefeld.de/rnahybrid/) tools, respectively, previously proposed for miRNA target prediction in plants (Dai et al., 2018) and animals (Kruger and Rehmsmeier, 2006). The collections of M. incognita and S. lycopersicum miRNAs were used as input for both tools, together with the transcriptome of the target organism (tomato and nematode, respectively). The two tools have been validated on intra-kingdom miRNA-target data in their original publications. However, it is worth mentioning that no large-scale validation is available for cross-kingdom predictions because of the very low number of validated miRNA-target pairs. Even though no gold standard tool or set of rules has been defined for cross-kingdom miRNA interactions, we assumed that such regulations follow the rules of the host organism, as it was assumed in other works (Shu et al., 2015; Chin et al., 2016; Zhang H. et al., 2016; Hou et al., 2018; Zhao et al., 2018; Bellato et al., 2019), thus motivating the use of plant- and animal-specific prediction tools. Consistent with previous in silico and in vivo studies (Liu et al., 2017; Li Z. et al., 2018; Lukasik et al., 2018; Mal et al., 2018; Wang et al., 2018), instead of restricting the search to the 3’UTRome, reported to be the preferential target of endogenous miRNA regulation in animals (Bartel, 2004), the full transcript sequences were used to search for targets in the RKN transcriptome. For the reasons above, the miRNA-target pairs found by the two tools represent putative regulations that may occur in nature, even though other herein neglected biological factors could play important roles.

RNAhybrid and psRNATarget were both run by setting a maximum of 50 targets per miRNA as previously done (Zhang H. et al., 2016; Bellato et al., 2019), to obtain a balanced set of miRNA-target pairs to be analyzed downstream. RNAhybrid was run using a Minimum Free Energy (MFE) threshold of -25 kcal/mol, corresponding to the upper bound of many experimentally found miRNA-target pairs (Zhang H. et al., 2016). Putative targets were ranked based on their MFE value (lowest to highest) for further selection of a subset of these genes, as required in the downstream steps (see 2.3). psRNATarget was run with an Expectation value of 3, corresponding to a slightly relaxed threshold in terms of prediction results stringency, as reported by Dai and Zhao (2011). Default values were used for the other psRNATarget parameters, as previously done in other studies (Bellato et al., 2019): Penalty for G:U pair = 0.5, Penalty for other mismatches = 1, Extra weight in seed region = 1.5, Seed region = 2-13 nucleotides, Mismatches allowed in seed region = 0, HSP size = 19.

In addition to the procedure described above (indicated as cross-kingdom hybridization), we followed a second search method that does not assume any hybridization rule in cross-kingdom interaction. This alternative method (indicated as seed region-based search) was carried out to find sly-miRNA targets in M. incognita transcriptome and included the following steps: (1) intra-kingdom hybridization (i.e., min-miRNAs vs. RKN transcriptome via RNA hybrid); (2) selection of all the sly-miRNAs that share a seed region (nucleotides 2-7 of the miRNA) with the collection of min-miRNAs; (3) association of the selected sly-miRNAs to the targets of the min-miRNAs with the same seed region. Location (CDS or UTRs) of the predicted miRNA binding and sequence similarity outside the seed region were also recorded to support further selection steps.

Statistically over-represented Gene Ontology (GO) terms in the Biological Process (BP) category were computed via enrichment analysis using the predicted miRNA targets obtained above. Duplicate genes in the target list were removed before analysis. The ClueGO application (v2.5.6) (Bindea et al., 2009) of Cytoscape (v3.7.3) (Shannon et al., 2003) was used as a GO analysis tool. A right-sided hypergeometric test with the Benjamini-Hochberg correction for multiple hypothesis testing and a P-value cutoff of 0.05 was used.

For M. incognita enrichment analysis, an ad hoc construction of the GO library was necessary for ClueGO analyses, as this organism was not previously included in this tool. To this aim, the GO biological processes list reported by Somvanshi et al. (2018) was used as a source, converted into a ClueGO-compatible format using a custom Python script, and integrated into ClueGO. In total, 5,508 and 6,098 unique GO terms in the BP category were present for S. lycopersicum and M. incognita, respectively. The BPs presented in this work after enrichment analysis refer to the most relevant term of the functionally grouped term networks provided by ClueGO. The number of input genes in enrichment analysis was set to obtain a comparable number of relevant terms between plant and RKN.

To further analyze the obtained biological processes, individual genes in the miRNA target list associated with the process were considered, and their function was searched in the literature. For S. lycopersicum, gene information was retrieved from the ITAG 2.4 annotations available in Phytozome v.12 (DOE JGI, https://jgi.doe.gov/more-intuitive-phytozome-interface/). For M. incognita, orthologous genes from other nematodes (mainly C. elegans) were obtained using WormBase ParaSite (https://parasite.wormbase.org/index.html) (Howe et al., 2017). This approach was considered because specific information on RKN genes is not directly available in public online resources. Finally, a subset of BP terms from Somvanshi et al. (2018) related to nematode development was selected and used to filter the sly-miRNA targets in the RKN transcriptome.

For the in vitro interaction assay, sly-miR166b, sly-miR169f, and sly-miR156a were selected taking into account the bioinformatics data obtained from the cross-kingdom hybridization and the seed region-based search approach. These microRNAs were obtained from Invitrogen™ Custom Primer Service (BMR Genomics, Padova, Italy) and the relative sequences are given: sly-miR166b (5’-UCGGACCAGGCUUCAUUCCCC-3’, STAR strand GGAAUGUUGUCUGGCUCGAGG), sly-miR156a (5’-UUGACAGAAGAUAGAGAGCAC-3’, STAR strand GCUCUCUAUGCUUCUGUCAU) sly-miR169f (5’-UAGGCGUUGUCUGAGGCUAAC-3’, STAR strand AUCCGUUACUGAGGAACCGAUAG).

Susceptible tomato seedlings (Roma cv.) and axenic cultures of phytoparasitic nematodes were prepared as described by Molinari and Miacola (1997). Freshly hatched M. incognita J2s (infective second-stage juveniles) larvae were used for soaking experiments, following a modified protocol (Danchin et al., 2013; Tan et al., 2013). About 8,000 J2s were soaked for 24 hours in a 40µl final volume of mineral water containing different solutions: (1) a siRNA designed to have no similarity in the M. incognita genome (Dalzell et al., 2010) defined as negative control solution (C-J2); and (2) three different solutions of 0.05 mg/ml sly-miRNAs corresponding to each tested tomato miRNA. J2s from each soaking experiment were washed twice with water by centrifugation at 10,000 g for 3 min, and re-suspended in 100 µl of water. Subsequently, the miRNA-soaked J2s were observed using a Leica M125 stereomicroscope (Leica Microsystem S.r.l, Buccinasco, Milano, Italy) to confirm their vitality and split in two Eppendorf tubes: one used for infection assay and another used for RNA extraction (frozen at -80°C).

For the interaction assay, the sly-miRNAs soaked J2 larvae (approx. 50 J2s/root apex) were loaded in wells placed on agar plates at 0.1 mm from the tomato root. Each plate contained three susceptible tomato seedlings, growth in axenic conditions. The infection was observed for six weeks, during which the penetration in the roots was evaluated in terms of gradual enlargements of the root tip caused by rapid cell division and proliferation. The giant cell induction and galls formation was monitored using a modified LEICA Software image analysis (unpublished), and the following scoring system was used: “+” ranging between 0-30%, “++” ranging between 31-60%, and “+++” ranging between 61-90% (Melillo et al., 2006; Cabrera et al., 2015). The parameters referring to both root enlargements and formation of galls (91-100%), were counted and scored as the number of symptoms normalized to the number of roots in which the phenomena were observed. All experiments were conducted in triplicates coming from two independent trials.

RNA was extracted using the TRIzol Reagent (Invitrogen, CA, USA) method, as indicated by the manufacturer. For this, 500 J2s soaked for 24 h in the control solution (C-J2) or the respective sly-miRNAs (miR156a-J2, miR169f-J2) solutions, were used. For the reverse transcription, 1 μg of total RNA was used along with the QuantiTect Reverse Transcription Kit (QIAGEN S.r.l, Milano) following the manufacturer’s instructions.

The qRT-PCR reactions were carried out using the StepOnePlus (Applied Biosystems, Life Technologies, Zurich, Switzerland) system and assembled in a reaction with 1.5 μl cDNA, 10 μl SYBR^® Select Master Mix (Applied Biosystem, Life Technologies, Zurich, Switzerland), 0.2 μl each of 100 μM of forward and reverse primers, and RNAse free water to 20 μl (total volume). Thermocycling was carried out with one cycle at 95°C for 5 min, followed by 40 cycles of 95°C for 45 s and 58°C for 1 min and 72°C for 45 s. The dissociation curve of the final products was checked to ascertain the presence of a single amplification product. The relative quantification was carried out using the 18S ribosomal RNA gene (NCBI accession HE667742.1, WormBase accession Minc3s09153g42974) as a reference gene. The Minc11367 (WormBase Accession Minc3s00025g01614) gene, was selected and tested as a putative target for sly-miR156a, while Minc00111 (WormBase Accession Minc3s00001g00015) was selected as a putative target of sly-miR169f. Oligonucleotide sequences to amplify the genes of interest were designed with Primer3Plus (https://primer3plus.com) and further validated through the online software Oligo Analyzer (https://eu.idtdna.com/calc/analyzer). All the oligonucleotide sequences are provided in Supplementary Table 1 . The ΔΔCt method was used to quantify gene expression. The reactions were performed in triplicate samples of each cDNA and using two independent replicates.

Results of experimental data are shown as mean ± standard deviation (SD) obtained from two independent experiments each with three replicates. Statistical analyses were conducted using the two-way analysis of variance (ANOVA), along with the heteroscedastic Student’s t-test (where *, P ≤ 0.05; **, P ≤ 0.01), available within the Microsoft Excel package.

In this work, a bidirectional bioinformatics workflow was carried out to predict miRNAs cross-kingdom potential in the host-parasite interaction between tomato and RKNs. These predictions are based on sequence complementarity between miRNAs and cross-species targets, as well as sequence similarity between tomato and nematode miRNAs. A total of 523 miRNA-target pairs were found in the cross-kingdom hybridization between S. lycopersicum transcripts and min-miRNAs, corresponding to 469 unique genes and 105 unique min-miRNAs ( Supplementary Dataset 1 ). The majority of genes were targeted by a single miRNA, but putative targeting by two miRNAs occurred as well ( Figure 1A ). All the target genes were adopted for subsequent enrichment analysis that yielded 43 biological processes (BPs), divided into 10 functional groups, as described below in Section 3.2. Conversely, 8,926 transcripts corresponding to 6,428 unique genes were found in M. incognita by cross-kingdom hybridization with the sly-miRNAs ( Supplementary Dataset 2 ). Each gene was putatively targeted by one or more miRNAs ( Figure 1B ). Based on the MFE of the targets, the 290 top-ranked genes were used for the enrichment analysis that resulted in 86 BPs in the 10 functional groups, described further in Section 3.3.

The seed region-based search resulted in the identification of 7 sly-miRNAs having homologies with 9 min-miRNAs; using these sly-miRNAs, a total of 450 putative target genes were identified in M. incognita ( Supplementary Dataset 3 ).

To analyze the obtained datasets, we first looked at the potential effects that min-miRNAs would have on the development of tomato plants providing discussions based on the function of putatively targeted genes. The biological processes found to be enriched among the putative cross-species targets of RKN miRNAs against S. lycopersicum transcripts are reported in Supplementary Dataset 4 and graphically represented in Figure 2 . These processes clustered in several major GO terms and the most abundant networks included brassinosteroid-mediated signaling pathway, response to herbivores, and positive regulation of protein catabolic process. Other processes like inositol-lipid mediated signaling, co-translational protein targeting to membrane, non-recombinant repair, rRNA methyl transferase, and cellular response to starvation, were less abundant. Considering the min-miRNAs putatively targeting genes in the tomato dataset, the enriched terms list includes 26 miRNAs and 46 unique targets, with the most interesting ones being summarized in Table 1 . Within the brassinosteroid-mediated signaling pathway, different phosphatases, phosphodiesterase, and proteases were predicted to be targeted by NOVEL-18-1, min_miRNA6, and min_miRNA98. Many protease gene families are well-known to be involved in plant immune responses (Balakireva and Zamyatnin, 2018). For instance, aspartyl proteases are specifically linked to systemic acquired resistance (SAR) induced in response to local infections (Breitenbach et al., 2014; Molinari et al., 2014; Molinari and Leonetti, 2019). Other miRNAs, like NOVEL-8, miR-50, and min_miRNA29, are predicted to target genes playing important roles in the response to herbivores. The S-receptor kinase-like genes, aside from being involved in stress responses and host-pathogen defense, have also important functions in cell signaling and development (Takasaki et al., 2000; Becraft, 2002; Pastuglia et al., 2002). Similarly, genes involved in the cuticle and cell wall development and composition (e.g., sterols, mannans) represent important defense lines against many types of pathogens (Wang et al., 2013; Kazan and Gardiner, 2017). Other min-miRNAs (min_miRNA206, NOVEL-18-1) were predicted to target different E3 ubiquitin-protein ligases in tomatoes; these proteins play important roles in the regulation of cell homeostasis and thus they are key regulators of plant growth and stress responses (Liu et al., 2021). Importantly, they are involved in the regulation of plant innate immunity (Duplan and Rivas, 2014).

Similarly, miRNAs (e.g., NOVEL-44, NOVEL-6-1, miR-76) affecting translational regulation can play critical roles in the plant defense against pathogen infection. In addition to the essential role of DNA repair in maintaining genome stability, recent works are discussing the involvement of DNA repair proteins in plant-pathogen interactions and SAR (Fu and Dong, 2013; Camborde et al., 2019). Moreover, the miRNA-mediated control over DNA damage responses is starting to gain more interest from both an endogenous (Gualtieri et al., 2021; Macovei et al., 2021) and cross-kingdom fashion (Bellato et al., 2019). An interesting finding in this sense is the case of let-7, one of the most abundant miRNAs found in nematodes, putatively targeting the nuclear pore complex protein Nup98-Nup96 in tomatoes. This complex plays a role in nuclear-cytoplasmic trafficking and mRNA export, being involved in several important biological events such as mitotic checkpoints. Nup98-deficient mutants share pleiotropic phenotypes (decreased root elongation, accelerated floral transition, reduced fertility, and robustness), indicating that it has a critical role in plant development (Parry, 2013).

Other miRNAs worth mentioning are miR-81, min_miRNA112, and min_miRNA206 predicted to putatively target genes involved in sugar metabolism, like sucrose-phosphate phosphatase (SPP) and trehalose-p6-phosphate synthase (TPS). SSP is an important regulator of carbon partitioning and loss-of-function of this gene leads to altered carbohydrate distribution resulting in a reduced growth rate (Chen et al., 2005, 2008). TPS leads to the formation of the trehalose-6-phosphate (T6P), a metabolic intermediate acting as a signaling molecule that regulates sugar metabolism (Paul, 2008), and its silencing in the tomato caused dysfunction in ROS accumulation and decreased expression of genes responsible for defense against pathogenic infections (Suárez et al., 2008).

To conclude, the prediction analyses show that the tomato genes putatively targeted by M. incognita miRNAs have essential roles in plant development and stress response, and their silencing can have negative repercussions for the plant. This is in agreement with the parasitic relation between the two organisms, where M. incognita tries to hijack the plant systems to promote its development.

When tomato miRNAs were evaluated against the nematode, the enrichment analysis of target genes resulted in BPs related to regulation of Wnt protein secretion, striate muscle contraction, sterol transported activity, ubiquinol-cytochrome-c reductase activity, [2Fe-2S] cluster assembly, glycogen biosynthetic process, defense response to fungus, serine-type exopeptidase, and apoptotic mitochondrial changes ( Figure 3 ; Supplementary Dataset 5 ).

Among the most interesting examples investigated are sly-miR156, sly-miR166, and sly-miR319, given that they have been previously identified as highly abundant in roots during M. incognita infection (Kaur et al., 2017). The abundance of these miRNAs may favor their uptake in high amounts during feeding. Table 2 presents the biological processes and putatively cross-kingdom targeted genes in relation to these three miRNAs. The data shows that sly-miR166(b,c) has the most abundant number of putative targets, distributed in GO terms related to response to stress and developmental processes. Although M. incognita genome has been sequenced (Abad et al., 2008), its annotation is not completed. To understand the roles of these putative targets we have looked at orthologues in related species. Among the genes involved in neuromuscular developmental processes, rGCs (guanylyl cyclases) plays a role in sensory processing (Maruyama, 2017), lev-11 (LEVamisole resistant) encodes a conserved CUB-domain containing transmembrane protein (Gally et al., 2004), snt-1 (SyNapTotagmin) functions as a Ca²⁺ sensor (Li L. et al., 2018), and nep-2 (NEPrilysin metallopeptidase family) is a homolog of the extracellular peptidase neprilysin whose loss-of-function leads to movement anomalies (Soh et al., 2020). Impaired movement can be also caused by dysfunctions in the genes coding for proteins that are part of the collagen extracellular matrix. This may be the case of dpy-17 (DumPY: shorter than wild-type), involved in cuticle development (Lang and Lundquist, 2021), and emb-9 (abnormal EMBroygenesis), an alpha1(IV) collagen gene with embryo lethal effects when mutated (Guo et al., 1991; Li-Leger et al., 2021). Other putative target genes of sly-miR166 are involved in cell cycle regulation, like cdc48 (Franz et al., 2014), rpt-2 (Papaevgeniou and Chondrogianni, 2014), and flub-2 (Haskell and Zinovyeva, 2021), nutrient availability (rpb-7, Collins et al., 2016), and gonad development (gon-1, Blelloch et al., 1999). Regarding sly-miR156a, this was predicted to target dhc1, encoding for the dynein heavy chain protein, and smrc-1, belonging to the Swi/snf chromatin remodeling complex. Dynein is an ATP-powered microtubule-based molecular motor, whose function includes the transport of cargo around the cell, while the loss-of-function of this gene inhibits apoptosis (Harders et al., 2018). The smrc-1 gene is involved in the protection against DNA replication stress and its loss-of-function leads to the accumulation of chronic replication stress (Yang et al., 2019). Finally, sly-miR319a was predicted to target the dlat-1 gene, encoding an enzyme with acetyltransferase activity; its lack of function leads to an early embryonic arrest (Li-Leger et al., 2021).

In the subsequent analysis, tomato miRNAs that share sequence homology with M. incognita miRNAs were analyzed via the seed region-based approach. Annotated orthologues of some predicted targets are collected in Table 3 . Among the targeted biological processes, neurological development is much represented by putative targets such as tgs-1 (trimethyl guanosine synthase), gcy-9 (receptor-type guanylate cyclase), mig-6 (abnormal cell MIGration, papilin), kal-1 (human KALlmann syndrome homolog), as well as chemosensory genes like Mi-odr-1 (Minc3s00015g01026, Minc3s00056g02910). A recent study indicated that Mi-odr-1 is present in two copies in M. incognita, and was found to be expressed in the cell bodies of amphidal neurons and phasmids (Shivakumara et al., 2019). Silencing the Mi-odr and Mi-gpa genes could affect the nematode perception and infestation of the tomato root system (Li et al., 2022). Loss of tgs-1 function in C. elegans leads to neurological phenotypes similar to those caused by the survival motor neuron (SMN) deficiency (Chen et al., 2022) whereas gcy-9 mutants have different physiology in relation to adaptation and plasticity (Rossillo and Ringstad, 2020). The C. elegans kal-1 gene affects epidermal morphogenesis by regulating the development of the substrate neuroblasts, and kal-1 mutants show delayed migration of the ventral neuroblasts (Hudson et al., 2006). Loss-of-function mutants for the extracellular matrix molecule mig-6 result in defects in dendrite formation (Ramirez-Suarez et al., 2019). Other important genes, like unc-52 and cbp-1,2,3, are involved in larval development (Merz et al., 2003) and embryogenesis (Shi and Mello, 1998). Inhibition of these genes results in defects in myofilament assembly, larval movement deficiencies, or developmental arrest (Rogalski et al., 1995; Shi and Mello, 1998).

Hence, most M. incognita genes putatively targeted by sly-miRNAs have important roles in nematode development, leading to adverse effects on digestion, mobility, and reproduction, often with lethal outcomes. This finding is of utmost importance for the agricultural sector in view of developing plant miRNA-based technologies to control nematode diffusion.

When considering the use of different computational methods to predict miRNA targets in a cross-kingdom manner, such as the ones used in this work, it is important to underline that these rely on different assumptions on targeting rules and are still necessary to face our currently limited knowledge in miRNA trans-species interactions. Experimental confirmations are therefore needed to understand the actual targeting roles of the illustrated miRNAs. The availability of these computational methods is however useful to guide researchers in the selection of miRNA candidates for further investigation, despite the target prediction algorithm outcomes could be different due to the different underlying assumptions.

To investigate the hypothesized cross-kingdom miRNAs transfer along with its potential to control the RKN infection, an in vitro experimental system was designed based on soaking assay, nematode larvae phenotyping, and gene expression profiling. To this purpose, J2 larvae were fed with solutions containing tomato miRNAs (sly-miR166b, sly-miR156a, sly-miR169f) selected from the bioinformatics data. The sly-miR166b - Minc3s01925g27230 (formerly named Minc10092) pair was chosen from the cross-kingdom hybridization approach ( Table 2 ). No C. elegans homolog was found for this gene but a protein orthologue was identified as guanylate cyclase (UniProtKB/TrEMBL accession A0A1I8B0V8_MELHA) in M. hapla. The sly-miRNA156a - Minc3s00025g01614 (formerly named Minc11367) pair was selected from the seed-based approach ( Table 3 ). This miRNA has a complete seed-region homology with min_miRNA51 targeting the Minc11367 gene, as shown in Supplementary Figure 1 . In plants, it is well known that miR156 targets the SQUAMOSA PROMOTER BINDING PROTEIN-LIKE (SPL) transcription factor, controlling genes involved in the regulation of reactive oxygen species (ROS) (Yin et al., 2019). In the RKN trans-kingdom approach, this sly-miRNA putative target was predicted as the C. elegans homologous gcy-9 (guanylyl cyclase, WBGene00001536). A homologue of this gene was also identified in the parasitic nematode Haemonchus contortus, a close relative of M. incognita (Winter et al., 2012). Lastly, sly-miR169f was predicted to target Minc3s00001g00015 (formerly named Minc00111, Supplementary Figure 2 ), an orthologue of the C. elegans unc-52 (WBGene00006787) gene. In the GO analysis, this accession was related to functions connected to cell junction organization or muscle system.

Following the selection of tomato miRNAs to be tested in the nematode system, an in vitro soaking assay was performed as described in Section 2.4. The motility and vitality of the J2 larvae were observed and compared with those of larvae kept in the control solution. Subsequently, these larvae were used to infect tomato seedlings. Figure 4 shows the effects of the infection symptoms classified as “+”, “++”, “+++”, along with galls formation on the roots. The values represent the percentage of symptoms normalized to the number of roots where the phenomena were observed. When the effects of the control solution and sly-miRNA156a soaked J2 larvae were compared, the best statistically significant reduction was detected in terms of gall formation. The amount of “+/roots”, “++/roots” and “+++/roots” was also reduced up to 1/3, while the number of “galls/roots” was reduced up to 1/4. A significant reduction of the same parameters was observed also for the J2s soaked with sly-miR169f, where root enlargements (“+/roots”, “++/roots”, “+++/roots”) were reduced up to 60% compared to control, and the number of “galls/roots” was halved. In the case of sly-miR166b, the observed reduction was less prominent, ranging from 15% (“+/roots”) to 25% (“galls/roots”).

Because the most prominent reduction of the galling process was observed for the sly-miRNA156a and sly-miRNA169f soaked larvae, these were selected for further molecular analysis. Quantitative RealTime-PCR was employed to measure the relative expression of the Minc11367 (Minc3s00025g01614) and Minc00111(Minc3s00001g00015) genes, putatively targeted by these miRNAs. The obtained data show a downregulation of 54% in the miR156a-J2 and 29% in the miR169f-J2 treated larvae compared to the control ( Figure 5 ). This result indirectly indicates that, in the in vitro experimental setup applied in this study, sly-miRNA156a and sly-miRNA169f have a cross-kingdom influence on the expression of Minc11367 and Minc00111 nematode genes.

To summarize these findings, Figure 6 shows a schematic representation of how the in vitro sly-miRNAs soaking experiments performed on J2 larvae led to a decrease in the M. incognita infection symptoms along with the downregulation of putative cross-kingdom targets. To our knowledge, this is the first evidence showing that tomato miRNAs can be used to alter the infection ability of M. incognita larvae. Other studies related to the trans-kingdom transfer of tomato miRNAs to pathogens focused mainly on the Botrytis cinerea fungal infection. For instance, sly-miR1001 was shown to inhibit fungal virulence and conidiospore germination by targeting genes encoding for an ATP-dependent metallopeptidase and a cysteine-type endopeptidase (Meng et al., 2020). More recently, a genome-wide study identified multiple sRNAs and miRNAs with antifungal properties (Wu et al., 2023). The same study demonstrated that exogenous application of sly-miR396a was able to suppress the virulence of B. cinerea. Therefore, the data provided in our study together with other experimental evidence form the literature, evidence that tomato miRNAs can be effectively used to limit plant pathogen infection. However, given that most studies provide in vitro evidence to support this fact, in the future it would be required to further investigate how this transfer occurs and how is it conditioned by concentration and bioavailability in an in vivo system.