Identification of jasmonic acid-associated microRNAs and characterization of the regulatory roles of the miR319/TCP4 module under root-knot nematode stress in tomato

Highlight Screening revealed that the action of miR319/TCP4 in serving as a systemic defensive responder and regulator that modulated the RKN systemic defensive response was mediated via JA.


Introduction
Small RNAs (sRNAs) are a class of short (20-30 nucleotide), endogenous, non-coding RNAs that play important roles in plant growth, development, and adaptation to environmental stress (Lu et al., 2008;Ruth et al., 2010). sRNAs primarily include microRNAs (miRNAs) and small interfering RNAs (siRNAs). In plants, miRNAs have been shown to regulate plant growth and development under various biotic and abiotic stresses, such as drought stress in Oryza sativa and Populus euphratica (Lu et al., 2008;Zhou et al., 2010;Li et al., 2011), heat stress in Brassica rapa (Yu et al., 2012), and heavy metal stress from copper (Waters et al., 2012), aluminium (Chen et al., 2012), and cadmium (Ding et al., 2011). miRNAs were also reported to be involved in plant-parasite interactions. In plants, miR393 was first reported to play a role in plant antibacterial PTI (pattern-triggered immunity) by regulating the auxin signalling pathway (Navarro et al., 2006). High-throughput sequencing identified a range of miRNAs, such as miR156, miR159, miR172, miR319, miR393, and miR396, in response to Pseudomonas syringae (Zhang et al., 2011). Recently, a study of tomato infection by Cucumber mosaic virus (CMV) revealed 79 miRNAs and 40 predicted candidate miRNAs that were responsive to CMV infection, including miR156, miR159, miR172, miR319, mi393, and miR396 (Feng et al., 2014). Shen et al. (2014) revealed that 62 miRNAs are responsive to Verticillium longisporum infection, including the conserved miR319 family. This growing body of evidence strongly suggests that miR-NAs play roles in plant defences against viruses (Kasschau et al., 2003;Bazzini et al., 2007;Navarro et al., 2008), fungi (Lu et al., 2007), and bacteria (Jagadeeswaran et al., 2009;Xin et al., 2010;Zhang et al., 2011). In addition, a negative correlation between miRNA abundance and their targets was observed in Arabidopsis roots after infection with cyst nematode, leading to the down-regulation of miR156, miR159, miR172, and miR396 (Hewezi et al., 2008). Further studies have shown that the miR396/GROWTH RESPONDING FACTOR 1 (GRF1)/GRF3 regulatory module acts as a developmental regulator in the formation of syncytia during cyst nematode infection (Hewezi et al., 2012). Since cyst nematodes and root-knot nematodes (RKNs) feed on syncytia and giant cells, respectively, the potential mechanisms of formation of feeding sites are distinct (De Almeida Engler et al., 2011;Kyndt et al., 2013). To date, however, studies of the link between RKN invasion and miRNA-modulated defence responses are extremely limited.
Jasmonic acid (JA) as a systemic signalling molecule is effective against tomato RKN disease and can reduce the number of root knots from nematode invasion (Cooper et al., 2005;Fujimoto et al., 2011;Nahar et al., 2011;Fan et al., 2014). JA is synthesized via the octadecanoid pathway in leaves, which involves the translocation of lipid intermediates from chloroplast membranes to the cytoplasm and later into peroxisomes (León, 2013). However, JA-mediated RKN resistance occurs in roots. Therefore, synthetic JA and methyl jasmonate (MeJA), which are considered to be long-distance signalling compounds, can be transported from shoots to roots via the vasculature to activate a series of defence responses against RKN invasion (Heil and Ton, 2008). JA has been shown to be transported in the phloem in response to pathogen infection (Schilmiller and Howe, 2005;Lough and Lucas, 2006;Thorpe et al., 2007;Truman et al., 2007). Therefore, the phloem is considered to be a hub that connects shoots and roots and in which miRNAs potentially play crucial roles.
Recent studies indicate that miRNAs are responsive to JA. For example, exogenous MeJA down-regulates miR156, miR168, miR169, miR172, miR172, miR396, miR480, and miR1310 and up-regulates miR164 and miR390 in Chinese yew (Qiu et al., 2009). In one study, 49 known miRNAs, 15 novel miRNAs, and three tasiRNA (trans-acting small interfering RNA) families were induced in JA-treated wild-type (WT) Arabidopsis, whereas one new miRNA, one tasiRNA family, and 22 known miRNAs were repressed in the JA-deficient aos mutant . Additionally, in Nicotiana attenuata, insect herbivores have been shown to alter sRNA transcriptomes related to phytohormone (JA and JA-Ile) signalling (Pandey et al., 2008), which implies that sRNAs probably serve as regulators in JA-mediated biotic stress responses.
The present work was planned to identify RKN-responsive JA-mediated miRNAs and interpret the roles of candidate miRNA. To the authors' knowledge, it is the first exploration of the responsive miRNAs under RKN stress. Sequencing of sRNAs was performed on phloem tissue after RKN inoculation in WT and spr2 (suppressor of prosystemin-mediated response 2, JA-deficient) tomatoes. A total of 263 known and 441 novel differentially expressed miRNAs were identified, and the roles of the JA-mediated miR319/TEOSINTE BRANCHED1/CYCLOIDEA / PRO-LIFERATING CELL FACTOR 4 (TCP4) module in the RKN-defensive response were demonstrated. Previous evidence has demonstrated that miR319s and their targets participate in abiotic stress responses and play positive roles in high salinity and drought stress resistances (Sunkar and Zhu, 2004;Liu et al., 2008;Lv et al., 2010;Zhou et al., 2010;Thiebaut et al., 2012). The outcomes of this study would provide novel insight into the functions of the miR319/TCP module in plants.

Plant materials and biotic stress treatment
Tomato cultivar CM (Solanum lycopersicum var Castlemart, WT) and its mutant spr2 (deficient in JA biosynthesis because Spr2 encodes a chloroplast fatty acid desaturase involved in JA biosynthesis; Li et al., 2003) were grown in a greenhouse with a day/night temperature of 25/18 °C, an air relative humidity (RH) of 60/70%, and a photosynthetic photon flux density (PPFD) of 500 μmol -2 s -1 for 10 h d -1 . RKN isolates were clonal populations established from a single female. They were maintained on tomato UC82 plants. When the seedlings had a total of four spread true leaves, nematodes were inoculated at 5000 per plant near the root. Leaves, phloem (stem), and roots were harvested from the WT and spr2 at 6, 12, 24, and 72 h after inoculation. For hormone treatment, leaves were collected from the WT after JA treatment at 0 h and 6 h. These samples were used for quantitative reverse transcription-PCR (qRT-PCR) analysis of miRNAs and their targets. The infected seedlings were harvested after 20 d for high-throughput sequencing, and the primary stems were cut 10 cm above the bottom of the root. The phloem tissue was quickly peeled off with forceps and a blade treated with DEPC (diethylpyrocarbonate), and the phloem tissue was placed in RNA-free tubes, flash-frozen in liquid nitrogen, and stored at -80 °C.
To investigate the role of miR319 in response to RKN (Meloidogyne incognita) stress, PCR was performed to confirm that the genomes of the plants under study did not contain Mi-1 ( Supplementary Fig. S1 availbale at JXB online), which confers effective resistance against RKNs in tomato (S. lycopersicum L.) (Seah et al., 2004). The transgenic tomatoes overexpressing Arabidopsis pre-miR319a or tomato TCP4 (LA) were in the M82 background. To overexpress miR319 or LA, the op:gene plants (opmiR319 or opLA m ) were crossed with FIL:LhG4 (pFIL), and hybrid F 1 seedlings were used for further study. The relative expression levels of JA biosynthetic genes in the WT and transgenic tomatoes (miR319-oe, LA m -oe) were analysed at 0, 6, 12, and 24 h after inoculation using qRT-PCR.
Small RNA library construction and high-throughput sequencing Total RNA was extracted with the TRIzol reagent according to the manufacturer's instructions. RNA quality was assessed with an Agilent Technologies 2100 Bioanalyzer. The RIN (RNA integrity number) was 7.2 for the WT and 6.2 for the spr2 mutant. Total RNA from WT and spr2 plants was prepared for sRNA sequencing-by-synthesis according to the protocols and standards for Illumina preparation at the Beijing Genomics Institute (BGI; Shenzhen, Guangdong, China). Briefly, RNA fragments of 18-30 nucleotides were isolated from the total RNA by electrophoretic separation via 15% TBE-urea denaturing PAGE. The 5ʹ RNA adaptor (5ʹ-GUUCAGAGUUCUACAGUCCGACGAUC-3ʹ) and 3ʹ RNA adaptor (5ʹ-pUCGUAUGCCGUCUUCUGCUUGUidT -3ʹ; p, phosphate; and idT, inverted deoxythymidine) were ligated to the isolated sRNAs using T4 RNA ligase (TaKaRa). The products were separated and purified at each step by TBE-urea PAGE. The 62-75 nucleotide ligated products were subsequently transcribed into cDNA and amplified by PCR, followed by purification and separation. The purified DNA fragments were used for sequencing with an Illumina 1G Genome Analyzer, an sRNA digital analysis system.

Small RNA annotation
Raw sequence reads were processed into clean full-length reads by discarding low-quality reads and corrupted adaptor sequences [reads <18 nucleotides, reads >30 nucleotides, reads with a poly(A), 3ʹ adaptor, or 5ʹ adaptor contaminants, and reads without a ligated adaptor]. Adaptors were then trimmed from the remaining highquality sequences. The distribution of small RNA lengths was analysed, and the two samples were compared to identify common and genotype-specific transcripts. The Short Oligonucleotide Analysis Package (SOAP; http://soap.genomics.org.cn) was used to annotate the sRNA sequences, which were then mapped to tomato TIGR reference sequences. After comparing the unique sequences with the Rfam 9.1 (http://ftp.selab.janelia.org/pub/Rfam) and NCBI GenBank databases (http://ftp.ncbi.nlm.nih.gov), all rRNAs, scR-NAs, snoRNAs, snRNAs, and tRNAs were discarded. The NCBI GenBank annotation has priority over Rfam (9.1). The sRNAs were then classified and annotated. The total rRNA content served as a quality standard.

Identification of known and novel differentially expressed miRNAs
The remaining sequences were analysed by BLAST searches of miRBase16.0 (http://www.mirbase.org/index.shtml) for matches to precursors or mature miRNAs; these known miRNAs came from all the plants. To evaluate the expression patterns of known miRNAs, a statistical analysis was used to determine the significance of known differences in miRNA expression between the WT and spr2 strains. Log 2 ratios and scatter plots were used to compare miRNA expression, and fold change=log 2 (treatment/control). Novel miRNAs were predicted according to the precursor hairpin secondary structure in Mireap (https://sourceforge.net/projects/mireap/). Briefly, potential novel miRNAs in the tomato genome were predicted using the following criteria: (i) the sequences of miRNA precursors folded into stem-loop structures that contain an ~21 nucleotide mature miRNA sequence on one arm and the miRNA* derived from the opposite arm to form a duplex with two nucleotides for the 3ʹ overhang; (ii) the Dicer PAZ structural domain, which binds the end of double-stranded RNA, recognizes the two-nucleotide overhang of the stem-loop structure, and cuts it into miRNAs; and (ii) the miR-NAs have lower folding free energies than random sequences with the same nucleotide content, so the stem-loop structures have a folding free energy of at most -18 kcal mol -1 (the lowest free energy is Mfe ≤ -18 kcal mol -1 ). χ 2 tests were used to determine the statistical significance of the differences between the two libraries. Compared with the WT group, the known and novel differentially expressed miRNAs with fold changes >2.0 and a P<0.01 were selected.

Identification and KEGG enrichment analysis of miRNA target genes
The targets of known and novel miRNAs were predicted using psRNATarget (Dai and Zhao, 2011). All the predictive targets were assigned KEGG terms (Moriya et al., 2007), and a search was carried out for significantly enriched KEGG terms compared with the entire transcriptome background.

Real-time PCR assay
A time-dependent expression analysis of selected miRNAs and their targets was performed by stem-loop RT-PCR and qRT-PCR. For stem-loop RT-PCR, total RNA was reverse-transcribed into cDNA using the Super-Script first-strand synthesis system (Invitrogen) according to the manufacturer's instructions. The cDNA was used as a template to perform real-time PCR with gene-specific primers and SYBR Green Mix (TaKaRa). Target expression levels were normalized by using tomato U6 as an internal reference. Stem-loop reverse transcription was performed according to a previous protocol (Varkonyi-Gasic et al., 2007). For qRT-PCR, reverse transcription was performed using TransScript First-Strand cDNA Synthesis SuperMix (TransGen) following the manufacturer's instructions. The resulting cDNA was analysed by relative quantitative PCR in the presence of SYBR Premix Ex Taq II (TaKaRa) in a Bio-Rad iCycler, with β-actin as the internal control. After the PCR, a melting curve was generated by gradually increasing the temperature to 95 °C to test the amplicon specificity. The primers for stem-loop RT-PCR and qRT-PCR are listed in Supplementary Table S1 at JXB online. All reactions were performed in triplicate.

Acid fuchsin staining
The root systems of each plant were harvested at 35 d after RKN infection and then submerged in a 15% solution of McCormick's red food colour (Thies et al., 2002) for 15-20 min to stain the egg masses. The root systems were carefully rinsed under running tap water and evaluated for gall severity and egg mass production.

JA level determination
Plant tissue was homogenized in liquid nitrogen, and ~100 mg of fresh leaves was sealed in 2 ml tubes (if >100 mg, in 5 ml tubes). A 0.5 ml aliquot of extraction buffer (2:1:0.005, isopropanol:water:concentrated HCl) was added to each sample with d5-JA (40 ng CDN Isotopes) as internal standards. The samples were homogenized for 45 s, followed by agitation for 30 min at 4 °C. CH 2 Cl 2 (1 ml) was added to the samples, followed by agitation for another 30 min (the CH 2 Cl 2 extraction step was performed on a vortexer with a 30 sample attachment at room temperature) and then centrifuged at 13 000g for 5 min. After centrifugation, two phases formed, and the plant debris was in the middle of the two layers. The aqueous phase was discarded, and the lower layer was collected, concentrated in a drying machine, and resolubilized in MeOH+water. If the solution contained precipitates, it was transferred to a glass tube and centrifuged at 13 000 g for 30 min. Afterwards, 5-10 μl was injected into the column for analysis (Pan and Wang, 2009).

An overview of sRNA sequencing
To identify the miRNAs involved in JA-mediated RKN resistance in tomato, two sRNA libraries from WT and spr2 mutant tomatoes were constructed. Solexa sequencing technology was used to investigate the expression abundance of sRNA in the two libraries, which generated a total of 17 753 883 and 11 251 489 reads in the WT and spr2, respectively. After removing the adaptors and low-quality tags, 16 694 025 (representing 6 195 892 unique reads) and 10 439 048 (representing 2 520 926 unique reads) clean reads remained for the WT and spr2 libraries, respectively. In total, 5 019 256 (WT) and 1 921 123 (spr2) reads mapped perfectly to the S. lycopersicum genome. Thereafter, the non-coding RNAs, including rRNAs, tRNAs, snRNAs, and snoRNAs, were annotated and removed. Querying the remaining sequences against miR-Base 16.0 identified 38 738 (WT) and 17 546 (spr2) unique reads that matched known miRNAs (Table 1).
The 21 and 24 nucleotide RNAs were the most abundant sRNA species (Fig. 1). The sRNAs ranged from 21 to 24 nucleotides in length and comprised 81.46% of the WT and 76.94% of the spr2 transcripts, which was consistent with evidence found in the conifers Pinus contorta and Taxus chinensis, for which >50% of the sequenced sRNAs were 21 or 24 nucleotides (Qiu et al., 2009). In contrast to the results found in T. chinensis (Qiu et al., 2009), the percentage of 24 nucleotide sRNAs was much higher than the percentage of 21 nucleotide sRNAs in the WT in this study. The sRNAs had a strong bias for 21 nucleotide lengths in the spr2 strain, almost twice the rate in the WT strain. In addition, 24 nucleotide sRNAs dominated the WT strain, which was consistent with previous deep sequencing studies in which Hi-JA Arabidopsis exhibited a bias for 24 nucleotide sRNA populations .

Identification of differentially expressed known and novel miRNAs
To identify the differential expression of known miRNAs from the two libraries for the WT and spr2 strain, clean reads were used to compare known plant miRNA precursors or mature miRNA sequences using miRBase 16.0. Novel miR-NAs were predicted according to the precursors' hairpin secondary structure in Mireap and by using the miRNA/ miRNA* criteria. The miRNA expression was considered to be significantly up-regulated or down-regulated with fold changes >2.0 or fold changes < -2.0, respectively, with P≤0.001. In total, 263 known and 441 novel miRNAs were differentially expressed (Supplementary Tables S2, S3 at JXB online). Scatter plots were used to compare the expression of known and novel miRNAs ( Fig. 2A, B).

Target prediction and KEGG enrichment of differentially expressed miRNAs
miRNAs regulate the expression of specific genes via hybridization to mRNA transcripts to promote RNA degradation, inhibit translation, or both (Krol et al., 2010). Known and novel miRNA sequences were searched against tomato genomic sequences on the psRNATarget webserver. Among the 263 known miRNAs, 131 could be searched for their potential targets, yielding 473 genes (Supplementary Table  S4 at JXB online). Of the 441 novel miRNAs, 161 miRNAs had 359 predicted targets (Supplementary Table S5). KEGG pathway analysis indicated that the significantly enriched pathways were 'plant-pathogen interaction' (ko04626) and 'ascorbate and aldarate metabolism' (ko00053) for known miRNAs and 'plant-pathogen interaction' (ko04626) and 'RNA polymerase' (ko03020) for novel miRNAs (Fig. 2C, D; Supplementary Table S6). Therefore, the focus of further study was on miRNAs in the 'plant-pathogen interaction' category (Table 2).

Expression profiling of selected miRNAs and their targets under RKN invasion
To investigate the potential role of selected miRNAs in response to RKN, the spatio-temporal expression profiling of predicted targets of miR156a, miR159a, and miR172a was also examined in the WT after RKN infection. The squamosa promoter-binding protein gene (Solyc10g078700. 1.1, SPL) is the target of miR156 (Lu et al., 2005(Lu et al., , 2008Bazzini et al., 2007). An opposite expression pattern between miR156a and SPL was observed in leaf (0-24 h), stem (0-48 h), and root (0-72 h) ( Fig. 4A-C). The MYB transcription factor is reportedly targeted by miR159 (Achard et al., 2004;Sunkar et al., 2004;Reyes and Chua, 2007;Liu et al., 2008), and a serine/threonine protein kinase (Solyc06g008320.2.1) was also predicted to be targeted by miR159a in the present analysis. Compared with miR159a, the MYB transcription factor (Solyc01g009070.2.1) showed a reverse expression pattern after 48 h in leaf and root (Fig. 4D, F), and from 24 h to 72 h in stem (Fig. 4E). A negative correlation was found between the expression of the serine/threonine protein kinase gene and miR159a from 24 h to 72 h in stem (Fig. 4H), and no negative correlation was observed in leaf and root (Fig. 4G, I). The AP2-like ethylene-responsive transcription factor is the target of miR172 (Aukerman and Sakai, 2003;Chen et al., 2004;Schwab et al., 2005), and miR172a and AP2-like transcription factor gene (Solyc04g049800.2.1) expression showed a reverse pattern from 0 h to 72 h in leaf  and from 24 h to 72 h in stem (Fig. 4J, K), and analogous patterns occurred in root (Fig. 4L). The fact that these miRNAtarget pairs exhibited inverse expression at particular times and places suggests that miRNAs potentially play roles in self-adaption to RKN invasion through the regulation of their target genes. In the present study, miR319b was responsive to RKN invasion (Fig. 3) and was predicted to be involved in plantpathogen interaction (Table 2). TCP4, the target of miR319 (Palatnik et al., 2003;Koyama et al., 2007;Ori et al., 2007;Efroni et al., 2008), showed a significant up-regulation in leaf, though no obvious expression changes in miR319b were observed from 0 h to 48 h; however, miR319b expression dramatically declined from 0 h to 24 h (Fig. 5A). The reverse expression trends between TCP4 and miR319b were observed in stem and root (Fig. 5B, C). Although the fold change of miR319b expression in sRNA sequencing was not the highest, the plausible miR319b-targeted inhibition of TCP4 occurred in both shoots and roots. For JA treatment, the expression of miR319b declined, and TCP4 increased (Fig. 5D, E). Based on all the above, an attempt was made to identify the function of miR319 during RKN invasion. LA (LANCEOLATE) encodes a TCP-family transcription factor (TCP4) that contains a miR319-binding site in tomato, which could be cleaved and down-regulated by ath-miR319a from Arabidopsis (Ori et al., 2007). A sequence comparison of ath-miR319a (Arabidopsis) and sly-miR319s (tomato) showed that ath-miR319a and sly-miR319b are identical (Fig. 5F), indicating that ath-miR319a could function in place of sly-miR319b in tomato. To study the potential function of miR319b in response to RKN stress in tomato, the transactivation system was used to express ath-miR319a and TCP4 (LA) under the regulation of the FILAMENTOUS FLOWER (FIL) promoter in the transgenic M82 tomato line. The FIL promoter is active primarily in tomato young primordia and later in initiating leaflets (Lifschitz et al., 2006). FIL>>LA m , LA m -oe). The LA m contains a mutation in the LA miRNA-binding site that is complementary to miR319 and is thus a miR319-resistant version of LA (Fig. 5G). Finally, three miR319 overexpression types were obtained, and the expression levels of miR319 and TCP4 in hybrid F 1 were analysed by qRT-PCR (Fig. 5H). Based on the significance of the expression level, type 3 was chosen for the following tests.

miR319 overexpression reduces resistance to RKN
Previous studies reported that the overexpression of miR319 plays a positive role in various abiotic stress responses. For example, miR319 overexpression leads to enhanced cold tolerance in rice . Zhou et al. (2013) revealed that miR319 overexpression enhances salt and drought tolerance in transgenic creeping bentgrass. In this study, WT and transgenic tomatoes were examined for their resistance to RKN infection. Tomato roots were inoculated with RKN and then stained with acid fuchsin at 35 d after inoculation, producing obvious red staining of the galls, and the number of galls per plant was then evaluated. The results indicated that the number of galls on miR319-oe roots was much higher than on WT roots; however, the LA m -oe plants had few galls (Fig. 6). This finding indicated that miR319 and TCP4 had opposite effects on resistance to RKN infection.
miR319 overexpression reduces endogenous JA levels JA is a plant signalling compound that induces resistance to pathogens (Cooper et al., 2005). Schommer et al. (2008) showed that JA biosynthesis is associated with the miR319/ TCP regulatory module in Arabidopsis. Moreover, the JA biosynthesis and signalling pathways play key roles in the RKN resistance of the rice root system (Cooper et al., 2005;Fujimoto et al., 2011;Nahar et al., 2011). In the present study, the expression levels of the pivotal genes involved in JA biosynthesis, including LOXD, AOS1, AOC1, and OPR3, were monitored in WT and transgenic tomatoes at 0, 6, 12, and 24 h after RKN inoculation. LOXD expression in the WT line peaked at 12 h after inoculation, but no obvious change was observed in miR319-oe (Fig. 7A). In the LA m -oe plants, LOXD expression peaked at 6 h after inoculation (Fig. 7B). Moreover, similar expression patterns for AOS1, AOC1, and OPR3 were found in the WT; they first decreased and then peaked at 12 h after inoculation. However, no obvious changes were observed in the miR319-oe plants (Fig. 7C, E, G). In LA m -oe, similar to LOXD, the expression levels of AOS1 and AOC1 peaked at 6 h after inoculation (Fig. 7D, F), and the expression level of OPR3 peaked at 12 h after inoculation (Fig. 7H). These results showed that the expression levels of JA biosynthetic genes in miR319-oe were low, and no obvious change was observed after RKN inoculation. However, those genes accumulated earlier and to a greater extent in LA m -oe plants, suggesting that the miR319/TCP4 module plays crucial roles in modulating JA biosynthesis induced by RKN invasion. In addition, the co-expression of LOXD, AOS1, AOC1, and OPR3 in WT and LA m -oe suggested that those genes belonging to the LOX pathway are probably under the control of the same regulatory mechanism. The levels of endogenous JA in WT and transgenic tomatoes were further detected at 0 h and 12 h after inoculation. The results showed that RNK invasion increased the endogenous JA levels in the WT but not in miR319-oe plants. LA m overexpression led to a remarkable increase in basal and induced JA levels (Fig. 7I).

Changes in expression of miR396 in roots under RKN stress
Previous studies reported that miR396/GRF functioned in nematode resistance in the roots of Arabidopsis (Hewezi et al., 2012). The sequencing data showed that both miR396a and miR396b were differentially expressed between the two libraries (Supplementary Table S2 at JXB online). miR396a expression levels declined in root; however, no obvious expression differences were observed for miR396b (Fig. 8A). The miR396 targets GRF1, GRF3, and GRF4, but not GRF2, were transcriptionally up-regulated (Fig. 8B), suggesting that the miR396/GRF module is responsive to RKN invasion in tomato.
Discussion miR319 negatively regulates RKN resistance by affecting the JA level in plants miR319 is one of the most highly conserved miRNAs in plants and was also one of the first to be identified (Weigel et al. 2000;Palatnik et al. 2003;Schommer et al. 2012). Previous studies revealed that the miR319 family is involved in biotic stress response (Zhang et al., 2011;Feng et al., 2014;Shen et al., 2014). In the present work, miR319 was identified between two libraries, WT+N and spr2+N (Supplemetnary Table S2 at JXB online), and the expression analysis indicated that the response of miR319/TCP4 to RKN invasion occurred in both shoots and roots ( Fig. 5A-C). miR319 has been reported to target TCP genes, which encode plantspecific transcription factors (Palatnik et al., 2003(Palatnik et al., , 2007Schwab et al., 2005;Koyama et al., 2007Koyama et al., , 2010Ori et al., 2007;Schommer et al., 2008Schommer et al., , 2012Nag et al., 2009) that are involved in JA biosynthesis and senescence (Schommer et al., 2008). Microarray experiments comparing the shoot apical meristem transcriptomes of WT and miR319-oe plants have revealed a clear decrease in the levels of all miR319targeted TCP genes (Palatnik et al., 2003;Efroni et al., 2008;Schommer et al., 2008). Consistent with these studies, miR319 overexpression led to a decrease in the transcriptional level of TCP4 (Fig. 5H). Schommer et al. (2008) and  Hao et al. (2012) demonstrated that TCP positively regulates the JA level in plants. Similarly, it was also observed here that LA m overexpression resulted in an increase in the basal and RKN-induced levels of JA-synthetic gene expression and endogenous JA; however, these levels were decreased in the miR319-oe plants (Fig. 7). Previous studies revealed that the JA pathway is a crucial player in maintaining and defending RKN in plants (Cooper et al., 2005;Fujimoto et al., 2011;Nahar et al., 2011;Fan et al., 2014). In the present study, RKN resistance increased in LA m -oe but decreased in the miR319-oe plants (Fig. 6). Taken together, these results suggested that overexpressing miR319 inhibited TCP4, which in turn regulated JA biosynthetic genes (Fig. 7A-H), resulting in a lower endogenous JA level (Fig. 7I). Consequently, the resistance to RKN infection was affected in miR319-oe tomatoes (Fig. 6).
In the WT background, JA treatment led to the decrease of expression of miR319b and increase of TCP4 (Fig. 5D, E) in leaves, which is consistent with the observations in RKN stress (Fig. 5A). Combined with the results in spr2 (Fig. 3), these findings implied that JA acts as the regulator of miR319b in early RKN response (Fig. 9). miR319b was responsive to RKN invasion and was down-regulated in shoot, with the expression of TCP4 subsequently increasing (Fig. 5A, B), such that JA synthesis in shoots may become activated (Fig. 7A, C, E, G, I), leading to JA-mediated systemic resistance (Fig. 9). Combined with the results in miR319-oe and LA m -oe, this suggested that JA mediated miR319 serving as a systemic defensive responder and modulator that functioned at least partially via TCP4 under RKN stress (Fig. 9).

miR396 probably participates in RKN resistance in tomato roots
miR396 is a negative regulator of mitotic cell division that acts through the down-regulation of GRF genes in shoot meristems, leaves, and roots (Liu et al., 2008;Rodriguez et al., 2010;Wang et al., 2011;Hewezi et al., 2012). The miR396/GRF module was reported to be involved in cyst nematode resistance in Arabidopsis (Hewezi et al., 2008(Hewezi et al., , 2012. The accumulated evidence demonstrates that miR319 and miR396 are the miRNAs with the most molecular connections to phytohormone signalling pathways (Curaba et al., 2014). The overexpression of an miRNA-resistant form of AtTCP4, a target of miR319, results in increased miR396 accumulation and a corresponding decrease in AtGRF gene expression, and AtTCP4 also regulates GRF activity independently of miR396 (Rodriguez et al., 2010;Fig. 9). Recently, Schommer et al. (2014) reported that miR319regulated TCP4 positively regulates miR396, thereby repressing GRFs and cell proliferation in Arabidopsis. In the present study, the expression patterns between miR319b and miR396a were diametrically opposite (Fig.5A), hence a similar regulation between miR396a and miR319b/TCP4 appeared to occur in tomato roots under RKN stress (Figs 8A,9). Previous studies demonstrated that miR396 was significantly down-regulated in response to cyst nematodes, which is consistent with the present results ( Fig. 8A; Hewezi et al., 2008), and miR396 overexpression reduced the syncytium size and arrested cyst nematode development by repressing GRFs in Arabidopsis roots ( Fig. 9; Hewezi et al., 2012).
Based on the above data, a model is proposed whereby the miR319/TCP module influences RKN resistance via two pathways in tomato (Fig. 9): TCP-regulated JA biosynthesis in shoots and cell proliferation regulated by the miR396/ GRF module in roots.

Supplementary data
Supplementary data are available at JXB online. Figure S1. Mi-1 gene test of plant materials used in this study. Table S1. Primers used for stem-loop RT--PCR and qRT-PCR. Table S2. Differentially expressed known miRNAs between the WT and spr2 after RKN inoculation. Table S3. Differentially expressed novel miRNAs between the WT and spr2 after RKN inoculation. Table S4. Predicted targets of differentially expressed known miRNAs. Table S5. Predicted targets of differentially expressed novel miRNAs. Table S6. KEGG enrichment of known and novel miRNA targets.