Characterization of microRNAs from Arabidopsis galls highlights a role for miR159 in the plant response to the root-knot nematode Meloidogyne incognita
Summary
- Root knot nematodes (RKN) are root parasites that induce the genetic reprogramming of vascular cells into giant feeding cells and the development of root galls. MicroRNAs (miRNAs) regulate gene expression during development and plant responses to various stresses.
- Disruption of post-transcriptional gene silencing in Arabidopsis ago1 or ago2 mutants decrease the infection rate of RKN suggesting a role for this mechanism in the plant-nematode interaction. By sequencing small RNAs from uninfected Arabidopsis roots and from galls 7 and 14 d post infection with Meloidogyne incognita, we identified 24 miRNAs differentially expressed in gall as putative regulators of gall development. Moreover, strong activity within galls was detected for five miRNA promoters.
- Analyses of nematode development in an Arabidopsis miR159abc mutant had a lower susceptibility to RKN, suggesting a role for the miR159 family in the plant response to M. incognita. Localization of mature miR159 within the giant and surrounding cells suggested a role in giant cell and gall. Finally, overexpression of miR159 in galls at 14 d post inoculation was associated with the repression of the miR159 target MYB33 which expression is restricted to the early stages of infection.
- Overall, these results implicate the miR159 in plant responses to RKN.
Introduction
Small RNAs are noncoding RNAs of 20–24 nucleotides in length. In plants, they act as key transcriptional and post-transcriptional regulators of gene expression in diverse biological processes (Chen, 2009; Bologna & Voinnet, 2014; Matzke & Mosher, 2014). Plant small RNAs are classified into two predominant classes, according to their biogenesis: microRNAs (miRNAs) and small interfering RNAs (siRNAs) (Axtell, 2013). MiRNAs are generated by the transcription of miRNA genes by RNA polymerase II to produce an RNA precursor that forms a double-stranded RNA hairpin structure, known as the pri-miRNA. This is then processed by DICER to first generate a shorter secondary RNA precursor called the pre-miRNA, and a 20–22 nt-long miRNA:miRNA* duplex. The mature miRNA of the duplex is then incorporated into the RNA-induced silencing complex (RISC) and guides the major protein of the RISC, ARGONAUTE1 (AGO1), to the targeted messenger RNA (mRNA) on the basis of sequence complementarity. The AGO1-bound miRNA hybridizes with the target mRNA, inducing its degradation or inhibiting its translation (Bologna & Voinnet, 2014). Most plant miRNA genes are members of gene families, where mature miRNAs of each family may differ from one another by several nucleotides. Despite this, most miRNAs from a given family are predicted to regulate the same mRNA targets, as shown for the miR159 family (Palatnik et al., 2007).
Plant miRNAs play a key role in regulating gene expression in various processes, including development (Lelandais-Brière et al., 2010; Chitwood et al., 2014), metabolism (Liang et al., 2015; Paul et al., 2015), responses to abiotic stresses (Barciszewska-Pacak et al., 2015), hormone signalling (Curaba et al., 2014) and plant responses to pathogens or symbiotic micro-organisms (Pumplin & Voinnet, 2013; Weiberg & Jin, 2015). For example, Arabidopsis thaliana miR393 was the first miRNA shown to play a role in plant immunity to bacteria. This miRNA, induced by a bacterial pathogen-associated molecular pattern (PAMP) peptide, flg22, represses auxin signalling by targeting the mRNAs encoding the auxin receptor responsible for restricting the growth of Pseudomonas syringae pv tomato DC3000 (Navarro et al., 2006). Furthermore, miR396 has been shown to be involved in successful interactions between Medicago truncatula and the symbiotic fungus Rhizophagus irregularis, and in root growth (Bazin et al., 2013). Plants overexpressing miR396 display lower levels of fungal colonization than wild-type plants and have a lower root biomass and weaker expression of six growth factor genes (MtGRF) and two bHLH79-like target genes.
Root knot nematodes (RKN) Meloidogyne spp. are highly polyphagous sedentary plant parasites capable of infesting most crop species (Blok et al., 2008; Abad & Williamson, 2010). After penetrating host roots, RKN larvae migrate toward the vascular cylinder and reprogram gene expression in several vascular root cells, to induce their development into hypertrophied multinucleate giant feeding cells (GCs). These GCs are metabolically overactive, and serve as the sole source of the nutrients required for RKN development. The growth of the GCs and divisions of the surrounding cells lead to a root deformation known as a knot or gall. GCs are initiated during the first 3 d post infection and they endure successive nuclear divisions coupled with cell expansion, during the first 10 d post inoculation (dpi) in Arabidopsis (Caillaud et al., 2008). By 21 dpi, differentiated GCs have formed and undergone extensive endoreduplication (Wiggers et al., 1990; de Almeida Engler & Gheysen, 2013). The redifferentiation of vascular cells into GCs results from the extensive reprogramming of gene expression in root cells, in response to RKN signals (Escobar et al., 2015). The expression of genes encoding proteins involved in metabolism, the cytoskeleton, cell cycle, cell rescue, defence, hormones, cellular communication and cellular transport are modified in galls and giant cells from various plant species (Favery et al., 2016). We started to decipher the genetic pathways that are modified in infected roots, leading to the formation of GCs, but little is known about the regulators of this reprogramming of gene expression.
Changes in the root miRNA repertoire in response to infection with sedentary plant parasitic nematodes have been reported. The first studies focused on plant responses to cyst nematodes (CN). Cyst nematodes are another subgroup of sedentary parasitic nematodes that induce a multinucleate feeding structure, the syncytium, in plant roots. GCs and syncytia differ in terms of their biogenesis, with the syncytium resulting from the induction of an initial syncytial cell within the root parenchyma that then integrates several hundred of the surrounding cells through cell wall dissolution (Sobczak & Golinowski, 2008). Sequencing approaches identified 30 miRNAs in A. thaliana and 40 miRNAs in soybean that are differentially expressed in response to CN infection (Hewezi et al., 2008; Li et al., 2012). Functional analyses then revealed key roles for two of these miRNAs, miR396 and miR827, in the reprogramming of root cells to form syncytial cells and the suppression of plant defences, respectively, in A. thaliana (Hewezi et al., 2012, 2016). The expression profile of miRNAs during RKN infection has been investigated more recently. Zhao et al. (2015) described the expression profiles of five miRNA genes (MIR156, MIR159, MIR172, MIR319 and MIR396) during the systemic response to RKN infection. They provided evidence for very early regulation of miR319b and its target, TCP4, in the root, within the first 24 h, and showed that TCP4-overexpressing mutants were more resistant to RKNs, whereas miR319-overexpressing plants were more susceptible to RKNs. They showed that miR319b acts as a systemic signal modulating the systemic defence response to RKN via jasmonic acid. The only study reported so far on the global sRNAs profiles within the RKN feeding sites is the sequencing of small RNAs from hand-dissected Meloidogyne javanica galls at early infection times, 3 dpi in A. thaliana (Cabrera et al., 2016). Sixty-two miRNAs were identified as having differential expression between the uninfected roots and early developing galls. Furthermore, a role of the miR390/TAS3 regulatory module in plant responses to RKN was demonstrated, with miR390 KO mutant lines infected with M. javanica displaying significantly lower rates of infection. Gall formation requires miR390/TAS3, through the effects of auxin-responsive factors. In the present paper, we aimed to obtain a complete miRNA profile throughout gall development, by sequencing the population of A. thaliana small RNAs expressed in galls induced by the model RKN species Meloidogyne incognita at two key points during GC/gall development: 7 dpi, corresponding to the phase of successive nuclear divisions without cytokinesis; and 14 dpi, corresponding to an increase of DNA amounts through endoreduplication without nuclear divisions (de Almeida Engler & Gheysen, 2013). The present analysis adds also crucial data to understand the global changes on the sRNA population along the course of RKN infection. We identified 24 miRNAs as differentially expressed in galls 7 dpi and/or 14 dpi as compared to uninfected roots, and established a role for miR159 in the plant–nematode interaction.
Materials and Methods
Biological material, growth conditions and nematode inoculation
Seeds of Arabidopsis thaliana L. Heynh. (ecotype Wassilewskija) or Solanum lycopersicum L. cv ‘Saint Pierre’ were surface-sterilized and sown on Murashige & Skoog (M&S; Duchefa, Haarlem, the Netherlands) medium agar plates (0.5 × MS, 1% sucrose, 0.8% agar, pH 6.4). Plates were kept at 4°C for 2 d, and then transferred to a growth chamber (20°C with 8 h : 16 h, light : dark). Meloidogyne incognita strain ‘Morelos’ was multiplied on tomato plants in a growth chamber (25°C, 16 h : 8 h, light : dark). For in vitro nematode infection, J2 larvae were surface-sterilized with HgCl2 (0.01%) and streptomycin (0.7%) as described elsewhere (Caillaud & Favery, 2016). We inoculated 25-d-old seedlings grown in vitro individually with 200 sterilized J2s each, resuspended in Phytagel. At 7 and 14 dpi, galls were hand-dissected from the infected roots. We also dissected internodes from uninfected roots (without apical and lateral root meristems) at the same time point than gall samples for use as a negative control. Samples were immediately frozen in liquid nitrogen and stored at −80°C. Three independent biological replicates were established for each set of conditions.
A Meloidogyne javanica population was maintained in vitro on cucumber (Cucumis sativus cv Hoffmanns Giganta) and egg masses hatching according to Díaz-Manzano et al. (2016).
All mutant and transgenic lines tested for susceptibility assays with root knot nematodes (RKNs) are listed in Supporting Information Table S1 with references.
Construction and sequencing of small RNA libraries
Total RNA, including small RNAs (< 200 nt long), was isolated from galls or uninfected roots at 7 and 14 dpi. Approximately 150 galls or internode fragments from uninfected roots were independently ground into powder in liquid nitrogen, with a mortar. Total RNA was extracted from these samples with the miRNeasy Mini Kit (Qiagen), according to the manufacturer's instructions, with three additional washes in RPE buffer. The quality and integrity of the RNA were assessed with a Bioanalyzer (Agilent, Santa Clara, CA, USA). Small RNA libraries were generated by ligation, reverse transcription and amplification (11 cycles) from total RNA (2 μg), with the reagents of the NEBNext Small RNA Library Prep Set for SOLiD. Libraries were then quantified with the Bioanalyzer High Sensitivity DNA Kit (Agilent) and sequenced on a SOLiD 5500 wildfire sequencer (Life Technologies, Carlsbad, CA, USA) at the Nice-Sophia Antipolis functional genomics platform (France Génomique, IPMC, Sophia Antipolis, France). The full raw sequencing data were submitted to the GEO database (http://www.ncbi.nlm.nih.gov/geo/) with the accession no. GSE100498.
Bioinformatic miRNA analysis
For each library, adapters were trimmed and reads matching ribosomal RNA, mitochondrial RNA and repeat sequences were removed by performing Blast analyses with the sequences listed in the Rfam database (Nawrocki et al., 2015). The Bowtie aligner was then used to align the trimmed reads (Langmead et al., 2009) on a virtual concatenated genome generated from the A. thaliana genome (TAIR10.21) and the M. incognita genome (Abad et al., 2008). Each read was attributed to the A. thaliana and/or M. incognita genome on the basis of the best alignment obtained (two colour mismatches allowed). Reads with identical best alignment hits for the two genomes were attributed to both the nematode and the plant. If one read aligned to multiple genomic locations, a single read was attributed to each locus.
Low-quality mapped reads were removed (more than four colour mismatches, zero sequence mismatches and zero unknown nucleotides ‘N’), and reads corresponding to molecules of between 20 and 24 nt in size were retained for further analysis. The htseq-count package (Anders et al., 2014) was used to count reads mapping perfectly onto the predicted A. thaliana mature microRNA (miRNA) sequence (TAIR10). Reads mapping to multiple loci were counted for each of the loci concerned. The counts for mature miRNAs from each replicate were used for differential expression analysis with the R package edgeR (Robinson et al., 2010). Differentially expressed miRNAs, identified with a false discovery rate of 5% (adjusted P < 0.05; Benjamini–Hochberg adjustment), were retrieved.
Infection assays
For infection assays, 20–29 A. thaliana plants for each line (mutants and wild-type (WT)) were grown in a mixture of 50% sand and 50% soil, in a growth chamber. We inoculated each plant with 200 M. incognita J2s 21 d after germination. Six weeks post infection, the roots were collected, washed and stained with eosin. Stained roots were weighed and galls and egg masses were counted for each root under a binocular microscope. Mann–Whitney U-tests (α = 5%) were performed to determine the significance of the differences in the numbers of egg masses and galls per root observed between mutants and WT.
Localization of miRNA expression
We localized miRNA promoter activity in A. thaliana lines expressing various reporter genes (GUS, VENUS or YFP) fused to the promoter of one of the selected miRNA genes: MIR167a, MIR167b, MIR167c, MIR167d, MIR408, MIR164c and MIR394b (Table S1). We inoculated 21-d-old seedlings in soil and in vitro, as described above. We collected inoculated roots and washed them in water 7, 14 and 21 dpi. GUS staining was performed as described previously (Favery et al., 1998), and the roots were observed under a Zeiss Axioplan 2 microscope. For promoter activity of pMYB33::MYB33::GUS, GUS expression analysis was performed as described in Cabrera et al. (2016). VENUS/YFP-expressing whole galls and primary and secondary uninfected roots were observed with a Zeiss LSM880 confocal microscope. For GUS lines, stained galls were dissected, fixed in 1% glutaraldehyde and 4% formaldehyde in 50 mM sodium phosphate buffer pH 7.2, dehydrated, and embedded in Technovit 7100 (Heraeus Kulzer, Wehrheim, Germany) as described by the manufacturer. Sections were cut and mounted in DPX (VWR International Ltd, Poole, UK), and observed under a Zeiss Axioplan 2 microscope (Zeiss, Jena, Germany). For YFP- and VENUS-expressing lines, galls and root fragments were dissected, embedded in 6% agarose and sliced (100 μm) with a HM650V vibratome (Microm Microtech, Brignais, France), as described by Caillaud & Favery (2016).
In situ hybridization of miRNAs
Plant miRNAs were localized in galls in situ, by applying a published protocol (Díaz-Manzano et al., 2016) to tomato galls. ‘Saint Pierre’ tomato plants were grown in vitro on M&S (Duchefa) medium agar plates (0.5 × MS, 1% sucrose, 0.8% agar, pH 6.4). Tomato roots were inoculated with 200 sterilized J2 10 d post germination, as described above for Arabidopsis. Galls were dissected out by hand 14 d post infection and fixed by incubation in 2.5% glutaraldehyde (Sigma-Aldrich in 0.1% Tween in PBS for 7 d at 4°C. The galls were dehydrated through ethanol bath and Histo-Clear II then embedded in Paraplast (Sigma-Aldrich) for the cutting of sections. The paraffin was then removed and the 10-μM sections were prehybridized as described by Díaz-Manzano et al. (2016), before being hybridized overnight at 55°C with labelled miR159-LNA probes (Exiqon, Vedbæk, Denmark) at a concentration of 10 nM. Digoxigenin let-7 LNA probes (Exiqon) were used as a negative control, because this miRNA has no orthologue within the tomato genome.
Results
Post-transcriptional gene silencing disruption alters plant–RKN interaction
We first investigated a possible role for miRNAs in the plant–RKN interaction, by analysing the response to RKN of A. thaliana plants with mutations of genes encoding two of the main components of the post-transcriptional gene silencing machinery (PTGS): the ARGONAUTE (AGO) 1 and 2 proteins. AGO1, the main component of the RISC complex, plays an essential role in the miRNA-mediated regulation of development and stress responses (Zhang et al., 2011), but it also binds 21 nt small interfering RNAs (siRNAs). AGO1 mutants have highly impaired post-transcriptional gene silencing (PTGS) (Carbonell et al., 2012). In Arabidopsis, AGO2 also binds miRNAs, but with a bias for 5′ adenosine residues and functions in antibacterial and antiviral immunity (Zhang et al., 2011). We selected the hypomorphic mutants ago1-27 and ago1-25, which have been reported to lack PTGS but to display few developmental defects (Morel et al., 2002), the ago2-1 mutant, and the ago1-27 ago2-1 double mutant. These mutants have dark, highly serrated leaves (Morel et al., 2002), but they have well-developed root systems (Fig. 1a). We quantified the infection rate by RKN of these mutants by counting the number of mature females with egg masses per root, as an indicator of the complete development and reproduction of the nematode, comparing the results obtained with those for WT plants, used as a control. The two ago1 mutants, ago1-27 and ago1-25 (Figs 1b, S1), had 55–76.5% fewer egg masses than the WT. The ago2-1 mutant showed a decreased infection rate also, but this decrease was not as strong as that of the ago1 mutants (Figs 1b, S1). Finaly, the ago1-27/ago2-1 double mutant had fewer egg masses than the WT, the decrease in the number of egg masses being similar to that for the single ago1-27 mutant (Figs 1b, S1). The decrease of infection rate induced by hypomorphic ago1 mutations clearly supports a role for PTGS in the plant–RKN interaction.
Overview of the miRNAs detected in galls and control roots
We sequenced the small RNAs expressed in roots and galls at different time points in the plant–RKN interaction. Twelve small RNA libraries corresponding to three independent replicates of galls at 7 and 14 dpi and uninfected roots were sequenced (Fig. 2). Between 31 and 70 million reads were obtained for gall and root libraries at 7 and 14 dpi (Fig. S2). Filtering reads against adaptors, ribosomal RNA, mitochondrial RNA and repeat sequences removed only 0.09–0.4% of the total number of reads, demonstrating the high quality of the sequenced libraries (Fig. S2). Galls consist of a mixture of plant and nematode tissues and therefore contain a mixture of A. thaliana and M. incognita reads. Between 67.2 and 80.2% of the filtered reads in all samples mapped to the concatenated genome constructed from the A. thaliana and M. incognita genomes, indicating a high degree of homogeneity between the replicate samples (Fig. S2). An analysis of the size distribution of the sequenced reads revealed a major peak at 24 nucleotides corresponding to the canonical size of heterochromatic-siRNAs (Figs 2, S3). Of the 427 predicted mature miRNAs and the 325 precursors of miRNAs from A. thaliana listed by miRbase v.21 (Kozomara & Griffiths-Jones, 2014), 235 precursors and 215 mature miRNAs were identified in at least one library at 7 and 14 dpi, respectively (Tables S2, S3). Between 5% and 10% of filtered gall reads mapped to the genome of M. incognita (Table S2). Within the M. incognita genome, a significant number of gall reads mapped to eight loci that were previously predicted as putative MiRNA genes on M. incognita genome v.1 (Abad et al., 2008) (Table S3). Among these eight MiRNA genes, four were conserved in other organisms inclluding the free-living nematode Caenorhabditis elegans or the lepidopteran Bombyx morii, and four had no homology with MiRNA genes listed in miRBase v.21 (Kozomara & Griffiths-Jones, 2014).
Identification of miRNAs differentially expressed in galls
Statistical analyses performed with edgeR (Robinson et al., 2010) identified 10 miRNAs differentially expressed (DE) in galls at 7 dpi from eight miRNA families. Five of these miRNAs were more abundant and the other five were less abundant in galls (Table 1). At 14 dpi, 20 DE miRNAs from 15 miRNA families were identified: 16 DE miRNAs were more abundant and the other four were less abundant. Six DE miRNAs (miR164c, miR319c, miR390a, miR390b, miR398a, miR408) had similar expression profiles at 7 and 14 dpi (Table 1). Galls are complex organs composed of a mixture of plant and nematode tissues. To limit counting errors due to miRNAs conserved between M. incognita and plants, only reads strictly identical to mature Arabidopsis miRNA sequence were counted. We checked the conservation of the 24 DE Arabidopsis miRNAs in M. incognita and none had sequences identical to a sequence from Meloidogyne, excluding the possibility of differences in read counts due to the inclusion of nematode reads. These DE miRNA belonged to 12 miRNA families including 18 miRNA genes that are strongly conserved in the plant kingdom (miRBase v.21; Kozomara & Griffiths-Jones, 2014). Most of the DE miRNAs identified were well supported by a significant number of reads with an average read number superior to 25 in at least one condition. However, five DE miRNAs (miR156 h, miR163, miR2111a-3p, miR831, miR861-5p) were eliminated from further analyses because of their very low coverage. Based on conservation and expression support, we selected 10 of the 18 families of miRNAs DE in galls for further functional analysis. These families contained 16 miRNA genes that are conserved and display robust expression: MIR159a-c, MIR164c, MIR167d, MIR319c, MIR390a-b, MIR394a-b, MIR398a-b-c, MIR399b-c and MIR408.
| 7 dpi | 14 dpi | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Average of normalized reads | Average of normalized reads | ||||||||
| miRNA | Gall | Root | logFC | P-value | Gall | Root | logFC | P-value | Conservation |
| miR156 h | 3 | 0 | 3.94 | 3.813E-02 | 1 | 0 | – | ns | Yes |
| miR157d | 28 | 10 | – | ns | 144 | 21 | 2.12 | 2.552E-03 | Yes |
| miR159a | 327 | 75 | – | ns | 211 | 37 | 1.87 | 5.576E-03 | Yes |
| miR159c | 754 | 149 | – | ns | 653 | 99 | 2.21 | 8.196E-03 | Yes |
| miR163 | 17 | 15 | – | ns | 0 | 6 | −5.09 | 1.062E-05 | No |
| miR164c | 78 | 163 | −2.09 | 1.637E-03 | 92 | 185 | −1.61 | 1.378E-02 | Yes |
| miR167d | 17991 | 4753 | – | ns | 29188 | 4111 | 2.26 | 6.772E-05 | Yes |
| miR319c | 15 | 29 | −2.06 | 2.152E-02 | 7 | 17 | −2.02 | 3.130E-03 | Yes |
| miR390a/b | 12866 | 1185 | 2.50 | 1.637E-03 | 14336 | 1706 | 2.52 | 2.552E-03 | Yes |
| miR394a/b | 401 | 176 | – | ns | 341 | 79 | 1.55 | 2.093E-02 | Yes |
| miR398a | 578 | 61 | 2.26 | 4.136E-03 | 641 | 92 | 2.30 | 2.552E-03 | Yes |
| miR398b/c | 10267 | 1011 | – | ns | 19894 | 1491 | 3.29 | 6.772E-05 | Yes |
| miR399b/c | 90 | 168 | −2.08 | 4.136E-03 | 223 | 211 | – | ns | Yes |
| miR408 | 9029 | 1087 | 2.04 | 4.164E-02 | 9184 | 1220 | 2.41 | 1.022E-03 | Yes |
| miR822 | 2385 | 1422 | – | ns | 397 | 1302 | −2.17 | 1.181E-02 | No |
| miR831 | 6 | 2 | – | ns | 6 | 0 | 2.97 | 3.130E-03 | No |
| miR833b | 106 | 26 | – | ns | 126 | 25 | 1.78 | 2.552E-03 | No |
| miR861-5p | 0 | 1 | −5.79 | 1.410E-02 | 0 | 1 | – | ns | No |
| miR2111a-3p | 0 | 1 | – | ns | 4 | 0 | 3.85 | 1.533E-02 | Yes |
| miR2934-5p | 22 | 10 | – | ns | 30 | 6 | 1.77 | 5.369E-03 | No |
- miRNAs differentially expressed between galls and uninfected roots are shown, with the mean numbers of reads for the three gall and root libraries at 7 or 14 dpi. Gall/root expression fold change (FC) and statistical significance (P-value) for the differentially expressed miRNAs are indicated. The presence of miRNA in species other than Arabidopsis is also mentioned. ns, not significant (P > 0.05). Conservation: ‘yes’ indicated that the miRNA is conserved in Arabidopsis and in at least one other plant species based on plant miRNAs repertoire available in miRbase v.21 (Kozomara & Griffiths-Jones, 2014).
We investigated the expression pattern of biologically validated targets for the miRNAs that were shown as DE in galls at 7 and or 14 dpi by screening results from previous transcriptomic analyses (Jammes et al., 2005). Among the 49 genes listed as targeted by the DE miRNA in plants (Table S4), eight are DE galls according to microarrays (Table S5). Among these eight genes, four displayed an anticorrelated expression profile with their corresponding miRNAs supporting a gene regulation by these miRNAs.
Temporal and spatial expression profiles for miRNAs differentially expressed in galls and roots
Previous analyses of pMIR390a activity showed this promoter to be highly active in galls induced by RKNs (Cabrera et al., 2016), consistent with our sequencing results. We investigated the tissue-specific expression of miRNA genes DE in gall and root, by analysing the pattern of promoter activation in A. thaliana reporter lines. Seven lines carrying the miRNA promoter fused to various reporter genes were tested (Table S1). The activation patterns of the promoters of the miR167, miR408, miR394 and miR164 families clearly showed activity in roots and/or galls.
The miR167 family includes four miRNA genes in A. thaliana: MIR167a, MIR167b, MIR167c and MIR167d. We followed the activity of the four MIR167 promoters in the A. thaliana lines described by Wu et al. (2006). We found that pMIR167b and pMIR167c had no detectable activity in galls (Figs S4, S5), whereas pMIR167a and pMIR167d were highly active in uninfected roots and in galls at 7 dpi, but weakly active in galls at 14 dpi (Fig. 3a–h). The pMIR167a and pMIR167d promoters were both highly active in galls at 7 dpi, but these two promoters had different expression profiles in uninfected roots: pMIR167a was strongly expressed in secondary roots (Fig. 3a), whereas pMIR167d activity was restricted to the root tip (Fig. 3d). These expression patterns are consistent with our sequencing results, which showed MIR167d to be the only MIR167 gene significantly more abundant in galls, but with a time lag, as statistical analyses of sequenced reads identified a specific overexpression of mature miR167d at 14 dpi. We further investigated the cellular distribution of pMIR167a and pMIR167d activity in semi-thin 7 dpi gall sections, by GUS staining (Fig. 3g,h). GUS activity was observed in the whole central part of galls for pMIR167a, whereas it was restricted to the GCs and neighbouring cells for pMIR167d.
Unlike the other miRNAs analysed to date, miR408 is encoded by a single MIR408 gene. Analysis of the pMIR408::GUS line showed that this promoter was highly active in galls at both 7 and 14 dpi (Fig. 3i,j). In uninfected root, pMIR408 was activated specifically in secondary roots (Fig. 3k).
Arabidopsis thaliana has two miR394 genes, MIR394a and MIR394b, with identical mature sequences, and therefore both were identified, by sequencing, as being more abundant in galls. The distribution of MIR394b expression was investigated in an A. thaliana line carrying the c. 3-kb promoter region of MIR394b fused to NLS-YFP (Knauer et al., 2013). We detected pMIR394b in galls and in the vascular cylinder of both infected and uninfected roots. In uninfected roots, the nuclear YFP signal was observed in root vascular cylinder cells of both primary and secondary roots (Fig. 4a–c). Only a few YFP-stained nuclei were observed at 7 dpi galls (Fig. 4d), but larger numbers of YFP-stained nuclei were observed at 14 dpi (Fig. 4e) and 21 dpi, when numerous stained nuclei were observed throughout the gall (Fig. 4f).
The miR164 family has three genes in A. thaliana: MIR164a, MIR164b and MIR164c. However, only miR164c was found to be less abundant at 7 and 14 dpi, according to sequencing data. We investigated the activation of pMIR164c in A. thaliana plants carrying the promoter of MIR164c fused to the VENUS reporter gene and the N7 nuclear localization signal (Sieber et al., 2007). In uninfected roots, a strong nuclear VENUS signal was detected in vascular cylinder cells, with a stronger signal in secondary roots than in primary roots (Fig. 5a–c). In infected roots, we observed a clear nuclear VENUS signal within galls at 7, 14 and 21 dpi, and in surrounding roots (Fig. 5d–f). The pattern of pMIR164c activation was not consistent with the repression of miR164c suggested by statistical analysis of the sequencing reads.
We identified five promoters of miRNA genes (MIR167a, MIR167d, MIR164c, MIR394b and MIR408) as activated within galls at 7 and/or 14 dpi. The activation patterns of these promoters in galls and uninfected roots confirmed the upregulation of MIR167d, MIR394b and MIR408 highlighted by sequencing analyses.
A role for miR159 in the plant response to RKN
We investigated the role of five conserved DE miRNA families (miR408, miR159, miR398, miR319 and miR399) in gall development, using available transgenic mutant lines. Given the redundancy between miRNA family members, wherever possible, we chose to use mutants in which all the members of a miRNA family were affected: overexpressing (OE) lines or multiple knockout (KO) lines. WT, KO and OE lines were inoculated with M. incognita J2s and the susceptibility of these lines was quantified by counting the galls and egg masses produced by the adult females at the root surface. OE and KO miR408, OE miR398, OE miR319 and OE miR399 mutants had similar numbers of egg masses to WT plants (Figs S6, S7). By contrast, a strong and significant (P < 0.05) decrease, of between 40 and 50%, in the number of galls and egg masses was observed for the miR159abc triple mutant (Allen et al., 2010) relative to WT (Fig. 6).
Infection assays suggested a role for the miR159 family in the plant response to RKN infection. We further investigated the role of miR159 in plant responses to RKNs, by analysing the cellular distribution of the mature forms within galls by in situ hybridization. As the miR159 family has been conserved throughout evolution in the plant kingdom (Axtell & Bartel, 2005), we were able to perform in situ hybridization studies on sections of 14 dpi tomato galls. Hybridization were performed on tomato instead of Arabidopsis gall sections because preservation of tomato galls tissues after sectioning and hybridization is enhanced as compared to Arabidopsis and therefore resolution is improved. Hybridization with the miR159 LNA probe corresponding to the tomato miR159 sequence gave an intense specific signal within the giant cells and in the neighbouring cells (Fig. 7a), whereas a very faint signal was detected in any other part of the infected roots (Fig. 7b), and no signal was detected with the negative control probe, miRNA let-7 (Fig. 7c). This distribution suggests with a role for miR159 in gall development during the plant response to RKN. The members of the miR159 family control several transcription factors of the MYB family, by inducing the cleavage of their mRNA (Palatnik et al., 2007). We investigated the expression profile of the transcription factor MYB33, one of two major miR159 targets, in Arabidopsis lines expressing a pMYB33::MYB33::GUS translational fusion (Millar & Gubler, 2005). This construct was susceptible to post-transcriptional gene silencing by miR159. MYB33::GUS expression was clearly observed within galls at early stages of gall development, at 3 and 7 dpi (Fig. 8a,b), at 14 dpi no signal was observed (Fig. 8c). This gradual decrease in MYB33::GUS expression during gall development was correlated with the significant overexpression of miR159a and -c in galls observed at 14 dpi in sequencing analyses (Table 1). Overall, these results suggest that the miR159 family is involved in plant responses to RKN, at medium–late infection stages, probably through the repression of MYB33 transcription.
Discussion
Differentially expressed miRNAs in galls
Plant parasitic nematodes have a striking ability to manipulate normal parenchyma root cells and turn them into hypermetabolic feeding cells. Multiple transcriptomic analyses have been carried out and we are now beginning to decipher the massive reprogramming of transcription underlying the formation of these feeding structures. By contrast, the molecular mechanisms regulating this reprogramming of gene expression are still poorly understood. Two recent analyses identified nematode-responsive microRNAs (miRNAs) very early (3 dpi) in the plant–root knot nematode (RKN) interaction in tomato and Arabidopsis revealing a role for tomato miR319 and its target TCP4 and for the Arabidopsis miR390/TAS3 module in response to RKNs (Zhao et al., 2015; Cabrera et al., 2016). Initial functional analyses revealed a role for the Arabidopsis miR390/TAS3 module and for tomato miR319 and its target TCP4 in response to nematodes. In this study, we aimed to extend these findings by identifying miRNAs expressed in Arabidopsis galls and roots at key later stages of giant cell (GC) formation: 7 and 14 d post inoculation (dpi). Expression studies were also performed at three time points that correspond to different phases of gall formation: 3 dpi corresponds to the first nuclear divisions without cytokinesis; the polyploidy of GCs then increases due to successive divisions, until 7 dpi (Starr, 1993), and, by 14 dpi, nuclear division has stopped but the nuclei and GCs continue to increase in size, due to endoreduplication. By sequencing whole small RNAs from Arabidopsis galls and uninfected roots, we identified 24 miRNAs as differentially expressed (DE) in galls at 7 and/or 14 dpi. We established the expression profiles of five of these miRNAs in galls and provided evidence supporting a role for miR159 in galls.
Repressed miRNAs were over-represented among the 62 miRNAs DE at 3 dpi (Cabrera et al., 2016), whereas the proportions of overexpressed- and over-repressed miRNAs in galls were similar at 7 dpi and overexpressed miRNAs were over-represented (16 of 20) at 14 dpi. A similar switch in miRNA expression profile between early and later stages of feeding site formation was observed in Arabidopsis, in the syncytia induced by the cyst nematode Heterodera schachtii, in which a global repression of miRNAs was observed at 4 dpi whereas overexpressed miRNAs were more numerous in syncytia at 7 dpi (Hewezi et al., 2008). Compared to miRNAs DE at 3 dpi in galls (Cabrera et al., 2016), only two miRNA families, miR390 and miR319, displayed significant differential expression with the same profile at the three time points of gall development considered (Fig. S5). The other miRNA families were DE at a specific single time point or at only two time points, as for miR399, which was differentially regulated at 3 and 7 dpi, miR408, miR398 and miR164, which were differentially regulated at 7 and 14 dpi, and miR167 and miR159, which were differentially regulated at 3 and 14 dpi. Most of the miRNAs DE in galls (19 of 24) have been conserved throughout evolution and have orthologues in multiple plant species. As RKNs induce similar galls and GCs in a wide range of plant species, conserved miRNAs are of particular interest as possible regulators of the gene networks underlying gall formation.
Thirteen miRNA families in Arabidopsis have been identified as DE at 4 or 7 dpi in syncytia induced by the cyst nematode H. schachtii (Hewezi et al., 2008). Four miRNA families – miR157, miR164, miR167 and miR398 – were DE in galls and syncytia at 7 dpi, but, with the exception of miR164, these miRNAs had opposite expression profiles in galls and syncytia. This finding highlights the differences in biogenesis between these two structures, despite their morphological similarities.
We analysed the expression patterns of 71 genes that were defined as biologically validated targets of the miRNAs DE in galls. We identified five genes with an expression profile in galls anticorrelated with their corresponding miRNAs. This result also underlines that the use of microarrays is not the best approach to investigate mRNA targeted by miRNAs partly because of its lack of sensitivity for transcription factors and partly because RNA cleavage products may hybridize with arrays. Analysis of cleavage product resulting from miRNA activity by high-throughput sequencing, the ‘degradome’ method, would represent an accurate approach toward identifying miRNA targets that need to be developed for the plant–nematode interaction.
Functional analysis of DE miRNAs in galls
The in vivo expression patterns of the DE miRNAs in galls were investigated in Arabidopsis lines carrying reporter genes fused to the miRNA promoters. The expression patterns we observed are consistent with the strong expression in galls suggested by the sequencing results obtained for miR394b, miR408 and miR167d. However, there was a time lag between promoter activity and the detection of mature miRNA for miR167d, with strong promoter activity observed in galls at 7 dpi that subsequently decreased, eventually disappearing at 14 dpi, whereas sequencing analyses identified the mature form of miR167d as being statistically more abundant only in galls at 14 dpi. The expression profiles obtained for the four different genes of the miR167 family were consistent with the sequencing results, which showed only miR167d to be overexpressed in galls, with miR167a strongly expressed in both galls and roots. Surprisingly, we observed strong miR164c promoter activity in galls and uninfected roots, whereas sequencing identified miR164c as less abundant in galls. In uninfected roots, the miR164c promoter was much more active in secondary roots than in primary roots. Most of the gall samples were collected from primary roots, whereas healthy internodes were collected from both primary and, mostly, secondary roots. Thus, the repression in galls suggested by sequencing analysis for miR164c may thus be due to differences in expression between primary and secondary roots, rather than a difference in expression specific to galls. We investigated the susceptibility to RKNs of plants with modified patterns of expression for several miRNAs. There is only one copy of MIR408 in the Arabidopsis genome, but the miR398, miR319 and miR159 multi-miRNA gene families each contain several genes with mature sequences differing by one or a few nucleotides, often targeting the same mRNAs (Palatnik et al., 2007; Bouché, 2010). Hence, we overcame the issue of redundancy due to multigene families, by selecting Arabidopsis lines in which the whole miRNA family was affected: multiple insertion mutant (miR159abc) or OE lines. No change in susceptibility was observed for miR408, miR398c, miR319a or miR399c lines. Thus, these miRNAs may not be essential for the Arabidopsis–RKN interaction although overexpression of MIR319a in tomato increased susceptibility to RKN infection (Zhao et al., 2015).
MiR159, a new player in the plant–nematode interaction
The miR319 and miR159 families are both ancient and closely related, with 17 of 21 nucleotides in common (Palatnik et al., 2007). These two families have different targets, with miR159 regulating MYB transcription factors and miR319 guiding the cleavage of TCP transcription factors. However, some of these mRNAs may be targeted by members of both families. For example, MYB33 and MYB65 can also be targeted by miR319 (Palatnik et al., 2007; Zhao et al., 2015). Strong resistance to RKNs was observed in the miR159abc triple mutant, indicating a role for the miR159 family in the formation of galls induced by RKN in Arabidopsis. The miR159 family contains three genes in A. thaliana, in which miR159a and miR159b are the most abundant forms, with very low levels of miR159c expression observed in flowers, in a narrow range of cell types (Allen et al., 2010). Results of in situ hybridization confirmed the expression of the miR159 family within galls and GCs. However, this technique does not discriminate between the different members of the miR159 family that have highly similar mature sequences. Production and analysis of A. thaliana lines expressing reporter gene fused to the promoters of MIR159a, MIR159b and MIR159c and infected with M. incognita should complete our in situ hybridization results and shed light on the member(s) of this family expressed within gall cells. Most analyses of the miR159 family have been performed in flowers in which miR159c has been considered obsolete due to its very weak expression, the apparently low level of selection pressure on this gene and the absence of phenotypic defects and target deregulation in miR159c mutants (Allen et al., 2010; Alonso-Peral et al., 2012). Although miR159b has been shown to be repressed in galls at 3 dpi (Cabrera et al., 2016), miR159a and c were more abundant in galls at 14 dpi. Comparisons of the normalized abundances of the various members of the miR159 family (Tables S2, S3) showed miR159c to be the most abundant form in roots and galls, at both 7 and 14 dpi. This finding suggests that miR159c may play a specific role in roots. Our infection assays on the miR159abc triple mutant demonstrated that miR159 played a key role in the plant response to RKN, but further specific functional analyses for each individual gene of the miR159 family are now required to determine the specific role of miR159c in the plant–nematode interaction. The observation of gall sections and measurement of GC size in the various miR159 mutants are required to determine if these microRNAs are involved in the giant cell formation. Predicted miR159 cleavage sites have been identified in seven gibberellin- and abscisic acid (ABA)-regulated MYB, GAMYB-like, transcription factors (Palatnik et al., 2007), with MYB33 and MYB65 identified as the major targets. Gibberellic acid (GA) has been implicated in plant response to RKN. GA promotes cell division and elongation, and a general induction of genes involved in GA biosynthesis and response was observed in galls and/or GCs from tomato (Bar-Orl et al., 2005) and rice (Kyndt et al., 2012; Ji et al., 2013). In galls, MYB33 expression has been shown to be inversely correlated with the expression of miR159 family members, as assessed by sequencing. Whereas miR159 was repressed at 3 dpi and overexpressed at 14 dpi, MYB33 was strongly expressed and translated at 3 dpi, and its MYB33::GUS protein concentration decreased to undetectable at 14 dpi. Although MYB33 transcripts have been detected by microarrays in galls at 7 and 14 dpi (Jammes et al., 2005), the undetectable concentration of MYB33 protein at 14 dpi showed by our results suggest an inhibition of MYB33 translation at 14 dpi. Altogether, these results suggest that miR159 may play a role during gall development by inhibiting MYB33 translation as it was previously demonstrated (Li et al., 2012). However, identification of miR159 cleavage products by degradome sequencing (Addo-Quaye et al., 2008) is required to identify the various targets of miR159 in galls. The members of the miR159 family also target several other MYB transcription factors. Further functional analyses are now required to investigate the other targets of miR159 in galls, to decipher the miR159 pathway involved in the plant–nematode interaction and the role of miR159 in root development.
Despite the strong resistance to RKN observed in hypomorphic ago1 mutants, suggesting a role for small RNAs and post-transcriptional gene silencing machinery (PTGS) in plant–nematode interactions, most of the miRNAs, except miR159, we have investigated to date were found to have no effect on the plant response to Meloidogyne. ARGONAUTE1 (AGO1) is a key protein of the RNA-induced silencing complex (RISC), a multiprotein complex involved in both miRNA and small interfering RNA (siRNA) pathways that is targeted by pathogens (Weiberg et al., 2013). In plants, siRNAs regulate gene expression at the transcriptional and post-transcriptional levels, and siRNA involvement in plant responses to pathogens is increasingly being reported (Katiyar-Agarwal et al., 2007; Pumplin & Voinnet, 2013; Cabrera et al., 2016). The sequencing results obtained in this work should now be processed with a dedicated algorithm to identify siRNAs produced in galls and to investigate siRNA-mediated regulation pathways and their role in gall formation.
Acknowledgements
The authors thank Lionel Navarro (ENS, Paris, France) for fruitful discussions and for providing the ago1 and ago2 mutant lines. The authors thank Christine Lelandais (IPS2, Paris-Saclay, France) and Nicolas Bouché (IJPB, Versailles, France) for helpful discussions, Jason W. Reed (University of North Carolina, USA) for providing the pMIR167::GUS lines, Thomas Laux (University of Freiburg, Switzerland) for providing the pMIR394b:NLS:YFP lines, Lei Li (Beijing University) for providing the pMIR408::GUS lines lines, Tzyy-Jen Chiou (Agricultural Biotechnology Research Center, Taiwan) for providing the mir399 mutant lines, Olivier Pierre (ISA, France) for providing help with microscope observation and Marc Bailly Béchet (ISA, France) for his advice in statistics. C.M. was supported by ‘Santé des Plantes et Environnement’ INRA department and Provence Alpes Côte d'Azur fellowship. This work was funded by INRA and by the French Government (National Research Agency, ANR) through the ‘Investments for the Future’ LabEx SIGNALIFE: program reference #ANR-11-LABX-0028-01 and by the Plant-KBBE program NESTOR (ANR-13-KBBE-0003-06). Work was also supported by the Spanish government (grants AGL2013-48787; AGL2016-75287-R to CE). ACS was supported by a fellowship from the University of Castilla-La Mancha, co-founded by the European Social Fund. This Work was supported by France Génomique National infrastructure, funded as part of ‘Investissement d'avenir’ program managed by Agence Nationale pour la Recherche (contract ANR-10-INBS-09).
Author contributions
C.M. performed the research, data analysis and interpretation; M.M. collected biological samples and performed ISH experiments; M.d.R., K.L. and C.M. performed the bioinformatic data analysis; N.M. produced the nematodes; V.M. prepared libraries for sequencing; A.R. performed infection assays; B.R. performed GUS analyses; A.M. provided the pMYB33:MYB33:GUS lines and helped to write the paper; C.E., J.C., M.B. and A.C.S. performed pMYB33:MYB33:GUS analysis; S.J-P. and B.F. supervised the experiments and data analyses; S.J-P, B.F. and P.A. supervised the research; and C.M., S.J-P. and B.F. wrote the paper.
