•Root-knot nematodes (RKNs) are obligate endoparasites that maintain a biotrophic relationship with their hosts over a period of several weeks and induce the differentiation of root cells into specialized feeding cells. Nematode effectors synthesized in the oesophageal glands and injected into the plant tissue through the syringe-like stylet certainly play a central role in these processes.
•In a search for nematode effectors, we used comparative genomics on expressed sequence tag (EST) datasets to identify Meloidogyne incognita genes encoding proteins potentially secreted upon the early steps of infection.
•We identified three genes specifically expressed in the oesophageal glands of parasitic juveniles that encode predicted secreted proteins. One of these genes, Mi-EFF1 is a pioneer gene that has no similarity in databases and a predicted nuclear localization signal. We demonstrate that RKNs secrete Mi-EFF1 within the feeding site and show Mi-EFF1 targeting to the nuclei of the feeding cells.
•RKNs were previously shown to secrete proteins in the apoplasm of infected tissues. Our results show that nematodes sedentarily established at the feeding site also deliver proteins within plant cells through their stylet. The protein Mi-EFF1 injected within the feeding cells is targeted at the nuclei where it may manipulate nuclear functions of the host cell.
Root-knot nematodes (RKNs) are among the most economically important crop pests and can infest > 5500 plant species (Blok et al., 2008; Djian-Caporalino, 2012). A fascinating singularity of RKNs, Meloidogyne spp., is the ability of these obligate parasites to hijack plant cell fate and induce the differentiation of root cells into specialized feeding cells called giant cells that provide nutrients to the parasite (Caillaud et al., 2008a). A second peculiarity of RKNs is their ability to maintain a biotrophic interaction with the host for the 3–6 weeks necessary for completion of their life cycle. The interaction starts when preparasitic second-stage juveniles (J2s) penetrate the root apex and migrate between cells until they reach the vascular cylinder. There, parasitic J2s settle in the intercellular space, perforate the plant cell walls with a protrusible stylet and induce the differentiation of the giant cells. Giant cells are hypertrophied, multinucleated and metabolically hyperactive cells that result from several rounds of karyokinesis with aborted cell divisions (Caillaud et al., 2008b). After giant cell induction, parasitic J2s start feeding and develop into parasitic juvenile stages J3 and J4 and finally into adults. Females remain sedentarily fixed and feed from five to seven giant cells until eggs are released on the root surface. It is anticipated that effectors secreted by the stylet contribute to giant cell induction and plant defence suppression during infection reviewed by Davis et al. (2008), Gheysen & Mitchum (2011) and Rosso et al. (2011). Nematode effectors are produced by three oesophageal gland cells and secreted through the stylet in the plant tissue, but their nature and molecular functions are still largely obscure. The genome sequencing of two RKN species, Meloidogyne incognita and Meloidogyne hapla (Abad et al., 2008; Opperman et al., 2008), provided the first genome data available for plant parasitic nematodes and a fabulous opportunity to further explore the genes involved in the success of infection. In a search for new candidate effector genes, we conducted a comparative analysis of expressed sequence tags (ESTs) from various M. incognita developmental stages and retrieved genes specifically expressed during the early steps of infection. For this purpose we generated 43 504 new ESTs that we added to the 20 312 publicly available sequences. A selection pipeline allowed retrieval of the RKN genes specifically expressed in parasitic juveniles that encoded predicted secreted proteins. We identified three candidate effector genes whose expression was restricted to the oesophageal glands of M. incognita parasitic juveniles. We further analysed one gene that we could validate as new effector by in planta immunolocalization and demonstrated that this effector is injected into the feeding site and targeted at the nuclei of giant cells.
Materials and Methods
EST production and clustering
Meloidogyne incognita, population Morelos, were reared on glasshouse-grown tomatoes (Solanum lycopersicum L. cv St Pierre). Eggs and freshly hatched J2s were collected as described previously (Rosso et al., 1999). Stressed J2s were obtained by soaking freshly hatched J2s for 1 h in 0.2 M NaCl, for 30 min in 1% EtOH, for 1 h at 40°C or for 1 h in 0.1% acetic acid. The different batches of stressed J2s were pooled before RNA extraction. Total RNA was extracted using TRIzol reagent (Invitrogen) and mRNA was purified by polyA purification using the Qiagen OligoTex kit (Qiagen). cDNA libraries were obtained with the CloneMiner cDNA Library Construction Kit (Invitrogen). EST sequencing was done at the French Sequencing Center, Génoscope (Evry, France) and newly generated sequences were trimmed as described in Supporting Information, Methods S1. In total 63 816 ESTs were assembled into unisequences using the CAP3 sequence assembly program (Huang & Madan, 1999) with overlap percentage identity cutoff argument set to 95%.
Bioinformatics and sequence analyses
Gene models were attributed to EST contigs using BLASTX searches against the predicted proteins from the genome of M. incognita (http://www.inra.fr/meloidogyne_incognita). Alignments were inspected manually and predicted proteins that showed a minimum of 80% identity and covered a minimum of 50% of the translated unisequence length with E-value < 10−20 were attributed to the respective unisequence. The same criteria were used to attribute gene models from the M. hapla genome (http://www.pngg.org/cbnp/index.php). We checked for the absence of ESTs from homologous genes in eggs, J2s or females of other Meloidogyne species, using BLASTN searches on the dbEST database restricted to Meloidogyne (Taxid: 189290). Alignments were inspected manually and sequences that showed > 95% identity and covered a minimum of 50% of the query with an E-value < 10−20 were removed. We searched for homologs in the National Center for Biotechnology Information (NCBI) NRdb using BLASTN searches and an E-value < 10−10 as threshold. Putative orthologs were searched by reciprocal best-hit comparison with OrthoMCL (Li et al., 2003) using gene models from the genomes of M. incognita (MincV1A1), M. hapla (Mh10g200708), Brugia malayi (bma1), Caenorhabditis elegans (wormpep.ws203.fa), Caenorhabditis briggsae (brigpepws203), Pristionchus pacificus (Ppapepws203) and Drosophila melanogaster (BDGP5.4.54.). Predicted functional domains were identified using InterproScan (Quevillon et al., 2005) restricted to HMMPfam (Punta et al., 2011). From the Interpro accession we mapped Gene Ontology entries (Hunter et al., 2009). In addition, we searched for putative protein functions by similarity searches using BLASTP in the NRdb and Swiss prot databases using a threshold expect value E < 10−10.
In situ hybridizations
For in situ hybridizations, DNA probe templates were amplified from parasitic juvenile cDNAs with gene-specific primers (Table S1). Sense and antisense probes were synthesized in separate reactions with digoxigenin-11-UTP (Roche). In situ hybridizations were performed as described in Methods S1. Bound probes were detected by alkaline phosphatase immunostaining. Stained nematode sections were examined with differential interference contrast microscopy.
Immunolocalization of Mi-EFF1 in tomato infected roots
A serum directed against Mi-EFF1 was raised by rabbit immunization with two synthetic peptides (Eurogentec, Liege, Belgium). Galls from infected tomato roots were dissected, fixed, dehydrated and embedded in methacrylate as described in Vieira et al. (2011). Five micrometre sections were incubated with 1 : 50 primary anti-Mi-EFF1 serum and 1 : 300 secondary fluorescein isothiocyanate (FITC)-conjugated anti-rabbit antibodies. Slides were mounted with ProLong antifade medium (Invitrogen) and observed with an epifluorescence microscope (Axioplan 2; Carl Zeiss).
Subcellular localization in plant cells
The coding sequence of Mi-EFF1 (Minc17998) without secretion signal peptide was amplified by PCR using specific primers from the EST clone mij306d10t1.1 (Table S1) and cloned into the Gateway donor vector pDON207 (Invitrogen) before insertion into pK7FWG2.0 or pK7GWF2.0 green fluorescent protein (GFP) fusion vectors (Karimi et al., 2002). A mutated form of Mi-EFF1 was cloned in which the nuclear localization signal (NLS) PLKRGRE was changed to PLAAGAE. For this purpose, two independent PCR reactions were performed with primers that carried the mutations and a third overlapping PCR was performed using extremity primers (Table S1). For transient expression in plant cells, fully expanded leaves from 3- to 4-wk-old Nicotiana tabacum plants were infiltrated with a solution of transformed Agrobacterium tumefaciens GV3101 (Holsters et al., 1980) (OD600 = 0.5 in 10 mM MgCl2, 10 mM MES (2-(N-morpholino)ethanesulfonic acid), pH 5.6, 200 μM acetosyringone). Infiltrated leaf patches were mounted in water and GFP fluorescence was monitored in Lambda mode with a 499–550 nm beam path (488 nm excitation line) on an inverted confocal microscope (Axiovert 200 M, LSM510 META; Zeiss).
Identification of three new candidate effectors in M. incognita
Before our study, 20 312 M. incognita EST sequences obtained from eggs, J2s, parasitic juveniles or females were available in the NCBI’s dbEST database. For this study, we generated a total of 43 504 new ESTs, among which 35 732 ESTs were generated from eggs and J2s and 7772 ESTs were generated from stressed J2s treated with various abiotic stresses, including salt, oxidative and heat stresses (GenBank accession numbers JN376805–JN377341 and JK266916–JK309882; Table S2). Clustering the NCBI’s and newly generated ESTs yielded a total of 22 350 unisequences that covered 38% of the 19 212 protein-coding gene models identified in the genome of M. incognita (Abad et al., 2008). The newly generated ESTs produced 13 417 new unisequences, providing substantial enrichment to identified transcript sequences in M. incognita. For comparative genomics, ESTs were classified in five datasets containing ESTs obtained from eggs, J2s, stressed J2s, parasitic juveniles and females. In order to identify parasitism genes involved in the early steps of infection, we focused our analysis on the 593 transcript unisequences specifically present in M. incognita parasitic juveniles. Remarkably, the 10 most abundantly represented unisequences within this dataset had no significant hit in NCBI databases and could not be assigned a predicted function, highlighting the lack of knowledge on the nematode functions activated during early parasitic stages (Table S3). We checked for the absence of homologs in egg, J2 or female developmental stages from other Meloidogyne species and retrieved 256 M. incognita unisequences qualified as parasitic juvenile-specific (Fig. 1). We looked for the corresponding gene models in the predicted proteomes of M. incognita and M. hapla. Among these unisequences, 108 had a gene model in M. incognita or in M. hapla. Potentially secreted proteins were identified by the presence of an N-terminal secretion signal peptide (SignalP v 3.0 program; Bendtsen et al., 2004) and the absence of a transmembrane domain (TMHMM v.2.0; Krogh et al., 2001). We identified a total of 28 predicted secreted proteins. Remarkably, overall, 13 proteins encoded by pioneer genes had no similarity with protein sequences available to date in the NRdb database, suggesting that they could result from specific adaptations to parasitism (Table S4). The identification of two predicted collagen proteins in our dataset was consistent with the well documented cuticle modifications that occur during the early steps of parasitism (Johnstone & Barry, 1996). Predicted functions could be assigned to six deduced proteins based on the presence of functional domains and significant similarities with characterized proteins from databanks, including a putative metallopeptidase (Minc00108), a putative Nudix hydrolase (Minc07457) and a putative galectin (Minc12424). (Table S4).
Because nematode effectors are injected into the plant tissue during infection, they have to be expressed in secretory organs of parasitic stages. We localized the expression of 11 genes in parasitic juveniles using whole-mount in situ hybridization (Table S4). Interestingly, we observed the confined expression of the putative metallopeptidase gene (Minc00108) and two pioneer genes (Minc08146 and Minc17998) in the oesophageal gland cells of parasitic juveniles (Fig. 2), indicating that these three genes encode candidate effectors potentially secreted through the stylet into plant tissues during parasitism (protein sequences available at http://www.inra.fr/meloidogyne_incognita). The strong induction of transcription of these three genes in parasitic juveniles was confirmed by reverse transcription-quantitative PCR (Fig. S1).
We identified no putative ortholog to the three new candidate effectors in the animal parasitic or free-living nematodes for which a genome sequence is available. We identified four predicted recent in-paralogs for the putative metallopeptidase Minc00108 in M. incognita (Minc00107, Minc00121, Minc00122 and Minc1149). By contrast, the pioneer genes Minc08146 and Minc17998 appeared as single copy genes. In addition, we identified no putative homologs for the three candidate effector genes using TBLASTN and BLASTX searches in the genome of the vertebrate parasite Trichinella spiralis (Project GenBank ID: 12603) or in the genomes of the plant parasitic cyst nematodes Globodera pallida (http://www.sanger.ac.uk/cgi-bin/blast/submitblast/g_pallida) and Heterodera glycines (Project GenBank ID: 28939), suggesting that the three genes are RKN-specific.
Mi-EFF1 is injected into plant cells during parasitism
We selected Minc17998, whose coding sequence was supported by ESTs on its whole length, to perform further functional analyses. Minc17998 encodes a 122-amino-acid-long predicted protein that we called Mi-EFF1 (for M. incognita EFFECTOR 1). We analysed the secretion of Mi-EFF1 in plant tissues during parasitism by immunolocalization on sections of tomato roots infected with M. incognita. For this purpose, a serum was raised against the Mi-EFF1-specific peptides NGGPVGDYSNFKREC and VQPVNNTVTDTKNSS (Fig. S2). To have a complete view of Mi-EFF1 localization throughout parasitism, immunolocalizations were performed on galls sectioned at 7, 14 and 21 d after inoculation (DAI). At 7 DAI, the anti-Mi-EFF1 serum reacted specifically with a protein in the oesophageal glands of migratory juveniles. No signal was detected outside the nematode at this stage, suggesting that Mi-EFF1 was not secreted during migration (Fig. 3a–d). At 14 DAI, nematodes were sedentary and giant cells were formed. We observed accumulation of Mi-EFF1 into the feeding site outside the nematode (Fig. 3e–l). At this stage Mi-EFF1 was also expressed in the dorsal oesophageal gland of the nematode and a strong signal was observed at the stylet tip (Fig. 3m–p). Interestingly, intense signals were consistently observed in giant cell nuclei (Fig. 3m–t). At 21 DAI, nematodes had developed into females. No signal was observed on sections at this stage (data not shown). Control experiments using the preimmune serum showed no background fluorescence (Fig. S3).
The localization of Mi-EFF1 in the nuclei of giant cells was concordant with in silico prediction of a NLS (PLKRGRE) in its coding sequence (Fig. S2). We further analysed Mi-EFF1 targeting to the nucleus of plant cells using transient expression of GFP fusion proteins in tobacco leaf epidermal cells. To mimic the injection of Mi-EFF1 within plant cells by the nematode, Mi-EFF1–GFP and GFP–Mi-EFF1 fusion proteins were constructed without secretion signal peptide. Interestingly, both Mi-EFF1–GFP (Fig. 4a–c) and GFP–Mi-EFF1 (not shown) fusion proteins accumulated in the nucleus of tobacco leaf cells. As a control, a Mi-EFF1–GFP fusion protein was constructed with a mutated version of the NLS in which the lysine and arginine residues were replaced by alanine residues (PLAAGAE). Confocal microscopy imaging showed expression of the fusion protein in the cytoplasm of agroinfiltrated cells (Fig. 4d). A signal was also observed in the nuclei, probably as a result of passive diffusion from the cytoplasm, as the fusion protein (predicted molecular weight c. 40 kDa) was smaller than the nuclear exclusion size (c. 60 kDa) (Haasen et al., 1999). Of note, in both experiments, using overexpression in tobacco leaves and immunocytochemistry on infested roots, the signal was excluded from the nucleolus.
Plant parasitic nematodes use their stylet as a syringe to secrete proteins within plant tissues during infection. In planta localization studies have shown the secretion of several RKN proteins into the apoplasm, such as the cell wall-degrading and cell wall-modifying enzymes that soften the plant cell walls during root invasion (Vieira et al., 2011). Yet several lines of evidence suggest that other nematode-secreted proteins are injected into plant cells after the nematode has pierced the cell wall with its stylet. In the case of cyst nematodes, screens for plant proteins able to interact with nematode-secreted proteins have identified various intracellular putative targets, including a coiled-coil-nucleotide-binding site-leucine-rich repeat (CC-NBS-LRR) protein (Rehman et al., 2009) and defence- and stress-associated proteins (Hewezi et al., 2010; Patel et al., 2010). In addition, transcription factors of the SCARECROW family were identified as putative targets for an RKN-secreted peptide (Huang et al., 2006). Before our study, however, injection of a nematode protein into plant cells was only observed in a single case, for a CLAVATA3/ESR (CLE)-like protein secreted by the soybean cyst nematode Heterodera glycines that was observed within the cytoplasm of feeding cells (Wang et al., 2010). It was also shown that the protein can be redirected to the apoplasm where it supposedly interacts with a plant receptor at the plasma membrane (Replogle et al., 2011).
In order to better understand the molecular functions of nematode-secreted proteins during the interaction, we explored RKN transcriptomic data and retrieved nematode genes that encoded proteins potentially secreted during the early steps of infection. We retrieved 28 genes from M. incognita specifically expressed in early parasitic stages that encoded predicted secreted proteins. Interestingly, 13 genes (50%) were pioneer genes with no homolog in databanks, pinpointing potential specific adaptations to plant parasitism. Among them we identified three genes specifically expressed in the oesophageal glands of parasitic juveniles that encoded proteins with a predicted secretion signal peptide thus possibly injected via the stylet into the plant tissue during infection. The protein Minc00108 was similar to a predicted astacin metallopeptidase from C. elegans. However, the two proteins were not found in the same orthology group and no ortholog for Minc00108 was identified in the genome of other nematode species, strengthening the hypothesis that this astacin-like protein may have a specific role in the plant–nematode interaction. The candidate effectors Minc08146 and Minc17998 (Mi-EFF1) were pioneer genes with no homolog in databanks and no predicted function. Pioneer genes have previously been found to be abundant among RKN candidate effectors (Huang et al., 2004; Dubreuil et al., 2007). The absence of homologs in free-living or animal parasitic nematodes and in the closely related cyst nematodes further supported a specific role in the plant–RKN interaction.
Using immunolocalizations on infected tomato roots, we confirmed the secretion in planta of Mi-EFF1. Although we could detect the presence of Mi-EFF1 in the dorsal gland of migratory juveniles, no signal was detected along the migratory path of the nematode at this stage, suggesting that Mi-EFF1 is not secreted during migration. By contrast, Mi-EFF1 was secreted via the stylet by sedentary nematodes settled at the feeding site. Interestingly, the secreted Mi-EFF1 was localized in the nuclei of giant cells. During induction of the giant cells and feeding, sedentary nematodes perforate the cell wall with their stylet. The stylet comes into contact with the plasma membrane without perforation and delivers nematode-secreted proteins in the cytoplasm (Williamson & Hussey, 1996). Our observations suggested that Mi-EFF1 was redirected from the cytoplasm to the nuclei of plant cells. Potential targeting of nematode proteins to the nucleus of plant cells was previously deduced from overexpression assays using candidate effectors from cyst nematodes (Tytgat et al., 2004; Elling et al., 2007; Jones et al., 2009) but has never been observed in planta during infection. Our study shows that the targeting of nematode-secreted proteins to the nucleus is biologically relevant and suggests that RKNs manipulate functions of the plant cell nucleus during the interaction.
The nuclear compartment is a major platform for the activation of plant immunity (Rivas, 2012) and the reprogramming of host cells during compatible interactions (Hok et al., 2010). In connection with this, a growing number of effectors isolated from viruses, bacteria, phytoplasma, oomycetes or fungi have been shown to be targeted at the nucleus of plant cells where they manipulate plant functions to promote infection (see review by Deslandes & Rivas, 2011). In some cases, these effectors directly bind to plant promoters and act as transcription factors (Kay & Bonas, 2009; Scholze & Boch, 2011; Nissan et al., 2012) or modify histone proteins and thereby modify DNA accessibility for transcription factors (Tasset et al., 2010). In other cases, they interact with plant proteins to ultimately regulate plant gene expression (Bernoux et al., 2008; Kloppholz et al., 2011). Mi-EFF1 is a pioneer gene with no predicted homolog in databases and no known functional domain. As a consequence, it is difficult to predict the molecular function of this protein. Further functional studies, such as the identification of a plant target for Mi-EFF1, will help understanding the molecular function of Mi-EFF1.
M.J. received a grant from INRA and the Region Provence-Alpes-Côte d’Azur. P.V. received a grant from Fundação para a Ciência e para a Tecnologia, from Portugal (SFHR/BD/41339/2007).