Transcriptional reprogramming by root knot and migratory nematode infection in rice


Author for correspondence:

Godelieve Gheysen

Tel: +32 92645888



  • Rice is one of the most important staple crops worldwide, but its yield is compromised by different pathogens, including plant-parasitic nematodes. In this study we have characterized specific and general responses of rice (Oryza sativa) roots challenged with two endoparasitic nematodes with very different modes of action.
  • Local transcriptional changes in rice roots upon root knot (Meloidogyne graminicola) and root rot nematode (RRN, Hirschmanniella oryzae) infection were studied at two time points (3 and 7 d after infection, dai), using mRNA-seq.
  • Our results confirm that root knot nematodes (RKNs), which feed as sedentary endoparasites, stimulate metabolic pathways in the root, and enhance nutrient transport towards the induced root gall. The migratory RRNs, on the other hand, induce programmed cell death and oxidative stress, and obstruct the normal metabolic activity of the root. While RRN infection causes up-regulation of biotic stress-related genes early in the infection, the sedentary RKNs suppress the local defense pathways (e.g. salicylic acid and ethylene pathways). Interestingly, hormone pathways mainly involved in plant development were strongly induced (gibberellin) or repressed (cytokinin) at 3 dai.
  • These results uncover previously unrecognized nematode-induced expression profiles related to their specific infection strategy.


Rice is one of the most important crop plants worldwide, and it has proven to be an excellent model system for monocotyledonous plants. Estimates of annual yield losses as a result of plant-parasitic nematodes on this crop range from 10 to 25% worldwide, with the rice root knot nematode (RKN) Meloidogyne graminicola and the root rot nematode (RRN) Hirschmanniella oryzae being the predominant nematode species attacking this plant. M. graminicola is the most important pathogen in Southeast Asian aerobic rice culture, while H. oryzae is mainly found in areas where rice is grown in submerged conditions (Bridge et al., 2005).

Although they are both endoparasitic nematodes, M. graminicola and H. oryzae have very different lifestyles. Migratory nematodes such as the rice RRN H. oryzae penetrate the root, at any zone except the root tip, through a combination of mechanical action and secretion of cell wall-degrading or -modifying enzymes. Inside the root, H. oryzae can move freely in the air channels between the radial lamellae of the parenchyma and feeds upon the cortical cells. A few days after entry, the female lays eggs, which hatch in 4–5 d inside the roots. Under suitable conditions the life cycle is completed in c. 5 wk. Infected cells become necrotic, leading to large root lesions. Roots invaded by this nematode turn yellowish brown and rot, and this root damage causes significant growth retardation to the plant (Bridge et al., 2005). The second-stage juvenile of the rice RKN M. graminicola invades the plant root in the root elongation zone and moves towards the root tip, where it punctures some selected vascular cells with its stylet and injects pharyngeal secretions into it, ultimately leading to the reorganization of these cells into typical feeding structures called giant cells, from which the nematode feeds for the remainder of its sedentary life cycle (Gheysen & Mitchum, 2011). Already at 1 d after infection (dai), swelling of infected root tips can be observed. At 3 dai, hyperplasy and hypertrophy of the cells surrounding the giant cells result in the formation of terminal hook-like galls, a typical characteristic of the rice RKN (Bridge et al., 2005). After three moults, the females lay their eggs inside the galls, while most other RKNs deposit egg masses at the gall surface, and hatched juveniles can reinfect the same or adjacent roots. In well-drained soil at 22–29°C the life cycle of M. graminicola is completed in 19 d. By contrast to migratory nematodes, RKNs hardly cause any necrosis, as they migrate in between rather than through the plant cells by enzymatic softening of the middle lamella of the root cells (Gheysen & Mitchum, 2011).

During the infection process plant-pathogenic nematodes cause major changes in plant gene expression (Gheysen & Mitchum, 2011). Nematode-induced transcriptome changes have been mainly identified by microarray analysis on dicotyledonous plants upon infection with sedentary nematodes (Alkharouf et al., 2006; Ithal et al., 2007a,b; Klink et al., 2007; Szakasits et al., 2009; Barcala et al., 2010). Except for our own work on systemic defense pathways in H. oryzae-infected rice (Kyndt et al., 2012b), no data on the response of plants upon migratory nematode infection are available. The systemic impact on the rice defense system appeared to be quite different between RKN- and RRN-infected plants, exemplifying the different modes of action of these two nematodes (Kyndt et al., 2012b).

While the transcriptome response has been studied in detail for different plant species in a pathogenic interaction with bacterial and fungal pathogens, far less research has been performed on the molecular interaction between plants and nematodes (Gheysen & Mitchum, 2009). The comparative transcriptome analysis of rice upon infection with sedentary or migratory nematodes described in this paper provides important insights into general and specific defense strategies of the monocotyledonous rice plant upon root infection and about genes that are potentially targeted by effectors of nematodes with totally different lifestyles.

Materials and Methods


Oryza sativa L. cv ‘Nipponbare’ (GSOR-100, USDA) was germinated for 6 d at 30°C, transferred to synthetic absorbent polymer (SAP) substrate (Reversat et al., 1999) and further grown at 26°C under a 16 : 8 h light : dark regime. Nematodes were cultured and extracted as described in (Kyndt et al., 2012b). Twelve-day-old plants were inoculated with 250 nematodes per plant for M. graminicola and 400 nematodes per plant for H. oryzae. Control plants were mock-inoculated. One day after inoculation, the plants were transferred to a hydroponic culturing system with Hoagland solution (Reversat et al., 1999) to synchronize the infection process. Infected and control tissues were collected at 3 and 7 dai. In the case of RKN, galls were sampled. Since RKN-induced galls are located at the root tip, uninfected root tips, encompassing the meristem and the elongation zone (c. 0.3 mm long), were sampled as control material. For migratory RRN infection, the whole root tissue without these root tips, was sampled. For each treatment, two or three biological replicates, with each replicate containing a pool of six different plants, were analyzed by mRNA-Seq. For quantitative reverse transcription polymerase chain reaction (qRT-PCR) validation, another two independent biological replicates were taken. RNA was extracted using the Qiagen RNeasy Plant Mini Kit (Qiagen, Venlo, the Netherlands), with an additional sonication step after addition of buffer RLT (Qiagen). RNA integrity was checked using the Agilent BioAnalyzer 2100 (Agilent, Diegem, Belgium).

Library preparation and Illumina GAIIx sequencing

Approx. 2 μg of total RNA of each sample was used for mRNA-Seq library construction according to the manufacturer's recommendations (Illumina, San Diego, CA, USA). The denatured libraries were diluted to a final concentration of 6 pM and loaded on a single read flow cell. Multiplexing was applied to minimize lane effects. After cluster generation, the multiplexed library was sequenced on an Illumina Genome Analyzer II (76 cycles).

Mapping reads to genome data and annotated transcripts

Reads were mapped to the O. sativa ssp. japonica reference genome (build MSU6.0) in two phases using TopHat version 1.3.1 (Trapnell et al., 2009) and Cufflinks, version 1.0.3 (Trapnell et al., 2010). A detailed description of the workflow and settings used in the data analysis is given in Supporting Information, Notes S1.

Identification of novel transcriptionally active regions (nTARs)

The Cufflinks program generates a GTF file, including all transcripts annotated in MSU6 and putative novel transcripts derived from the data. All putative nTARs marked as splice variants of known genes or located within intronic regions were disregarded and the 4684 remaining nTARs were compared by BLASTN (E < 1e–4) with nTARs from our previous analysis on rice root tissue (Kyndt et al., 2012a). A more detailed description of this analysis is given in Notes S1. BLASTx searches were performed against the Swiss-Prot and trEMBL and all predicted rice proteins ( Homologs of the nTARs in rice expressed sequence tags (ESTs) were searched by tBLASTx (E < 1e–4).

Calculation, normalization and profiling of gene expression

Expression was quantified per sample and per annotated or unannotated locus as the sum of all reads mapped to the respective gene exons. Splice variants were treated as a single gene. Expression profiles were assessed using the R-package ‘baySeq’, version 1.5.1. (Hardcastle & Kelly, 2010). To compensate for artificial differences in read distributions, the original library sizes were multiplied by additional normalization factors calculated using the methods described in Robinson & Oshlack (2010), with standard settings as implemented in the edgeR package (version 2.0.3). For all further analyses the expression level of each transcript for each condition was estimated as the average number of reads detected across the biological replicates. The fold change (FC) is calculated as the ratio of the average number of reads + 1 (to avoid 0 values) in the different conditions. Log2-transformed values of FC were used.

Gene Ontology and enrichment analyses

Gene Ontology (GO) analysis and GO enrichment were performed using agriGO (Du et al., 2010). Parametric analysis of gene set enrichment (PAGE) (Kim & Volsky, 2005), based on differential gene expression levels (log2FC), was executed. Benjamini and Hochberg false discovery rate (FDR) correction was performed using the default parameters to adjust the P-value.

In addition, we used MapMan (Thimm et al., 2004) to visualize the expression of genes onto metabolic pathways and the Wilcoxon Signed Rank (WSR-test (with Benjamini and Hochberg correction) to test the statistical significance of differential expression of these pathways.

Validation of mRNA-Seq by qRT-PCR

Based on potential functional importance, 15 genes were selected per nematode and per time point for validation in independent samples by qRT-PCR. After removal of transcripts with unspecific amplification, at least 12 genes remained for validation. Chromosomal location of these transcripts and primer sequences are presented in Table S1. qRT-PCR was performed and analyzed as described in Kyndt et al. (2012a).

Detection of nematode transcripts in infected tissues

The nature of the sample preparation allowed us not only to examine the rice transcriptome, but also simultaneously to capture that of the invading nematodes. Since no reference genomes are available for these nematodes, we used a de novo assembly method employing only those reads that could not be mapped in the reference rice annotation approach (termed ‘unmapped fraction’). First, transcript contigs were assembled using Velvet (v1.1.07, Zerbino & Birney, 2008) with variable k-mer lengths (k = {43,47,53,57}). The contig files from each assembly were merged per nematode using the multi-K method described in Surget-Groba & Montoya-Burgos (2010). These contigs were BLASTed against the rice genome (MSU6.0) and against 454-EST contigs available in our laboratory (M. graminicola, A. Haegeman et al., unpublished; H. oryzae, L. Bauters et al., unpublished) (E < 1e–15). For more details, see Notes S1.


Root tissues were collected at two stages after nematode inoculation: 3 dai (‘early’) and 7 dai. For simplicity, we call the 7 dai data ‘late’, although a complete life cycle of M. graminicola takes 2–3 wk and 5 wk up to 2 months for H. oryzae. A total of 98 671 971 reads were acquired and made publicly available (GSE35843). The short reads were aligned against the whole reference genome sequence of cv Nipponbare (MSU6.0) and in total 49.20% of the sequenced reads could be mapped (Table 1). The total length of mapped reads was c. 3.69 billion bases, representing 9.5-fold the rice genome size and c. 36-fold coverage of the annotated transcriptome. The expression of a total of 42 640 different rice genes was detected in the analyzed tissues.

Table 1. Overview of the obtained sequencing data and mapping of these sequences on to the rice genome
 Total number of sequenced readsTotal number of mappingsNumber of reads with at least one mapping
  1. dai, days after infection.

Uninfected roots 3 dai10 431 41316 289 0114 540 703
Uninfected root tips 3 dai12 370 82983 510 5047 035 305
Uninfected roots 7 dai8 599 73070 245 3774 157 671
Uninfected root tips 7 dai17 737 509147 909 0767 990 824
Migratory nematode-infected roots 3 dai11 680 617201 021 7415 925 094
Migratory nematode-infected roots 7 dai7 082 550151 245 9454 431 705
Root galls 3 dai15 351 251270 299 1977 300 833
Root galls 7 dai15 418 072302 814 2707 184 996
Total98 671 9711 243 335 12148 567 131
Mapping percentage  49.20%
Coverage of the rice genome19.28242.919.49
Coverage of the rice transcriptome73.02920.1535.94

nTARs in infected tissue

TopHat and Cufflinks were used to detect potential exon junctions and unannotated transcribed regions. The locations of the resulting nTARs were matched against O. sativa MSU6 genes. Cufflinks also discovered a series of potential gene extensions (marked with ‘-ext’ in data supplements). All nTARs overlapping known exons were excluded, resulting in 8290 putative nTARs. After excluding overlaps in intronic regions, 4684 putative nTARs remained (Table S2). A total of 2070 sequences showed homology to an EST or cDNA record for ssp. japonica, thus supporting tentative transcripts not included in the MSU6.0 genome. A total of 2491 nTARs have a significant BLAST hit (BLASTn E < 1e–04) with nTARs detected in our previous transcriptome data of healthy rice root tissues (Kyndt et al., 2012a). Hence, an extra set of 2193 nTARs was detected in rice roots challenged by nematode infection.

To predict the function of these 4684 putative nTARs, different database searches were performed. First, the nTAR sequences were BLASTX-ed against all rice proteins and 1099 nTARs resulted in a significant transcript hit (Table S2; E < 1e–4); hence these nTARs are potential paralogs of previously annotated loci. A SwissProt/trEMBL search was executed to detect known signatures and, for 2062 nTARs, a hit was found with a known protein family (Table S2) (E < 1e–4). For 901 nTARs, a functional annotation could be predicted based on similarity with known genes from O. sativa or other plants. Based on this homology, putatively coding regions were found for protein kinases, transcription factors, disease or drug resistance proteins, cyclins, auxin response factors and lipoxygenases.

Genes responding to infection by RKNs in rice

Different comparative gene expression analyses were performed for nematode-infected tissues and their respective healthy control. The different comparisons that were made, the number of biological replicates per condition and the number of statistically significant differentially expressed genes (DEGs) that were identified are shown in Table 2.

Table 2. Comparative gene expression analyses executed on mRNA-Seq data obtained in the current study, showing the number of biological replicates, differentially expressed genes (DEGs, listed in Tables S4–S6) and the included novel transcriptionally active regions (nTARs)
 Biological replicates per treatmentNo. of DEGsNo. of DEGs up-regulated No. of DEGs down-regulated
  1. dai, days after infection; RKN, root knot nematode.

  2. a

    FDR < 0.1

RKN responseEarly (3 dai galls vs 3 dai healthy root tips)3156 (16 nTARs)7680
Late (7 dai galls vs 7 dai healthy root tips)3991 (56 nTARs)93556
General (galls vs healthy root tips)6510 (43 nTARs)374136
RKN-specific response6382 (63 nTARs)269113
RRN responseEarly (3 dai roots vs 3 dai healthy roots)a295 (12 nTARs)8015
Late (7 dai roots vs 7 dai healthy roots)a2111 (26 nTARs)7239
General (infected roots vs healthy roots)4114 (12 nTARs)7044
Root rot nematode-specific response5578 (144 nTARs)5780
General nematode responseAll infected root tissues vs healthy root tissues1094 (26 nTARs)6924

Transcriptome changes in young gall tissue (3 dai)

At 3 dai, the RKNs have penetrated the root and moved to the tip, where they induce root galls. Galls generally contain three to four J2 or J3 stage nematodes (Kyndt et al., 2012b). Gene set enrichment analysis on relative expression levels (log2FC) of all transcripts in the galls vs noninfected root tips (Fig. 1a, Table S3) showed that genes involved in ‘metabolic process’ were up-regulated at 3 dai, while GO terms ‘plant-type cell wall organization or biogenesis’, ‘transmembrane transport’ and ‘photosynthesis’ were down-regulated in the infected tissue. Pathway mapping with MapMan showed a significant up-regulation of the cell wall-degrading beta-1,3-glucan hydrolase gene family and down-regulation of biotic stress-related genes in 3 dai galls (Fig. 2). For example, a general down-regulation of the phenylpropanoid pathway was detected (results not shown).

Figure 1.

Results of parametric analysis of gene set enrichment of transcriptome data of root knot nematode (RKN; Meloidogyne graminicola)- or migratory root rot nematode (Hirschmanniella oryzae)-infected rice (Oryza sativa) roots at 3 (early) or 7 d (late) after infection (dai). (a) 3 dai with RKNs; (b) 7 dai with RKNs; (c) general response upon RKN infection; (d) 3 dai with root rot nematodes; (e) 7 dai with root rot nematodes; (f) general response upon root rot nematode infection. Z-scores of all secondary level Gene Ontology (GO) terms are shown. Dark gray bars, GO terms that are up-regulated in the infected tissue vs the corresponding control; light gray bars, GO terms that are down-regulated in the infected tissue vs the corresponding control.

Figure 2.

MapMan visualization of biotic stress-related genes, and genes involved in auxin and cytokinin pathways. The visualization shows the observed differential expression patterns, based on the log2 fold changes of mRNA levels, in root knot (Meloidogyne graminicola)- or migratory root rot nematode (Hirschmanniella oryzae)-infected rice (Oryza sativa) roots at 3 (early) or 7 d (late) after infection (dai). Red dots, gene is up-regulated in infected tissue vs the corresponding healthy control; blue, down-regulation.

One hundred and fifty-six transcripts were found to be significantly differentially expressed (FDR < 0.05) (Tables 2, S4). Twelve of them were independently validated by qRT-PCR, and the expression pattern was confirmed in all cases (Table 3).

Table 3. Differential expression patterns (log2FC) of a selected set of rice genes in root knot nematode (RKN)- or root rot nematode (RRN)-infected root tissues at 3 and 7 d after infection (dai), as detected by quantitative reverse transcription polymerase chain reaction (qRT-PCR) and mRNA-Seq data (down-regulated genes, green; up-regulated genes, red)Thumbnail image of

Among the strongest down-regulated genes are three thionin genes, a nodulin MtN3 family protein, a lipid phosphatase protein, a MYB family transcription factor and a putative dirigent gene, most probably involved in lignan biosynthesis (Davin & Lewis, 2000). Among the genes that are most strongly up-regulated in young gall tissue, there were three genes of the lipid transfer protein (LTP)-gene family, the gibberellin (GA) receptor GID1L2, a harpin-induced protein 1 domain-containing protein, an auxin response factor, isochorismatase, and many different protein kinases and chitinases.

Transcriptome changes in mature gall tissue (7 dai)

At 7 dai, the galls have matured, and giant cells within the gall tissue are fully developed (Engler et al., 1999). Nematodes developing within these galls are now at the J3 or J4 stage. In 7 dai galls, genes involved in ‘transcription’, ‘post-translational protein modification’, ‘peptide transport’, and ‘regulation of nitrogen compound metabolic process’ were up-regulated (Fig. 1b, Table S3). Genes involved in ‘translation’, ‘chromatin organization’, ‘ion transmembrane transport’, and ‘plant-type cell wall organization’ were generally expressed to a lesser degree than in uninfected root tips. MapMan pathway mapping showed a significant modification of the ATP-synthesis pathway, light reaction, Calvin cycle, cell wall modification, and minor CHO metabolism–trehalose in 7 dai galls vs noninfected root tips. While down-regulated at 3 dai, many genes involved in phenylpropanoid and jasmonate (JA) biosynthesis pathways were up-regulated in 7 dai gall tissue (results not shown).

At an FDR cutoff of 0.05, 991 DEGs were found in 7 dai galls vs healthy root tips (Tables 2, S4). Twelve of them were independently analyzed by qRT-PCR and the trend of expression was confirmed in all cases (Table 3).

The strongest up-regulation was found for genes encoding, for instance, a putative abscisic stress-ripening protein, OsLTPL116, OsLTPL124, GA receptor GID1L2, PR-genes chitinase CHIT14, thaumatin and defensin DEF8, and caffeoyl-CoA O-methyltransferase, a key enzyme in the biosynthesis of lignin cell wall precursors. Among the down-regulated genes there are eight genes belonging to the thionin gene family, two dirigent genes and cellulose synthase CSLF3.

General response of rice root tips upon RKN infection

We also compared the gene expression level in all sampled gall tissues (at both time points) with the six uninfected root tip samples to look at consistent trends in the root gall transcriptome, regardless of the time point after infection (Fig. 1c, Table S3).

Five hundred and ten rice loci were detected to be consistently and significantly differentially expressed. For instance, a consistent twofold (log2FC) up-regulation of OsPR10 (LOC_Os03g18850) was detected in the 3 and 7 dai galls vs the uninfected rice root tips. We also detected a significant and consistent activation of genes involved in gibberellin metabolism, with the two key oxidase enzymes that catalyze the penultimate steps in GA biosynthesis, GA 20-oxidase 1 and 2, and the receptor GID1L2 strongly induced in gall tissue at both time points (Fig. 3).

Figure 3.

MapMan visualization of the gibberellic acid biosynthesis pathway. The visualization shows the observed differential expression patterns, based on the log2 fold changes of mRNA levels, in root knot (Meloidogyne graminicola)-infected rice (Oryza sativa) galls vs healthy root tips. Red dots, gene is up-regulated in infected tissue vs the corresponding healthy control; blue, down-regulation.

Genes responding to infection by migratory RRNs in rice

Transcriptome changes early upon RRN infection

At 3 dai with 400 H. oryzae the root system generally contains 20–100 migratory nematodes (Kyndt et al., 2012b) and some cells within the roots are necrotic. Comparing the transcriptome of infected root tissue with noninfected roots at 3 dai (Table S3, Fig. 1d) revealed that genes involved in ‘electron transport’, ‘response to stress’, ‘post-translational protein modification’, ‘plant-type cell wall organization or biogenesis’, and ‘programmed cell death’ were induced upon infection. On the other hand, genes involved in ‘photosynthesis’, and ‘generation of precursor metabolites and energy’ are down-regulated at 3 dai after RRN infection. MapMan mapping showed a significant change in the phenylpropanoid pathway, as well as flavonoids, Calvin cycle, hemicellulose synthesis, lipid degradation, secondary metabolism-simple phenols, and cell wall degradation. Receptor kinases, the ethylene (ET) and JA metabolism, protein degradation and post-translational modification were significantly affected upon migratory nematode infection. In the JA metabolism, different paralogs of the biosynthetic enzymes allene oxide synthase and lipoxygenase are induced 3 d after RRN inoculation.

Only 25 genes were found to be significantly differentially expressed at a cutoff of FDR < 0.05. As this dataset contained fewer biological replicates (Table 2) and hence less statistical power, we decided to set the FDR cutoff at < 0.1. At this cutoff, 95 DEGs were found (Tables 2, S4). Twelve genes were validated by qRT-PCR and the trend of expression was confirmed in all but one case (Table 3).

OsLTPL124 (LOC_Os04g52260) was 5.3-fold (Log2FC) down-regulated at 3 d after migratory nematode infection. A gene encoding a CLAVATA1 (CLV1) protein kinase (LOC_Os05g51740) was 1.5-fold (log2FC) down-regulated in the infected root tissue. Transcription factors OsWRKY62, OsWRKY70 and OsWRKY11 were strongly up-regulated at 3 dai after migratory nematode infection. An induction of oxidative stress is exemplified by the up-regulation of three peroxidase precursors and two glutathione S-transferases.

Transcriptome changes late upon RRN infection

Seven days after migratory RRN infection, the roots show large necrotic lesions, and nematodes have laid eggs within the root tissue. When comparing the transcriptome of infected vs uninfected root tissues, (Table S3, Fig. 1e) we detected an up-regulation of ‘primary metabolic process’, ‘protein metabolic process’ and ‘defense response’, while ‘photosynthesis’ and ‘porphyrin metabolic processes’ were down-regulated in the infected roots. MapMan mapping showed a general down-regulation of genes involved in the Calvin cycle and light reactions, while mitochondrial electron transport was up-regulated at 7 d after migratory nematode infection. The JA pathway and phenylpropanoid pathways did not respond as strongly as at 3 dai (results not shown).

At a cutoff of FDR < 0.05, 38 DEGs were found at 7 d after migratory nematode infection. Considering that only two biological replicates per treatment were analyzed for these tissues (Table 2), the statistical analysis was also performed with less stringent conditions (FDR < 0.1), resulting in 111 DEGs (Tables 2, S4). The trend of expression was confirmed by qRT-PCR in 11 out of 14 cases (Table 3).

Among the strongest down-regulated genes, for instance, are those coding for phosphoribulokinase, glyceraldehyde-3-phosphate dehydrogenase and fructose-bisphospate aldolase, all involved in primary plant metabolism (glycolysis and Calvin cycle). Transcripts coding for OsWRKY62 and OsWRKY77 are highly up-regulated upon infection, as well as a nucleotide-binding site–leucine-rich repeat (NBS-LRR) disease resistance protein (LOC_Os09g34150) and patatin (LOC_Os08g28880).

General response of rice upon RRN infection

The gene expression profile in all RRN-infected rice root samples was compared with the transcriptome of uninfected root samples to detect consistent transcriptome changes regardless of the time point after infection (Table S3, Fig. 1f).

Differential expression analysis detected 114 loci to be significantly differentially expressed at both time points (FDR < 0.05) upon RRN infection in rice roots (Table 2). Three genes involved in carbohydrate metabolism, coding for fructose-1,6-bisphosphatase (LOC_Os03g16050), glyceraldehyde 3-phosphate dehydrogenase (LOC_Os04g38600) and fructose-bisphospate aldolase (LOC_Os11g07020), are strongly and consistently down-regulated in the infected roots compared with the uninfected root tissue. A significant differential expression was found for the WRKY family transcription factors and DUF26 and wall-associated receptor kinases. Among them, for instance, a strong up-regulation was detected in the transcripts coding for OsWRKY62 (LOC_Os09g25070) and OsWRKY59 (LOC_Os01g51690) in the infected roots. Further, OsSub13, a subtilisin-like serine proteinase belonging to a class of pathogenesis-related proteins (Jorda & Vera, 2000), and a protein-disulfide isomerase (PDI), involved in the biosynthesis of a group of plant defense peptides called cyclotides (Jennings et al., 2001), were strongly up-regulated.

General response of rice upon infection with nematodes

Differential gene expression analysis identified 94 loci to be significantly and consistently differentially expressed in rice root tissue upon infection with either migratory RRNs or RKNs (FDR < 0.05; Table S4). Among these genes that are generally responding upon root challenge with nematodes is OsWRKY62, which is 3.4-fold up-regulated in infected vs noninfected tissue. Peng et al. (2008) showed that overexpression of OsWRKY62 suppresses the activation of defense-related genes, implying that this gene functions as a negative regulator of innate immunity in rice. Seven of these general nematode-responsive DEGs were arbitrarily chosen to independently analyze their expression in infected and uninfected tissue at 3 dai by qRT-PCR. The trend of expression was confirmed in all but one case (Table 3).

Comparison between the response of rice upon RKN and RRN infection

A comparison between the biotic stress responses in rice root tissue after infection with RKNs vs RRNs provides insights into mechanisms that are specifically related to the different infection strategy of each nematode. In Table 4, an overview of the general trends observed at 3 dai with RKNs or RRNs is shown.

Table 4. Overview and comparison of the local physiological transcriptional response in rice root tissue upon infection with either Meloidogyne graminicola or Hirschmanniella oryzae
 M. graminicola-infected gall tissue (3 dai)H. oryzae-infected root tissue (3 dai)
  1. dai, days after infection; BR, brassinosteroid.

  2. ↑, generally induced in nematode-infected tissue; ↓, generally repressed in nematode-infected tissue; ↘, slightly repressed in nematode-infected tissue; ≈, generally unchanged in nematode-infected tissue; ↓ and ↑, both strongly induced and strongly repressed genes detected in nematode-infected tissue.

Nutrient transport
Antioxidant activity
Plant cell wall organization and biogenesis
Biotic stress
Hormone pathways mainly involved in plant defense
Salicylic acid pathway
Jasmonate pathway↓ and ↑
Ethylene pathway
Hormone pathways mainly involved in plant development
Auxin pathway↓ and ↑
Cytokinin pathway
GA pathway
BR pathway

To detect specific RKN-responsive genes, the RRN-infected tissue was also used as control tissue next to the uninfected root tissues, and vice versa.

In RRN-infected roots, biotic stress-related genes are strongly up-regulated at 3 dai, while this response is attenuated at the later time point after infection (Fig. 2). For RKN-infected galls, many genes involved in biotic stress response are down-regulated at 3 dai, but up-regulated at 7 dai. This up-regulation at 7 dai, however, is not as strong as is seen in the 3 dai RRN-infected roots.

Differential gene expression analysis identified 382 loci that are only significantly differentially expressed upon RKN infection, and not upon H. oryzae infection (FDR < 0.05; Table S5). Among them, for instance, the AP2-like transcription factor PLETHORA 2 (LOC_Os07g03250) is highly up-regulated in gall tissue. Two transcripts coding for WKRY-transcription factors (OsWRKY34 and OsWRKY36), two genes annotated as growth-regulating factors (LOC_Os02g47280 and LOC_Os04g51190) and one coding for osmotin (LOC_Os12g38150) are specifically induced in gall tissue. On the other hand, five loci coding for dirigent proteins (LOC_Os01g25030, LOC_Os07g44250, LOC_Os07g44280, LOC_Os07g44450 and LOC_Os10g18760), major regulators of lignan biosynthesis (Davin & Lewis, 2000), are specifically down-regulated in the gall tissue.

Five hundred and seventy-eight loci were detected to be significantly differentially expressed in migratory nematode-infected root tissue, but not affected in gall tissue (FDR < 0.05; Table S6). For instance, seven transcription factors of the WRKY-superfamily were up-regulated in H. oryzae-infected roots, WKRY11, 45, 51, 59, 66, 77 and an unclassified WRKY gene. Also, nine genes annotated as ‘disease-resistance proteins’ were up-regulated upon H. oryzae infection but not upon RKN infection. Three loci coding for flavonol synthase genes and one for flavonol-3-O-glycoside-7-O-glucosyltransferase 1 were found to be specifically more transcribed in RRN-infected root tissue, as well as four loci coding for peroxidases.

Detection of nematode transcripts in infected root tissue

As 51.8% of sequenced reads (Table 1) could not be mapped to the rice genome (the ‘unmapped fraction’), we evaluated whether some of these originated from the infecting nematodes. After de novo assembly, contigs were BLASTed against the rice genome (MSU6.0) again and against EST data from the nematodes used in this study.

In all, 3856 of the contigs assembled from gall tissue matched EST data from M. graminicola and another 137 matched EST data from H. oryzae. Among the contigs assembled from migratory nematode-infected roots, 3744 matched ESTs from H. oryzae, and another 1291 matched ESTs from M. graminicola. (Table 5). Hits with ESTs from the other nematode are most likely highly conserved genes that are present in both nematodes but which might not have been picked up during EST-sequencing of both nematodes. A SwissProt/trEMBL search was performed on the contigs with a nematode hit, and for 88% (M. graminicola) and 57% (H. oryzae) of them, a potential function can be predicted on this basis. Table 5 shows that the majority of these transcripts potentially code for housekeeping genes, but a few putative nematode secreted proteins were also found.

Table 5. Overview of results obtained from the unmapped fraction of mRNA-Seq reads
 Contigs obtained from unmapped fraction of
RKN-infected root tipsRoot rot nematode-infected roots
  1. EST, expressed sequence tag; OS, original species; RKN, root knot nematode.

Total number of contigs in the unmapped fraction 10 1059376
No. of contigs with BLAST-hit in
 Oryza sativa MSU6.02260779
EST data of Meloidogyne graminicola38563744
EST data of Hirschmanniella oryzae1371291
Hits with unknown function47154
Nematode housekeeping genes
Cuticle collagen1218
Ribosomal proteins285165
Eukaryotic translation initiation factors363
Elongation factors6917
Heat shock proteins4213
Predicted annotation of a selection of putative nematode-secreted proteins 32 kDa beta-galactoside-binding lectin OS = Haemonchus contortus32 kDa beta-galactoside-binding lectin OS = Haemonchus contortus
Peroxiredoxin OS = Ascaris suumPeroxiredoxin OS = Ascaris suum
Superoxide dismutase (Cu-Zn) OS = Brugia pahangiAncylostoma secreted protein OS = Ancylostoma caninum
Polyprotein ABA-1 (fragment) OS = Ascaris suumPolyprotein ABA-1 (fragment) OS = Ascaris suum
Putative cathepsin L protease OS = Meloidogyne incognitaGlucuronoxylanase xynC OS = Bacillus subtilis
Fatty-acid and retinol-binding protein 1 OS = Onchocerca gutturosaPectate lyase OS = Meloidogyne javanica
Glutathione peroxidase OS = Schistosoma mansoniBeta-1,4-endoglucanase OS = Pratylenchus penetrans
Programmed cell death protein 6 OS = Mus musculusBeta-glucanase OS = Rhodothermus marinus
 Endoglucanase celA OS = Streptomyces lividans
 Endoglucanase OS = Clostridium saccharobutylicum
 Endoglucanase Z OS = Dickeya dadantii
 Expansin B2 (fragment) OS = Globodera rostochiensis
 Expansin B1 protein OS = Globodera rostochiensis
 Expansin-like protein OS = Bursaphelenchus xylophilus


Using mRNA-Seq, we studied transcriptome changes induced in a compatible interaction between plant-parasitic nematodes and O. sativa. We acquired almost 100 million reads from infected and healthy root tissues, thereby obtaining count data for more than 40 000 rice transcripts. A good consistency (91%) between expression levels identified by mRNA-seq and qRT-PCR analyses was established (Table 3). The comparison between the transcriptome of the plant upon infection with two types of nematodes, exhibiting distinct feeding behaviors, provides an unprecedented insight into the compatible interaction between nematodes and the plant root.

Plants generally respond to nematode infection by differential expression of genes involved in stress and defense responses, cell wall alteration, metabolism and nutrient allocation, signal transduction and phytohormone action (Li et al., 2008; Escobar et al., 2011). In accordance with these previous observations, genes involved in those categories were differentially expressed in the current study. However, when comparing the response of rice roots upon infection with migratory and sedentary endoparasitic nematodes, many discrepancies arise. General trends, as deduced from the data presented in this paper, are shown in Table 4. Some of the more remarkable differences and similarities will be discussed in the following sections.

For the RKN analysis, whole gall tissues were used and thus gene expression differences reflect a combination of changes occurring in giant cells and surrounding gall tissues. The study of Barcala et al. (2010) showed that, although some genes have similar regulation in both galls and giant cells, the majority had different expression patterns.

Metabolic activity and nutrient transport

Feeding sites of sedentary endoparasitic nematodes, such as Meloidogyne spp. and cyst nematodes, are the only source of nutrients for these root parasites throughout their lives (Jammes et al., 2005; Szakasits et al., 2009). Hofmann et al. (2010) showed that cyst nematodes cause a major reprogramming of the primary metabolism in local and systemic tissues upon cyst nematode infection. The current study confirmed that phenomenon in root galls of rice roots (Fig. 1, Table S3). This observation was strongest at 3 dai, when genes involved in protein and sucrose biosynthesis were highly induced, and is in concordance with the view that the RKNs stimulate the cells within the gall to be more active and produce nutrients to the benefit of the nematode. An activation of the trehalose metabolism was observed in both galls and syncytia (Hofmann et al., 2010). Trehalose has multiple functions in plants, not only in carbohydrate storage and metabolism, but also as a stress protectant, and as a metabolic signaling molecule involved in many plant–pathogen interactions (Fernandez et al., 2010) and cell wall modification (Bae et al., 2005).

The nematode manipulates the plant to redirect nutrients to this new nutrient sink, mainly during the initial phase of giant cell formation, while at later stages, intracellular and transmembrane transport of lipids and proteins is also strongly reprogrammed (Hofmann et al., 2010). For instance, in gall tissue there was a significant up-regulation of transcripts encoding the nuclear export protein exportin 1, amino acid and peptide transporters, transmembrane nucleobase-ascorbate transporters, and plant lipid transfer proteins (LTPs, Tables S3 and S4). LTPs are a family of peptides that facilitate lipid exchange between cell membranes (D'Angelo et al., 2008), but for which a possible defense role in plants has also been suggested (Kim et al., 2006).

In the primary metabolism of the gall, a remarkable phenomenon is observed, which might seem counterintuitive: photosynthesis-related genes are highly up-regulated in 7 dai galls (Table S3), while roots are generally seen as photosynthetically inactive tissue. However, under light induction, roots can develop chloroplasts (Flores et al., 1993) and we have recently demonstrated (Kyndt et al., 2012a) that photosynthesis-related genes are activated in the root tissue in a hydroponic growth system even with minor light influence. Nevertheless, when infected with nematodes, only RKN-infected tissues showed this photosynthetic capacity, while RRN-infested roots revealed a strong suppression of photosynthesis (Table S3). This observation is in accordance with the observations of Szakasits et al. (2009) which highlighted the up-regulation of chloroplast genes in syncytia of cyst nematodes, but induction of photosynthesis has not been described before in galls.

In RRN-infected plants, on the other hand, the metabolic activity of the root is suppressed, while many transcripts attributed to cell death are induced, in line with the root necrosis caused by feeding of the migratory nematode.

Cell wall biosynthesis, degradation and modification

The plant cell wall, composed of cellulose microfibrils, hemicelluloses and pectin, represents a structural barrier to infection by a wide range of plant pathogens. Endoparasitic nematodes have evolved sophisticated mechanisms to breach this barrier and secrete a battery of cell wall-modifying enzymes into the plant root (Gheysen & Mitchum, 2009; Davis et al., 2011). The processes that give rise to feeding site formation also involve recruitment of plant genes encoding cell wall-modifying proteins. Our study revealed that the biosynthesis of new cell walls is up-regulated in migratory nematode-infected roots, most probably as a plant defense strategy. Conversely, cell wall organization and biosynthesis are repressed by RKN infection (Table S3).

While cell wall biosynthesis is reduced, plant genes coding for cell wall-degrading enzymes are activated in root galls (reviewed in Gheysen & Mitchum, 2009). In rice root galls, endo-β-1,4-glucanases belonging to the glycosyl hydrolase family 17 were strongly induced. By contrast, there was a down-regulation of xyloglucan endotransglycosylases (XETs), enzymes that show a specific activity on xyloglucans, the major component of hemicellulose. Cell expansion, which is required for the development of nematode feeding sites, is generally associated with structural changes in cell wall xyloglucans, which play an important role in cell wall elasticity and rigidity (Scheller & Ulvskov, 2010). Similar to our observations, transcriptome data from cyst nematode-infected soybean and Arabidopsis roots revealed suppression of XETs in the locally infected tissue (Puthoff et al., 2003; Ithal et al., 2007a; Barcala et al., 2010), while they have been described to be locally up-regulated in isolated syncytia (Ithal et al., 2007b) and giant cells (Barcala et al., 2010). As previously stated, the galls analyzed in this study were composed of giant cells and their surrounding cells, and we hypothesize that suppression of cell wall biosynthesis and XETs occurs mainly in these surrounding cells. Differences in cell wall composition of monocots vs dicots and the fact that M. graminicola-induced galls are located at the root tip, and hence must have some effect on cell multiplication in the root meristem, probably also play a role here. To further elucidate the importance and localization of these observations, studies on isolated giant cells from rice roots are needed.

Plant hormone pathways and defense

The expression of plant defense proteins and changes in plant development are mainly regulated by phytohormones. Nematode infection in plants results in a change in hormone homeostasis, through an interplay between active manipulation by nematode effectors secreted in the plant tissue, to promote plant susceptibility, and defense responses activated by the plant to reduce pathogen infection (Goverse & Bird, 2011).

In contrast to galls, the migratory nematode-infected roots showed an induction of many genes known to be responsive to external stimuli such as general, biotic and abiotic stress factors (Figs 1, 2). Microarray data from Arabidopsis and soybean have shown before that the plant defense response is up-regulated in cyst nematode-infected tissues (Alkharouf et al., 2006; Ithal et al., 2007a), but this and other studies (Jammes et al., 2005; Barcala et al., 2010) show instead an early down-regulation of defense pathways upon RKN infection.

Many genes involved in the phenylpropanoid pathway, responsible for the biosynthesis of different metabolites, such as lignin precursors, flavonoids and hydroxycinnamic acid esters and salicylic acid (SA; Boudet, 2000), are strongly suppressed in 3 dai galls in rice (this study) and Arabidopsis (Barcala et al., 2010), whereas they were reported to be induced at early time points after migratory nematode infection (this study) and cyst nematode infection in soybean (Ithal et al., 2007a). The fact that this response is attenuated in RKN-infected tissue further demonstrates the subtleness and manipulative power of the RKNs.

The JA and ET pathways are known to play synergistic roles in plant innate immunity (Pieterse et al., 2009). In the case of migratory nematode-infected roots, a trend of coregulation of genes involved in these pathways was also observed, with them generally being up-regulated, similar to their systemic induction upon RRN infection (Kyndt et al., 2012b). Although ET has been suggested to be critical for syncytium formation during cyst-nematode infections in Arabidopsis (Goverse et al., 2000), RKNs seem to suppress ET pathways, both locally (Barcala et al., 2010; Nahar et al., 2011; and this study) and systemically (Kyndt et al., 2012b). Ithal et al. (2007b), detected a strong and consistent suppression of the JA pathway in isolated syncytia after cyst nematode infection in soybean, but in RKN galls on rice, no such trend was observed (Nahar et al., 2011; this study). This is an interesting observation in view of the fact that the JA pathway, modulated by ET, is a key player in systemically induced defence against RKNs in rice (Nahar et al., 2011). From these results, we might deduce that the RKN does not specifically target the JA pathway, but rather focuses on down-regulation of the SA and ET pathways. This may explain why activating the JA pathway is the most effective strategy when trying to induce systemic defense against RKNs (Cooper et al., 2005; Nahar et al., 2011).

In galls, GA biosynthesis genes and the GID1L2 GA receptor are strongly induced (Fig. 3). GA plays an important role in stimulating cell division and elongation (Richards et al., 2001). Similarly, Bar-Or et al. (2005) and Klink et al. (2007) noticed the up-regulation of GA biosynthesis genes in tomato galls and syncytia on soybean.

Brassinosteroids (BRs) play important roles in physiological and developmental processes, and their impact on plant defense has also been demonstrated (Belkhadir et al., 2012). De Vleesschauwer et al. (2012) showed that the BR pathway is involved in rice root susceptibility to Pythium graminicola, by down-regulating the SA pathway. Similarly, BRs suppress rice defense against RKNs through antagonism to the JA pathway (Nahar et al., 2012). In light of this knowledge, it is interesting to notice that the BR pathway is generally slightly induced in nematode-infected tissues, and even significantly at 7 dai in RKN gall tissue (Table S4). The BR biosynthesis gene OsDWARF (LOC_Os03g40540) and different paralogs coding for the BRASSINOSTEROID INSENSITIVE 1-associated receptor kinase 1 (BAK1) are more highly expressed in nematode-infected rice roots. BAK1 is involved in BR signaling, together with BRASSINOSTEROID INSENSITIVE 1 (BRI1), for which the different homologs in the rice genome are also generally induced in nematode-infected vs uninfected root samples. An activation of the BR pathway might be important for the nematode to overcome root defense.

The auxin pathway is responsible for root initiation, development, and lateral root formation (De Smet et al., 2010). The transcript encoding indole-3-acetic acid-amido synthetase (OsGH3.1), which conjugates IAA to amino acids, was found to be strongly and consistently suppressed in rice roots infected with either RKNs or RRNs. This potentially results in higher auxin concentrations in the infected roots, which can lead to improved cell division and expansion and cell wall loosening, but probably also plays a part in plant defense suppression through its antagonism with the SA pathway (Domingo et al., 2009). Auxin manipulation is known to be an important process during initiation and development of the feeding sites of sedentary plant-parasitic nematodes (Grunewald et al., 2009b). Our observations indicate that not only sedentary nematodes, but also migratory nematodes, manipulate auxin homeostasis in the root (Fig. 2). OsYUCCA1, the only rice gene for which the role in IAA biosynthesis has been described in detail (LOC_Os01g46290; Yamamoto et al., 2007), is strongly up-regulated in root galls at 3 dai and also slightly at 7 dai, corroborating previous evidence of induced auxin concentrations in nematode feeding sites (Karczmarek et al., 2004; Grunewald et al., 2009a).

Auxin and cytokinin (CK) have contrasting roles in root development (Dello Ioio et al., 2007). CKs play an inhibitory role in root meristem size, root elongation and lateral root formation in Arabidopsis (Chapman & Estelle, 2009). In migratory nematode infected roots, the CK pathway was not strongly influenced. But in contrast to the enhanced CK amounts generally observed in leaves early upon infection with biotrophic fungi (Walters & McRoberts, 2006), a strong suppression of the CK pathway was detected in 3 dai root galls (Fig. 2). At 7 dai, however, the CK pathway is induced in galls, confirming previous observations of Lohar et al. (Lohar et al., 2004), who proved the requirement of CK to establish mature galls (7 dai) in Lotus japonicus. Interference with CK concentrations and signaling is probably important for the enlargement of the root tip during early root gall formation. Next to swelling of the root meristem, CK suppression and auxin induction will probably also lead to the induction of lateral root formation, a phenomenon which is frequently observed in the vicinity of galls and syncytia (Goverse et al., 2000; Karczmarek et al., 2004). Next to their developmental function, CK concentrations have also been shown to repress transporters of macronutriens in rice (Hirose et al., 2007), and reducing CK concentrations might also be necessary for the conversion of galls into nutrient sinks. Further research into the exact role of and interplay between the auxin, CK and other hormone pathways in root gall development and plant defense against nematodes will certainly prove to be worthwhile.


We acknowledge the financial and infrastructural support of Ghent University (‘Bioinformatics: from nucleotides to networks’, BOF_01GA0805, and Stevin Supercomputer Infrastructure). Nematodes were collected and kindly provided by Prof. Dirk De Waele (Catholic University Leuven, Belgium) and Zin Thu Zar Maung (Plant Protection Division, Yangon, Myanmar). We thank Jean-Pierre Renard, Sarah De Keulenaer and Lien De Smet for excellent technical assistance. Tina Kyndt, Annelies Haegeman and Tim De Meyer (in part) are supported by a FWO postdoctoral fellowship. Simon Denil is supported by an IWT doctoral grant.