Nematicides have been commonly used to control PPNs in agriculture, but some nematicides such as methyl bromide and aldicarb are currently banned from use in many countries due to their negative effects on the environment and human health . It has therefore become important to understand the molecular mechanisms of plant immunity against PPNs to provide a foundation for the development of new environmentally friendly and effective control methods. In general, the plant immune system is represented by two inter-related tiers . The first is governed by cell surface-localized pattern recognition receptors that perceive pathogen associated molecular patterns , leading to pattern triggered immunity . Successful pathogens secrete effector molecules into the apoplast ordirectly into plant cells, which interfere with PTI, resulting in successful infection. Resistant plants recognize cell-invading effectors through recognition by intracellular nucleotide-binding domain leucine-rich repeat -type immune receptors, which are encoded by resistance genes. Similar mechanisms are also conserved in plant-PPN interactions. For example, the well conserved nematode pheromone ascaroside has been identified as a PAMP , but the corresponding PRR has not yet been found. PPN genome sequence analyses identified a number of candidate virulence effectors , and a handful of NLR protein-encoding R genes involved in PPN recognition have been well-studied and characterized, including tomato Mi-1.2, Mi-9, and Hero-A; potato Gpa2 and Gro1-4; pepper CaMi; and prune Ma . Mi-1.2, Mi-9, CaMi, and Ma confer resistance against RKNs, black flower buckets whereas Hero-A, Gpa2, and Gro1-4 provide resistance against CNs.
Although the PPN perception mechanism is somewhat clearer at the molecular level, it is still largely unknown what kind of downstream responses are induced after the recognition of avirulent PPNs. It is also unclear what kind of host responses are induced after infection with virulent PPNs, leading to susceptibility and infestation. There are several difficulties in working on plant responses against PPNs. First and foremost, most model plants, such as Arabidopsis, are susceptible to PPNs and therefore cannot be used to study the cascade of responses leading to resistance. Second, PPNs migrate long distances inside roots, inducing complicated responses as they go, triggered by mechanical stress and wounding, among others, making it difficult to pinpoint the key genes involved in resistance or susceptibility by transcriptome analyses. Some studies have used comparative transcriptomics using susceptible and resistant plants infected with a single genotype of nematode . However, it is difficult to rule out the possibility that differences in gene expression were due to resistance or susceptibility rather than to differences in the genetic backgrounds of host plants. Lastly, susceptible responses such as the formation of feeding sites are induced in specific cells targeted by PPNs, and defense responses are likely to be induced in the cells directly impacted by PPN activity. Thus, cells responding to PPNs are rather limited, making the analysis technically challenging. Here we have introduced Solanum torvum Sw “Torvum Vigor” to overcome these problems. S. torvum has been widely used as a rootstock of eggplant to prevent disease caused by PPNs, as well as the soil-borne pathogens Ralstonia solanacearum, Verticillium dahliae, and Fusarium oxysporum f. melongenae n. f. . S. torvum Sw “Torvum Vigor” is resistant to Meloidogyne arenaria pathotype A2-O , but susceptible to M. arenaria pathotype A2-J .
By using S. torvum and avirulent or virulent isolates, we established an in vitro infection system and performed comparative transcriptome analyses to identify genes whose expressions were associated with either resistance or susceptibility by carefully collecting only root tips attacked by RKNs, which allowed us to detect gene expression only in cells directly affected by nematodes. In addition, observation of infected root tip morphology suggests that the success or failure of the immune system against PPNs is determined within a few days of invasion. Thus, we decided to focus on the transcriptional changes that occurred in the very early stages of infection, which has not been studied in previous transcriptomic analyses . Comparative clustering analyses of gene expression identified a large number of novel genes, especially those involved in susceptibility through cell wall modification and transmembrane transport; resistance through lignin and isoprenoid biosynthesis and fatty acid metabolism; and suberin biosynthesis in mechanical wounding. Consistent with the transcriptional up-regulation of lignin biosynthetic genes from A2-O invasion, lignin is accumulated at the root tip of S. torvum infected with avirulent A2-O but not with virulent A2-J, suggesting that S. torvum reinforces the cell wall as a defense response against the avirulent RKN. M. arenaria pathotypes A2-J and A2-O were propagated on Solanum lycopersicum cultivar “Micro-Tom” in a greenhouse. Nematode eggs were isolated from infected roots and then hatched at 25 ◦C. Freshly hatched J2s were collected and transferred to a Kimwipe filter placed on the top of a glass beaker filled with sterilized distilled water containing 100 µg/ml streptomycin and 10 µg/ml nystatin. Only active J2s pass through the filter. Filtered J2s were surface sterilized with 0.002 % mercuric chloride, 0.002 % sodium azide, and 0.001 % Triton X-100 for 10 min, and then rinsed three times with SDW . Eleven-day-old S. torvum seedlings grown on the MS-Gelrite in 6-well plates were inoculated with 200–300 J2s resuspended in SDW.
The plates were wrapped in aluminum foil for 2–3 days after inoculation to promote nematode infection. When mature giant cells were observed 18 days post-inoculation , we used the MS-Gelrite media without sucrose to prevent the formation of callus-like structures. The difference in the number of normal galls formed by A2-J or A2-O at 4 DPI was statistically tested using the Mann-Whitney U test with R software . Nematodes resident in root tissues were stained with acid fuchsin 2–4 DPI , photographed by light microscopy , and the photomicrographs were processed using cellSens . For the observation of giant cells and developing nematodes at 18 DPI, infection sites were fixed with glutaraldehyde and cleaned with benzyl-alcohol/benzyl-benzoate . We observed BABB-cleaned samples by confocal laser scanning microscopy . Photomicrographs were processed using LAS X software . S. torvum seedlings were grown on half-strength MS-Gelrite medium containing 1% sucrose. Eleven-day-old seedlings were treated with SDW as a mock infection or with 200–300 J2s of M. arenaria A2-J for susceptible infection or A2-O for resistant infection. Root tips attacked by the nematodes were checked under microscopy, and more than 50 root tips were cut and collected for each treatment . Root tip samples were collected at 1, 2, and 3 DPI with four biological replicates. Whole shoot and root samples were collected at 1, 3, 6, and 9 DPI with four biological replicates. Root tip samples were used for de novo assembly and differential gene expression analyses, and whole shoot and root samples were used only for de novo assembly. RNA-seq libraries were prepared from the collected samples using a high-throughput RNA-seq method . The 85-bp paired-end reads for the root tip samples, and the 85-bp single-end reads for the whole shoot and root samples were sequenced on an Illumina NextSeq 500 platform . The FASTX toolkit 0.0.13.2 was used for quality filtering. Low-quality nucleotides were removed from the 30 ends, french flower bucket and short reads were excluded. Reads with at least 95% of nucleotides with Phred scores > 20 were kept and used for the downstream analyses . Filtered reads were mapped to the genome assembly of M. arenaria A2-J or A2-O using HISAT2 to exclude reads of nematode origin. Unmapped reads were used for de novo transcriptome assembly . Three different transcriptome assemblers were used for de novo assembly: SOAPdenovo-Trans v1.03 , Velvet v1.2.10 /Oases v0.2.09 and Trinity package v2.4.0 . Unmapped paired-end and single end reads were normalized using Trinity and assembled independently . Oases assembled scaffolds were split at gaps into contigs before merging with contigs from the other assemblies with the EvidentialGene tr2aacds pipeline. The tr2aacds pipeline produces ‘primary’ and ‘alternate’ sequences of non-redundant transcripts with ‘primary’ transcripts being the longest coding sequence for a predicted locus. Next, we used the evgmrna2tsa program from EvidentialGene to generate mRNA, coding, and protein sequences. BUSCO v3.0.2 was applied for quantitative assessment of assembly completeness. This assembly and one previously reported for S. torvum by Yang et al.were compared to the Embryophyta odb9 dataset, which contains 1,440 BUSCO groups. The homology of the contigs from the final assembly was searched against the NCBI non-redundant database using BLASTX with an e-value threshold of 1E-05.
We also compared the contigs with Arabidopsis genome annotation using BLASTX at the e-value cutoff of 10. Results of the annotation are summarized in Supplementary Table 2. To group genes by expression pattern, we applied the self organizing map clustering method on genes within the top 25 % of the coefficient of variation for expression across samples as previously described . Scaled expression values, representing the average principal component values among each gene in a cluster were used for multilevel three-by-three hexagonal SOM . The final assignment of genes to winning units formed the basis of the gene clusters. The results of SOM clustering were visualized in a principal component analysis space where PC values were calculated based on gene expression across samples . We compared the contigs of our assembly with the NCBI non-redundant database using BLASTX with an e-value threshold of 1E-05. In addition, predicted amino acid sequences that begin with methionine were also annotated using InterProScan . BLASTX and InterProScan outputs were used for Blast2GO analysis to annotate the contigs with Gene Ontology terms . GO enrichment analyses of the sets of genes induced by A2-O infection at 1 DPI or that were assigned to each cluster generated by SOM was performed by comparison with all genes using GO terms generated by Blast2GO at the FDR cutoff of 1E-04 . We further used the “Reduce to most specific terms” option in Blast2GO to remove general GO terms and obtain only the most specific GO terms.Quantification of aliphatic suberin was performed as described previously . Eleven-day-old plants were treated with SDW or infected with A2-J or A2-O. At 4 DPI, root tips inoculated with nematodes were microscopically checked for infection, and more than 50 infected root tips were cut and collected for each treatment. To remove unbound lipids, samples were extracted in methanol for 24 h then in chloroform for 24 h, dried, and weighed. Samples were depolymerized and analyzed by gas chromatography-mass spectrometry for monomer identification and for quantitative analysis based on an internal standard using an identical gas chromatography system coupled with a flame ionization detector as described previously . To understand the differential responses of S. torvum to M. arenaria A2-J and A2-O, we first established an in vitro infection system. Seedlings of S. torvum were grown in MSGelrite plates for 11 days and then inoculated with 200–300 J2s of A2-J or A2-O. At 4 DPI, more than 90 % of root tips infected with A2-J induced the formation of gall-like structures ranging in size. These galls are classified here as “normal” galls, while the rest produced brown pigments. Normal galls lacked obvious brown pigment accumulation and were further classified based on the width of the gall into small , medium , and large . In contrast, about 60 % of A2-O-infected root tips accumulated at least some brown pigment. Some of these brownish root tips also had an abnormal appearance due to the formation of balloon-like structures, and others had many localized and highly pigmented spots. There were a very few small gall-like structures formed after infection with A2-O, but far fewer and smaller than in root tips infected with A2-J . RKN staining by acid fuchsin revealed that both A2-J and A2-O successfully invaded the roots . Interestingly, host cells invaded by A2- O uniformly accumulated brownish pigments, suggesting that the surrounding tissue is strongly responding to, and highly correlated with A2-O infection, a response that was absent from A2-J infected roots. It is generally known that browning of plant tissue is related to enzymatic or non-enzymatic oxidation of phenolic substances , but the identity of the brown pigments synthesized upon infection with A2-O is unknown. By 18 DPI, A2-J had induced the formation of mature multinucleate giant cells and developed into fourth stage juveniles . In contrast, A2-O did not induce the formation of giant cells nor develop past second stage juveniles.