Secretome Analysis of the Pine Wood Nematode Bursaphelenchus xylophilus Reveals the Tangled Roots of Parasitism and Its Potential for Molecular Mimicry

Since it was first introduced into Asia from North America in the early 20th century, the pine wood nematode Bursaphelenchus xylophilus has caused the devastating forest disease called pine wilt. The emerging pathogen spread to parts of Europe and has since been found as the causal agent of pine wilt disease in Portugal and Spain. In 2011, the entire genome sequence of B. xylophilus was determined, and it allowed us to perform a more detailed analysis of B. xylophilus parasitism. Here, we identified 1,515 proteins secreted by B. xylophilus using a highly sensitive proteomics method combined with the available genomic sequence. The catalogue of secreted proteins contained proteins involved in nutrient uptake, migration, and evasion from host defenses. A comparative functional analysis of the secretome profiles among parasitic nematodes revealed a marked expansion of secreted peptidases and peptidase inhibitors in B. xylophilus via gene duplication and horizontal gene transfer from fungi and bacteria. Furthermore, we showed that B. xylophilus secreted the potential host mimicry proteins that closely resemble the host pine’s proteins. These proteins could have been acquired by host–parasite co-evolution and might mimic the host defense systems in susceptible pine trees during infection. This study contributes to an understanding of their unique parasitism and its tangled roots, and provides new perspectives on the evolution of plant parasitism among nematodes.


Introduction
The pine wood nematode, Bursaphelenchus xylophilus, is a plant parasitic nematode (PPN) and one of the most notorious forest pests in the world. The disease caused by B. xylophilus is called ''pine wilt disease'' and was first discovered in the early 20 th century in Japan. In 1971, B. xylophilus was shown to cause the disease [1]. Then, B. xylophilus spread to neighboring East Asian countries, such as China and Korea in 1982 and 1988, respectively [2][3][4], and subsequently it was found in Portugal in 1999 [5] and in Spain in 2008 [6]. It is now known that B. xylophilus was first described in 1934 in Louisiana, and thus originated in North America. From there it was introduced into Japan [7]. Unlike other major herbaceous PPNs, such as root knot nematodes and cyst nematodes, which infect plant roots and induce the formation of specialized feeding cells to uptake nutrients, B. xylophilus infects the above-ground parts of trees and quickly kills its host. Once B. xylophilus enters a pine tree, it migrates through the resin canals of the tree, destructively feeding on the parenchymal cells. B. xylophilus is also able to feed on fungal growth after the plant cells are dead.
To obtain nutrients for development and reproduction from the cytoplasm of living cells, the fungal feeding nematodes and PPNs evolved a needle-like feeding structure, the stylet (Figure 1), as well as marked morphological and physiological modifications of the pharynx [8]. When feeding on plants and fungi, B. xylophilus uses the stylet to pierce the cell wall and ingest nutrients from the cytoplasm. The proteins secreted from the stylet are produced in the esophageal glands (subventral and dorsal glands) ( Figure 1). Furthermore, the proteins also secreted from the hypodermis or released from natural openings of the nematode [9]. These secretions would contain cell-wall degrading enzymes and other crucial molecules for migration within plant tissues and in the interaction of the nematode with its host plant.
Owing to the importance of secreted proteins during PPN infection, these proteins have been extensively studied [10][11][12][13]. Early investigations of these secreted proteins were performed by biochemical approaches [14,15]. Around the end of 20 th century, rapid advances in molecular biology techniques occurred and a large number of secreted proteins were identified by expressed gene characterization [10,[16][17][18]. To date, over 100 genes encoding secreted proteins have been cloned, and the host targets and functions of several of the effector proteins have been elucidated in root knot nematodes and cyst nematodes [13]. Recently, Bellafiore et al. [19] performed a proteome analysis and identified 486 secreted proteins of the root knot nematode Meloidogyne incognita. Their study represented a significant breakthrough in the research of PPN secreted proteins, because they identified a large set of secreted proteins and because the proteins can be directly identified. There is evidence to suggest that some proteins present in secretions of nematodes may lack a classical signal peptide for secretion [20]. Therefore, proteome analysis, at present, would be the most powerful and successful approach in identifying secreted proteins.
The secreted proteins of B. xylophilus have also attracted the attention of many researchers as pathogenicity factor candidates in pine wilt disease [21][22][23][24]. However, so far only a limited number of cell-wall degrading enzymes, such as endo-b-1,4-glucanases (cellulases), b-1,3-glucanases, pectate lyase, expansin-like protein and venom allergen-like proteins [24] have been cloned and characterized. The molecular mechanisms of pathogenicity of B. xylophilus continue to remain controversial. This may, in part, be attributed to the limited knowledge regarding the molecules involved in the pathology as well as the lack of techniques for functional analyses. Potential pathogenic molecules from the nematodes include surface coat proteins and secretions from the stylet or other natural openings ( Figure 1). So far, a profile of B. xylophilus surface coat proteins has been revealed by proteome analysis using the expressed sequence tag database of B. xylophilus [25]. Since the entire genome sequence of B. xylophilus was recently determined [26], it enables us to address the study of proteins on a larger scale. Thus, we conducted a large-scale identification of secreted proteins using nano-liquid chromatography coupled with tandem mass spectrometry (nanoLC-MS/MS) analysis in the present study.
The primary aim of this study was to obtain a complete picture of the proteins secreted by B. xylophilus using proteome analysis based on the genome sequence information [26] and to gain insight into the molecular basis of B. xylophilus parasitism. Furthermore, we compared the secretome profile among parasitic nematodes and reconsidered the evolution of parasitism in nematodes.

Nematodes
The virulent Ka4 isolate of B. xylophilus, which was the source for the inbred line Ka4C1 that was sequenced, was used in this study [26]. A mixed culture of the propagative forms, including 2 nd -(J2), 3 rd -(J3), and 4 th -stage juveniles (J4), adults and eggs, were propagated on the fungus Botrytis cinerea growing on potato dextrose agar medium (Nissui-seiyaku, Tokyo, Japan) containing 100 units/ml penicillin and 100 mg/ml streptomycin at 25uC. After incubation for 10 days, the nematodes were extracted for 6 h from the culture using the Baermann funnel technique [27], and washed 10 times in sterile water containing 100 units/ml penicillin, 100 mg/ml streptomycin and 0.25 mg/ml amphotericin B.
Preparation of the pine wood extract as a stimulant for the production of secreted proteins The pine wood extract was prepared from nematode noninoculated 3-year-old Japanese black pine, Pinus thunbergii, seedlings. The stems of the seedlings were cleaned and cut into small pieces, and 30 g of the wood pieces were soaked in 150 ml distilled water for 24 h at 4uC. Supernatant solution was then collected and passed through filter paper. To remove proteins larger than 3 kDa, which were derived from pine tissues, the extracted solution was centrifuged through an Amicon Ultra centrifugal filter (Millipore, Bedford, MA) with a 3 kDa cutoff. The solution, which passed through the membrane into the bottom of the centrifugation tube, was collected and refiltered using a 0.22-mmpore polyvinylidene fluoride (PVDF) membrane. The solution of pine extracts thus obtained was used in the following procedure to prepare B. xylophilus secreted proteins for use as a stimulant.

Preparation of secreted and whole body proteins of B. xylophilus
For preparation of secreted proteins, a nematode population of 1610 7 was soaked in 5 ml of pine wood extract on a 100 mm low attachment surface plate, EZ-BindShut dish (Iwaki, Tokyo, Japan), at 28uC for 16 h with agitation. The nematodes were then gently pelleted by centrifugation at 8 g for 5 min at 25uC. The supernatant was collected, passed through a 0.45-mm pore size hydrophilic PVDF Millex-HV syringe filter, and concentrated with Microcon YM-3 columns (Millipore) to 100 ml.
For preparation of whole B. xylophilus proteins, 100 ml of nematodes was solubilized in 200 ml of protein lysis buffer (9 M urea, 2% CHAPS, 20 mM Tris-HCl at pH 8.0). After one freeze (liquid nitrogen) and thaw (room temperature) cycle, the nematodes were homogenized using disposable homogenization pestle on ice. The crude lysate was then pelleted by centrifugation at 15,000 6 g for 20 min at 4uC. The supernatant was collected and concentrated using the same procedure as the preparation of secreted proteins.
The protein concentrations of sample solutions were determined using a Bradford assay kit (Nacalai Tesque, Kyoto, Japan) according to the manufacturer's protocol. A total of 45 mg of collected proteins was reduced with 10 mM Tris (2-carboxyethyl) phosphine for 60 min and alkylated with 20 mM iodoacetamide for 30 min at room temperature. After acetone precipitation, proteins were digested in 200 mM triethylammonium bicarbonate with 2.5 mg trypsin (Promega, Madison, WI) for 12 h at 37uC. Following enzymatic digestion, the peptides were applied to a proteome analysis system. Three independent biological replicates were prepared and used in the following proteome analysis.

NanoLC-MS/MS analysis
Proteome analysis was performed by liquid chromatography (LC; Prominence nano flow system; Shimadzu, Kyoto, Japan)/ mass spectrometry (MS; LTQ Velos orbitrap mass spectrometer; Thermo Fisher Scientific, Bremen, Germany). Ten microliters of proteolytic digests was injected and separated by reversed-phase chromatography using a custom-made monolithic silica capillary column, prepared from a mixture of tetramethoxysilane and methyltrimethoxysilane (300 cm long, 0.1 mm ID) as described in Motokawa et al. [28], at a flow rate of 500 nl/min. The gradient was provided by changing the mixing ratio of the two eluents: A, 0.1% (v/v) formic acid, and B, acetonitrile containing 0.1% (v/v) formic acid. The gradient was started with 5% B, increased to 45% B for 600 min, further increased to 95% B to wash the column, returned to the initial condition, and held for reequilibration. The separated analytes were detected on a mass spectrometer (with full scan range 350-1500 m/z). For datadependent acquisition, the method was set to automatically analyze the top 10 most intense ions observed in the MS scan. An ESI voltage of 2.4 kV was applied directly to the LC buffer distal to the chromatography column using a microTee (Upchurch Scientific, Oak Harbor, WA). The ion transfer tube temperature on the LTQ Velos ion trap was set to 300uC.
The mass spectrometry data were used for protein identification by Mascot software (Matrix Science, Boston, MA) with an annotated B. xylophilus protein database (v.1.2) obtained from GeneDB (http://www.genedb.org) [26]. MS/MS spectra were collected for secreted and whole B. xylophilus proteins (6,221,311 and 700,658 spectra, respectively). Since 47,835 MS/MS spectra were collected from the pine wood extract without B. xylophilus, these spectra were excluded and 6,173,476 MS/MS spectra were used for following B. xylophilus secreted protein identification analysis. The enzyme parameter was limited to full tryptic peptides with a maximum miscleavage default setting (carbamidomethylation of cysteines, +/-20 ppm for precursor ions, +/-0.6 Da for fragment ions). A decoy database was constructed to calculate the in situ false discovery rate (FDR) [29]. An identification filtering criteria of 1% FDR was used at the peptide level for every search. All protein matches were required to be detected in at least two of three biological replicates. Proteins with single peptide hits were limited to unique peptide sequences not matched in any higher ranked proteins.
To compare the relative abundance of proteins between secretome and whole B. xylophilus proteins, we used the spectral counting method described in Bellafiore et al. [19]. Total spectral counts were used for normalization of the spectral counts from each protein. The normalized spectral count ratio of each protein (secretome/whole B. xylophilus proteins) was used as the index of protein abundance in secretome and whole B. xylophilus proteins.
Comparison of secretome profiles among parasitic nematodes based on OrthoMCL cluster analysis and Gene Ontology (GO) analysis To detect the putative orthologs in the secretomes of other parasitic nematodes, we performed an OrthoMCL cluster analysis, using the default settings (E-value cutoff 1e-5 and percent identity 50%), on the 1,515 secreted proteins of B. xylophilus against the secreted proteins of two other parasitic nematodes, M. incognita and Brugia malayi [30]. The secretome sequence lists of M. incognita and B. malayi were obtained from Bellafiore et al. [19] and Bennuru et al. [20], respectively. In B. xylophilus, GO annotation made by Kikuchi et al. [26] was used. GO annotations of the secretomes of M. incognita and B. malayi were performed in the same manner as the annotation for B. xylophilus. In brief, GO annotations were initially derived using Blast2GO software [31] based on the BLAST match against NCBI non-redundant proteins with an Evalue cutoff of 1e-10. The GO annotations were also performed with InterProScan [32], and then these annotation data were merged together within the Blast2GO.

Detection of peptidases and peptidase inhibitors
To detect putative peptidases (also termed proteases or proteinases) and peptidase inhibitors, a MEROPS batch BLAST search was performed [33]. A total of 1,515 secreted protein sequences were subjected to the MEROPS BLAST search and classified into detailed MEROPS peptidase or peptidase inhibitor families (E-value cutoff of 1e-5) [33]. To compare the peptidases and peptidase inhibitors in the secretome of M. incognita and B. malayi, the same procedures were performed against each secretome profile.

Detection of plant/fungal cell-wall degrading enzymes
The CAZymes Analysis Toolkit [34] was used to detect B. xylophilus carbohydrate active enzymes (CAZymes) based on the CAZy database. An annotation method ''based on association rules between CAZy families and Pfam domains'' was used with an E-value threshold of 0.01, a bitscore threshold of 55 and rule support level 40. The putative plant/fungal cell-wall degrading enzymes were manually selected from the CAZymes according to Kikuchi et al. [26]. In addition, only for expansin, which would be involved in cell-wall modification, the candidate proteins were manually identified based on the InterProScan based annotation data.

Detection of proteins acquired by horizontal gene transfer
The potential horizontal gene transfers (HGT) specifically into B. xylophilus were manually identified based on the data in Kikuchi et al. [26]. Kikuchi et al. [26] performed BLASTP analysis to compare predicted B. xylophilus protein sequences against the NCBI nr database, producing a candidate set of horizontally transferred genes that had significant BLAST hits (E-value ,1e-10) to organisms other than nematodes and no significant BLAST hits (E-value #1e-5) to any nematode sequence except for genes from Aphelenchoidea. For each of HGT candidates, amino acid sequence data was extracted from the NCBI database, aligned using Muscle and generated Maximum Likelihood phylogenetic trees. The candidates supported by phylogenetic evidence were identified as HGT genes [26].
Detection of potential B. xylophilus proteins that mimic the host plant defense system To find proteins in the B. xylophilus secretome that mimic those of the host plant, we performed BLASTP analysis to compare B. xylophilus protein sequences against the NCBI nr database. The proteins that had a significant top BLAST hit to plant proteins (Evalue ,1e-4) and no significant BLAST hits (E-value ,0.1) to any nematode sequence were identified as potential proteins that mimic the host plant defense system.

Extraction and stimulation of the secreted proteins of B. xylophilus
In the preliminary test, the efficiency of three potential stimulants, 5-methoxy-N, N-dimethyl tryptamine oxalate (DMT), resorcinol, and pine wood extract, were examined for efficacy to induce the secretion of proteins of B. xylophilus. DMT and resorcinol had been reported as stimulants of protein secretion in other PPNs [35,36]. The pine wood extract showed the highest efficacy in induction of secreted proteins of the four test solutions (0.4% resorcinol, 400 mg/ml DMT, pine wood extract and phosphate buffered saline), and the efficacy was not lost even when high molecular weight (. 3,000 Da) molecules were removed from the extract (data not shown). This result suggested that the water soluble low molecular weight compounds in the pine extract could induce the secretion of proteins. Furthermore, three classes of pine seedlings, those that had been inoculated with nematodes three days prior (early stage of infection), those that had been inoculated with nematodes 2 weeks prior (late stage of infection), and nematode free (non-infested), were prepared and treated with B. xylophilus to determine the optimal conditions of pine seedlings as a stimulant. As a result, a significant decrease in the efficacy of secretion induction was observed only in the pine seedlings in the late stage of infection (data not shown). Since the mixed stages of nematodes in the propagative stages, which are the stages for propagation in the pine tree and the cause of the pine wilt, were used in this study, the secreted B. xylophilus proteins used for this proteome analysis should include those in the early stage of infection. In order to exclude the possibility of potential contamination of the pine-derived proteins in the B. xylophilus secretome, we employed the pine wood extract not infected with nematodes as a control. No proteins were detected in gels by silver staining, but using mass spectrometry, 47,835 MS/MS spectra and 15 peptides were found and overlapped with the B. xylophilus secretome samples. These proteins were derived from the pine extract and were excluded in the protein identification analysis.

Identification of the B. xylophilus secretome
In this secretome analysis, 1,515 distinct secreted proteins of B. xylophilus were identified. Kikuchi et al. [26] predicted that the B. xylophilus genome includes 18,074 protein-coding genes, meaning that at least 8.4% of the total B. xylophilus proteins are secreted. The complete list of identified secreted proteins is given in a concise form as supplemental information (Table S1, Dataset S1). These include all secreted proteins previously reported by Kikuchi et al. [22,23,37,38] and Lin et al. [24] indicating that the results obtained here both confirm and extend previous studies. The majority of secreted proteins of B. xylophilus (72.0%) were identified by detecting two or more peptides. Among 1,515 proteins identified in this study, 510 proteins were not able to be annotated. The draft genome data indicated that B. xylophilus produces 3,365 predicted secreted proteins in total. In this study, 625 of a total of 1,515 secreted proteins had putative secretion signal sequence(s), and therefore, the remaining 58.7% may have unknown secretory signals or may be secreted through nonclassical secretory pathways. The proportion of the identified proteins bearing a secretion signal in this study is similar to a previous secretome study on the filarial nematode B. malayi [20] and is higher than that of M. incognita [19], suggesting that the secreted proteins lacking known secretion motifs are common to the nematodes.
The whole B. xylophilus proteins were also identified to serve as a control for potential contamination of the secretome by somatic proteins (Table S2). The relative abundance of proteins between secreted proteins and whole B. xylophilus proteins were shown in Table S1 and Figure 2. Approximately 27% (407) of the secreted proteins were at least 10-fold more abundant than the whole B. xylophilus proteins. On the other hand, only 4% (66) of the secreted proteins were 10-fold less abundant than the whole B. xylophilus proteins. Most of proteins discussed in detail (cell-wall degrading enzyme, HGT, host mimicry etc.) were significantly less abundant or absent in the whole B. xylophilus proteome (Figure 2). These results provide further evidence that they are indeed secreted. The abundant proteins in the whole B. xylophilus proteins may be contaminations, but there is a possibility that these proteins were truly secreted.  Table S5). Furthermore, a GO analysis among nematodes clearly showed the expansion of peptidases in the secretome of B. xylophilus (Figure 4). In particular, a large number of cysteine and aspartic peptidases were detected in B. xylophilus ( Table 1). The genome sequence also revealed that B. xylophilus has a large number of predicted peptidase genes [26]. The expansion of peptidases was also reflected in the secretome of B. xylophilus. As a result of a MEROPS BLAST analysis, a total of 161 secreted peptidases were identified in the B. xylophilus secretome ( Table 1). The percentage of peptidases in the B. xylophilus secretome was 10.6%, whereas in M. incognita and B. malayi there were 6.4% and 5.2%, respectively. As a result of peptidase classification, C1A (papain) family cysteine peptidases and A1A (pepsin) family aspartic peptidases were both identified as major secreted peptidase groups by MEROPS BLAST. In addition, a GO analysis among nematodes also indicated the expansion of proteins with nucleotide binding functions in the M. incognita secretome and the expansion of proteins with cation (metal) binding functions in the B. malayi secretome. M. incognita induces the differentiation of root cells into specialized feeding cells called 'giant cells'. It has been suggested that the secreted proteins that target host nuclei manipulate nuclear functions of the cells and induce giant cells [19,39]. Therefore, the expansion of proteins with nucleotide binding functions would be a key event toward the establishment of a unique feeding strategy and highly specialized plant parasitism. In the secretome analysis of M. incognita, only the J2 stage was used [19], although the mixed stages of the propagative forms were used in B. xylophilus. It is possible that the differences in secretome profiles and the number of secreted proteins identified are due to the different life stages used in the secretome analysis. However, the J2 stage in M. incognita is an infective stage to host plants. Similarly, the propagative forms of B. xylophilus are stages that directly contact the host tree and cause disease ( Figure S1). Therefore, it is appropriate to compare the J2 stage of M. incognita and the mixed stages of the propagative forms of B. xylophilus to compare their plant parasitisms, because there is no evidence that only a specific stage in the propagative forms of B. xylophilus is responsible for the disease.
A phylogenetic tree of the C1A (papain) family cysteine peptidases showed that the C1A family cysteine peptidases have undergone the greatest expansion via gene duplication without HGT from ancestral cysteine peptidase genes ( Figure S2). However, a phylogenetic tree of the A1A (pepsin) family aspartic peptidases indicated that HGT is a primary driving force in the expansion of A1A family aspartic peptidases ( Figure S3). Why does B. xylophilus increase the number of peptidases by gene duplication and HGT? Peptidases could act in at least two ways to assist the survival of B. xylophilus: they may be required for nutritional purposes or utilized to degrade proteins in the plant cell wall, allowing for the spread of nematodes or the defeat of host defenses. One explanation for nutritional purposes is that B. xylophilus can uptake nutrients from a variety of food sources. B. xylophilus can feed both on fungi and living plant cells. More correctly, Bursaphelenchus spp. were originally a mycophagous group [40] that has adapted to feeding on living plant tissues. In nature, when living tree cells are no longer available after tree death, B. xylophilus feeds and reproduces on the fungal hyphae growing along the resin canals. On the other hand, Meloidogyne spp. are obligate plant parasites, and they can uptake nutrients only from host plant cells.  Kikuchi et al. [26] showed that both M. incognita and Meloidogyne hapla have fewer putative peptidase genes in their genome than other nematodes. Furthermore, very few peptidases were detected in the secretome of M. incognita (Table 1). Taking into account the evolutionary development of nematodes ( Figure S4) [41,42], B. xylophilus would gain peptidase genes by gene duplication and HGT, whereas the obligate plant parasitic groups, such as M. incognita, would lose many of the peptidase genes as they adapted to each food source and/or life cycle. In the PPNs, very few reports regarding secreted peptidases have been presented so far [19,43].
In addition, there are no reports of a functional analysis of the secreted peptidases in plant-nematode interactions. However, in the animal parasitic and entomopathogenic nematodes, there are many examples of secreted peptidases that are involved in tissue invasion/migration [44][45][46][47][48][49][50]. It has been indicated that these peptidases contribute to host specificity, host range and virulence [48,51]. Therefore, it is quite likely that the secreted peptidases of B. xylophilus have a key role in its successful parasitism of pine trees. However, further functional analyses are needed to determine an exact role of the peptidases in host-parasite interactions.

Cell-wall degrading enzyme
Enzymes involved in the carbohydrate metabolic process, including glycoside hydrolases (GH), carbohydrate esterases (CE), glycosyl transferases (GT), carbohydrate-binding modules (CBM) and polysaccharide lyases (PL), are referred to as carbohydrate active enzymes. Some of these are also known as cell-wall degrading enzymes. In total, 65 putative carbohydrate active enzymes were identified in the secretome of B. xylophilus in this study (Table S6). Furthermore, we manually detected four expansin proteins based on the InterProScan based annotation data. Among them, we detected 28 putative plant/fungal cell-wall degrading enzymes based on Kikuchi et al. [26] (Table 2). It has been suggested that GH45 cellulase, pectate lyase and expansin proteins are the major plant cell-wall degrading enzymes of B. xylophilus, because the principal carbohydrates of the primary cell wall of plant are cellulose, hemicellulose and pectin [52]. GH16 b-  1,3-glucanase and chitinases (GH18 and GH20) have also been suggested to play important roles in fungal cell-wall degradation by the basically mycophagous B. xylophilus because b-1,3-glucan and chitin are main components of the fungal cell wall [53]. The results of this study are the first evidence of the secretion of these enzymes by B. xylophilus. The profiles of the cell-wall degrading enzymes of B. xylophilus also suggested that B. xylophilus has quite different mechanisms of cell-wall degradation from M. incognita, because no putative fungal cell-wall degrading enzymes were identified in the secretome of M. incognita [19]. Such a difference would reflect the difference in their food sources. B. xylophilus can use both plant cells and fungal ones as food sources, while most other obligate PPNs feed on only plants. Therefore, the differences in cell-wall degrading enzymes would allow the unique life style of the facultative PPN B. xylophilus.

The secreted proteins of B. xylophilus acquired by horizontal gene transfer
The B. xylophilus genome includes some genes acquired from other organisms by HGT [25]. Kikuchi et al. [26], supported by phylogenetic evidence, indicated that a total of 24 genes were putatively horizontally transferred genes. In this study, we identified 16 of the 24 (66.7%) putatively horizontally transferred proteins in the secretome of B. xylophilus. There were three aspartic endopeptidase, one cysteine peptidase inhibitor, seven GH45 family members, and five GH16 b-1,3-glucanases (Table 3). They were acquired from ascomycete fungi and bacteria. Since the percentage of secreted proteins in the total predicted proteins was 8.6%, the ratio (66.7%) of secreted proteins to whole proteins acquired by HGT was significantly higher. Since these proteins acquired by HGT are essential for survival and give fitness advantages to B. xylophilus, they would be positively selected for over evolutionary time.

Evasion from the host defense response
It is worth highlighting proteins related to evasion from the host defense response, because continuous excess ROS generation is a distinctive characteristic of pine wilt disease and is suggested to be an important factor in the development of disease symptoms [54,55]. Thus, the effective anti-oxidant ability is of critical importance in establishing the infection. In this study, only the proteins that showed anti-oxidant activity, catalase activity or superoxide dismutase activity in the molecular functions of the GO analysis were regarded as anti-oxidant proteins. As a result, a total of 12 anti-oxidant proteins were identified in the secretome of B. xylophilus (Table 4), including peroxiredoxin, catalase, glutathione peroxidase, nucleoredoxin-like protein, superoxide dismutase, and thioredoxin. In addition, four homologs of glutathione-s-transferases, well known as a detoxifying enzyme, were identified. B. xylophilus also accumulates anti-oxidant and detoxifying enzymes on the body surface [25]. B. xylophilus mainly inhabits the resin canals and rays of the host pine tree. The resin canals are filled with several terpenes including thick and sticky ones. Furthermore, in the early stage of a B. xylophilus infection, abundant ROS is produced in the pine tree [56]. Therefore, the resin canal would be an extremely severe environment for B. xylophilus [57]. The secreted anti-oxidant and detoxifying enzymes would play a pivotal role in protecting B. xylophilus itself from oxygen free radicals and toxic compounds in the pine tree.
In addition to ROS generation in the plant defense system, plant peptidases also play key roles against pathogens and pests. The MEROPS BLAST analysis detected 47 putative peptidase inhibitors in the B. xylophilus secretome ( Table 5). The number of secreted peptidase inhibitors of B. xylophilus was significantly greater than other parasitic nematodes. In particular, the inhibitors in subfamily I25B and I29 were highly expanded in B. xylophilus secretome (Table 5). Inhibitor family I25B and I29 primarily contain inhibitors of cysteine peptidases (C1A). Some I25B family proteins also inhibit legumain (C13). It is known that the families of cysteine peptidases C1A and C13 play key roles in various physiological phenomena in plants [58][59][60], and they are involved in the regulation of the plant defense system [61]. A role in defense against pathogens by executing programmed cell death using the caspase activity observed for these cysteine peptidases has been proposed [62]. Therefore, the main role of the expanded I25B and I29 peptidase inhibitors in the B. xylophilus secretome could be to battle against host plant cysteine peptidases. Recently, Hirao et al. [63] revealed that the expression of peptidase genes in a susceptible family of Japanese black pines was induced more quickly and more significantly during B. xylophilus infection than in a resistant family. Such an overexpression of peptidase genes was one of the most intense reactions in the B. xylophilus infection at the gene expression level. Therefore, the interactions of peptidases with their substrates and inhibitors must be one of the most important molecular battle-fields at the B. xylophilus-pine interface. Since the number of secreted peptidase inhibitors in B. xylophilus was significantly greater than in other parasitic nematodes, the expansion of peptidase inhibitors could result from the competitive co-evolution of B. xylophilus and pines or between a close ancestor of B. xylophilus and its host.
Potential B. xylophilus proteins that mimic host plant defense systems In the B. xylophilus secretome, we identified three secreted proteins that have high sequence similarity to plant proteins, while no similar proteins were detected in other nematode species (Evalue ,0.1). These included two putative thaumatin-like proteins and a cystatin-like peptidase inhibitor (Table 6). In plants, thaumatin-like proteins are known as pathogenesis-related protein 5 (PR-5), and they have an antifungal activity that acts by permeabilizing fungal membranes [64]. Furthermore, they also appear to function by binding and hydrolyzing b-1,3-glucans [65], or inhibiting fungal xylanases [66]. Moreover, it was reported that thaumatin in plants has a possible role in activating other plant defense pathways, including phenylpropanoid and phytoalexin production [67]. So far, thaumatin-like proteins have been discovered in a wide range of organisms, including plants, nematodes [68], insects [69], and fungi [70,71]. However, the two thaumatin-like proteins in the secretome of B. xylophilus were more like those of plants than those of other nematodes. Surprisingly, the thaumatin-like proteins of B. xylophilus showed the highest level of similarity to the proteins of pine tree (the genus Pinus).
As mentioned above, cystatin mainly inhibits peptidases belonging to the C1 (papain family) and C13 (legumain family) peptidase families. Multiple roles have been attributed to cystatins in plants, including the control of endogenous cysteine peptidases, seed development, and programmed cell death [72][73][74]. Furthermore, plant cystatins also play a significant role in plant defenses against pathogenic microbes, herbivorous arthropods, and PPNs [73,75,76]. The cystatin-like peptidase inhibitor in the secretome of B. xylophilus was most similar to a cystatin-like peptidase inhibitor of the herbaceous plant Medicago truncatula. Molecular mimicry is one of the well-known strategies for parasite host defense system evasion and host manipulation. One well-known example of molecular mimicry is the secretion of proteins by animal parasitic nematodes [77]. The helminth parasite B. malayi secretes homologs of the human cytokine macrophage migration inhibitory factor (MIF) [78]. B. malayi-encoded MIFs have the same effect on human monocytes as the mammalian MIF. Molecular mimicry in PPNs has also been reported. Examples of these molecules include the CLAVATA3/ESR-like (CLE) peptides [79]. The nematode CLEs may play roles through molecular mimicry in the regulation of certain root meristematic cells that are essential for the establishment of feeding sites in the host, a critical step for nematode parasitism of plants [79][80][81]. In addition, Bellafiore et al. [19] identified several secreted proteins of the root knot nematode M. incognita that are homologous to plant proteins by proteome analysis, although the function of these proteins still remains unclear. In B. xylophilus, there has been no report of molecular mimicry until this finding. Recently, Hirao et al. [63] revealed that the expression of antimicrobial peptides and putative pathogenesis-related genes (e.g., PR-1b, 2, 3, 4, 5, and 6) was much higher in susceptible trees than in resistant trees, irrespective of the time after nematode infection. In particular, the genes PR-5 (thaumatin) and PR-6 (peptidase) were significantly overexpressed in susceptible trees during B. xylophilus infections. That is, the molecular mimicry candidates in B. xylophilus were expressed concurrently with the most noteworthy proteins in susceptible pine trees. This correspondence is an intriguing and compelling finding. Although it is difficult to determine the evolutionary origin of the potential mimicry proteins of B. xylophilus, these molecular mimicry proteins could have been acquired by host-parasite co-evolution. Only about 100 years have passed since the first report of pine wilt disease. From this viewpoint, co-evolution could have occurred through an interaction with native pine species in North America. The native pine trees should have evolved to overcome these molecules from the parasites during host-parasite co-evolution. In contrast, nonnative pine species to North America would not have evolved with respect to resistance against these mimicry molecules of newly emerged parasites and would be highly susceptible to their damage. Therefore, the abnormal responses in susceptible pine trees (mostly non-native pine species) might be induced by mimicry molecules secreted by B. xylophilus.

Concluding Remarks
This study provides a comprehensive profile of the secretome of B. xylophilus, comprising a large array of proteins, many of which have been shown to be important in nutrition, invasion/migration in host trees, modulating the host immune evasion, and potentially promoting their survival. In addition, many ''hypothetical proteins'' were detected by LC-MS/MS-based secretome analysis. In this secretome analysis the application of high-throughput proteomic techniques allowed the identification of low abundance proteins that might not be detected using gel electrophoresis-based methods. Typical SignalP analyses would not account for 58.7% of the secreted proteins identified here, and this suggests that new methods for these predictions need to be properly tailored to (parasitic) nematodes. Furthermore, the most important finding in this study is that B. xylophilus secretes a large number and various types of peptidases and peptidase inhibitors. The theories of gene duplication suggest that ancestral genes possessed multiple functions prior to duplication that were then preserved and refined in the duplicated genes by subsequent positive selection to evolve paralogs with distinct functions [82,83]. Therefore, it is quite likely that these secreted peptidases and peptidase inhibitors, exceptionally expanded, are essential molecules for parasitism by B. xylophilus. As for the cysteine peptidase inhibitors, we revealed not only their expansion but also the molecular mimicry potential to host pines. Taking into consideration that B. xylophilus is an exotic pathogen in Asia and Europe, these potential host mimicry proteins that co-evolved in relation to the native pine species of North America might be crucial molecules in the pathogenicity of B. xylophilus against the non-native susceptible pine trees.
The comparative functional analysis of the B. xylophilus secretome and the M. incognita secretome showed the unique expansion of peptidases and nucleotide binding proteins in B. xylophilus and M. incognita, respectively, although 51.4% of the secreted proteins of M. incognita were shared with those of B. xylophilus. These unique expanded molecules seem critical to each unique parasitism. In the study of pine wilt disease, researchers have sought molecules similar to or homologous with effector molecules of obligate PPNs for a long time. However, such an approach would not be effective for identifying crucial molecules in, and for understanding the parasitism of B. xylophilus because the strategy and the evolutionary processes of plant parasitism would be significantly different. The secretome information of B. xylophilus will aid us in addressing the issue of parasitism by B. xylophilus separately from those of the obligate PPNs. Future research that focuses on these unique secreted proteins of B. xylophilus will be useful in revealing mechanisms of pine wilt disease and in developing novel control strategies for pine wilt disease.     Dataset S1 Protein datasets used for secretome identification.