Unveiling the Secretome of the Fungal Plant Pathogen Neofusicoccum parvum Induced by In Vitro Host Mimicry

Neofusicoccum parvum is a fungal plant pathogen of a wide range of hosts but knowledge about the virulence factors of N. parvum and host–pathogen interactions is rather limited. The molecules involved in the interaction between N. parvum and Eucalyptus are mostly unknown, so we used a multi-omics approach to understand pathogen–host interactions. We present the first comprehensive characterization of the in vitro secretome of N. parvum and a prediction of protein–protein interactions using a dry-lab non-targeted interactomics strategy. We used LC-MS to identify N. parvum protein profiles, resulting in the identification of over 400 proteins, from which 117 had a different abundance in the presence of the Eucalyptus stem. Most of the more abundant proteins under host mimicry are involved in plant cell wall degradation (targeting pectin and hemicellulose) consistent with pathogen growth on a plant host. Other proteins identified are involved in adhesion to host tissues, penetration, pathogenesis, or reactive oxygen species generation, involving ribonuclease/ribotoxin domains, putative ricin B lectins, and necrosis elicitors. The overexpression of chitosan synthesis proteins during interaction with the Eucalyptus stem reinforces the hypothesis of an infection strategy involving pathogen masking to avoid host defenses. Neofusicoccum parvum has the molecular apparatus to colonize the host but also actively feed on its living cells and induce necrosis suggesting that this species has a hemibiotrophic lifestyle.


Introduction
Eucalyptus species are native to Australia but due to their enormous economic significance are planted in many countries around the world. Eucalyptus species were introduced in Portugal more than 100 years ago and are nowadays the most representative forest tree species. Eucalyptus globulus is the most abundant species in Portugal, occupying ca. 8500 km 2 , the equivalent to ca. 9% of the country (26% of the forest area of Portugal), mostly in Central and Northwest Portugal [1,2]. This species is well adapted to the Mediterraneanlike climate and is exploited mainly due to the commercial interests of the pulp and paper industries. Unfortunately, they are commonly susceptible to diseases/infections caused by various species of the family Botryosphaeriaceae.
Botryosphaeriaceae are well-known fungal opportunistic pathogens that elicit disease symptoms in plants under stress conditions, resulting in high economic losses [3][4][5]. In addition, these species are known to occur in asymptomatic plant tissues as commensals or latent pathogens in a variety of tree species including Eucalyptus [5][6][7][8]. Botryosphaeriaceae have been associated with Eucalyptus canker and dieback in Portugal [9,10] and are considered a significant threat to the productivity and sustainability of Eucalyptus spp. plantations. In a survey conducted in 2015, and again in 2018, the predominant isolates collected from Eucalyptus were identified as belonging to the genus Neofusicoccum [9,10]. Several studies have reported a diverse assemblage of Neofusicoccum species occurring on Eucalyptus spp. both as disease-causing agents and as commensals [6,11,12].
Neofusicoccum parvum is a vascular aggressive pathogen that causes severe decline and dieback symptoms in a wide range of hosts [7,13], being also common in many Eucalyptus species [4,6,9,14,15]. In general, the fungus penetrates through wounds and colonizes the host tissues, causing shoot dieback, stem canker, cane bleaching, bud necrosis, and graft failure. Neofusicoccum parvum is an endophyte (i.e., it colonizes the interior of plants) that switches from a 'no-or not visible' inducing host damage status to a clear pathogen. In fact, not much is known about the strategies that this fungus employs to infect its hosts, or about the molecules it expresses during infection. Several studies have suggested that N. parvum pathogenicity could be related to the ability of this fungus to colonize woody tissue combined with the production of several phytotoxins [16][17][18][19] and also the expression of extracellular proteins with phytotoxic properties [20]. A study of genes encoding necrosis and ethylene-inducing proteins (NLPs) in N. parvum showed that they are functional genes encoding proteins toxic both to plant and mammalian cells, being most probably involved in virulence or cell death during N. parvum infection [21]. Recent genomic and transcriptomic analyses have shown that this pathogen has evolved special adaptive mechanisms to infect woody plants [22,23]. These mechanisms include a significant expansion of gene families associated with virulence and nutrient uptake, including cellular transporters, cell-wall-degrading enzymes (CWDEs), cytochrome P450s, putative effectors, and biosynthesis of secondary metabolites. The interaction between grapevine and N. parvum was also studied at the transcriptomic level [22,24]. Host plant stems and leaves underwent extensive transcriptomic reprogramming, but woody stems reacted earlier than leaves to infection. Gene expression analysis showed that N. parvum co-expresses genes associated with secondary metabolism and plant cell wall degradation in a dependent manner on the growth substrate and the stage of plant infection. Overall, these studies have shed light on the interactions between plants and N. parvum. However, a full understanding of the pathogenicity mechanism is still far from being accomplished. To investigate the mechanisms of pathogenicity of this fungus, we centered our analysis on the secretome [25][26][27][28][29], due to its relevance to the infection mechanism and to fungus-plant interactions. Proteomics data from the species of the family Botryosphaeriaceae are limited. So far, the proteome of Diplodia seriata [30], Diplodia corticola [31], and, most recently, Lasiodiplodia theobromae [32][33][34] have been made available. Proteins identified in these studies suggest differences in the infection strategies of these fungi. Although the genome of N. parvum was sequenced and released in 2013 [13], no proteomics studies have been carried out until now.
The aim of this study was to characterize the secretome of N. parvum, evaluate its response to an in vitro host mimicry, and predict interactions of the secretome proteins with host proteins.

Fungal Strains, Plant Material, and Culture Conditions
The strain used in this study, N. parvum CAA704, was recovered from E. globulus displaying symptoms of dieback and decline in Portugal [9]. This strain also proved to be pathogenic to E. globulus in artificial inoculation trials [9]. The strain was grown on Potato Dextrose Agar (PDA, Merck, Germany) at 25 • C for 7 days prior to the inoculations. The 3-months-old E. globulus (MB43, obtained from Altri, SGPS, S.A.) seedlings were watered weekly and kept at room temperature under natural light.
Two conditions were tested: control and infection-like. For the control condition, two mycelium plugs (5 mm diameter) were inoculated into a 250 mL flask containing 50 mL of Potato Dextrose Broth (PDB, Merck, Germany) and incubated in triplicate at 25 • C for 12 days. For the infection-like condition, a sterilized piece of E. globulus stem (±2 g) was added to the PDB, as described elsewhere [31]. The culture supernatant of each condition was harvested through filter paper and immediately stored at −80 • C for extracellular protein extraction. Mycelia obtained from both conditions were collected by filtration, washed with sterile water, and frozen with N 2 (L) for DNA and RNA extraction.

RNA Extraction and cDNA Synthesis
Total RNA was extracted from 12-days-old mycelium ground in liquid nitrogen (three biological replicates from each condition) using the Spectrum Plant Total RNA kit (Sigma-Aldrich, St. Louis, MO, USA), according to the manufacturer's instructions. Samples were treated with DNase I digestion set (RNase-Free DNase Set, QIAGEN, Hilden, Germany) for 15 min to remove genomic DNA. The quality and quantity of RNA were checked by gel electrophoresis and NanoDrop™ 1000 Spectrophotometer (Thermo Scientific, Waltham, MA, USA). cDNA was generated using the Nzy First-Strand cDNA Synthesis Kit (Nzytech, Lisboa, Portugal), according to the manufacturer's instructions.

Quantitative PCR
Target genes were selected according to their pattern of expression and functional annotation ( Table 1). All reactions were performed in a CFX96 Real-Time thermocycler (BioRad, Hercules, CA, USA) using the NzySpeedy quantitative PCR (qPCR) Green Master Mix (2×) (NZYtech, Lisboa, Portugal). For each reaction, 5 µL of the Master Mix, 0.5 µL (10 µM) of each primer, 4.2 µL of nuclease-free water, and 0.5 µL of template cDNA were used. The PCR program used was: 95 • C-3 min, 40 cycles of 95 • C-15 s, and 60 • C-30 s. After this step, the fluorescence intensity was measured and, at the end of the program, the temperature was increased from 65 • C to 95 • C at a rate ramp of 0.1 • C/s, allowing the melting curves elaboration. Cq values were calculated with BIO-RAD CFX Manager software and used to compare the expression between reference and target genes.

Extracellular Protein Extraction
Secreted proteins were extracted using TCA/acetone according to the method described by Fernandes, et al. [31]. To discard precipitated polysaccharides, 35 mL of the culture supernatant was centrifuged (48,400× g, 1 h at 4 • C). One volume of cold TCA/acetone (20%/80% (w/v)) supplemented with 0.14% (w/v) dithiothreitol (DTT) was added to the supernatant and incubated at −20 • C for 1 h. Precipitated proteins were collected by centrifugation (15,000× g, 20 min, 4 • C) and the supernatant was removed. Precipitated proteins were washed with 10 mL of ice-cold acetone (twice) (15,000× g, 15 min, 4 • C) and 10 mL of ice-cold 80% acetone (v/v) (15,000× g, 15 min, 4 • C) to discard the excess of TCA from the precipitate. Residual acetone was air-dried, and the protein pellet was resuspended in 0.1 M Tris HCl pH 8 and stored at −80 • C.

Protein Sample Cleaning
To remove salts, detergents, and phenolic compounds, the protein extract was cleaned with the water/chloroform/methanol protein precipitation method (adapted from [36]). Briefly, a mixture of methanol, chloroform, and water (4:1:3 (v/v/v)) was added to the sample and thoroughly vortexed. Then, the mixture was centrifuged at 14,000× g for 1 min and the top aqueous methanol layer was removed (the proteins being in the interphase). Four volumes of methanol were added, and the mixture was vortexed and centrifuged at 14,000× g for 5 min. The supernatant was removed without disturbing the pellet. The air-dried pellet was finally resuspended in 0.1 M Tris HCl pH 8 and stored at −80 • C.

Protein Quantification
Protein concentration assay was carried out with the Pierce ® 660 nm Protein Assay kit (Thermo Scientific, Waltham, MA, USA) according to the manufacturer's instructions, using Bovine Serum Albumin (BSA) as standard. All samples were quantified in triplicate.

Protein Quality Evaluation by Electrophoresis
The quality of protein samples was assessed by SDS-PAGE. Briefly, 3 µg of protein were denaturated and separated in a 12.5% SDS-PAGE gel electrophoresis, for 120 min at 120 V, in a Mini-PROTEAN 3 Cell (Bio-Rad, Hercules, CA, USA), according to Laemmli's protocol [37]. The running buffer contained 100 mM Tris, 100 mM Bicine, and 0.1% (w/v) SDS. Gels were stained with Coomassie Brilliant Blue G-250. After staining, gels were scanned on a GS-800 Calibrated Densitometer (Bio-Rad, Hercules, CA, USA).

Tryptic Digestion, Mass Spectrometry Analysis, and Protein Identification
Ten µg of the protein sample were diluted in NH 4 HCO 3 50 mM buffer (in 30 µL). Twenty µL of BSA 0.002 µg/mL was added and the solution was incubated at 80 • C for 10 min. Samples were reduced with 5 µL of DTT 50 mM/NH 4 HCO 3 50 mM (incubation at 60 • C for 10 min) and alkylated with 5 µL of iodoacetamide (IAA) 150 mM/NH 4 HCO 3 50 mM (incubation in the dark for 20 min). Proteins were digested with 2 µL of trypsin 0.1 µg/µL. Afterward, samples were acidified with 1% formic acid and incubated at 37 • C for 30 min. After centrifugation (16,000× g, 30 min), the supernatant was transferred to new vials and a peptide purification step was performed using C 18 Omix tips. The peptides were dried in a vacuum concentrator (SpeedVac, ThermoFisher Scientific, Waltham, MA, USA) and stored at −20 • C until analysis.
Purified peptides were re-dissolved in loading solvent (0.1% trifluoroacetic acid (TFA) in water/acetonitrile (ACN) (98:2, v/v)) and injected into an Ultimate 3000 RSLC nano system in-line connected to a Q Exactive HF mass spectrometer (Thermo, Waltham, MA, USA). Trapping was performed at 10 µL/min for 4 min in loading solvent A on a 20 mm trapping column (made in-house, 100 µm internal diameter (I.D.), 5 µm beads, C 18 Reprosil-HD, Dr. Maisch, Ammerbuch, Germany) and the sample was loaded onto a 400 mm analytical column (made in-house, 75 µm I.D., 1.9 µm beads, C 18 Reprosil-HD, Dr. Maisch, Ammerbuch, Germany). Peptides were eluted by a non-linear gradient from 2 to 56% solvent B [0.1% formic acid in water/acetonitrile (2:8, v/v)] over 145 min at a constant flow rate of 250 nL/min, followed by a 10 min wash reaching 97% MS solvent B and reequilibration with solvent A (0.1% formic acid in water) for 20 min. The column temperature was kept constant at 50 • C by a column oven (Sonation COControl). The mass spectrometer was operated in data-dependent mode, automatically switching between MS and MS/MS acquisition for the 16 most abundant ion peaks per MS spectrum. Full-scan MS spectra (375-1500 m/z) were acquired at a resolution of 60,000 in the Orbitrap analyzer after accumulation to a target value of 3 × 10 6 . The 16 most intense ions above a threshold value of 1.3 × 10 4 were isolated for fragmentation at a normalized collision energy of 28% after filling the trap at a target value of 1 × 10 5 for a maximum of 80 ms. MS/MS spectra (200-2000 m/z) were acquired at a resolution of 15,000 in the Orbitrap analyzer.
The raw data generated from LC-MS was further inputted in Max-Quant (version 1.6.2.1, https://maxquant.net/maxquant/, accessed on 1 April 2017, Max Planck Institute, Martinsried, Germany), a quantitative proteomics software developed by Cox and Mann [38]. MS1 spectra were searched with the Andromeda peptide database engine [39] against a FASTA database of proteins from the N. parvum genome from UniProt (July, 2017) [40] and analyzed for label-free quantification of the peptides present in the samples. The peptide database was constructed from in-silico prediction of tryptic peptides with up to two missed cleavages, carbamidomethylation of cysteines as fixed modifications, and oxidation of methionines and N-terminal acetylation as variable modifications. Peptide spectral matches were validated using a percolator based on q-values at a 1% false discovery rate (FDR). Identified peptides were assembled into protein groups according to the law of parsimony and filtered to 1% FDR. Perseus software (version 1.6.1.3, https://maxquant.net/perseus/, accessed on 1 April 2017, Max Planck Institute, Martinsried, Germany) [41] enabled the affiliation of the protein groups into identified proteins. Identified proteins were filtered and only considered for analysis if present in 3 replicates and using at least 3 peptides for identification. Reverse proteins and proteins identified only by site were filtered out. A multi-scatter plot and hierarchical clustering were performed to assess the quality of the experiment. To identify interactor proteins, a two-sample t-test between control and infection-like samples was performed with minimal fold change (s0) of 1.8 and 1% FDR. A scatter plot, volcano plot, and profile plot were used to visualize the results ( Figures S2-S4).

Bioinformatic Analysis
Identified proteins were classified according to the GO (biological process). Whenever necessary, the protein's family and domain were determined by the identification of conserved domains in the InterPro database (http://www.ebi.ac.uk/interpro, accessed on 1 April 2017) [42]. Cell-wall-degrading enzymes were classified according to the carbohydrate-active enzymes database CAZy (http://www.cazy.org, accessed on 1 April 2017) [43].

Interactomics Analysis
The OralInt algorithm [47] was used to predict the interactions between all the proteins of Eucalyptus grandis reference proteome (Uniprot UP000030711, 44,150 proteins) with the differentially secreted proteins of N. parvum identified in this study (117 proteins). OralInt is a computational prediction method based on an ensemble methodology combining five distinct protein-protein interactions (PPI) prediction techniques, namely: literature mining, primary protein sequences, orthologous profiles, biological process similarity, and domain interactions [47]. Since the sequence is the feature with the best overall performance, this method of predicting interactions can be applied independently of the organisms under study.

Secretome Analysis
Prior to LC-MS, protein extracts were analyzed for quality control by SDS-PAGE ( Figure S1). The secretomes of N. parvum grown in the absence (control) and presence of a Eucalyptus stem (infection-like condition) were analyzed. In total, 471 proteins were identified in both control and infection-like secretomes, of which 131 proteins were significantly different in abundance between the two conditions (t-test, difference cutoff of 1.8). Most of these proteins are extracellular (Tables 2 and S1), except for 14 proteins predicted as intracellular proteins (10.7%, Table S1). Proteins were classified according to their gene ontology (GO) (Molecular Function), and into 10 protein families: CAZymes, hydrolases, proteases, oxidoreductases, lyases, protein-protein interaction, carbohydrate-binding proteins, RNA-binding proteins, and proteins with other functions and unknown functions (Figure 1).
A variety of proteases (endo and exoproteases) were also identified. The aspartic endopeptidase PEP1 (R1GM42) was more abundant in the infection-like condition (although its mRNA was as abundant as in control conditions, Figure S6) along with a variety of other metallopeptidases (M28, M35, and M43, Figure 1B and Table 2). In contrast, serine peptidases (S8 (R1EAW3) and S10 (R1FV38)) were less abundant in the infection-like conditions ( Figure 1 and Table 2).
A putative berberine-like protein (R1GD68)-an oxidoreductase with a FAD-binding domain-was the most abundant protein in the infection-like condition ( Table 2).
Other functional categories-proteins involved in carbohydrate binding (R1EYI5 and R1GAK8), RNA binding (R1ERG2, R1FZX2, and R1H1L9), protein-protein interactions (R1ENG6, R1E9S0, and R1GCJ5), and proteins with other functions (R1EGT1, R1GV87, R1EAF3, R1G1U2, R1FV21, R1FVG4, R1GBA7, R1EWZ5, R1GDV3, R1GKT0, R1E681)were also identified ( Figure 1 and Table 2).  For each category, the number of proteins is reflected by the size of the pie slice. The classification was obtained from the GO annotation at the UniProt database [40]. When lacking exact functional annotations in UniProt, the family and domain databases InterPro and Pfam [42,53] were used to reveal annotations of the identified proteins of conserved domains.

Protein-Protein Interaction
Due to the lack of data available on the proteome sequence of E. globulus, the reference proteome for E. grandis (a closely related species) was used. Protein-protein interaction (PPI) networks between all the proteins of E. grandis (44,150 proteins, blue) and the extracellular proteins of N. parvum (117 proteins, red) were predicted using the OralInt algorithm ( Figure 2). OralInt is based on high-quality experimental PPIs that feeds an artificial intelligence algorithm that is later validated. In previous studies, OralInt was applied to predict interactions between the Zika virus and the host [54] and between the oral microbiome and the host [55].
A total of 3201 interactions were predicted involving 76 proteins of N. parvum and 1591 proteins of E. grandis. Some proteins, both in Eucalyptus and in the fungus, showed a high number of interactions ( Figure 2B, Tables 3 and S3). The functional analysis of Eucalyptus proteins on which the fungus acts is provided in the Supplementary file ( Figure S5). Neofusicoccum parvum hub proteins-those that center a high number of interactions-include mainly enzymes (Table 3). Neofusicoccum parvum proteins interact mainly with proteins involved in biosynthetic processes, nucleobase-containing compound metabolic processes, signal transduction, cell communication, response to endogenous stimulus, and response to stress, which fits well with a response to a foreign attack. interaction (PPI) networks between all the proteins of E. grandis (44,150 proteins, blue) and the extracellular proteins of N. parvum (117 proteins, red) were predicted using the OralInt algorithm ( Figure 2). OralInt is based on high-quality experimental PPIs that feeds an artificial intelligence algorithm that is later validated. In previous studies, OralInt was applied to predict interactions between the Zika virus and the host [54] and between the oral microbiome and the host [55]. A total of 3201 interactions were predicted involving 76 proteins of N. parvum and 1591 proteins of E. grandis. Some proteins, both in Eucalyptus and in the fungus, showed a high number of interactions ( Figure 2B, Tables 3 and S3). The functional analysis of Eucalyptus proteins on which the fungus acts is provided in the Supplementary file ( Figure S5). Neofusicoccum parvum hub proteins-those that center a high number of interactions-include mainly enzymes (Table 3). Neofusicoccum parvum proteins interact mainly with proteins involved in biosynthetic processes, nucleobase-containing compound metabolic processes, signal transduction, cell communication, response to endogenous stimulus, and response to stress, which fits well with a response to a foreign attack.

Discussion
Plant infection by phytopathogens, such as N. parvum, is a complex process that starts with the attachment of the infective propagule to the plant surface followed by penetration and infection. The infection mechanism of Neofusicoccum species relies on a myriad of molecules, mainly secondary metabolites, and proteins. It is known that species like N. parvum are able to express metabolites with phytotoxic activities such as Cyclohexenones, 5,6-Dihydro-2-pyrones, Melleins, Isosclerone, Hydroxypropyl-and methyl-salicylic acid, Tyrosol, Ethyl linoleate, Stearic acid, and Naphthalenones (Botryosphaerones A and D and 3,4,5-Trihydroxy-1-tetralone (for a review on the metabolites produced by Neofusicoccum spp. see [19]). These compounds were identified in isolates pathogenic of Vitis vinifera, and although it is expected that some of them will be present in other pathogen-host systems, it is not known. Understanding how other pathogen-host function is vital to fully understanding the mechanism of infection of N. parvum. The choice of the Eucalyptus-N. parvum system was based on the following reasons: (1) there are no reports on any molecules involved in the infection of Eucalyptus by N. parvum and (2) Eucalyptus is an economically vital crop in many countries, such as Portugal, and (3) there are no reports on the proteins involved in the infection mechanism of N. parvum.
Neofusicoccum parvum is a common pathogen of grapevine that infects many other hosts. In fact, the strain used in this work, N. parvum CAA704, was recovered from E. globulus displaying symptoms of dieback and decline and later it was shown to be pathogenic to E. globulus [9]. We compared the protein profiles of N. parvum in the control and infection-like conditions and identified and quantified proteins whose abundance changes in response to the Eucalyptus stem, to highlight proteins involved in the interaction with Eucalyptus during fungal infection. Although simple, and lacking the influence that molecules expressed by the host in response to the fungus attack have on the infection mechanism, the in vitro infection-like system used has been successfully used for other systems to mimic the infection mechanism of fungal pathogens [56,57].
Secreted proteins were visualized in a volcano plot (Figure 3) to have a quick visual identification of proteins that display large magnitude fold change and high statistical significance. The most significant proteins are discussed below. Most of the proteins (86.3%) were predicted to contain a Signal P motif and are supposed to traverse the classical Golgi and endoplasmic reticulum secretion pathway. The possible implementation of the non-classical pathway for the proteins lacking signal peptide (13.7%) was confirmed using the SecretomeP predictor [46] (Table 2). Such proteins, known as leaderless-secreted proteins (LCPs), have been identified in several other studies involving the secretome [30,58,59]. Of these LCPs proteins, a putative ethanolamine utilization protein (R1G1U2) and a chitin-binding protein (R1EW80) showed low Secre-tomeP scores (NN score = 0.223 and 0.417, respectively) ( Table 2). However, the NN score of a chitin-binding protein is relatively close to the 0.5 threshold, suggesting that the protein may in fact be secreted. The presence of intracellular proteins in the secretome is common and can result from cell death during culture, cell lysis during protein extraction, or secretion through non-common mechanisms. The number of cellular proteins identified in this study is similar to that identified in the secretome of D. seriata (16 proteins, [30]), D. corticola (12 proteins, [56]), and L. theobromae (16 proteins, [32]) the closest Botryosphaeriaceae species whose secretomes were studied.

Proteins Involved in Carbohydrate Metabolic Processes
Being a plant pathogen, it is not surprising that the N. parvum strain CAA704 secretome is mainly constituted by proteins involved in carbohydrate metabolic processes (18% Most of the proteins (86.3%) were predicted to contain a Signal P motif and are supposed to traverse the classical Golgi and endoplasmic reticulum secretion pathway. The possible implementation of the non-classical pathway for the proteins lacking signal peptide (13.7%) was confirmed using the SecretomeP predictor [46] (Table 2). Such proteins, known as leaderless-secreted proteins (LCPs), have been identified in several other studies involving the secretome [30,58,59]. Of these LCPs proteins, a putative ethanolamine utilization protein (R1G1U2) and a chitin-binding protein (R1EW80) showed low SecretomeP scores (NN score = 0.223 and 0.417, respectively) ( Table 2). However, the NN score of a chitin-binding protein is relatively close to the 0.5 threshold, suggesting that the protein may in fact be secreted. The presence of intracellular proteins in the secretome is common and can result from cell death during culture, cell lysis during protein extraction, or secretion through non-common mechanisms. The number of cellular proteins identified in this study is similar to that identified in the secretome of D. seriata (16 proteins, [30]), D. corticola (12 proteins, [56]), and L. theobromae (16 proteins, [32]) the closest Botryosphaeriaceae species whose secretomes were studied.

Proteins Involved in Carbohydrate Metabolic Processes
Being a plant pathogen, it is not surprising that the N. parvum strain CAA704 secretome is mainly constituted by proteins involved in carbohydrate metabolic processes (18% and 30% of the proteins identified in control and in the presence of Eucalyptus, respectively) and in catabolic processes (8% and 9.7%; Figure 4). As a plant pathogen, N. parvum uses its host as a nutrient source and for that the carbohydrate-degrading enzymes are essential (Figure 4). Although there are very few secretomes of Botryosphaeriaceae fungi available, data shows that the trend is similar with a major component of the secretomes being carbohydrate-related enzymes [30,31,60,61]. Botryosphaeriaceae species can shift between a pathogenic and non-pathogenic lifestyle when triggered (by conditions not yet fully understood) and therefore the characterization of the secretomes under controlled in vitro conditions is of the utmost relevance to understanding the nature of these organisms. Plant cell-wall-degrading enzymes (PCWDEs) play significant roles in plant colonization and are typical of necrotrophic life-style fungi [62], allowing them to perceive weak regions of plant epidermal cells and penetrate the plant's primary cell wall. Our results indicate that N. parvum is equipped with an army of extracellular PCWDEs expressed even in the absence of plant tissue, but induced by the presence of the Eucalyptus stem ( Table 2). Pectic enzymes (in multiple forms) are the first cell-wall-degrading enzymes induced by pathogens when cultured on isolated plant cell walls and the first produced in infected tissues [63,64]. Pectic enzymes induce the modification of the cell wall structure, exposing cell wall components for degradation by other enzymes [65]. Pectin is also present in Eucalyptus cell walls (15.2-25.8 mg g −1 pectin, [66]) and the secretion of pectindegrading enzymes by N. parvum upon interaction with the Eucalyptus stem surely promotes the close interaction between the fungus and plant. All identified pectin-degrading enzymes [pectinases (GH53 and CE12) and pectate lyases (PL1, P3, PL4), Tables 2 and S1] are more abundant in the presence of host material, suggesting that this fungus is more adapted to degrade intact or living plants than decaying biomass (where pectin is not present and is already decayed), which is in consonance with the fungus being a biotroph. Kang and Buchenauer [67] and Tomassini, et al. [68] demonstrated that wheat infection by Fusarium culmorum and F. graminearum depends on the production of CWDE at the early stages of infection which results in the facilitation of a rapid colonization of wheat spikes. Moreover, the up-regulation of these enzymes in lethal isolates of Verticilium albo- Plant cell-wall-degrading enzymes (PCWDEs) play significant roles in plant colonization and are typical of necrotrophic life-style fungi [62], allowing them to perceive weak regions of plant epidermal cells and penetrate the plant's primary cell wall. Our results indicate that N. parvum is equipped with an army of extracellular PCWDEs expressed even in the absence of plant tissue, but induced by the presence of the Eucalyptus stem (Table 2). Pectic enzymes (in multiple forms) are the first cell-wall-degrading enzymes induced by pathogens when cultured on isolated plant cell walls and the first produced in infected tissues [63,64]. Pectic enzymes induce the modification of the cell wall structure, exposing cell wall components for degradation by other enzymes [65]. Pectin is also present in Eucalyptus cell walls (15.2-25.8 mg g −1 pectin, [66]) and the secretion of pectin-degrading enzymes by N. parvum upon interaction with the Eucalyptus stem surely promotes the close interaction between the fungus and plant. All identified pectin-degrading enzymes [pectinases (GH53 and CE12) and pectate lyases (PL1, P3, PL4), Tables 2 and S1] are more abundant in the presence of host material, suggesting that this fungus is more adapted to degrade intact or living plants than decaying biomass (where pectin is not present and is already decayed), which is in consonance with the fungus being a biotroph. Kang and Buchenauer [67] and Tomassini, et al. [68] demonstrated that wheat infection by Fusarium culmorum and F. graminearum depends on the production of CWDE at the early stages of infection which results in the facilitation of a rapid colonization of wheat spikes. Moreover, the up-regulation of these enzymes in lethal isolates of Verticilium albo-atrum compared to mild isolates was also described by Mandelc and Javornik [28], having implied its hypothetical contribution for plant vascular system colonization. We also identified cellulose-degrading enzymes mainly in the presence of the Eucalyptus stem. Putative GH12 protein (R1GQP5) raises special attention due to a high increase in response to Eucalyptus (3.9-fold up, Table 2). Recently, the xyloglucan-specific endo-β-1,4-glucanase (GH12 family) isolated from P. sojae culture filtrates induced cell death in dicot plants [69]. Gui,et al. [70] demonstrated that two of the six GH12 proteins in the fungus Verticillium dahliae Vd991 (VdEG1 and VdEG3) acted as virulence factors and as Pathogen-Associated Molecular Patterns (PAMPs), inducing cell death and triggering PAMP-triggered immunity in Nicotiana benthamiana. A glucanase (R1GZN3, GH7) was more abundant in the presence of the Eucalyptus stem than in control conditions. Cellulases belonging to GH6 and GH7 families are related to fungal virulence in the phytopathogenic fungus Magnaporthe oryzae, where they seem to be involved in the penetration of the host epidermis and further invasion [71].

EER REVIEW 15 of 24
Hemicellulases are generally involved in the degradation of hemicellulose from plant cell walls, helping in the colonization and in the acquisition of nutrients during infection. The up-regulation of two endoxylanases [beta-xylanase GH10 (R1FWZ0) and endo-1,4beta-xylanase GH11 (R1GCT8)] was observed in N. parvum secretome in response to the Eucalyptus stem. GH10 and GH11 endoxylanases play significant roles in both vertical penetration of cell walls and horizontal expansion of the rice pathogen M. oryzae in infected leaves [72]. A recent study showed that two genes encoding GH10 xylanases are crucial for the virulence of the oomycete plant pathogen Phytophthora parasitica [73]. In B. cinerea, Xyn11A encodes an endo-β-1,4-xylanase Xyn11A, and the disruption of this gene resulted in reduced virulence of the pathogen [74]. However, several reports failed to show an essential role of endoxylanases in fungal pathogenicity [75][76][77]. Therefore, the role of xylanases in fungal pathogenesis may vary depending on the characteristics of the pathosystem and awaits further investigation.
The most abundant CAZy proteins in the control secretome of N. parvum are involved in lignin degradation (AA1, AA5, AA7) ( Table 2 and Table S1). These enzymes belong to the oxidoreductase family which can produce the H 2 O 2 required for the action of extracellular peroxidases. Usually, N. parvum is not considered a major lignin-depolymerizing fungus, like white-rot fungi [78], but, in our study, several extracellular lignin-degrading enzymes were identified. However, most of those enzymes (five out of six) were less abundant in the presence of the Eucalyptus stem, indicating that they may have another role in N. parvum rather than a direct role in lignocellulose deconstruction during infection. A similar observation was described for the white-rot fungus Lentinula edodes when exposed to microcrystalline cellulose, cellulose with lignosulfonate, and glucose [79]. CAZymes involved in lignin degradation were repressed by cellulose. Cai, Gong, Liu, Hu, Chen, Yan, Zhou, and Bian [79] suggested that laccases may have a role in increasing fungal resistance to oxidative stress rather than being involved in lignocellulosic degradation.

Defense from Host
Oxidoreductases, important virulence factors induced during plant infection [80,81], are over-represented in the secretome of the N. parvum supplemented with the Eucalyptus stem (5-fold), where they may contribute to the alkaloid biosynthesis and production of hydrogen peroxide through the oxidation of metabolites [82].
Besides host degradation, fungal cell wall degradation plays a fundamental role in fungal development during infection, facilitating fungal branching and elongation [83]. The putative chitin-binding protein (R1EW80) contains a chitin deacetylase domain which catalyzes the conversion of chitin into chitosan required for appressorium formation [84]. Interestingly, the up-regulation of this protein in response to the host mimicry reinforces the hypothesis that this protein might play a role in host colonization. Some pathogens produce chitin-binding proteins that mask chitin, avoiding host recognition by shielding, or by modifying it [85,86]. Similarly, we identified a putative chitin deacetylase (CE4, R1E7G7) which is significantly up-regulated (5.6-fold) in the presence of the Eucalyptus stem. Chitin deacetylases are also involved in the protection of fungi from host plant chitinases by converting the fungal cell wall chitin into chitosan [87,88]. An endo-chitosanase (GH75, R1GTL6) is also up-regulated (4.1-fold) during the interaction with the Eucalyptus stem, suggesting that chitosan generation by chitin deacetylase enhanced the chitosanolytic activity of the fungus. We predicted that this endo-chitosanase (GH75, R1GTL6) has numerous interactions with other Eucalyptus proteins (69 proteins , Tables 3 and S4), some of which share important functions in Eucalyptus [auxin response factor (A0A059ACB3), delta-1-pyrroline-5-carboxylate synthase (A0A059BIT2), alpha-1,4 glucan phosphorylase (A0A059D8I8)], suggesting a central role for this protein in N. parvum infection mechanism of Eucalyptus (Tables 3 and S4).

Virulence
Most extracellular proteases identified-several of which were described as virulence factors in fungal necrotrophs [89][90][91][92][93]-were more abundant upon induction by Eucalyptus. Specifically, metalloproteases such as deuterolysin are also induced in a virulent strain of D. corticola upon challenge by the host (Quercus suber) [56]. It has been suggested that deuterolysin targets proteins in the plant cell wall [94], being directly involved in the infection mechanism.
The putative ricin B lectin protein (R1GAK8), involved in carbohydrate binding, contains a pectin_lyase_fold/virulence domain (InterPro IPR011050) considered a virulence factor in several species [95][96][97]. Ricin B lectins inhibit protein synthesis [98] and are highly expressed during infection [99,100]. In N. parvum, the putative ricin B lectin protein was induced in response to the host mimicry (4.3-fold).
Proteins containing ribonuclease/ribotoxin domains are also more abundant in the secretome of N. parvum supplemented with the Eucalyptus stem when compared to the axenic culture. Ribonucleases perform a variety of functions, serving as extra-or intracellular cytotoxins, and modulating host immune responses [101,102]. Ribotoxins are fungal extracellular ribonucleases that are highly toxic due to their ability to enter host cells and their effective ribonucleolytic activity against the ribosome [103]. Extracellular ribonucleases have been related to biotrophic fungi defenses, inhibiting the action of plant ribosome-inactivating proteins that would otherwise lead to host cell death, and pathogen death [102]. Secretion of low-molecular-weight guanyl-preferring ribonucleases (RNases) has also been reported in the secretome of the D. corticola [56]. Nonetheless, we predicted that the uncharacterized protein with ribonuclease activity (R1FZX2) establishes more than 360 interactions with Eucalyptus proteins.
Neofusicoccum parvum secretome contains necrotic elicitors like the necrosis-inducing protein (R1FZC0) and the putative epl1 protein (R1G1Q3) containing cerato-platanin domain, both in the control and infection-like conditions (Table S2), suggesting that these phytotoxins are constitutively expressed by N. parvum. Interestingly, earlier we showed that NLP genes coding necrosis-inducing proteins in N. parvum are functional genes. NLP genes of N. parvum encode proteins toxic both to plant and mammalian cells, most probably involved in virulence or cell death during infection by N. parvum [21].
Proteins from the fasciclin family have been identified as cell adhesion molecules in various organisms [104][105][106][107]. In this study, the accumulation of the putative fasciclin domain family protein (R1EWZ5) in the infection-like conditions could be responsible for the attachment of fungal hyphae to the host material. In the rice blast fungus, Magnaporthe oryzae, MoFLP1 null mutants generated by targeted fasciclin gene disruption showed a significant reduction of conidiation, conidial adhesion, and appressorium turgor, resulting in overall decreased fungal pathogenicity [108]. But the knowledge on the role of fasciclin domain-containing proteins on fungi pathogenesis is still scarce and, according to Seifert [109], "cell adhesion" might be a result of turgor pressure and a buildup of adhesive materials such as pectin and not a direct function of fasciclin domain family proteins [109].

Protein-Protein Interactions
We predicted PPIs between N. parvum secreted proteins and proteins of E. grandis (used as a reference, since E. globulus genome sequence is not available). Some proteins identified display a high number of PPIs with host proteins, suggesting that those proteins may function as cross-talkers in biological functions between the fungus and the host. Among these, an auxin response factor (A0A059ACB3) in Eucalyptus raises special attention: not only it has a high number of interactions (33 , Table S5) but most proteins (27 out of 33 proteins), with whom the auxin response factor interacts, are more abundant in the infection-like condition. Furthermore, virulence factors like ricin B lectin protein (R1GAK8), putative exo-beta protein (PL3, R1H382), glucanase (R1GZN3, GH7), and putative pectate lyase (R1H2U7) were identified as the proteins interacting with auxin response factor (Table S5). Auxin response factors (ARFs) family proteins are key players in auxin signaling [110]. Indole-3-acetic acid (IAA), the major form of auxin in plants, is the most important phytohormone with the main effects on plant growth, development, and on the regulation of plant senescence [111]. Pathogens may promote auxin accumulation or auxin signaling in the host through the action of virulence factors that have evolved to modulate host auxin biology. So far, studies on Arabidopsis imply that auxin reduces (hemi) biotroph resistance but enhances plant defenses toward necrotrophic pathogens [112]. Consistent with IAA promoting hemibiotroph susceptibility, auxin, and, more specifically, IAA, also act as virulence factors of the hemibiotrophic rice pathogens Magnaporthe oryzae, Xanthomonas oryzae pv. oryzae, and X. oryzae pv. oryzicola [113][114][115]. Like many other microorganisms, these pathogens produce and secrete IAA themselves and increase IAA biosynthesis and signaling upon infection [115].

Conclusions
Multi-omics approaches to pathogen-host interactions are becoming more common, but still largely rely on the existence of genome data. Nonetheless, even in the absence of such information, using similar genome data (e.g., from the same genus) has been widely used with success. We used proteomics (focused on the secretome) and computational interactomics to shed light on the infection mechanism of N. parvum on Eucalyptus. Most of the secretome proteins induced under host mimicry are, as in the case of other Botryosphaeriaceae fungi, cell-wall-degrading enzymes (CWDEs), especially those targeting pectin and hemicellulose, allowing the fungus to invade host tissues and extract nutrients for its own growth. Additionally, the degradation of xylan (hemicellulose) and pectin is required for fungal pathogens to invasively penetrate and proliferate inside host cells. We also found the up-regulation of chitosan synthesis and chitin degradation proteins during interaction with the Eucalyptus stem, suggesting that the pathogen masks itself to avoid plant defenses.
Surely not of negligible relevance, N. parvum proteins predicted to be involved in the largest number of interactions with Eucalyptus proteins are degradative enzymes. But besides attacking the integrity of the host, N. parvum appears to mask or modify its own cell surface avoiding plant defenses, which would allow the fungus to colonize the host while actively releasing enzymes and toxins (such as proteins containing ribonuclease/ribotoxin domains, putative ricin B lectins, putative epl1 proteins containing cerato-platanin domain and necrosis inducing proteins). The isolate used in this study is able to infect and cause disease in E. globulus [9]. Like many other species of Botryosphaeriaceae fungi, N. parvum can change between a commensal and a pathogenic lifestyle.
The distinction between an organism that colonizes living plants without causing symptoms of disease and that, given a certain stimulus, becomes a pathogen and an hemibiotroph (organisms that switch from an initial biotroph to necrotroph behavior) is not easy. Neofusicoccum parvum can colonize (infect?) its hosts without causing (visible) damages (compatible with being a non-pathogenic endophyte or-we believe this is a more accurate adjective-a commensal organism), but it can also become a pathogen, causing necrosis and ultimately the death of its host. The physiology of this shift is not known, although the literature refers that host stress can induce the shift.
It is our belief that the main issue is: is N. parvum a plant commensal that shifts to a pathogen? Or is it a pathogen that, for some time, does not express its pathogenic traits? Or is it a biotroph, feeding on its host, but without causing major harm?
What do we know for sure? We know that N. parvum can cause necrosis, so, at a given point of its life cycle, it is a necrotroph: it can express active necrosis elicitors able to induce phytotoxicity and necrosis [21]. The secretome of N. parvum in the presence of eucalyptus is also compatible with a necrotrophic lifestyle: it expresses CAZymes that are as important in establishing infection as in accessing nutrients during necrotrophic growth [116].
But is N. parvum a commensal or a (non-obligate) biotroph? The molecular evidence that we have indicates ("suggests") that N. parvum is in fact a biotroph: it has the molecular machinery that allows it to colonize, spread, and feed on living plants (e.g., the overexpression of chitosan's synthesis proteins). Furthermore, N. parvum expresses pathogenesis-related proteases and proteins containing ribonuclease/ribotoxin domains (known as toxic due to their ability to enter host cells and their effective ribonucleolytic activity against the ribosome) even in the absence of the plant host.
Last, the N. parvum genome contains some of the proteins involved in the shift between the biotroph and the necrotrophic phases of a typical hemibiotroph. One example is the 4-phosphopantetheinyl transferase protein (CgPPT1). CgPPT1 is functionally involved in and required for the biotrophy-necrotrophy transition of Colletotrichum graminicola [117]. In accordance with the hypothesis that N. parvum is an hemibiotroph, its genome contains a CgPPT1 gene. Also, in Colletotrichum, the shift from biotrophy to necrotrophy is defined by induction of the degradome, mirroring necrotrophic pathosystems [118]. Neofusicoccum parvum has the enzymes (CAZYmes, proteases, . . . ) typical of these 'degradomes'.
There are still many questions to be answered, but N. parvum secretome is more closely related to that of an hemibiotroph than to a plant commensal. It lives inside plants, feeds on them, and causes damage, eventually killing them. Why it remains "dormant" or why initial damages are not seen are questions that still need to be addressed.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jof8090971/s1, Figure S1: SDS-PAGE of N. parvum-secreted proteins; Figure S2: Histograms of the LFQ intensity values of the nine samples under analysis; Figure S3. Multi-scatter plot with Pearson correlation values of the nine samples against each other; Figure S4: Hierarchical clustering of the nine samples under analysis; Figure S5: Enriched functions of Eucalyptus grandis proteins on which the fungus acts; Figure S6: Relative quantification by RT-qPCR of mRNA; Table S1: Differentially expressed proteins in the control and infection-like secretome of N. parvum; Table S2: Common proteins identified in the control and infection-like secretome of N. parvum; Table S3: PPI prediction involving proteins of N. parvum and proteins of E. grandis; Table S4: List of proteins of Eucalyptus interacting with the endo-chitosanase protein of N. parvum; Table S5: List of proteins of N. parvum interacting with auxin response factor (ARF) protein of Eucalyptus.