Chitosan modulates Pochonia chlamydosporia gene expression during nematode egg parasitism

Summary Climate change makes plant‐parasitic nematodes (PPN) an increasing threat to commercial crops. PPN can be managed sustainably by the biocontrol fungus Pochonia chlamydosporia (Pc). Chitosan generated from chitin deacetylation enhances PPN parasitism by Pc. In this work, we investigate the molecular mechanisms of Pc for chitosan resistance and root‐knot nematode (RKN) parasitism, using transcriptomics. Chitosan and RKN modify the expression of Pc genes, mainly those involved in oxidation–reduction processes. Both agents significantly modify the expression of genes associated to 113 GO terms and 180 Pc genes. Genes encoding putative glycoproteins (Pc adhesives) to nematode eggshell, as well as genes involved in redox, carbohydrate and lipid metabolism trigger the response to chitosan. We identify genes expressed in both the parasitic and endophytic phases of the Pc lifecycle; these include proteases, chitosanases and transcription factors. Using the Pathogen—Host Interaction database (PHI‐base), our previous RNA‐seq data and RT‐PCR of Pc colonizing banana we have investigated genes expressed both in the parasitic and endophytic phases of Pc lifecycle.

Chitosan is a linear polymer of β-(1-4)-linked Nacetyl-2-amino-2-deoxy-D-glucose (acetylated) and 2-amino-2-deoxy-D-glucose (deacetylated) (Kaur and Dhillon, 2014). This polymer is an elicitor of plant defences (Benhamou and Thériault, 1992;Lafontaine and Benhamou, 1996;Yin et al., 2016;Suarez-Fernandez et al., 2020) and has antifungal activity (Shih et al., 2019), inhibiting or killing fungal pathogens. Chitosan also promotes the growth of resistant fungi such as Pc and entomopathogenic fungi (Palma-Guerrero et al., 2007). The molecular mechanisms that determine whether a fungus is resistant or sensitive to chitosan remain to be determined. Pc is resistant to chitosan and can use it as a nutrient source (Palma-Guerrero et al., 2010). Chitosan-resistant fungi produce valuable bioproducts from chitosan degradation due to their chitinases and chitosanases (Kaczmarek et al., 2019). The Pc genome encodes a high number of chitosanases that are induced during nematode egg parasitism (Aranda-Martinez et al., 2016). Chitosan improves efficiency in reducing nematode pests by nematophagous fungi Mwaheb et al., 2017). Therefore, combining Pc and chitosan could be a good strategy to manage PPN infections in plants.
Global unbiased transcriptomic analyses are a useful tool for determining genes involved in the response of fungi to elicitors such as chitosan (Zhang et al., 2020). These analyses also show which genes are involved in biological processes, such as pathogenicity to nematodes (Balestrini et al., 2019). The activation of specific genes can trigger the transition from endophytism to parasitism and vice versa in fungi (Fesel and Zuccaro, 2016;Zhang et al., 2018). In view of the multitrophic interaction of chitosan and Pc with plants and PPN, it is relevant to study genes in common between endophytism and pathogenicity. This could reveal which genes are key for interacting with other organisms.
This work aims to analyse the molecular mechanisms that are activated in Pc when interacting with chitosan, RKN or both as well as to unravel which mechanisms would increase RKN parasitism by Pc. These mechanisms could explain what makes a fungus resistant or sensitive to chitosan. In this work we focus on identifying which genes are shared in both lifestyles: RKN parasitism and plant root endophytism.

Chitosan and RKN eggs modify Pc gene expression
Twelve samples were processed (four treatments per triplicate), with average total read bases (total reads × read length) of 17 894 300 929 yielding approximately 100 GB of transcriptomic raw data. When using DESeq2 in a genome-guided analysis to Pc123 genome 39 009 transcripts in 20 502 loci were found. In addition, 5746, 710 and 6595 genes were expressed in Pc treated with chitosan (PcQ), Pc treated with nematodes (PcRKN) and Pc treated with chitosan and nematode eggs (PcRKNQ) respectively (all data are available in Supplementary Information File). Considering all treatments together, 80 upregulated and 99 downregulated genes can be found with a threshold of ±2 in log 2 fold change values ( Fig. 1; Tables 1-3). In PcQ there are 136 differentially expressed genes (DEG, up and downregulated), in PcRKNQ 75 DEG, while in PcRKN only 10. Therefore, chitosan is a stronger modulator of Pc gene expression than RKN eggs are. Two genes encoding proteins (RZR63781.1, ribonuclease H-like protein and RZR63940.1, ankyrin repeat protein) are downregulated in all treatments. These results are supported by qRT-PCR ( Supplementary Fig. 1).

Chitosan favours oxidation-reduction and associated processes in Pc
We assessed the sets of differentially expressed Pc genes for enrichment in Gene Ontology (GO) terms ( Fig. 2). We considered terms from all three GO domains (Biological Processes, BP; Molecular Function, MF; Cellular Component, CC) for upregulated ( Fig. 2A) and downregulated genes (Fig. 2B). For upregulated genes, 'oxidation-reduction' (GO:0055114) and 'polysaccharide catabolism' (GO:0000272) are the most enriched BP in chitosan treatments. The importance of oxidationreduction and polysaccharide catabolism is also reflected in MF (e.g. 'oxidoreductase activity acting on paired donors, with incorporation or reduction of molecular oxygen' (GO:0016705) and 'monooxygenase activity' (GO:0004497) for oxidation-reduction processes; and 'chitosanase activity' (GO:0004568) for polysaccharide metabolism). Proteolysis (reflected in MF in 'aspartic-type endopeptidase' (GO:0004190) and 'endopeptidase' (GO:0004175) activities) is also an enriched GO term in chitosan treatments. Transmembrane transport is a BP enriched in all treatments. 'Carbohydrate transport' (GO:0008643) and 'carbohydrate derivative metabolic process ' (GO:1901135) are also enriched GO terms in chitosan treatments. Taken together, this would indicate a high chitosan turnover by Pc.
The presence of RKN represses nucleoside metabolic process (e.g. reflected in MF by 'nucleic acid binding' (GO:0003676), 'catalytic activity' (GO:0003824) and 'DNA binding' (GO:0003677)). These GO terms are enriched in both PcRKN and PcRKNQ treatments. 'Oxidation-reduction process' (GO:0055114) is the most enriched GO term for downregulated genes in chitosan treatments (reflected in MF by 'oxidoreductase' (GO:0016491) and 'hydrolase' (GO:0016787) activities). 'Plasma membrane' (GO:0005886) is the most enriched CC GO term for downregulated genes associated with chitosan treatments. PcRKN does not display any GO term associated with downregulated genes linked to CC GO domain. Chitosan and RKN modify P. chlamydosporia gene expression. Venn diagrams show a total of 80 upregulated (A) and 99 downregulated (B) genes when a threshold of ±2 in log 2 fold change is set and adjusted P-value is taken into account (44.7% vs. 55.3%). Treatments: PcQ, P. chlamydosporia with chitosan (1 mgÁml −1 ); PcRKN, P. chlamydosporia with M. javanica eggs (1 eggÁμl −1 ); PcRKNQ, P. chlamydosporia with M. javanica eggs (1 eggÁμl −1 ) and chitosan (1 mg ml −1 ). All treatments were applied for 4 days.   Table 1). These can be classified into nine clusters according to their behaviour with chitosan ( Fig. 3). Pc genes included in these significantly enriched GO terms are shown in Supplementary Table 2. Behaviour of the individual GO term is shown in Supplementary Fig. 2. Chitosan promotes the expression of Pc genes associated with 38 GO terms ( Fig. 3A) when nematodes are absent. Twenty-four of them are related to the cell cycle, 12 to protein synthesis and modification and two to sugar metabolism. In the absence of nematodes, chitosan represses Pc genes involved in metal transport and redox metabolism GO terms ( Fig. 3B and C; Supplementary Table 1). Nematode eggs reverse this behaviour. Conversely, chitosan induces genes associated with energy GO terms (Fig. 3D). Chitosan also induces the expression of genes associated with lipid metabolism GO terms (specially sphingomyelin (GO:0004767, GO:0006684 and GO:0006685) metabolism; Fig. 3E). Chitosan increases expression of genes in GO terms related to chitin and chitosan degradation ('chitosanase activity' (GO:0016977) and 'exo-1,4-beta-D-glucosaminidase' activity (GO:0052761)). This is further enhanced when RKN eggs are present (Fig. 3F). 'Phosphate pentose shunt' (GO:0006098) and 'NADPH regeneration' (GO:0006740) ( Fig. 3G) are not affected by chitosan. Both are repressed by the presence of nematodes. 'Metallocarboxypeptidase activity' (GO:0004181) is overexpressed with chitosan in the absence of nematode eggs (Fig. 3H). 'Structural constituent of cell wall' (GO:0005199) is induced with chitosan ( Fig. 3I).

Chitosan and RKN significantly modify Pc gene expression dynamics
In response to nematodes and chitosan, 180 Pc genes cluster into nine significant unique gene trends (Fig. 4,     Supplementary Fig. 3. Chitosan activates the expression of the genes associated with clusters 1-3 (Fig. 4A, B and C  (Fig. 4H), but chitosan represses them. The most significant encodes a salicylate hydroxylase (RZR59707.1). Nematode eggs and chitosan induce the expression of genes in Cluster 9 (Fig. 4I). Genes encoding the following proteins belong to this cluster: RZR65223.1 WSC domain-containing protein, RZR69865.1 Cytochrome P450, RZR64444.1 GMC oxidoreductase, RZR64079.1 3-beta hydroxysteroid dehydrogenase/isomerase. These genes would play a crucial role for RKN egg parasitism by Pc with chitosan.
Eight Pc genes are expressed in endophytism, pathogenicity and response to chitosan To investigate Pc genes shared between endophytism and parasitism, we combined PHI-base (Urban et al., 2020; www.phi-base.org), Pc upregulated genes found in this work (RKN parasitism and chitosan metabolism) and Pc genes expressed during Barley colonization (Larriba et al., 2014). The intersection between these datasets includes eight genes encoding the following proteins (Fig. 5): RZR66026.1 (secreted aspartic proteinase precursor, overexpressed in PcQ and PcRKNQ), RZR62042.1 (helix-loop-helix DNA-binding protein, overexpressed in PcRKNQ), RZR63158.1 (peptidase A1, overexpressed in PcQ), RZR64159.1 (major facilitator superfamily domain, general substrate transporter, overexpressed in PcRKNQ), RZR62046.1 (polyketide synthase (PKS), overexpressed in PcRKNQ), RZR61625.1 (cytochrome P450 ClCP1, overexpressed in PcQ and PcRKNQ), RZR69240.1 (APSES transcription factor, overexpressed in PcQ) and RZR61845.1 (glycoside hydrolase family 75 protein (chitosanase) overexpressed in PcQ and PcRKNQ). We have investigated the expression of these genes in banana plants colonized by Pc. Banana plants modify the expression of these Pc genes ( Supplementary Fig. 4). Both barley and banana are monocots, and the former has been  Pc acidic peptidases modify their expression when banana plants are present. Secreted aspartic proteinase precursor gene is induced, while the A1 peptidase gene A. Cluster related to cell cycle and associated processes. B. Redox and transport. C. Redox and transport metabolism. D. Includes GO terms related to the previous step to energy production. E. Lipid metabolism. F. All GO terms related to chitin/chitosan degradation. G. Phosphate pentose shunt and NADPH regeneration GO terms. H. This cluster is represented by one GO term: metallocarboxypeptidase activity. I. Associated with cell wall-associated processes. Legend is shown in Fig. 3A. is slightly repressed. A chitosanase gene is expressed whenever chitosan is present. The fungus does not modify the expression of this gene when it grows endophytically on bananas, but the maximum expression occurs in the 'artificial rhizosphere' (liquid surrounding roots) with chitosan. One Pc PKS protein, involved in the secondary metabolism of the fungus, is slightly induced by the presence of the plant. Secondary metabolism plays an essential role during plant-host interaction (Macheleidt et al., 2016). Endophytism in Pc also modifies the expression of transcription factors. Helixloop-helix DNA binding protein is overexpressed in all treatments. Cytochrome P450 ClCP1 is related to oxidative metabolism, which is also involved in this process.

Discussion
In this work, we have carried out a transcriptomic study, using RNA-seq, to determine Pc genes involved in the response to chitosan and RKN parasitism. We have found that chitosan on its own modifies the expression of more Pc genes than RKN does. This may be because chitosan permeabilizes plasma membrane in fungi triggering the activation of reactive oxygen species (ROS) and cell death (Lopez-Moya et al., 2019). Besides, chitosan solutions reach all fungal cells, whereas nematodes are in contact with only specific parts (mainly appressoria) of the fungus. Chitosan enhances redox processes, proteolysis and carbohydrate metabolism in Pc. Redox metabolism is the most affected process by chitosan. To this respect, the gene encoding CipC1 (concanamycin induced protein C), a ROS-related protein, is the most induced by chitosan. This protein has also been found induced in citrus fruits infected by the fungal pathogen Penicillium digitatum (López-Pérez et al., 2015). CipC1 is involved in hyphal branching and upregulated in Laccaria bicolor in response to Pseudomonas fluorescens (Deveau et al., 2007). This gene could be overexpressed in Pc to overcome chitosan-induced ROS. Other genes overexpressed in chitosan-induced ROS are Cytochromes P450 and b561, L-ascorbate oxidase and aldo/keto reductase. This enhanced activation of redox metabolism-related genes connects resistance with chitosan-sensitive fungi (Jaime et al., 2012;Lopez-Moya et al., 2016). Pc has to fight plasma membrane oxidation generated by chitosan-induced ROS. Chitosan permeabilizes the plasma membrane of fungi in an energy-dependent manner (Palma-Guerrero et al., 2009). In our work, Pc activates energy and lipid metabolism genes with chitosan. This was previously found on chitosan-sensitive yeasts and filamentous fungi (Jaime et al., 2012;Lopez-Moya et al., 2016). Lipid oxidation generates oxylipins (Gabbs et al., 2015), which can modify fungal morphology, promoting appressoria differentiation (Niu et al., 2020). We then may speculate that Pc oxylipins could be involved in chitosan induction of Pc appressoria found previously . ROS increase in the cell could lead to Pc methylation (Wu and Ni, 2015), a highly represented GO term in chitosan treatments. This indicates chitosan may also trigger Pc epigenetic mechanisms (Razin and Cedar, 1991;Phillips, 2008). ROS also causes damage to proteins, probably activating protease induction (Schieber and Chandel, 2014). To this respect, chitosan induces Pc genes encoding all families of peptidases. In our study, chitosan enhances Pc subtilisin S8/S53 expression. This protease may help the fungus to adapt to the ecological niches, facilitating nutrition (Li et al., 2017). Pc subtilisins have been detected in both the parasitic  and endophytic  phases of the fungal lifestyle. Peptidases A1 and A4 (fungal family G1), involved in plant parasitism and host adaptation (Kirshnan et al., 2018), are also upregulated in Pc with chitosan. Finally, Pc metallo-endopeptidases are also upregulated with chitosan. This wide range of biological processes and molecular functions affected by chitosan indicates the effort made by the cellular machinery of Pc to overcome stress. Further work should quantify the response of individual Pc cells against chitosan, since antibiotics generate dead, dying and resistant active and inactive cells (Bamford et al., 2017).
Pc has mechanisms to efficiently degrade chitosan and thus reduce its damage (mainly caused by ROS) on cells. Pc overexpresses chitosanases (GH75) with chitosan, degrading it into monosaccharides. One WSC domain-containing protein, a stress responsory related to carbohydrate binding (Tong et al., 2016;Oide et al., 2019), is also upregulated with chitosan. Genes coding for membrane transporters, such as the major facilitator superfamily domain, general substrate transporter, oligopeptide transporter OPT-like protein, maltose permease and sugar transporter family protein, are also overexpressed with chitosan. They may be involved in the assimilation of monosaccharides generated upon chitosan degradation by Pc. Neurospora crassa, a chitosan-sensitive fungus, also shows the activation of monosaccharide transport genes (Lopez-Moya et al., 2016) upon chitosan treatment but not as many as Pc. This may be a further reason to explain why Pc is more resistant to chitosan than N. crassa. Overexpression of Pc glucokinase in the presence of chitosan is probably related to the final catabolism of this polymer (Maitra and Lobo, 1983). Other glycoside hydrolases (GH2 and 3) are also overexpressed in chitosan treatments. This may indicate the high metabolic potential of Pc to degrade and assimilate chitosan. RKN eggs stimulate Pc to overexpress rather than repress genes. RKN also induce gene expression of nematode-trapping fungi, such as Arthrobotrys conoides (Pandit et al., 2017), indicating the activation of parasitic pathways. In this work, the addition of chitosan to nematode eggs displays the highest values of Pc gene expression in all treatments tested. This shows that chitosan is a strong elicitor of genes potentially involved in RKN egg parasitism by Pc. In cluster analyses, genes encoding proteins for adhesion (FLO1), chitosan and sugar degradation (GH2, GH3, GH75), membrane transport (MFStransporters) and carbohydrate metabolism (glucokinase) are overexpressed in chitosan and RKN treatments. In a previous study (Lin et al., 2018), RKN eggs in Minimal Medium (MM) were found to induce Pc adhesives (CFEM), GHs enzymes and proteases. In our GO cluster analysis, we have found GO terms related to chitin/ chitosan degradation with an enhanced expression upon addition of nematode eggs and chitosan. Genes that share this behaviour may explain why RKN eggs parasitism by Pc increases in the presence of chitosan. The trend pattern (increase or decrease in expression by adding chitosan to treatments) is best observed in a gene rather than in GO cluster analyses. This is because GOs share genes and when some have opposing trends, the average GO behaviour does not show up in the cluster analysis. Pc deploys its machinery to putatively attach to RKN eggshell by binding peptides, lipids and carbohydrates. FLO1 proteins are flocculation proteins present in yeasts, related to adhesion to hyphae (Moreno-García et al., 2018). FLO1 is a mannose-binding glycoprotein, which could be a determinant for hyphal adhesion to the nematode eggshell. A CRAL/TRIO domain protein is upregulated when Pc is in contact with RKN. This domain is related to the binding to small lipophilic molecules (Panagabko et al., 2003). It could also be involved to the attachment to the eggshell lipid layer (Johnston and Dennis, 2012). We hypothesize that once the fungus is attached to the eggshell, its degradation begins by transforming the RKN egg chitin layer (Johnston and Dennis, 2012) to chitosan using chitindeacetylases (Aranda-Martinez et al., 2016) and degrading the resulting chitosan mainly using chitosanases (GH75). Other glycosyl hydrolases (GH2 and GH3) may also contribute to degradation. Sugars may be introduced to the cells of the fungus through transporters (MFS). All molecular processes involved in egg parasitism are enhanced by the addition of chitosan. The increase in ROS could explain this behaviour.
Pc can act both as an endophyte (Bordallo et al., 2002) or RKN egg parasite . In this work, we have explored the common 'gene toolbox' involved in endophytism (Larriba et al., 2014), pathogenicity (PHI-base) and response to chitosan. We have found eight candidate genes, among them proteases, chitosanases, redox-related proteins and transcription factors.
Proteases modify their expression in endophytic processes. This is related to parasitism in other fungi (Druzhinina et al., 2012) and to endophytism in insect pathogenic fungi (Moonjely et al., 2016). In works of gene expression during endophytism, it has been found that Pc overexpresses ribosomal proteins, proteases, secreted proteins and heat shock proteins, among others probably related to transitions in the lifestyle of the fungus (Pentimone et al., 2019). The fact that chitosan increases the secretion of Pc proteases could explain why the fungus colonizes the plant efficiently when chitosan is present. Pc does not modify the expression of GH75 when it grows endophytically on bananas, but the maximum expression of this gene occurs in the rhizosphere with chitosan. Endophytic fungi have enzymes that modify chitin and chitosan (Govinda Rajulu et al., 2011;Venkatachalam et al., 2015). This could mean that the fungus enhances its chitin-and chitosanolytic metabolism in order to start plant root colonization. Chitosan and its derivatives are highly related to redox metabolism (Sarangapani et al., 2018;Ivanova and Yaneva, 2020), as we commented before. Cytochrome P450 ClCP1 is also involved in oxidative metabolism (Korzekwa, 2014), and it is expressed in endophytes (Chadha et al., 2018). This is consistent with studies which prove that root colonization processes activate ROS (Segal and Wilson, 2018). One Pc PKS protein is slightly induced by the presence of the plant. PKS are expressed in biocontrol fungus such as Clonostachys rosea during fungalfungal interactions (Fatema et al., 2018) and virulence events (Tsai et al., 1998). This gene is also related to melatonin biosynthesis (Knapp et al., 2018). Melatonin is a precursor of plant hormones (Arnao and Hernández-Ruiz, 2018). It has been shown that endophytic organisms secrete melatonin and derivatives during root colonization (Jiao et al., 2016). This could mean that Pc is secreting metabolites homologous to plant hormones in order to facilitate plant root colonization. This could be one of the reasons why Pc increases banana plant growth and development (Mingot-Ureta et al., 2020). Finally, endophytism in Pc modifies the expression of transcription factors. Helix-loop-helix DNA binding protein is overexpressed in all treatments. Related to previous results, Pc seems to be activating metabolic routes in order to colonize the plant properly hiding from plant defences.
In conclusion, Pc modifies its gene expression to parasite RKN eggs, colonize plant roots or resist chitosan. When Pc is growing with chitosan, it activates metabolic pathways to avoid chitosan-induced ROS damage. Pc enzymes capable of degrading it into sugars. Pc sugar carriers introduce them into the cell for catabolism. This process increases ROS content, and Pc activates lipid metabolism. Metal membrane transporters putatively carry out these reduction and oxidation reactions. During this process genes related to plant endophytism or nematode egg parasitism are activated, which could indicate that, in combination with the real stimulus, chitosan could be a non-toxic additive to increase plant colonization by Pc and sustainably reduce plant-parasitic nematodes, such as RKN, in banana and other agroecosystems.

Pc, chitosan and RKN: RNA-seq experimental design
Pc conidia (final concentration 10 6 conidiaÁml −1 ) were inoculated into 100 ml flasks each containing 20 ml Czapek Dox broth medium (Ward et al., 2012). Flasks were incubated at 25 C with shaking at 120 rpm. After 5 days, mycelia were recovered by filtration through Miracloth (Calbiochem) and washed twice with sterile distilled water. Pc mycelia (ca. 0.2 g) were inoculated axenically into 100 ml flasks each containing 20 ml MM (Aranda-Martinez et al., 2016) and amended with either: (i) chitosan (PcQ) (0.1 mg ml −1 final concentration), (ii) surface-sterilized RKN eggs (PcRKN) (1 egg μl −1 final concentration), or (iii) both (PcRKNQ). Controls consisted of Pc mycelium growing in MM (Pc). All treatments were carried out in triplicate. Flasks were incubated for 4 days as before. Samples were then filtered through Miracloth, frozen in liquid N 2 , lyophilized and stored at −80 C until used. The experiment was performed three times.

RNA extraction and RNA-seq performing
Total RNA extractions were performed using TRIzol reagent (Life Tech), following the manufacturer's protocol. The quality of all RNA samples was determined using a bioanalyzer (Agilent 2100 Bioanalyzer System) to confirm it was adequate for RNA-seq analysis (Supplementary Table 4). Three replicates per treatment were then selected at random to perform RNA-seq analysis. cDNA synthesis, library construction and Illumina sequencing were carried out by Macrogen (Seoul, South Korea). TruSeq Stranded mRNA LT Sample Prep Kit (Illumina) was used as Library Kit and TruSeq Stranded mRNA Sample Preparation Guide, Part #15031047 Rev. E as Library Protocol. The reagent used was NovaSeq 6000 S4 Reagent Kit (Illumina) and sequencing protocol NovaSeq 6000 System User Guide Document #1000000019358 v02.

Bioinformatic analyses
Raw reads were trimmed and filtered with Trimmomatic (Bolger et al., 2014) to remove adapters with up to two mismatches that had a palindrome read alignment accuracy of 30 and a sequence match accuracy of 10. Leading and trailing low-quality or N bases (<3) were removed by using a 4-base-wide sliding window. Where average base quality was low (<15) reads were trimmed and any short reads (<80 bp) were removed. The quality of the reads was then checked with FASTQC (http://www. bioinformatics.babraham.ac.uk/projects/fastqc/).
Reads were quantified using Salmon mapping against the P. chlamydospora 123 genome (GenBank accession: GCA_000411695.2) using a wrapper script (align_and_estimate_abundance.pl) from the Trinity software package (Grabherr et al., 2011). DEG were determined with the Bioconductor package DESeq2 (Love et al., 2014) using the likelihood ratio test. Genes were filtered so only those with at least three samples with >10 counts were analysed. P values were adjusted using the Benjamini-Hochberg (BH) correction (Benjamini and Hochberg, 1995). Genes were considered differentially expressed if they had a log 2 fold change of ±2 and a BH-FDR-adjusted P-value of ≤0.05. This process selected genes whose expression varied the most with treatments with respect to the control and characterized the main genes involved in the response of Pc to chitosan and RKN.

Pc-Chitosan-Musa: bioassays
Pc (10 6 conidiaÁml −1 ) was inoculated in Cz liquid medium and incubated for 5 days as described above. Thirty-six banana plantlets were placed individually in Magenta Boxes™ (Sigma) each containing 50 ml of MM supplemented or without chitosan (final concentration 0.1 mg ml −1 ). Half of the plants were inoculated with 0.2 g Pc mycelium and half were left uninoculated. Plants were maintained at 24 C, 60% relative humidity and 16:8 h light/darkness photoperiod, with 100 rpm shaking, for 4 days. Control replicates of 0.2 g of mycelium were inoculated in 100 ml flasks each containing 20 ml MM supplemented or not with chitosan (final concentration 0.1 mg ml −1 ). Controls with and without chitosan were made by triplicate to obtain replicability among the samples. To extract RNA, three plant roots from the same treatment were collected for each extraction. In this way, three replicates were obtained per treatment, each replicate with three whole roots from three different plants. RNA was extracted as described above. Final treatments were: Pc (Pc mycelium in MM), PcQ (Pc mycelium in MM amended with 0.1 mg ml −1 chitosan), PcB (Pc mycelium growing in MM close to banana roots), PcBQ (Pc mycelium growing in MM amended with 0.1 mg ml −1 chitosan close to Banana roots), BPc (banana roots colonized by Pc) and BPcQ (Banana roots colonized by Pc in medium amended with 0.1 mg ml −1 chitosan). To identify potential genes related to endophytism and pathogenicity, a whole-genome blast search was conducted against the Pathogen-Host Interaction database v. 4.9 (PHIbase; www.phi-base.org; Urban et al., 2020). These candidate genes were intersected with expressed genes when Pc colonizes Barley (Larriba et al., 2014) and upregulated genes (log 2 fold change values ≧2) in all treatments in this work (data available in Supplementary Information File). Selected genes were evaluated by qRT-PCR in banana treatments.

qRT-PCR
RNA was treated twice with DNase (Turbo DNA-free, Ambion) to remove any remnants of DNA in the samples. Then, cDNA was obtained using NZY First Strand cDNA Synthesis Kit (NZYtech). Finally, qRT-PCRs were performed in a StepOnePlus™ Real-Time PCR System machine, using SYBR Green with ROX (Roche) and a ΔΔCt methodology. qRT-PCR analyses included three biological replicates with three technical replicates each. ΔΔCt method was used to calculate the relative fold gene expression of samples and statistical analyses were performed using ANOVA in GraphPad Prism 7.0 Software (www.graphpad.com). Primers used for qRT-PCRs are shown in Supplementary Table 5. Pochonia chlamydosporia allantoate permease (RZR69578.1; Rosso et al., 2014), glyceraldehyde-3-phosphate dehydrogenase (RZR61537.1; Escudero et al., 2016) and ßtubulin (RZR65128.1; Ward et al., 2012) were used as housekeeping genes.

Statistical analyses and figures
Statistical analysis for differential gene expression was determined using DESeq2 (Love et al., 2014). Statistics for gene set enrichment analysis were performed using OmicsBox (BioBam, Spain) and figures generated using GraphPad Prism version 7.00 for Mac, GraphPad Software, La Jolla, California, USA, (www.graphpad.com).