A haustorial‐expressed lytic polysaccharide monooxygenase from the cucurbit powdery mildew pathogen Podosphaera xanthii contributes to the suppression of chitin‐triggered immunity

Abstract Podosphaera xanthii is the main causal agent of cucurbit powdery mildew and a limiting factor of crop productivity. The lifestyle of this fungus is determined by the development of specialized parasitic structures inside epidermal cells, termed haustoria, that are responsible for the acquisition of nutrients and the release of effectors. A typical function of fungal effectors is the manipulation of host immunity, for example the suppression of pathogen‐associated molecular pattern (PAMP)‐triggered immunity (PTI). Chitin is a major component of fungal cell walls, and chitin oligosaccharides are well‐known PAMP elicitors. In this work, we examined the role of PHEC27213, the most highly expressed, haustorium‐specific effector candidate of P. xanthii. According to different computational predictions, the protein folding of PHEC27213 was similar to that of lytic polysaccharide monooxygenases (LPMOs) and included a conserved histidine brace; however, PHEC27213 had low sequence similarity with LPMO proteins and displayed a putative chitin‐binding domain that was different from the canonical carbohydrate‐binding module. Binding and enzymatic assays demonstrated that PHEC27213 was able to bind and catalyse colloidal chitin, as well as chitooligosaccharides, acting as an LPMO. Furthermore, RNAi silencing experiments showed the potential of this protein to prevent the activation of chitin‐triggered immunity. Moreover, proteins with similar features were found in other haustorium‐forming fungal pathogens. Our results suggest that this protein is a new fungal LPMO that catalyses chitooligosaccharides, thus contributing to the suppression of plant immunity during haustorium development. To our knowledge, this is the first mechanism identified in the haustorium to suppress chitin signalling.


| INTRODUC TI ON
Podosphaera xanthii is the main causal agent of cucurbit powdery mildew, a disease that causes important yield losses in cucurbit crops (Bellón-Gómez et al., 2015;Fernández-Ortuño et al., 2006;Pérez-García et al., 2009;del Pino et al., 2002). Like all powdery mildew fungi, P. xanthii is dependent on living plant cells to complete its asexual life cycle (Martínez-Cruz et al., 2014;Vogel & Somerville, 2002;Weßling et al., 2012). In this cycle, conidia transported by wind are deposited onto the leaf of a susceptible host plant. Subsequently, conidial adhesion, germination, and penetration are necessary steps for disease establishment (Spanu et al., 2010). After penetration, the fungus forms a specialized parasitic structure inside plant epidermal cells called the haustorium, which is responsible for the exchange of factors with the plant, such as the acquisition of nutrients (Bindschedler et al., 2009;Both et al., 2005;Martínez-Cruz et al., 2014;Micali et al., 2011). However, to complete this cycle, the pathogen needs to avoid the action of plant defence elements, such as enzymes (van den Burg et al., 2007;Delaunois et al., 2014) and receptors, that recognize pathogen-associated molecular patterns (PAMPs) (de Jonge et al., 2011;Tanaka et al., 2010), activating the so-called PAMP-triggered immunity (PTI) (Pieterse et al., 2009).
For this reason, fungal pathogens have had to evolve and adapt to their hosts by developing several strategies to overcome plant defence responses. In this way, they counter with effectors, small proteins acquired during the coevolution of plant-pathogenic fungi and their hosts (Pieterse et al., 2009), which, among other functions, prevent the recognition of PAMPs by plant receptors (Jones & Dangl, 2006;Lo Presti et al., 2015;Sánchez-Vallet et al., 2013).
A major component of the fungal cell wall and a well-known PAMP is chitin (de Jonge et al., 2010;Kombrink & Thomma, 2013;Pieterse et al., 2009;Tanaka et al., 2013), which is a long-chain polymer of β-1,4-N-acetylglucosamine, a derivative of glucose (de Jonge et al., 2010;Liu, Li et al., 2012;Wan et al., 2008;Young et al., 2005). Chitin provides structural rigidity to the fungal cell wall and acts as the first line of defence of pathogenic fungi against plant-secreted enzymes, such as chitinases (Kombrink & Thomma, 2013;Wan et al., 2008). As a consequence of the enzymatic activity of plant chitinases, chitin oligosaccharides are released from the fungal cell wall and can be recognized by the plant receptor CERK1, a transmembrane LysM-containing receptor with an intracellular kinase domain, thereby promoting chitinspecific signalling (Cao et al., 2014;Miya et al., 2007;Sánchez-Vallet et al., 2013). This signalling induces the activation of several plant defence mechanisms, including the accumulation of reactive oxygen species (ROS) and cell wall deposits, such as lignin and callose, that provide cell wall reinforcements (van den Burg et al., 2007;Doehlemann & Hemetsberger, 2013;Kaku et al., 2006;Mentlak et al., 2012).
To counter chitin-triggered immunity, several effectors have been described in phytopathogenic fungi that play roles in avoiding chitin oligosaccharide recognition by plant receptors. One of these effectors is Avr4, a Cladosporium fulvum apoplastic effector, which protects fungal cell wall chitin from the action of plant chitinases released during the infection process (Bolton et al., 2008;van den Burg et al., 2007).
Other effectors include Ecp6 and Slp1, proteins secreted by C. fulvum and Magnaporthe oryzae, respectively. These proteins sequester the free chitin oligosaccharides released as a consequence of the activity of plant chitinases, thus avoiding their recognition by the host. Another mechanism involved in the suppression of chitin-triggered immunity is the action of the chitin deacetylase (CDA) enzyme. CDA is a widely conserved enzyme that catalyses the hydrolysis of the acetamido groups of N-acetylglucosamine in chitin, promoting their conversion to chitosan, a glucosamine polymer and deacetylated chitin derivative that shows a considerably lower degree of immune elicitation than chitin (Mochizuki et al., 2011;Sánchez-Vallet et al., 2013;Xi et al., 2014).
The suppression of PAMP-triggered immunity by powdery mildew fungi has been a poorly investigated issue despite the fact that their nature as obligate biotrophs implies that the suppression of host defensive response activation should be a key aspect of their physiology.
The first evidence in this regard has been the recent identification of effectors with chitinase activity (EWCAs), a family of conserved chitinases, secreted mainly by hyphae, that suppress chitin signalling by catalysing immunogenic chitooligosaccharides (Martínez-Cruz et al., 2021). In the haustorial transcriptome of P. xanthii, a unigene encoding a small secreted protein without an annotated function but predicted by protein modelling to be a putative lytic polysaccharide monooxygenase (LPMO) with a chitin-binding domain, was found among the top 50 expressed genes; it was the most highly expressed gene among those encoding proteins specifically expressed in the haustorium (Polonio, Seoane, et al., 2019). LPMOs are a class of recently characterized enzymes that are able to oxidize different recalcitrant polysaccharides, including chitin Vaaje-Kolstad et al., 2010). Chitin LPMOs act on the crystalline chitin surface, introducing chain breaks and generating oxidized chain ends (Vaaje-Kolstad et al., 2010). These enzymes are part of a pool of enzymes that different organisms secrete to obtain energy from dead biomass (Hemsworth et al., 2014). However, to date, LPMOs have not been reported in plant-pathogenic or biotrophic fungi. Considering the putative role of this protein in chitin modification, as well as its high and exclusive expression in the haustorium, in this work we analysed the role of this putative LPMO in powdery mildew pathogenesis using computational approaches, experiments with purified recombinant proteins, and RNAi silencing experiments. Our results suggest that this effector could play a role in the catalysis of the chitin oligosaccharides released during the development of haustoria by plant endochitinases, thus avoiding the perception of chitin by the host plant and thereby allowing the development of haustoria inside plant epidermal cells.

| The protein folding of PHEC27213 is similar to an LPMO and it contains a putative chitin-binding domain
The unigene PHEC27213 was selected from the P. xanthii haustorial transcriptome because it was the most highly expressed, haustorium-specific gene encoding a secreted protein. However, because PHEC27213 lacks an annotated function or domains, the F I G U R E 1 Main features predicted for the PHEC27213 protein. (a) Predicted three-dimensional (3D) models of Podosphaera xanthii PHEC27213 performed by the I-TASSER, IntFOLd, and Phyre2 servers, as well as the most similarly folded proteins. The structural alignments of both 3D models are also shown. (b) Amino acid sequence of the putative chitin-binding domain of PHEC27213 predicted by MotifScan. Conserved residues are shown in the box. (c) Sequence alignment of PHEC27213 and several AA10 chitin lytic polysaccharide monooxygenase (LPMO) proteins retrieved from the Protein Data Bank, 5FJQ from Cellvibrio japonicus, 4A02 from Enterococcus faecalis, and 2YOW from Bacillus amyloliquefaciens. The amino acid similarity is shown in grey, with the darkest grey amino acids being the most similar. Conserved amino acids are marked with an asterisk. The putative chitin-binding domain, corresponding to several conserved residues from CBM33 of the AA10 chitin LPMOs, is shown in a red box amino acid sequence corresponding to the mature protein, without the signal peptide, was used to perform 3D modelling using the I-TASSER, Phyre2, and IntFOLD servers to elucidate the putative function of PHEC27213. In all cases, the resulting 3D models were similar ( Figure 1a). These models were used as templates to perform protein fold recognition analyses. The I-TASSER model matched with high confidence with an AA11 LPMO from Aspergillus oryzae (PDB code 4MAH) (Figure 1a, Table 1), whereas the Phyre2 and IntFOLD models matched with high confidence with an LPMO-like protein from Laetisaria arvalis (PDB code 6IBH) ( Figure 1a, Table 1). However, there was only 15.4% and 21.8% sequence identity between PHEC27213 and the A. oryzae and L. arvalis proteins, respectively ( Figure S1).
On the contrary, it was not possible to detect any complete carbohydrate-binding module (CBM) characteristic of the LPMO proteins using the Pfam or dbCAN2 servers. However, using MotifScan software, a putative chitin-binding domain 3 (Pfam ID = Chitin_ bind_3) was located from amino acids 115 to 128, corresponding to some residues of CBM33 from the AA10 LPMO (Figure 1b,c).
Thus, PHEC27213 appears to be a protein with a typical LPMO histidine brace and folding similar to LPMO proteins, while it lacks the full CBM domain that is present in canonical LPMOs; instead, it has only a few residues of this domain, which seem to be related to chitin binding.
The presence of this putative chitin-binding domain and the specific expression of this protein in the haustorium suggest that PHEC27213 could interact with the chitin from the haustorial cell wall of P. xanthii.

| PHEC27213 shows chitin-binding activity and breaks chitin into small oligosaccharides
To validate the function of PHEC27213, the protein was expressed in vitro as a 6 × His-tagged fusion in the Escherichia coli Bl21-CodonPlus-RIL. PHEC27213 was predominantly found in the soluble fraction and was obtained after purification by immobilization on a nickel affinity column with a yield of 1.34 mg/ml of soluble protein (Figure 2a). The polysaccharide-binding ability of the PHEC27213 protein was studied via a binding assay using colloidal chitin and cellulose as polysaccharides and bovine serum albumin (BSA) as a negative control. Only approximately 20% of the total soluble protein was present in the supernatant after exposure to chitin and approximately 70% was present after exposure to cellulose, indicating a high binding of PHEC27213 to chitin. By contrast, BSA was fully recovered from the supernatant, indicating the absence of binding ( Figure 2b). These samples were also analysed by western blot analysis. In the presence of colloidal chitin, most of the PHEC27213 was retained in the pellet. However, while PHEC27213 was also observed in the pellet in the presence of cellulose, the protein was mainly detected in the supernatant (Figure 2c).
The putative function of PHEC27213 was also studied in terms of chitin and cellulose degradation. For this purpose, the protein was incubated overnight with colloidal chitin or cellulose and ascorbic acid as a reducing agent, and after incubation the presence of free oligosaccharides was analysed by matrix-assisted laser desorption ionization-time of flight-mass spectrometry (MALDI-TOF-MS).
In the reaction of PHEC27213 with chitin, small oligosaccharides were detected (Figure 3a), whereas in the same reaction without PHEC27213 (Figure 3b) or without ascorbic acid (Figure 3c), it was not possible to detect small chitin oligosaccharides; because of their low solubility, longer oligosaccharides were never observed. In the reaction with chitin, among the small oligosaccharides, those with a degree of polymerization of 5 (DP5) were the most highly represented ( Figure 3a). In addition, smaller peaks corresponding to DP4 and very small peaks corresponding to DP6 and DP7 were also detected. In the cases of DP6 and DP7, they presented a difference c TM-score values are known standards for measuring the structural similarity between two structures, which are usually used to measure the accuracy of structure modelling when the native structure is known. In I-Tasser and IntFOLD, TM-scores are in the range of 0 to 1, being 1 a perfect match between models. In Phyre2, TM-scores range between 0% and 100%. d Coverage represents the coverage of the alignment by TM-align and is equal to the number of structurally aligned residues divided by length of the query protein.  Figure S2). To check the number of oxidations that may occur during chitin catalysis by PHEC27213, the unoxidized standard oligosaccharide corresponding to the most predominant peak detected after enzymatic assays with PHEC27213, that is, DP5, was also analysed by MALDI-TOF-MS, which showed an m/z difference that corresponded with a single oxidation ( Figure 3d).
Moreover, to test the ability of PHEC27213 to catalyse oligosaccharides, a reaction with unoxidized DP7 (Figure 3e) was also carried out. An analysis of the reaction products showed that indeed PHEC27213 was able to release, predominantly, DP5 oligosaccharides ( Figure 3f). In the case of cellulose, no peaks corresponding to oligosaccharides released from cellulose could be detected, indicating that PHEC27213 did not catalyse cellulose ( Figure S3). These results revealed the ability of the PHEC27213 protein to bind and catalyse chitin via a single oxidation. With such activity and its predicted structure as an LPMO, the protein was renamed PxLPMO1 (P. xanthii lytic polysaccharide monooxygenase 1).

| Molecular docking suggests a binding site for the chitin heptamer near the histidine brace active site
The putative chitin-binding site was computationally assessed by molecular docking using the best 3D model of PHEC27213 (PxLPMO1), that is, that predicted by IntFOLD (Table 1), and the

| PxLPMO1 is coexpressed with two plant endochitinase genes
To elucidate the physiological function of PxLPMO1, its expres-

| RNAi silencing of the PxLPMO1 gene reduces P. xanthii development and activates an oxidative burst
The high expression of PxLPMO1, which was the 13th most highly expressed gene in the haustorium and the most highly expressed  191, 190, 189, and 118 (yellow), which allow the interaction of the chitin heptamer with residues 1 and 52 of the histidine brace active site (green) gene among the genes encoding haustorium-specific secreted proteins (Polonio, Seoane, et al., 2019), suggested that it plays an important role in P. xanthii biology. To validate the role of PxLPMO1, the Agrobacterium tumefaciens-mediated host-induced gene silencing (ATM-HIGS) assay was used. To quantify fungal growth after gene silencing, two approaches were used: haustorial counts by light microscopy and a molecular approach by quantitative PCR (qPCR). The activation of plant defence responses was also studied. In this regard, the production of reactive oxygen species, such as hydrogen peroxide (H 2 O 2 ), was histochemically examined. The efficacy of ATM-HIGS was studied by RT-qPCR, showing that the levels of the CmMLO1 (the positive control for RNAi-induced resistance) and PxLPMO1 transcripts decreased by approximately 50% during RNAi gene silencing experiments ( Figure S4). As shown in Figure 6, after silencing PxLPMO1, the development of P. xanthii was clearly altered and delayed, and this decrease was evident 72 hr after inoculation compared with the negative control (empty vector). In parallel, a strong accumulation of H 2 O 2 was observed that was even higher than that observed in the samples corresponding to the CmMLO1 RNAi-positive control (e) Effect of the infiltration of the products from the PxLPMO1 (PHEC27213) enzymatic reaction on the oxidative burst. The supernatants from the reaction products described in Figure 3a were infiltrated into melon cotyledons, and then the samples were processed for detection of H 2 O 2 by the DAB uptake method. As positive and negative controls for the production of oxidative burst in leaf tissue, the products of the same reactions without PxLPMO1 or without colloidal chitin were used, respectively the supernatant of the reaction mixture without PxLPMO1 ( Figure 6e).

| RNAi silencing of the PxLPMO1 gene activates chitin-triggered immunity
The ability of the purified PxLPMO1 protein to catalyse chitin combined with the fact that the supernatant of PxLPMO1 reaction products was not capable of triggering an oxidative burst in leaf tissue suggested that the activity of PxLPMO1 was related to preventing the activation of chitin-triggered immunity. To

| PxLPMO1 orthologues and putative LPMOs are present in many pathogenic ascomycete fungi and in other haustorium-forming fungal pathogens
The presence of PxLPMO1 orthologues was studied by BLASTp (BLAST + v. 2.7.1; E value < 1E−5) analysis using the deduced amino acid sequence of PxLPMO1 as a query sequence. As shown in In all powdery mildew fungi examined, including P. xanthii, as well as in the poplar rust fungus M. larici-populi, at least one PxLPMO1-like protein was found ( Figure S5 and Table 3). However, in the case of species of the genus Puccinia, none was found. These proteins showed common features with PxLPMO1 (similar folding, small size, presence of a signal peptide, the typical histidine brace, and the presence of conserved residues corresponding to chitin-binding domain 3), but they only had sequence identities with PxLPMO1 of 20%-36% ( Figure S5b and Table 3). This lack of identity is largely due to the high sequence vari-  (Table 3). Interestingly, phylogenetic analysis of these proteins showed that the LPMO-like proteins from haustorium-forming fungal pathogens form an independent clade and are phylogenetically separated from bacterial AA10 and fungal AA11 chitin LPMOs ( Figure S5d).
Finally, BLASTn searches of the P. xanthii genomes showed the presence of two copies of PxLPMO1 that were identical in the two available genomes. The paralog A (PxLPMO1A) that was the object of this study was found in the JACSEY010001314.1 and JAAAXZ010000022.1 scaffolds, while the paralog B (PxLPMO1B) was found in the JACSEY010001350.1 and JAAAXZ010000055.1 scaffolds. Compared to PxLPMO1A, PxLPMO1B showed seven nucleotide changes that caused four amino acid substitutions, including one in the signal peptide (F2S) and two in the highly variable C-terminus (R188G and G194W) ( Figure S5e).

| D ISCUSS I ON
As an obligate biotroph, P. xanthii requires living cells to complete its asexual life cycle, which ultimately implies that it must suppress the activation of plant defence responses. The mechanisms by which P. xanthii avoids recognition by the host remain largely unknown.
For this purpose, P. xanthii, among its other abilities, has to hide its PAMPs or manipulate their detection to suppress the activation of PTI. These activities are most likely carried out by the secretion of effectors. In other fungal pathogens, such as C. fulvum or M. oryzae, effector proteins with the ability to suppress chitin-triggered immunity have been described (Bolton et al., 2008;van den Burg et al., 2007;Jonge et al., 2010;Sánchez-Vallet et al., 2013;Xi et al., 2014). In the case of P. xanthii, the recent discovery of an effector family with chitinase activity (EWCAs) has revealed a new mechanism to suppress chitin signalling that consists of the degradation of immunogenic oligosaccharides, effectors that are also present in other powdery mildews and many fungal pathogens (Martínez-Cruz et al., 2021). However, in the case of haustorium-forming pathogens, to date no specific mechanisms associated with the haustorium have been identified to suppress chitin-triggered immunity, despite the fact that this ability should be essential for the survival of this cell because it is the fungal structure that maintains the most intimate relationship with the host cells (Bindschedler et al., 2009;Both et al., 2005;Martínez-Cruz et al., 2014;Micali et al., 2011). In this regard, the absence of any known mechanism related to this purpose in the most extensively studied powdery mildew species, Blumeria graminis, is striking, even though several studies have indicated that numerous candidate effectors expressed in haustoria become engaged in a coevolutionary arms race with the innate immune system of the host (Hacquard, 2014;Pedersen et al., 2012). Thus, proteomic studies in B. graminis revealed how its haustoria are enriched with small secreted proteins, including proteins involved in carbohydrate metabolism (Bindschedler et al., 2009(Bindschedler et al., , 2011Godfrey et al., 2009) that can act in a similar way. However, PHEC27213 displayed a putative chitin-binding domain (Chitin_bind_3) that corresponded to several residues of the CBM33 from AA10 LPMO. The chitin-binding activity of PHEC27213 was confirmed by a binding assay, which also showed its ability to bind cellulose to a lesser extent. On the contrary, an enzymatic assay showed that PHEC27213 was active only against chitin, catalysing it into small chitooligosaccharides. Therefore, the PHEC27213 protein was renamed PxLPMO1 (P. xanthii LPMO1). Furthermore, MALDI-TOF-MS analysis of the enzymatic reaction products showed that the product masses were consistent with the cleavage of the primary chain by C1 oxidation, producing predominantly aldonic acid oligosaccharides, as has been previously described for AA10 and AA11 LPMOs ( predominantly showed a degree of polymerization of 5, which is different from that previously described for AA10 and AA11 LPMOs.  Lee et al., 2013;Miya et al., 2007;Petutschnig et al., 2014). released from the haustorial cell wall by plant endochitinases, catalysing them into small oligosaccharides and thus avoiding chitin recognition by CmCERK1, thereby suppressing the activation of chitin-triggered immunity (Figure 8). In this model, we anticipate that CmCERK1 should be located in the extrahaustorial membrane, as previously described for other plant receptors responsible for pathogen recognition, such as RPW8.2, which enhances hydrogen peroxide accumulation and callose encasement of the haustorial complex (Kim et al., 2014;Wang et al., 2009) of LPMO proteins that could be specifically related to the suppression of chitin recognition by the host. These LPMO-like proteins present low sequence identities with PxLPMO1; however, they have the same structure, that is, they are small proteins with a signal peptide, they have the histidine brace, they have a similar fold, they have a putative chitin-binding domain, and they have a highly variable C-terminal domain. Although it is tempting to speculate that LPMO-like proteins from haustorium-forming fungal pathogens could act similar to PxLPMO1, the putative membrane localization and predicted GPI-anchors of some of them differs from PxLPMO1, which suggests that they could carry out different functions according to their C-terminal domains.
This fact could explain the binding of chitin by the C-terminal domain of PxLPMO1 that the LPMO-like proteins located in the membrane could not perform. This is the case of other LPMO-like proteins, such as the LPMO-like proteins from the ectomycorrhizal fungus Laccaria bicolor (Labourel et al., 2020) or from the yeast pathogen Cryptococcus neoformans (Garcia-Santamarina et al., 2020), both belonging to the X325 family, as well as the 6BIH protein from L. arvalis, which has a protein fold similar to PxLPMO1. These proteins are essential for symbiosis and pathogenesis, respectively, but they are GPI-anchored proteins and carry out functions different from canonical LPMOs. In the case of L. bicolor, the protein acts as a chitin-reorganization protein, but it is incapable of catalysing chitin, while the C. neoformans protein acts primarily as a copper acquisition protein and is incapable of catalysing cellulose (Garcia-Santamarina et al., 2020;Labourel et al., 2020). However, these observations do not exclude the possibility that they exhibit enzyme activity on other yet unidentified polysaccharide substrates.
To conclude, our findings show the presence of a novel class of chitin LPMOs in P. xanthii with orthologues in different ascomycete ectomycorrhizal and fungal pathogens. We also suggest F I G U R E 8 Schematic representation of the proposed role of PxLPMO1. (a) In the absence of PxLPMO1, plant-secreted endochitinases release chitin fragments from the haustorial cell wall that can be recognized by the CmCERK1 receptor, activating chitin-triggered immunity (PTI). (b) When PxLPMO1 is released into the extrahaustorial matrix, the enzyme catalyses the transformation of immunogenic chitin fragments into small chitooligosaccharides, predominantly DP5 oligosaccharides, that cannot induce CmCERK1 dimerization, thereby suppressing the activation of PTI. To reduce complexity, other components of the fungal cell wall have been omitted, such as β-glucans and cell wall proteins the existence of LPMO-like proteins in haustorium-forming fungal pathogens that can carry out different functions related to fungal development or pathogenesis, according to their C-terminal domains. Regarding pathogenesis, we demonstrate the involvement of LPMO enzymes in the suppression of chitin signalling, which reinforces the idea that the evolution of molecular strategies to disarm the activation of chitin-triggered immunity is mandatory for the successful colonization of plant environments by fungi, especially haustorium-forming fungal pathogens.

| Sequence analysis, protein modelling, and molecular docking
The unigene PHEC27213 was selected from the previously published P. xanthii secretome (Polonio, Seoane, et al., 2019). This unigene was the most expressed gene among the expressed haustorium-specific proteins; however, PHEC27213 had no annotated function. To analyse the signal peptide and select the mature protein sequence, the SignalP v. 4.1 server (Petersen et al., 2011) was used. To elucidate the putative function of PHEC27213, the mature protein sequence of PHEC27213 was employed to construct 3D models using the I-TASSER (Zhang, 2008), Phyre2 (Kelly et al., 2015), and IntFOLD (McGuffin et al., 2019;Roche et al., 2011) servers. The identification of domains in the mature protein was carried out by the Pfam (Sonnhammer et al., 1998), dbCAN2 (Zhang et al., 2018), and MotifScan (Pagni et al., 2007) servers, whereas the UniProt server (Apweiler et al., 2004) was used to perform protein alignments with several AA10 LPMO sequences retrieved from Protein Data Bank (PDB).
To identify the putative site of chitin binding to the 3D predicted model of PHEC27213, the chitin heptamer molecule (DP7) was taken from the PDB model 5GQB, a Lepidoptera-exclusive insect chitinase from Ostrinia furnacalis (Liu et al., 2017), and used to perform an automated molecular docking analysis using the SwissDock server (www.swiss -dock.ch/docking) (Grosdidier et al., 2011). The docking was performed using the "Accurate" parameter with default parameters otherwise and no region of interest defined (blind docking).
To identify melon leaf chitinases differentially expressed in response to P. xanthii infection, a previous RNA-Seq analysis of the early stages of melon powdery mildew disease was used (Polonio, Pineda, et al., 2019). In this study, the raw reads from the melon plants infected with P. xanthii were trimmed and aligned to the melon reference transcriptome to perform an expression analysis. Uniquely localized reads were used to calculate those differentially expressed genes between the control and infected plants. A p value <.05 and log 2 (fold change) >1 were considered the significance threshold for each gene. Two differentially expressed plant endochitinases, an acidic endochitinase (XM_008439336.2) and an EP3-like endochitinase (XM_008446389.2), were selected and used for expression analysis in conjunction with the PHEC27213 unigene.

| DNA and RNA isolation and cDNA synthesis
To isolate DNA and RNA from P. xanthii-infected zucchini cotyledons, the cotyledons were frozen in liquid nitrogen and ground with a mortar and pestle. Genomic DNA was isolated using the

| Construction of protein expression and RNAi silencing vectors
The plasmids used in this work are listed in Table S1 and are schematically represented in Figure S6.  (Karimi et al., 2002) and Gateway cloning technology (Invitrogen) were used as previously described (Martínez-Cruz et al., 2018). Specific primers with attB1 or attB2 tails (Table S2) were used to amplify a 315-bp fragment of PxLPMO1 from the previously obtained cDNA (see above). The resulting plasmid, pPxLPMO1-RNAi, was checked by PCR and sequencing. For RNAi silencing experiments, the pCmMLO1-RNAi plasmid, containing a 412-bp fragment of the melon CmMLO1 gene, which encodes a plant transmembrane protein involved in abiotic and biotic stresses and whose loss-of-function mutation protects plants from powdery mildew infection (Kusch & Panstruga, 2017), was used as a positive control (Martínez-Cruz et al., 2018), and the empty pB7GWIWG2(II) vector was used as a negative control (Table S1). In addition, for RNAi silencing of the melon chitin receptor kinase gene CmCERK1, the plasmid pCmCERK1-RNAi (Table S1)

| Protein expression and purification
For in vitro production of the PHEC27213 protein, E. coli BL21-

| Binding activity assays
Prior to binding assays, cellulose, and colloidal chitin solutions were prepared. In the case of cellulose, Avicel PH-101 crystalline cellulose was used. In the case of chitin, a suspension of colloidal chitin was prepared as described by Souza et al. (2009). Briefly, 5 g of chitin powder from shrimp shells (Sigma-Aldrich) was added to 60 ml of a solution of concentrated HCl and incubated with stirring overnight. This mixture was then added to 200 ml of previously cooled 95% ethanol and incubated overnight with stirring again.
The precipitate was centrifuged at 4 °C for 20 min at 5,000 × g and then filtered using filter paper. The resulting colloidal chitin was washed several times until reached a pH of 7 and was then stored at 4 °C in dark.
To perform binding assays, 250 μl of 1% solutions of colloidal chitin or cellulose were centrifuged at 13,000 × g for 5 min at 4 °C.
The resulting pellet was resuspended in 250 μl of 1 mg/ml purified PHEC27213 protein in 0.1 M sodium phosphate buffer (pH 7) or bovine serum albumen (BSA) (negative control). The mixtures were incubated at 4 °C for 1 hr with gentle manual agitation every 15 min.
Later, the mixtures were centrifuged at 13,000 × g for 5 min at 4 °C.
After centrifugation, the supernatants and pellets were separated, and the proteins present in the supernatant were visualized using Mini-PROTEAN Stain-Free Precast Gels (Bio-Rad) in a ChemiDoc XRS + system (Bio-Rad) and quantified as described above as an indicator of the proteins unbound to colloidal chitin or cellulose.

| LPMO activity assay
To validate the LPMO activity predicted for the PHEC27213 protein, chitin, and cellulose degradation assays were performed (Hemsworth et al., 2014). In these assays, 1 μM purified PHEC27213 was added to a reaction mixture containing 2 mg/ml colloidal chitin, hepta-N-

| RT-qPCR and qPCR
The expression analysis of P. xanthii and melon genes and the molecular estimation of P. xanthii biomass were carried out by RT-qPCR and qPCR, respectively. The primers used for these analyses (Table S2) were designed using Primer3 software (Koressaar & Remm, 2007;Thornton & Basu, 2011). For gene expression analysis, total RNA was extracted and used to synthetize cDNA as described above. As normalization reference genes, the P. xanthii translation elongation factor 1-alpha gene PxEF1 (MK249653) and the C. melo actin-7 gene CmACT7 (XM_008462689.2) were used (Polonio, Pineda, et al., 2019;Polonio, Seoane, et al., 2019). For the molecular estimation of fungal biomass, total DNA was isolated from agroinfiltrated and infected melon cotyledons as described above. For this purpose, the P. xanthii β-tubulin gene PxTUB2 (KC333362) and the C. melo actin-7 gene CmACT7 (XM_008462689.2) were quantified and the P. xanthii/C. melo genomic DNA ratio was calculated as previously described (Vela-Corcía et al., 2016). RT-qPCR and qPCR assays were carried out in a CFX384 Touch Real-Time PCR detection system (Bio-Rad) using SsoFast EvaGreen Supermix (Bio-Rad) according to the manufacturer's indications with the following conditions: enzyme activation step at 95 °C for 30 s, followed by 40 cycles at 95 °C for 5 s and 65 °C for 5 s. After amplification, the data were processed by CFX Manager Software (Bio-Rad), and the amplicon sizes were confirmed by visualization on 2% agarose gels.

| Agrobacterium tumefaciens-mediated hostinduced gene silencing (ATM-HIGS) assay
To study the role of PxLPMO1 in P. xanthii development, an ATM-HIGS assay was carried out as previously described (

| Haustorial counts and visualization of fungal development and oxidative bursts
For the quantification of the number of haustoria after RNAi silencing, the visualization of fungal development and to analyse the activation of oxidative burst (e.g., the accumulation of H 2 O 2 ), the 3,3′-diaminobenzidine (DAB) method (Thordal-Christensen et al., 1997) proposed by Martínez-Cruz et al. (2018) was performed.
Briefly, discs of 1 cm in diameter were taken from agroinfiltrated and P. xanthii-infected cotyledons and incubated in 1 mg/ml DAB (pH 3.8) overnight in the dark and at room temperature. After incubation, the discs were decoloured in boiling ethanol and observed by light microscopy using an Eclipse E800 microscope (Nikon). With these preparations, the haustoria can be visualized as black spots, whereas the hyphae are stained brown. In the same preparations, epidermal cells with brown-red precipitates are reactive cells with H 2 O 2 accumulation.

| Statistical analysis
When required, statistical analysis of data was carried out by IBM SPSS v. 20 software (SPSS) using Pearson's correlation coefficient or Fisher's least significant difference test (LSD).

ACK N OWLED G EM ENTS
We thank Irene Linares (University of Malaga, Spain) for her technical assistance. We are also grateful to Mercedes Martín from the Proteomics Service (University of Malaga) for the excellent technical support provided in MALDI-TOF-MS analysis. We also thank Jesús Hierrezuelo (University of Malaga) for his help in interpreting the MALDI-TOF-MS data. This work was supported by a grant from the Agencia Estatal de Investigación (AEI) (AGL2016-76216-C2-1-R), cofinanced by FEDER funds (European Union). A.P. was supported by a PhD fellowship (BES-2014-068602) from the former Ministerio de Economía y Competitividad (MINECO).

CO N FLI C T O F I NTE R E S T
Authors declare no competing financial interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The sequence of PxLPMO1 can be found in the GenBank database at https://www.ncbi.nlm.nih.gov/genba nk/ with the accession no.