Nep1-like Proteins from Valsa mali Differentially Regulate Pathogen Virulence and Response to Abiotic Stresses

Necrosis and ethylene-inducing peptide 1(Nep1)-like protein (NLP) is well known for its cytotoxicity and immunogenicity on dicotyledonous, and it has attracted large attention due to its gene expansion and functional diversification in numerous phytopathogens. Here, two NLP family proteins, VmNLP1 and VmNLP2, were identified in the pathogenic fungus Valsa mali. We showed that VmNLP2 but not VmNLP1 induced cell death when transiently expressed in Nicotiana benthamiana. VmNLP2 was also shown to induce cell death in apple leaves via the treatment of the Escherichia coli-produced recombinant protein. VmNLP1 and VmNLP2 transcripts were drastically induced at the early stage of V. mali infection, whereas only VmNLP2 was shown to be essential for pathogen virulence. We also found that VmNLP1 and VmNLP2 are required for maintaining the integrity of cell membranes, and they differentially contribute to V. mali tolerance to salt- and osmo-stresses. Notably, multiple sequence alignment revealed that the second histidine (H) among the conserved heptapeptide (GHRHDWE) of VmNLP2 is mutated to tyrosine (Y). When this tyrosine (Y) was substituted by histidine (H), the variant displayed enhanced cytotoxicity in N. benthamiana, as well as enhanced virulence on apple leaves, suggesting that the virulence role of VmNLP2 probably correlates to its cytotoxicity activity. We further showed that the peptide among VmNLP2, called nlp25 (VmNLP2), triggered strong immune response in Arabidopsis thaliana. This work demonstrates that NLPs from V. mali involve multiple biological roles, and shed new light on how intricately complex the functions of NLP might be.


Introduction
In plant-pathogen interactions, phytopathogens produce a variety of molecules, such as cell wall-degrading enzymes, effectors, and toxins, in order to establish effective infection on their host plants. Among them, effectors are usually produced to interfere with plant immunity by manipulating plant cellular processes [1,2]. On the contrary, plants can detect effectors, via either cell surface-localized or intracellular immune receptors, to activate immune responses and counteract pathogen infection [3,4]. Remarkably, there are quite a few molecules that can not only promote virulence, but also stimulate immune response on plants, of which, one of the most known examples represents the necrosis-and ethyleneinducing-like proteins (NLPs) [5,6].
Originally identified from culture filtrates of Fusarium oxysporum [7][8][9], NLPs are a superfamily of effector proteins widely distributed across bacteria, fungi and oomycetes. NLPs can be divided into three types, i.e., type 1, type 2 and type 3, which respectively contain two, four, and six conserved cysteine residues [8,9]. The typical bioactivity of NLPs is to cause cell death in dicotyledons [8,10]. Thus far, NLPs with cell death-inducing activity have been identified in various pathogens. Ottmann and colleagues have proven that NLPs can function as cytolytic toxins that promote the virulence of Pythium aphanidermatum [11].
Recently, the receptor for NLPs acting as toxins has been found to be plant sphingolipid glycosylinositol phosphorylceramides (GIPC) [12]. Notably, not all NLPs exhibit cell death-inducing activity in plants, and many characterized NLPs are noncytotoxic. For example, 5 out of 7 NLPs from Verticillium dahliae, 11 out of 19 NLPs from Phytophthora sojae, and all 10 tested NLPs from Hyaloperonospora arabidopsidis are not capable of inducing plant cell death [13][14][15]. Intriguingly, both cytotoxic and noncytotoxic NLPs can serve as immune elicitors. The cytotoxic NLP Pcc from the phytopathogenic bacterium Pectobacterium carotovorum can activate the expression of immune marker genes HIN1 and PAD3 in tobacco [11]. The 10 noncytotoxic NLPs from H. arabidopsidis unexceptionally trigger immune responses when ectopically expressed in Arabidopsis [16]. Moreover, a conserved immunogenic peptide among NLPs has been identified, which can be recognized by the receptor-like protein RLP23 in Arabidopsis [17,18].
During the past two decades, increasing evidence shows that NLPs play essential roles in diverse plant-pathogen interactions. For instance, VdNLP1 and VdNLP2 from V. dahliae, both of which display cytotoxic activity, are induced during pathogen infection and are required for pathogen virulence [15]. VdNLP1 also affects pathogen vegetative growth and conidiospore production [15]. Besides, most HaNLP genes from H. arabidopsidis are upregulated at the early infection stages, and the expression is accompanied by appressorium formation, suggesting that these HaNLP proteins may still be functioning in pathogen-plant attachment [13]. In addition, the NLPs from Phytophthora sojae are under strong positive selection, implying possible alternative roles in these proteins [14]. In spite of these, the function of noncytotoxic NLPs and cytotoxic NLPs outside pathogen-plant interactions remains largely undiscovered.
Apple Valsa canker caused by the ascomycete fungus Valsa mali, a weak parasitic pathogenic fungus, annually results in huge economic losses for the apple industry [19,20]. Though much efforts have been made to elucidate the infection mechanism of V. mali, the major virulence determinants of this fungus still remain largely puzzled. Much evidence has demonstrated that toxins play important roles during the V. mali infection of its apple host [21][22][23][24]. Considering that NLPs frequently function as cytolytic phytotoxins on dicotyledonous plants, we suspect that this member of proteins may play important roles in apple-V. mali pathosystem.
In this study, we successfully identified two NLP family proteins in V. mali, respectively named VmNLP1 and VmNLP2. We found that VmNLP2 but not VmNLP1 could induce cell death in the model plant Nicotiana benthamiana, and VmNLP2 can also cause cell death in the apple host. Both of these two genes were highly induced at early stage of V. mali infection. Target gene deletion revealed that only VmNLP2-deficient strain exhibited reduced virulence on apple host. Further analysis showed that VmNLP1 and VmNLP2 are differently involved in the maintenance of cell membrane integrity and pathogen tolerance to salt and osmotic stresses. Moreover, the immunogenic peptide among VmNLP2 displayed a potent activity to inducing immune response in A. thaliana, whereas the peptide among VmNLP1 showed a much weaker activity. Collectively, our results revealed that VmNLP2 is an essential virulence factor of V. mali, and we emphasized that the two NLP proteins from V. mali are functionally diversified in multifaceted biological roles, including their cytotoxic activity, contribution to pathogen virulence, and response to abiotic stresses.

Strains and Plant Growth Conditions
The Valsa mali wild-type strain 03-8 and Sclerotinia sclerotiorum strain 1980 were stored in 20% glycerol at 4 • C. Escherichia coli strain DH5α used for plasmid construction was cultured on lysogeny broth (LB) medium at 37 • C, and Agrobacterium tumefaciens strain GV3101 used for agroinfiltration of plants was cultured on LB medium at 28 • C. Micropropagated apple (Malus domestica) rootstock plants GL-3 and Arabidopsis thaliana were cultured in an artificial climate chamber, 12 h/12 h light/dark at 22 • C. Nicotiana benthamiana was grown in an artificial greenhouse, 16 h/8 h light/dark at 22 • C.

Plasmid Construction
The open reading frames of VmNLP1, VmNLP2 or VmNLP2 without signal peptide were amplified from complementary DNA (cDNA) library of V. mali with gene-specific primers using Phanta Max Super-Fidelity DNA Polymerase (Vazyme, Nanjing, China). The hemagglutinin (HA)-tagged pCAMBIA1300 vector carrying kanamycin resistance gene was digested with KpnI and was subsequently ligated with the amplicons using ClonExpress II One-Step Cloning Kit (Vazyme, Nanjing, China). For generation of sitemutated proteins, the site-mutated fragments were amplified using mismatched base primers, which were subsequently cloned into pCAMBIA1300 vector using ClonExpress MultiS One-Step Cloning Kit (Vazyme, Nanjing, China). To generate gene complemented constructs, the PDL2 vector was linearized by XhoI digestion and further ligated with VmNLPs gene fragments that were amplified from V. mali genomic DNA containing their native promoters. The primers used for plasmid construction were listed in Supplementary Table S1. All PCR products were excised and purified from gel using Gel extraction Kit (OMEGA Biotech, Guangzhou, China). All constructs were sequence-verified by standard vector primers (Sangon Biotech, Shanghai, China).

A. tumefaciens-Mediated Transient Expression and Trypan Blue Staining
Four-to five-week-old N. benthamiana plants were used for agroinfiltration assays. The constructs were transformed into Agrobacterium tumefaciens GV3101 via electroporation. The colonies were selected by kanamycin (100 µg/mL), rifampicin (100 µg/mL), and confirmed by PCR. The individual correct transformant was cultured in LB medium at 28 • C with 220 rpm for 48 h. Cells were collected by centrifugation and were suspended in MES buffer (10 mM Magnesium chloride, 10 mM 2-(N-morpholino) ethane sulfonic acid (MES), 200 µM acetosyringone, pH 5.7). The suspended A. tumefaciens cells were incubated at room temperature for 3 h and were then mixed with P19 silencing suppressor. The final OD 600 of suspensions were adjusted with MES buffer to 0.6. The suspensions were infiltrated into tobacco leaves using a needleless 1-mL syringe. Symptoms were observed 4-6 d post agroinfiltration. A. tumefaciens carrying INF1 and green fluorescent protein (GFP) were used as positive and negative control. Trypan blue staining was performed as described [29]. The experiment was repeated at least three times.

Western Blotting
N. benthamiana leaves transiently expressing indicated proteins were collected 36 h after infiltration. Lysing buffer (50 mM Tris, 150 mM NaCl, pH 7.5, 5 mM dithiothreitol (DTT), 0.5% TritonX-100, 1 mM phenylmethanesulfonyl fluoride (PMSF), 1% proteinase inhibitor cocktail (Sigma-Aldrich, St Louis, MO, USA)) was used to extracted total proteins. Ten minutes after centrifuge at 15,000× g (4 • C), the supernatant was collected to a new tube. The protein samples were boiled for 6 min in 2× sodium dodecyl sulfate (SDS) loading buffer and were then subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Subsequently, the protein was transferred from gel to a polyvinylidene difluoride (PVDF) membrane under the transfer buffer (20 mM Tris, 150 mM glycine). The membrane was washed by Tris buffered saline (TBS), and TBST (TBS with 0.1% Tween 20) containing 5% nonfat dry milk was used to block the membrane at 50 rpm shaking for 2 h at room temperature. The membrane was then incubated overnight with mouse anti-HA (Abcam, Cambridge, UK, ab18181) monoclonal antibody and rinsed by TBST for three times, followed by staining with goat-anti mouse IgG (Abways, Shanghai, China) secondary antibody. Proteins were detected using ECL subtracted kit (GE Healthcare, RPN2235), as recommended by the manufacturer.

RNA Extraction and qRT-PCR Analysis
Total RNA from collected samples, including leaves of N. benthamiana and A. thaliana, apple twigs infected with V. mali, as well as mycelia from V. mali, was extracted using Quick RNA isolation Kit (Huayueyang, Beijing, China) according to the manufacturer's instruction. RNA concentration and purity were quantified by NanoDrop Micro Photopeter (NanoDrop, Wilmington, DE, USA). The RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific, Waltham, MA, USA) was used for first-strand cDNA synthesis from 1 µg of total RNA, followed by quantitative reverse transcription-PCR (qRT-PCR) using Real Star Green Mixture (GenStar, Beijing, China) with specific primers. G6PDH in V. mali, Actin2 in A. thaliana or MdH in M. domestica were used as internal controls. Relative fold changes were calculated using the 2 −∆∆CT method [30].

Generation of Deletion Mutants, Complementation and Overexpression Transformants
To create deletion cassettes, the upstream and downstream flanking sequences amplified from V. mali genetic DNA were fused with NEO fragments amplified from vector pFL2, using double-joining PCR method [31]. The replacement fragment was transformed into protoplasts of V. mali via the polyethylene glycol (PEG)-mediated method [32]. The transformants were selected by G418-resistance followed by four types of PCR verifications to confirm the gene was knocked out. To obtain complementation transformants, pDL2 vector carrying VmNLPs or site-directed mutation variants (VmNLP1 H127Y or VmNLP2 Y137H ) were transformed into protoplasts of VmNLPs deletion mutants. The hygromycin-resistance transformants were ascertained by PCR with primer pair VmNLP-5F/6R and sequencing. The overexpression transformants were achieved through fusing the fragments with the pDL2 vector containing the Magnaporthe grisea ribosomal protein 27 promoter, followed by transformation into the V. mali protoplasts. The transformants were filtered by hygromycin and total RNA was isolated to assess transcriptional levels of VmNLPs using qRT-PCR analysis. All primers used for gene knockout and overexpression are listed in Table S1.

Virulence and Phenotypic Analysis of Transformants
For virulence tests, fresh mycelial plugs collected from V. mali wild strain (WT), deletion mutants, complementation transformants and overexpression transformants were inoculated on leaves and twigs detached from Malus domestica Borkh. cv. Fuji [33]. Five millimeters mycelium plugs used in this assay were taken from the edge of growing colonies cultured on potato dextrose agar (PDA, 20% potato extract (w/v), 2% dextrose (w/v), 1.5% agar (w/v)) medium. Disease lesions were measured 2-3 days post-inoculation (dpi) for apple leaves and 4-5 dpi for apple twigs, followed by a data analysis. For vegetative growth detection, colony diameters of all strains cultured on PDA plates were calculated 2 days post-cultivation at 25 • C. For abiotic stress assays, PDA media were supplemented with congo red (CR, 400 µg/mL), sorbitol (1 M), sodium dodecyl sulfate (SDS, 0.01%) or KCl (0.3 M). Each experiment was conducted with three biological replicates, and all assays were repeated at least three times.

Expression and Purification of Recombinant VmNLP2 Protein
The coding sequence of VmNLP2 without signal peptide was amplified and cloned into pET28a vector (Novagen Inc., Madison, WI, USA) with NdeI and BamHI restriction sites, followed by transformation into E. coli BL21 (DE3). Cells were cultivated in LB medium at 37 • C until OD 600 = 0.6~0.8. Subsequently, cells were introduced with 0.3 mM isopropyl-β-D-thiogalactopyrandoside (IPTG) for 24 h at 16 • C, and were then harvested by centrifugation at 5000× g. The collected bacteria cells were suspended in lysis buffer (20 mM sodium hydrogen phosphate, 300 mM NaCl, pH 7.4) containing 1 mM phenylmethanesulfonyl fluoride (PMSF), 1.98 mM β-mercaptoethanol and 1 mg ml −1 lysozyme followed by sonication. The supernatant of cells was collected via centrifugation at 4 • C and 10,000× g for 10 min and were subsequently loaded to Ni-NTA resin (Thermo Scientific, Waltham, MA, USA). VmNLP2 recombinant protein purification was finished by affinity chromatography according to the manufacturer's instructions. The samples were subjected to silver staining using the Pierce™ Silver Stain for Mass Spectrometry Kit (Thermo Scientific, Waltham, MA, USA) following the manufacturer's instructions, and were further analyzed by western blotting detection using anti-His monoclonal antibody (Abways, Shanghai, China).

Measurement of Reactive Oxygen Burst
Reactive oxygen species (ROS) generation was detected as described [34]. Briefly, leaf discs were collected from A. thaliana seedlings and floated on 0.1 mL ddH 2 O overnight in a 96-well plate. Right before luminescence detection on a Varioskan LUX multimode microplate reader (Thermo Scientific), water was replaced with a 0.1 mL reaction solution (100 µM L-012 (Wako Chemical, Osaka, Japan), 20 µg mL −1 peroxidase (Solarbio, Beijing, China)), and 1 µM peptides. Eight biological replicates were used for each sample, and the experiments were repeated at least three times.

VmNLP2 but Not VmNLP1 Displays Cytotoxic Activity on N. benthamiana
To identify the NLP gene family in V. mali, a blast search of the V. mali genome [24] was performed with the FoNep1 from Fusarium oxysporum [25,26] in NCBI database. As a result, two different NLP homologues were identified and named VmNLP1 and VmNLP2, respectively (Table S2). Sequence analysis indicated that both VmNLP1 and VmNLP2 contain a predicted N-terminal signal peptide and two conserved cystine residues ( Figure S1), for which they are classified as type 1 NLPs.
To test whether these two proteins exhibit cell death-inducing activity, hemagglutinin (HA)-tagged VmNLP1 and VmNLP2 were transiently expressed in N. benthamiana through agroinfiltration, with mCherry-tagged INF1 [35] and HA-tagged GFP used as positive and negative controls, respectively. As shown in Figure 1A, both VmNLP2 and INF1 induced visible cell death 5 days post agroinfiltration (dpa), whereas VmNLP1 and GFP control did not cause any necrosis symptoms. This indicates that VmNLP2 but not VmNLP1 displays cytotoxic activity in N. benthamiana. To test whether the signal peptide (SP) is needed for the cytotoxic activity of VmNLP2, VmNLP2 without SP (VmNLP2 ∆SP ) was transiently expressed in N. benthamiana. It showed that VmNLP2 ∆SP cannot induce necrosis as full length VmNLP2 did ( Figure 1A), suggesting that VmNLP2 requires its SP for cell death activation. Western blotting analysis showed that all proteins tested were successfully expressed in N. benthamiana ( Figure 1B).

VmNLP2 Shows Cytotoxicity on Malus Domestica
To examine whether VmNLP2 displays cytotoxic activity on its natural apple host, we firstly produced VmNLP2 recombinant protein in E. coli. N-terminal His-tagged VmNLP2 was verified through silver staining and western blotting detection ( Figure 2A). Next, M. domestica leaves punctured by sterile syringes at the treatment site were dropped with 5 µM VmNLP2 purified protein, with buffer and 10% methanol used as negative controls, and 40 mM protocatechuic acid [22] used as a positive control. It illustrated that visible necrosis could be observed 24 h after treatment by VmNLP2 purified protein or protocatechuic acid, whereas neither buffer nor 10% methanol control caused any apparent symptoms ( Figure 2B). Therefore, VmNLP2 exhibits cytotoxic activity on its apple host as well.

VmNLP2 Is Required for V. mali Virulence
To determine the potential virulence roles of VmNLP1 and VmNLP2, we first analyzed their expression profiles during V. mali infection by qRT-PCR. As was shown, both VmNLP1 and VmNLP2 were markedly induced between 6 to 36 h post inoculation (hpi), and both of their transcripts peaked at 6 hpi ( Figure 3A). This indicates that VmNLP1 and VmNLP2 are probably involved in the pathogenic process of V. mali, particularly during the early stage of pathogen infection.
We next tested whether VmNLP1 and VmNLP2 are required for pathogenicity of V. mali, for which, gene knockout mutants ( Figure S2A) and corresponding complementation transformants ( Figure S2B) of these two genes were generated via PEG-mediated protoplast transformation. Phenotypic analysis manifested that there was no obvious difference on filamentous growth in all deletion mutants ( Figure S3A,B). Inoculation assays on apple host showed that ∆VmNLP1-10 and ∆VmNLP1-66 were as virulent as the WT strain ( Figure S4A). Nevertheless, compared to WT strain, the virulence of ∆VmNLP2-36 was significantly impaired ( Figure 3B). Moreover, by introducing VmNLP2 into ∆VmNLP2-36, the complementation transformant ∆VmNLP2-C1 caused comparable disease lesions to the WT strain ( Figure 3B). In order to further verify the above results, overexpression transformants were generated and their virulence were tested on detached apple leaves. The overexpression transformants of VmNLP1 (VmNLP1-OE6) and VmNLP2 (VmNLP2-OE17) were verified to show six-fold and eight-fold enhanced transcript levels, respectively (Supplementary Figure S5). Inoculation assays showed that lesion diameter caused by VmNLP1-OE6 was not conspicuously different from that caused by WT strain ( Figure S4B). However, VmNLP2-OE17 caused larger disease lesion on apple leaves than that of the WT strain ( Figure 3C). We also tested the virulence of these transformants on the detached twigs of apple. As expected, VmNLP1-deletion mutants did not result in apparent alterations in virulence compared to the WT strain ( Figure S4C). Whereas, ∆VmNLP2-36 displayed markedly reduced lesion length compared with the WT strain, and the VmNLP2 complementation transformant was as virulent as the WT strain ( Figure 3D). Consistently, the lesion length developed by VmNLP2-OE17 was greater than those that of the WT strain ( Figure 3D). Taken together, these results demonstrated that VmNLP2 but not VmNLP1 is required for V. mali full virulence on the apple host.

VmNLP1 and VmNLP2 Differentially Contribute to Salt Tolerance, Osmotic and Membrane Stresses
NLPs have also been demonstrated to occur in both pathogenic and non-pathogenic microorganisms [6,9], and it was assumed that potential function outside plant-pathogen interactions may exist for these proteins [10,15]. Therefore, apart from virulence, whether VmNLP1 and VmNLP2 are involved in V. mali responsiveness to other environmental factors such as abiotic stresses were subsequently tested. For this, V. mali WT strain, gene deletion mutants, and corresponding complementation transformants were cultured in PDA plants supplemented with 0.3 M KCl (salt stress), 1 M Sorbitol (osmotic stress), 0.01% SDS (cell membrane damaging agent), or 400 µg ml −1 CR (Congo red, cell wall inhibitor). We found that the vegetative growths of ∆VmNLP1-66 and ∆VmNLP2-36 were similar to that of the WT under the resistance to cell wall stress caused by CR. Interestingly, ∆VmNLP1-66 and ∆VmNLP2-36 displayed smaller vegetative growth under the treatment of SDS compared to the WT ( Figure 4A), suggesting that both VmNLP1 and VmNLP2 participate in maintaining pathogen cell membrane integrity. In addition, under 1 M Sorbitol, the vegetative growth of ∆VmNLP1-66 but not ∆VmNLP2-36, was smaller than that of WT ( Figure 4A), indicating that VmNLP1 also plays a protective role against osmo-tolerance. Unexpectedly, under 0.3 M KCl, the colony of ∆VmNLP1-66 was smaller, whereas ∆VmNLP2-36 colony was larger than that of the WT (Figure 4A), indicating that these two NLPs contrastingly regulate V. mali tolerance to salt stress. All the complement transformants showed similar tolerance levels to the WT under all tested abiotic stresses ( Figure 4A,B). Collectively, VmNLP1 and VmNLP2 differentially participate in V. mali response to salt, osmotic, and membrane stresses. and 400 µg/mL Congo red (CR) at 25 • C for 3 days, respectively. The experiment was performed in triplicate, with five petri dishes in each repetition. (B) Colony diameter calculation for growth of the stains 3 days after inoculation on tested medium. Statistical analyses were performed by Student's t-test (**, p ≤ 0.01; ***, p ≤ 0.001). Bars indicate ± SD of the mean of three replicates.

The Second Histidine Residue among the Conserved Heptapeptide Is Mutated to Tyrosine in VmNLP2
The difference between VmNLP1 and VmNLP2 in multiple biological roles presented above prompted us to analyze their amino acid sequence. Fifteen fully studied NLPs derived from bacterial, fungal and oomycete species were selected for multiple sequence alignment. It revealed that several residues known to be essential for phytotoxic activity of NLPs, including K116, D117, as well as two cystine residues, are conserved in VmNLP1 and VmNLP2 ( Figure S1). The characteristic heptapeptide motif "GHRHDWE" that are considered to be essential for necrosis-inducing activity of NLPs [9,11], are also conversed in almost all selected NLPs ( Figure 5). Notably, however, for VmNLP2, the motif is "GHRYDWE", in which the second histidine (H) among the heptapeptide is mutated to tyrosine (Y) (Figure 5), which may account for the distinct functions between VmNLP1 and VmNLP2. Figure 5. The second histidine residue among the conserved heptapeptide of VmNLP2 is mutated to tyrosine. A total of 17 NLPs, including VmNLP1 and VmNLP2 from Valsa mali, NLP pp from Parasitic phytophthora [18], FoNep1 from Fusarium oxysporum [25], SsNLP1 and SsNLP2 from Sclerotinia sclerotiorum [36], VdNLP1 and VdNLP2 from Verticillium dahlia [15], Bcnep1 and Bcnep2 from Botryis cinerea [37], MpNLP1 and MpNLP2 from Moniliophthora perniciosa [38], HaNLP3 from Hyaloperonospora arabidopsidis [13], PiNPP1.1 from Phytophthora infestans [39], PsojNIP from Phytophthora sojae [40], NLP pya from Pythium aphanidermatum [41], and NLP pcc from Pectobacterium carotovorum [42] were used for multiple sequence alignment by Clustal W. Part of the alignment results are shown. Conserved residues are shaded on a red background. Amino acid mutation among conserved heptapeptide is indicated by a black arrow.

Site-Directed Mutagenesis of Tyrosine to Histidine Promoted VmNLP2 Cytotoxicity and Virulence
To assay whether the natural residue mutation among the heptapeptide of VmNLP2 affect its the cytotoxicity, constructs with this tyrosine substituted by histidine (VmNLP2 Y137H ) were generated. For comparison, we also replaced the Histidine among VmNLP1 with tyrosine (VmNLP1 H127Y ). When transiently expressed in N. benthamiana, neither VmNLP1 nor VmNLP1 H127Y was able to induce plant cell death ( Figure 6A). Interestingly, however, VmNLP2 Y137H apparently exhibited enhanced cytotoxic activity than that of VmNLP2 ( Figure 6A). All proteins expressed in N. benthamiana were verified by Western blotting ( Figure 6B). Thus, this result suggests that the natural residue mutation among VmNLP2 can attenuate its cytotoxicity.
We next tested whether VmNLP1 H127Y and VmNLP2 Y137H can influence the virulence of V. mali. Site-directed mutation transformants ∆VmNLP1-C H127Y and ∆VmNLP2-C Y137H were generated by introducing VmNLP1 H127Y and VmNLP2 Y137H fragments into the corresponding deletion mutants. All mutation transformants were ascertained by PCR analysis (Supplementary Figure S2B). Two of them were randomly selected for a virulence test. As was shown, ∆VmNLP1-C H127Y caused similar disease lesion as the complementation transformant ∆VmNLP1-C1 did, either on the detached apple twigs or leaves (Supplementary Figure S6A,B). However, compared with ∆VmNLP2-C1, ∆VmNLP2-C Y137H showed an obvious increment in virulence when tested in detached apple leaves ( Figure 6C). To be noted, ∆VmNLP2-C Y137H showed no apparent change in virulence on detached twigs ( Figure S7), which may be attributed to apple twigs being less sensitive. Altogether, the natural residue mutation among VmNLP2 can also attenuate its virulence, and VmNLP2 contribution to virulence is probably correlated to its cytotoxicity.

nlp25 among VmNLP2 Triggers Strong Immunity Response in A. thaliana
Since both cytotoxic and noncytotoxic NLPs possess a conserved immunogenic peptide [16,18], we next explored whether VmNLP1 and VmNLP2 contain such immunogenic peptides. Therefore, the peptides nlp20 derived from VmNLP1 and nlp25 derived from VmNLP2 ( Figure S1) corresponding to nlp20 (NLP Pp ) [18] were synthesized. The peptides were subsequently used for measurement of immune responses, including ROS burst, expression of immune marker genes, and activation of disease resistance in Arabidopsis thaliana, a model plant that recognizes the nlp20 (NLP Pp ) peptide [17]. The well characterized flg22 peptide [43] was used as a positive control. The luminol-based chemiluminescence assay revealed that both nlp25 (VmNLP2) and nlp20 (VmNLP1) can elicit ROS burst in A. thaliana ( Figure 7A). Notably, nlp20 (VmNLP1)-induced ROS level was not so strong as that of nlp25 (VmNLP2). We further treated with A. thaliana with 1 µM peptides and collected the samples 24 h later to analyze the expression of defense genes by qRT-PCR. As was shown in Figure 7B, the transcript levels of PR1, WRKY33 and FRK1 were markedly induced by treatment of nlp25 (VmNLP2), but not nlp20 (VmNLP1). In accordance, the disease lesions caused by the pathogenic fungus Sclerotinia sclerotiorum on Arabidopsis pre-treated with nlp25 (VmNLP2) but not nlp20 (VmNLP1) peptide were smaller than that of water control ( Figure 7C). These results collectively indicate that nlp25 (VmNLP2) possesses a strong immunogenic activity, whereas nlp20 (VmNLP1) shows a much weaker and even no apparent immunogenic activity in A. thaliana. We also tested whether the synthetic peptides exhibited immunogenic activities in N. benthamiana. As a result, none of them triggered apparent ROS bust (Supplementary Figure S8A), which is consistent with the fact that N. benthamiana lacks the RLP23 receptor for detection of nlp20 [17,18]. Most recently, two NLPs from Pythium oligandrum have been demonstrated to activate immune responses in N. benthamiana [44]. We next tested whether VmNLP recombinant proteins function similarly. As was shown, neither of them obviously induced ROS burst (Supplementary Figure S8A), nor did they activate the expression of immune marker genes like NbPR1, NbPR2 and NbPR4 in N. benthamiana (Supplementary Figure S8B). Thus, VmNLPs and the synthetic peptides therein display no immunogenic activity in the model plant N. benthamiana.

Discussion
The function of NLPs is varied due to the broad taxonomic distribution and rapid evolution [8,14]. In this study, we found that two NLPs from Valsa mali participate in a variety of biological roles and exhibit functional diversification. We have shown that only one of the two NLP proteins in V. mali, VmNLP2, possesses cytotoxic activity and contributes to virulence of V. mali. The non-cytotoxic VmNLP1 does not contribute to pathogenicity. However, we found that VmNLP1 positively regulates V. mali tolerance to salt, osmotic stress and cell membrane damaging agent. VmNLP2 was also shown to be involved in maintaining the integrity of cell membrane, but VmNLP1 and VmNLP2 played a distinct role in response to salt stress. Moreover, we showed that the immunogenic peptide nlp25 (VmNLP2) triggered strong immune responses in Arabidopsis thaliana, whereas nlp20 (VmNLP1) only elicited much weaker immune responses in A. thaliana.
Based on the ability to induce necrosis, NLPs can be divided into cytotoxic and non-cytotoxic NLPs. The cytotoxic NLPs can permeabilize plant plasma membranes of dicotyledons and lead to necrosis [7,11]. The non-cytotoxic NLPs are not capable of inducing cell necrosis, and their biological functions are largely mysterious [13,14]. Most necrotrophic phytopathogens have only one or two NLPs, and a majority of them, if not all, have been demonstrated to be cytotoxic. For example, Botrytis elliptica contains two NLPs which exhibit cytotoxic activity in many tested dicots [45], and two NLPs from Sclerotinia sclerotiorum can also induce necrosis when transiently expressed in tobacco leaves [36]. Here, we found that VmNLP2 but not VmNLP1 from the weakly parasitic V. mali exhibited cytotoxic activity in plants (Figures 1 and 2). Actually, there are also some fungal species that contain necrotrophic phase encoding noncytotoxic NLPs on account of gene expansion, such as Verticillium dahlia [15] and Colletotrichum higginsianum [46]. It is worth noting that the mature protein of Botryis cinerea NLPs fused with the signal peptide of tobacco PR1a can trigger necrotic responses in N. benthamiana, whereas the BcNEP1 and BcNEP2 carrying their endogenous SP did not cause necrosis [37]. Likewise, it is possible that the inability of VmNLP1 to induce plant cell death may attribute to the potential inefficiency of its endogenous SP. Alternatively, the tertiary structure of VmNLP1 may not be conductive to its toxic function, similar to the case of HaNLP3, a typical noncytotoxic NLP from Hyaloperonospora arabidopsidis [13].
The transcripts of VmNLP1 and VmNLP2 were found to be highly induced at early stage of infection ( Figure 3A), and we therefore hypothesized that these two genes may be involved in pathogenicity. Deletion of VmNLP2 reduced virulence on apple leaves and twigs ( Figure 3B,D), however, the knocking out of VmNLP1 posed no apparent effect on pathogen virulence (Supplementary Figure S4A,C). In accordance, the overexpression of VmNLP2 in V. mali promoted pathogen infection in apple leaves ( Figure 3C). This evidence together demonstrated that VmNLP2 is an important virulence factor of V. mali. This is similar to several characterized NLPs from other phytopathogens, which serve as virulence factors as well. For example, targeted deletion nip gene in E. carotovora subsp. carotovora compromised virulence in potato [42]; V. dahliae lacking either of the two cytotoxic NLPs showed significantly reduced virulence on tomato as well as on Arabidopsis plants [15]; overexpression of FoNep1 in the fungus Colletotrichum coccodes improved fungal virulence on Abutilon theophrasti [47]. Importantly, by multiple sequence alignment, we observed a residue mutation among the conserved heptapeptide motif of VmNLP2, in which the second histidine was mutated to tyrosine ( Figure 5). We further found that the cytotoxic activity and virulence of VmNLP2 were simultaneously enhanced when the tyrosine residue was replaced with histidine residue (Figure 6), indicating a potential positive correlation between VmNLP2 cytotoxicity and its contribution to virulence. This result is reminiscent of NLP Pcc from Pectobacterium carotovorum. It has been reported that the complementation of less cytotoxic NLP Pcc variants in NLP Pcc -deficient P. carotovorum strain resulted in correspondingly declined virulence [11]. These findings reinforce that NLPs can act as toxin-like virulence factors during the interaction of plant and pathogens.
Previous studies reveled that the large expansion of Nep1-like proteins family not only in oomycetes but also in fungi [6]. Thus, we were also interested in exploring the biological roles of VmNLPs apart from virulence and cytotoxic activity. In this study, we found that both VmNLP1 and VmNLP2 are involved in the salt tolerance and maintenance of cell membrane integrity of V. mali (Figure 4). In addition, VmNLP1 also participated in response to osmotic stress ( Figure 4). Meanwhile, considering that NLPs are secreted proteins, we cannot exclude that VmNLP1 and VmNLP2 serve as 'guards' to differentially protect V. mali against abiotic stresses. A role of NLP in growth and conidiospore production has also been confirmed in the fungus V. dahlia [15]. Furthermore, the high expression of HaNLPs at an early infection stage led researchers to hypothesize it may play a role in zoospore attachment or primary contact between pathogen and host [13]. Our results here indicate that the two NLPs from V. mali are functionally diversified, not only in cytotoxicity and virulence, but also in regulation of pathogen responses to abiotic stresses.
In this study, we found that both the peptide nlp25 (VmNLP2) and nlp20 (VmNLP1) can elicit ROS burst in A. thaliana, however, ROS-triggered by nlp20 (VmNLP1) was much lower than that of nlp25 (VmNLP2) ( Figure 7A). Additionally, nlp25 (VmNLP2) but not nlp20 (VmNLP1) activated the expression of defense-related genes and induced Arabidopsis resistance against Sclerotinia sclerotiorum ( Figure 7B,C), though nlp20 (VmNLP1) is more similar to nlp20 derived from NLP Pp [18]. This result further highlights the functional diversification of VmNLPs. We have also tested the response of apple leaves to these two peptides. Nevertheless, both nlp20 (VmNLP1) and nlp25 (VmNLP2) failed to trigger the expression of defense-related genes on apple, nor could they promote apple resistance to the V. mali infection (Supplementary Figure S9). It is possible that apple lacks a receptor like RLP23 from Arabidopsis [17] for detection of the immunogenic peptide among NLPs.

Conclusions
In summary, our study revealed that VmNLP2 is a cytotoxic NLP and it serves as an essential virulence factor of V. mali. Surprisingly, we showed that VmNLP1 and VmNLP2 play contrasting roles in V. mali response to cell membrane damaging agent, salt-and osmic-tolerance. The two NLPs in V. mali play functionally diversified roles in cytotoxicity, virulence, and regulation of pathogen tolerance to abiotic stresses, which offer a new understanding of the biological roles of the Nep1-like proteins.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/jof7100830/s1, Figure S1: Multiple sequence alignment of VmNLPs with its homologous sequences from other pathogen species using Clustal W. Figure S2: Validation of gene deletion and complementation mutants. Figure S3: Deletion of VmNLP1 or VmNLP2 posed no obvious influence on V. mali filamentous growth. Figure S4: Deletion of VmNLP1 showed no apparent influence on V. mali virulence. Figure S5: Relative transcript levels of VmNLP1 and VmNLP2 overexpression transformants. Figure S6: Substitution of the second histidine with tyrosine in the conserved heptapeptide among VmNLP1 did not affect virulence. Figure S7: Substitution of the tyrosine with histidine showed no apparent contribution to VmNLP2 virulence on apple twigs. Figure S8: VmNLP recombinant proteins and synthetic peptides are unable to trigger immune responses in N. benthamiana leaves. Figure S9: The peptide nlp20 (VmNLP1) and nlp25 (VmNLP2) are unable to trigger immune responses in apple leaves. Table S1: Primers used in this study.