RLP23 is required for Arabidopsis immunity against the grey mould pathogen Botrytis cinerea

Necrosis- and ethylene-inducing-like proteins (NLPs) are secreted by fungi, oomycetes and bacteria. Conserved nlp peptides derived from NLPs are recognized as pathogen-associated molecular patterns (PAMPs), leading to PAMP-triggered immune responses. RLP23 is the receptor of the nlp peptides in Arabidopsis thaliana; however, its actual contribution to plant immunity is unclear. Here, we report that RLP23 is required for Arabidopsis immunity against the necrotrophic fungal pathogen Botrytis cinerea. Arabidopsis rlp23 mutants exhibited enhanced susceptibility to B. cinerea compared with the wild-type plants. Notably, microscopic observation of the B. cinerea infection behaviour indicated the involvement of RLP23 in pre-invasive resistance to the pathogen. B. cinerea carried two NLP genes, BcNEP1 and BcNEP2; BcNEP1 was expressed preferentially before/during invasion into Arabidopsis, whereas BcNEP2 was expressed at the late phase of infection. Importantly, the nlp peptides derived from both BcNEP1 and BcNEP2 induced the production of reactive oxygen species in an RLP23-dependent manner. In contrast, another necrotrophic fungus Alternaria brassicicola did not express the NLP gene in the early infection phase and exhibited no enhanced virulence in the rlp23 mutants. Collectively, these results strongly suggest that RLP23 contributes to Arabidopsis pre-invasive resistance to B. cinerea via NLP recognition at the early infection phase.

www.nature.com/scientificreports/ in oomycetes. Type III NLPs are mainly observed only in a limited number of ascomycete species 19 . Subsequent structural analysis of an NLP from the phytopathogenic oomycete Phytium aphanidermatum revealed that NLPs have two opposing antiparallel β-sheets, called β-sandwich, in the central part of the protein, similar to lectins in fungi and actinoporins in sea anemones 20 . Moreover, it was recently reported that NLPs specifically bind to glycosyl inositol phosphoryl ceramide (GIPC) 21 , as actinoporins bind to sphingomyelin. Although GIPC is the most abundant class of sphingolipids in plants, GIPCs also occur in fungi and protozoa 22 . GIPC is composed of inositol phosphoceramide and is anchored in the membrane 23 . It is speculated that the conformational changes of GIPC-NLP complexes induce pore formation and lead to cell death 21,24 . Importantly, two independent articles reported that the amino acids that are conserved in the central part of NLPs (a 20-amino-acid pattern termed nlp20 and a 24-amino-acid pattern termed nlp24) act as a PAMP. The nlp24 peptide contains two conserved regions; conserved region I starts with the AIMY sequence of amino acids, which are highly conserved in type I NLPs, whereas conserved region II starts with the heptapeptide motif GHRHDWE, which is highly conserved in all NLPs 6,18 . Treatment with the synthetic nlp20/nlp24 peptide induces defence responses, including ethylene production, in A. thaliana 2,6 . The recognition of nlp20 is also observed in other plants of the Brassicaceae family, i.e., Arabis alpina, Thlaspi arvense and Draba rigida, and the Asteraceae family, i.e., Lactuca sativa 2 . In contrast, it was reported that cucurbits recognize the C-terminal amino acids of NLPs 25 . Recently, an LRR-receptor protein (LRR-RP) termed RLP23 was identified as the receptor for nlp20 in A. thaliana 26 . It was also reported that RLP23 constitutively interacts with the LRR-RK called SUPPRESSOR of BIR1-1 (SOBIR1) and forms a complex with BAK1 after nlp20 perception. The complex triggers the subsequent activation of the nlp peptide-induced immunity [26][27][28] .
However, the actual contribution of RLP23 to plant immunity remains unclear. Here, we report that RLP23 was required for Arabidopsis immunity against the necrotrophic fungal pathogen Botrytis cinerea, which is the causal agent of grey mould. Microscopic observation revealed that the invasion ratio of B. cinerea was higher in Arabidopsis rlp23 mutants compared with wild-type (WT) plants, suggesting the involvement of RLP23 in pre-invasive immunity, which inhibits pathogen invasion. B. cinerea carries two NLP genes termed BcNEP1 and BcNEP2. In the interaction with Arabidopsis, BcNEP1 was preferentially expressed before/during pathogen invasion, whereas BcNEP2 was expressed at the late infection phase. We also discovered that the nlp peptides derived from BcNEP1 and BcNEP2 activated the PAMP-triggered immune response in an RLP23-dependent manner. Together with further studies on A. brassicicola, these results strongly suggest that NLP perception via RLP23 in the early infection phase contributes to Arabidopsis immunity against B. cinerea.

Results
Arabidopsis rlp23 mutants show enhanced susceptibility to the necrotrophic pathogen Botrytis cinerea. Arabidopsis rlp23 mutants were inoculated with the Botrytis cinerea IuRy-1 strain via dropping of a conidial suspension of the pathogen. We found that the Arabidopsis rlp23-1 plants showed enhanced susceptibility to B. cinerea at 7 days post-inoculation (7 dpi) (Fig. 1A); enhanced susceptibility was also observed in Arabidopsis rlp23-2 plants (Fig. 1A). Lesion size was significantly increased in the rlp23-1 and rlp23-2 mutants compared with the parental WT plant (Col-0) (Fig. 1B). A quantitative analysis of the growth of B. cinerea in planta also supported this finding. The amount of B. cinerea in inoculated Arabidopsis plants was quantified by quantitative PCR (qPCR) of the B. cinerea cutinase A gene (BcCutA) at 6, 12 and 16 h post-inoculation (hpi) 29 . The expression of BcCutA increased gradually in the inoculated Arabidopsis plants from 6 to 16 hpi (Fig. 1C). Importantly, the amount of the amplified BcCutA at 12 hpi in the rlp23-2 mutant was significantly higher than that detected in the WT plants (Fig. 1C). These results suggest that RLP23 contributes to Arabidopsis immunity against the necrotrophic fungal pathogen B. cinerea.
The Arabidopsis rlp23 mutant has a defect in pre-invasive resistance against B. cinerea. To identify the pathogen infection steps that are affected by the loss of RLP23 function, we performed microscopic observation of the B. cinerea infection behaviour in Arabidopsis plants. At 6 hpi, most conidia had germinated and started to elongate germ tubes, but did not develop invasive hyphae in WT Col-0 or the rlp23-2 mutant. At 12 hpi, some of the germinating conidia exhibited invasive hyphae inside plants ( Fig. 2A). Specifically, about 25% of germinating conidia in WT plants displayed invasive hyphae at 12 hpi (Fig. 2B). In contrast, about 45% of germinating conidia in the rlp23-2 mutant showed invasive hyphae at the same time point (Fig. 2B). Thus, the invasion ratio of B. cinerea was clearly higher in the rlp23-2 mutant compared with WT plants at 12 hpi. Considering the higher amount of B. cinerea detected at 12 hpi in the rlp23-2 mutant compared with the wild type (Fig. 1C), these results suggest that the enhanced susceptibility to B. cinerea observed in the Arabidopsis rlp23 mutants is at least partially attributable to the reduction in pre-invasive resistance against the pathogen. nlp peptides derived from two types of B. cinerea NLPs induce Arabidopsis immunity in an RLP23-dependent manner. Because RLP23 is required for the perception of nlp peptides derived from NLPs and the subsequent activation of immunity 26 , next we investigated whether the reduced immunity against B. cinerea observed in the rlp23 mutants is related to the RLP23-dependent recognition of NLP proteins secreted by B. cinerea. It was previously reported that B. cinerea secretes two types of NLPs, i.e., BcNEP1 and BcNEP2, during infection of tomato (Fig. 3) 12 . To examine whether B. cinerea also secretes these two NLPs during infection of A. thaliana, we sprayed a conidial suspension of B. cinerea on Arabidopsis plants and quantified the expression level of the BcNEP1 and BcNEP2 genes by quantitative reverse transcription PCR (RT-qPCR). BcNEP2 was preferentially expressed in the late phase of B. cinerea infection, which is in accordance with the expression pattern of NLP genes in other fungal pathogens [30][31][32] (Fig. 4A). The expression of BcNEP2 was clearly induced at 72 hpi and was highest at 96 hpi. In contrast, the expression level of BcNEP1 was highest at 12 hpi, a time point BcCutA was quantified by qPCR in the extracted genomic DNA. The Arabidopsis ⍺-shaggy kinase gene (AtASK) was used as a reference. Means and SDs were calculated from three independent samples. The statistical analysis was conducted using two-tailed Student's t-tests. The level of BcCutA was compared at the same time points between Col-0 and the rlp23-2 mutant plants (*P < 0.05). www.nature.com/scientificreports/ We then wondered whether BcNEP1 was actually recognized by Arabidopsis RLP23. Focusing on the amino acid sequences of BcNEP1 and BcNEP2, we performed multiple alignment of the amino acid sequences of BcNEP1, BcNEP2, and ChNLP1 of Colletotrichum higginsianum, PpNPP1 of Phytophthora parasitica and BsNLP1 of Bacillus subtilis using Clustal/Omega 2,18,33 . This analysis showed that these proteins had a high similarity in their amino acid sequences (Fig. 3). However, in the case of the predicted nlp24 region, BcNEP1 exhibited a difference vs. the other four NLPs of microorganisms in three kingdoms. Although the nlp24 region of the five NLPs, including BcNEP1, commonly possessed two conserved regions, i.e., conserved region I and conserved region II, the corresponding sequence of BcNEP1 contained additional three amino acids (NVV) between conserved region I and conserved region II (Fig. 5A).
The 20 and 24 amino acids derived from the nlp sequence of BcNEP2 were reported to activate Arabidopsis immunity 2,6 ; however, it remains unknown whether the corresponding 27 amino acids derived from BcNEP1 (termed Bcnlp27 BcNEP1 here) can activate immunity. Here, we named the 24 amino acids derived from the BcNEP2 nlp sequence as Bcnlp24 BcNEP2 . Because Bcnlp27 BcNEP1 contains three additional amino acids in the nlp24 sequence, we were not able to exclude the possibility that the insertion blocks its recognition by Arabidopsis RLP23. To test whether Bcnlp27 BcNEP1 can be recognized by RLP23 and trigger immunity, we measured the amount of ROS after treatment of Arabidopsis leaf discs with each synthetic nlp peptide. We found that Bcnlp24 BcNEP2 induced ROS production, as reported previously (Fig. 5B). Notably, Bcnlp27 BcNEP1 also induced ROS production, to the same extent as did Bcnlp24 BcNEP2 (Fig. 5B). In contrast, ROS production was suppressed in the rlp23 mutants (Fig. 5B). These results demonstrate that not only Bcnlp24 BcNEP2 but also Bcnlp27 BcNEP1 triggers the PAMP-triggered immune response in A. thaliana in an RLP23-dependent manner, supporting the idea that the recognition of BcNEPs, especially BcNEP1, by RLP23 at the early infection phase contributes to Arabidopsis pre-invasive immunity against B. cinerea.  www.nature.com/scientificreports/ necrotrophic fungal pathogen that induces lesions in Arabidopsis leaves, similar to B. cinerea 34 . Therefore, next we investigated whether the A. brassicicola strain Ryo-1 also expresses its NLP gene at the early infection phase in Arabidopsis. Via the alignment of the A. brassicicola genome sequence with the A. alternata mRNA sequence using Exonerate 35 , we found that A. brassicicola carries only one NLP homologue in its genome sequence (Supplementary Fig. 1). We designated this gene AbNLP1 and quantified its expression by RT-qPCR during A. brassicicola infection. We observed that AbNLP1 was expressed at very low levels at 24 hpi, increased to 48 hpi and then decreased towards 72 hpi (Fig. 6A). Importantly, AbNLP1 expression was not detected at 4 and 12 hpi. We also found that the A. brassicicola strain Ryo-1 starts to invade Arabidopsis at around 12 hpi 36 . Collectively, these results indicate that AbNLP1 of A. brassicicola is not expressed before/during invasion, in contrast to BcNEP1 of B. cinerea. Next, to determine whether the lack of RLP23 also reduces the immunity of Arabidopsis against A. brassicicola, a conidial suspension of the A. brassicicola strain was drop-inoculated on the rlp23 mutants. The rlp23-1 and rlp23-2 mutants exhibited immunity against A. brassicicola at 4 dpi, similar to that observed in the WT plants (Fig. 6B) and in contrast to the case of B. cinerea inoculation. The size of the lesions caused by A. brassicicola in the rlp23-1 and rlp23-2 mutants was not significantly different from that of Col-0 (Fig. 6C), whereas the cyp71A12 cyp71A13 mutant exhibited enhanced susceptibility to the pathogen, consistent with our recent finding 36 . Thus, RLP23 is not essential for the Arabidopsis immunity against A. brassicicola.

The necrotrophic pathogen
We also investigated the possible role of RLP23 toward a fungal pathogen taking infection strategies that are distinct from those of B. cinerea. C. higginsianum is a hemi-biotrophic fungal pathogen, i.e., the pathogen initially establishes a biotrophic infection in host cells, which is followed by a necrotrophic phase that leads to cell death and the emergence of pathogenic lesions 37 . The host plant species of C. higginsianum is Brassica rapa, and it also infects A. thaliana 38 . We inoculated C. higginsianum (MAFF305635) on the Arabidopsis rlp23 plants by dropping a conidial suspension of the pathogen and measured the lesion size at 5 dpi. The rlp23-1 and rlp23-2 mutants did not exhibit enhanced susceptibility to C. higginsianum, and there was no significant difference compared with Col-0 regarding the size of the lesions caused by C. higginsianum (Supplementary Fig. 2). These results showed that RLP23 is of particular importance to the immunity against B. cinerea, whereas it is dispensable for the immunity against A. brassicicola and C. higginsianum. www.nature.com/scientificreports/ Bcnlp peptides enhance the Arabidopsis immunity against A. brassicicola. The results described above strongly suggest that the perception of Bcnlp27 BcNEP1 of BcNEP1 at the early infection phase contributes to Arabidopsis immunity against B. cinerea. To address this contention further, we inoculated A. brassicicola on Arabidopsis WT plants by dropping a conidial suspension together with the Bcnlp27 BcNEP1 peptide. We found that Arabidopsis plants that were inoculated with the pathogen and Bcnlp27 BcNEP1 concomitantly showed enhanced resistance against A. brassicicola compared with plants that were inoculated with the pathogen alone (Fig. 7A). The lesion size caused by A. brassicicola was significantly reduced in the Bcnlp27 BcNEP1 -treated plants compared with the plants that were inoculated with A. brassicicola alone (Fig. 7B). Similar results were obtained for the co-inoculation of A. brassicicola and the Bcnlp24 BcNEP2 peptide (Fig. 7). These results demonstrate that the activation of the nlp-induced immunity in the early infection phase strengthens the Arabidopsis immunity against A. brassicicola.

Discussion
RLP23 encodes an RLP that is essential for the perception of the nlp24 peptides and subsequent activation of immune responses. However, the actual contribution of RLP23 to the Arabidopsis immunity against pathogens remains poorly understood. In this study, we revealed that RLP23 was involved in Arabidopsis immunity against the necrotrophic fungal pathogen B. cinerea (Fig. 1). In contrast, RLP23 was dispensable for the immunity against the necrotrophic fungus A. brassicicola and the hemi-biotrophic fungus C. higginsianum (Fig. 6B,C and Supplementary Fig. 2). Interestingly, microscopic observation revealed that the invasion ratio of B. cinerea was increased in the rlp23 mutant compared with WT plants (Fig. 2B). Furthermore, BcNEP1 was expressed before/during invasion, in contrast to that observed for BcNEP2, which was expressed at the late infection phase (Fig. 4A). Furthermore, not only the nlp sequence of BcNEP2 (Bcnlp24 BcNEP2 ) but also that of BcNEP1 (Bcnlp27 BcNEP1 ) was recognized by RLP23, although the Bcnlp27 BcNEP1 sequence contained three additional amino acids that were absent in BcNEP2 and in the NLPs of C. higginsianum, P. parasitica (oomycete) and B. subtilis (bacterium) (Fig. 5A). Furthermore, we found that A. brassicicola expressed its NLP gene (AbNLP1) preferentially at the late infection phase, rather than before/during pathogen invasion (Fig. 6A). We also showed that the co-inoculation of A. brassicicola with the Bcnlp27 BcNEP1 peptide strengthened the immunity against A. brassicicola (Fig. 7). www.nature.com/scientificreports/ Collectively, these results revealed that RLP23 contributes to the Arabidopsis pre-invasive resistance to B. cinerea via the recognition of the NLP proteins, mainly BcNEP1, which are secreted at the early infection phase.
In hemi-biotrophic pathogens, including Colletotrichum fungi, it was proposed that NLP proteins function in the transition from the biotrophic phase to the necrotrophic phase and in the subsequent maintenance of the necrotrophic phase 32 . This idea is consistent with the findings that NLP expression is restricted to the late infection phase of hemi-biotrophic pathogens 31,39,40 . However, our previous work on C. orbiculare-cucurbit interactions suggested another explanation for the restriction of NLP expression to the late phase 25 . We generated a transgenic strain of C. orbiculare that expressed the NLP gene not only at the late infection phase but also at the early infection phase, and found that the generated C. orbiculare transformants failed to infect cucurbits by activating their pre-invasive resistance 25 . In this case, cucurbits did not recognize the nlp24 region; rather, they recognized the C-terminal region of the NLP protein 25 .
The results reported here provide the first example of the activation of pre-invasive resistance by a WT fungal pathogen, rather than an artificial transgenic pathogen 25 , via plant recognition of the pathogen NLP protein, which further supports the relationship between the restriction of NLP expression to the late phase and the avoidance of pathogen NLP recognition by plants at the early phase. Our study also provides evidence that supports the idea that PRR-dependent PAMP recognition contributes to pre-invasive resistance to fungal pathogens in higher plants. This finding is also consistent with the knowledge that non-host plant resistance against a broad range of fungal pathogens largely relies on pre-invasive immune responses [41][42][43][44][45] .
Because Arabidopsis recognizes BcNEP1 via RLP23, the expression of BcNEP1 at the early infection phase has a negative impact on B. cinerea infection. Why does B. cinerea express BcNEP1 at the early infection phase? Does B. cinerea benefit from BcNEP1 expression? We consider that B. cinerea expresses BcNEP1 to induce cell death in host plants at the early infection phase, thus facilitating its infection, because B. cinerea is a necrotrophic fungal pathogen. Interestingly, BcNEP1 reportedly induces necrosis more efficiently compared with BcNEP2 13 . However, a single-gene knockout mutant of BcNEP1 was previously generated and the diameters of the lesions caused by the ΔBcnep1 mutant were not significantly different from those caused by the B. cinerea WT strain in detached tomato and Nicotiana benthamiana leaves 12 . Thus, there is currently no direct evidence that BcNEP1 actually contributes to the B. cinerea virulence. Also, it is noteworthy that the nlp24 peptide is recognized by www.nature.com/scientificreports/ a limited number of plant species, including several Brassicae species 2 . Therefore, we speculate that B. cinerea evolved to express BcNEP1 at the early phase to cause cell death in host plants that are not able to recognize BcNEP1, because B. cinerea fungi generally exhibit a broad host range. The RLP23-triggered antifungal immune pathways that are crucial for pre-invasive resistance against B. cinerea also remain to be elucidated. A recent report revealed that the nlp24 peptide strongly induces ethylene production in an RLP23-dependent manner compared with flg22 26 . Ethylene is a plant hormone that regulates diverse developmental and physiological processes, including fruit ripening, seed germination, abscission, senescence and immunity to pathogens 46 . In particular, ethylene contributes to immunity against necrotrophic pathogens, including B. cinerea, compared with other types of pathogens 47 . For example, ethylene insensitive 2 (EIN2) is an endoplasmic reticulum (ER) membrane-localized positive regulator of ethylene, and the Arabidopsis ein2 mutants are more susceptible to B. cinerea infection 48,49 . Constitutive expression of ERF1, an early ethylene responsive gene 50 , also enhances the Arabidopsis resistance against B. cinerea 51 . Therefore, we speculate that the induction of ethylene production by BcNEP1 might be involved in the activation of antifungal immune responses that restrict the entry of B. cinerea. Further studies are necessary to elucidate at the molecular level the pre-invasive immune responses that are triggered by the RLP23-dependent nlp recognition.
In contrast to that observed for B. cinerea, the rlp23 mutants showed no enhanced susceptibility to A. brassiciacola (Fig. 6B,C). It is likely that RLP23 did not contribute to the immunity against A. brassiciacola because of (i) the late timing of AbNLP1 expression and (ii) the low expression level of AbNLP1 (too low to activate immunity). Interestingly, the ein2 mutants did not exhibit enhanced susceptibility to A. brassicicola, unlike that observed for B. cinerea, even though A. brassicicola is also a necrotrophic pathogen 52 . Alternatively, the Arabidopsis immune www.nature.com/scientificreports/ responses that are effective against A. brassicicola might be distinct from those that are effective against B. cinerea. However, we showed that the co-inoculation of A. brassicicola with the Bcnlp peptides enhanced the Arabidopsis immunity to this pathogen (Fig. 7). Therefore, we assume that the immune responses triggered by the nlp peptides are commonly effective against B. cinerea and A. brassicicola.

Materials and methods
Plant growth. Seeds of A. thaliana were sown on soil or rockwool (Grodan), incubated for 2 days at 4 °C in the dark and then grown at 22 °C in 16 h daylight with water for soil or Hoagland medium for rockwool.
Arabidopsis T-DNA insertion lines. All mutant lines used in this study were derived from Col-0. The For the analysis of lesion development after the inoculation assay, four drops of 5 µl of a conidial suspension of each pathogen were placed onto each leaf without wounds, and at least 24 lesions were evaluated in each experiment. The developed lesions were quantified using the ImageJ image analysis software (https ://image j.net). For pathogen invasion assay, trypan blue staining of inoculated leaves was conducted according to Koch and Slusarenko (1990) 55 . For the trypan blue assay, at least 200 conidia were investigated in each treatment area. www.nature.com/scientificreports/ Quantitative RT-PCR analysis and quantitative PCR analysis. For the quantification of the amount of B. cinerea, 4-5-week-old plants grown on rockwool were drop-inoculated with B. cinerea (1 × 10 5 conidia/mL) and three leaves were collected. Total DNA was extracted using a DNeasy Plant Mini Kit (Qiagen). For the quantification of gene expression levels during pathogen infection, 4-5-week-old plants grown on rockwool were spray-inoculated with B. cinerea (3 × 10 5 conidia/mL) or A. brassicicola (5 × 10 5 conidia/mL) and three leaves were collected. Total RNA was extracted using an RNeasy Plant Mini Kit (Qiagen). The Takara Prime Script RT Master Mix (Takara Bio Inc.) was used for cDNA synthesis. Takara TB Green Premix Ex Taq I and a Thermal Cycler Dice Real Time System TP800 (Takara) were used for both quantitative RT-PCR and quantitative PCR using the primers listed in Supplementary Table 1. For quantitative RT-PCR, the Botrytis ubiquitin (BcUBQ, XM_001556819.1), Arabidopsis ubiquitin-conjugating enzyme 2 (AtUBC, At5g25760) and A. brassicicola elongation factor 1 (AbEF1, GEMY01015044) genes were used as internal controls to normalize the levels of cDNA. For quantitative PCR, Arabidopsis α-shaggy kinase (AtASK, At5g26751) was used as the control. The primers used for the amplification of BcCutA 29 , AtASK 29 , AtUBC 56 , BcUBQ 57 and AbEF1 58 were as described previously. Relative gene expression was calculated with ∆Ct method, subtracting Ct values of internal control gene from that of target gene, and represented as 2^(−∆Ct). Relative fold change was calculated with ∆∆Ct method, normalizing ∆Ct values with ΔCt value of Col-0 at 0 h, and represented as 2^(−∆∆Ct) 59 .
Synthetic peptides. The synthetic peptides used in this study were as follows: Bcnlp27 BcNEP1 , GIMYAW-YFPKDQPAAGNVVGGHRHDWE; and Bcnlp24 BcNEP2 , AIMYSWYMPKDEPSTGIGHRHDWE. Each peptide was dissolved in sterile distilled water.

ROS measurements.
The ROS assay was performed as described previously, with some modifications 60 .
Leaf discs from true leaves of 4-5-week-old plants grown on soil were kept in the dark overnight with 50 µl of sterile distilled water in a 96-well plate. A reaction solution (50 μl) including 400 µM luminol, 20 µg/ml of horseradish peroxidase and 1 µM synthetic peptides (Bcnlp27 BcNEP1 and Bcnlp24 BcNEP2 ) was added into each well just before measurement. Luminescence was measured for about 60 min every 2 min using Luminoskan Ascent (Thermo Fisher Scientific). At least eight leaf discs were measured per experimental plot.