The Arabidopsis thaliana LysM‐containing Receptor‐Like Kinase 2 is required for elicitor‐induced resistance to pathogens

Abstract In Arabidopsis thaliana, perception of chitin from fungal cell walls is mediated by three LysM‐containing Receptor‐Like Kinases (LYKs): CERK1, which is absolutely required for chitin perception, and LYK4 and LYK5, which act redundantly. The role in plant innate immunity of a fourth LYK protein, LYK2, is currently not known. Here we show that CERK1, LYK2 and LYK5 are dispensable for basal susceptibility to B. cinerea but are necessary for chitin‐induced resistance to this pathogen. LYK2 is dispensable for chitin perception and early signalling events, though it contributes to callose deposition induced by this elicitor. Notably, LYK2 is also necessary for enhanced resistance to B. cinerea and Pseudomonas syringae induced by flagellin and for elicitor‐induced priming of defence gene expression during fungal infection. Consistently, overexpression of LYK2 enhances resistance to B. cinerea and P. syringae and results in increased expression of defence‐related genes during fungal infection. LYK2 appears to be required to establish a primed state in plants exposed to biotic elicitors, ensuring a robust resistance to subsequent pathogen infections.

These early signalling events are followed by a more delayed induction of responses, including the expression of defence-related genes, the biosynthesis of antimicrobial compounds and the deposition of callose in the cell wall, resulting in a so-called pattern-triggered immunity (PTI) effective against a broad range of pathogens (Boller & Felix, 2009;Macho & Zipfel, 2014;Tang, Wang, & Zhou, 2017).
Most studies on the role and mode of action of PRRs in plant immunity have focused on responses occurring within minutes or hours upon elicitation (MAPK activation; oxidative burst; early gene expression). However, the establishment of a lasting status of enhanced resistance in response to treatments with MAMPs or DAMPs likely depends on long-term responses that are at least partially dependent on the accumulation of phytohormones, such as salicylic acid (SA), ethylene and jasmonates (Berens, Berry, Mine, Argueso, & Tsuda, 2017;Broekgaarden, Caarls, Vos, Pieterse, & Van Wees, 2015;De Vleesschauwer, Xu, & Höfte, 2014;Wasternack, 2007). For instance, resistance against the necrotrophic fungus Botrytis cinerea induced in Arabidopsis by pre-treatments with flg22 or OGs requires the biosynthesis of camalexin (Ferrari et al., 2007) and an intact ethylene signalling pathway (Gravino, Savatin, Macone, & De Lorenzo, 2015).
Inducible defences are costly, and their improper activation might reduce plant fitness (Huot, Yao, Montgomery, & He, 2014). Plants have therefore evolved the ability to acquire a pre-conditioned state of defence after specific stimulation by microbial infections or environmental stresses. Exposure to MAMPs or DAMPs might increase the activation of defence responses upon subsequent perception of the same or a different elicitor, ensuring a robust resistance with a low fitness cost. However, the relative contribution of the perception and signalling mediated by specific PRRs in elicitor-induced resistance is not well understood. In this work, we have investigated the role of LYK proteins in Arabidopsis basal and elicitor-induced resistance to pathogens. Our results indicate that basal resistance to B. cinerea does not require an intact chitin perception system, but resistance to this pathogen induced by exogenous chitin is impaired by mutations in CERK1, LYK2 and LYK5. Notably, LYK2 is required for enhanced resistance to fungal infection observed after pre-treatments with different elicitors and contributes to basal and induced resistance to Pst DC3000. Our results suggest that LYK2 is largely dispensable for chitin perception and early signalling and has a more general role in the regulation of elicitor-induced priming of defence responses during pathogen infection.

| Plant material and growth conditions
Arabidopsis thaliana (L.) Heynh plants used in this work were all in the Columbia-0 (Col-0) background. Seeds of cerk1-2 (Miya et al., 2007) were a kind gift of Dr. Naoto Shibuya (Meiji University, Japan); lyk5-2 (Salk_131911C) seeds were kindly provided by Dr Elena Petutschnig (Georg-August-University of Göttingen, Germany). The lyk2-1 (Salk_152226) and lyk2-2 (Salk_012441) lines were obtained from the Nottingham Arabidopsis Stock Centre (NASC) and brought to homozygosity before further characterization. For seedlings assays, Arabidopsis seeds were sterilized in 1.6% sodium hypochlorite and 0.01% sodium dodecyl sulphate (SDS) and stored at 4 C in the dark for 2 days. Then, seeds were evenly distributed on 12-well-plates containing 0.5x Murashige-Skoog (MS) basal salts, 0.5% sucrose, pH 5.6, and transferred to a growth chamber with 12 hr day/12r h night cycle at 22 C with 50% relative humidity. For leaf assays, seeds were distributed in plates containing 0.5x MS basal salts, 0.5% sucrose, 0.7% plant agar, pH 5.6. After sowing, plates were stored at 4 C in the dark for 2 days, then transferred to a growth chamber with 12 hr day/12 hr night cycle at 22 C. Ten-day-old seedlings were transferred to soil and grown in the growth chamber under 12 hr day/12 hrnight cycle at 22 C with 50% relative humidity. For
Before plant inoculation, spores were suspended at a final concentration of 5 Â 10 5 conidiospores ml À1 in 24 g l À1 potato dextrose broth (Difco, Detroit, USA) and incubated for 2-3 hr at room temperature (RT). Four-week-old Arabidopsis plants were inoculated placing 5 μl drops of the spore suspension on each side of the middle vein of fully expanded rosette leaves. Plants were covered with a clear plastic dome to ensure high humidity and incubated at 22 C with a 12 hr photoperiod. Lesion areas were determined 48 hr postinfection by measuring water-soaked lesions, using ImageJ software (https:// imagej.nih.gov/ij/). For elicitor-induced protection, plants were sprayed with water, 200 μg ml À1 OGs, 1 μM flg22 or 100 μg ml À1 colloidal chitin 24 hr before inoculation, as previously described (Ferrari et al., 2007).
Pst DC3000 was inoculated in LB liquid medium containing 25 μg ml À1 rifampicin and grown under agitation (200 rpm) for 8-12 hr at 28 C, until OD 600 = 0.6-1.0. Fully expanded rosette leaves of four-week-old plants were syringe-infiltrated with a suspension of bacteria [1 Â 10 6 colony-forming units (cfu) ml À1 ] as previously described (Katagiri, Thilmony, & He, 2002). Leaf discs were collected right after inoculation, 2 and 3 days after infiltration, and ground in water to collect bacteria. For each sample, a 1:10 dilution series was plated on solid LB medium containing rifampicin, and colonies were counted after incubation at 28 C for approximately 3 days.
Proteins were transferred to a nitrocellulose membrane using TransBlot Turbo (Bio-Rad). The filter was stained for 10 min with As controls, primary antibodies against actin (Sigma) were used.

| Oxidative burst assays
Hydrogen peroxide production was measured by a luminol-based assay as previously described (Galletti, Ferrari, & De Lorenzo, 2011).
Leaf discs (0.2 cm 2 ) from four-week-old plants were washed for 2 hr with water and incubated overnight in a 96-well plate (one disc per well). Water was then replaced with a solution of luminol (Sigma-Aldrich; 30 mg ml À1 ) and horseradish peroxidase (Sigma-Aldrich; 20 mg ml À1 ) containing 100 nM flg22. For chitin elicitation (100 μg ml À1 ), discs were vacuum infiltrated with the chitin solution for 2 min before the addition of the luminol/peroxidase solution.
Plates were analysed for 40 min using a GloMax 96 microplate luminometer (Promega) and a signal integration time of 1 s. Luminescence was expressed in relative light units (RLUs). (v/v) glycerol and examined by a UV epifluorescence microscope (Nikon, Eclipse E200) equipped with a cooled charge-coupled device camera (DS-Fi1C). Images were acquired with the Nis Elements AR software (Nikon). Fluorescence intensity in each image was calculated using ImageJ (https://imagej.nih.gov/ij/). UBIQUITIN 5 (UBQ5, At4G05320). Three technical replicates were performed for each sample, and data analysis was done, with minor modifications, as previously described (Redwan et al., 2016) using LinRegPCR software (Ruijter et al., 2013). Sequences for all primers used for quantitative PCR and identifiers of the corresponding genes are listed in Table S1. Analysis of the expression of LYK genes from publicly available microarray data was performed using Genevestigator (Hruz et al., 2008).

| Generation of constructs and transgenic plants
The Red Fluorescent Protein (RFP) coding sequence was amplified by PCR from the pSAT6-mRFP-N1 plasmid (Invitrogen) with a highfidelity DNA polymerase (Roche), using the following primers: ATCG ATCTAGAGTCGACGGTACCG (RFP-FW) and ATCGAGAGCTCTTA GGCGCCGGTG (RFP-REV). The PCR product was purified, digested with XbaI and SacI (whose sites were introduced with the PCR reaction) and ligated to a pBI-121 vector (Invitrogen), after removal of the GUS cassette with XbaI and SacI. The ligation product was introduced in E. coli DH10B cells by electroporation, and transformed bacteria were selected on LB agar medium containing 50 μg ml À1 kanamycin.
The obtained plasmid (pBI-RFP) was purified and used for the generation of the LYK2-RFP construct. The full-length LYK2 coding sequence was amplified by PCR with a high-fidelity DNA polymerase (Roche) from Col-0 genomic DNA, using the following primers: CATC TCCCTTCTGAGGACCA (LYK2gFw) and GATGAGTTTAGGGCCATGA TGC (LYK2gRev). The PCR product was cloned into the pGEM T-Easy vector (Invitrogen) and used to transform E. coli One Shot ® Omni-MAX™ 2 T1R (Invitrogen) cells by electroporation. The obtained plasmid (pGEM-LYK2) was purified and digested with KnpI and SmaI and the insert was cloned in frame upstream the RFP coding sequence in the pBI-RFP plasmid previously digested with the same enzymes and dephosphorylated. To generate constructs for the overexpression of the untagged version of LYK2, the insert of pGEM-LYK2 was ligated with pBI121, previously digested with KnpI and SmaI. The obtained plasmids (pBI121-LYKL2-RFP and pBI121-LYK2) were introduced into E. coli One Shot ® OmniMAX™ 2 T1R (Invitrogen) and transformed bacteria were selected on LB agar medium containing 50 μg ml À1 kanamycin. To generate LYK2-GFP, the full-length coding sequence of LYK2 was cloned into the pDONR-Zeo plasmid by BP cloning using a Gateway-based system (Invitrogen). The obtained plasmid was then used for LR cloning with the destination plasmid pGWB5 . For generation of 35S:CERK1-GFP, 35S:CERK1-myc and 35S:LYK5-myc constructs for CoIP experiments, the full-length coding sequences of LYK5 and CERK1 were cloned into the pDONR201 plasmid (Invitrogen) using the following primers: for LYK5, and pGWB17 (myc-tag), respectively

| Co-immunoprecipitation assays
A. tumefaciens was grown overnight in LB medium containing the appropriate antibiotics, collected by centrifugation, and then suspended in 10 mM MgCl 2 containing 100 mM acetosyringone. After incubation at RT for at least 2 hr, the cultures were diluted to an OD 600 = 0.5.
Leaves of four-week-old N. benthamiana plants were agroinfiltrated using a needleless syringe and plants were returned to the greenhouse for 72 hr. Samples from agroinfiltrated leaves were lysed in a buffer containing 50 mM Tris (PH 7.6), 150 mM NaCl, 0.5% Triton X-100 and protease inhibitor cocktail P9599 (Sigma, MO). Extracts were centrifuged at 14,000 Â g for 15 min at 4 C. Anti-Myc or anti-GFP traps (Chromotek) were used for co-immunoprecipitation experiments according to the manufacturer's instructions. Immunoblot analysis was performed as previously described (Willmann et al., 2011), using antimyc or anti-GFP antibodies (Sigma-Aldrich) at a dilution of 1:3000.

| LYK2 is required for resistance to Botrytis cinerea induced by different elicitors
To investigate the role of LYK2 in Arabidopsis immunity, two homozygous insertional lines were obtained: lyk2-1 (SALK_152226), carrying a predicted T-DNA insertion in the second exon, corresponding to the extracellular domain of LYK2, and lyk2-2 (SALK_012441), with a predicted insertion at the very beginning of the first exon ( Figure S1a, b). Untreated seedlings of both mutants showed significantly decreased levels of LYK2 transcripts, compared to the wild type ( Figure S1c). In addition, we generated two independent homozygous lines (35S:LYK2 line 1.1 and 5.15) overexpressing LYK2 ( Figure S1d).
Both lines accumulated high levels of LYK2 transcripts in the absence of any treatment ( Figure S1d).
Available microarray data indicate that basal LYK2 transcript levels are quite low in seedlings and leaves of WT plants, compared to CERK1, LYK3 and LYK5, are similar to those of LYK4 ( Figure S2a).
We examined the expression of LYK2 in WT and mutant seedlings treated with different elicitors. Compared to water-treated seedlings, LYK2 transcripts in the wild type increased two-to four-fold after 1 hr of treatment with OGs, flg22 or chitin ( Figure S2b). In contrast, expression of LYK2 in both lyk2 mutants did not significantly differ after elicitation (Figure 2b). These results indicate that lyk2-1 and lyk2-2 have impaired basal and elicitor-triggered expression of LYK2. Basal expression of CERK1 and LYK5 was comparable in WT and lyk2 seedlings; transcripts for both genes slightly increased to a similar extent after 1 h of elicitation with flg22 or chitin, though expression levels were quite variable ( Figure S2c,d).
Notably, compared to the WT, lyk2 mutants accumulated greater transcript levels of both CERK1 and LYK5 after 3 h of treatment with chitin, but not with flg22 ( Figure S2c,d), suggesting that reduced expression of LYK2 might trigger, in response to this MAMP, a compensatory response resulting in the enhanced expression of other LYK genes.
To investigate the role of Arabidopsis LYKs in basal and chitininduced resistance to fungal infection, we evaluated the severity of symptoms caused by B. cinerea in lyk2-1 and lyk2-2, as well as in cerk1-2 and lyk5-2, which carry loss-of-function mutations in CERK1 and LYK5, respectively, and are impaired in chitin perception Miya et al., 2007;Wan et al., 2008). 3.2 | LYK2 is dispensable for early chitin perception and signalling but is required for chitin-induced callose deposition It was previously reported that lyk2 mutants do not show defects in chitin-induced production of ROS  and F I G U R E 1 Impact of lyk mutations on basal and elicitor-induced resistance to B. cinerea. Leaves of four-week-old WT, lyk2-1, lyk2-2, lyk5-2 and cerk1-2 plants were sprayed with water or with 100 μg ml À1 chitin (a), 200 μg ml À1 OG or 1 μM flg22 (b). After 24 h, leaves were inoculated with a B. cinerea spore suspension (5 Â 10 5 spores ml À1 ). (c) Leaves of four-week-old WT, lyk2-1, 35S:LYK2 1.1 and 15.5 (OE 1.1 and OE 5.15) plants were inoculated with B. cinerea. Lesion areas were measured 48 hours after inoculation. Data are means ± SE (n = 12); asterisks indicate statistically significant differences between WT and mutants, according to Student's t-test (**, p > .01; ***, p < .001). These experiments were repeated three times with similar results expression of WRKY53 and MPK3 (Wan et al., 2008). It is however possible that LYK2 regulates specific subsets of responses to this MAMP. After 5 and 10 min of treatment with two different doses of chitin (10 and 25 μg ml À1 ), phosphorylation of MPK3, MPK4 and MPK6, one of the earliest responses to this elicitor (Pitzschke & Hirt, 2009;Ramonell et al., 2005), was similar in WT, lyk2-1 and lyk2-2 seedlings (Figure 2a,b). ROS production in response to chitin and flg22 was comparable in WT, lyk2-1 and lyk2-2 seedlings (Figure 2c-f). Expression of two elicitor-responsive marker genes, PHYTOALEXIN DEFICIENT 3 (PAD3) (Asai et al., 2002;Ferrari et al., 2007;Zhou, Tootle, & Glazebrook, 1999) in response to chitin was also unaffected in lyk2-1 and lyk2-2 seedlings, compared to the wild type (Figure 2g,h). We also analysed if a later response to chitin, namely callose deposition, requires LYK2. Adult rosette leaves were infiltrated with water, chitin (100 μg ml À1 ) or flg22 (100 nM) and the presence of callose was revealed by aniline blue staining. Leaves of both lyk2 mutants, in comparison to WT plants, showed a significant reduction of the intensity and number of callose deposits in response to chitin infiltration, but not in response to flg22 Early chitin-induced responses are unaffected in lyk2 mutants. (a-b) 10-day-old WT, lyk2-1 and lyk2-2 seedlings were treated for 5 (a) and 10 min (b) with water (W) or chitin at the concentration of 10 μg ml À1 (C10) or 25 μg ml À1 (C25). Phosphorylated MPK3, MPK4, MPK6 and MPK11 were detected by immunoblot using an anti-P44/P42 antibody. Total MPK3 and MPK6 were detected by immunoblot using an anti-MPK3 and anti-MPK6 antibodies. Arrows indicate the molecular weight (in kDa) of the marker bands. Equal loading was evaluated by Ponceau-S Red staining and using an anti-Actin antibody. (c-f) Leaf discs of four-week-old WT, lyk2-1 and lyk2-2 plants were treated for the indicated times with 100 μg ml À1 chitin (c,d) or 100 nM flg22 (e,f). H 2 O 2 production was measured with a luminolbased assay and expressed in relative light units (RLU s À1 ). Data points represent the average of at least 12 discs ± SE. Bars in (d) and (f) represent the average of total H 2 O 2 production ± SE. Differences between total RLUs in WT and lyk2-1 or lyk2-2 were not significant (ns), according to Student's t-test (p > .05). (g, h) FRK1 (g) and PAD3 (h) expression in WT and lyk2 seedlings treated with water or chitin (5 and 25 μg ml À1 ) for 1 (g) and 3 h (h) was analysed by qRT-PCR using UBQ5 as control. Data are means ± SE (n = 3 biological replicates). Asterisks indicate statistically significant differences according to Student's t-test (***, p < .001) These experiments were repeated three times with similar results [Colour figure can be viewed at wileyonlinelibrary.com] (Figure 3a,b). These data suggest that LYK2 does not contribute to the initial perception of chitin and has a minor, if any, role in early signalling events that lead to MAPK activation, oxidative burst, and early gene expression, but is necessary for full activation of callose deposition, a late response to this elicitor. Moreover, lyk2 mutants are normally responsive to flg22, though LYK2 is required for flg22-induced resistance to fungal infection.

| LYK2 is required for enhanced defence responses to chitin or fungal infection in plants pretreated with elicitors and contributes to resistance to Pst DC3000
Basal expression of LYK2 is very low but increases in seedlings treated with different MAMPs and DAMPs ( Figure S2b Figure S5). Chitin-induced MAPK phosphorylation in water-pre-treated seedlings was comparable in WT and lyk2-2 seedlings at both 10 and 20 min after elicitation (Figure 4a,b) and was slightly reduced in lyk2-1 after 20 min of elicitation (Figure 4b and Figure S5). After flg22 pre-treatment, chitininduced phosphorylation at 10 min was comparable in all genotypes ( Figure 4a and Figure S5a) but was strongly reduced at 20 min in lyk2-1 and lyk2-2 seedlings, compared to the wild type (Figure 4b and Figure S5b). As expected, chitin failed to induce any detectable activation of MAPKs in cerk1-2, even after pre-treatment with flg22 (- Figure S5a,b). These results confirm that LYK2 is dispensable for initial MAPK phosphorylation triggered by chitin treatment but suggest that, in plants previously exposed to an elicitor, MAPKs are dephosphorylated more rapidly if LYK2 is not functional.
Pre-treatments of WT plants with flg22 significantly enhanced chitin-induced expression of FRK1 and PAD3 ( Figure 5); this enhanced expression was significantly impaired in lyk2-1 and lyk2-2 plants ( Figure 5). In the absence of pre-treatments with flg22, lyk5-2 and cerk1-2 did not show induction of PAD3 and FRK1 in response to chitin, as expected; however, expression of both marker genes was induced by chitin in flg22-pre-treated lyk5-2 seedlings, though to a lesser extent than in the wild type, whereas cerk1-2 was almost completely unresponsive to chitin, even after previous exposure to flg22 ( Figure 5). Taken together, the results described above confirm that CERK1 and LYK5, but not LYK2, are required for basal responsiveness to chitin in plants and that CERK1 is absolutely required for chitin responsiveness, and suggest that Arabidopsis, upon flg22 elicitation, acquires the ability to partially respond to chitin also in the absence of a functional LYK5. Moreover, these data indicate that LYK2 is necessary to enhance responses downstream chitin perception in plants pre-treated with flg22.
To further investigate the role of LYK2 in elicitor-induced resistance, the expression of two defence-related genes, PAD3 and PR-1, F I G U R E 5 Priming of chitininduced gene expression in lyk mutants. 10-day-old WT, lyk2-1, lyk2-2, lyk5-2 and cerk1-2 seedlings were pre-treated with water or 10 nM flg22 for 24 h and subsequently treated with water or chitin (25 μg ml À1 ) for 1 h. FRK1 and PAD3 expression was measured by qRT-PCR and normalized using UBQ5. (a) Bars represent mean expression ± SE (n = 3 biological replicates), asterisks indicate statistically significant differences between WT and mutants, according to Student's t-test (*, p < .05; **, p < .01; ***, p < .001). (b) Expression of FRK1 and PAD3, relative to WT seedlings pretreated with water and then treated with water, of the same samples as in (a). Bars represent mean fold-change ± SE (n = 3 biological replicates). Different letters indicate statistically significant differences, according to one-way ANOVA followed by Tukey's HSD test (p < .01). This experiment was repeated three times with similar results (a) (b) F I G U R E 4 Chitin-triggered MAPK activation in lyk2 mutants pre-treated with elicitors. 10-day-old WT, lyk2-1 and lyk2-2 seedlings were pre-treated with water or flg22 (10 nM) for 24 hr and subsequently treated with water or chitin (25 μg ml À1 ) for 10 (a) or 20 min (b). Phosphorylated MPK3, MPK4, MPK6 and MPK11 were detected by immunoblot using an anti-P44/P42 antibody. Total MPK3 and MPK6 were detected by immunoblot using anti-MPK3 and anti-MPK6 antibodies. Antibodies against actin were used as controls. The arrows indicate the molecular weight of marker bands (in kDa). This experiment was repeated three times with similar results was evaluated in WT and mutant plants treated with water or flg22 and subsequently inoculated with B. cinerea. Camalexin production, catalysed by the cytochrome P450 CYP71B15 encoded by PAD3 (Schuhegger et al., 2006), is necessary for elicitor-induced resistance to B. cinerea (Ferrari et al., 2007) and is primed by elicitor treatments Savatin et al., 2015). PR-1 is a well-known marker for SA-dependent responses (Delaney et al., 1994;Gaffney et al., 1993), which are also important for basal resistance to B. cinerea (Ferrari et al., 2003) and for long-lasting systemic acquired resistance to several pathogens Delaney et al., 1994;Gaffney et al., 1993). After 24 hr of infection, in water-pre-treated plants, a higher expression of PAD3 in lyk2 mutants, and of PR-1 in lyk2-1 and cerk1-2 mutants was observed, compared to the wild type, (Figure 6). In the wild type pre-treated with flg22 or OGs, expression of PAD3 and PR-1 was induced by the fungus to a greater extent than in control plants, indicating a priming effect of the elicitors ( Figure 6 and Table S2), with a stronger effect observed for flg22 compared to OGs. The priming effect of the elicitors on fungal-induced PAD3 expression was also observed in cerk1-2 and lyk5 plants, though transcript levels in these mutants reached lower levels than in the WT ( Figure 6 and Table S2). In infected lyk2 mutants, transcript levels of PAD3 after elicitor pre-treatments were even lower than in waterpre-treated plants, possibly because of the already high levels in the control, and were however comparable to those observed in lyk5 and cerk1-2 ( Figure 6 and Table S2). Elicitor-primed expression of PR-1 during B. cinerea infection was observed in all tested genotypes but was strongly reduced in lyk2-1 and lyk2-2 plants, whereas it was comparable to the WT in lyk5 and cerk1-2 ( Figure 6 and Table S2). These data indicate that LYK2, CERK1 and LYK5 are all required for full priming of PAD3 expression upon infection after elicitor pre-treatments, which is likely dependent on an increased ability of the plant to respond more efficiently to chitin, whereas priming of PR-1 expression is dependent on LYK2 but largely independent of chitin perception.
We subsequently tested if elevated LYK2 expression levels affect responses to chitin treatment or pathogen infection. Expression of PAD3 in response to low doses of chitin was only moderately enhanced in transgenic seedlings overexpressing LYK2 (Figure 7a).
Consistently, MAPK activation in response to chitin was comparable in WT and 35S:LYK2 plants (Figure 7b). Notably, expression of both PAD3 and PR-1 during B. cinerea infection was significantly increased in both water-and elicitor-pre-treated 35S:LYK2 plants (Figure 7c,d).
Moreover, PR-1 expression was enhanced in transgenic plants even before fungal inoculation, regardless of whether they were pretreated with water or elicitors ( Figure 7d). These data suggest that LYK2 overexpression does not affect chitin perception or early signalling, but it can increase basal (in the case of PR-1) and fungal-induced (both PR-1 and PAD3) defence gene expression.

| LYK2 localizes in the plasma membrane and constitutively interacts with LYK5
We next investigated the subcellular localization of LYK2 and its possi- Since LYK5 and CERK1 reside in the PM and physically interact upon chitin elicitation to induce an immune response , we hypothesized that LYK2 might associate with one or both proteins. For this purpose, co-immunoprecipitation (Co-IP) assays were performed in N. benthamiana leaves transiently coexpressing LYK2-GFP and a myc-tagged version of either LYK5 or CERK1 (LYK5-myc and CERK1-myc, respectively). Beside the expected 110 kDa band corresponding to the full-length protein, a band of about 70 kDa, reactive to the anti-GFP antibody, was detectable by immunoblot upon expression of LYK2-GFP (Figure 9a), supporting the hypothesis of a partial cleavage of the cytoplasmic portion of the protein. Anti-GFP beads could immunoprecipitate both forms of LYK2-GFP and could also co-purify LYK5-myc when both proteins were co-expressed in N. benthamiana (Figure 9a), suggesting that LYK2 and LYK5 can physically interact. When CERK1-myc was transiently co-expressed with LYK2-GFP in N. benthamiana, no detectable signal for CERK1-myc could be observed by immunoblot on total protein extracts using the anti-myc antibody, though immunoprecipitation using anti-myc beads revealed that the protein was indeed expressed (Figure 9b). However, under these conditions, no interaction between LYK2-GFP and CERK1-myc could be detected (a) Rosette leaves of four-week-old WT, lyk2-1, lyk2-2 and cerk1-2 plants were sprayed with water (H 2 O) or flg22 and, after 24 hr, infiltrated with Pst DC3000. Bacterial growth was measured at the indicated times (days postinfection, dpi). (b) Rosette leaves of fourweek-old WT and 35S:LYK2 lines 1.1 and 15.5 plants were infiltrated with Pst DC3000, and bacterial growth was measured at the indicated times. Bars indicate mean log 10 of colony-forming units (CFUs) per cm À2 (± SE, n = 12). For each time point, statistically significant differences, according to Student's t-test, between similarly treated WT and mutants are indicated by asterisks (*, p < .05; ***, p < .01); in (a), differences between water-and flg22-treated plants of the same genotype are indicated by pound signs (###, p < .01). The results are representative of three independent experiments and G. cichoracearum, but not to Colletotrichum higginsianum (Miya et al., 2007;Wan et al., 2008). Here we have shown that mutants for CERK1 or LYK5 do not display enhanced susceptibility to B. cinerea. This is not unexpected, since this pathogen can release other MAMPs (Poinssot et al., 2003;Yang, Yu, Liang, Anderson, & Cao, 2018;Zhang, Qiu, Zeng, Guo, & Yang, 2015) and DAMPs Voxeur et al., 2019), whose recognition is probably sufficient to confer a WTlike degree of resistance also in the absence of chitin recognition. On the other hand, activation of a strong immune response occurs when plants are exposed to purified chitin, effectively protecting against subsequent Botrytis infections (Aziz et al., 2006). We found that, in Arabidopsis, this protection requires an intact chitin perception system, since it is abolished in cerk1-2 and lyk5-2 mutants. Unexpectedly, also lyk2 mutants were impaired in chitin-induced resistance, prompting us to hypothesize that LYK2 might play a role in chitin perception and/or signalling, despite previous work failed to reveal a defect in chitin responses in mutants for this gene Wan et al., 2012). Indeed, we found that chitin-induced MAPK phosphorylation, ROS production and early marker gene expression are largely unaffected in lyk2 plants, whereas callose deposition, detected 24 hr after chitin treatment, is reduced in the mutants. These results suggest that LYK2 is not important for chitin perception and early signalling under basal conditions, though it might contribute to ensure proper induction of long-term defence responses triggered by this MAMP.
The minor role of LYK2 in early chitin responses might be due to its low basal expression levels, since lyk2 mutations impair only responses to this MAMP, like callose deposition and enhanced resistance to fungal infection, that are observed several hours after chitin treatment, when LYK2 transcripts have significantly increased above basal levels. In addition, lack of LYK2 results in greater transcript levels of CERK1 and LYK5 at 3 hr after chitin treatment, suggesting the existence of some compensatory mechanism that might partially mask a possible role of this gene in chitin signalling. We therefore hypothesized that increased LYK2 expression after elicitation might increase the ability of the plant to respond to chitin, either exogenously provided or released during fungal infection. This hypothesis was initially corroborated by the observation that gene expression in response to chitin elicitation after flg22 pre-treatment is enhanced only in WT plants, but not in plants lacking LYK2. Since the extracellular domain of LYK2 cannot be pulled down by chitin beads when overexpressed in protoplasts, in contrast to that of CERK1, LYK4 and LYK5 , it is unlikely that LYK2 might directly bind chitin.
Moreover, overexpression of LYK2 did not per se increase MAPK activation and only slightly increased gene expression induced by chitin, suggesting that this protein likely modulates responses downstream of, or independently of the early signalling events triggered by the initial activation of the chitin perception complex. On the other hand, CoIP and BiFC experiments suggest that LYK2 might physically interact with LYK5, hinting to a possible function of LYK2 in modulating the activity of the chitin perception complex. Interestingly, LYK4 also constitutively heterodimerizes with LYK5, and chitin treatments induce the formation of a tripartite complex comprising CERK1, LYK5 and LYK4 (Xue, Li, Xie, & Staehelin, 2019). LYK2, like LYK4, might act as a scaffold protein for LYK5, contributing, in primed plants, to increase the extent or duration of some responses downstream of the activation of the chitin perception complex. This hypothesis is suggested by the more rapid dephosphorylation of MAPKs in lyk2 mutants pre-treated with flg22. However, since lyk2 mutants, but not lyk5-2 or cerk1-2, are not protected against B. cinerea after pretreatments with OG or flg22 and do not show priming of PR-1 expression during fungal infection, LYK2 appears to mediate elicitor-induced resistance also independently of chitin perception.
LYK2-mediated induction of resistance against fungal infection seems to involve the regulation of multiple defence responses. We have previously observed that exogenous OGs or flg22 induce a transient increase of PAD3 transcript levels, that return to basal levels within 12 hr of treatment (Denoux et al., 2008;Ferrari et al., 2007).
However, camalexin accumulates to higher levels during B. cinerea infection when plants are pre-treated with elicitors , suggesting that a previous elicitation primes plants to produce this phytoalexin more rapidly upon pathogen attack. Indeed, we observed elicitor-mediated priming of PAD3 expression in response to fungal infection, which requires an intact chitin perception complex, beside LYK2, since it is also reduced in mutants lacking LYK5 or CERK1. Therefore, in plants pre-treated with MAMPs or DAMPs and subsequently attacked by a fungus, enhanced activation of responses triggered by chitin (or other elicitors) released during infection might increase PAD3 expression and camalexin accumulation, enhancing resistance. We have previously observed that the Arabidopsis bak1-5 mutant, which is strongly impaired in flg22-induced responses Schwessinger et al., 2011), is also compromised in both basal and flg22-induced resistance to B. cinerea (Gravino et al., 2017). Notably, BAK1 is required for responses to different MAMPs, including flg22, PGN and lipopolysaccharides, but not to chitin (Shan et al., 2008), confirming that chitin perception per se is not necessary for basal resistance to the fungus. It was subsequently reported that co-inoculation of B. cinerea with flg22 results in increased resistance to the fungus, and that this resistance depends on the BAK1-mediated phosphorylation of the juxtamembrane domain of CERK1, which in turn increases sensitivity to chitin (Gong et al., 2019). Flg22-induced priming of PAD3 expression requires both LYK2 and an intact chitin perception system, comprising CERK1 and LYK5. The lack of evidence for a role of LYK2 in direct chitin perception, and the observation that LYK2, but not CERK1 or LYK5, is required for enhanced resistance to B. cinerea and for priming of PR-1 expression when plants are pre-treated with elicitors 24 hr before inoculation indicate that LYK2 regulates, upon elicitation, a longlasting ability of the plant to prime some defence responses, beside PAD3 expression, independently of chitin perception. Interestingly, mutations in LYK2 reduce basal resistance to Pst DC3000, similar to what observed for cerk1-2, but, in contrast to the latter, also strongly impair flg22-induced resistance against this pathogen. This is in agreement with the hypothesis that LYK2 might positively regulate long-term defence responses downstream of recognition of different elicitors, though it cannot be ruled out that it might directly participate to PGN perception/signalling, as previously demonstrated for CERK1 and OsCERK1 (Ao et al., 2014;Buist, Steen, Kok, & Kuipers, 2008;Gust, Willmann, Desaki, Grabherr, & Nurnberger, 2012;Willmann et al., 2011). The PM-associated Ca 2+ -binding protein PCaP1 was recently shown to mediate OG-and flg22-induced resistance to B. cinerea and to be required for elicitor-triggered priming of defence gene expression during infection . Further investigation is however required to determine whether LYK2 and PCaP1 act in the same pathway.

Modulation of responses triggered by MAMPs and DAMPs is
crucial to ensure proper protection against invading pathogens without excessive cost in terms of growth. In this context, LYK2 appears to be a key regulator of priming of defence responses, ensuring that plants display enhanced resistance after pre-exposure to an elicitor.
Previous work suggests that LYK3, another Arabidopsis LYK, exerts an opposite role, negatively regulating defence responses, as lack of a functional protein causes constitutive expression of defence responses, including PAD3 expression, and increased resistance to B. cinerea (Paparella et al., 2014). The negative role of LYK3 in immunity might be mediated by the phytohormone abscisic acid (ABA), as this protein is required for both ABA-mediated repression of elicitor responses and for some physiological responses to this hormone (Paparella et al., 2014). Interestingly, loss of LYK3 also results in increased sensitivity to salt stress, suggesting that suppression of defence responses might contribute to properly counteract abiotic stresses (Paparella et al., 2014). Increasing evidence indeed suggests that CERK1 might regulate responses to abiotic stresses, as cerk1 mutants also show increased sensitivity to salt stress (Espinoza, Liang, & Stacey, 2017), and overexpression of a fungal chitinase causes increased tolerance to salt in a CERK1-dependent manner (Brotman et al., 2012). These observations lead to speculate that LYK proteins might have a more general function in balancing responses against different stresses beyond their role in MAMP perception. The possible function of LYK2 and other LYKs in modulating responses to different stresses therefore deserves future investigation.
The ability to mount stronger defence responses after a previous exposure to a MAMP or DAMPs might avoid overactivation of plant immunity, not only preventing growth-defence trade-offs that might be triggered by the numerous microorganisms present in the environment (Yu, Pieterse, Bakker, & Berendsen, 2019), but possibly favouring the interaction with beneficial microorganisms. Increasing evidence indicates that LYKs play multiple functions in plant immunity and symbiosis. For instance, OsCERK1 is required for mycorrhizal colonization (Miyata et al., 2014;, and its homologs in L. japonicus (LjLYS6) and M. truncatula (MtLYK9) also play a dual role in immunity and symbiosis (Bozsoki et al., 2017;Gibelin-Viala et al., 2019). Notably, rhizobial LCOs inhibit flg22 responses in Arabidopsis in a LYK3-dependent manner , and M. truncatula mutants for MtLYK9 are less colonized by the arbuscular mycorrhiza Rhizophagus irregularis but are more susceptible to the oomycete Aphanomyces euteiches (Gibelin-Viala et al., 2019). Future work will help determine if LYK2 is involved in the fine-tuning of plant responses to different microorganisms.
In conclusion, our results indicate that LYK2 has a limited role in chitin perception, but it is necessary to ensure a robust and durable resistance to pathogens after elicitor pre-treatments, priming activation of defence responses downstream of and/or independently of chitin perception. Future studies will help elucidate the molecular mechanism of action of LYK2, providing novel clues about how plants modulate immunity. This knowledge might be important not only to improve crop resistance to microbial diseases, but also to increase our understanding of the complexity of the interactions between plants and microbes in the environment.

ACKNOWLEDGMENTS
We thank Naoto Shibuya (Meiji University, Japan) for providing cerk1-2 seeds and Heini M. Grabherr (University of Tuebingen) for assistance with CERK1-GFP and LYK5-GFP cloning. We are grateful to Volker

CONFLICT OF INTEREST
The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT
All data generated or analysed during this study are included in this published article and its supplementary information files.