The phylogenetic tree was inferred by using the Maximum Likelihood method and JTT matrix-based model (Jones et al., 1992) with a bootstrap of 1000 replications. The tree with the highest log likelihood (-28734.46) is shown. The percentage of trees in which the associated taxa clustered together is shown next to the branches. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using a JTT model, and then selecting the topology with superior log likelihood value. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. This analysis involved 22 amino acid sequences. There were a total of 1230 positions in the final dataset. Phylogenetic analysis was conducted in MEGA X (Kumar et al., 2018).

RNAseq expression data on VvLYK genes were extracted from public data hosted on the NCBI website (https://www.ncbi.nlm.nih.gov/). VvLYK expression profiles of Vitis vinifera cv. Carignan leaf tissues following an infection with Erysiphe necator were extracted from the list of differentially expressed genes from Amrine et al. (2015) (BioProject Accession: PRJNA279229). VvLYK expression profiles of V. vinifera cv. Victoria leaf tissues following an infection with Coniella diplodiella were extracted from the study of Su et al. (2019) (BioProject Accession: PRJNA476839) using the GRape Expression ATlas (GREAT) application (https://great.colmar.inrae.fr/).

Arabidopsis thaliana wild-type (WT) Columbia (Col-0), atlyk4 mutant (WiscDsLox297300_01C), atlyk5 mutant (SALK_131911C, allele atlyk5-2), atlyk4/5 double mutant (WiscDsLox297300_01C x SALK_131911C; Cao et al., 2014) or transgenic lines atlyk4/5-p35S::VvLYK5-1/-2 were grown under a 10/14-h day/night cycle at 20/18°C. Transgenic lines were obtained by floral-dip transformation (Clough and Bent, 1998) of the atlyk4/5 double mutant line with coding sequences of VvLYK5-1 or VvLYK5-2, amplified from complementary DNA (cDNA) of Vitis vinifera cv. Marselan leaves and cloned in the pFAST_R02 overexpression vector (Shimada et al., 2010). Transgenic seeds were selected by their red fluorescence using the n°63-HE filter on a Zeiss Axio Zoom V16 microscope. Homozygous seeds of the T3 generation were genotyped before use in the experiments thereafter ( Figure S1A ).

V. vinifera cv. Marselan cells were cultivated, under continuous light, in liquid Nitsch-Nitsch medium (Nitsch and Nitsch, 1969) supplemented with 1 g/L casein hydrolysate, 400 µg/L 1-naphthaleneacetic acid and 40 µg/L 6-benzylaminopurine. They were maintained in suspension by continuous shaking (120 rpm at 24°C) and subcultured every 7 days using a 1:6 dilution in 120 mL of fresh liquid medium. For the experiments, grapevine cells were used 24h after diluting twice the 7-day-old culture in fresh medium.

Purified hexamer (DP6) of chitin (GLU436) and chitosan (GLU426) were provided by Elicityl (Crolles, France). These chitooligosaccharides were dissolved in sterile ultrapure water and used at a final concentration of 0.1 g/L in all experiments.

Leaves of 4 weeks old Arabidopsis plants were cut, pre-infiltrated with ultrapure water then equilibrated, abaxial face on ultrapure water, for 4h in a 6-well plate. They were then treated by substitution of water with elicitors or water (mock treatment) and harvested 10 min after. Proteins were extracted using a buffer containing 50 mM Hepes (pH 7.5), 5 mM EGTA (pH 8.1), 5 mM EDTA, 1 mM Na₃VO₄, 50 mM β-glycerophosphate, 10 mM NaF, 1 mM phenylmethanesulfonyl fluoride, 5 mM dithiothreitol, and 1X cOmplete™ EDTA-free Protease Inhibitor Cocktail (Roche). MAPKs activation was detected after immunoblotting of the extracted proteins (20 µg) using an anti-p42/44-phospho-ERK antibody (Cell Signaling). The revealing step was performed on an Amersham™ ImageQuant™ 800 (Cytiva) using ECL™ Prime as a western blotting detection reagent. Transfer quality and homogeneous loading were checked by Ponceau red staining. Three independent experiments were performed.

Arabidopsis thaliana leaves were treated in the same way as for assessing MAPK activation and harvested 1h post-treatment (hpt). After grinding, total RNAs were extracted using the SV Total RNA Isolation System with DNAse treatment (Promega). First-strand cDNA was synthesized from 1 μg of total RNA using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems). Real-time qPCR was performed in a ViiA™ 7 Real-Time PCR system (Applied Biosystems) with 10 ng cDNA and GoTaq^® qPCR Master Mix (Promega). Relative gene expression was assessed according to the Common Base Method (Ganger et al., 2017) taking into consideration the efficiency (E) of each reaction calculated by the LinRegPCR quantitative PCR data analysis program (Ruijter et al., 2009). For each gene of interest (GOI), the mean of resulting technical duplicate data (efficiency-weighted C q ( w ) values = C _q ×log(E) ) was normalized by the mean C q ( w ) data of AtRHIP1 (AT4G26410) and AtPTB1 (AT3G01150) housekeeping genes (HKG) ( Δ C q ( w ) =   C q ;   G O I ( w ) − C q ;   H K G 1 ( w ) +   C q ;   H K G 2 ( w ) 2 ) before being normalized to the water control treatment (Fold-change = 10 − Δ Δ C q ( w ) , with Δ Δ C q ( w ) =   Δ C q ;   C O S ( w ) − Δ C q ;   M o c k ( w ) ). All primers used are listed in Table S1 . Four independent experiments were conducted.

For grapevine cells, suspensions were divided into 100 mL flasks then equilibrated for 1.5h on a rotary shaker with the same conditions as for their culture. They were next treated with chitin (final concentration of 0.1 g/L in a final volume of 20 mL) or water (as control) and harvested at 1h, 3h, 6h, and 24h post-treatment (hpt). The qPCR protocol and calculation method were the same as described above except that VvVATP16 (Vitvi03g04022), VvRPL18B (Vitvi05g00033) and VvVPS54 (Vitvi10g01135) were used as housekeeping genes. Five independent experiments were conducted.

Four-week-old Arabidopsis plants were sprayed on both sides with a solution of chitin or water as control. Four days post-treatment, one leaf from each of the two plants used for each modality was harvested and bleached in pure ethanol. They were then cleared overnight in 1 g/L chloral hydrate and washed three times with 0.1 M phosphate buffer (pH 8.0) before being stained in 0.1 M phosphate buffer (pH 8.0) containing 0.01% (w/v) aniline blue for a day. After staining, leaves were mounted on slides in 80% (v/v) glycerol. Callose deposition was observed on the adaxial side by epifluorescence microscopy under UV (λ_exc = 340-380 nm, λ_em = 425 nm, long pass filter, magnification x100, Leica DMRB). Ten representative images of the callose deposits were acquired for each condition using the Nis Elements BR software (Nikon) with the DS-5Mc-U1 digital photomicrographic camera (Nikon) equipped on the microscope. Image analysis was then performed on the Fiji application (Schindelin et al., 2012) as described in Mason et al. (2020). Three independent experiments were conducted.

Erysiphe necator was grown on detached leaves of V. vinifera cv. Marselan previously disinfected in a solution of 2% (v/v) sodium hypochlorite, maintained on sterile agar plates and subcultured every two weeks. Four-week-old A. thaliana plants were used to assess powdery mildew penetration efficiency. Two leaves per plant were infected with E. necator spores using a fine paintbrush. Detached leaves from two plants per line were sampled 48h post-inoculation (hpi) and stained with trypan blue according to Koch and Slusarenko (1990). Fungal structures were visualized using a Leica (Wetzlar, Germany) DME light microscope (magnification x400). A minimum of 100 germinated spores were scored for each treatment. Successful penetration of epidermal cells (% of penetrated cells) was indicated by the presence of a haustorium within the cell or the development of a secondary hyphae from the appressorium as described in Figure S2 . Three independent experiments were performed with the same results.

Leaves of Nicotiana benthamiana were transformed by agroinfiltration (Norkunas et al., 2018) with the Agrobacterium tumefaciens strain GV3101 containing an overexpression cassette for the expression of VvLYK5-1_YFP or VvLYK5-2_YFP (see below), MtNFP_mCherry (coding for a protein previously described to be localized at the PM; Lefebvre et al., 2012), and the RNA silencing suppressor p19 (Voinnet et al., 2003), with OD600 of each strain adjusted to 0.5, 0.5 and 0.2 respectively. After 72 hours, the subcellular localization of fusion proteins was studied and 3 leaf discs (7 mm of diameter) were harvested for immunoblot analysis. Subcellular localization was analyzed with a confocal laser scanning microscope (SP8, Leica) using a x25 water immersion objective lens. Leaf discs were ground using a Restch MM400 mixer mill. Total proteins were extracted in 2X Laemmli buffer and boiled at 95°C for 5 minutes and separated on SDS-PAGE, and then transferred to a nitrocellulose membrane. Immunodetection of the fusion proteins ( Figure S3 ) was performed using the rabbit anti-GFP (Sigma-Aldrich) and anti-mCherry polyclonal antibodies (Lefebvre et al., 2012), and polyclonal goat anti-rabbit antibodies fused to HRP (Millipore).

Leaves of Nicotiana tabacum cv. Xanthi were transformed by agroinfiltration (Norkunas et al., 2018) with the Agrobacterium tumefaciens strain GV3101 containing an overexpression cassette for VvLYK1-1, VvLYK1-2, VvLYK5-1, or VvLYK5-2 fused with either a C-terminal Cyan Fluorescent Protein (C_ter-CFP in the pH7CWG2 plasmid, Karimi et al., 2005) or a C-terminal Yellow Fluorescent Protein (C_ter-YFP in the pH7YWG2 plasmid, Karimi et al., 2005). Five days after (co-)infiltration, the abaxial side of N. tabacum leaves transiently expressing the different constructs was infiltrated with chitin or water (mock treatment). Observations were performed 30 min post-treatments (according to Cheval et al., 2020) using a Nikon A1-MP multiphoton microscope with an Apo IR x60 objective (NA: 1.27, water immersion, Nikon). Fluorescence lifetime imaging (FLIM) images were collected using a time-correlated single-photon counting (TCSPC) module (Picoquant). CFP excitation (820 nm with two-photon excitation) was provided by an IR laser (Chameleon, Coherent) delivering femtosecond pulses at a repetition rate of 80 MHz. Its resulting fluorescence emission was collected with a single photon avalanche diode (SPAD), using a band-pass emission filter FF01-494/20 (Semrock). TCSPC lifetime recording was performed over 200 temporal channels (final resolution: 0.64 ps). The FLIM analysis was performed on regions of interest (ROIs) drawn on the plasma membrane using the SymPhoTime (PicoQuant) software. Fluorescence lifetime values were calculated by fitting the tail of the CFP fluorescence decay with a bi-exponential model. Among the two generated lifetime constants (τ₁ and τ₂), only τ₁ was reported, as it was the most sensitive to the Förster resonance energy transfer (FRET) (Bègue et al., 2019; Rosnoblet et al., 2021). The efficiency of the energy transfer was given by the following equation: E ( % ) = ( 1 − F D A F D ) × 100 , with FDA and FD representing the relative fluorescence lifetime of the donor in the presence or the absence of the acceptor, respectively. Three independent experiments were conducted.

In contrast to A. thaliana that possesses 5 LysM-RLKs (Wan et al., 2008), the most recent annotation of the grapevine genome predicts 16 VvLYK genes encoding 16 LysM-RLKs. To narrow down the number of candidates for COS perception, we performed a phylogenetic analysis of this multigene family in these two species. An actualized maximum-likelihood phylogenetic tree was constructed in order to highlight protein sequence similarities between grapevine and Arabidopsis LYKs ( Figure 1A ). Two sequences, named VvLYK5-1 and VvLYK5-2 according to the previous naming convention proposed by Brulé et al. (2019), were found unambiguously located in the same clade as the major Arabidopsis chitin receptor AtLYK5 (Cao et al., 2014) since they were clustered together in 100% of the trees constructed during the analysis ( Figure 1A ).

Expression profiles following inoculation with fungal pathogens of each predicted VvLYK gene were then analyzed using publicly available RNAseq data (Amrine et al., 2015; Su et al., 2019) to identify genes that are transcriptionally regulated during the plant defense reaction against these chitin-containing organisms. Figures 1B, C indicate that VvLYK5-1 was up-regulated in response to infection by both E. necator and Coniella diplodiella (grapevine white rot). Interestingly, neither VvLYK5-2 nor the previously identified genes VvLYK1-1 and VvLYK1-2 encoding COS receptors (Brulé et al., 2019) appeared to be differentially expressed in grapevine leaves in response to these two pathogens. To confirm publicly available transcriptomic data, grapevine cells were treated with purified chitin hexamer (DP6, thereafter referred as chitin treatment). Our results confirmed that only VvLYK5-1 transcripts were significantly upregulated 1h post-treatment ( Figure 1D ).

Further sequence analysis of VvLYK5-1 and VvLYK5-2 was performed after cloning and sequencing their respective cDNAs from leaves of V. vinifera cv. Marselan. In silico analysis revealed the presence of a signal peptide, three extracellular LysM domains, a transmembrane domain, and a cytoplasmic serine/threonine kinase domain ( Figure 2A , Figure S4 ). As expected, VvLYK5-1_YFP was detected at the PM ( Figure 2B ) as it co-localized with the PM marker MtNFP_mCherry (Lefebvre et al., 2012). VvLYK5-2_YFP mainly accumulated at PM and also appeared partly in the endoplasmic reticulum ( Figure 2B ). Moreover, the analysis of the predicted amino acid sequences of VvLYK5-1 and VvLYK5-2 ( Figure S4 ) indicated that most of the amino acids known to be important for the kinase activity (Hanks et al., 1988; Johnson et al., 1996; Petutschnig et al., 2010; Klaus-Heisen et al., 2011; Suzuki et al., 2016) were absent, as in AtLYK5 (Cao et al., 2014). Thus, in the Gly-rich loop, three out of the four glycines present in AtLYK1/VvLYK1-1 and MtLYK3 were absent in VvLYK5-1/-2, as in AtLYK5. Similarly, in the VAIKK motif, the ATP binding site K³⁵⁰ of AtLYK1 (X⁴⁵¹ of the consensus sequence) which is mandatory for the kinase activity (Petutschnig et al., 2010), was mutated in AtLYK5 and VvLYK5-1/-2. In the catalytic loop, the amino acids RD found in AtLYK1, VvLYK1-1 and MtLYK3 were replaced by the KN residues in AtLYK5, VvLYK5-1 and VvLYK5-2. This difference is crucial knowing that a single mutation of D⁴⁴¹ in MtLYK3 (X⁵⁴⁵ on the consensus sequence) leads to a loss of its in vitro kinase activity (Klaus-Heisen et al., 2011). In the activation segment, the aspartic acid of the DFG motif, that is well conserved across kinases (Hanks et al., 1988), is also mutated in AtLYK5, VvLYK5-1 and VvLYK5-2. Finally, the phosphorylation site T⁵⁷³ of AtLYK1 (X⁶⁸⁶ of the consensus sequence), required for the activity of the AtLYK1 kinase domain (Suzuki et al., 2016) and also conserved in MtLYK3 (Klaus-Heisen et al., 2011), was absent in the two VvLYK5-1/-2 protein sequences. When considered together, all of these differences suggest that VvLYK5-1 and VvLYK5-2 do not possess a functional kinase domain.

In order to compare the two related elicitors based on their degree of acetylation (DA), we chose to use COS with the same degree of polymerization (DP). Since COS with a low DP (i.e. from 6 to 8 glucosamine (GlcN) or N-acetyl-glucosamine (GlcNAc) residues) are known to induce the most effective defense reactions in different plant species, including grapevine and Arabidopsis (Aziz et al., 2006; Miya et al., 2007; Petutschnig et al., 2010; Brulé et al., 2019), we also used COS with this DP in this study.

To find out whether VvLYK5-1 or VvLYK5-2 were able to trigger immune responses in response to COS, we overexpressed their encoding genes in the atlyk4/5 double mutant, known to be insensitive to the chitin oligomers (Cao et al., 2014). Three independent lines of each construct overexpressing VvLYK5-1 or VvLYK5-2 ( Figures S1B, C ) were first evaluated on their ability to induce MAPKs activation in response to a chitin oligomer treatment. As expected, chitin induced the phosphorylation of two MAPKs, MPK3 and MPK6 (at 43 and 47 kDa, respectively) in WT Col-0 plants (Cao et al., 2014) but not in the atlyk4/5 double mutant ( Figure 3A ). This chitin-induced MAPKs activation was restored in the three transgenic atlyk4/5-p35S::VvLYK5-1 lines. Weaker response in line #7.8 was correlated with lower transgene expression ( Figure S1B ). In contrast, MAPKs activation was not restored in transgenic atlyk4/5-p35S::VvLYK5-2 lines ( Figure 3A ). To note, a weak chitin-induced activation of MPK3 in atlyk4/5-p35S::VvLYK5-2 lines could be detected in only one out of three experiments and was therefore not considered. Next, the expression of the defense gene encoding flagellin-induced receptor kinase 1 (FRK1), which has been previously shown to be a chitin-responsive gene (Brulé et al., 2019), was monitored. Figure 3B shows that AtFRK1 gene expression was transcriptionally upregulated upon chitin treatment in WT Col-0 plants but not in the atlyk4/5 line. Overexpression of VvLYK5-1 in the atlyk4/5 mutant line restored AtFRK1 expression in each of the three evaluated transgenic lines ( Figure 3B ). In contrast, overexpression of its paralogous gene, VvLYK5-2, resulted in a weak chitin-triggered AtFRK1 expression in only one of the three evaluated lines ( Figure 3B ).

The atlyk4/5 double mutant was also unable to elicit defense responses after chitosan hexamer treatment ( Figure S5 ). Interestingly, in transgenic lines overexpressing VvLYK5-1 or VvLYK5-2, chitosan treatment did not appear to clearly restore the MAPKs activation compared to WT ( Figure S6A ), and the level of AtFRK1 transcripts was not statistically different compared to water treatment ( Figure S6B ). In sum, these results suggest that VvLYK5-1 can complement chitin-triggered immune responses in the atlyk4/5 double mutant line, but probably not those triggered by chitosan. Moreover, VvLYK5-2 does not seem to play a major role in grapevine for the perception of chitin or chitosan oligomers.

As chitin is known to induce callose deposition in A. thaliana leaves (Giovannoni et al., 2021), we investigated whether the expression of VvLYK5-1 could also complement this late defense response event. Figure 4 indicates that the atlyk4/5 double mutant did not accumulate callose deposits in response to chitin treatment. Indeed, the number of callose deposits, close to 200 per mm² in the WT in response to chitin treatment, decreased to a level similar to a mock treatment in the atlyk4/5 double mutant. This is in agreement with previous reports showing that atlyk4 and alyk5 mutants do not trigger any chitin-mediated callose deposition in plasmodesmata (Cheval et al., 2020). By evaluating our transgenic lines, we could observe that VvLYK5-1 overexpression was sufficient to restore the chitin-induced callose deposition in the atlyk4/5 double mutant ( Figure 4 ). Levels of callose deposition similar to the WT were detected in all the three independent atlyk4/5-p35S::VvLYK5-1 lines with a number of deposits per mm² ranging from 100 to 300 depending on the transgenic line and not in the three independent atlyk4/5-p35S::VvLYK5-2 lines ( Figure 4 ). Together, these data indicate that only the expression of VvLYK5-1 can restore this chitin-induced late defense response in the atlyk4/5 double mutant.

We have previously demonstrated that VvLYK1-1 plays an important role in host resistance to penetration by a non-adapted powdery mildew species (Brulé et al., 2019). We therefore next investigated whether also VvLYK5-1 or VvLYK5-2 are required for host resistance to powdery mildew penetration. First, we evaluated if the atlyk4/5 double mutant line was more sensitive to powdery mildew than the WT, as it was previously reported for atcerk1/lyk1 (Brulé et al., 2019). Our results indicated that the atlyk4/5 double mutant was significantly more sensitive to the non-adapted fungus E. necator than the WT Col-0, with a percentage of epidermal penetrated cells increasing from 45% in the WT to 71% in the double mutant ( Figure 5 ). However, penetration resistance was restored back to WT levels in atlyk4/5 mutant lines overexpressing VvLYK5-1. This observation suggests that VvLYK5-1 might be also important for grapevine defense against powdery mildew. Remarkably, this restored resistance to the Arabidopsis non-adapted pathogen was also found in the three independent transgenic lines atlyk4/5-p35S::VvLYK5-2 ( Figure 5 ).

Browsing through the STRING database (Szklarczyk et al., 2019) to identify putative VvLYK5-1 and VvLYK5-2 interacting partners, we found that both were predicted to interact with VvLYK1-1, VvLYK1-2, and VvLYK1-3 according to reported interactions from orthologous proteins in other species ( Figure S7 ). Based on the findings that (a) AtLYK1 is known to directly interact with AtLYK5 for chitin binding (Cao et al., 2014), (b) VvLYK5-1 is probably lacking a functional kinase domain ( Figure S3 ), and (c) VvLYK5-2 (this study) and VvLYK1-3 (Brulé et al., 2019) are probably not involved in COS-triggered immunity, we hypothesized that putative interactions might take place between VvLYK5-1 and VvLYK1-1 or VvLYK1-2 for COS perception.

According to basal expression data obtained from Grape eFP Browser (Fasoli et al., 2012), it is unlikely that VvLYK5-1 and VvLYK1-2 can interact as VvLYK5-1 is globally well expressed in all tissues/organs whereas VvLYK1-2 is not (e.g. in adult leaves) or weakly expressed except in wooden stems and flowers ( Figure S8 ). In contrast, VvLYK1-1 and VvLYK5-1 show high co-expression levels in the whole plant ( Figure S8 ) which is consistent with a putative function of VvLYK1-1 as the main co-receptor interacting with VvLYK5-1.

This hypothesis was supported by a FRET-FLIM protein-protein interaction experiment which did not reveal any significant interaction between VvLYK1-2_CFP and VvLYK5-1-YFP before or after chitin treatment ( Figure S9 ). FRET-FLIM measurements performed for VvLYK1-1_CFP and VvLYK5-1_YFP did not reveal any interaction between these receptors at the basal state ( Figure 6 ). However, 30 min after chitin treatment, a significant decrease of CFP fluorescence lifetime (τ₁) was observed when VvLYK1-1_CFP was co-transformed with VvLYK5-1_YFP, with a recorded mean τ₁ decreasing from 2.13 ns for the donor when expressed alone to 1.88 ns when co-expressed with the VvLYK5-1_YFP acceptor (E = 11.70%; p-value< 0.001, Figure 6 ). To note, no interaction between VvLYK1-2_CFP or VvLYK1-1_CFP and VvLYK5-2_YFP before or after chitin treatment could be demonstrated ( Figure S9 , Figure 6 ). Altogether, these results reveal that only VvLYK1-1_CFP and VvLYK5-1_YFP can physically interact upon chitin perception.