Interaction of Medicago truncatula Lysin Motif Receptor-Like Kinases, NFP and LYK3, Produced in Nicotiana benthamiana Induces Defence-Like Responses

Receptor(-like) kinases with Lysin Motif (LysM) domains in their extracellular region play crucial roles during plant interactions with microorganisms; e.g. Arabidopsis thaliana CERK1 activates innate immunity upon perception of fungal chitin/chitooligosaccharides, whereas Medicago truncatula NFP and LYK3 mediate signalling upon perception of bacterial lipo-chitooligosaccharides, termed Nod factors, during the establishment of mutualism with nitrogen-fixing rhizobia. However, little is still known about the exact activation and signalling mechanisms of MtNFP and MtLYK3. We aimed at investigating putative molecular interactions of MtNFP and MtLYK3 produced in Nicotiana benthamiana. Surprisingly, heterologous co-production of these proteins resulted in an induction of defence-like responses, which included defence-related gene expression, accumulation of phenolic compounds, and cell death. Similar defence-like responses were observed upon production of AtCERK1 in N. benthamiana leaves. Production of either MtNFP or MtLYK3 alone or their co-production with other unrelated receptor(-like) kinases did not induce cell death in N. benthamiana, indicating that a functional interaction between these LysM receptor-like kinases is required for triggering this response. Importantly, structure-function studies revealed that the MtNFP intracellular region, specific features of the MtLYK3 intracellular region (including several putative phosphorylation sites), and MtLYK3 and AtCERK1 kinase activity were indispensable for cell death induction, thereby mimicking the structural requirements of nodulation or chitin-induced signalling. The observed similarity of N. benthamiana response to MtNFP and MtLYK3 co-production and AtCERK1 production suggests the existence of parallels between Nod factor-induced and chitin-induced signalling mediated by the respective LysM receptor(-like) kinases. Notably, the conserved structural requirements for MtNFP and MtLYK3 biological activity in M. truncatula (nodulation) and in N. benthamiana (cell death induction) indicates the relevance of the latter system for studies on these, and potentially other symbiotic LysM receptor-like kinases.

All four genes encode receptor-like kinases (RLKs) with an extracellular region (ExR) predicted to contain three LysM domains, a transmembrane helix, and a protein kinase domain (KD) within the intracellular region (InR) [5,[20][21][22]. Remarkably, in contrast to MtLYK3 and LjNFR1, which both display kinase activity, MtNFP and LjNFR5 seem to function as pseudokinases that neither show nor rely on the intrinsic kinase activity to signal [9,[20][21][22]. LjNFR5 is hypothesized to form a receptor complex with LjNFR1: a notion consistent with their demonstrated cofunctioning during the determination of RL specificity [23]. Similarly, a receptor complex composed of MtNFP and a yetunidentified LysM-RLK or MtLYK3 is predicted to initiate the pre-infection responses and the infection process, respectively [12,15]. Since mutagenesis studies in Medicago have not identified alterations in genes other than MtNFP that lead to complete lack of responsiveness to NFs, a function of this additional LysM-RLK in the pre-infection stage is most likely redundant. In addition, MtNFP has been implicated in Medicago interactions with pathogens (Aphanomyces euteiches and Colletotrichum trifolii), and with beneficial arbuscular mycorrhiza (AM) fungi [24][25][26][27]. However, it remains to be shown whether MtNFP functions in these processes alone or in co-operation with (an)other RLK(s).
We are interested in NF-induced signalling mediated by MtNFP and MtLYK3, focusing on their postulated interaction in situ. However, our attempts to visualize these proteins in Medicago root have been unsuccessful, presumably due to stringent regulation of their accumulation (even in the situation of an attempted overproduction). Nicotiana benthamiana (Nicotiana) has proved to be a useful model for heterologous production and structure-function studies on multiple proteins, providing invaluable insights that guided their subsequent analyses in the respective homologous systems [39]. Therefore, we employed an Agrobacterium tumefaciens (Agrobacterium)-mediated transient transformation of Nicotiana leaves [40], which allowed us to produce both proteins to levels suitable for fluorescence microscopy. Remarkably, we found that heterologous co-production of MtNFP and MtLYK3 resulted in the induction of defence-like responses that are typically observed upon treatment with pathogen-derived molecules. As the apparent (functional) interaction of these LysM-RLKs in Nicotiana activated defence-like responses, similar to these mediated by AtCERK1, our results indicate the existence of parallels between NF-induced and COs-induced signalling.

Constructs for Plant Expression
The cDNA of MtNFP, MtLYK3, and MtDMI2 (in the latter case the first intron was included in the cDNA), and the genomic sequence of AtCERK1 were PCR amplified and cloned in a pMON999 vector containing a CaMV 35Sp:: (sYFP2, mCherry, 3xFLAG or empty) 35S terminator cassette. The stop codon was removed from the coding sequences during cloning (except when generating untagged MtNFP and MtLYK3 constructs) to allow a translational fusion. Sequences of the primers and linkers are given in Table S1. All point mutations were introduced as described [20]. All constructs were sequenced to verify the correct insert sequence. Constructs generated in pMON999 vector were subsequently recloned into a pBin+ (all MtNFP constructs) or pCambia1390 (all MtLYK3, AtCERK1, and MtDMI2 constructs) vector using HindIII and SmaI sites. MtNFP[DInR], MtNFP-YFP N , MtLYK3-YFP N , MtLRRII.1-YFP C , AtBRI1-YFP C (where YFP N or YFP C encode, respectively, the N-and C-terminal part of split YFP used in BiFluorescence Complementation assay) constructs are described [22,41].

Nicotiana Transformations
Agrobacterium tumefaciens GV3101::pMP90 and LBA4404 strains were transformed with the respective constructs via electroporation. The LBA4404 strain was used only in the experiments that compared the effect of different Agrobacterium strains on cell death (CD) induction upon MtNFP and MtLYK3 co-production. All results presented in Figures 1-5, Figure S1 and S3, and Tables 1-3 were obtained with Agrobacterium GV3101::pMP90 strain. Agrobacterium-mediated transformation of Nicotiana was performed essentially as described [42], except that Agrobacterium cultures were grown in LB medium supplemented with 25 mg/mL of rifampicin and 50 mg/mL of kanamycin. Resuspended cells were incubated at room temperature for at least 1 h before being infiltrated into fully expanded leaves of green house-grown plants using needleless syringes. Agrobacterium transformants carrying the respective construct were resuspended in the infiltration medium to desired OD 600 : all MtNFP and MtNFP[DInR]-sYFP2 constructs -OD 600 = 0.4; all MtLYK3 and AtCERK1 constructs -OD 600 = 0.7; MtDMI2-sYFP2 -OD 600 = 1.0. Then, they were mixed 1:1 with GV3101::pMP90 transformants carrying: pCambia1390 vector with an empty CaMV 35Sp::35S terminator cassette (for separate expression), a desired MtNFP construct or a desired MtLYK3 construct before being infiltrated into Nicotiana leaves. All experiments included mock infiltration with GV3101::pMP90 transformants carrying pCambia1390 vector with an empty CaMV 35Sp::35S terminator cassette, and a positive control (coexpression of WT MtNFP-FP and WT MtLYK3-FP constructs). Cell death induction upon separate expression or co-expression of each (pair of) constructs was analyzed between 24 and 72 hai in at least three independent experiments, every time using three different plants. In case of no macroscopic symptoms, three leaves were stained with Evans blue to confirm the lack of CD.
To confirm efficient accumulation of MtLRRII.1-YFP C and AtBRI1-YFP C fusions in Nicotiana, Agrobacterium transformants carrying the respective constructs were co-infiltrated at high optical densities (final OD 600 = 0.5) with Agrobacterium transformants carrying either MtNFP-YFP N or MtLYK3-YFP N constructs (final OD 600 = 0.5). The observed complementation of YFP fluorescence reported on efficient accumulation, and even unspecific oligomerization of the respective encoded fusions.

Stereoscopic Analysis
Blue light-excitable autofluorescence and far-red chlorophyll autofluorescence in intact Nicotiana leaves were imaged using 430/ 40 excitation and 485/50 emission BP filters, or 480/40 BP excitation and 510 LP emission filters, respectively. Images were captured using CMOS USB DCC1645C camera (THORLabs, Newton NJ, USA) implemented on a Leica MZ FLIII stereoscope. Evans blue staining was performed as described [42]. Leaves were cleared by boiling in acidic lactophenol/ethanol solution (10 g phenol in 10 ml lactic acid, mixed 2:1 with 96% ethanol) until the complete removal of chlorophyll (approximately 3 min per leaf). Ethanol-inextractable autofluorescence was excited with 312 nm wavelength. Images were captured using a Cool Snap CF camera (Photometrix, Tucson AZ, USA).

qRT-PCR Analysis
RNA extraction and qRT-PCR were performed as described [17] except that cDNA was prepared from 500 ng of total RNA (see Table S1 for primer sequences). Two technical replicates from two biological replicates were analyzed and results were collated.

Co-production of MtNFP and MtLYK3 in Nicotiana Leaves Induces Cell Death
MtNFP and MtLYK3 cDNA sequences were fused at their 39 ends to the sequence encoding a fluorescent protein (FP); either a super yellow fluorescent protein 2 (sYFP2) or mCherry [43][44], and were expressed from a CaMV 35S promoter in Nicotiana leaves, where they were delivered by Agrobacterium-mediated transformation. These and similar MtNFP and MtLYK3 constructs were shown to complement Mtnfp and Mtlyk3 mutants, respectively [17][18]22], and are therefore suitable for studying the encoded LysM-RLKs. Confocal laser-scanning microscopy analysis demonstrated co-localization of both MtNFP-sYFP2 and MtLYK3-sYFP2 fusions with a plasma membrane (PM) marker (the hypervariable region [HVR] of maize [Zea mays] ROP7 fused to the Cterminus of mCherry; [20,22]), hence indicating PM localization of MtNFP and MtLYK3 fusions in Nicotiana leaf epidermal cells. Surprisingly, co-infiltration of Agrobacterium transformants carrying MtNFP-FP or MtLYK3-FP constructs, leading to the co-production of the encoded fusions, resulted in collapse and subsequent desiccation of the infiltrated region within 48 hours after infiltration (hai) (Fig. 1A), regardless of the Agrobacterium strain used (i.e. GV3101::pMP90 or LBA4404; unpublished data). This cell death (CD) response was not dependent on the tag attached to either protein, since an identical response was observed upon coproduction of FP-tagged, 3xFLAG-tagged or untagged MtNFP and MtLYK3 (Fig. 1A, Table 1). Importantly, production of the separate MtNFP or MtLYK3 (fusions) did not induce CD as confirmed with an exclusion dye (Evans blue) staining (Fig. 1A), which reflects compromised membrane permeability attributed with cell death.
To investigate whether a similar CD response could be triggered by heterologous production of other plant RLKs, we analysed the Nicotiana response to expression of Medicago Doesn't Make Infection 2 (DMI2; [45]), MtLRRII.1 [41]), and Arabidopsis Brassinosteroid Insensitive 1 (BRI1; [46]), all driven by the CaMV 35S promoter. Notably, none of these RLKs, alone or in combination with either MtNFP or MtLYK3 fusions, induced CD (Fig. 1B, Table 1), despite being efficiently produced in Nicotiana leaves, as confirmed with fluorescence microscopy (see Materials & Methods). Thus, the Nicotiana CD response was not a general response to a heterologous production of RLKs but rather a specific response to MtNFP and MtLYK3 co-production.

Production of AtCERK1 also Induces Cell Death in Nicotiana Leaves
A rapid tissue collapse at the site of pathogen attack, termed the hypersensitive response (HR), is frequently observed in incompatible plant-pathogen interactions where it is thought to contribute to pathogen restriction and to generate a signal that activates systemic plant defence mechanisms [47][48]. The apparent phenotypic similarity of the CD response to MtNFP and MtLYK3 co-production with the HR elicited by various pathogen-derived components (MAMPs and so-called effectors) [49][50], prompted us to investigate whether co-production of the symbiotic LysM-RLKs might activate defence signalling similar to that mediated by LysM-RLKs functioning in innate immunity. AtCERK1 mediates signalling upon the perception of COs or PGN, although, to our knowledge, CD induction in response to these MAMPs has not been reported so far in any plant species. We therefore investigated the Nicotiana response to heterologous production of wild-type (WT) AtCERK1 or its kinase-inactive variant carrying a substitution in the catalytic lysine (Lys 349 in a kinase subdomain II). Both AtCERK1 and AtCERK1[K349E] constructs were generated as fusions to the 59 end of the sYFP2 sequence, and their expression in Nicotiana leaves was driven by the CaMV 35S promoter.
Notably, heterologous production of AtCERK1-sYFP2 fusion resulted in tissue collapse and desiccation of (nearly) the entire infiltrated region in 20 out of 22 infiltrations 36 hai ( Fig. 2A). This CD induction abolished our attempts of precisely characterizing the subcellular localization of AtCERK1 fusion in Nicotiana leaf epidermal cells, although we could detect sYFP2 fluorescence at the cell boundary (unpublished data). On the contrary, we observed clear co-localization of AtCERK1[K349E]-sYFP2 fusion with the PM marker using confocal laser-scanning microscopy analysis (Fig. S1). The PM localization of AtCERK1[K349E] fusion in Nicotiana leaf epidermal cells is in agreement with the reported subcellular localization of AtCERK1 fluorescent fusion in onion (Allium cepa) epidermal cells [31]. Importantly, production of the kinase-inactive variant of AtCERK1 did not result in CD induction, as confirmed with Evans blue staining (Fig. 2B), indicating that biological activity of AtCERK1 in Nicotiana leaves was dependent on its kinase activity.

Cell Death Induction Upon MtNFP and MtLYK3 Coproduction, and AtCERK1 Production in Nicotiana Leaves Requires an Influx of Extracellular Ca 2+
An influx of extracellular Ca 2+ causes an increase in the cytosolic [Ca 2+ ] that is required for MAMP (including COs)induced activation of a MAPK cascade, ROS production, and gene expression. Thus, Ca 2+ influx is postulated to occur very early in the plant defence signalling pathway [51][52], possibly immediately upon the activation of the PM-localised MAMP receptors [53] We wanted to know whether an influx of extracellular Ca 2+ was similarly involved in CD induction upon MtNFP and MtLYK3 co-production or separate production of AtCERK1. To this end, MtNFP-3xFLAG and MtLYK3-3xFLAG fusions or AtCERK1-3xFLAG fusion were (co-)produced in adjacent regions in Nicotiana leaves. Twelve hours later, parts of the infiltrated leaf regions were syringe-infiltrated with 5 mM lanthanum chloride (an established inhibitor of the PM-localized calcium channels) or water, and the CD development was monitored between 24 and 72 hours after the first infiltration (with Agrobacterium). Notably, in 24 out of 30 leaf regions coproducing MtNFP and MtLYK3 fusions, compromised membrane permeability and tissue collapse were first (i.e. between 36 and 42 hai) localized only (or mostly) outside the lanthanum chloride-treated regions (Fig. 3A). Later on (i.e. 60 hai), 26 out of 30 parts of leaf regions treated with lanthanum chloride showed confluent death of the entire infiltrated region (unpublished data). Similar delay of the CD development was observed 33 hai in 11 out of 21 leaf regions producing AtCERK1 fusion and treated with lanthanum chloride (Fig. 3C). On the contrary, control treatment with water did not affect the development of confluent CD upon (co-)production of MtNFP and MtLYK3 fusions or AtCERK1 fusion (Fig. 3B, D).

Cell Death Upon MtNFP and MtLYK3 Co-production, and AtCERK1 Production in Nicotiana Leaves is Associated with an Induction of Defence-like Responses
Subsequently, we investigated whether co-production of MtNFP and MtLYK3 or production of AtCERK1 in Nicotiana leaves was associated with an accumulation of phenolic compounds and/or induction of defence-related gene expression, two established hallmarks of plant defence response, including that induced by COs and/or PGN [54][55][56]. We started by analysing the kinetics of CD development. To this end, Agrobacterium transformants carrying MtNFP-3xFLAG, MtLYK3-3xFLAG or AtCERK1-3xFLAG construct were (co-) infiltrated at different time-points in adjacent circles in Nicotiana leaves, and CD development was monitored between 24 and 48 hai. In case of co-production of MtNFP and MtLYK3 fusions, macroscopic symptoms of CD were first observed around  MtLRRII.1-YFP C 0/9 0/9*** 0/9*** AtBRI1-YFP C 0/7 0/9*** 0/9*** # 2 unless stated differently: with -3xFLAG (*) untagged (**), or 2YFP N (***) tagged construct. Indicated constructs were expressed alone or co-expressed with either MtNFP or MtLYK3 in Nicotiana leaves, and the infiltrated regions were marked. Macroscopic symptoms of cell death were scored 48 hai: only infiltrations that resulted in confluent death of (nearly) the entire infiltrated region were scored and are presented as a fraction of total infiltrations performed. doi:10.1371/journal.pone.0065055.t001 36 hai (Fig. 4A) as a type of flaccidity and the appearance of small patches of collapsed tissue (these were more pronounced on the abaxial side of the leaf). Forty-eight hai, 30 out of 31 infiltrations showed pronounced tissue desiccation of the entire infiltrated region (Fig. 1A). Compromised membrane permeability preceded tissue collapse and often occurred over the entire infiltrated region approximately 33 hai (Fig. 4A). Compromised membrane permeability of leaf regions producing AtCERK1 fusion was observed already approximately 27-30 hai, and pronounced macroscopic symptoms of CD developed 36 hai ( Fig. 2A, Fig. 3 C, D). In addition, co-production of MtNFP-3xFLAG and MtLYK3-3xFLAG fusions resulted in accumulation of blue light-excitable autofluorescence (Fig. 4B) approximately 36 hai. This was not observed upon separate production of either fusion, or upon coproduction of MtNFP-3xFLAG and kinase-inactive MtLYK3[G334E]-3xFLAG fusions (unpublished data). Accumulation of ethanol/lactophenol-inextractable and UV-excitable autofluorescence, indicative of phenolic compounds, was detected approximately 36 hai and 30 hai in leaf regions (co-)producing MtNFP-3xFLAG and MtLYK3-3xFLAG fusions or AtCERK1-3xFLAG fusion, respectively (Fig. 4C). Mock infiltration, separate production of MtNFP-3xFLAG or MtLYK3-3xFLAG fusion, production of kinase-inactive AtCERK1[K349E]-3xFLAG fusion or co-production of MtNFP-3xFLAG and kinase-inactive MtLYK3[G334E]-3xFLAG fusions did not result in the accumulation of similar autofluorescence (Fig. 4C).
Taken together, the indication of a localized accumulation of phenolic compounds and induction of defence-related gene expression suggested that the Nicotiana response to MtNFP and MtLYK3 co-production triggered responses that were qualitatively similar to the responses to heterologous production of AtCERK1. In addition, the fact that in both cases the impairment of Ca 2+ influx delayed the CD development suggested that the apparent functional interaction of two symbiotic LysM-RLKs in Nicotiana leaves mimics the action of the MAMP receptor, AtCERK1, and triggers defence-like responses.

Cell Death Induced in Nicotiana Leaves Upon MtNFP and MtLYK3 Co-production is a NF-independent Response
Perception of NFs results in triggering host symbiotic programme mediated by MtNFP and/or MtLYK3 [2]. In contrast, co-production of these LysM-RLKs in Nicotiana leaves apparently triggered some signalling cascade in the absence of NFs. Therefore, we investigated the effect of NF produced by Sinorhizobium meliloti, a microsymbiont of Medicago, on this CD response. To this end, Agrobacterium transformants carrying either MtNFP-sYFP2 or MtLYK3-mCherry construct were co-infiltrated into Nicotiana leaves at varying concentrations (as measured with OD 600 ). Then, purified SmNF at 10 27 M (in diluted DMSO) or diluted DMSO alone was applied between 9 and 24 hai to parts of the leaf regions co-producing MtNFP and MtLYK3 fusions, and CD development was monitored between 24 and 72 hours after the first infiltration (with Agrobacterium) using Evans blue staining. For all bacterial concentrations and time-points of SmNF/DMSO application tested, compromised membrane permeability in leaf regions co-producing MtNFP and MtLYK3 fusions was observed at similar time irrespective of the SmNF or DMSO treatment (Fig. 5), indicating similar kinetics of CD development. Therefore, we did not obtain evidence for any stimulatory or inhibitory effect of the SmNF on the CD development upon MtNFP and MtLYK3 co-production.

The Intracellular Region of MtNFP and Kinase Activity of MtLYK3 are Required for Cell Death Induction in Nicotiana Leaves
The independence of CD induction upon MtNFP and MtLYK3 co-production from the SmNF perception prompted us to compare structural requirements of CD induction and nodulation with regard to these LysM-RLKs. In case of MtNFP, a recent structurefunction study in Medicago [22] showed that loss-of-function mutations located in the ExR could be attributed to retention of the mutated protein in the endoplasmic reticulum (ER), whereas most substitutions located in the InR were found not to have an effect on the MtNFP function in nodulation. Therefore, we decided to limit our analysis of MtNFP to three point-mutated variants carrying: Ser 67 Phe (encoded by the Mtnfp-2 allele), Ser 67 Ala, and Gly 474 Glu substitution; and a truncated variant with almost the entire InR deleted, termed MtNFP[DInR] (amino acids: 1-283) (see Table 2). Based on structure-function studies on MtLYK3 and LjNFR1, respectively in Medicago and in Lotus ( [12,[20][21], Table 3 in this study), we decided to test the effect of 16 point mutations (listed in Table 3 Table 3). All truncated/mutated variants were prepared as fusions to the Nterminus of sYFP2, and their production and correct PM localization in Nicotiana leaf epidermal cells was confirmed, except for two MtNFP variants: MtNFP[S67F]-sYFP2 fusion was retained in the ER, and MtNFP[G474E]-sYFP2 fusion showed a partial PM localization, ( [20,22] and Fig. S1). Additionally, we ruled out a possibility that the presence of WT MtNFP-FP or WT MtLYK3-FP fusion might affect stability/localization of the truncated/mutated fusions by confirming their efficient production and PM localization in Nicotiana leaf epidermal cells also upon co-production with MtLYK3 or MtNFP fusions (unpublished data).
Subsequently, we analyzed the ability of truncated/mutated MtNFP and MtLYK3 variants to induce CD upon either their separate production or co-production with WT MtLYK3-mCherry or WT MtNFP-mCherry fusion, respectively. In order to compare the CD induction ability of truncated/mutated variants with WT proteins, concomitant co-infiltration with Agrobacterium transformants carrying either WT MtNFP-FP or WT MtLYK3-FP construct was done on every leaf. Development of CD was monitored between 36 and 72 hai, and in case of the absence of or weakly pronounced macroscopic symptoms, the occurrence of CD was further scrutinized with Evans blue staining. None of the truncated/mutated variants was able to induce CD in Nicotiana leaves on its own ( Table 2, 3). Co-production of MtNFP[S67A]-sYFP2 and MtLYK3 fusions resulted in a confluent death of (nearly) the entire infiltrated region in 6 out of 9 infiltrations, and compromised membrane permeability that could be observed in the entire infiltrated region ( Table 2). In contrast, co-production of MtLYK3 fusion with MtNFP[S67F]-sYFP2, excited autofluorescence (right panel) within the infiltrated region. Bar is 1 cm. C, Accumulation of phenolic compounds. The following fusions were (co-)produced in Nicotiana leaves: MtNFP-3xFLAG (1); MtLYK3-3xFLAG (2); MtNFP-3xFLAG+MtLYK3-3xFLAG (3); MtNFP-3xFLAG+MtLYK3[G334E]-3xFLAG (4); AtCERK1-3xFLAG (5); or AtCERK1[K349]-3xFLAG (6). Macroscopic observations (left panel) and subsequent UV-excited autofluorescence of ethanol/lactophenol-cleared (right panel) leaf regions are depicted 36 hai (except for 5-30 hai). Bars are 1 cm. D, Induction of NbHIN1, NbPR1 basic, NbACRE31, and NbACRE132 expression in response to separate production or co-production of: MtNFP-3xFLAG (NFP), MtLYK3-3xFLAG (LYK3), MtLYK3[G334E]-3xFLAG (LYK3[G334E]), and AtCERK1-3xFLAG (CERK1). Leaf samples were collected 24 hai and induction of gene expression was analyzed using qRT-PCR. Histograms represent induction of NbHIN1 (white columns), NbPR1 basic (grey columns), NbACRE31 (hatched columns), and NbACRE132 (black columns) normalized by one reference gene, MtEF1 a. Induction of each gene was normalized to that caused by mock infiltration, and then calculated as % induction relative to the induction observed upon co-production of MtNFP and MtLYK3 fusions. Bars represent standard deviation of the mean. At least two technical replicates from two biological replicates were analyzed. doi: 10 (Table 3). In contrast, co-production of MtNFP fusion with all seven MtLYK3-sYFP2 mutated variants affected for their autophosphorylation activity in vitro did not induce CD in Nicotiana leaves (Table 3). In case of mutations that do not affect autophosphorylation activity of MtLYK3 kinase, we found that MtLYK3[T285A S286A T300A]-sYFP2 and MtLYK3[S471A]-sYFP2 fusions were as active as WT MtLYK3-sYFP2 fusion for CD induction upon their co-production with MtNFP fusion, whereas MtLYK3[K464A]-sYFP2 fusion induced compromised membrane permeability (but no macroscopic symptoms of cell death) upon co-production with MtNFP fusion (Fig. S3, Table 3). Co-production of MtNFP fusion with MtLYK3[T433A]-sYFP2, MtLYK3[T472A]-sYFP2 or MtLYK3[T512A]-sYFP2 fusion resulted in a confluent death of (nearly) the entire infiltrated region in, respectively, 7 out of 20, 5 out of 11, and 12 out of 20 infiltrations, whereas the remaining leaf regions displayed only (a) small patch(es) of dead tissue (Fig. S3, Table 3).
Taken together, most of the structural requirements regarding the MtNFP and MtLYK3 InR, and the autophosphorylation activity of the MtLYK3 KD, appeared to be identical for biological activity of these LysM-RLKs in both Medicago and Nicotiana. More specifically, we found out that both nodulation and CD induction displayed the same requirements for 11 out of 15 residues located in the MtLYK3 InR. On the contrary, a single mutation in the MtLYK3 ExR tested (that does not affect the PM localization of the fusion) was found to be crucial for MtLYK3 function in nodulation but not in CD induction. In case of MtNFP, the substitution of Ser 67 similarly abolished (S67F) or did not have an effect (S67A) on MtNFP function in nodulation and CD induction, which seemed to correlate with, respectively, the absence or presence of MtNFP fusion at the PM.    [20,22] and oligomerization status in vivo (manuscript in preparation), and led to the surprising observation of a CD induction (Fig. 1A). This response phenotypically and kinetically (Fig. 4A) resembled the HR elicited in Nicotiana spp. by pathogen-derived components [49], or CD induced by (co-) production of certain defence-related proteins, e.g. Pto [58]. Notably, a similar CD response was observed in Nicotiana leaves upon production of AtCERK1 ( Fig. 2A), a MAMP receptor. Although CD induction was not demonstrated in any plant species in response to COs or PGN [54][55][56], it has been observed upon deregulation of various stress/defence-related signalling components [59][60], including a mitogen-activated protein kinase kinase (OsMKK4) implicated in COs-induced signalling in rice [61]. In addition, expression of AtCERK1 from the CaMV 35S promoter (used also in our studies) in Arabidopsis was shown to result in a ligand-independent dimerization of AtCERK1 [38]. We hypothesize that overproduction of this LysM-RLK in Nicotiana analogously leads to its dimerization and enhanced (deregulated) kinase activity, which in turn is required and sufficient for triggering CD; a response not observed upon ligand-induced activation of AtCERK1 in Arabidopsis.
Cell death induction upon MtNFP and MtLYK3 co-production in Nicotiana is in agreement with the Nicotiana [21] and Arabidopsis [19] response to LjNFR1 and LjNFR5 co-production. However, in these studies the associated induction of putative defence-related responses has not been investigated. We here showed that both MtNFP and MtLYK3 co-production and AtCERK1 production in Nicotiana leaves triggered local accumulation of phenolic compounds, and a similar induction of expression of 4 out of 6 tested defence-related genes (Fig. 4C, D). We speculate that the two other genes might display different kinetics of induced expression (here analyzed only 24 hai) or might undergo suppression by Agrobacterium [62]. Importantly, COs-and/or PGN-induced expression of PR1, ACRE31, ACRE132 and a member of a HIN1 gene family was reported previously [32,52,54], linking these genes to MAMP-induced gene regulation mediated by AtCERK1 in Arabidopsis and/or NbCERK1 in Nicotiana. In addition, we found that the lanthanum chlorideinduced impairment of a Ca 2+ influx similarly delayed the CD development upon (co-)production of the LysM-RLKs in our study (Fig. 3A, C). Therefore, we speculate that the signalling triggered upon MtNFP and MtLYK3 co-production in Nicotiana mimics AtCERK1-mediated signalling and thereby results in an induction of defence-like responses.

Similarities and Differences between Symbiotic and Defence Signalling Mediated by LysM-RLKs
The similarity between Nicotiana response to MtNFP and MtLYK3 co-production and AtCERK1 production suggests a possible overlap in signalling mediated by these LysM-RLKs. Several NF-induced processes, such as: a transient increase of reactive oxygen species (ROS) production; activation of phospholipase C (PLC) and PLD; and prolonged oscillations of (peri)nuclear [Ca 2+ ] are implicated in switching on the symbiotic programme [63][64][65][66][67][68][69], whereas a Ca 2+ influx is postulated to act as a signal for infection thread formation [70]. Interestingly, (CERK1-mediated) COs-and/or PGN-induced responses also *-see [20], except for the T480A (Fig. S2), **-see [20], except for the P87S [12], involve a Ca 2+ influx, an elevated ROS production, and PLC activation [33,35,[37][38][52][53][71][72][73]. We speculate that these similar processes might be activated/regulated by related molecular components, hence allowing two Medicago LysM-RLKs to activate signalling components present in Nicotiana leaf. Remarkably, Nakagawa and associates [74] demonstrated that swapping of the AtCERK1 ExR and a certain part of the AtCERK1 InR for the corresponding regions from LjNFR1 conferred on AtCERK1 a competence, albeit inefficient, for symbiotic signalling during Lotus-Mesorhizobium loti interaction. Conversely, our results demonstrate that MtNFP and MtLYK3, when co-produced in Nicotiana, are capable of signalling in a similar manner to AtCERK1. We hypothesize that due to the absence of symbiosis-specific ''decoders'' or ''modulators'' in Nicotiana, MtNFP-and MtLYK3-mediated signalling might be differently interpreted in this species, resulting in the induction of defence-like responses. Importantly, NF-induced host responses are partially contradictory. On one hand, NFs are postulated to suppress the production in legume roots of salicylic acid and ROS, two potent signals implicated in defence signalling, upon rhizobia perception [63,75]. This differential induction of the symbiotic competence or defence could be regulated quantitatively via NPR1 (Non-expressor of Pathogenesis-related genes 1)-mediated signalling [76]. On the other hand, even perception of compatible NFs leads to the induction of (some) defence-related gene expression and phosphorylation of defence-related proteins in the initial stage of symbiosis [7,11,68,74]. An even more striking example of a NF-induced defence-like response comes from Sesbania rostrata. NF produced by its microsymbiont, Azorhizobium caulinodans, triggers the production of hydrogen peroxide and ethylene that eventually lead to local CD required for the formation of cortical infection pockets [77]. Therefore, our results agree with the hypothesized stimulation of (initial) host's defence responses via LysM-RLKs that mediate NFinduced signalling.

Possible Co-functioning of MtNFP and MtLYK3 in Nicotiana and Medicago
To explain a possible signalling mechanism employed by the kinase-inactive MtNFP, it has been proposed to form a receptor complex with MtLYK3 and another LYK protein during, respectively, the infection thread growth and pre-infection stage of symbiosis [12,15]. In contrast, it is not known whether MtNFP functions alone during arbuscular mycorrhiza symbiosis [24][25][26] or resistance towards fungal and oomycete pathogens [27] or requires a similar co-functioning with (an)other RLK(s). We here demonstrated that MtNFP required MtLYK3 to induce CD in Nicotiana, and that neither of these LysM-RLKs could be substituted by an unrelated RLK (Fig. 1B, Table 1). We propose that this Nicotiana response reflects a functional interaction between these LysM-RLKs. A rather limited heteromerization of MtNFP and a kinase-inactive MtLYK3 variant observed in the PM of Nicotiana leaf epidermal cells (manuscript in preparation) does not exclude either hypothesized mechanism: a direct, phosphorylation-dependent physical interaction between MtNFP and WT MtLYK3; or indirect (functional) interaction between MtNFP and MtLYK3 that requires independent activation of different molecular components by either LysM-RLK, and a later convergence of such putative signalling pathways. Interestingly, MtLYK3, but not the homologous [78] AtCERK1, required the presence of MtNFP for CD induction in Nicotiana ( Fig. 2A), indicating the specific requirement for MtNFP to potentiate the MtLYK3-mediated signalling. This observation agrees with the hypothesized specialization of the LysM-RLKs mediating NF-induced signalling during the co-evolution of legumes with rhizobia [28].

Cell Death Induction in Nicotiana and Nodulation in Medicago Share Certain Structural Requirements Regarding MtNFP and MtLYK3
Curiously, CD induction in Nicotiana was independent from the NF perception (Fig. 5), and the presence of Pro 87 in the MtLYK3 ExR (Table 3), in contrast to MtNFP and MtLYK3 function in nodulation [12,15]. Further mapping of crucial amino acid residues, and detailed characterization of their exact role in signalling would be required to clarify in the future whether or not nodulation and CD induction indeed hold different structural requirements with regard to the ExRs of these LysM-RLKs. On the contrary, the biological activity of MtNFP in Nicotiana was dependent on its PM localization, the presence of its InR, and the Gly 474 (Table 2), thus mimicking the structural requirements of RL symbiosis regarding MtNFP [22], and proteins encoded by MtNFP orthologs in Lotus [79] and pea (Pisum sativum) [5]. The overlap between structural requirements of nodulation and CD induction was even more pronounced with respect to the MtLYK3 InR. Out of 16 residues whose role in nodulation was identified, 9 residues (Thr 319, Gly 334, Lys 349, Glu 362, Thr 433, Asp 441, Thr 472, Thr 475, and Thr 480) were found to be equally important, and 2 residues (Thr 285, Ser 286) were equally dispensable for MtLYK3 biological activity both in Medicago and in Nicotiana (Table 3). In addition, the K464A and T512A substitutions had a negative effect of MtLYK3 biological activity in both Nicotiana and Medicago, although this effect was more (K464A) or less (T512A) severely pronounced during CD induction assays than during nodulation (Table 3). Various mutations abolishing MtLYK3 autophosphorylation activity ( [20] and Fig. S2) similarly abolished its biological activity in Medicago [20] and in Nicotiana (Table 3), supporting the hypothesis that autophosphorylation of MtLYK3 is crucial for its signalling function. Importantly, as the role of Thr 480 in nodulation has not been described so far, our results revealed its importance for MtLYK3 function in vivo. Notably, the shared structural requirements of nodulation and CD induction were also confirmed with regard to several (putative) phosphorylation sites that do not abolish MtLYK3 autophosphorylation activity in vitro (Table 3). Phosphorylation within the InR of a RLK is often required for activation and regulation of its catalytic activity, and for generation of docking sites for (downstream) signalling components [80][81][82]. The shared importance of three (Thr 433, Thr 472, and Thr 512) out of five such phosphorylation sites for MtLYK3 biological activity in both plant species suggests that some of these phosphorylation-dependent functions required for MtLYK3-mediated signalling are conserved in Nicotiana leaf. Demonstrated significant overlap between structural requirements of nodulation and CD induction regarding the MtNFP and MtLYK3 InRs supports our notion of the relevance of the Nicotiana system for studies on these, and potentially other (symbiotic) LysM-RLKs. This system presents certain practical advantages over the legume root system, in terms of rapidity and ease of expression of multiple constructs. In view of hypothesized similarities between NF (i.e. lipo-chitooligosaccharide)-induced and COs-induced signalling, analyzing known molecular components/processes involved in the CERK1-mediated signalling [35][36][37][38]83] might provide information on the yet-unidentified players implicated in the perception and/or transduction of the NF signal. This would be especially important as still very little is known about the identity of interactors of these symbiotic LysM-RLKs [17][18]21,84]. Possible candidate signalling molecule(s) function-ing in co-operation with, or downstream from the LysM-RLKs, and identified in this heterologous system should then be evaluated in legume root in order to confirm their involvement in symbiosis.

Supporting Information
Figure S1 Subcellular localization of various protein fusions in Nicotiana leaf epidermal cells. The plasma membrane marker, mCherry-HVR, was co-produced with the designated fusions in Nicotiana leaf epidermal cells, and the fluorescence (viewed from abaxial side) was imaged 24 hai using confocal laser scanning microscopy. From left to right: green fluorescence of sYFP2; orange fluorescence of mCherry; superimposition of green, orange, and far-red (chlorophyll) fluorescence with the differential interference contrast (DIC) image. Bars are 20 mm. Note 1: in case of subcellular localization of MtNFP[G474E]-sYFP2 fusion, strong fluorescent puncta (indicated with an arrowhead) at the cell boundary of many cells (sometimes in association with nuclei), and pronounced ER localization (indicated with an arrow) of the fusion were still visible at 48 hai. Nevertheless, some cells showed a more uniform pattern of fluorescence at the cell boundary, and this observation, together with a partial insensitivity of this mutated variant to the PNGaseF treatment [22], indicated that at least some MtNFP[G474E] fusion had reached the PM. Note 2: as all kinase-inactive MtLYK3 variants were produced and correctly localized to the plasma membrane in Nicotiana leaf epidermal cells, their lack of biological activity can be attributed to the general abolishment of kinase activity rather than to an individual effect of a particular mutation.  Materials and Methods S1. (DOC)