An EFR‐Cf‐9 chimera confers enhanced resistance to bacterial pathogens by SOBIR1‐ and BAK1‐dependent recognition of elf18

Summary The transfer of well‐studied native and chimeric pattern recognition receptors (PRRs) to susceptible plants is a proven strategy to improve host resistance. In most cases, the ectodomain determines PRR recognition specificity, while the endodomain determines the intensity of the immune response. Here we report the generation and characterization of the chimeric receptor EFR‐Cf‐9, which carries the ectodomain of the Arabidopsis thaliana EF‐Tu receptor (EFR) and the endodomain of the tomato Cf‐9 resistance protein. Both transient and stable expression of EFR‐Cf‐9 triggered a robust hypersensitive response (HR) upon elf18 treatment in tobacco. Co‐immunoprecipitation and virus‐induced gene silencing studies showed that EFR‐Cf‐9 constitutively interacts with SUPPRESSOR OF BIR1‐1 (SOBIR1) co‐receptor, and requires both SOBIR1 and kinase‐active BRI1‐ASSOCIATED KINASE1 (BAK1) for its function. Transgenic plants expressing EFR‐Cf‐9 were more resistant to the (hemi)biotrophic bacterial pathogens Pseudomonas amygdali pv. tabaci (Pta) 11528 and Pseudomonas syringae pv. tomato DC3000, and mounted an HR in response to high doses of Pta 11528 and P. carotovorum. Taken together, these data indicate that the EFR‐Cf‐9 chimera is a valuable tool for both investigating the molecular mechanisms responsible for the activation of defence responses by PRRs, and for potential biotechnological use to improve crop disease resistance.


I N T RO D U C T I O N
Plants have evolved an innate immune system that relies on the recognition of potential pathogens by a defined pool of membrane and cytosolic receptors. The first layer of plant immunity comprises plasma membrane pattern recognition receptors (PRRs) that recognize microbe-associated molecular patterns (MAMPs), thereby mounting MAMP-triggered immunity (MTI) (Cook et al., 2015;Couto and Zipfel, 2016;Dangl et al., 2013;Dodds and Rathjen, 2010). Successful pathogens often produce effector proteins capable of suppressing MTI and preventing an effective immune response. As a countermeasure, plants have evolved pathogen strain-specific immune receptors, so-called resistance (R) proteins, which are capable of recognizing particular effectors and subsequently activate effector-triggered immunity (ETI). Although leading to outputs that are qualitatively similar to those of MTI, ETI is generally stronger and faster, and often includes a rapid programmed cell death (PCD) of the host cells at the site of infection, referred to as hypersensitive response (HR) (Chisholm et al., 2006;Jones and Dangl, 2006).
Interfamily transfer of PRRs is an effective strategy for improving resistance of plants to a wide range of infectious microbes (Dangl et al., 2013). For example, EFR confers broad-spectrum resistance to bacterial pathogens in tomato and Nicotiana benthamiana (Nb) (Lacombe et al., 2010), and also in the monocot rice (Schwessinger et al., 2015). The rice resistance protein Xa21, which confers resistance to Pst DC3000 upon transfer to Arabidopsis , has also been transferred to crops like citrus (Mendes et al., 2010), tomato (Afroz et al., 2011 and banana (Tripathi et al., 2014), to confer resistance to Xanthomonas species. Strategies based on stacking multiple resistance genes are also pursued to avoid that the R gene-mediated resistance might be overcome by pathogens. For example, transgenic potato (Solanum tuberosum) plants expressing three R genes against Phytophthora infestans show robust resistance (Chen and Ow, 2017;Halpin, 2005;Zhu et al., 2012).
Recent research provides evidence that engineering novel recombinant PRRs by domain swapping to obtain chimeric receptors that combine useful features of different PRRs is a promising option to breed for durable and wide spectrum resistance (Boutrot and Zipfel, 2017;De Lorenzo et al., 2011). Significant progress has been achieved, leading to the indication that the ectodomain of a chimeric receptor retains the ligand perception ability, while the endodomain maintains the output intensity (Albert et al., 2010;Brutus et al., 2010;He et al., 2000;Holton et al., 2015;Kishimoto et al., 2010;Kouzai et al., 2013;Schwessinger et al., 2015;Thomas et al., 2017). For instance, we previously reported that the chimera FLS2-EFR, carrying the ectodomain of FLS2 and the kinase domain of EFR, recognizes flg22 and induces an EFR-like response in N. tabacum (Nt, tobacco), which lacks an endogenous EFR receptor (Lacombe et al., 2010). This response typically results in a higher ethylene production than that induced by FLS2 upon perception of flg22 (Brutus et al., 2010). WALL-ASSOCIATED KINASE 1 (WAK1) is a receptor of oligogalacturonides (OGs) and triggers defence responses upon perception of this so-called damage-associated molecular pattern (DAMP). Upon OG treatment, a chimera comprising the ectodomain of WAK1 and the EFR kinase domain triggers an EFR-like response (Brutus et al., 2010). On the other hand, a reciprocal EFR-WAK1 chimera recognizes elf18 to induce a stronger OG/WAK1-like oxidative burst response than that triggered by EFR upon perception of elf18 (Brutus et al., 2010). However, the downstream signalling components employed by the transferred PRRs and chimeras are largely unknown, and the signalling pathways that are triggered in the plant from which the PRR originates and those in the recipient plant might not be conserved.
In this work, we aimed to obtain a MAMP-dependent robust resistance response in tobacco, by exploiting the EFR-mediated recognition of the broad-spectrum bacterial MAMP elf18 and the ability of Cf-9 to trigger a fast HR upon activation (Hammond-Kosack et al., 1998;Stergiopoulos and de Wit, 2009). Tobacco SR1 plants transiently expressing the chimera EFR-Cf-9, harbouring the ectodomain of EFR and the TM and cytoplasmic domain of Cf-9 (hereafter referred to as the endodomain), indeed mount a strong elf18-triggered HR. We show that chimeric EFR-Cf-9 protein constitutively interacts with SOBIR1, and requires both SOBIR1 and BAK1 for functionality. Transgenic tobacco plants expressing EFR-Cf-9 activate an HR upon elf18 treatment and show enhanced resistance to the (hemi)biotrophic bacterial pathogens Pta 11528 and Pst DC3000.

R E S U LT S
EFR-Cf-9 recognizes elf18C to trigger a strong HR in tobacco plants We previously reported that the ectodomain of EFR (Lys649 EFR ) is functional in an EFR-WAK1 chimera and perceives elf18, while the ectodomain of FLS2 (Arg806 FLS2 ) is functional in an FLS2-EFR chimera (eJMC in the original article) and perceives flg22 (Brutus et al., 2010). Here, we fused the ectodomains of EFR or FLS2 to the endodomain (Trp811 Cf-9 ) of Cf-9 (Jones et al., 1994), to obtain EFR-Cf-9 and FLS2-Cf-9 chimeras, respectively ( Fig. 1). Both chimeras were in turn fused to eGFP and transiently expressed by agroinfiltration in leaves of tobacco plants stably overexpressing Avr9 (Hammond-Kosack et al., 1998) or transiently co-expressing Avr9. In parallel, EFR-eGFP, FLS2-eGFP and Cf-9-eGFP fusions were agroinfiltrated as controls. Leaf areas transiently expressing eGFP-tagged EFR, EFR-Cf-9, FLS2 and FLS2-Cf-9 did not exhibit any symptom, while leaves co-expressing Cf-9-eGFP and Avr9, as expected, displayed a clear HR (Figs S1 and S2). These results indicate that the chimeras are not responsive to Avr9 and do not per se induce HR-like symptoms. Next, whether EFR-Cf-9 and FLS2-Cf-9 are functional and trigger a ligand-dependent HR-like response was tested. Upon infiltration with 100 nM elf18C, leaf areas transiently expressing EFR-Cf-9-eGFP showed a strong HR, similar to the Cf-9-eGFP/Avr9 control, in two different tobacco cultivars (SR1 and Samsun) (Fig. 2). In contrast, leaf areas transiently expressing FLS2-Cf-9-eGFP did not show visible symptoms upon treatment with 100 µM flg22 (Fig. S2), suggesting that FLS2-Cf-9-eGFP is not functional in tobacco.
Together, these data show that EFR-Cf-9 recognizes elf18C to trigger a strong Cf-9/Avr9-like HR, indicating that while the ectodomain of EFR retained the ability to perceive the elf18C peptide, its fusion to the endodomain of Cf-9 resulted in an output similar to that of the native Cf-9.
EFR-Cf-9 functionality requires SOBIR1 and BAK1 It has been reported that the TM domain of SOBIR1 is required for its interaction with Cf-4 (Bi et al., 2016) and that the co-receptors SOBIR1 and BAK1 are both required for the function of Cf-4 (Liebrand et al., 2013;Postma et al., 2016). Cf-9 shares an identical endodomain with Cf-4 ( Thomas et al., 1997) and also interacts with SOBIR1 (Liebrand et al., 2013). We therefore investigated whether EFR-Cf-9 also interacts with tomato SOBIR1 (henceforth indicated as SlSOBIR1). EFR-Cf-9-Myc was generated and co-expressed with SISOBIR1-eGFP in N. benthamiana. In parallel, FLS2-Cf-9 (as the non-functionality of this chimera might be due to a possible lack of interaction with SOBIR1), Cf-4-eGFP, FLS2-eGFP and EFR-eGFP were agroinfiltrated as controls. Immunoprecipitation (IP) of SlSOBIR1 using GFP-Trap beads demonstrated that Cf-was co-precipitated with SlSOBIR1 Fig. 1 Overview of the structure of the chimeric receptors. The coding regions of EFR, Cf-9 and FLS2 are indicated in dark grey, white and light grey, respectively. The arrows indicate the junction between the EFR or FLS2 ectodomain and the Cf-9 transmembrane (TM) domain, with the corresponding residues included in the chimeras. SP, signal peptide; LRR, leucine-rich repeat; eJM, external juxtamembrane; iJM, internal juxtamembrane; K, kinase.
Fig. 2 EFR-Cf-9 recognizes elf18C and triggers an HR. EFR-Cf-9, EFR and Cf-9, all fused to eGFP, were transiently expressed in wild type (WT) N. tabacum SR1 (left) and N. tabacum Samsun (right) plants (n = 4). After 2 days post agroinfiltration, leaves were treated with Milli-Q water (MQ) or 100 nM elf18C and pictures were taken after 2 days. As a positive control, eGFP-tagged EFR-Cf-9, EFR and Cf-9 were also co-infiltrated with an Agrobacterium strain driving expression of Avr9, and pictures were taken at 12 days post-infiltration (dpi). At these time points, all leaves agroinfiltrated with EFR-Cf-9 and subsequently infiltrated with el18C, and all leaves co-expressing Cf-9 and Avr9 showed necrosis of at least half of the infiltrated area, whereas none of the other leaves showed any HR-like symptoms. Experiments were repeated three times with similar results. Representative images are shown. , which is consistent with our previous finding (Liebrand et al., 2013). In addition, FLS2-Cf-9 and EFR-Cf-9 were also copurified with SlSOBIR1, indicating that both chimeras constitutively interact with SlSOBIR1, similar to the Cf proteins ( Fig. 3A).
Kinase activity of both SOBIR1 and BAK1 is required for the function of Cf-4 (Liebrand et al., 2013;Postma et al., 2016;Van Der Burgh et al., 2018). The AtBAK1-5 mutant contains a point mutation in its kinase domain, which causes this protein to have a slightly lower kinase activity than BAK1 itself (Schwessinger et al., 2011). BAK1-5 has been reported to have a dominant-negative effect on the BAK1-dependent immune response mediated by FLS2 and EFR (Schwessinger et al., 2011). In order to study whether kinase activity of BAK1 is also required for EFR-Cf-9 functionality and whether the kinase-inactive variant AtSOBIR1-RD/N (Bi et al., 2016) displays a dominant-negative effect similar to the hypoactive kinase AtBAK1-5 and kinase-dead AtBAK1-RD/N mutants (Schwessinger et al., 2011), EFR-Cf-9-eGFP was either transiently expressed in combination with AtBAK1-RD/N, AtBAK1-5, AtSOBIR1-RD/N-eGFP or GUS-eGFP, followed by treatment with 100 nM elf18C. The EFR-Cf-9-dependent elf18C-triggered HR was not affected by GUS-eGFP, but was reduced by AtBAK1-RD/N and even more compromised by , indicating that both BAK1 mutants have a dominant-negative effect on the EFR-Cf-9triggered HR, possibly because the overexpressed BAK1 mutants compete away the endogenous NtSERK3a/b from the activated EFR-Cf-9-containing signalling complex. We conclude that the EFR-Cf-9-dependent elf18C-triggered HR depends on the kinase activity of BAK1. Unlike AtBAK1-RD/N and AtBAK1-5, AtSOBIR1-RD/N-eGFP did not affect the HR (Fig.  5). Considering that RLPs constitutively interact with SOBIR1, whereas the recruitment of BAK1 to the signalling complex is ligand-dependent, it is possible that transiently expressed AtSOBIR1-RD/N-eGFP may fail to displace the endogenous NtSOBIR1/SOBIR1-like protein that is bound to EFR-Cf-9. Fig. 3 Both FLS2-Cf-9 and EFR-Cf-9 interact with SOBIR1. (A) eGFP-tagged tomato (Sl) SOBIR1 was transiently co-expressed with Myc-tagged Cf-4, FLS2, EFR, FLS2-Cf-9 and EFR-Cf-9. (B) Myc-tagged SlSOBIR1 was transiently co-expressed with eGFP-tagged FLS2, EFR, FLS2-Cf-9 and EFR-Cf-9.Agrobacteria driving expression of the various constructs were infiltrated at a final OD 600 of 1.0. After 2 days, proteins were extracted and immunoprecipitated (IP) using GFP-Trap beads. Proteins were detected by Western blot/immunoblot (WB) using α-GFP and α-Myc antibodies. Equal loading is indicated by Rubisco band. Experiments were repeated three times and representative images are shown.
Taken together, these data indicate that both SOBIR1 and BAK1 are required for the EFR-Cf-9-dependent elf18C-triggered HR, and that BAK1 has to be kinase-active.
EFR-Cf-9-transgenic tobacco plants are differentially responsive to elf18 variants An untagged version of EFR-Cf-9 was stably expressed in tobacco SR1 plants. Fourteen primary transformants were obtained, of which five had detectable expression of EFR-Cf-9 ( Fig. 6A) and were therefore propagated. Two independent lines (K1A and K5A) showing a 3:1 segregation ratio for kanamycin resistance, and therefore likely carrying an insertion in a single locus, were selected for further characterization. Transgenic plants were morphologically identical to the parental plants (Fig.  6B). To verify that the chimera was properly expressed and localized, trypsin digestion of microsomal leaf proteins from wild type (WT) and K1A plants was performed, followed by Liquid chromatography-tandem mass spectrometry (LC-MS/MS). Six peptides corresponding to the ectodomain of EFR (coverage of 15.49; score of 16.31) were found only in extracts from the transgenic plants (Fig. S3), confirming that EFR-Cf-9 is expressed and is likely membrane-localized. No peptide corresponding to the chimera was found in WT extracts.
Subsequently, the functionality of EFR-Cf-9 in K1A and K5A plants was assessed. Only leaves from both transgenic lines mounted an HR upon elf18C treatment, at concentrations ranging from 1 nM to 10 μM (Fig. 6C). Responsiveness to different elf18 variants was also tested in line K1A, revealing that an HR was triggered also by elf18B and elf18G, though to a lesser extent when compared with elf18C (Fig. S4). This suggests that, although the EFR ectodomain was fused to the endodomain of Cf-9, the structure of the ligand still determines the output. To obtain a more quantitative evaluation of the response mediated by EFR-Cf-9, and to assess whether the presence of EFR-Cf-9 might cause an elevated basal defence response, the expression of the marker genes Avr9/Cf-9 RAPIDLY ELICITED-132 (ACRE-132) (Durrant et al., 2000) and HAIRPIN INDUCED 1 (HIN1) (Gopalan et al., 1996) was analysed. Transcript levels of both genes were similar in MQ water-treated WT and K1A plants, indicating that EFR-Cf-9 does not affect basal defence (Fig. S5). In addition, as expected, WT plants did not respond to any of the elf18 variants tested, whereas they showed a significantly increased expression of ACRE-132 and HIN1 after flg22 elicitation (Fig. S5). In contrast, treatment of K1A plants with elf18C induced a significant increase in transcript levels for ACRE-132 and HIN1, which are required for the EFR-Cf-9-dependent elf18C-triggered HR. eGFP-tagged EFR, Cf-9 and EFR-Cf-9 were either transiently expressed with Avr9, or expressed alone, in N. tabacum Samsun plants silenced for NtSOBIR1/SOBIR1-like, NtSERK3a/b or GUS (n = 4). Agrobacteria driving expression of the constructs were infiltrated at a final OD 600 of 1.0, except for the positive control Rx (D460V), which was infiltrated at an OD 600 of 0.1. At 2 days postinfiltration, leaves were treated with Milli-Q water (MQ) or 100 nM elf18C. Pictures were taken at 2 days after treatment with elf18C or, in the case of Cf-9-Avr9 interaction, at 12 days after co-infiltration of Cf-9 and Avr9. Experiments were repeated three times with similar results and representative images are shown.
were even higher than in plants treated with flg22 (Fig. S5). Moreover, treatment with elf18B and elf18G also resulted in increased transcript levels for ACRE-132 and HIN1, but to a lesser extent than for elf18C (Fig. S5), which supports our finding that elf18C triggers a stronger HR than the elf18 variants ( Fig. S4). On the other hand, WT and K1A plants displayed similar expression levels of both genes in response to flg22 (Fig.  S5), indicating that EFR-Cf-9 does not affect the endogenous NtFLS2-mediated response to flg22.
Together, these data indicate that transgenic tobacco plants expressing EFR-Cf-9 are not altered in their basal defence and in their responsiveness to flg22. Moreover, the EFR-Cf-9-transgenic plants respond more efficiently to elf18C than to the different elf18 variants.
Transgenic tobacco plants expressing EFR-Cf-9 show enhanced resistance to (hemi)biotrophic bacterial pathogens To study whether the EFR-Cf-9-mediated immune response eventually leads to resistance against bacterial pathogens, WT and transgenic K1A and K5A plants were inoculated with the (hemi)biotrophic bacterial pathogens Pta 11528 and Pst DC3000 at a dose of OD 600 = 0.002 (corresponding to 10 4 CFU/cm 2 ). Compared to WT, a clear reduction in bacterial colonization was found in both transgenic lines inoculated with Pta 1152 and, to an even greater extent, Pst DC3000 ( Fig. 7A and B). To verify that EFR-Cf-9 indeed triggers an HR not only in response to the purified elf18 peptides, but also to bacterial infection, we inoculated different doses (OD 600 = 0.002, 0.02 and 0.2) of Pta 11528 in WT, K1A and K5A plants. Within 48 h, leaf sectors of the transgenic plants inoculated with the highest doses of bacteria (OD 600 = 0.02 and 0.2) displayed HR-like symptoms, whereas WT plants did not show any symptoms with all doses of bacteria (Fig. 8A).
We also tested the susceptibility to the necrotrophic bacterium Pectobacterium carotovorum subsp. carotovorum strain DSMZ 30169, which is the causal agent of bacterial soft rot. Inoculation with P. carotovorum is able to induce an EFRdependent response in Arabidopsis, indicating the presence of an elf18-like MAMP (Lacombe et al., 2010). Leaves inoculated with P. carotovorum at a density of OD 600 = 0.02 (corresponding to about 10 5 CFU/cm 2 ) showed a ten-fold increase in bacterial count at 24 h, with no significant differences between WT and the transgenic lines (Fig. 7C). Notably, also in this case we observed a rapid and strong HR after inoculation with bacteria in K1A and K5A but not in WT plants (Fig. 8B). This suggests that EF-Tu of P. carotovorum is recognized by EFR-Cf-9 and triggers an HR that is not able to restrict colonization by this pathogen.
The fungal pathogen Botrytis cinerea does not produce an elf18-like MAMP and its colonization in Arabidopsis is not affected by the lack of EFR (Brutus et al., 2010). Consistently, when B. cinerea was inoculated onto WT and transgenic K1A and K5A leaves, no differences in disease severity was observed (Fig. S6), indicating that the ectodomain of EFR-Cf-9 does not recognize any MAMP from B. cinerea and that basal defence is not enhanced by the presence of the chimera.
Taken together, these data indicate that pathogen-derived elf18-like MAMPs activate EFR-Cf-9-dependent immunity resulting in an HR, which restricts (hemi)biotrophic bacteria colonization, but does not compromise colonization by necrotrophic bacteria or fungi.

D I SC U SS I O N
The chimera EFR-Cf-9 recognizes elf18 to trigger a strong immune response Interfamily transfer of PRRs is a promising strategy to confer broad-spectrum resistance to pathogens, and it has been successfully employed with EFR (Lacombe et al., 2010; Schwessinger  (Song et al., 2017). Engineering chimeric receptors that combine the properties of two separate PRRs is another effective strategy to improve host resistance. Current research on chimeric PRRs has revealed that the ectodomain determines ligand specificity, while the endodomain determines output intensity (Brutus et al., 2010;He et al., 2000;Holton et al., 2015;Kishimoto et al., 2010;Kouzai et al., 2013;Schwessinger et al., 2015). Here we generated the chimera EFR-Cf-9 (Fig. 1), combining the ectodomain of EFR, which provides broad-spectrum recognition of bacterial pathogens, and the endodomain of Cf-9, which induces a strong HR-associated immune response. We demonstrated that, when expressed in tobacco plants, EFR-Cf-9 recognizes elf18 leading to an HR (Fig.  2). We also showed that EFR-Cf-9 interacts with SOBIR1, similar to the Cf-9 protein itself (Fig. 3), and requires both SOBIR1 and BAK1 for its function (Fig. 4). In addition, EFR-Cf-9 also recognizes pathogen-derived elf18 or elf18-like MAMPs to activate immunity (Fig. 7) and triggers HR during bacterial infection (Fig.  8). Moreover, EFR-Cf-9 retains the recognition feature of EFR as elf18C, compared to elf18 variants with lower EFR-eliciting activity, triggers a stronger response, (Figs S4 and S5).
In contrast to EFR-Cf-9, FLS2-Cf-9 failed to trigger an HR upon treatment with flg22 (Fig. S2), although FLS2-Cf-9 also interacts with SOBIR1 in N. benthamiana (Fig. 3). Although the FLS2 ectodomain is functional in FLS2-EFR (Brutus et al., 2010) and the Cf-9 endodomain is functional in EFR-Cf-9 (Fig. 2), it is still possible that the selected point of junction between the ectodomain of FLS2 and the endodomain of Cf-9 might not be optimal for functionality of this specific chimera.   addition, specific downstream signalling components may act in an opposite manner in different species. For instance, the Xa21-binding (XB) protein OsXB24 is an ATPase that negatively affects both Xa21-and EFR-mediated immunity in rice (Chen et al., 2010;Schwessinger et al., 2015), while its orthologue plays a positive role in EFR-mediated immunity in Arabidopsis . Thus, the components required for functionality of a transferred PRR are not easily predictable. Ectopic expression of chimeric receptors allows to address the requirement of specific protein domains for physical and functional interaction with partners participating in downstream signalling. For example, both EFR-Xa21 and native Xa21 form a constitutive complex with OsSERK2, indicating that OsSERK2 interaction with Xa21 does not specifically require the Xa21 ectodomain. In Arabidopsis, both Xa21 and EFR-Xa21 interact with BAK1 in a ligand-dependent manner (Schwessinger et al., 2015). In this study, we show that EFR-Cf-9 constitutively interacts with SOBIR1 and requires SOBIR1 for its function (Figs 3 and 4). EFR-Cf-9 is anticipated to also form a complex with BAK1 upon elf18 treatment, as the chimera also requires kinase-active BAK1 for its function (Fig. 5). It should be noted that AtBAK1-5 had a stronger effect than AtBAK1-RD/N in suppressing EFR-Cf-9-mediated HR (Fig. 5). The interaction between AtBAK1 and EFR is kinase activity-independent and AtBAK1-5 has a higher affinity to EFR than WT AtBAK1 (Schwessinger et al., 2011), explaining why AtBAK1-5 is more efficient than AtBAK1-RD/N in suppressing EFR-Cf-9 activity. SOBIR1 constitutively interacts with Cf-4, which shares an identical endodomain with Cf-9 (Liebrand et al., 2013;Thomas et al., 1997), whereas it does not interact with the RLKs FLS2 and EFR and is not required for RLK-mediated immunity (Gust and Felix, 2014). The external juxtamembrane region of EFR-Cf-9 does not carry the typical stretch of acidic amino acids that are thought to play a role in the interaction of Cf-9 with SOBIR1 (Bi et al., 2016;Gust and Felix, 2014). However, EFR-Cf-9 does interact with SOBIR1, suggesting that the TM of Cf-9, carrying an extensive GxxxGxxxGxxxG dimerization motif, is sufficient for interaction with SOBIR1 (Bi et al., 2016). The requirement of the GxxxG motif for the interaction between SOBIR1 and EFR-Cf-9 will be investigated by performing site-directed mutagenesis of this motif in future studies. Together, this indicates that the endodomain, and in particular the TM, of Cf-9 provides the chimera EFR-Cf-9 with the features necessary for Cf-9/Cf-4 signalling, enabling constitutive interaction with SOBIR1 and ligand-dependent BAK1 recruitment to the EFR-Cf-9/SOBIR1 complex, eventually mounting an HR.

Transgenic expression of EFR-Cf-9 affects resistance against bacterial pathogens
In tomato, Cf proteins mediate a strong resistance to the C. fulvum strains carrying the proper avirulence genes (Joosten et al., 1997;Stergiopoulos and de Wit, 2009). It is widely accepted that this resistance is largely based on the HR resulting from the activation of the Cf protein, which leads to localized cell death that restricts pathogen spread. However, the strong resistance mediated by Cf genes has a narrow specificity, and is easily overcome by mutations in the corresponding Avr genes of the pathogen. On the other hand, PRR-mediated immunity triggered upon recognition of MAMPs is effective against a wide range of microbes but is weaker than that of Cf proteins. Hence, the generation of a chimera containing the ectodomain of a PRR and the endodomain of a Cf protein might combine the beneficial features of PRRs and Cf proteins, and enable the chimera to recognize MAMPs and mount an HR, leading to a strong immunity to a wide range of microbes.
EFR-Cf-9 recognizes elf18 to mount an HR (Fig. 2), suggesting that it might also recognize pathogen-derived elf18 to activate immunity. Indeed, transgenic tobacco plants expressing EFR-Cf-9 showed a significantly reduced susceptibility to two (hemi)biotrophic bacterial pathogens, Pta 11528 and Pst DC3000 ( Fig. 7A and B). In addition, inoculation of a high dose of Pta 11528 causes HR in the transgenic plants (Fig. 8A), indicating that EFR-Cf-9 recognizes both the purified elf18 peptide and the pathogen-derived EF-Tu, and that this recognition effectively restricts bacterial growth. However, compared to Pst DC3000, EFR-Cf-9-mediated resistance to Pta 11528 is less efficient (Fig.  7A and B), suggesting that Pta 11528 might partially suppress the EFR-Cf-9-mediated immune response, as Pta 11528 has been shown to suppress salicylic acid-mediated defence responses (Lee et al., 2013).
Inoculation of the necrotrophic bacterium P. carotovorum in the EFR-Cf-9-transgenic plants also resulted in HR-like symptoms (Fig. 8B), since P. carotovorum carries a form of EF-Tu that is able to activate EFR (Lacombe et al., 2010). However, no significant difference in P. carotovorum growth between WT and transgenic plants was found (Fig. 7C).

CO N C LU S I O N
This work indicates that the EFR-Cf-9 chimera is functional in tobacco plants, since it recognizes elf18 to trigger an HR, which fits our current view on the ectodomain of a PRR determining the ligand specificity, and the endodomain determining the output intensity. Moreover, EFR-Cf-9 constitutively interacts with SOBIR1, and the EFR-Cf-9-mediated HR triggered by elf18 is dependent on SOBIR1 and BAK1, which is reminiscent of our working model for Cf-4/Cf-9 and indicates that the endodomain of Cf-9 confers EFR-Cf-9 the repertoire of Cf-9/Cf-4-like signalling. In addition, transgenic tobacco plants expressing EFR-Cf-9 do not show altered basal defence but are resistant to (hemi)biotrophic bacteria, suggesting a potential biotechnological use of this chimeric receptor to improve crop disease resistance. Furthermore, our construction of the EFR-Cf-9 chimera reveals that it is possible to customize crop immunity by the generation of chimeric receptors containing features of a varying ligand recognition range and immune response intensity. E X PE R I M E N TA L PROC E D U R E S Plant materials and growth conditions N. benthamiana, N. tabacum Samsun, N. tabacum SR1, Avr9transgenic N. tabacum SR1 (Hammond-Kosack et al., 1998), and EFR-Cf-9-transgenic N. tabacum SR1 plants were grown under 16 h of light at 25 °C, and 8 h of darkness at 21 °C in climate chambers with a relative humidity of 75%.

Generation of transgenic tobacco plants
Transformation of tobacco plants was performed according to established methods (Horsch et al., 1988). Briefly, A. tumefaciens carrying the 35S:EFR-Cf-9 plasmid were suspended in an infection medium composed as follows: Murashige-Skoog (MS) basal medium, 3% w/v sucrose, 200 µM acetosyringone, 0.001% v/v Silwet. Leaf discs of 1 cm of diameter from 4-week-old tobacco SR1 leaves were incubated in the infection medium for 30 min at 25 ºC. Leaf discs were then placed on co-culturing MS medium containing Gamborg's vitamin mix, 1 mg/L 6-BAP, 0.1 mg/L NAA, 200 µM acetosyringone at 25 °C in the dark for 2 days. Infected explants were then transferred to fresh solid MS medium supplemented with vitamin mix, 1 mg/L 6-BAP, 0.1 mg/L NAA, 200 mg/L timentin, 200 mg/L cefotaxime and 200 µg/mL kanamycin at 25 °C in the light for 30 days for shoot regeneration. Regenerated shoots were transferred to fresh solid MS medium supplemented with vitamin mix, 0.1 mg/L NAA, 200 mg/L timentin, 200 mg/L cefotaxime and 200 mg/L kanamycin at 25 °C in the light for 30 days for root regeneration. From about 1200 co-cultivated leaf explants, 14 primary transformants were obtained and transferred to soil for propagation. Five transformants had detectable transgene expression and were propagated; of these, two (K1A and K5A) showed a 3:1 segregation ratio for kanamycin resistance, and were therefore selected for further characterization.

Hypersensitive response assays
The elf18 (elf18D) and flg22 peptides were synthesized by EZBiolab (Carmel, IN, USA). The other elf18 peptides were kind gifts of Cyril Zipfel (The Sainsbury Laboratory, Cambridge, UK). For HR assays in N. tabacum SR1 and N. tabacum Samsun plants, fully expanded leaves were infiltrated with Agrobacterium suspensions at an OD 600 of 1.0, except for Rx (D460V), which was infiltrated with an OD 600 of 0.1 (Liebrand et al., 2013;Postma et al., 2016). MQ water and elf18 and/or flg22 peptides were infiltrated at 2 days post-agroinfiltration (dpi) at the indicated concentrations. Leaves were examined for development of an HR between 2 dpi and 12 dpi. For each treatment, at least three leaves, taken from separate plants, were agroinfiltrated.

Pathogenicity assays
P. amygdali pv. tabaci (Pta) 11528 and P. syringae pv. tomato (Pst) DC3000 were grown in 5 mL of low salt Bertani-Luria (LSBL) medium at 28 °C for 24 h at 200 rpm until OD 600 was 1.0. Bacteria were centrifuged for 10 min at 2500 × g, the supernatant was discarded and bacteria were rinsed twice with sterile water. Bacteria were suspended in an infiltration medium containing 10 mM MgCl 2 at the indicated doses.
P. carotovorum subsp. carotovorum strain DSMZ 30169 was obtained from DSMZ GmbH (Germany). Bacteria were grown in liquid Luria-Bertani (LB) broth at 28 °C for 16 h, centrifuged at 8000 × g for 10 min, and washed with sterile water. Bacteria were then suspended in 50 mM potassium phosphate buffer, pH 7.0 and inoculated at the indicated doses.
Bacteria were syringe-infiltrated into leaf sectors of 4-weekold tobacco plants (six sectors per leaf, three leaves per genotype). Inoculated plants were returned to greenhouse and bacterial count was determined after 0 h and 120 h (for Pta 11528 and Pst DC3000) or 24 h (for P. carotovorum). Briefly, inoculated leaf sectors were sterilized for 1 min in EtOH 70%, washed for 1 min in sterile water, and leaf discs (diameter = 0.4 mm) were cut from each sector. Discs were ground with a pestle in 100 μL of sterile water, then additional 900 μL of water were added; samples were vortexed, and serial dilutions were plated on LB solid medium. Plates were incubated at 28 ºC for two days, and colonies were counted for each dilution.

Mass spectrometry
Membrane-enriched protein fractions from one leaf of a WT SR1 plant and of a transgenic K1A plant were extracted as previously described (Mattei et al., 2016). Proteins were separated by SDS-PAGE, and lanes were cut into ten slices, that were subjected to in-gel trypsin digestion. Peptides were analysed by LTQ liquid chromatography (LC) Orbitrap MS/MS, and protein identification was performed by MaxQuant platform and Proteome Discoverer as previously described (Mattei et al., 2016).

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of this article at the publisher's web site: Fig. S1 EFR-Cf-9 does not trigger an HR in tobacco plants expressing Avr9. eGFP-tagged EFR, Cf-9 and EFR-Cf-9 were transiently expressed in Avr9-transgenic N. tabacum SR1 plants at a final OD 600 of 1.0. For each construct, at least three leaves, taken from separate plants, were agro-infiltrated. Pictures were taken at three days post-infiltration (dpi). The infiltrated areas are indicated by white dashed lines. Under these conditions, all leaves agroinfiltrated with Cf-9-eGFP, and none of the other samples, showed necrosis of at least half of the infiltrated area. This experiment was repeated three times with similar results. Representative images are shown. Fig. S2 FLS2-Cf-9 does not trigger an HR in tobacco plants upon treatment with flg22. eGFP-tagged FLS2, Cf-9 and FLS2-Cf-9 were either expressed alone (first and third column) or transiently co-expressed with Avr9 (second column), or, followed by treatment with Milli-Q (MQ) water (first column) or 100 μM flg22 (third column) after two days. Agrobacteria driving expression of the various constructs were infiltrated at a final OD 600 of 1.0. Pictures were taken at four days post-infiltration (dpi) with the elicitor or at seven days of co-expression with Avr9. The infiltrated areas are indicated by white dashed lines. For each treatment, at least three leaves per construct, taken from separate plants, were infiltrated. Under these conditions, all leaves co-expressing Avr9 and Cf-9-eGFP showed necrosis of at least half of the infiltrated area, whereas none of the other samples showed HR-like symptoms. This experiment was repeated three times with similar results. Representative images are shown. Fig. S3 Identification of EFR Cf 9 in microsomal fractions of transgenic tobacco plants. Amino acid sequence of the EFR Cf 9 chimeric receptor. The five peptides identified by mass spectrometry in microsomal protein fractions extracted from transgenic EFR-Cf-9 expressing K1A leaves are underlined. SP, signal peptide; LRR, leucine rich repeat; eJM, external juxtamembrane; TM, transmembrane; iJM, internal juxtamembrane. Fig. S4 EFR Cf 9 transgenic plants recognize different elf18 variants. Fully expanded leaves of 4-week-old untransformed wild type (WT) and transgenic tobacco plants expressing EFR Cf 9 (K1A) plants were treated with Milli-Q (MQ) water or with flg22, elf18C, elf18B or elf18G at a concentration of 100 nM. For each peptide, at least four leaves per genotype, taken from independent plants, were infiltrated. Pictures were taken at 40 h post-infiltration. The infiltrated area is indicated by the white dashed line. At this time point, all leaves from K1A plants infiltrated with el18C, elf18B or elf18C showed necrosis of at least half of the infiltrated area, whereas those infiltrated with flg22 or water, and all leaves from WT plants, failed to display any HR-like symptoms. This experiment was repeated three times with similar results, and representative images are shown. Fig. S5 Defence-related genes ACRE-132 and HIN1 are differentially up regulated in transgenic plants expressing EFR-Cf-9 upon treatment with elf18 variants. Quantitative Reverse Transcription Polymerase Chain Reaction (qRT PCR) was used to determine ACRE-132 (A) and HIN1 (B) expression levels in WT and EFR-Cf-9 expressing transgenic K1A plants upon treatment with Milli-Q (MQ) water or with elf18C, elf18B, elf18G and flg22 at a concentration of 100 nM. RNA was extracted from infiltrated leaf sectors obtained from three separate plants for each genotype. Bars represent average ± standard deviation (SD) of three replicates. Different letters indicate statistically significant differences, according to one-way analysis of variance (ANOVA) followed by Tukey's significance test (P < 0.05). This experiment was repeated twice with similar results. Fig. S6 Transgenic plants expressing EFR-Cf-9 show unaltered susceptibility to Botrytis cinerea. Leaves of soil grown 4-weekold wild type (WT) and transgenic plants expressing EFR-Cf-9 (K1A and K5A) were inoculated with B. cinerea spore suspension, and lesion areas were measured at 48 hpi. Bars indicate average lesion area ± standard error (SE) (n > 18). No significant difference was observed between WT and transgenic lines (K1A and K5A), according to Student's t-test (P > 0.5)