Summary
The first layer of plant immunity is activated by cell surface receptor-like kinases (RLKs) and proteins (RLPs) that detect infectious pathogens. Constitutive interaction with the RLK SUPPRESSOR OF BIR1 (SOBIR1) contributes to RLP stability and kinase activity. As RLK activation requires transphosphorylation with a second associated RLK, it remains elusive how RLPs initiate downstream signaling. To address this, we investigated functioning of Cf RLPs that mediate immunity of tomato against Cladosporium fulvum.
We employed live-cell imaging and co-immunoprecipitation in tomato and Nicotiana benthamiana to investigate the requirement of associated kinases for Cf activity and ligand-induced subcellular trafficking of Cf-4.
Upon elicitation with the matching effector ligands Avr4 and Avr9, BRI1-ASSOCIATED KINASE 1 (BAK1) associates with Cf-4 and Cf-9. Furthermore, Cf-4 that interacts with SOBIR1 at the plasma membrane, is recruited to late endosomes after elicitation. Significantly, BAK1 is required for Avr4-triggered endocytosis, effector-triggered defenses in Cf-4 plants and resistance of tomato against C. fulvum.
Our observations indicate that RLP-mediated immune signaling and endocytosis require ligand-induced recruitment of BAK1, reminiscent of BAK1 interaction and subcellular fate of the FLAGELLIN SENSING 2 RLK. This reveals that diverse classes of cell surface immune receptors share common requirements for signaling initiation and endocytosis.
Introduction
The innate immune system of higher eukaryotes senses a wide range of pathogens through distinct pattern recognition receptors (PRRs). The engagement of these receptors leads to the activation of signaling pathways resulting in the induction of defense responses. Plants rely on plasma membrane-resident receptor-like kinases (RLKs) and receptor-like proteins (RLPs) as a first layer of the immune system against phytopathogenic microbes, many of them exploiting the extracellular space of plant tissues for their growth (Faulkner & Robatzek, 2012). Therefore, decoding the mechanisms underlying the perception of pathogen-derived signals by RLKs and RLPs at the host cell surface and subsequent activation of defense signaling is an important aspect for understanding how plant immunity is initiated. The Cf resistance genes encode leucine-rich repeat (LRR)-RLPs conferring immunity to specific races of the pathogenic fungus Cladosporium fulvum causing leaf mold disease of tomato (Solanum lycopersicum) (Joosten & de Wit, 1999; Rivas & Thomas, 2005; Stergiopoulos & de Wit, 2009). Distinct members of the Cf receptor family are activated upon recognition of their matching ligands, which are the so-called avirulence (Avr) proteins that are secreted by the fungus upon invasion of tomato leaves. The founding member of the Cf proteins and LRR-RLPs in general is Cf-9, which mediates resistance to Avr9-producing strains of C. fulvum, whereas for example Cf-2 and Cf-4 recognize Avr2 and Avr4, respectively (Rivas & Thomas, 2005). The Cf receptors mediate race-specific immunity to C. fulvum, which is typically associated with the hypersensitive response (HR), a form of programmed cell death.
Well-characterized PRRs are FLAGELLIN SENSING 2 (FLS2) and ELONGATION FACTOR-TU RECEPTOR (EFR), which are LRR-RLKs initiating broad-spectrum immunity against bacterial infection upon detection of the pathogen-associated molecular patterns (PAMPs) flagellin (flg22) and EF-Tu, respectively (Boller & Felix, 2009). To activate immune signaling, FLS2 and EFR form ligand-induced complexes with members of the SOMATIC EMBRYOGENESIS RECEPTOR KINASE (SERK) LRR-RLK family, of which BRI1-ASSOCIATED KINASE 1 (BAK1)/SERK3 plays a prominent role in FLS2-mediated immunity (Heese et al., 2007; Roux et al., 2011). BAK1 was initially found to interact with the LRR-RLK brassinosteroid receptor BRASSINOSTEROID INSENSITIVE 1 (BRI1), and is involved in a number of immune pathways and developmental processes, in line with its broader function as a co-receptor (Santiago et al., 2013; Sun et al., 2013; Liebrand et al., 2014). All receptors mentioned above are plasma membrane-localized proteins allowing them to detect extracellular microbe-derived ligands. In order to reach their destination after their maturation in the endoplasmic reticulum (ER), they have to be targeted to the plasma membrane by secretory trafficking (Beck et al., 2012a). As part of the immune response, plasma membrane-resident FLS2, when activated by flg22, is recruited to ARA7/RabF2b- and ARA6/RabF1-positive endosomes, which are compartments of the late endosomal trafficking pathway (Robatzek et al., 2006; Beck et al., 2012c). This process depends on BAK1/SERK3 (Chinchilla et al., 2007; Beck et al., 2012c), which itself undergoes constitutive endocytosis (Russinova et al., 2004).
In contrast to RLKs, RLPs lack an intracellular signaling domain and therefore it has been long suggested that these receptors require the interaction with a protein kinase to initiate signaling (Jones et al., 1994; Joosten & de Wit, 1999; Liebrand et al., 2014). For example, the Toll-like receptors (TLRs) of the mammalian innate immune system require interaction with the MyD88 adaptor protein to recruit the kinase IRAK and subsequently activate immune signaling (Moresco et al., 2011; Broz & Monack, 2013). Following the paradigm of the involvement of BAK1/SERK3 in RLK-mediated signaling, there is genetic evidence for a role of BAK1/SERK3 also in plant defense responses mediated by RLPs. Both resistance to the fungal pathogen Verticillium dahliae, conferred by the tomato LRR-RLP Ve1 and immunity elicited by Sclerotinia sclerotiorum through Arabidopsis thaliana RLP30, are dependent on BAK1/SERK3 (Fradin et al., 2009; Fradin et al., 2011; Zhang et al., 2013). Furthermore, another SERK family member, SERK1, was also shown to be genetically required for Ve1 and Cf-4 antifungal immunity in Arabidopsis and tomato, respectively (Fradin et al., 2011). However, no molecular interaction between these LRR-RLPs and the SERKs has been reported, leaving the possibility that the observed phenotypes are indirect, for example through BAK1/SERK3 function in cell death control or damage-induced immunity (Kemmerling et al., 2007; Gao et al., 2009; Schwessinger et al., 2011; Liu et al., 2013; Tintor et al., 2013). In addition to its positive regulatory role, BAK1/SERK3 was also described to negatively regulate immunity in tomato (Sl) mediated by SlEix2, an LRR-RLP recognizing fungal xylanase (Bar et al., 2010). In a BAK1/SERK3-dependent manner, the close homologue SlEix1 impairs SlEix2 endocytosis and HR (Bar et al., 2010).
Besides genetic evidence for a role of SERK1, no direct involvement in Cf-mediated immunity of the SERKs has been reported to date. Instead, another LRR-RLK, referred to as SUPPRESSOR OF BIR1-1/EVERSHED (SOBIR1/EVR), has recently been identified as a critical component in LRR-RLP-mediated immunity (Jehle et al., 2013; Liebrand et al., 2013; Zhang et al., 2013; Zhang et al., 2014). SlSOBIR1 interacts specifically with RLPs, including the Cf-2, Cf-4 and Cf-9 proteins, and this interaction occurred independently of the presence of their matching Avr ligands. The constitutive association of these LRR-RLPs with SlSOBIR1 is required for RLP stability and has been proposed to allow activation of a cytoplasmic signaling cascade upon perception of an Avr by the interacting Cf protein (Liebrand et al., 2013; Liebrand et al., 2014). It however remains to be elucidated how ligand-dependent activation of LRR-RLP signaling actually occurs.
Colonization of tomato leaf tissue by C. fulvum is fully confined to the apoplast and the fungus does not form specialized feeding structures like haustoria. Consistent with this life-style, the Avr proteins secreted by the fungus appear to solely accumulate in the apoplast (Joosten & de Wit, 1999; Thomma et al., 2005; Stergiopoulos & de Wit, 2009; Joosten, 2012). This suggests that the Cf receptors perceive the various secreted Avr proteins of C. fulvum at the plasma membrane of the host cells. However, the exact subcellular localization of Cf proteins has remained unclear ever since their first identification (Jones et al., 1994). Initially it was shown that Cf-9 is present both at the ER and at the plasma membrane (Piedras et al., 2000). A subsequent study demonstrated a role of a C-terminal dilysine motif of Cf-9 in ER targeting (Benghezal et al., 2000), which however was later found to be dispensable for Cf-9 function (van der Hoorn et al., 2001). More recently, ER-resident chaperones were identified as interacting proteins of Cf-4 and were shown to be important for Avr4-triggered immunity (Liebrand et al., 2012), which provided further evidence for ER localization of these RLPs.
Here, we have used live-cell imaging of transiently and stably expressed fluorescent protein fusions in N. benthamiana to investigate the subcellular localization and dynamics of Cf-4 in planta. Our data suggest that Cf-4, together with SlSOBIR1, perceives Avr4 at the plasma membrane of the host cells, a process that subsequently induces the association of Cf-4 with BAK1/SERK3, and likewise with SERK1. We furthermore demonstrate a role of these SERK family members in Cf-mediated immunity and show that the Cf-4 receptor undergoes BAK1/SERK3-dependent endocytosis upon Avr4 perception.
Materials and Methods
Plant materials and constructs
Nicotiana benthamiana plants were grown under 16 hours of light at 24°C and 45 - 65 % humidity. Various constructs used in this study have been described before (in all cases GFP refers to enhanced (e)GFP); Cf-4-GFP, Cf-9-GFP, SlSOBIR1-GFP, SlSOBIR1-like-GFP, AtSOBIR1-GFP, AtSOBIR1-Myc and AtSOBIR1D489N-Myc (Liebrand et al., 2012; Liebrand et al., 2013); FLS2-YFPn/c (Frei dit Frey et al., 2012); MEMB12-mCherry (Geldner et al., 2009); VHA-a1-RFP, RFP-ARA7 and ARA6-RFP (Dettmer et al., 2006); AtBAK1 (Schwessinger et al., 2011). Cf-4-YFPn/c and SlSOBIR1-YFPn/c were obtained by PCR-amplifying the coding regions of the respective genes, which were cloned into the pENTR/D-TOPO® system (Invitrogen) and subsequently recombined into pAM-PAT-GWPro35S (GenBank AY436765). The 35S::ACA8-mCherry construct was generated by Gateway recombination (Invitrogen) using pDONR201 containing the coding sequences of ACA8 devoid of stop codon (Frei dit Frey et al., 2012) and a 35S::GW-mCherry destination vector (provided by R. Panstruga, RWTH Aachen University, Germany). SlSOBIR1-HA was obtained by recombining pENTR/D-TOPO®-SlSOBIR1 (Liebrand et al., 2013) to pGWB14, carrying the HA coding sequence (Nakagawa et al., 2007). SlSERK1-Myc and SlSERK3a-Myc were obtained by recombining pENTR/D-TOPO®-SlSERK1 and pDONR201®-SlSERK3a (Liebrand et al., 2013) to pGWB20 carrying the Myc coding sequence (Nakagawa et al., 2007), respectively. pUB::AtBAK1D416N was obtained by PCR amplification from pGWB14-BAK1D416N-HA (Schwessinger et al., 2011) using primers BAK1-TOPO-F (5’-CAC CAT GGA ACG AAG ATT AAT GAT C-3’) and BAK1-TOPO-R (5’-TTA TCT TGG ACC CGA GGG GTA TTC-3’) introducing a stop codon, directional cloning into pENTR/D-TOPO® and subsequent recombination of pENTR/D-TOPO-BAK1D416N with pUB-DEST (Grefen et al., 2010) All recombinations were performed using the classical Gateway LR clonase reaction (Invitrogen). The correctness of all constructs was confirmed by sequencing.
Transient, stable transformation, and Virus-Induced Gene Silencing (VIGS)
Transient transformation of N. benthamiana epidermal cells was done as described before (Choi et al., 2013). Briefly, two-day cultures of Agrobacterium tumefaciens GV3101 carrying the respective expression constructs in liquid LB medium supplemented with antibiotics were washed in water prior to N. benthamiana leaf infiltrations. For single protein localization and co-localization purposes, final OD600=0.25 and OD600=0.5, respectively, were used for agro-infiltrations. Microscopy was performed at 2-3 dpi. N. benthamiana lines stably expressing Cf-4-GFP were obtained by incubating N. benthamiana leaf explants with Agrobacterium strain GV3101 carrying plasmid pBIN-KS-35S::Cf-4-GFP (Liebrand et al., 2012). Selection of transformed plants was done as described before (Horsch et al., 1985; Gabriëls et al., 2006). Using segregation analysis based on kanamycin resistance, a single locus insertion line was selected. Tobacco Rattle Virus (TRV)-mediated Virus-Induced Gene Silencing (VIGS) in N. benthamiana was performed as described before (Liebrand et al., 2012). Briefly, A. tumefaciens GV3101 carrying TRV-RNA1 at OD600 = 0.4 and GV3101 carrying TRV-RNA2 containing the respective target sequences at OD600 = 0.2 were mixed and co-infiltrated into N. benthamiana leaves of two-weeks old plants, and three weeks later leaves were used for further analysis. TRV alone was used as a control. The effects on Cf-4-GFP accumulation of VIGS targeting various endogenous N. benthamiana genes in combination with co-expression of A. thaliana genes were observed via immunoblotting and RT-PCR (Supporting Information Figs. 6c, 9c,d).
Bioassay for Avr4-induced immunity
TRV constructs targeting NbSERK3a/b (Chaparro-Garcia et al., 2011), TRV2::NbSOBIR1/NbSOBIR1-like (Liebrand et al., 2013) and TRV2::Cf-4 (Gabriëls et al., 2006), alongside with TRV2::GUS (Tameling & Baulcombe, 2007) controls, were agro-infiltrated into leaves of one-week-old N. benthamiana:Cf-4-GFP plants at O.D.600 = 0.5. After three weeks, Avr4 (O.D.600 = 0.03, twice) (Gabriëls et al., 2007), RxD460V (O.D.600 = 0.1) (Bendahmane et al., 2002) and BAX (O.D.600 = 0.5) (Lacomme & Santa Cruz, 1999), were agro-infiltrated into mature, fully expanded leaves of the TRV-inoculated plants. HR scores were recorded three days after agro-infiltration. In addition, three-week old NbSERK3a/b-silenced leaves were transiently transformed with Cf-4-GFP and 3 days later infiltrated with 300μM Avr4 protein. HR was observed 6 days after Avr4 infiltration. VIGS in tomato, followed by inoculation with conidia of C. fulvum race 5-pGPD:GUS and subsequent GUS staining and quantification were performed as described before (Liebrand et al., 2013).
Immunoblot analysis and co-immunoprecipitation
Immunoblot analysis with anti-GFP and anti-Myc antibodies and pull-down experiments were carried out as previously reported (Liebrand et al., 2012; Liebrand et al., 2013), with the following modifications: Transiently transformed N. benthamiana leaves were infiltrated with Avr4, Avr2, or Avr9 proteins or flg22 peptide, at the indicated concentrations, and total proteins were extracted. For detection of the HA-epitope tag the anti-HA-Biotin High Affinity Antibody (clone 3F10; Roche Applied Science) was used. In contrast to the theoretical mass of Cf-4-GFP (app. 123 kDa) Immunoblot analysis revealed a specific band at about 140 kDa in accordance with what was observed in previous studies (Liebrand et al., 2012).
qRT-PCR analysis
For qRT-PCR, total RNA was isolated from N. benthamiana:Cf-4-GFP leaf material, at 2 weeks after agro-inoculation with the various VIGS constructs. RNA extraction, cDNA synthesis and qRT-PCR were performed as described (Liebrand et al., 2012). NbSERK3a/b expression was investigated using primers rbo16 (5’-TGC GCT GAA GAC CAA CTT GGC T-3’) and rbo17 (5’-CTG AAG CTT GCC CAA TGT GTC G-3’). NbSERK1 expression was investigated using primers rbo20 (5’-ATT GCA CAG TCT GCG T-3’) and rbo21 (5’-CGA AGG AAT CTC AAT TTA GTC-3’). Expression of NbSOBIR1 was investigated using primers to266 and to267 (Liebrand et al., 2013). Expression of endogenous actin was used to calibrate the expression level of the query genes, as previously described (Liebrand et al., 2012). qRT-PCRs to determine the expression levels of AtBAK1 and AtBAK1D416N were performed using primers AtBAK1-F (5’-TGG ACT TGC AAA ACT CAT GG-3’) and AtBAK1-R (5’-GAT CAAAAG CCC TTT GTC CA-3’).
Confocal microscopy and image analysis
Confocal laser microscopy was performed using a Leica SP5 laser point scanning microscope (Leica, Germany) mounted with hybrid detectors (HyD™) as described previously (Beck et al., 2012c). Briefly, GFP and RFP/mCherry fluorophores were excited using the 488-nm argon laser and the 561 nm diode, respectively, and fluorescence emission was captured between 500 and 550 nm for GFP and between 580 and 620 nm for RFP/mCherry. For GFP only images, chloroplast autofluorescence was captured between 700 and 800 nm. Sequential scan mode was used for simultaneous imaging of GFP and RFP/mCherry. Abaxial sides of N. benthamiana leaf discs were imaged. Images were taken using a 63X water immersion objective and processed using the Leica LAS-AF and FIJI (ImageJ) software packages. Spot detection and quantification on confocal micrographs were performed using EndoQuant, which is a modification of EndomembraneQuantifier suitable for standard confocal images (Beck et al., 2012c).
Results
Cf-4 interacts with SlSOBIR1 at the plasma membrane
To address the localization and dynamics of the Cf-4 receptor in antifungal immunity, we used functional fluorescently-tagged Cf-4 (Cf-4-GFP) mediating Avr4-triggered HR in N. benthamiana (Supporting Information Fig. S1; (Liebrand et al., 2012)). We monitored Cf-4-GFP subcellular localization and revealed its presence at the plasma membrane of leaf epidermal cells by co-localizing Cf-4-GFP with the plasma membrane autoinhibitory calcium ATPase ACA8, fused to mCherry (Fig. 1, Supporting Information Fig. S2; (Frei dit Frey et al., 2012)). The cell surface localization of this Cf protein is in agreement with the described apoplastic localization of its ligand, Avr4, (Joosten et al., 1994; van den Burg et al., 2006) and the plasma membrane-localization of its constitutive interactor, SOBIR1 (Fig. 2.; (Leslie et al., 2010; Liebrand et al., 2013)). SlSOBIR1-GFP, its close homolog SlSOBIR1-like fused to GFP and also AtSOBIR1-GFP were all detected at the plasma membrane, co-localizing with ACA8-mCherry, and internal vesicles (Figs. 1, 2, Supporting Information Figs. S2, S3, S4).
To investigate the subcellular localization of the Cf-4-SlSOBIR1 complex, we used bimolecular fluorescence complementation (BiFC). We observed reconstitution of the YFP molecule by the detection of fluorescence when transiently co-expressing Cf-4 and SlSOBIR1 that were C-terminally fused to respectively the C- and N-terminal halves of YFP (respectively YFPc and YFPn), indicative of Cf-4-SlSOBIR1 heterodimerization (Fig. 1). Cf-4-SlSOBIR1 heterodimerization was found at the plasma membrane, which was confirmed by ACA8-mCherry co-localization (Fig. 1). In this assay, we did not detect positive BiFC when co-expressing FLS2 and SlSOBIR1 fused to the YFP halves, whereas BiFC did occur when co-expressing FLS2-YFPc and FLS2-YFPn (Supporting Information Fig. S5). These observations indicate homodimerization of FLS2 but no heterodimerization between FLS2 and SlSOBIR1, which is consistent with previous findings and thus supporting the specificity of BiFC in this system (Frei dit Frey et al., 2012; Sun et al., 2012; Liebrand et al., 2013). All these results show that Cf-4 localizes at the plasma membrane, where this LRR-RLP interacts with the RLK SlSOBIR1.
SOBIR1 localizes to endosomes
In addition to its localization at the plasma membrane, and in accordance to what was shown in previous studies, AtSOBIR1 was also observed at internal vesicles that showed co-labeling with the endocytic tracer FM4-64, indicative of endosomal localization (Leslie et al., 2010). To examine the identity of these SOBIR1-positive vesicles, we transiently co-expressed SlSOBIR1-GFP, SlSOBIR1-like-GFP and AtSOBIR1-GFP with described fluorescent markers of the endomembrane trafficking pathways. These include MEMB12-mCherry for labeling Golgi compartments, VHA-a1-RFP for labeling the trans-Golgi network (TGN) and the Rab5 GTPases RFP-ARA7/RabF2b and ARA6/RabF1-RFP for labeling endosomes (Geldner et al., 2009; Beck et al., 2012c). No co-localization between the SOBIR1-positive vesicles and MEMB12- and VHA-a1-labeled compartments was detected (Fig. 2, Supporting Information Fig. S3a,b). By contrast, we found clear localization of the SOBIR1 receptors to RFP-ARA7/RabF2b- and ARA6/RabF1-RFP-positive endosomes (Fig. 2, Supporting Information Fig. S3c,d). These observations are in agreement with the described FM4-64-positive endosomal localization of AtSOBIR1-YFP in Arabidopsis (Leslie et al., 2010), and suggest that constitutive endocytosis of SOBIR1 receptors takes place. Furthermore, co-localization with ARA6/RabF1 indicates that the SOBIR1 receptors might enter the late endosomal pathway (Beck et al., 2012c).
Avr4 triggers endocytosis of the Cf-4/SOBIR1 complex
The observation that Cf-4 interacts with SlSOBIR1 at the plasma membrane, together with the finding that SOBIR1 receptors are present at endosomes, prompted us to investigate whether Cf-4 is endocytosed when activated by Avr4. To test this, we treated N. benthamiana leaves transiently and stably expressing Cf-4-GFP with purified Avr4 and Avr2 proteins, of which the latter is specifically recognized by the Cf-2 receptor. When elicited with Avr4, this triggered internalization of Cf-4-GFP (Fig. 3a, Supporting Information Fig. S6). Co-expression with ARA6/RabF1-RFP showed a strong overlap of this marker with the Avr4-induced Cf-4-GFP-positive vesicles, thereby revealing endosomal localization of activated Cf-4 (Fig. 3, Supporting Information Videos S1, S2). Cf-4-As expected, GFP maintained plasma membrane localization upon control treatment with Avr2, indicating that endocytosis of Cf-4 is ligand-specific and depends on the activation of the Cf-4 receptor (Fig. 3a, Supporting Information Fig. S6). Consistent with SOBIR1 localizing to endosomes (Fig. 2, Supporting Information Fig. S3), we observed co-localization of Avr4-induced Cf-4-GFP-positive endosomes and endosomal SlSOBIR1-mCherry (Fig. 3).
As both, SOBIR1 and Cf-4 underwent endocytosis and localized at late endosomal compartments we questioned whether they might get co-internalized. Therefore we investigated whether Cf-4 is endocytosed together with SlSOBIR1. Transient co-expression of Cf-4 and SlSOBIR1, C-terminally fused to the C- and N-terminal halves of YFP, respectively, results in positive BiFC at the plasma membrane (Fig. 1) and when triggering with Avr4, BiFC was also observed at internal vesicles (Fig. 3b). These vesicles also showed co-localization with ARA6/RabF1-RFP, demonstrating endosomal localization of the reconstituted Cf-4-SlSOBIR1 BiFC heterodimer (Fig. 3b), and in agreement with Avr4-induced co-localization of Cf-4-GFP and SlSOBIR1-mCherry at endosomes (Fig. 3a). Treatment with Avr2 did not trigger re-localization of the plasma membrane-resident reconstituted Cf-4-SlSOBIR1 BiFC heterodimer (Fig. 3b), further supporting ligand-specific internalization of this Cf-4 together with SlSOBIR1. These results indicate that the receptors can be endocytosed as heterodimers, an observation that was previously made for the BRI1-BAK1 complex in Arabidopsis protoplasts (Russinova et al., 2004). In contrast to this finding, we did not observe flg22-induced endocytosis of the FLS2 homodimer as the reconstituted BiFC signal was maintained at the plasma membrane after elicitation (Supporting Information Fig. S5b). Endocytosis of activated FLS2 depends on BAK1/SERK3 (Chinchilla et al., 2007; Beck et al., 2012c), and this may indicate that the reconstituted FLS2 BiFC homodimer could be affected in complex formation with BAK1.
Cf-4 endocytosis requires BAK1/SERK3
Because flg22-activated FLS2 traffics via the ARA6/RabF1-positive late endosomal pathway and depends on functional BAK1/SERK3 (Beck et al., 2012c; Choi et al., 2013), this raises the possibility that Avr4-induced endocytosis of Cf-4 also involves BAK1/SERK3. To test this, we applied virus-induced gene silencing (VIGS), using a tobacco rattle virus (TRV) construct targeting the two NbSERK3 homologues (NbSERK3a and b) in N. benthamiana (Chaparro-Garcia et al., 2011). Monitoring Cf-4-GFP localization in leaves of TRV::NbSERK3a/b-inoculated N. benthamiana revealed a strongly reduced amount of Avr4-induced Cf-4-GFP-positive endosomes, as compared to the situation in control leaves from plants that had been inoculated with TRV only (Fig. 4, Supporting Information Fig. S10b), while Cf-4 protein levels were unaltered (Supporting Information Fig. S10c). By using a heterologous functional complementation approach, Avr4-induced endocytosis of Cf-4-GFP was partially restored when AtBAK1 was transiently expressed in NbSERK3a/b-silenced leaves. No reduction was observed in the amount of ARA6/RabF1-labelled endosomes when NbSERK3a/b was silenced (Fig. 4, Supporting Information Fig. S10b,c). Importantly, this suggests that, as is the case for the RLK FLS2, ligand-induced endocytosis of the RLP Cf-4 is BAK1/SERK3-dependent, and that this RLK plays a role in Cf-4 function. This finding is further strengthened by expression of a kinase-inactive AtBAK1 variant (AtBAK1-KD) in NbSERK3a/b-silenced leaves that did not restore Cf-4-GFP endocytosis upon triggering with Avr4 (Fig. 4, Supporting Information Fig. S10b,c).
As a next step, we investigated whether SOBIR1 is required for Avr4-induced Cf-4 endocytosis using a similar heterologous functional complementation approach. In agreement with previous findings revealing that Cf-4 abundance is dependent on SOBIR1 (Liebrand et al., 2013), Cf-4-GFP levels in N. benthamiana were increased by transient expression of AtSOBIR1-Myc and its kinase-inactive variant (Supporting Information Fig. S7c). Importantly, transient expression of AtSOBIR1-Myc but not kinase-inactive AtSOBIR1-KD-Myc restored Avr4-induced endocytosis of Cf-4-GFP in NbSOBIR1/-like-silenced leaves (Supporting Information Fig. S7a,b.). Altogether, these data indicate that the kinase activities of both BAK1 and SOBIR1 are required for endocytosis of Cf-4 upon Avr4 recognition, suggesting the active removal of triggered receptors from the plasma membrane.
Cf-4 associates with SERK members in a ligand-depending manner
We addressed a possible role of BAK1/SERK3 in Cf-4-mediated immunity by co-immunoprecipitation experiments. Cf-4-GFP was purified from N. benthamiana leaves transiently co-expressing Cf-4-GFP, SlSOBIR1-HA and SlSERK3a-Myc, that had either been mock-treated or challenged with Avr4, Avr2 or flg22 at two days after agro-infiltration of the three constructs. While C-terminally tagged BAK1/SERK3 fusion proteins are not signalling competent after flg22 and elf18 trigger during immunity their recruitment behavior to the FLS2 receptor complex is still intact, which suggests suitability for interaction studies with other receptors (Ntoukakis et al., 2011). In all cases, SlSOBIR1-HA was detected upon immunoprecipitation of Cf-4-GFP, independent of the treatment conditions (Fig. 5, Supporting Information Fig. S8). This confirms the constitutive interaction between these two receptors as shown by BiFC (Fig. 1) and as was previously reported (Liebrand et al., 2013). By contrast, SlSERK3a-Myc was only revealed upon immunoprecipitation of Cf-4-GFP from agro-infiltrated leaves that had first been infiltrated with the matching ligand Avr4, but not when infiltrated with Avr2 or flg22, which was included as control treatment (Fig. 5, Supporting Information Fig. S8). Importantly, this demonstrates that ligand-dependent hetero-complex formation of RLPs as the Cf-4 receptor with SlSERK3a takes place, in a way that is similar to what has been shown for RLK-type PRRs (Monaghan & Zipfel, 2012; Liebrand et al., 2014).
Because of the previous genetic evidence for a role of SlSERK1 in Cf-mediated immunity (Fradin et al., 2011), we examined whether Cf-4 also interacts with this co-receptor by co-immunoprecipitation experiments after transient co-expression of the various receptors in N. benthamiana leaves. Unlike what is the case for SlSERK3a-Myc, SlSERK1-Myc was revealed after immunoprecipitation of Cf-4-GFP, independent of elicitation of Cf-4 by Avr4 (Supporting Information Fig. S8). However, the amount of co-immunoprecipitated SlSERK1- Myc was strongly enhanced in the Avr4-treated leaves, as compared to the leaves treated with Avr2. We cannot exclude the possibility that this differential pattern between Cf-4 interaction with SlSERK1 and SlSERK3a is caused by the higher steady state expression levels of SlSERK1-Myc compared to those of SlSERK3a-Myc (Supporting Information Fig. S8; see input). However, following the recent notion of the presence of BRI1-BAK1/SERK3 preassembled heterodimers (Bücherl et al., 2013), our data could also indicate that Cf-4 exists in a preformed complex with SlSOBIR1 and SlSERK1, which becomes stabilized and/or recruits increased amounts of SlSERK1 upon elicitation with Avr4.
We next explored the possibility that ligand-induced recruitment of SlSERK1 and -3 could exist as a more general mechanism of Cf receptor complex activation. To this end, we purified Cf-9-GFP from N. benthamiana leaves transiently co-expressing this RLP together with SlSOBIR1-HA and either SlSERK1-Myc or SlSERK3a-Myc. Similar to what we observed for activated Cf-4, both SlSERK1-Myc and SlSERK3a-Myc were strongly co-immunoprecipitated with Cf-9-GFP when elicited with its matching ligand effector Avr9 but, as expected, not significantly with Avr4 that was included as control treatment (Supporting Information Fig. S9). This supports the idea that additional SOBIR1-dependent RLPs, such as Ve1, RLP30 and ReMax (Fradin et al., 2009; Fradin et al., 2011; Zhang et al., 2013), also recruit members of the SERK family to form active receptor complexes.
SERK members mediate Avr4-triggered immunity
Avr4 triggers HR in Cf-4-expressing N. benthamiana, which can be diminished by silencing of the gene encoding the Cf-4 receptor itself or NbSOBIR1/-like ((Liebrand et al., 2013); Fig. 6a). To determine whether BAK1/SERK3 is involved in Cf-4-mediated immunity, we examined the Avr4-triggered HR in Cf-4-GFP-expressing N. benthamiana leaves (transiently and stably), without and with silencing of NbSERK3a/b. We consistently found that the Avr4-triggered HR was significantly reduced in NbSERK3a/b-silenced Cf-4-GFP-expressing leaves, as compared to the leaves of TRV control plants (Fig. 6a, Supporting Information Fig. S10a). NbSERK3a/b-silencing did not generally affect programmed cell death, as this was still induced by the auto-active variant of the nucleotide binding-LRR immune receptor RxD460V and the pro-apoptotic factor BCL2-ASSOCIATED PROTEIN X (BAX) (Lacomme & Santa Cruz, 1999; Bendahmane et al., 2002). The capability of the NbSERK3a/b-VIGS construct to knock down the expression of both NbSERK3a/b gene homologues was previously shown (Chaparro-Garcia et al., 2011). We further investigated the specificity of NbSERK3a/b-silencing within the SERK family and found that also a likely NbSERK1 homologue was knocked down upon expression of the NbSERK3a/b-VIGS construct (Supporting Information Fig. S10b). Thus, it could be that Avr4-triggered HR in N. benthamiana requires not only BAK1/SERK3 but also SERK1, an observation that is consistent with the earlier finding that VIGS of SlSERK1 compromises Cf-4-mediated immunity to C. fulvum in tomato (Fradin et al., 2011).
To examine whether BAK1/SERK3, in addition to SERK1, is also involved in Cf-4-mediated resistance to Avr4-secreting C. fulvum strains, their encoding genes were silenced in Cf-4-expressing tomato plants. A strain secreting Avr4 and expressing the GUS reporter gene exhibited strong colonization of tomato leaves lacking the Cf-4 receptor (MM-Cf-0) and when the Cf-4 gene was silenced (Fig. 6b). Colonization by this fungus was significantly more abundant in Cf-4-expressing tomato leaves silenced for SlSERK3, as increased numbers of successful colonization attempts were observed compared to the GUS-silenced negative control. This phenotype is reminiscent of the effect of silencing of SlSOBIR1(-like) on Cf-4-mediated resistance to C. fulvum (Liebrand et al., 2013). We furthermore silenced SlSERK3 together with SlSERK1, which also compromised resistance to the pathogen. Taken together, these results show that BAK1/SERK3 is a key positive regulator of full Cf-4-mediated immunity induced upon Avr4 perception.
Discussion
The Cf signaling pathway is essential for the immune response of tomato to C. fulvum (Rivas & Thomas, 2005; Stergiopoulos & de Wit, 2009). A number of Cf signaling components were identified through genetic and proteomic approaches, but the mechanism by which Cf-4 initiates downstream signaling remained unclear (Liebrand et al., 2014). Here, we show that both SERK1 and BAK1/SERK3 are recruited to the Cf-4 receptor in an Avr4-dependent manner, a mechanism we additionally confirmed for Avr9-induced activation of Cf-9. Consistently, we show that Cf-4 requires these RLKs for its function. Our observations imply that Avr4 induces the formation of a complex of Cf-4 and SlSERK1/3 to induce Cf-4 signaling. Furthermore, we confirm that Cf-4 interacts with SOBIR1 at the plasma membrane and, more importantly demonstrate ligand-induced SERK3 dependent late endocytic trafficking of the Cf-4 RLP together with SlSOBIR1 as a novel pathway in Cf-mediated downstream events.
The requirement of BAK1/SERK3 hints at clear similarities between the Cf-4/9 effector receptors and FLS2 MAMP receptor pathways, further evidenced by overlapping transcriptional reprogramming upon activating FLS2- and Cf-mediated immune responses (Navarro et al., 2004). This overlap potentially involves the regulation of downstream responses through similar components which, in addition to the SERK family members, include E3 ligases (PUB12/13 for FLS2 and CMPG1 for Cf-4 immunity; (Gilroy et al., 2011; Lu et al., 2011)) and receptor-like cytoplasmic kinases (BIK1/PBL1 for FLS2, ACIK1 for Cf-4 and CAST AWAY for SOBIR1 (EVR) signaling; (Burr et al., 2011; Monaghan & Zipfel, 2012; Liebrand et al., 2014)). Given these similarities, it is conceivable that the constitutive association between the RLP Cf-4 and the RLK SlSOBIR1 represents a PRR complex, in which the Cf-4 ectodomain mediates specific ligand recognition and the SlSOBIR1 kinase domain is regarded as the signaling part. This is in contrast to PRRs exemplified by FLS2, in which both functions are present within the same molecule. Ligand-induced interaction of PRRs with SERK member RLKs is subsequently required to trigger downstream signaling by both RLP- and RLK-type PRRs (Fig. 7).
Regarding Cf-4 as a PRR is in agreement with the observation that Cf-4 not only recognizes the Avr4 protein secreted by C. fulvum, but is also activated by the Avr4 homologue produced by the banana pathogen Mycosphaerella fijiensis (Stergiopoulos et al., 2010). Both C. fulvum and M. fijiensis Avr4 proteins bind to chitin (van den Burg et al., 2003; Stergiopoulos et al., 2010) and it was suggested that, reminiscent of PAMP recognition, the chitin-binding motif of Avr4 is the pattern that is recognized through interaction with the LRRs of Cf-4 (Thomma et al., 2011). As elicitation with Avr4 appears to maintain the Cf-4 association with SlSOBIR1 ((Liebrand et al., 2013); Fig. 5) and induces the interaction of Cf-4 with SlSERK1/3, this could suggest the possibility that a complex consisting of Cf-4, SlSOBIR1 and SlSERK3a is formed. In turn, this could result in increased phosphorylation events between the SlSOBIR1 and SlSERK1/3 kinases, which then trigger downstream signaling (Fig. 7), similar to what has been proposed for the FLS2-BAK1 model (Schwessinger et al., 2011). Furthermore, our data show that SlSERK1 and BAK1/SlSERK3 are also recruited to the Cf-9 receptor upon its activation with Avr9, supporting the idea of a general mechanism by which SOBIR1-dependent RLPs form active receptor complexes. Although no molecular interactions have been reported so far, a similar scenario could be proposed for RLP30, which also genetically depends on SOBIR1 and BAK1/SERK3 (Zhang et al., 2013). A role for SERK receptors is less clear for SlEix2, because in this case BAK1 was found to negatively regulate this receptor through interaction with its close homologue SlEix1 (Bar et al., 2010). Future studies should determine whether, similar to the trans-phosphorylation of FLS2 by BAK1/SERK3 (Schwessinger et al., 2011), the phosphorylation status of the kinase domain of SOBIR1 also alters upon SERK1/3 recruitment.
Receptor-mediated endocytosis is part of the eukaryotic immune response and for example is found for the FLS2 receptor (Husebye et al., 2006; Robatzek et al., 2006; Spallek et al., 2013). An important role of receptor-mediated endocytosis is to control the abundance of receptor (complexes) at the plasma membrane, a process that is well established in animals and involves lysosomal/vacuolar degradation (Lemmon & Schlessinger, 2010). Activated FLS2 traffics into the late endosomal pathway and is a cargo of multivesicular bodies localizing to the lumen of these late endosomes for delivery to the vacuole (Beck et al., 2012c; Spallek et al., 2013). This pathway could be responsible for the flg22-induced FLS2 degradation, because chemicals affecting endosomal trafficking inhibit the degradation of this receptor (Smith et al., 2014). Our co-localization data strongly suggest that activated Cf-4 is also targeted for vacuolar degradation through the late endosomal pathway, consistent with the observation that Cf-4-GFP protein levels are reduced upon Avr4 elicitation (Supporting Information Fig. S1a). Endosomal sorting for vacuolar degradation requires the transfer of ubiquitin to the plasma membrane cargo and subsequent deubiquitination of the cargo at multivesicular bodies (Beck et al., 2012c). The ubiquitin E3 ligase CMPG1 and the deubiquitinating enzyme UBP12 are known positive and negative regulators of Cf-4- and Cf-9-mediated HR, respectively (Ewan et al., 2011; Gilroy et al., 2011), and given their biochemical function could also be involved in the endocytosis and degradation of Cf proteins.
Ligand-induced activation triggers the formation of a hetero-complex consisting of Cf-4 and SlSERK1/3. This differs from the situation with TLRs from mammals, which recruit cytoplasmic kinases to initiate signaling, but in plants is reminiscent of the FLS2 pathway that also depends on interaction with the co-receptor BAK1/SERK3 to signal. Endocytic removal of activated PRRs, which is also known for TLR4 (Husebye et al., 2006), might provide a mechanism to regulate receptor presence at the plasma membrane and associated events (Felix et al., 1998; Beck et al., 2012a; Smith et al., 2014). Understanding how plant cells regulate PRR subcellular localization is essential as pathogens are likely to target components of the trafficking pathway to suppress plant defenses, as was recently found for the Phytophthora infestans Avr3a effector (Chaparro-Garcia et al.) and bacterial HopM1 (Nomura et al., 2011). This latter effector of Pseudomonas syringae targets a host ADP ribosylation factor guanine nucleotide exchange factor, referred to as AtMIN7/BEN1, resulting in its breakdown. AtMIN7/BEN1 is required for PAMP-triggered immunity and regulates endosomal trafficking possibly at the TGN, where HopM1 can also be found (Nomura et al., 2006; Nomura et al., 2011). It remains to be seen whether C. fulvum also secretes effectors that enter inside plant cells and are able to alter host subcellular trafficking to promote infection success.
Acknowledgements
We thank the members of the Robatzek laboratory for fruitful discussions, Brande Wulff for reading the manuscript and Cyril Zipfel for providing AtBAK1 constructs. Zeinu Mussa is acknowledged for help with N. benthamiana transformations.