Tomato receptor FLAGELLIN-SENSING 3 binds flgII-28 and activates the plant immune system

Plants and animals detect the presence of potential pathogens through the perception of conserved microbial patterns by cell surface receptors. Certain solanaceous plants, including tomato, potato and pepper, detect flgII-28, a region of bacterial flagellin that is distinct from that perceived by the well-characterized FLAGELLIN-SENSING 2 receptor. Here we identify and characterize the receptor responsible for this recognition in tomato, called FLAGELLIN-SENSING 3. This receptor binds flgII-28 and enhances immune responses leading to a reduction in bacterial colonization of leaf tissues. Further characterization of FLS3 and its signalling pathway could provide new insights into the plant immune system and transfer of the receptor to other crop plants offers the potential of enhancing resistance to bacterial pathogens that have evolved to evade FLS2-mediated immunity. FLS2 is the well-known plasma membrane receptor for flg22, a specific region of bacterial flagellin. But Solanaceae can also detect flagellin through another epitope, flgII-28, thanks to the novel receptor-like kinase FLS3 now identified in tomato.

T he recognition of conserved microbe-associated molecular patterns (MAMPs) by pattern recognition receptors (PRRs) is one of the initial events that activates pattern-triggered immunity (PTI) in both plants and animals [1][2][3][4] . This immune response leads to the rapid generation of reactive oxygen species (ROS), activation of mitogen-activated protein kinases (MAPKs) and extensive changes in the transcriptome that together hinder the infection process 1,4-6 . The first plant PRR-MAMP pair, consisting of FLAGELLIN-SENSING 2 (FLS2) and its ligand the flagellin epitope flg22, was identified 15 years ago, and works in concert with the co-receptor BAK1 to activate intracellular immune signalling [7][8][9][10] . Since then, approximately ten additional receptor-ligand pairs involved in immunity, either through perception of MAMPs or damage-associated molecular patterns (DAMPs), have been identified, with direct binding being demonstrated for only a subset of these pairs 2 .
Recently, a subset of solanaceous species, including tomato, potato and pepper, but not Nicotiana spp., was found to recognize a second epitope of flagellin termed flgII- 28 (refs 11,12). FlgII-28 perception occurs independently of FLS2 (ref. 12), but the molecular basis of its recognition is unknown. The discovery that tomato recognizes a second flagellin MAMP, combined with extensive natural variation and recent availability of the genome sequence for this species, offered the opportunity to identify the flgII-28 receptor using a genetic approach. Here, we use natural variation in tomato heirloom varieties and a mapping-by-sequencing approach to identify a receptor-like kinase gene, named FLAGELLIN-SENSING 3 (FLS3), which confers responsiveness to flgII-28. We demonstrate that FLS3 is the flgII-28 receptor and show that FLS3-mediated immunity enhances resistance to a bacterial pathogen. FLS3 represents an orthogonal means for flagellin perception, and therefore expression of this solanaceousspecific PRR in crop plants that are normally unable to detect flgII-28 could be used to combat pathogens that have evolved to evade or subvert flg22 detection.

Results
We have previously reported natural variation for perception of flagellin epitopes among tomato heirloom varieties 13 . Further screening of ∼100 accessions using an assay to detect ROS production identified eight varieties and two tomato wild species accessions with a strongly reduced response to flgII-28 (Fig. 1a,b and Supplementary Fig. 1a-d). To identify the responsible gene by map-based cloning, segregating populations were generated by crossing accessions that are flgII-28 sensitive (LA1589 and Rio Grande) or insensitive (Yellow Pear, Matt's Wild Cherry and Galapagos). The resulting F1 plants were responsive to flgII-28 ( Supplementary Fig. 1c,d), indicating the allele responsible for the sensitivity is dominant. Testing of F2 plants with the ROS assay revealed a segregation ratio of 3:1 (sensitive/insensitive) in two of the three populations ( Supplementary Fig. 1e), indicating that flgII-28 sensitivity can be conferred by a simply inherited locus.
To identify the genomic region linked to flgII-28 sensitivity, we generated DNA libraries for next-generation Illumina sequencing using flgII-28 non-responsive F2 plants from the LA1589 × Yellow Pear cross because genome sequences were available for these lines 14,15 . Analysis of the sequencing data showed that only chromosome 4 had a notable deviation from the expected 1:1 LA1589: Yellow Pear single nucleotide polymorphism (SNP) ratio ( Supplementary Fig. 2), with one region in particular having very few LA1589-specific SNPs. This region, spanning 2.619-5.486 Mb from the end of the chromosome, contains 322 annotated genes including nine leucine-rich repeat, receptor-like kinases (LRR-RLKs).
In parallel, we analysed SNP data generated independently from 75 tomato cultivars 16 . Relationships among the varieties based on genome-wide SNPs revealed no similarity among three known flgII-28-insensitive cultivars (Yellow Pear, Gold Ball Livingston and San Marzano). A separate analysis using only SNPs between 1 and 10 Mb on chromosome 4 identified a close relationship of the insensitive cultivars ( Supplementary Fig. 3). These analyses further supported this region as the location of the flgII-28 sensitivity locus.
We next performed fine mapping using DNA markers (Supplementary Table 1) and succeeded in delimiting a <0.6 Mb region that co-segregated with flgII-28 sensitivity; this region contained one receptor-like kinase gene (Solyc04g009640), which we tentatively designated FLAGELLIN-SENSING 3 (FLS3) (Fig. 1c). We confirmed that this region was linked to the sensitive phenotype observed in our other two segregating F2 populations ( Supplementary Fig. 1f ). Analysis of FLS3 in non-responding accessions identified two alleles that were different from the allele in the flgII-28-responsive accessions Heinz1706, Rio Grande and LA1589. Remarkably, one of these alleles, fls3-1, was present in eight insensitive tomato cultivars and one S. pimpinellifolium accession; this allele has a single nucleotide deletion that causes an aberrant stop codon (Fig. 1d, Supplementary Table 2 and Supplementary Fig. 4a). The other allele, fls3-2, was found in one accession of S. pimpinellifolium, LA0373, and encodes a full-length protein with four amino acid changes (Fig. 1d, Supplementary Table 2 and Supplementary Fig. 4a).
The expression of many PRR-encoding genes is induced by MAMPs 17 . Using available RNA sequencing data (Rosli et al. 6 and the Tomato Functional Genomics Database, http://ted.bti.cornell. edu/), we found that expression of FLS3 is induced by flg22 and flgII-28 treatment similar to that observed for FLS2.1 (Supplementary Fig. 1g). The tomato bacterial pathogen Pseudomonas syringae pv. tomato DC3000 (Pst DC3000) has two well-characterized effector proteins, AvrPto and AvrPtoB, which are known to suppress PTI-related immune responses, including the induction of PTIrelated gene expression 18 . Similar to FLS2.1, transcript abundance of FLS3 increased after inoculation with a Pst DC3000 strain that lacks avrPto and avrPtoB (ΔavrPtoΔavrPtoB), and this increase was inhibited by Pst DC3000, indicative of effector suppression ( Supplementary Fig. 1g). Thus, FLS3 belongs to a subset of tomato genes referred to as FIRE genes (Flagellin-induced, repressed by effectors) 6 .
The FLS3 gene encodes a class XII RLK (ref. 19) with 27 LRRs, although it lies in a sub-clade of this class that is distinct from EF-Tu receptor (EFR), FLS2 and XA21 (ref. 19). Typical of many immune receptors 20 , FLS3 has a non-arginine-aspartate (RD) intracellular kinase domain ( Supplementary Fig. 4a). We identified potential FLS3 orthologues from sequenced accessions of potato and pepper, but not from Nicotiana benthamiana or petunia ( Supplementary Fig. 4b). Certain varieties of pepper and potato were previously shown to be sensitive to flgII-28 (ref. 12) and we found the sequenced accessions were also sensitive, whereas N. benthamiana and petunia are not (data not shown and Supplementary Fig. 4c,d). These observations suggest that the FLS3 gene likely arose, possibly by duplication of a related gene, after the divergence of Capsicum and Solanum from other solanaceous species.
The strong selection for weakly immunogenic flgII-28 alleles in Pst field populations, which originally led to the identification of this MAMP (ref. 11), is compelling evidence of its importance in natural plant-bacterial interactions. To investigate whether flgII-28 plays a role in tomato resistance to Pseudomonas strains under controlled laboratory conditions, we utilized the pathogen Pseudomonas cannabina pv. alisalensis ES4326 (Pcal ES4326; formerly P.s. maculicola 21 ), which has a flg22 sequence that is not recognized by tomato 12 but a flgII-28 which is recognized (Fig. 2a, b). We used this pathogen because it allowed us to specifically test the contribution of flgII-28 perception to plant immunity independent of flg22 recognition by FLS2. These experiments were performed with the LA1589 × Yellow Pear F2 population used previously to map FLS3 (Fig. 1), which segregates for FLS3 (FLS3/FLS3, FLS3/fls3-1, fls3-1/fls3-1). Pcal ES4326 grew twofold more and caused more disease symptoms in fls3-1/fls3-1 plants than in FLS3/FLS3 or FLS3/fls3-1 plants ( Supplementary Fig. 5a,b). As this bacterial growth difference in the F2 plants was modest, we wanted to confirm that this difference was due to flgII-28 recognition. Therefore we developed a PTI induction assay that used heat-killed Pst strains expressing FliC variants from either Pst DC3000 (in which both flg22 and flgII-28 are active) or Pcal ES4326 (in which only flgII-28 is active) as a source of flagellin. We first induced PTI by infiltrating this solution into plants and then challenged them 16 hours later with a virulent bacterial pathogen, either a Pst strain lacking flagellin or Pcal ES4326. Initially we used LA1589 and Yellow Pear to test this system and observed that the plants pretreated with heat-killed DC3000 fliC had significantly less bacterial growth and symptom production than those pretreated with the empty vector control ( Fig. 2c and Supplementary Fig. 5c), which is likely to be due to PTI induction by FLS2 recognition of flg22. However, for plants pretreated with heat-killed ES4326 fliC, only LA1589 plants had significantly less bacterial growth than the empty vector control ( Fig. 2c and Supplementary Fig. 5c), suggesting that since Yellow Pear lacks FLS3, it does not recognize the flgII-28 epitope of the ES4326 flagellin protein. Subsequently, we tested the F2 plants segregating for FLS3 and fls3-1, and observed that pretreatment using heat-killed ES4326 fliC led to significant differences in bacterial growth depending on the genotype of the plants. Plants that lacked FLS3 ( fls3-1/fls3-1) showed higher bacterial growth and more severe symptoms similar to the Yellow Pear parent than to F2 plants that had at least one copy of FLS3 (FLS3/FLS3 or FLS3/fls3-1) or to LA1589 (Fig. 2d,e). These differences were not observed when the plants were pretreated with heat-killed DC3000 fliC ( Supplementary Fig. 5d,e), suggesting that the observed bacterial virulence differences are attributable to PTI induction by FLS3 perception of flgII-28. All together these observations indicate that FLS3 is associated with modestly enhanced resistance to bacterial infection, although because we have used F2 plants we cannot rule out that another locus closely linked to FLS3 could also function in flgII-28 recognition.
To determine if ectopic expression of FLS3 is able to confer flgII-28 sensitivity, we transfected Yellow Pear protoplasts with an FLS3 construct and tested for phosphorylation of MAPK proteins, which occurs in the responsive Rio Grande cultivar on flgII-28 treatment (data not shown). Expression of FLS3 resulted in an increase in phosphorylated MAPKs specifically after flgII-28 treatment, similar to levels observed either with the control treatment flg22, or with expression of the unrelated PRR EFR and treatment with its cognate ligand elf18 (Fig. 3a). Treatment of FLS3-expressing protoplasts with elf18 did not cause MAPK activation, indicating the response to flgII-28 treatment was specific. To confirm the protoplast results, we transiently expressed FLS3 in normally insensitive N. benthamiana leaves followed by treatment with flgII-28 and observed the production of ROS, similar to that observed with EFR and elf18 treatment ( Supplementary Fig. 6a,b). As a control, we showed that treatment with flgII-28 in leaves expressing yellow  fluorescent protein (YFP) did not induce ROS production compared with the water controls ( Supplementary Fig. 6a).
To gain insight into whether key residues in the kinase domain of FLS3 are important for the flgII-28-mediated response, we generated several FLS3 kinase domain variants, transiently expressed them in N. benthamiana leaves, and measured ROS production after treatment with different concentrations of flgII-28. FLS3 (K877Q) has a glutamine substituted for the conserved lysine in the ATP binding site (Supplementary Fig. 4a) which is responsible for phosphotransfer 22 and is known to be required for downstream signalling events in other plant PRRs 5,23 . Unlike wild-type FLS3, expression of FLS3(K877Q) did not lead to production of ROS after flgII-28 treatment but instead was similar to the background levels observed in the YFP control (Fig. 3b,d), indicating a loss of FLS3 responsiveness to flgII-28 treatment. Next, we examined the fls3-2 allele present in LA0373 that has a proline substitution for a conserved threonine residue in the kinase P + 1 loop region ( Supplementary Fig. 4a), which we hypothesized might be responsible for the reduced response of FLS3-2 to flgII-28 treatment (Fig. 3c,d and Supplementary Fig. 1b). To test the importance of this residue for FLS3 activity, we generated two different FLS3 variants. FLS3(T1011P) has the 'wild-type' threonine residue changed to a proline, and this variant showed reduced ROS production compared with FLS3 after flgII-28 treatment (Fig. 3b,d), and FLS3-2 (P1011T), which has the proline residue changed back to the 'wild-type' threonine, had increased ROS production compared with FLS3-2 (Fig. 3c,d). FLS3-2 and the two P + 1 loop variants responded more strongly to higher concentrations of flgII-28 (Fig. 3d). These results indicate the importance of certain key residues in the kinase domain for the response of FLS3 to flgII-28; however, it is important to note that they are from transient Agrobacterium-mediated expression and may not be reflective of wild-type FLS3 sensitivity (in Rio Grande or LA1589 plants) where the gene is expressed in the proper genomic context and under its native promoter. We observed similar accumulation of all FLS3 proteins ( Supplementary Fig. 6d-f ). Although these results suggest that kinase activity is important for FLS3 signalling, so far we have been unable to detect in vitro or in vivo kinase activity for FLS3 (similar to a recent report for FLS2 activity 23 ), and so it is presently unknown whether these substitutions affect enzymatic properties or alter interactions with signalling components independent of kinase activity.
To investigate if flgII-28 directly and specifically binds FLS3, we developed a photo-affinity labelling strategy similar to that used to demonstrate direct binding of brassinosteroids to the BRI1 receptor 24 . Synthetic samples of the flgII-28 and flg22 peptides were converted into photo-affinity probes flgII-28* and flg22*, respectively, by means of selective addition of a bifunctional chemical tag to the amino (N)-terminal amine of each peptide. This chemical tag includes a methyl trifluorodiazirine photo-crosslinker moiety, enabling UV irradiation-triggered covalent attachment of the peptide probe to its cognate PRR, in addition to an alkyne handle, permitting installation of a biotin reporter tag using 'click chemistry' (Fig. 4a). Importantly, the modified peptide probes, flgII-28* and flg22*, retained the ability to elicit an immune response ( Supplementary Fig. 7a-e). Next, we treated purified plasma membrane preparations from N. benthamiana leaves expressing FLS3-GFP with flgII-28* or flg22* and subsequently UV-irradiated them for photo-crosslinking ( Supplementary Fig. 7f). FLS3-GFP was then immunoprecipitated using the GFP tag and subsequently biotinylated using click chemistry. Only plasma membranes treated with flgII-28*, but not those treated with flg22*, showed FLS3-GFP biotinylation (Fig. 4b), demonstrating the affinity of FLS3 for flgII-28*. In parallel, we performed the same experiment using FLS2-GFP, and only observed FLS2-GFP biotinylation when samples were treated with flg22* but not with flgII-28*. In addition to demonstrating that FLS3 is the receptor for flgII-28, these data establish the specificity of peptide-probe binding and UV-crosslinking between the ligands and their cognate receptors.
To determine the specific affinity of FLS3 for flgII-28, a series of flgII-28* concentrations was used for binding and crosslinking; FLS3-flgII-28* complexes were clearly detected at low nanomolar concentrations of flgII-28* (Fig. 4c). Binding could also be weakly observed at sub-nanomolar concentrations, but only when the blots were exposed for a long period (1 h compared to less than 10 min; for comparison the exposures for the anti-GFP blots to demonstrate equal loading were equivalent at 1 min each between the left and right panels of Fig. 4c). We performed similar experiments for flg22* binding to FLS2, and observed binding consistently at 50 nM flg22* (Fig. 4d). This concentration is higher than the EC 50 value of ∼0.03 nM that was reported previously 25 . This could be due to the propensity of flg22 to stick to surfaces 26 ; during purification of flg22*, we observed that addition of the hydrophobic crosslinking moiety further increased this tendency to surface deposit and aggregate.
We observed that simultaneous treatment of flgII-28* with a large excess (40-fold) of unmodified flgII-28 eliminated biotinylation of FLS3-GFP, indicating that these two peptides compete for the same binding site (Fig. 4e). When decreasing concentrations (20-fold to 0.4-fold) of unmodified flgII-28 were added, the amount of biotinylated FLS3-GFP increased (Fig. 4e). As expected, treatment with flgII-28* and a 40-fold excess of flg22 did not prevent biotinylation of FLS3-GFP, indicating that flg22 does not compete with flgII-28* in binding to FLS3-GFP (Fig. 4e). Collectively, these results show that FLS3 binds directly and specifically to flgII-28 and represents a bona fide receptor of this ligand.
To gain further insight into FLS3 signalling, we investigated whether the co-receptor BAK1 (also known as NbSERK3 in N. benthamiana 10,27 ) was necessary for flgII-28 responsiveness as it is for both FLS2 and EFR signalling 9,10,28 . Since we were unable to detect FLS3 protein accumulation in either transfected Arabidopsis protoplasts or stably transformed Arabidopsis plants (data not shown), we instead used N. benthamiana plants knocked down for BAK1 expression by virus-induced gene silencing (VIGS). Silencing of BAK1 was confirmed by quantitative PCR with reverse transcription (RT-PCR) (Supplementary Fig. 8a) and we observed reduced ROS production upon flgII-28 treatment in BAK1-silenced leaves expressing FLS3 compared with control-silenced plants (Fig. 5a), although FLS3 protein accumulation was comparable ( Supplementary Fig. 8b).
Overexpression of Arabidopsis BAK1 along with FLS3 in BAK1silenced plants increased ROS production upon flgII-28 treatment (Supplementary Fig. 8c). We next investigated whether FLS3 and BAK1 could physically associate in plant cells, and whether this interaction was ligand dependent, as is the case for the FLS2-BAK1 interaction 29 . Epitope-tagged versions of each protein were co-expressed in N. benthamiana leaves using agroinfiltration; the infiltrated areas were treated either with solutions of 1 µM flgII-28 or flg22 peptides or buffer alone before harvesting, and the proteins were immunoprecipitated for analysis by immunoblotting. FLS3 co-immunoprecipitated with BAK1 specifically in the presence of flgII-28, similar to FLS2 but a small amount of flg22-independent interaction between FLS2 and BAK1 could be observed, possibly because of the better expression of FLS2 protein in the samples (Fig. 5b and Supplementary Fig. 8d). This interaction was specific, as the YFP control could not pull down FLS3 (Supplementary Fig. 8e). Collectively, these data indicate that BAK1 is required for the FLS3-dependent response to flgII-28, though it is possible that additional host proteins may also contribute to the response to this MAMP.
Finally, to reveal possible connections with the known PTI suppression activities of pathogen effectors, we investigated whether the Pst effector proteins AvrPto and AvrPtoB, which suppress FLS2-and EFR-mediated signalling 18 , also suppress FLS3 signalling. Co-expression of AvrPto with FLS3 in N. benthamiana leaves caused a reduction in ROS production after flgII-28 treatment (Fig. 5c). Similar results were obtained with co-expression of AvrPtoB 1-387 , which lacks the effector E3 ligase domain 30 (Fig. 5d); accumulation of FLS3 protein was not substantially altered in the presence of either effector ( Supplementary Fig. 8f,g). These results demonstrate that the effectors target FLS3 signalling, although the specific mechanisms of suppression remain to be investigated.

Discussion
The discovery that FLS3 acts in addition to FLS2 to detect flagellin represents a pioneering example in plants where two receptors have been identified which recognize different MAMPs within the same pathogen protein 11,12 . However, this phenomenon has been reported in mammals, where extracellular TLR5 and intracellular NAIP5/6 receptors perceive different epitopes of flagellin 31,32 . The presence of multiple MAMPs within the same microbial feature may not be unique to flagellin, because it was recently reported that an epitope of EF-Tu in a region distinct from elf18, called EFa50, is able to induce PTI responses in rice 33 ; however, the receptor for the second EF-Tu MAMP is unknown. FLS3 and FLS2 belong to divergent sub-clades of class XII RLKs 19 and it is possible that, despite their mutual dependence on BAK1, the two receptors act with some different host components to promote PTI. This possibility might explain why FLS3 causes a more sustained production of ROS than does FLS2 in response to their respective ligands (Fig. 1a). It is unknown to what extent FLS3 and FLS2 might contribute additively or redundantly to the host response to flagellin-derived MAMPs. This question is addressable by generating in the same genetic background single and double mutants in the receptor genes using CRISPR technology. Given the difference between flg22 and flgII-28 and in the LRR domains of the receptors, it seems likely that FLS3 binds flgII-28 in a manner distinct from FLS2. Future comparisons of FLS3 and FLS2 have the potential to reveal new insights into the evolution, structural biology and mechanisms underlying PTI in tomato.
There are several mechanisms by which bacteria evade recognition of their flagellin. One tactic deployed by several Pseudomonas spp. involves degradation of excess flagellin monomers by an alkaline protease secreted from the bacteria 34,35 . Another more broadly employed strategy is the attenuation of flagellin recognition through the alteration of MAMPs important for recognition by a host receptor 36 . There are now several reports of the transfer of PRRs from one plant species to another conferring broadened resistance to pathogens [37][38][39][40][41][42][43][44] . Therefore, it is possible that FLS3 could be used in the development of plants which have resistance against bacterial pathogens that have evolved to evade recognition of their flg22 region; Pcal ES4326 is one example of such a bacterium. In addition, some heirloom tomato varieties are known to generate very high levels of ROS in response to flgII-28 (ref. 13) and it is possible that such genetic variation might be useful in the breeding of cultivars with enhanced resistance to bacterial pathogens.

Methods
Plant materials and growth conditions. Seeds of S. lycopersicum 'Yellow Pear', S. pimpinellifolium accession LA1589, F1 hybrids and F2s were provided by Esther van der Knapp at the University of Georgia. Seeds of S. lycopersicum 'Heinz 1706' were provided by James Giovannoni at the Boyce Thompson Institute. Other accessions of S. pimpinellifolium were obtained from the Tomato Genetics Resource Center (http://tgrc.ucdavis.edu/). Seeds of S. lycopersicum cv. 'Matt's Wild Cherry' and 'Galapagos' used for generating the F2-segregating populations were obtained from Good Mind Seeds (http://goodmindseeds.org/). Tomato and N. benthamiana plants were grown as previously described 6,45 .
Oxidative burst bioassay. The production of ROS was detected using a luminolbased assay as previously described 12 . Measurements were taken every 2 min for 32 min, and the average ROS production for each plant was the mean of three or four leaf discs. Total ROS production was determined by summing the average relative light unit (RLU) values for the time points between 0 and 32 min after treatment.
Virulence assays in tomato. Pseudomonas cannabina pv. alisalensis ES4326 (formerly P. syringae pv. maculicola 12,21 ) and Pseudomonas syringae pv. tomato DC3000ΔavrPtoΔavrPtoBΔhopQ1-1ΔfliC strains were grown on King's B solid medium, and bacterial suspensions were prepared in 10 mM MgCl 2 and 0.02% Silwet L-77. Plants were vacuum infiltrated as previously described 48 . The PTI induction assays used Pst DC3000ΔavrPtoΔavrPtoBΔhopQ1-1ΔfliC strains that were transformed with constructs that allowed for expression of fliC variants from either Pst DC3000 or Pcal ES4326 (ref. 12). Bacterial suspensions of 10 8 c.f.u. ml -1 were boiled for 5 min to kill the bacteria, and the resulting solutions were used for vacuum infiltration to induce PTI. After 16 h, plants were vacuum infiltrated with either Pst DC3000ΔavrPtoΔavrPtoBΔhopQ1-1ΔfliC or Pcal ES4326 at 5 × 10 4 or 1 × 10 5 c.f.u. ml -1 . Inoculated plants were kept in a growth chamber until sampled to determine bacterial populations two or three days after infiltration.
Plasma membrane enrichment, binding and photo-crosslinking with MAMP probes and biotinylation using click chemistry. Membrane protein enrichment was performed according to Broghammer et al. 49 with the following modifications. Frozen N. benthamiana leaves were homogenized in extraction buffer (30 g fresh weight tissue in 200 ml buffer) consisting of 50 mM MOPS-KOH, pH 7.5, 500 mM D-sorbitol, 5 mM DTT, 5 mM EDTA, 1% polyvinylpyrrolidone (PVP), and 1 mM phenylmethylsulphonyl fluoride (PMSF). After miracloth filtration and initial centrifugation, the homogenate was centrifuged at 100,000g for 75 min at 4°C. The resulting microsomal pellet was suspended in 25 mM Tris-HCl, pH 7.5, 250 mM sucrose, 10 mM potassium phosphate, pH 7.5 and 28.8 mM NaCl. Using aqueous two-phase partitioning, plasma membrane-enriched microsomes were purified using a bulk phase with 6% Dextran M r 450,000-650,000 (Sigma-Aldrich) and 6% polyethylene glycol 3350 (Sigma-Aldrich). The upper phase was centrifuged at 100,000g for 2 h at 4°C, and the plasma membrane-enriched microsome pellet was suspended in 2 ml of cold binding buffer consisting of 25 mM MES, pH 6.0, 3 mM MgCl 2 and 10 mM NaCl, and the sample was equally divided before the addition of peptides. Plasma membrane-enriched microsomes were incubated with peptides for 15 min in the dark at 4°C to allow for binding to occur before irradiating for 15 min using a UV lamp (Blak-Ray B-100AP 100-watt lamp, UVP Ultraviolet Products) at a working distance of 2.5 cm. The plasma membrane-enriched microsomes were solubilized in binding buffer containing 1% Triton X-100 and 0.1% SDS (v/v). GFP-tagged receptors were immunoprecipitated overnight at 4°C using 10 µl of GFP-Trap_A slurry per sample (ChromoTek GmbH). The resin was washed twice with 1× PBS pH 7.4 and twice with radioactivity immune precipitation assay (RIPA) buffer consisting of 1× PBS pH 7.4, 1% Triton X-100, 0.5% sodium deoxycholate and 0.1% SDS. Next, the resin was suspended in click chemistry buffer consisting of RIPA buffer, 500 µM BTTAA, 250 µM CuSO 4 pentahydrate, 2 mM sodium ascorbate and 100 µM azide-PEG4-biotin conjugate (Click Chemistry Tools) in a total volume of 250 µl and placed on a rotatory shaker at 4°C for 2-6 h. The resin was washed as described above before the immunoprecipitated material was eluted by boiling for 5 min in 3× Laemmli sample buffer.
Sequences were submitted to NCBI as project number PRJNA263381. Data and output from this study can be accessed through the Solgenomics ftp site: ftp://ftp.solgenomics.net/.