EFR allosterically activates BAK1.

The kinase domains of coRK BAK1 and the RKs BRI1, EFR and FLS2 were tagged with RiD domains and purified from E. coli λPP cells. A) Recombinantly expressed RiD-tagged kinase domains were mixed together at equimolar ratios (2 µM), with or without addition of 10 µM Rap as well as 10 mM MgCl2 and 1 mM AMP-PNP. B) RK and coRK were mixed at an equimolar ratio at 50 nM, kinase-dead BIK1D202N substrate was added at 500 nM. Reactions were carried out at RT for 10 min with 0.5 μCi [γ-³²P]ATP, 100 μM ATP and 2.5 mM each of MgCl2 and MnCl2. Addition of 1 μM Rap enhanced transphosphorylation of BIK1 by EFR and BRI1 but not by FLS2. Kinase-dead BRI1 and FLS2 failed to enhance BIK1 transphosphorylation, but kinase-dead EFR retained some ability to do so. A similar trend was observed for (auto)phosphorylation of BAK1 itself.C) Quantification of band intensities over three independent experiments of which a representative is shown in B.

EFRY836F and EFRSSAA impair the active kinase conformation, which is required for signaling function.

A, (left) HDX-MS results for unphosphorylated EFR and EFRY836F protein. The difference in percent H/D exchange in wild type EFR and EFRY836F is expressed as the Δ%EX (wild type EFR – EFRY836F), with the positive and negative Δ%EX indicating more stabilized and destabilized regions in EFRY836F, respectively, compared to wild-type EFR. The Δ%EX values at different labeling time points are shown as colored lines, as indicated in the figure. The horizontal dotted black lines indicate the 98% confidence interval for the Δ%EX data (±7.18%, corresponding to ±0.4 Da difference between wild type and Y836F percent exchange) calculated as described previously (Houde et al., 2011). Regions with Δ%EX values that exceed this confidence limit are indicated as colored bars in the figure, including the β3-αC loop (orange), the catalytic loop plus part of αE (purple), and the A-loop (blue). These regions are colored in the AlphaFold2-derived model of the EFR kinase domain shown at right, in which Y836 is shown as a purple sphere and labeled. All data are the average of three independent biological repeats (n=3) with three technical repeat experiments each. A summary of the HDX-MS analysis is presented in Table 2. B,C) Secondary site mutation EFR F761[H/M] partially restores function of EFRY836F (B) and EFRSSAA (C). Full length EFR and its variants were expressed transiently in N. benthamiana and their function was tested in an oxidative burst assay. EFR F761H partially restored oxidative bursts of EFRY836F and EFRSSAA. Outliers are in indicated by asterisk in addition to the outlier itself and are included in statistical analysis; Statistical test: Kruskal-Wallis test (p < 2.2*10-16 in B, p = 1.163*10-7 in C), Dunn’s post-hoc test with Benjamin-Hochberg correction (p <= 0.05) Groups with like letter designations are not statistically different.

Homology-based design of putative intragenic suppressor mutations for EFR.

The list contains the residue number of EFR and the analogous oncogenic mutation in BRAF, as well as a short description of the mode of action of the oncogenic mutation. See Figure 2 – Supplement 1B for structural locations.

EFRF761H/Y836F and EFRF761H/SSAA recover receptor complex activation.

A) In infection assays, GUS activity was high in the positive control efr-1 line. GUS activity level was reduced in the EFRWT and EFRF761H complementation lines, but much less so in the EFRY836F and EFRSSAA complementation lines. By contrast, EFRF761H/Y836F and EFRF761H/SSAA complementation lines displayed substantially repressed GUS activity. Each experiment was repeated three times with similar results. Outliers are indicated by an additional asterisk and included in statistical analysis. Statistical test: Kruskal-Wallis test (p = 5.704*10-7), Dunn’s post-hoc test with Benjamin-Hochberg correction (p <= 0.05) Groups with like letter designations are not statistically different. B) In IP kinase assays, ligand-induced interaction of EFRWT and EFRF761H with BAK1 increased transphosphorylation of BIK1D202N, but this was abolished for EFRY836F and EFRSSAA. Both EFRF761H/Y836F and EFRF761H/SSAA showed partially restored BIK1D202N trans-phosphorylation as well as BAK1 S612 phosphorylation (across four replicates for EFRF761H/SSAA and in two out of four replicates for EFRF761H/Y836F). Samples were also probed for MAPK phosphorylation for effective ligand treatment. Treatment: 100 nM elf18 for 10 min. C) Quantification of BIK1D202N band intensity observed in autoradiographs from the four independent replicates performed. Dotted red line indicates unchanged band intensity in mock vs. elf18 treatment.

EFRF761H recovers BAK1Y403F function.

The cytoplasmic domains of BAK1 and EFR variants with fused RiD-tags were transiently expressed in N. benthamiana and leaf discs were treated with Rap to induce dimerization. EFR and EFRF761H induced a similar total oxidative burst when BAK1 was co-expressed. The co-expression of BAK1Y403F and EFR diminished the oxidative burst, which was restored partially when EFRF761H was co-expressed. Outliers are indicated by an additional asterisk and included in statistical analysis. Statistical test: Kruskal-Wallis test (p < 8.516 *10-7), Dunn’s post-hoc test with Benjamin-Hochberg correction (p <= 0.05) Groups with like letter designations are not statistically different.

Related EFR kinases from LRR-RK XIIa in the Arabidopsis genus can function independent of their calatytic activity.

A) Phylogenetic analysis of LRR-RK subfamily XIIa. Selected LRR-RK XIIa kinase domains are labeled and highlighted with purple points. The EFR-like clade contains all Arabidopsis XIIa kinases except FLS2 and XIIa2 and also selected XIIa kinases from Arabidopsis lyrata and Brassica rapa. B,C) The ectodomain of EFR was fused to the transmembrane and intracellular domain of selected LRR-RK XIIa members to create elf18-responsive chimeras for testing the immune signaling function and catalytic dependency of the related kinase domains. The chimeras were transiently expressed in N. benthamiana and tested in oxidative burst assays. All Arabidopsis LRR-RK XIIa members induced an oxidative burst except XIIa2, the closest FLS2 related kinase in the subfamily. Catalytic dependency of the kinase domains appears to vary from kinase to kinase, with catalytically dead versions of EFR, FEXL1 and XIIa5 inducing a WT-like oxidative burst and XPS1 and XIIa6 displaying a reduced oxidative burst. FLS2 kinase dead exhibited a diminished oxidative burst. Experiments were repeated three times with similar results.

Summary table of HDX-MS analysis
Protein expression conditions
Extinction coefficients and molecular weights retrieved from ProtParam and used for determination of protein concentration.
Antibodies used for immunodetection of Western blotted proteins.

Source and dilutions of antibodies are indicated.

List of primer used for molecular cloning in this study.
List of plasmids generated and used in this study.

CZLp number indicates the stock number of the plasmid, in case materials are requested. Source information and bacterial resistance markers are indicated. Plasmid maps are available through a Zenodo repository (link).

Transgenic Arabidopsis used in this study

EFRY836F compromises ligand-induced receptor complex activation.

Two-week old seedlings were mock treated or treated with 1 µM elf18 for 10 min. Immunoprecipitation was then performed with anti-GFP, and the resulting immunoprecipitates probed for BAK1 S612 phosphorylation. Phosphorylated BAK1 was found to co-immunoprecipitate with WT EFR-GFP but not with EFRY836F-GFP, indicating ligand-induced receptor complex activation.

VIa-Tyr forms H-bonds with the αC-β4 loop in various predicted and solved structures.

Solved structures were retrieved from PDB. AlphaFold2 models for kinases in their active conformation were retrieved from (Faezov and Dunbrack, 2023). BAK1 and EFR models were predicted by AlphaFold2, using the complete intracellular domain. H-bonds were predicted in ChimeraX and distances are indicated.

Rational design of activating mutations in EFR and screen for functional recovery of EFRY836F.

A) Alignments of EFR with human kinases containing oncogenic, kinase activating mutations that stabilize the αC-helix-in active-like conformation (described in Foster et al., 2016; Hu et al., 2015). Homologous sites in EFR are indicated by arrows with the residue number. B) Structural model of the EFR kinase domain from AlphaFold2 with homologous sites identified in the sequence alignment from A highlighted in teal (missense mutation) or red (deletion). EFR Y836 at the C-terminal end of the αE-helix is colored purple. C) Screening of the homology-based putatively activating EFR mutations for restoration of EFRY836F function in N. benthamiana. All putative activating mutations were functional at WT-like level except EFRΔNLLKH. Only EFRF761[HM] could functionally recover EFRY836F as the oxidative burst was partially restored. D) EFRL873E showed a WT-like oxidative burst but EFRL873E/Y836F did not restore the oxidative burst. Outliers are in indicated by asterisk in addition to the outlier itself and are included in statistical analysis; Statistical test: Kruskal-Wallis test (p = 9.319*10-6 in C, p = 0.01242 in D), Dunn’s post-hoc test with Benjamin-Hochberg correction (p <= 0.05) Groups with like letter designations are not statistically different.

EFR A-loop phosphorylation sites may coordinate with basic residues from the β3-αC loop and αC-helix.

A) In PKA (1ATP), the A-loop phosphorylation on T197 coordinates with H87 from the αC-helix. B) In EFR (AlphaFold2 (AF2) model), there are two basic residues extending downwards from β3-αC loop (H748) and αC-helix (K752) that may coordinate with A-loop phosphorylation on S887 or S888.

Multiple immune signaling branches are partially restored in EFRF761H/Y836F and EFRF761H/SSAA.

Stable transgenic complementation lines in the Arabidopsis efr-1 background were generated and physiological experiments conducted in the T3 generation (except for EFRF761H /Y836F#5, which is a double insertion line in T2 generation). A) In the oxidative burst assay, EFR F761H restored oxidative burst in EFRF761H/Y836F and EFRF761H/SSAA complementation lines. Two independent experiments were merged into one graph as WT controls showed comparable total oxidative burst. A third independent experiment was performed with similar results. Outliers are indicated by an additional asterisk and included in statistical analysis. Statistical test: Kruskal-Wallis test (p < 2.2*10-16), Dunn’s post-hoc test with Benjamin-Hochberg correction (p <= 0.05) Groups with like letter designations are not statistically different. Alike oxidative burst assays, EFR F761H restored SGI (B) and MAPK activation (C) in EFRF761H/Y836F and EFRF761H/SSAA complementation lines. For SGI assays, four independent experiments wtih 5 nM elf18 treatment are shown. Outliers are indicated by an additional asterisk and included in statistical analysis. Statistical test: Kruskal-Wallis test (p < 2.2 *10-16), Dunn’s post-hoc test with Benjamin-Hochberg correction (p <= 0.05) Groups with like letter designations are not statistically different. For MAPK activation assays, a representative experiment is shown. Similar results were obtained in three more experiments.

Function of BAK1 Y403F is partially recovered by a regulatory spine mutation (I338H) that stabilizes the active-like conformation.

A) The RiD system was utilized to test the recovery of BAK1 Y403F by transient expression in N. benthamiana. The Y403F mutation in BAK1 diminished the oxidative burst, whereas BAK1 I338H displayed near WT-like responses. Combining the I338H and Y403F mutations, however, led to a partial recovery of oxidative burst. Outliers are indicated by an additional asterisk and included in statistical analysis. Statistical test: Kruskal-Wallis test (p < 2.247 *10-11), Dunn’s post-hoc test with Benjamin-Hochberg correction (p <= 0.05) Groups with like letter designations are not statistically different.B) SDS-PAGE analysis of protein levels in experiments 2 and 3 of A is shown. Protein accumulation data for experiment 1 was not collected.

EFR F761H accelerates the onset of the oxidative burst but requires the catalytic activity of BAK1.

A) Quantification of the time until the oxidative burst reaches its half maximum from experiments presented in Figure 2B. Both putative activating mutations, F761H and F761M accelerate the onset of the oxidative burst. B) Time resolved oxidative burst assay. Presented curves are from replicate number three as a representative example. Graphs in A and B are based on data presented in Figure 2B. Error bars represent standard error of the mean. C) EFR F761H requires the catalytic activity of BAK1 to induce the oxidative burst. Data from three independent experiments is merged in one graph. D) Protein accumulation of the RiD-tagged protein related to panel C. Statistical analysis in A and C: Outliers are indicated by an additional asterisk and included in statistical analysis. Statistical test: Kruskal-Wallis test (p = 1.686*10-8 in A, p = 5.89910-8 in C), Dunn’s post-hoc test with Benjamin-Hochberg correction (p <= 0.05) Groups with like letter designations are not statistically different.

Protein accumulation for the oxidative burst assay in Figure 4.

Leaf discs were collected after the oxidative burst assay and protein were extracted by boiling in SDS-loading buffer followed by immunoblotting. Non-infiltrated leaf discs served as negative control.

XIIa5D839N exhibits largely XIIa5WT-like characteristics.

A,B) Catalytic site mutation of XIIa5 exhibited the least delayed onset of oxidative burst. In A, quantification of the time to reach the half maximum of the oxidative burst is shown. The underlying data are the same as used for total oxidative burst in the main figure. In B, the actual oxidative burst curves are presented as average of six individual plants transiently expressing the indicated chimeric protein. Error bars represent standard error of the mean. C) A catalytic site mutation of XIIa5 did not negatively affect BIK1 trans-phosphorylation or BAK1 autophosphorylation. Experiments were performed as described in Figure 1. D) Quantification of band intensities on autoradiographs from three independent experiments are shown. BRI1D1009N and EFRD849N displayed results similar to Figure 1B,C. In contrast to EFRD849N, for which BIK1 and BAK1 relative band intensities slightly decreased compared to wild type EFR, BIK1D202N and BAK1 relative band intensities were wild-type-like for XIIa5D839N.

Protein accumulation of EFR-XIIa chimeras in N. benthamiana.

All constructs exhibited detectable protein accumulation in transiently transformed N. benthamiana leaves. Similar protein accumulation was observed in 3 replicates for A. Protein accumulation for constructs in B was tested once.