Reprogramming of flagellin receptor responses with surrogate ligands

Abstract

Receptor kinase (RK) families process information from small molecules, short peptides, or glycan ligands to regulate core cellular pathways in plants. To date, whether individual plant RKs are capable of processing signals from distinct types of ligands remains largely unexplored. Addressing this requires the discovery of structurally unrelated ligands that engage the same receptor. Here, we focus on FLAGELLIN-SENSING 2 (FLS2), an RK that senses a peptide of bacterial flagellin to activate antibacterial immunity in Arabidopsis. We interrogate >20,000 potential interactions between small molecules and the sensory domain of FLS2 using a large-scale reverse chemical screen. We discover two small molecules that interact with FLS2 in atypical ways. The surrogate ligands weakly activate the receptor to drive a functional antibacterial response channeled via unusual gene expression programs. Thus, chemical probes acting as biased ligands can be exploited to discover unexpected levels of output flexibility in RKs signal transduction.

Fig. 1. Identification of a functional peptidomimetic-based ligand of FLS2 using large-scale reverse chemical genomics.

a Chemical array interaction pipeline used in the study. b Bar plots distribution of FRK1 expression in leaves treated for 2 h with FICs at 1 µM or with flg20 at 100 nM (black arrowhead). FICs are ordered from left to right by increasing FRK1 expression. Bars represent the normalized mean of qPCR quantifications (n = 3). The color gradients indicate the level of FRK1 expression levels. The top 10 hits are indicated in bracket. c Dendrogram showing chemical relatedness between FICs. The color gradients indicate the level of FRK1 expression in (b). d Chemical structures of FIC32/Maya1 and FIC31. e Quantification of interaction between Maya1 and FLS2^ECD using MST. Data points in the fitted dose-response binding curve indicate the fraction of labeled FLS2^ECD bound to Maya 1 (mean ± SE; n = 4). f qPCR analysis, showing relative expression levels of FRK1 from wildtype (WT) and fls2 leaves treated for 2 h with Maya1 (1 µM; n = 8) or flg22 (10 nM; n = 4). The box plot represents the first and third quartiles, centered by the median, and whiskers include the min-max of the data points. Letters above the boxes (a–c) indicate the results of one-way ANOVA followed by a Tukey’s multiple comparison test. Groups with the same letter are indistinguishable at >95% confidence. p-values are provided in the source data file.

Fig. 2. Maya2 interacts with and partially activates FLS2 in roots.

a Bar plots distribution of FRK1 expression in roots after treatments of proFRK1:NLS-3mVenus seedlings for 48 hours with FICs at 1 µM or with flg20 at 100 nM (black bar). FICs are ordered from left to right by increasing mVenus fluorescence intensity. Bars represent the normalized mean fluorescence intensity from at least 5 independent roots. The color gradients indicate the level of fluorescence intensity. The top 10 hits are indicated in bracket. b Dendrogram showing chemical relatedness between FICs. The color gradients indicate the level of FRK1 expression in (a). c Chemical structures of Maya2. d Quantification of interaction between FLS2^ECD and Maya2 using MST. Data points in the fitted dose-response binding curves indicate the fraction of labeled FLS2^ECD bound to Maya2 (mean ± SE; n = 8). e FRK1-induction patterns in proFRK1:NLS-3mVenus roots treated for 48 hours with flg20, flg22 or increasing concentrations of Maya2. Concentrations indicated on top. The nuclear-localized mVENUS signals (green) are co-visualized with the plant cell wall marker propidium iodide (PI, magenta). Scale bar, 100 µm. f Quantitative analysis of mVENUS fluorescence intensity after treatment with the ligands indicated at the bottom. The box plot represents the first and third quartiles, centered by the median, and whiskers include the min-max of the data points. g, h same as (e, f) in the fls2 mutant background. Statistical significance was assessed by linear mixed effect models; letters above the boxes (a–d) indicates the results of a one-way ANOVA followed by a Tukey’s multiple comparison test (two-sided, adjusted for multiple comparisons, p < 0.05). Groups with the same letter are indistinguishable at >95% confidence. The numbers of biologically independent observations (n) are indicated on below the letters on the graph. p-values are provided in the source data file.

Fig. 3. Maya1/2 act as biased ligands to partially activate FLS2 signaling.

a Heatmap of differential interactomes induced by 10 mins treatments of proFLS2:FLS2-GFP seedlings with Maya1 (10 µM), Maya2 (10 µM), or flg22 (10 nM). The pixels corresponding to BAK1 are indicated on the right. The color scale shows log₂ fold changes in protein abundances compared to the DMSO control. b Immunoblotting (IB) analyses of FLS2–BAK1 co-immunoprecipitations in Arabidopsis transgenic seedlings (proFLS2:FLS2-GFP) treated as described in (a). Antibodies were used for blotting as indicated. An independent biological replicate is in Supplementary Fig. 8a. c FRK1-induction measured by qPCR analysis of WT, bak1-5, and fls2 leaves infiltrated with DMSO, Maya1 (1 µM), Maya2 (1 µM), or flg22 (10 nM) for 2 hours. The box plot represents the first and third quartiles, centered by the median, and whiskers include the min-max of the data points. d Seedling Growth inhibition (SGI) percentage values after 6 days of incubation with DMSO, Maya1 (5 µM), Maya2 (5 µM), and flg22 (1 nM). Genotypes are indicated at the bottom. SGI percentage indicates the relative reduction in growth compared between the average fresh weight of the mock treatment controls with that of the treated samples, where 100% indicates no inhibition. The Box plot displays the first and third quartiles, split by the median. Whiskers extend to include the 10^th and 90th percentile. e IB analysis of FLS2-BAK1 co-immunoprecipitation in proFLS2:FLS2-GFP seedlings untreated, treated with flg22, or co-treated with Maya1 or Maya2. Concentrations are indicated on top. This experiment was repeated twice with similar results. f Quantification of relative band intensity (Co-IP/IP) from (e) with two independent experiments using Fiji (mean ± SE; n = 3). g Similar to (e), but for PEPR1-BAK1 in proPEPR1:PEPR1-GFP/pepr1/pepr2 seedlings untreated, treated with pep2 alone, or co-treated with Maya1 or Maya2. This experiment was repeated twice with similar results. h Same as (f), but from (g) with two independent experiments (mean ± SE; n = 3). For (c, d, f, h), statistical significance was assessed by linear mixed effect models; letters above the boxes (a–e) indicate the results of a one-way ANOVA followed by a Tukey’s multiple comparison test (two-sided, adjusted for multiple comparisons, p < 0.05). Groups with the same letter are indistinguishable at >95% confidence. Numbers of biologically independent observations (n) are indicated below the letters. p-values are provided in the source data file.

Fig. 4. Analysis of potential Maya1/2 ligand binding sites from molecular docking simulation.

a Representative snapshot showing the most probable binding pose of Maya1 (left) and Maya2 (right) on the FLS2^ECD structure (PDB ID: 4MNA). The residues predicted to be involved in interaction with Maya1 (turquoise) and Maya2 (green) are shown in pink. FLS2 LRR numbers are shown in gray on the right of the snapshots. b Maya1-mediated induction of FRK1 in leaves of fls2 plants left untransformed (--) or transformed with either wildtype or mutant FLS2 variants (residue changes indicated at the bottom). The box plot represents the first and third quartiles, centered by the median, and whiskers include the min-max of the data points. c Same as (b) but with flg22. d Maya2-mediated induction of FRK1 in roots of proFRK1:NLS-3mVenus/fls2 seedlings. The box plot represents the first and third quartiles, centered by the median, and whiskers include the min-max of the data points. e Same as (d) but with flg22. For (b–e), statistical significance was assessed by linear mixed effect models; letters above the boxes (a–e) indicate the results of a one-way ANOVA followed by a Tukey’s multiple comparison test (two-sided, adjusted for multiple comparisons, p < 0.05). Groups with the same letter are indistinguishable at >95% confidence. Numbers of biologically independent observations (n) are indicated below the letters. p-values are provided in the source data file.

Fig. 5. FLS2 activation by Maya1/2 induces minimal gene expression programs that are sufficient to promote antibacterial immunity.

a Heatmap of differentially expressed genes (DEGs) in wildtype (WT) and fls2 leaves treated for 2 hours with Maya1 (1 µM) or flg22 (10 nM). The color scale bar at the bottom shows the log₂ fold change values of Maya1- or flg22-responsive genes compared to DMSO controls. b Gene Ontology analysis showing enrichment in biological process categories upon treatments of wildtype leaves with Maya1. Pie charts show the fractions of Maya1-responsive gene functions enriched in each category. c, d Venn diagrams showing the overlap of flg22- (left) and FLS2-dependent Maya1-responsive (right) DEGs. Upregulated and downregulated DEGs are shown at the top and bottom, respectively. DEGs numbers are indicated in the circles. e–h same as (a–d) but for 1-h treatments of wildtype and fls2 roots with Maya2 (10 µM) or flg22 (10 nM). i Growth of Ps DC3000 in wildtype and fls2 leaves primed with DMSO, Maya1 (1 µM), Maya2 (1 µM), or flg22 (10 nM) a day prior bacterial inoculation. The number of bacteria per area of leaf (cfu/cm²) is plotted on a log₁₀ scale for day 0 and day 3. The box plot represents the first and third quartiles, centered by the median, and whiskers include the min-max of the data points. Numbers of biologically independent observations (n) are indicated below the letters. Statistical significance was assessed using linear mixed effect modeling; letters above the boxes (a–c) indicate the results of one-way ANOVA followed by a Tukey’s multiple comparison test (two-sided, adjusted for multiple comparisons, p < 0.05). Groups with the same letter are indistinguishable at >95% confidence. Numbers of biologically independent observations (n) are indicated below the letters. p-values are provided in the source data file. j Representative pictures of leaves taken 5 days after inoculation with Ps DC3000. Pretreatments and genotypes are indicated at the bottom and the top, respectively.