Coordinated actions of NLR-assembled and glutamate receptor–like calcium channels in plant effector-triggered immunity

Significance

A major barrier to microbial infection in plants is mediated by large families of intracellular nucleotide-binding/leucine-rich repeat (NLR) immune receptors recognizing pathogen attack. Significant advances have been made in understanding mechanisms of NLR activation and early signaling, with some NLRs functioning as pathogen-induced oligomeric Ca²⁺ permeable ion channels which promote host transcriptional changes and cell death at pathogen infection sites. The next steps in immune response execution remain obscure. Here, we identify and characterize a pair of glutamate receptor–like (GLR) Ca²⁺ ion channels as one important NLR-controlled transcriptional immune output. Our analysis shows that NLR-assembled and canonical Ca²⁺ ion channels cooperate to confer robust resistance against disease.

Abstract

The plant immune system utilizes nucleotide-binding/leucine-rich repeat (NLR) proteins to detect pathogen virulence factors (effectors) inside host cells and transduce recognition to rapid defense. In dicotyledenous plants, pathogen activated Toll-like/interleukin-1 receptor-containing NLRs (TNLs) establish a signaling network of enhanced susceptibility 1 (EDS1)-family dimers with RPW8-type coiled-coil (CC_R) domain NLRs (RNLs) to stimulate transcriptional reprogramming leading to host cell death and pathogen restriction. Evidence suggests that TNL- and EDS1-activated RNLs function as oligomeric Ca²⁺ permeable ion channels at the plasma membrane. However, the downstream processes for immunity execution are poorly understood. Here, we studied pathogen effector-triggered immunity conferred by Nicotiana benthamiana TNL (Roq1) which signals almost exclusively through the EDS1-senescence associated gene101 (SAG101)-N required gene 1 (NRG1) RNL module. We identify a pair of glutamate receptor–like Ca²⁺ ion channels (GLR2.9a and GLR2.9b) which, unlike most other pathogen-induced GLRs, are highly up-regulated by the EDS1-SAG101-NRG1 module in the TNL immune response. We show that oligomeric NRG1 Ca²⁺ channel activity is necessary for GLR2.9a and GLR2.9b induced expression. Consequently, GLR2.9a and GLR2.9b proteins contribute to NRG1-dependent Ca²⁺ accumulation in host cells, and to pathogen resistance and host cell death. We establish that GLR2.9a localizes mainly to the plasma membrane/cytoplasm whereas GLR2.9b accumulates preferentially at the nuclear envelope. The data show that transcriptionally up-regulated canonical Ca²⁺ ion channels GLR2.9a and GLR2.9b are a functional output of the EDS1-SAG101-NRG1 module for TNL-triggered immunity.

Plants have evolved sophisticated innate immune systems which deploy cell-surface and intracellular receptors to detect pathogen attack and induce defense programs (1). Cell-surface pattern recognition receptors (PRRs) recognize microbe- or danger-associated molecular patterns (MAMPs or DAMPs) which activate pattern-triggered immunity (PTI) (2). PTI is a low-level immune response that prevents nonadapted microbes from colonization. Host-adapted (virulent) pathogens deliver virulence factors (called effectors) which manipulate plant cells and suppress PTI, thereby facilitating infection (3). A major immune barrier to disease caused by such microbes is mediated by intracellular nucleotide-binding/leucine-rich repeat (NLR) receptors which detect particular effectors or their activities to induce effector-triggered immunity (ETI) (4). ETI is an amplified immune response involving the transcriptional boosting of PTI defenses and often localized host cell death at pathogen infection sites (1, 5).

Pathogen-detecting NLR receptors, referred to as sensor NLRs, are divided into two major functional classes defined by their N-terminal signaling domains: a coiled-coil (CC) domain in CC-NLRs (CNLs) and a Toll-Interleukin-1 Receptor (TIR) domain in TIR-NLRs (TNLs) (4). Recent studies showed that pathogen-activated sensor NLRs of both types can form signaling-active oligomeric complexes called resistosomes, which are, in principle, similar to pathogen induced NLR inflammasomes conferring innate immunity and cell death in mammals (4, 6). Certain plant CNL-type resistosomes transmit pathogen recognition autonomously to defense pathways (7, 8). Other sensor CNLs require a network of downstream oligomerizing “helper” CNLs to induce pathogen resistance and host cell death (9–13). An emerging biochemical function of sensor and helper CNL resistosomes is as Ca²⁺permeable ion channels associated with the plasma membrane (8, 13–16). Increased Ca²⁺ influx into the cytoplasm mediated by CNL- or CNL-like resistosomes is proposed to stimulate Ca²⁺ decoding systems and nuclear transcription factors which reprogram cells and tissues for defense (17, 18). However, the steps downstream of activated sensor and helper CNL resistosomes in executing immunity and cell death remain obscure.

TNL-type sensor NLRs in dicotyledenous (dicot) plants combat diverse pathogens (19). Pathogen effector-activated sensor TNL resistosomes resistance to peronospora parasitica (RPP1) in Arabidopsis thaliana (20) and recognition of XopQ1 (Roq1) in Nicotiana benthamiana (21), as well as pathogen-responsive TIR-only proteins (22–25), signal as NAD⁺ hydrolyzing enzymes producing cyclic and noncyclic ribosylated nucleotide products. The TNL/TIR-only generated noncyclic nucleotide signals pRib-AMP/pRib-ADP and ADPr-ATP/di-ADPR are plant immune second messengers which, respectively, activate enhanced disease susceptibility 1 (EDS1)—phytoalexin-deficient 4 (PAD4) and EDS1—senescence-associated gene 101 (SAG101) dimeric receptors (26, 27). Nucleotide-bound EDS1-PAD4 and EDS1-SAG101 dimers are recognized by a small conserved group of helper NLRs (called RNLs) with an N-terminal CC-like 4-helical bundle HET-S/LOP-B (HeLo) (CC_HeLo or CC_R) signaling domain (16, 28, 29). Arabidopsis pRib-AMP/pRib-ADP bound EDS1-PAD4 dimers interact specifically with activated disease resistance 1 (ADR1) family RNLs, and ADPr-ATP/di-ADPR bound EDS1-SAG101 dimers with N requirement gene 1 (NRG1) family RNLs, as two distinct immune signaling modules (26, 27, 30–34). In both modules, EDS1 dimer binding results in RNL conformational activation through a similar mechanism as pathogen effector activation of sensor CNLs and TNLs, except that RNLs recognize host small molecule-modified EDS1 complexes (35–38). In vivo evidence points to ADR1- and NRG1-type RNL proteins forming CNL-like resistosomes with Ca²⁺ ion channel activity at the plasma membrane (15, 16, 39). A number of in vivo check points in helper NLR resistosome assembly have been observed as potentially important constraints on activity (13, 37–39). If CNLs and TNLs (the latter via RNLs) converge on Ca²⁺ influxes into cells, a further question remains whether and how NLR-assembled ion channels are coordinated with canonical Ca²⁺ ion channels that regulate plant immune responses and homeostasis (17, 18).

In Arabidopsis, signaling via the EDS1-PAD4-ADR1 or EDS1-SAG101-NRG1 module varies for different sensor TNLs, probably due to availability of their stimulating nucleotides (4). In N. benthamiana, sensor TNL Roq1 recognition of Xanthomonas euvesicatoria (Xe) 85-10 outer protein Q (XopQ) promotes transcriptional reprogramming, pathogen resistance, and host cell death in leaves almost exclusively via the EDS1-SAG101-NRG1 module (40–42), with the EDS1-PAD4-ADR1 module stimulating stomatal defenses (43). Here, we have exploited the NbRoq1-XopQ system to explore processes in ETI execution downstream of activated EDS1-SAG101-NRG1. In a transcriptome analysis we identified a pair of glutamate receptor–like (GLR) Ca²⁺ ion channel proteins GLR2.9a and GLR2.9b which, unlike most GLRs that are mobilized in PTI, are further strongly up-regulated in the NbRoq1 ETI response. By testing NRG1 structure-guided variants, we establish that an EDS1-SAG101-activated NRG1 resistosome with Ca²⁺ ion channel activity is necessary for GLR2.9a and GLR2.9b transcriptional induction in TNL-triggered ETI. We further show that both GLRs, although with different subcellular accumulation profiles, contribute to increased intracellular Ca²⁺ and to pathogen restriction and cell death in Roq1-mediated ETI. Therefore, GLR Ca²⁺ ion channels link TNL-activated EDS1-SAG101-NRG1 mediated transcription to immunity and cell death.

Results

EDS1-SAG101-NRG1 Control of TNL-Triggered Transcriptional Reprogramming in N. benthamiana.

To examine immune-related gene expression changes controlled by the EDS1-SAG101-NRG1 node in N. benthamiana (Nb), we first compared TNL (Roq1) induced cell death and disease resistance phenotypes of Nb wild-type (WT) and previously characterized Nb nrg1-5 (nrg1-5) single (44) and Nb eds1 pad1 sag101a sag101b (epss) quadruple (30) mutants with two newly generated independent sextuple Nb epssna mutant lines (ln1 and ln2) (SI Appendix, Fig. S1A). Conductivity (ion leakage) assays in leaves of these lines at 24 h post infiltration (hpi) with Pseudomonas fluorescens 0-1 (45) (Pf0-1) delivering Xe effector XopQ (recognized by Nb Roq1) (46) were used to quantify host cell death. Pathogen resistance assays were performed by counting Xe bacteria in leaves at 6 days post infiltration (dpi). The epss and epssna mutants were equally defective in promoting Roq1-triggered cell death and pathogen resistance (SI Appendix, Fig. S1 B and C). These data support cooperative functions of EDS1-SAG101 dimers with NRG1 and EDS1-PAD4 dimers with ADR1 in Nb immunity signaling. The Nb nrg1-5 single mutant displayed strongly reduced host cell death and full loss of bacterial resistance (SI Appendix, Fig. S1B), consistent with a major role of the EDS1-SAG101-NRG1 node and minor contribution of the EDS1-PAD4-ADR1 node in Nb TNL-triggered immunity in leaves (41).

We next infiltrated leaves of Nb WT, nrg1-5, epss, and epssna_ln1 with a mock solution (10 mM MgCl₂), Pf0-1 containing an empty vector (Pf0-1 EV) to elicit PTI, or Pf0-1 delivering XopQ to elicit TNL (Roq1) ETI, and processed samples for RNA-sequencing (RNA-seq) analysis at 6 hpi (SI Appendix, Fig. S1D). The 6 hpi time point was chosen to capture early TNL-triggered defense gene expression changes after initial detection of TNL-induced EDS1-SAG101-NRG1 association at 4 hpi with Pf0-1 XopQ (31). A principal component analysis (PCA) of three replicates for each treatment showed that mock treatments in all lines clustered together and away from the PTI and ETI treatments (SI Appendix, Fig. S1E). The Nb WT response to Pf0-1 XopQ (ETI), but not to Pf0-1 EV (PTI), separated from that of the mutants (SI Appendix, Fig. S1E), indicating that WT strongly differentiates from the nrg1-5, epss, and epssna mutant lines in TNL ETI but not PTI responses at 6 hpi. The RNA-seq data, represented as a heatmap of scaled gene expression values (Fig. 1A) or as total numbers of differentially expressed genes (DEG) per genotype (Fig. 1B), further showed that the EDS1-SAG101-NRG1 node promotes gene expression changes in ETI but not in PTI. Transcripts of EDS1a, EDS1b, SAG101a, and NRG1 genes comprising the EDS1-SAG101-NRG1 signaling module (40), but not EDS1-PAD4-ADR1 module components PAD4 and ADR1, were among strongly up-regulated genes in Pf0-1 XopQ-treated WT leaves (SI Appendix, Fig. S1F).

Fig. 1.

Analysis of PTI- and ETI-associated transcriptional changes and their dependence on the EDS1-SAG101-NRG1 node in N. benthamiana. (A) A heatmap representation of transcripts changes in different Nb genotypes, as indicated, at 6 h post MgCl₂ (mock), Pf0-1 EV (EV), or Pf0-1 XopQ (XopQ) infiltration. Transcript values relative to mean normalized counts of all RNA-seq samples are shown on the scale of red (higher than the mean), yellow (close to the mean), and blue (lower than the mean). See Materials and Methods for details. (B) Total numbers of up- and down- DEG in RNA-seq samples (EV vs. mock, XopQ vs. mock, XopQ vs. EV) as indicated. DEG were selected using |log2FC| ≥ 1 and FDR < 0.05 criteria. (C) A Venn diagram showing overlaps between total DEG in Nb WT (XopQ vs. EV [8063]), nrg1-5 (XopQ vs. EV [47]), epss (XopQ vs. EV [3]), and epssna (XopQ vs. EV [14]).

Further comparison of WT and mutant up- and down- regulated genes (up and down DEG) showed that nrg1-5 retained a low number (47) of DEG compared to WT (8063) in ETI vs. PTI tissues (|log2FC| ≥1 and false discovery rate (FDR) < 0.05) which were lost in the epss and epssna mutants, as represented in a Venn diagram (Fig. 1C). These residual DEG in nrg1-5 we attributed to EDS1-PAD4-ADR1 regulation of TNL ETI. Of eight ETI-specific up-DEG in nrg1-5 (Dataset S1), WRKY-family transcription factor NbWRKY40e was reported to bind the Isochorismate synthase 1 (NbICS1) promoter leading to increased salicylic acid biosynthesis and defense amplification impacting stomatal immunity (43, 47). Another WRKY gene, NbWRKY55, is a homolog of AtWRKY70 which confers resistance to Pseudomonas syringae infection through cooperation with AtWRKY46 and AtWRKY53 (48, 49). These data indicate that the EDS1-SAG101-NRG1 node is chiefly responsible for TNL-triggered transcriptional defense leading to host cell death and bacterial resistance in Nb leaves and that the EDS1-PAD4-ADR1 node compensates only to a small degree in transcriptional defense reprogramming when the EDS1-SAG101-NRG1 node is disabled in Nb ETI.

Boosted Expression of GLR 2.9a and 2.9b in Nb ETI.

Analysis of RNA-seq GO term categories enriched among DEG for tissues treated with Pf0-1 EV (PTI) and Pf0-1 XopQ (ETI) showed an enrichment of calcium-related processes (SI Appendix, Fig. S2 and Dataset S2), consistent with EDS1-SAG101 dimer assisted NRG1 assembly into a resistosome-like Ca²⁺ permeable ion channel for defense reprogramming and cell death in ETI (15, 16).

We found that expression of 19Nb GLR genes encoding a family of canonical Ca²⁺-permeable transmembrane ion channels (50) changed in an EDS1-SAG101-NRG1 independent manner in Pf0-1 EV (PTI) vs. mock treatments (Fig. 2A). Therefore, the PTI response in Nb involves a general mobilization of GLR Ca²⁺ channel activities, consistent with earlier observations in Arabidopsis (51). The expression of GLR genes NbGLR2.9a, NbGLR2.9b, and NbGLR3.1 was further significantly boosted by the EDS1-SAG101-NRG1 module in ETI, as indicated in Pf0-1 XopQ vs. Pf0-1 EV treatments (Fig. 2 A and B and SI Appendix, Fig. S3 A and B). We reasoned that this small group of GLRs might therefore be recruited to amplify TNL (Roq1) immunity. To test this further, we focused analysis on NbGLR2.9a and NbGLR2.9b because these genes were most prominent among the TNL (Roq1) ETI-related GLRs (Fig. 2B and SI Appendix, Fig. S3 A and B). Also, NbGLR2.9a and NbGLR2.9b, but not NbGLR3.1, were among NRG1-dependent induced genes in a previous study (41). qRT-PCR analysis of Nb WT, nrg1-5, and epssna Pf0-1 XopQ infiltrated leaves at 4 and 6 hpi showed that the EDS1-SAG101-NRG1 node is the major, but not exclusive, driver of induced NbGLR2.9a and NbGLR2.9b expression in TNL-triggered immunity (Fig. 2 C and D). We further noted that induced expression of NbPR5 in ETI was EDS1-family but not NRG1 dependent (SI Appendix, Fig. S3C), consistent with an EDS1-PAD4-ADR1 contribution to Nb defense (41).

Fig. 2.

Immunity induction and functional features of NbGLR2.9a and NbGLR2.9b. (A) Volcano plot showing transcripts (DEG) responsive to Pf0-1 EV (PTI) in Nb WT (|log2FC| ≥ 1 and FDR < 0.05 for EV vs. Mock). (B) Volcano plot showing transcripts (DEG) responsive to Pf0-1 XopQ (ETI) in Nb WT (|log2FC| ≥ 1 and FDR < 0.05 for XopQ vs. EV). Purple (Right) and green (Left) dots represent up- and down-DEG respectively. Differentially expressed GLR transcripts are highlighted in red and listed on the plots. (C and D) qRT-PCR analysis of GLR2.9a (C) and GLR2.9b (D) transcripts in Nb WT, nrg1-5, and epssna at 4 h and 6 h post infiltration with Pf0-1 XopQ. Samples are normalized to Nb WT at 4hpi with Pf0-1 EV (PTI). Data are from three biological replicates. The Nemenyi test with Bonferroni correction for multiple testing was applied (α = 0.05). Different lowercase letters indicate significant differences. (E) Schematic diagram of AtGLR3.4 Ca²⁺ channel gating mechanism based on protein structure analysis. Monomeric AtGLR3.4 (Left). Without agonist binding GLRs adopt an inactive conformation in which the transmembrane channel (green) is closed (Middle). It is proposed that bound agonists (blue and black circles) to ATD (blue) and LBD (orange) induce conformational changes (black arrows) leading to ion channel opening and Ca²⁺ influx (purple arrow) (Right). TMD, transmembrane domain; S1 and S2 (segment 1 and segment 2); M1 to M4 (membrane-spanning domains 1 to 4), (F) Diagram of predicted Nb GLR2.9a and GLR2.9b protein domains.

In animals, ionotropic glutamate receptors (iGluRs) are ion channel pores activated by glutamate binding to mediate excitatory neurotransmission in the central nervous system (52). In plants, AtGLR3.3 was shown to have a role in amino acid-induced cytosolic Ca²⁺ accumulation and is a key player in the glutamate-mediated wound response (53–55). iGluRs are composed of four subunits, each consisting of an extracellular amino-terminal domain (ATD) and ligand-binding domain (LBD), a transmembrane domain (TMD), and an intracellular carboxyl-terminal (CT) domain (52). The LBD consists of two segments, S1 and S2, which together create a clamshell-like structure around the ligand. Agonist binding to the LBDs typically drives conformational activation of iGluRs and plant GLRs (52, 56). In Arabidopsis, the best characterized GLRs, AtGLR3.3, and AtGLR3.4, exhibit permeability to Ca²⁺ upon binding of glutamate or other amino acids to the LBD (53, 54, 56) (Fig. 2E). The AtGLR3.4 tetrameric structure (PDB: 7LZH) resolved by cryogenic electron microscopy (cryo-EM) revealed that its ATDs bind glutathione (GSH), which further promotes conformational changes necessary for GLR3.4 Ca²⁺ channel activity (56). The AtGLR3.4 TMD consists of transmembrane helices M1, M3, and M4 and a reentrant intracellular loop of M2 and M3 helices forming a gate for regulated Ca²⁺ transport (Fig. 2E) (56). Thus, AtGLR3.4 has structural and functional similarities to animal iGluRs (52, 56).

Based on the Sol Genomics Network Annotation file (57), the Nb genome encodes 36 GLR genes, none of which have been functionally characterized (SI Appendix, Fig. S4). BLASTP analysis with NbGLR2.9a and NbGLR2.9b in Phytozome gave AtGLR2.9 as the closest hit. We performed a phylogenetic analysis using the full-length amino acid sequences of GLRs from Arabidopsis and Nb, and found that the closest GLRs in Arabidopsis are “AtGLR2.8 and AtGLR2.9,” based on tree-distance calculations. However, NbGLR2.9a and NbGLR2.9b did not map to a direct Arabidopsis counterpart (SI Appendix, Fig. S4). We therefore think it likely that NbGLR2.9a/b form a group of GLRs with no clearly identifiable orthologs in Arabidopsis. In this study, we have kept the accepted genome annotation of NbGLR2.9a and NbGLR2.9b from the Sol Genomics Network to avoid confusion.

Both NbGLR2.9a and NbGLR2.9b possess a putative ATD, LBD, and TMD as in AtGLR3.4. However, GLR2.9b which is shorter than GLR2.9a, lacks ATD M4 helices and the LBD S2 region that stabilizes ligand (serine, glutamate, or methionine) binding to AtGLR3.4 (56) (Fig. 2F). The predicted NbGLR2.9b protein sequence has 92.8% identity to NbGLR2.9a (Fig. 2F). Modeling of NbGLR2.9a and NbGLR2.9b onto the active AtGLR3.4 tetramer (56) showed that both NbGLR2.9 proteins are predicted to form a similar tetrameric assembly (SI Appendix, Fig. S5 A and B) (56). Protein sequence alignments and structural modeling showed that, although NbGLR2.9b lacks part of the LBD S2 region and TMD M4 helices, key binding residues for agonists in the LBD S1 region are shared with NbGLR2.9a and AtGLR3.4 (Fig. 2F and SI Appendix, Fig. S5C). Moreover, the extracellular portion of the active tetrameric AtGLR3.4 ion conduction pore lined by M3 helices (aa 671 to 708) and gate-forming residues (56) are conserved in NbGLR2.9a and NbGLR2.9b (SI Appendix, Fig. S5C). These results are consistent with NbGLR2.9a and NbGLR2.9b both being Ca²⁺ permeable ion channels.

Collectively, the data show that NbGLR2.9a and NbGLR2.9b are transcriptionally induced by the TNL-triggered EDS1-SAG101-NRG1 module and their encoded proteins have features of agonist-gated Ca²⁺ permeable ion channels.

An NRG1 Oligomeric Ca²⁺ Ion Channel Drives GLR2.9a and GLR2.9b Induced Expression.

Recent in vitro and in vivo studies showed that an AtEDS1-AtSAG101-AtNRG1 stable trimeric complex forms after TNL activation (37–39). It remained unclear whether the EDS1-SAG101-NRG1 trimer has a defense function or is an inactive intermediate for assembling the signaling-active NRG1 pentamer. We therefore tested whether assembly of the NbEDS1-NbSAG101-activated NbNRG1 resistosome with Ca²⁺ permeable ion channel activity is necessary for transcriptional up-regulation of GLR2.9a and GLR2.9b in TNL mediated ETI. By modeling NbNRG1 onto the AtZAR1 CNL pentameric resistosome (PDB: 6J5T) (7), we identified and mutated equivalent residues (NbNRG1^D16A/D24A) to AtZAR1^E11A/E18A (Fig. 3 A and B). These amino acids are located at the inner surface of an N-terminal α1-helical funnel formed by the resistosome CC-domains and are essential for AtZAR-mediated Ca²⁺ influx into plant cells (14) but not AtZAR1 oligomerization or plasma membrane association (7, 14). Additionally, we made an NbNRG1^G226A/K227A ADP/ATP-binding (GKT, p-loop) mutant in the NbNRG1 NB-domain to disable oligomerization (Fig. 3 A and B) (16, 31).

Fig. 3.

NbNRG1-mediated calcium influx is necessary for XopQ-triggered host cell death and GLR2.9a/GLR2.9b up-regulation. (A) Alignment of NbNRG1 and AtZAR1 at the N-terminal α1-helix and P-loop region. NbNRG1 amino acids D16 and D24 aligned with residues E11 and E18 in AtZAR1 are indicated with red arrow heads. The conserved NbNRG1 G226 K227 P-loop motif is indicated with purple arrow heads. (B) NbNRG1 resistosome modeled on the AtZAR1 pentamer structure (PDB: 6J5T). Upper Right: Magnified α1-helical funnel with amino acids D16 and D24 in red. Bottom Right: Magnified P-loop region with amino acids G226 and K227 in purple. (C) Conductivity measurements of Roq1 triggered cell death in Nb WT, nrg1-5, and stable transformant lines of nrg1-5, as indicated (n = 18 from three independent experiments). Different lowercase letters indicate significant differences by one-way ANOVA followed by the post hoc test (α = 0.05). (D) Ribbon plots show time-course analysis of relative GCaMP3 fluorescence intensity reporting cytoplasmic Ca²⁺ levels in leaves of nrg1-5 stable transgenic lines, as indicated, after Pf0-1 XopQ infiltration (n = 4). GCaMP3-specific fluorescent signals were recorded at 5 min intervals from 0 to 20 hpi with Pf0-1 XopQ. (F_t−F₀): absolute change in GCaMP3 fluorescence signal at time t relative to the baseline at t = 0. (F_t−F₀)/F₀: GCaMP3 fluorescence signal change relative to the baseline. (E and F) qRT-PCR analysis of GLR2.9a, GLR2.9b in Nb WT, nrg1-5 transgenic lines, as indicated, at 6 h post Pf0-1 XopQ infiltration. Samples were normalized to NRG1^WT-myc at 6 hpi mock infiltration. Data are from three biological replicates. The Nemenyi test with Bonferroni correction for multiple testing was applied (α = 0.05). Different lowercase letters indicate significant differences.

We tested the functionality of native promoter-driven C-terminally 4x myc-tagged NRG1^WT, NRG1^D16A/D24A, and NRG1^G226A/K227A in an Nb epssna transient reconstitution assay for TNL (Roq1) resistance to Xe bacteria. The NRG1 proteins were each coexpressed with NbEDS1a-FLAG and NbSAG101b-GFP (or GFP as control) by agroinfiltration. Xe was coinfiltrated into Nb leaf sectors and Xe bacterial growth measured at 6 dpi. Both NRG1^D16A/D24A and NRG1^G226A/K227A failed to limit Xe whereas NRG1^WT restricted Xe infection, although not as strongly as WT plants (SI Appendix, Fig. S6A). This is consistent with a previous finding that transient expression of the At EDS1-SAG101-NRG1 module does not fully complement resistance (30). The NRG1^D16A/D24A and NRG1^G226A/K227A variants transiently expressed in nrg1-5 leaves also failed to promote XopQ-triggered cell death compared to NRG1^WT (SI Appendix, Fig. S6B). These experiments demonstrated that an NRG1 C-terminal 4× myc-tag does not impair its TNL-triggered immunity function, whereas disabling EDS1-SAG101-dependent NRG1 oligomerization and ion channel activity does disrupt its function.

We next made stable transgenic lines expressing the pNRG1: NRG1-myc, pNRG1:NRG1^D16A/D24A-myc, pNRG1:NRG1^G226A/K227A-myc transgenes or pNRG1:GUS-myc (as negative control) in nrg1-5 and selected individual lines with similar NRG1-myc protein accumulation for analysis (SI Appendix, Fig. S6C). While NRG1^WT rescued Pf0-1 XopQ triggered cell death in nrg1-5, NRG1^D16A/D24A showed a partial and NRG1^{G226A/ K227A} a complete loss of rescue (Fig. 3C).

To compare NRG1^WT and the two NRG1 mutant variants in TNL (Roq1) ETI-stimulated calcium influx into cells, we transiently expressed a fluorescent Ca²⁺ sensor GCaMP3 (13, 58) in the nrg1-5 complementation lines and infiltrated Pf0-1 XopQ into the same leaf sectors after 2 d to elicit ETI. NRG1^WT produced a steady increase in [Ca²⁺]_cyt from 4 to 5 hpi corresponding to the timepoint when EDS1-SAG101-NRG1 association is detected in IP experiments (31, 39, 59). This was followed by a steeper rise in Ca²⁺from ~10 to 20 hpi which we attributed to progression of cell death (31). By contrast, the NRG1^G226A/K227A p-loop mutant and the GUS negative control exhibited no early (3 to 6 h) or late (from 10 h) Ca²⁺ increase. The NRG1^D16A/D24A mutant exhibited strongly reduced Ca²⁺ influx from 4 hpi (Fig. 3D). GLR2.9a and GLR2.9b mRNA levels measured by qRT-PCR at 6 hpi with Pf0-1 XopQ showed that NRG1^WT induced GLR2.9a and GLR2.9b up-regulation, whereas NbNRG1^D16A/D24A and NbNRG1^G226A/K227A were defective in promoting expression of these genes (Fig. 3 E and F). Put together, these results suggest that an EDS1-SAG101-activated NRG1 resistosome with ion channel activity, and not the EDS1-SAG101-NRG1 trimer, is responsible for transcriptional induction of GLR2.9a and GLR2.9b in Nb TNL triggered immunity.

GLR2.9a and GLR2.9b Contribute to TNL-Triggered Cell Death and Pathogen Resistance.

We used CRISPR-Cas9 mutagenesis to assess whether GLR2.9a and GLR2.9b are required for TNL-triggered immunity in Nb. Two small guide (sg)RNAs were designed to target each of the GLR2.9a and GLR2.9b genes simultaneously in Nb WT and two independent Cas9-free glr2.9ab homozygous double mutant lines (ln1 and ln2) were selected (SI Appendix, Fig. S7A). We compared TNL (Roq1) ETI responses of glr2.9ab plants to those of Nb WT and nrg1-5. Host cell death elicited in leaves after infiltration of Pf0-1 XopQ was again quantified at 1 d by conductivity assays. The glr2.9ab mutants showed a partial loss of cell death, lying between WT and nrg1-5 plants (Fig. 4A). In TNL Roq1 resistance assays to Xe delivering XopQ, restriction of bacterial growth at 6 dpi in leaves of glr2.9ab was also intermediate between Nb WT and nrg1-5 (Fig. 4B). By contrast, growth of a nonrecognized XopQ deletant strain of Xe (XeΔXopQ) was unaffected in glr2.9ab and nrg1-5 mutants (Fig. 4C). We measured responses to the bacterial PAMP flg22 in Nb WT, glr2.9ab, and nrg1-5 plants and found that a flg22-induced ROS burst and expression of CYP71D20 (a PTI marker gene in Nb) (60) were similar between WT and mutant plants (Fig. 4D and SI Appendix, Fig. S7 B and C). Hence, GLR2.9a and GLR2.9b are required for full TNL-triggered, NRG1-driven ETI but not for basal immunity or flg22-triggered immune responses in Nb.

Fig. 4.

NbGLR2.9a and NbGLR2.9b promote TNL Roq1 triggered host cell death and pathogen resistance. (A) Conductivity measurements of Roq1-mediated cell death in Nb WT, nrg1-5, glr2.9ab_ln1, and glr2.9ab_ln2 at 1 d with Pf0-1 XopQ (n = 12 from three independent experiments). (B) Xe growth assay in leaves of the same lines as (A) at 6 dpi with Xe (n = 12 from three independent experiments). (C) XeΔXopQ growth assays in leaves of the same lines as (A) at 6 dpi (n = 12 from three independent experiments). (D) Nb WT, nrg1-5, glr2.9ab_ln1, and glr2.9ab_ln2 show comparable ROS production in response to 1 μm flg22 treatment. ROS production was monitored using a chemiluminescence assay. Data are presented as average signals from three independent experiments. Ribbon plots represent mean value ± SE (n = 8 biologically independent discs) with all individual data points. (E) Conductivity measurements of RPS4 TIR triggered cell death in leaves of the same lines as (A) at 24 h post agroinfiltration (n = 9 from three independent experiments). (F) Conductivity measurements of CNL Rx triggered cell death in leaves of the same lines as (A) at 24 h post agroinfiltration of Rx with or without PVX coat protein effector (CP) (n = 12 from three independent experiments). (A–F) Genotypes with different letter codes are significantly different. The one-way ANOVA followed by the post hoc test was applied (α = 0.05).

Additionally, transient overexpression of the TIR domain from Arabidopsis TNL receptor Resistance to Pseudomonas syringae 4 (RPS4), which has NADase- and EDS1-SAG101-NRG1-dependent cell death activity in Nb (26, 61), produced lower cell death in the glr2.9ab mutants compared to WT Nb (Fig. 4E). Therefore, GLR2.9a and GLR2.9b can stimulate Nb defenses downstream of TNL and TIR NADases. The glr2.9ab mutants were not compromised in cell death elicited by a CNL-class sensor NLR potato Rx activated by its coexpression with Rx-recognized potato virus X coat protein (CP) (Fig. 4F) (62) or by the barley CNL receptor MLA13 activated by effector AVR_A13-1 (63) in Nb leaves (SI Appendix, Fig. S7D). These data suggest that GLR2.9a and GLR2.9b are downstream components specifically of the TNL- or TIR-triggered EDS1-SAG101-NRG1 node to promote host cell death and pathogen restriction in Nb.

GLR2.9a and GLR2.9b Display Different Subcellular Locations.

In transient expression assays we tested whether fluorescent protein-tagged NbGLR2.9a and NbGLR2.9b expressed under control of their respective native promoter (pGLR2.9a:gGLR2.9a-mCherry and pGLR2.9b:gGLR2.9b-GFP) could rescue Pf0-1 XopQ-triggered cell death in glr2.9ab ln1 (Fig. 5A). In these assays expression of GLR2.9a-mCherry and GLR2.9b-GFP alone or together by agroinfiltration did not induce cell death at 2 dpi (SI Appendix, Fig. S7E), after which Pf0-1 XopQ was infiltrated into leaf sectors to activate TNL Roq1. GLR2.9a-mCherry and GLR2.9b-GFP individually or coexpressed, but not GUS-GFP (p35S:GUS-GFP), complemented the Nb glr2.9ab ln1 mutant in TNL (Roq1) cell death, although GLR2.9b-GFP alone routinely produced a stronger TNL-triggered cell death response than GLR2.9a-mCherry (Fig. 5A). The GLR2.9a and GLR2.9b proteins were detected on a Western blot probed, respectively, with α-mCherry and α-GFP antibodies (SI Appendix, Fig. S7F). We concluded that NbGLR2.9a-mCherry and NbGLR2.9b-GFP are each able to rescue Nb glr2.9ab ln1 defects in TNL-triggered cell death to Pf0-1 XopQ and that NbGLR2.9b alone can fully compensate for combined loss of GLR2.9a and GLR2.9b.

Fig. 5.

GLR2.9a and GLR2.9b proteins display different subcellular distributions. (A) Transient expression of GLR2.9a-mCherry together with GLR2.9b-GFP and GLR2.9a-mCherry or GLR2.9b-GFP alone, but not GUS-GFP, can rescue XopQ triggered cell death in glr2.9ab_ln1. At 2 d post agroinfiltration of the constructs, Pf0-1 XopQ was infiltrated into the same leaf sectors and samples were collected at 6 hpi for conductivity measurements. Nb WT infiltrated with or without Pf0-1 XopQ serves as positive and negative control (n = 12 from three independent experiments). Lowercase letters indicate significant differences measured by one-way ANOVA followed by the post hoc test (α = 0.05). (B and C) Western blot analysis of GLR2.9a-mCherry (B) and GLR2.9b-GFP (C) protein accumulation in Nb glr2.9ab_ln1 or Nb epssna leaves at 6 h post treatment with MgCl₂ (mock), Pf0-1 EV, or Pf0-1 XopQ. GLR2.9a-mCherry and GLR2.9b-GFP proteins were detected using α-mCherry and α-GFP antibodies, respectively. Three independent experiments showed the same trends. (D and E) Subcellular localizations of constitutively expressed GLR2.9a-mCherry (D) and GLR2.9b-GFP (E) at 2 d post agroinfiltration into Nb glr2.9ab leaves, imaged by confocal microscopy and analyzed in three independent experiments (five cells imaged/experiment). Representative images of GFP (green) and mCherry (red) signals are shown. Yellow arrows point to the nucleus. (Scale bar, 25 μm.) (F and G) Isolation of total (Input), plasma membrane/cytoplasm (PM/C), and nuclear (N) enriched fractions at 2 d post infiltration of Nb glr2.9ab leaves with 35S:GLR2.9a-mCherry (F) and 35S:GLR2.9b-GFP (G). α-PEPC and α-Histone H3 were used as cytosolic and nuclear markers, respectively. Three independent experiments showed similar results.

We monitored GLR2.9a-mCherry or GLR2.9b-GFP protein accumulation at 2 d after agroinfiltration in Nb glr2.9ab (ln1) or epssna leaves followed by 6 h mock, Pf0-1 EV, or Pf0-1 XopQ treatment. In the glr2.9ab background, levels of both GLR2.9 proteins increased in response to Pf0-1 EV vs. mock, and further in response to Pf0-1 XopQ vs. Pf0-1 EV (Fig. 5 B and C). The Pf0-1 XopQ (ETI) vs. Pf0-1 EV (PTI) increment in GLR2.9a and GLR2.9b protein accumulation was lost in epssna leaves (Fig. 5 B and C). These data suggest that GLR2.9a and GLR2.9b protein accumulation after TNL triggering is strongly EDS1-SAG101-NRG1-dependent, consistent with their respective mRNAs (Fig. 2 C and D). Increased GLR2.9a accumulation in Pf0-1 EV vs. mock treatments did not require EDS1-RNL signaling (Fig. 5B), fitting with GLR2.9a gene expression data (SI Appendix, Fig. S3B) and suggesting that a different mechanism underlies GLR2.9a and GLR2.9b accumulation in Nb PTI.

We tested the subcellular localizations of native promoter-controlled GLR2.9a-mCherry or GLR2.9b-GFP transiently expressed in glr2.9ab ln1 using confocal microscopy at 6 hpi mock, Pf0-1 EV, or Pf0-1 XopQ treatment. For GLR2.9a-mCherry, a weak signal consistent with GLR2.9a protein accumulation at the plasma membrane/cytoplasm (PM/C) compartment was detected after Pf0-1 EV, and a stronger signal after Pf0-1 XopQ treatment (SI Appendix, Fig. S8A). We did not detect NbGLR2.9b-GFP in these assays even with a Pf0-1 XopQ stimulus (SI Appendix, Fig. S8B). We therefore used biochemical fractionation assays followed by Western blotting to monitor subcellular accumulation of the same native promoter-driven GLR2.9a-mCherry and GLR2.9b-GFP constructs in glr2.9ab leaves at 6 h post mock, Pf0-1 EV, or Pf0-1 XopQ infiltration. In Pf0-1 EV and Pf0-1 XopQ elicited tissues, most GLR2.9a-mCherry accumulated in the PM/C fraction with a minor nuclear-enriched pool also detected (SI Appendix, Fig. S8C). By contrast, a NbGLR2.9b-GFP signal was detected predominantly in the nuclear-enriched fraction and was strongest after Pf0-1 XopQ treatment (SI Appendix, Fig. S8D). These data suggest that GLR2.9a and GLR2.9b proteins have different subcellular preferences in TNL ETI-elicited Nb leaf cells.

Because we could not detect transiently expressed native promoter-driven GLR2.9b-GFP in glr2.9ab ln1 leaves using confocal microscopy imaging (SI Appendix, Fig. S8B), we monitored fluorescence signals from GLR2.9a-mCherry or GLR2.9b-GFP transiently expressed under control of a CaMV-35S constitutive promoter in glr2.9ab ln1 leaves without an immune trigger. We found that GLR2.9a-mCherry localized mainly to the PM/C compartment but also to the nuclear envelope (NE), with part of NE signal also possibly coming from the endoplasmic reticulum proximal to the nucleus (Fig. 5D). GLR2.9b-GFP signals were detected predominantly at the NE and weakly at the PM/C, and appeared to form puncta in both compartments (Fig. 5E). We did not observe cell collapse or death at 2 dpi when images were taken (SI Appendix, Fig. S7E), suggesting that GLR enhanced expression alone is insufficient to elicit cell death. Overexpression of GLR2.9a and/or GLR2.9b did not rescue the loss of cell death in nrg1-5 after Roq1 activation or in response to RPS4-TIR (SI Appendix, Fig. S9 A and B). Also, flg22-induced cell death in nrg1-5 was not enhanced by GLR2.9a and/or GLR2.9b overexpression (SI Appendix, Fig. S9C). These data support the notion that up-regulated GLR2.9a and GLR2.9b do not signal without NRG1-induced defense reprogramming. Biochemical fractionation of the constitutively expressed GLR2.9a-mCherry and GLR2.9b-GFP proteins in glr2.9ab supported different PM/C vs. NE distributions (Fig. 5 F and G), as observed for the native promoter-driven GLR2.9a-mCherry and GLR2.9b-GFP proteins in Pf0-1 XopQ-triggered tissues (SI Appendix, Fig. S8 C and D). Collectively, the imaging and biochemical fractionation data suggest that while GLR2.9a and GLR2.9b are both strongly induced in TNL ETI responding tissues, they have different subcellular preferences, with GLR2.9a accumulating mainly at the PM/C and GLR2.9b at the NE.

GLR2.9a and GLR2.9b Promote Ca²⁺ Influx into Cells in TNL-Triggered ETI.

To assess the contributions of GLR2.9a and GLR2.9b to NRG1-dependent calcium influx into cells in TNL triggered ETI, we crossed a previously characterized stable transgenic Nb WT line expressing GCaMP3 (pCaMV35S:GCaMP3) (58) with nrg1-5 and glr2.9ab ln1. Homozygous lines were selected and used to quantify changes in cytoplasmic Ca²⁺ levels ([Ca²⁺]_cyt). Leaves of GCaMP3 sensor-expressing Nb WT plants were infiltrated with Roq1-recognized Xe or nonrecognized virulent XeΔXopQ bacteria and [Ca²⁺]_cyt dynamics were measured over 20 h as changes in GCaMP3 fluorescence over time. These assays showed a TNL ETI-specific cytoplasmic Ca²⁺ increase between 7 and 12 hpi (Fig. 6A). We excluded possible interference from autofluorescence due to host cell death because Xe induced macroscopic cell death in these assays was first observed between 1 and 2 dpi (SI Appendix, Fig. S10). We next compared Nb WT with nrg1-5 and glr2.9ab leaves after Xe infiltration. In contrast to the prominent Ca²⁺ peak between 7 and 12 hpi in Xe-elicited WT, the glr2.9ab mutant had reduced cytoplasmic Ca²⁺ accumulation and nrg1-5 displayed no Ca²⁺ peak (Fig. 6B). We concluded that part of the EDS1-SAG101-NRG1-driven TNL ETI response in Nb is mediated by up-regulated GLR2.9a and GLR2.9b Ca²⁺ channel activities contributing to host cell death and pathogen resistance.

Fig. 6.

NbNRG1, NbGLR2.9a, and NbGLR2.9b promote Ca²⁺ influx in TNL triggered ETI. (A) Ribbon plots show time-course analysis of relative GCaMP3 fluorescence intensity reporting cytoplasmic Ca²⁺ levels in leaves of Nb WT expressing GCaMP3, after XeΔXopQ or Xe infiltration (n = 3). GCaMP3-specific fluorescent signals were recorded at 5 min intervals from 0 to 20 hpi. (B) Ribbon plots show time-course analysis of relative GCaMP3 fluorescence intensity reporting cytoplasmic Ca²⁺ levels in leaves of Nb WT, glr2.9ab, and nrg1-5 expressing GCaMP3 upon infiltration with Xe. GCaMP3-specific fluorescent signals were recorded at 5 min intervals from 0 to 20 hpi. (C) A working model of signaling events in Nb TNL triggered cell death and pathogen immunity. Xe delivers effector XopQ into the host cell, which is recognized by TNL Roq1. The XopQ-activated Roq1 TNL resistosome is an NADase enzyme producing ADPr-ATP and di-ADPR nucleotide small molecules which specifically bind to and promote activation of an EDS1-SAG101-NRG1 complex. The resulting NRG1 pentameric resistosome forms a Ca²⁺ permeable ion channel at the plasma membrane which promotes Ca²⁺ influx into the cell, leading to transcription factor mediated defense reprogramming in the nucleus. GLR genes GLR2.9a and GLR2.9b are up-regulated downstream of NRG1 resistosome ion channel activity and, through their further promotion of Ca²⁺ influx into cells, contribute to TNL effector-triggered immunity. GLR2.9a and GLR2.9b proteins accumulate at the PM and NE to confer host cell death and pathogen resistance.

Discussion

To investigate processes operating downstream of the EDS1-SAG101-NRG1 node in TNL (Roq1) mediated ETI, we focused analysis on GLR2.9a and GLR2.9b Ca²⁺ channel genes and their proteins. These were among numerous PTI induced (EDS1-SAG101-NRG1 independent) GLRs in Nb, indicative of a general mobilization of GLR channel activities in PTI (51, 54). GLR2.9a and GLR2.9b were exceptional in being further boosted in expression through the EDS1-SAG101-NRG1 node in TNL ETI (Fig. 2 A and B). Reduced TNL-triggered NRG1-dependent Ca²⁺ influxes into cells recorded by a GCaMP3 fluorescence reporter (Fig. 6B), as well as compromised Xe resistance and host cell death in Nbglr2.9ab mutant lines (Fig. 4 A, B, and D), show that the transcriptionally induced GLR2.9a and GLR2.9b Ca²⁺ ion channels contribute to ETI, as depicted in Fig. 6C. Therefore, EDS1-SAG101-NRG1 mobilized canonical GLR2.9a and GLR2.9b Ca²⁺ ion channel activities are an important Nb TNL immunity output.

We used Nb combinatorial mutants to assess relative contributions of the EDS1-PAD4-ADR1 and EDS1-SAG101-NRG1 modules to PTI (Pf0-1 EV) and TNL ETI (Pf0-1 XopQ) transcriptomes (Fig. 1 A and B). As previously reported (41), there was an almost exclusive requirement for the EDS1-SAG101-NRG1 node in TNL-triggered gene expression changes in Nb leaves, with the EDS1-PAD4-ADR1 node playing a very minor role (Fig. 1 A and B). It emerged recently that the Nb EDS1-PAD4-ADR1 node is important for transcriptional reprogramming in stomatal immunity (43). In our analysis, loss of both EDS1-SAG101-NRG1 and EDS1-PAD4-ADR1 signaling branches (in Nb epss and Nb epssna mutants) only mildly impaired PTI-related gene expression in response to Pf0-1 EV bacteria vs. a mock treatment (Fig. 1B). This result tallies with an earlier report that EDS1 dimers and their respective RNLs are not required for PTI signaling from tested PRRs in Nb leaves (42). Therefore, transcriptional mobilization of defenses including multiple GLR genes in PTI (Fig. 1 A and B) appears to rely on different, presumably TNL/TIR-independent, mechanisms. The observed dispensability of GLR2.9a and GLR2.9b for limiting growth of virulent XeΔXopQ bacteria (Fig. 4C) or for PAMP flg22 induced ROS burst and CYP71D20 defense gene expression in leaves (Fig. 4D and SI Appendix, Fig. S7 B and C) might be due to redundancy with other induced GLRs (Fig. 2A) and probably other classes of Ca²⁺ ion channel, such as CNGCs which are involved in PTI and ETI regulation (17, 18).

The biochemical mechanism of NRG1 conformational activation by association with ADPr-ATP/di-ADPR bound EDS1-SAG101 was resolved in two recent structural studies of the Arabidopsis EDS1-SAG101-NRG1 protein complex (37, 38). These analyses identified interfaces formed between TNL-triggered EDS1-SAG101 C-terminal domains and the extreme C-terminal portion of NRG1 to promote NRG1 activation and host cell death. Assembly of an NRG1 stable pentamer required for the N-terminal α1-helix Ca²⁺ ion channel activity (15, 16) was not achieved in insect cell expression assays (37, 38). Instead, a stable EDS1-SAG101-NRG1 trimeric complex formed, which was also prominent in Arabidopsis tissues engineered to elicit ETI without PTI (39). Notably, accumulation of oligomeric AtNRG1 putative resistosomes at the plasma membrane required additional PTI signals with ETI (39). Also, pathogen-induced Arabidopsis and Nb EDS1-SAG101-NRG1 protein pools were detected in nuclei (39, 42). Put together, these characteristics raised the possibility of an alternative EDS1-SAG101-NRG1 trimer function in transcriptional defense. To critically test this we examined pathogen immunity, cell death, and Ca²⁺ influx phenotypes of NbNRG1 mutant variants disabled in EDS1-SAG101-mediated oligomerization (p-loop G226A/K227A) or N-terminal CC_R α1-helix ion channel function (D14A/D16A) (Fig. 3 C and D and SI Appendix, Fig. S6 A and B) (16, 31). Our data show that an NbNRG1 resistosome with Ca²⁺ ion channel activity is the chief driver of GLR2.9a and GLR2.9b up-regulation (Fig. 3 E and F) and of TNL ETI resistance and host cell death (Fig. 3C and SI Appendix, Fig. S6 A and B). Whether pathogen-activated NRG1 resistosomes signal exclusively at the plasma membrane (15, 16) or more broadly at endomembrane compartments and/or the NE requires further analysis. The CC_R domain is present in several key membrane-associating immune regulators with different architectures to NLRs (29, 64). Arabidopsis CC_R containing mixed-lineage kinase domain-like (MLKL 1) protein contributes to TNL-activated EDS1-dependent immune potentiation in leaves and, as part of this process, forms puncta at the plasma membrane (59, 65). The MLKL N-terminal CC_R domains within an activated tetramer mediated sustained Ca²⁺ influx into cells (59), thus adding to the network of canonical and noncanonical ion channels orchestrating plant immune responses (17, 34).

Emerging evidence suggests that plant-specific factors or processes provide check points in the assembly of signaling-active helper NLR resistosomes (13, 37–39), probably to prevent undesirable spread of defenses and cell death. A similar principle might apply to the GLR2.9a and GLR2.9b ion channels because their transient overexpression in Nb leaves without an immune trigger did not elicit cell death (SI Appendix, Fig. S7E). Also, overexpressed GLR2.9a and/or GLR2.9b failed to alleviate the loss of TNL (Roq1) or TIR (RPS4)-triggered cell death in nrg1-5 (SI Appendix, Fig. S9 A and B). We conclude that other signals induced by the EDS1-SAG101-activated NRG1 resistosome converge on up-regulated GLR2.9a and GLR2.9b proteins to stimulate their Ca²⁺ ion channel activities. Binding of reduced glutathione (GSH) within the ATD of AtGLR3.4 contributed to Ca²⁺ ion channel gating (56). Also, exogenously supplied GSH promoted Ca²⁺cytosolic influx and defense gene expression in Arabidopsis (54). Besides GSH, glutamate and methionine were reported to bind to the LBDs of AtGLR3.4 and AtGLR3.3 and stimulate Ca²⁺ ion transport (56, 66). GSH provision slowed pathogen growth while glutamate enhanced the wound response to insect herbivores through an AtGLR3.3-dependent pathway (54, 67). Although NbGLR2.9b protein lacks an LBD S2 portion which in AtGLR3.4 contains three residues (F624, E668, and Y671) for agonist binding and two residues (Q620, D621) to stabilize the ligand (56), conserved binding (D570, T572, R577) and stabilizing residues (R501, N549, Y552) present in LBD S1 might be sufficient to activate GLR2.9b (SI Appendix, Fig. S5C). Moreover, while the M4 membrane spanning segment present in structurally resolved AtGLR3.4 and the modeled NbGLR2.9a is missing in NbGLR2.9b (Fig. 2F and SI Appendix, Fig. S5 A–C) (56), GLR2.9b retains key GLR ion channel functional motifs (Fig. 2F and SI Appendix, Fig. S5 B and C). Also, GLR2.9b complemented TNL-triggered cell death when transiently expressed in Nbglr2.9ab_ln1 (Fig. 5A), suggesting that this shorter GLR2.9 isoform is a functional Ca²⁺ ion channel. Further investigation is required to determine how GLR2.9a and GLR2.9b ion channel activities are regulated, beyond their transcriptional induction by EDS1-SAG101-activated NRG1 in TNL mediated ETI.

Data showing where plant GLRs accumulate in cells are sparse, although a plasma membrane localization was proposed for Arabidopsis GLR3.4 where it can be activated by amino acids to transport Ca²⁺ into the cytoplasm (68). Our confocal microscopy imaging and subcellular fractionation assays suggest that NbGLR2.9a-mCherry and NbGLR2.9b-GFP accumulate at the PM/C and NE to different ratios when transiently expressed in Nbglr2.9ab leaves (Fig. 5 D–G and SI Appendix, Fig. S8). Especially striking was a prominent NbGLR2.9b-GFP pool enriched with nuclei after TNL triggering of leaves (SI Appendix, Fig. S8D), whereas NbGLR2.9a-mCherry accumulated preferentially at the PM/C (SI Appendix, Fig. S8C). Since overexpressed NbGLR2.9a-mCherry and NbGLR2.9b-GFP without an immune trigger (Fig. 5 F and G) displayed similar subcellular profiles as their native-promoter controlled counterparts in TNL immune-triggered cells (SI Appendix, Fig. S8 C and D), we speculate that GLR2.9a and GLR2.9b isoforms are preferentially targeted and/or stabilized, respectively, at the PM/C and NE. Puncta detected for NbGLR2.9b-GFP in both compartments (Fig. 5E) resemble those formed at the plasma membrane by an autoactive AtNRG1.1 D485V variant and effector-activated AtNRG1.1 WT in plant cells (15, 16). A loss-of-oligomerization mutant AtNRG1.1 L134E failed to produce puncta compared to AtNRG1.1 WT and abolished TNL-triggered cell death (15). Hence, puncta might reflect biologically active RNL and GLR protein foci. Our data provide a snapshot of GLR2.9a and GLR2.9b subcellular accumulation. It remains to be tested—especially for native promoter regulated GLR2.9b-GFP which we failed to detect by fluorescence imaging even under TNL ETI conditions (SI Appendix, Fig. S8B)—whether dynamic GLR2.9 protein relocalization to particular sites occurs after their TNL-triggered upregulation. In Medicago truncatula, three CNGC15 ion channel members were found to localize to the NE of root cells in order to modulate nuclear Ca²⁺ release which was necessary for host accommodation of nutrient-provisioning endosymbionts (69). In the Nb leaf response to pathogens it is possible that GLR2.9a activity at the PM/C and GLR2.9b at the NE help to optimize cytoplasmic and nuclear Ca²⁺-dependent decoding for rapid defense.

Materials and Methods

Plant Accession Numbers.

NbGLR2.9a (Niben101Scf01212g02011); NbGLR2.9b (Niben101Scf08670g00020); NbNRG1 (Niben101Scf02118g00018); NbADR1 (Niben101Scf02422g02015).

Plant Materials and Growth Conditions.

The epss and nrg1-5 mutants are published (30, 44). epssna In1 and ln2 and glr2.9ab In1 and ln2 mutants were generated in this study. Guide RNA sequences were designed to target NbGLR2.9a, NbGLR2.9b, NbNRG1, and NbADR1, and the corresponding oligonucleotides were annealed and ligated into a pDONR-based entry plasmid containing the Cas9 gene and a U6-26 promoter for guide RNA expression. An LR reaction (Life Technologies) was used to move the guide and Cas9 cassette into a Gateway compatible version of the pCambia2300 vector (70). Primers used for CRISPR Cas9 mutagenesis constructs are listed in SI Appendix, Table S1. Simultaneous CRISPR-Cas9 knockout mutations of NbNRG1 and NbADR1 were selected in the epss background; mutations of NbGLR2.9a and NbGLR2.9b were selected in the Nb WT. T3 homozygotes without Cas9 were used. Stable transformants of nrg1-5 expressing native promoter-driven NRG1-4xmyc WT or NRG1-4xmyc variants were generated by Agrobacterium-mediated transformation (71). Transgenic Nb WT harboring CaMV35S:GCaMP3 (58) was crossed with nrg1-5 and glr2.9ab_ln1. Kanamycin served as a selection marker. T3-generation homozygous mutant lines with the CaMV35S:GCaMP3 transgene were used for cytosolic calcium measurements. Nb plants were grown in 16 h photoperiod-controlled environment rooms at 22 to 24 °C and 65% relative humidity for 4 to 5 wk.

Construct Generation by Golden Gate Cloning.

Constructs for expressing NRG1 WT and variants were prepared using the Golden Gate MolClo kit (72).

The promoter+3’UTR of Nb NRG1 (830 bp) was cloned into level 0 plasmid pICH41295. NRG1 CDS was cloned into level 0 plasmid pAGM1287. NRG1^D16A/D24A and NRG1^G22A6/K227A were generated using the QuikChange II Site-Directed mutagenesis (SDM) protocol (#200555, Agilent). To obtain level 1 NRG1 constructs, level 0 constructs of NRG1/variants were combined with pICH41295 containing NRG1pro, 4xmyc (pICSL50010), CaMV 35S terminator (pICH41414), and backbone pICH47732. The promoter+3’UTR of Nb GLR2.9a (1,916 bp) and GLR2.9b (1,579 bp) were cloned into level 0 plasmid pICH41295. Genomic GLR2.9a and GLR2.9b were separately cloned into level 0 plasmid pAGM1287. Expression level 1 constructs for GLR2.9a (35S:gGLR2.9a-mCherry and pGLR2.9a:gGLR2.9a-mCherry) and GLR2.9b (35S:gGLR2.9b-GFP and pGLR2.9b:gGLR2.9b-GFP) were cloned using the above strategy except the modules were cauliflower mosaic virus (CaMV) 35S promoter (pICH51288), GLR2.9apro or GLR2.9bpro, gGLR2.9a or gGLR2.9b, GFP tag (pICSL50008) or mCherry tag (pICSL50004). Level 1 Golden Gate expression constructs were transformed via electroporation into Rhizobium radiobacter (hereafter Agrobacterium tumefaciens or Agrobacteria) strain GV3101 pMP90RK or pMP90 for transient expression in Nb or stable transformation of Nb. Primers for cloning and SDM are listed in SI Appendix, Table S1.

Agrobacterium-Mediated Transient Expression in N. benthamiana.

A. tumefaciens strains were grown on YEB plates containing appropriate antibiotics and incubated 1 d at 28 °C. Cells were collected from plates and resuspended in induction buffer (10 mM MES pH 5.3, 10 mM MgCl₂,150 nM acetosyringone) at OD₆₀₀ = 0.2. Silencing suppressors p19 and CMV2b (OD600 = 0.1) were coinfiltrated into leaves of 4- to 5-wk-old Nb leaves with a 1 mL needleless syringe. In cell death assays, leaves were spot-infiltrated and evaluated for macroscopic tissue collapse 2 d post infiltration. Leaf discs (7 mm) were punched out of infiltrated zones at various time points for conductivity (ion leakage) measurements and frozen in liquid nitrogen for western blot analysis.

Pf0-1 Infiltration into N. benthamiana.

Leaves of Nb were infiltrated with type III secretion system-equipped Pseudomonas fluorescens 0-1 (Pf0-1) strains resuspended at OD₆₀₀ = 0.3 in 10 mM MgCl₂. Mock infiltration used 10 mM MgCl₂ only. Leaves for protein assays and RNA-seq were harvested at 6 hpi, flash-frozen in liquid nitrogen and stored at −80 °C. Leaves were visualized for macroscopic tissue-collapse and cell death measurement 1 d after the Pf0-1 strains infiltration.

Cell Death Conductivity Assays.

A. tumefaciens or Pf0-1 (45) infiltrated Nb plants were kept at 22 °C with a 16-h light period. Eight 7-mm leaf discs were taken at specified timepoints, washed in 20 mL of milliQ water for 60 min, transferred to a 48-well plate with 0.5 mL milliQ in each well, and incubated at room temperature under light. Ion leakage (conductivity) was measured at 0 h and 24 h with a conductometer, Horiba Twin ModelB-173. For statistical analysis, results of measurements for individual wells (each leaf disc in one well represents a technical replicate) were combined from independent experiments (biological replicates).

In Planta Bacterial Growth Assays.

Xe 85-10 and XeΔXopQ were resuspended in 10 mM MgCl₂ at OD₆₀₀ = 0.001 and syringe-infiltrated into Nb leaves. Plants were kept in a long-day chamber (16 h light/ 8 h dark at 24 °C/22 °C). Bacteria were isolated at 0 dpi and 6 dpi (each with four 7-mm leaf discs representing four technical replicates), and dilutions were dropped onto NYGA plates supplemented with rifampicin 100 mg/L and streptomycin 150 mg/L. In statistical analysis of Xe and XeΔXopQ titers at 6 dpi, results from independent experiments (biological replicates) were combined.

Nuclear/Cytoplasm Fractionation.

Leaf tissue (1.5 g fw) was homogenized in 3 mL Honda buffer (2.5% Ficoll 400, 5% Dextran T40, 0.4 M sucrose, 25 mM Tris-HCl, pH7.4, 10 mM MgCl₂ (with 5 mM DTT, 1× protease inhibitor cocktail added before use). The lysate was filtered through a 62 µm nylon mesh, by centrifuging for 3 min at 1,000 rpm. After lysate collection, all steps were conducted at 4 °C or on ice. Triton X-100 was added to the lysate (final concentration of 0.5%) and mixed slowly followed by 15 min incubation on ice [− the total fraction aliquot (Input) was taken here]. The lysate was then centrifuged at 1,500 g for 5 min, the pellet retained, and the supernatant centrifuged at 13,000 g for 15 min, and a 500 mL aliquot taken as PM/C sample. The pellet was resuspended in 2.5 mL Honda buffer with 0.1% Triton X-100 and centrifuged again at 1,500 g for 5 min. After discarding the supernatant, the pellet was resuspended in 2.5 mL Honda buffer without Triton X-100. This suspension was transferred to two 1.5 mL Eppendorf tubes and centrifuged at 100 g for 5 min to remove remaining debris. The suspension was then centrifuged at 2,000 g for 5 min to pellet nuclei. The pellet was resuspended in 300 µL buffer G (1.7 M sucrose, 10 mM Tris-HCl pH 8.0, 0.15% Triton X-100, 2 mM MgCl2, 5 mM DTT, 1× protease inhibitor cocktail) and layered onto 300 µL buffer G in a new 1.5 mL Eppendorf tube followed by centrifugation at 16,000 g for 1 h. The supernatant was removed and the pellet resuspended in 100 µL Honda buffer and 100 µL 2× SDS loading buffer (N sample). Loading equal volumes of both PM/C and N fractions (10 µL) on a gel resulted in a 16-fold overrepresentation of the nuclear-enriched vs. PM/C sample.

Western Blot Analysis.

Two 9 mm leaf discs were frozen in liquid nitrogen and homogenized at 30 Hz for 60 s with 1 mm metal beads in a TissueLyser II (Qiagen, Netherlands). Total protein was extracted from the fine powder in 100 µL 2× Laemmli buffer (125 mM Tris-HCl pH6.8, 4% [v/v] SDS, 20% [v/v] glycerol, 200 mM DTT, 0.02% [w/v] bromophenol blue). Samples were boiled at 95 °C for 7 min and separated by SDS-PAGE (Any kD™ Mini-PROTEAN® TGX™ Precast Protein Gels, BioRad). Antibodies used for immunoblotting were α-myc (71D10, Cell Signaling), α-mCherry (E5D8F, Cell Signaling), α-GFP (11814460001, Roche), α-PEPC (Rockland 100-4163), α-H3 (ab1791, Abcam), HRP-conjugated antibodies (Anti-mouse IgG-HRP: sc2005, Santa Cruz Biotechnology; Anti-rabbit IgG-HRP: 31460, Invitrogen; Anti-rat IgG-HRP: sc2006, Santa Cruz Biotechnology). The dilution of primary and secondary antibodies was 1:5,000 (2% nonfat dry milk in TBST), except for α-PEPC, which was used at a dilution of 1:1,000. Membranes were developed using ClarityWestern ECL Substrate (Bio-Rad, USA) for low sensitivity detection and Clarity Max Western ECL Substrate (Bio-Rad, USA) for high sensitivity detection in a ChemiDoc (Bio-Rad, USA) device. For loading control, membranes were stained with Ponceau S (09276-6X1EA-F, Sigma-Aldrich).

Confocal Microscopy and Imaging.

A. tumefaciens strains carrying 35S:gGLR2.9b-GFP or 35S:gGLR2.9a-mCherry were syringe-infiltrated into leaves of Nb glr2.9ab for live cell imaging at 2 dpi. Agrobacteria carrying pGLR2.9b:gGLR2.9b-GFP or pGLR2.9a:gGLR2.9a-mCherry were syringe-infiltrated into leaves of Nb glr2.9ab for live cell imaging. At 2 dpi, the leaf zone was infiltrated with OD₆₀₀ = 0.3 Pf0-1 XopQ or Pf0-1 EV in 10 mM MgCl₂, or 10 mM MgCl₂ (mock). At 6 hpi, leaf discs were harvested for confocal microscopy imaging on a Zeiss LSM780 confocal laser scanning microscope. Wavelengths of 488 nm and 510 nm were used for excitation and detection of GFP. Wavelengths of 580 nm and 610 nm were used for excitation and detection of mCherry. Objectives used were 20× (0.8 NA, water). Confocal images were compiled using Fiji 1.53c. Two 9 mm leaf discs per sample were collected for western blot analysis.

ROS Burst Assay.

Leaf discs (~0.5 cm) were collected from 4-wk-old Nb leaves and incubated overnight in 200 μL sterile water in a 96-well plate. The discs were treated with a reaction solution containing 100 μM luminol and 1 mg/mL horseradish peroxidase supplemented with or without 1 μM flg22 peptide. Luminescence was measured using the GLOMAX96 microplate luminometer (Promega, Madison, WI).

Cytoplasmic Calcium Measurements in Nb Leaves.

GCaMP3 fluorescence signals were recorded in a temperature-controlled (22 °C) dark room using a motorized fluorescence stereo microscope Nikon SMZ25 with 0.5 × objective lens (SHR Plan Apo, WD:71, Nikon) and a high-resolution Nikon DS-Ri2 camera. A Nikon GFP-BP filter cube (EX: 470/40 and EM: 525/50) was used for excitation. GCaMP3 signals were captured every 5 min and analyzed with NIS-Elements Advanced Research microscope imaging software. For time-lapse measurements, 3-wk-old Nb were used. Immediately after the infiltration of mock, Pf0-1 EV, or Pf0-1 XopQ, detached leaves were placed on black cardboard over wet tissue and sealed in a plastic Petri dish. Measurements started after setting the exposure time to 7 s. Fluorescence was visualized using a “Plot Z-axis profile.” Fiji 1.53c was used for GCaMP3 signal analysis (59). Formula (F_t−F₀)/F₀ were used for GCaMP3 fluorescence signal analysis. F_t: signal intensity at time t. F₀: baseline signal intensity measured at t = 0.

Phylogenetic Tree Construction.

A list and amino acid sequences of Nb GLRs was obtained from Sol Genomics Network annotation file Niben101_annotation.functional.txt in the FTP database (https://solgenomics.net/). Sequences of A. thaliana GLRs were obtained from The Arabidopsis Information Resource (https://www.arabidopsis.org/). Amino acid sequences were aligned by MAFFT (v7.505) with the auto parameter, using the L-INS-i strategy. From the aligned files, the tree files were generated by IQ-TREE2 (2.2.2.1) (73, 74) AUTO parameter with the option SH-alrt and -B to bootstrap 1,000 times (75) for branch support. The phylogenetic tree was plotted with iTOL from the generated tree file.

RNA Extraction, Reverse Transcription, and qRT-PCR.

Two 9 mm leaf discs were frozen in liquid nitrogen and homogenized at 30 Hz for 60 s with 1 mm metal beads in a TissueLyser II (Qiagen, Netherlands). RNA was extracted using a my-Budget Plant RNA Kit (Bio-Budget Technologies, Germany) according to instructions. Extracted RNA was treated with DNase I to remove residual DNA. For reverse transcription, 1 µg total RNA was treated according to RevertAid First Strand cDNA Synthesis Kit instructions (Thermo Fisher Scientific, USA). One µL 12.5 ng/µL cDNA was mixed with 0.5 µL 10 mM forward primer, 0.5 µL 10 mM reverse primer and 5 µL SYBR Green (Bio-Rad iQ SYBR® Green Supermix) for qPCRs. Quantitative RT-PCR assays were performed using a CFX Connect™ Real-Time System Thermal cycler (Bio-Rad). Gene expression levels were quantified with the 2^−ΔΔCq method and NbACT2 as reference house-keeping gene.

RNA-seq Analysis.

Mock, Pf0-1 EV, and Pf0-1 XopQ (OD₆₀₀ = 0.3) were inoculated into leaves of Nb WT, nrg1-5, epss, and epssna. Nine 8 mm leaf discs were collected from three leaves of three plants per treatment to constitute one biological replicate. Three biological replicates were collected and frozen in liquid nitrogen 6 h after inoculation. Total RNA was extracted using TRIzol (15596018, Invitrogen). RNA purification was performed using a ReliaPrep™ RNA Miniprep System (Promega Z6112) following the manufacturer’s instructions. Eluted RNA was quantified on a NanoDrop spectrophotometer. Each sample was sent to Novogene for library preparation and RNA sequencing. RNA sequencing was performed using a Novaseq 6000 Illumina platform with V1.5 reagent. Raw FASTQ files were quality-controlled by the FastQC tool (v0.11.9). Quality controlled reads from raw FASTQ files were aligned to the N. benthamiana reference genome version v1.0.1 (https://solgenomics.net/). DEG were identified by using R package DESeq2 (v1.38.0). Only genes with DESeq2 normalized counts ≥ 10 in at least 3 out of 12 samples were analyzed to reduce the number of false positive DEG. Threshold cut-offs for DEG were |log2 fold change| ≥ 1 (fold change of 2) and a false discovery rate corrected P-value < 0.05.

GO Term Enrichment Analysis.

Functional annotations of genes including Gene Ontology terms (GO terms) and orthologs in other species were obtained from the Phytozome database (https://phytozome-next.jgi.doe.gov/). Shiny GO (76) was used for GO term enrichment analyses of the DEG clusters, with an enrichment threshold of P-value ≤ 0.05.

Nb GLR2.9a and GLR2.9b Protein Structure Modelling.

NbGLR2.9a and NbGLR2.9b structure models were prepared with SWISS-MODEL based on the cryo-EM structure of AtGLR3.4 (PDB: 7LZH). An NbNRG1 structure model was prepared with SWISS-MODEL based on the AtZAR1 pentamer (PDB: 6J5T). Visualization of the structures was performed in Chimera (version 1.17.1).

Software Used for Visualization and Statistical Analysis of Data.

Statistical analyses were performed in R (4.2.3). Data were first checked for normality of residuals distribution and homogeneity of variance by examining the plots and Shapiro–Wilk and Levene tests (P > 0.05). If both conditions were met, ANOVA was followed by a Tukey honestly significant difference test (α = 0.05). Otherwise, The Nemenyi test with Bonferroni correction for multiple testing was applied (α = 0.05). Heatmaps were generated with the ComplexHeatmap package. Venn diagrams were plotted by the VennDiagram (1.7.3) package. PCA was performed with the vegan (2.6.4) package. Boxplots and Ribbon plots were generated with the ggplot2 (3.4.2) package. All packages were installed in R (4.2.3).

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Data, Materials, and Software Availability

RNA-seq data generated in this study are deposited at the European Nucleotide Archive database (submissions 7-9 April, 2025) under accession code: PRJEB88173 (https://www.ebi.ac.uk/ena/browser/view/PRJEB88173) (77). Materials generated in the paper will be made available to academic researchers under an MTA. All other data are included in the manuscript and/or supporting information.