Activation of an atypical plant NLR with an N-terminal deletion initiates cell death at the vacuole

Abstract

Plants evolve nucleotide-binding leucine-rich repeat receptors (NLRs) to induce immunity. Activated coiled-coil (CC) domain containing NLRs (CNLs) oligomerize and form apparent cation channels promoting calcium influx and cell death, with the alpha-1 helix of the individual CC domains penetrating the plasma membranes. Some CNLs are characterized by putative N-myristoylation and S-acylation sites in their CC domain, potentially mediating permanent membrane association. Whether activated Potentially Membrane Localized NLRs (PMLs) mediate cell death and calcium influx in a similar way is unknown. We uncovered the cell-death function at the vacuole of an atypical but conserved Arabidopsis PML, PML5, which has a significant deletion in its CC_G10/GA domain. Active PML5 oligomers localize in Golgi membranes and the tonoplast, alter vacuolar morphology, and induce cell death, with the short N-terminus being sufficient. Mutant analysis supports a potential role of PMLs in plant immunity. PML5-like deletions are found in several Brassicales paralogs, pointing to the evolutionary importance of this innovation. PML5, with its minimal CC domain, represents the first identified CNL utilizing vacuolar-stored calcium for cell death induction.

Keywords: Cell Death and Calcium Influx; N-myristoylation and S-acylation; NLR; Resistosome; Tonoplast.

Figure 1. PML5 functions as a canonical cell death and calcium influx inducing CNL.

(A) Sequence alignment of the first 23 N-terminal amino acids of PML5 and related NLRs, highlighting the potential N-myristoylation sites in PML5. (B) Scheme of PML5 domain composition. Indicated are the potential S-acylation and N-myristoylation site and the P-loop as well as the MHD motifs. (C, D) Cell death induced by transiently expressed wild-type and mutant PML5 variants in N. benthamiana leaves (left) and corresponding protein blots (right). Leaves are shown in false color, red indicates cell death, and green healthy/alive tissue. Photos was taken 3 days after induction with 20 µM estradiol. Proteins were extracted 6 h after induction with 20 µM estradiol and detected with an anti-HA (α-HA) antibody (C) and anti-RFP (a-RFP) antibody (D). Citrine-HA was used as negative control. Bottom blot in (C) shows Citrine-HA (about 30 kDa). Ponceau S (PS) staining is shown as loading control. WT = wild-type PML5; D362V = MHD mutant; K74R = P-loop mutant; GCC(2,3,4)A = N-myristoylation and S-acylation (PTM) mutant; GCC/2,3,4)A,D362V = PTM/MHD quadruple mutant. (E) Cell death in two independent transgenic Arabidopsis lines conditionally overexpressing PML5-GFP. A 35::GFP plant was used as negative control. Four-week-old plants 30 h after induction with 20 µM estradiol are shown. Scale bars, 1 cm. (F) PML5 induces calcium ion influx in GCamp3 transgenic N. benthamiana. PML5-D362V-mCherry, PML5-K74R-mCherry, ADR1-D461V-mCherry (positive control), and P19 (negative control) expressing constructs were agroinfiltrated. Calcium influx was measured over 20 h. Plotted values are averages of 24 leaf discs from 6 leaves of individual plants, with Standard Error represented. The corresponding protein blot (bottom) with anti-RFP (a-RFP) antibody is shown on the right. Total proteins were extracted 4 h post induction with 80 µM estradiol. P19 was used as negative control. Ponceau S (PS) staining is shown as loading control. (G) PML5 cell death is blocked by LaCl₃. 2 mM LaCl₃ were co-infiltrated with PML5 and ADR1 Citrine-HA-tagged constructs. Expression was induced 20 h after infiltration and samples for protein blots were collected 5–6 h later. Protein expression was detected with an anti-HA (α-HA) antibody. Ponceau S (PS) staining is shown as loading control. Cell death was imaged 24 h after inducing the expression with 80 µM estradiol. Data information: In (F), data are represented as mean ± SEM. Source data are available online for this figure.

Figure 2. PML5 N-terminal 60 amino acids self-associate and are sufficient for cell death induction.

(A) Scheme of full-length PML5 and the generated PML5 fragments. Numbers indicate amino acid positions and length of fragments. (B) Cell death induced by transiently expressed full-length PML5 and PML5 fragments in N. benthamiana (top) and the corresponding protein blot (bottom). Leaves are shown in false color, red (top) or black/dark gray (bottom) indicates cell death and green (top) or light gray (bottom) healthy/alive tissue. Note: only full-length WT PML5 and the CC^1–60 fragment were inducing cell death symptoms. Photo was taken 2 days after induction with 20 µM estradiol. Fusion proteins were detected with an anti-HA (α-HA) antibody. Ponceau S (PS) staining is shown as loading control. Citrine-HA was used as a negative control. (C) PML5 CC^1–60 self-associates, as shown by transient expression in N. benthamiana. C-terminal Myc- and YFP-CC^1–60 fusions were co-expressed. Proteins were immunoprecipitated using anti-myc (α-MYC) beads and detected using anti-myc (α-MYC) and anti-GFP (α-GFP) antibodies. Ponceau S (PS) staining is shown as loading control. Source data are available online for this figure.

Figure 3. Full-length PML5 self-associates and forms high-molecular weight complexes.

(A–D) PML5 WT (A), D362V MHD mutant (B), K74R P-Loop mutant (C), and GCC(2,3,4)A mutant (D) self-associate, as seen after transient expression in N. benthamiana. Proteins were extracted 4 h after induction with 20 µM estradiol and immunoprecipitated using anti-RFP (α-RFP) beads and detected using anti-RFP (α-RFP) and anti-HA (α-HA) antibodies. Ponceau S (PS) staining is shown as loading control. (E) PML5 wild-type and D362V mutant form high-molecular weight complexes. Total protein from N. benthamiana leaves transiently expressing PML5 WT and PML5 mutants was extracted 6 h after induction with 20 µM estradiol, separated by BN-PAGE, and detected with an anti-HA (α-HA) antibody. Protein extracts were also separated by SDS-PAGE and proteins detected with anti-HA (α-HA) antibody. Ponceau S (PS) staining is shown as loading control. Red stars indicate potential PML5-containing complexes of different sizes. Source data are available online for this figure.

Figure 4. PML5 localizes to different cellular compartments.

(A–E) Confocal laser scanning microscopy indicates localization of transiently expressed PML5-Citrine-HA to Golgi membranes (A), the vacuolar membrane (B) and (C), and cytosol (D), but not to the plasma membrane (E). White dotted lines indicate the area used for colocalization profile analysis and the corresponding profiles are shown on the right, except for (A), where a line crossing the three circled puncta was used to extract the intensity profile. White dotted circle in (B) indicates vacuolar fragmentation. Scale bars, 10 µm; see Methods for compartment marker proteins. PML5-Citrine-HA expression was induced with 20 µM estradiol and confocal microscopy was performed 6–8 h after induction: Golgi (6 hpi), vacuolar membrane (7 hpi), cytosol (7 hpi), plasma membrane (8 hpi). (F) PML5-GFP localization in Arabidopsis elongating root cells of a transgenic line. Scale bar, 20 µm. PML5-GFP expression was induced with 20 µM estradiol and confocal microscopy was performed 18 h after induction. (G) PML5-GFP colocalizes with FM4-64 at the tonoplast. PML5-GFP expression was induced with 20 µM estradiol. Sixteen hours after induction seedlings were stained with 4 µM FM4-64 for 3 h before washed out. Scale bar, 20 µm. Confocal microscopy was performed 23 h after induction. Source data are available online for this figure.

Figure 5. pml5 mutants are slightly more susceptible to Pst DC3000 infection.

(A) Diagram of pml5 alleles. SALK T-DNA insertion (pml5-1; SALK_0691926) and CRISPR/Cas9 induced 4192 bp deletion/6 bp insertion in pml5-2c are shown. (B) Rosette of 8-week-old Arabidopsis pml5-1 and pml5-2c mutants compared to wild-type Col-0 grown in short day condition. (C) Fresh weight of 8-week-old wild-type, pml5-1, and pml5-2c plants from panel (B). Data is shown as boxplots. Data points represent 1 biological replicate with each 21 technical replicates (n = plants = 21). (D) Early flowering of 4-week-old pml5-1 and pml5-2c mutants compared to same age Col-0 wild-type. Plants were grown in long day condition. Scale bars, 2 cm. (E) No detectable PML5 transcripts in pml5-2c and reduced transcript levels in pml5-1, as determined by RT-PCR (30 cycles). ACTIN2 (25 cycles) is used as positive control and genomic DNA (gDNA) (30 cycles) as negative control. (F) Six-week-old plants, hand infiltrated with Pst DC3000 (OD₆₀₀ = 0.001) support modestly enhanced bacterial growth in pml5-1 and pml5-2c. Infiltrated leaves were collected on day 0 and day 3 post-infection for bacterial colony counts. Data are shown as boxplots and data points (colony forming units per square cm—cfu/cm²) represent two biological replicates with four technical replicates each (n8, each with six leaf samples). (G) Pst DC3000-induced disease symptoms on pml5-1 and pml5-2c. Col-0 and the eds1-12 mutants served as controls. Data information: For (C) and (F), individual data points are shown in boxplots (center line, median; bounds of box, the first and the third quartiles; whiskers, 1.5 times the interquartile range; error bar, minima and maxima. Data points with different letters indicate significant differences of P ≤ 0.05 (one-way ANOVA with a post hoc Tukey’s HSD test). Exact P values for all experiments are provided in Dataset EV1. Source data are available online for this figure.

Figure EV1. PML5 functions as a canonical cell death inducing NLR.

(A) AlphaFold 2 structural prediction of PML5. CC, NB-ARC, and LRR domains are indicated. A potential alpha 1 helix is predicted between Asp/D13 and Ile/I 27. (B) Sequence alignment of the first 26 N-terminal amino acids of Col-0 PML5, ZAR1 and N. benthamiana (Nb) NRC4, highlighting the hydrophobic and polar uncharged amino acids mutated in PML5 (see Fig. 1D). (C, D) Cell death induced by transiently expressed wild-type PML5 and mutant variants (C) constitutively under 35S promoter (top left) or conditionally expressed under a 35S promoter-controlled estradiol inducible system (bottom left). GFP (C) and Citrine-HA (D) served as negative controls. Total protein extracts were immunoblotted and detected with an anti-GFP (α-GFP) antibody. Protein expression analyses of constructs infiltrated are shown on right side with single GFP (C) and Citrine-HA (D) in the respective lower blot around 30 kDa. Cell death images were taken 2 days post-infiltration (C) or 3 days post-induction (D). Leaves are shown in false color, red indicates cell death and green healthy/alive tissue. WT = wild-type PML5; D362V = MHD mutant; K74R = P-loop mutant; GCC(2,3,4)A = N-myristoylation and S-acylation (PTM) mutant; GCC/2,3,4)A,D362V = PTM/MHD quadruple mutant. (E) Cell death phenotype in a single rosette leaf of 4-week-old Arabidopsis plant 18 h post-induction with 20 µM estradiol is shown. Leaves are shown in false color, red indicates cell death in wild-type PML5 and D362V lines, and green healthy/alive tissue in PML5 K74R and GFP negative control lines. (F) Growth restriction/cell death phenotype of two independent transgenic Arabidopsis lines conditionally overexpressing PML5-GFP (WT), D362V-GFP. A 35::GFP plant line was used as a control. 10-day-old Arabidopsis seedlings grown on ½ MS plates supplemented with or without 20 µM estradiol are shown. Scale bars, 1 cm.

Figure EV2. PML5-induced cell death is independent of EDS1, RNL helper NLRs, and NDR1.

(A) Cell death induced by transiently expressed wild-type PML5 (WT) and D362V mutant in WT, eds1, adr1, nrg1, adr1 nrg1 N. benthamiana mutant lines. The RPM1 D505V auto-active mutant was used as a positive control and Citrine-HA as a negative control. Images were taken 1 day post-induction with 20 µM estradiol. Leaves are shown in false color, red indicates cell death and green healthy/alive tissue. (B) Cell death in three independent transgenic Arabidopsis helperless mutant lines conditionally expressing PML5-GFP. PML5-GFP in Col-0 (PML5-GFP) served as the positive control. Protein blot (bottom) shows the expression of PML5-GFP detected with an anti-GFP (α-GFP) antibody in the positive control and three independent transgenic PML5-GFP helperless lines. Ponceau S (PS) staining is shown as a loading control. Images and samples for protein blot were taken 48 h post-induction with 20 µM estradiol. Leaves are shown in false color, red indicates cell death and green healthy/alive tissue. (C) Cell death in transgenic Arabidopsis eds1-12 and ndr1-1 mutant conditionally expressing PML5-GFP. PML5-GFP in Col-0 (PML5-GFP) served as the positive control. Protein blot (below) shows the expression of PML5-GFP detected with anti-GFP (α-GFP) antibody in the positive control and transgenic PML5-GFP eds1-12 and ndr1-1 mutants. Ponceau S (PS) staining is shown as a loading control. Samples for protein blot were taken 18 h post-induction (hpi) with 20 µM estradiol, cell death images were taken 48 hpi. Leaves are shown in false color, red indicates cell death and green healthy/alive tissue.

Figure EV3. PML5 N-terminal 60 amino acids are sufficient for cell death induction.

(A) Combination of PML5 CC^1–60, NB^61–383, and LRR^384–762 domains is not sufficient to induce a WT-like cell death response. PML5 domains were transiently expressed alone or in co-infiltrations in N. benthamiana. Image was taken 2 days post-induction with 20 µM estradiol. (Right panel) Total protein extracts were immunoblotted and detected with an anti-HA (α-HA) antibody. Ponceau S (PS) staining is shown as a loading control. The red asterisks indicate the different fragments expressed. Leaves are shown in false color, red indicates cell death and green healthy/alive tissue. (B) PML5 CC^1–60 self-associates in N. benthamiana transient expression. C-terminally YFP and MYC tagged CC^1–60 were co-expressed. Total proteins were immunoprecipitated using anti-GFP (α-GFP) beads and immunoblotted using an anti-GFP (α-GFP) and anti-Myc (α-myc) antibody. Ponceau S (PS) staining is shown as a loading control. (C) Bimolecular fluorescence complementation (BiFC) experiment of PML5 CC^1-60 fusion proteins, CC^1–60-cYFP^HA and CC^1–60-nYFP^MYC, transiently expressed in N. benthamiana. No YFP complementation different to the negative control could be detected for the PML5 CC^1–60 domain. RPM1 CC^1–155 served as positive control showing YFP complementation. Co-expression of cYFP^HA and nYFP^MYC with PML5 CC^1-60-nYFP^MYC and PML5 CC^1-60-cYFP^HA, respectively, served as negative controls. Scale bars, 20 µm. (D) Protein-blot analysis of total proteins from the transiently expressed proteins of the BiFC assay shown EV3C. Proteins were detected using anti-Myc (α-MYC) and anti-HA (α-HA) antibodies. Ponceau S (PS) staining is shown as a loading control. (E) Subcellular fractionation of transiently expressed PML5-Citrine-HA protein indicates a strong microsomal/membrane association. Total protein extracts were immunoblotted and detected with an anti-HA (α-HA) antibody for PML5, anti-UGPase (α-UGPase) as a cytosolic marker, and anti-Histone H3 (α-Histone H3) as a microsomal marker. T = total protein fraction; S = soluble protein fraction; M = microsomal protein fraction. (F–K) Confocal laser scanning microscopy showing subcellular localization of transiently expressed PML5-Citrine-HA mutants in comparison to different marker proteins: (F) D362V with nucleo-cytoplasmic mRFP; (G) K74R with nucleo-cytoplasmic mRFP; (H, I) GCC(2,3,4)A with nucleo-cytoplasmic mRFP and tonoplast mRFP; (J) GCC(2,3,4)A,D362V with nucleo-cytoplasmic mRFP; (K) T15E,L16E,F20E (EEE) with Golgi localized mRFP. White dotted lines indicate the area used for colocalization profile analysis and the corresponding profiles are shown on the right. n = nucleus. Scale bars, 10 µm; for information on compartment marker proteins, see the Methods section. Protein expression was induced with 20 µM estradiol and confocal imaging was performed at 6–8 h post-induction.

Figure EV4. PML5 does not contribute to resistance against avirulent Pst DC3000.

(A–M) Resistance and cell death induced upon infection with different avirulent Pst DC3000 strains is not altered in pml5 mutants. Rosette leaves of 6-week-old Arabidopsis plants were hand infiltrated with various Pst DC3000 strains using an OD₆₀₀ = 0.001: (A) Pst DC3000 AvrRpm1, (D) Pst DC3000 AvrPphB, (G) Pst DC3000 HopZ1a, (J) Pst DC3000 AvrRps4, (M) Pst DC3000 D36E and bacterial growth was determined on day 0 and day 3 post infiltration. For (A, D, G, J, M), data are shown as boxplots and data points (colony forming units per square cm—cfu/cm²) are indicated as black dots and represent 1 biological replicate with 4 technical replicates each (n = 4, with 6 leaf samples each). (B, E, H, K) Induction and (C, F, I, L) strength of cell death is not affected in pml5 mutants. The right side of the leaves was hand infiltrated with Pst DC3000: AvrRpm1 (B, C), AvrPphB (E, F), and HopZ1a (H, I) at an OD₆₀₀ = 0.1 and with Pf0-1 AvrRps4 (K, L) at an OD₆₀₀ = 0.2. Infiltrated leaves were imaged with a Typhoon laser scanner 5 h post infiltration (hpi) for AvrRpm1 (B), 22 hpi for AvrPphB (E), 24 hpi for HopZ1a (H), and AvrRps4 (K). The leaf images are shown in false color: Purple/blueish parts indicate non-infiltrated healthy tissue, and orange/yellowish dead cells. (C, F, I, L) Quantification of cell death intensity measured of infiltrated leaves similar to leaves shown in (B, E, H, K). Details on the methodology can be found in Methods section. Data are presented as boxplots. Results shown are from 1 biological replicate with 20 technical replicates for (L), 2 biological replicates with 24 technical replicates for (F) and 3 biological replicates with 28 technical replicates for (C, I). Data points of the different biological replicates are indicated by differently colored triangles. (N) Infiltration of Pst DC3000 D36E does not induce visible disease symptoms on pml5-1, pml5-2c or Col-0 and eds1-12. Data information: For (A, D, G, J, M) data are presented as boxplots (center line, median; bounds of box, the first and the third quartiles; whiskers, 1.5 times the interquartile range; error bar, minima and maxima), similarly for (C, F, I, L). There, black dots additionally represent outliers. Data points with different letters indicate significant differences of P ≤ 0.05 (one-way ANOVA with a post hoc Tukey’s HSD test). Exact P values for all experiments are provided in Dataset EV1.

Figure EV5. Loss of PML5 has no effect on PRR-induced responses and TRV resistance.

(A) Total ROS production in leaf discs of Col-0 and pml5-2c mutant treated with water (mock), 1 µM nlp20, or 100 nM flg22 over 60 min. RLU, relative light unit. Data points are indicated as gray dots from four biological replicates each with seven technical replicates (n = 28) and plotted as boxplots. The gray color box indicates mock control, green indicates nlp20 treatment and blue indicates flg22 treatment in Col-0 and pml5-2c plants. (B) Ethylene accumulation in Col-0 and pml5 mutant after 4 h treatment with water (mock), 1 µM nlp20, or 1 µM flg22. Data points are indicated as gray dots from five biological replicates each with 3 technical replicates (n = 15). (C) TRV viral accumulation was determined by qRT-PCR at 6 days post infection (dpi) (left) and 8 dpi (right) in infected Col-0, pml5-1c, and pml5-2c plants. Col-0 infected with an Empty Vector (EV) control served as negative control. Data points are from 10 different independent plants and the experiment was repeated 3 times (3 technical replicates) with similar results. Data information: Data in (A–C) are represented as boxplots (center line, median; bounds of box, the first and the third quartiles; whiskers, 1.5 times the interquartile range; error bar, minima and maxima). Statistical differences compared to the same treatment (A, B) in Col-0 were analyzed by two-sided Student’s t-test (α = 0.05) and are indicated with letters (a, P < 0.01, ‘n.s.’ not statistically different). In (C), data points with different letters indicate significant differences of P ≤ 0.05 (one-way ANOVA with a post hoc Tukey’s HSD test). Exact P values for all experiments are provided in Dataset EV1.