Transcriptomics and functional genomics of ROS-induced cell death regulation by RADICAL-INDUCED CELL DEATH1

Abstract

Plant responses to changes in environmental conditions are mediated by a network of signaling events leading to downstream responses, including changes in gene expression and activation of cell death programs. Arabidopsis thaliana RADICAL-INDUCED CELL DEATH1 (RCD1) has been proposed to regulate plant stress responses by protein-protein interactions with transcription factors. Furthermore, the rcd1 mutant has defective control of cell death in response to apoplastic reactive oxygen species (ROS). Combining transcriptomic and functional genomics approaches we first used microarray analysis in a time series to study changes in gene expression after apoplastic ROS treatment in rcd1. To identify a core set of cell death regulated genes, RCD1-regulated genes were clustered together with other array experiments from plants undergoing cell death or treated with various pathogens, plant hormones or other chemicals. Subsequently, selected rcd1 double mutants were constructed to further define the genetic requirements for the execution of apoplastic ROS induced cell death. Through the genetic analysis we identified WRKY70 and SGT1b as cell death regulators functioning downstream of RCD1 and show that quantitative rather than qualitative differences in gene expression related to cell death appeared to better explain the outcome. Allocation of plant energy to defenses diverts resources from growth. Recently, a plant response termed stress-induced morphogenic response (SIMR) was proposed to regulate the balance between defense and growth. Using a rcd1 double mutant collection we show that SIMR is mostly independent of the classical plant defense signaling pathways and that the redox balance is involved in development of SIMR.

Figure 1. Gene expression of Col-0 and rcd1 mutant in clean air and O₃-treated plants.

A) Experimental design with each sample hybridized against a reference RNA (solid arrows) enables multidirectional comparisons between genotypes and treatments with a linear mixed model (dotted arrows). The experiment described was repeated at each time point (0, 1, 2, 4, 8 and 24 h) and lists of differentially expressed transcripts (log₂ ratio ±1, q<0.05) were made from each comparison. Number of differentially expressed unique genes at all time points is shown. The 0 h comparison of O₃-treated rcd1 and O₃-treated Col-0 was analyzed together with rcd1 and Col-0 grown in clean air to comprise the list of genes differentially regulated in rcd1 in normal growth conditions (rcd1-Col-0). Simultaneously, this 0 h comparison was omitted from the rcd1-Col-0 O₃ data set. B) Venn diagram showing the overlap of gene lists. Transcripts at least two-fold differentially regulated between rcd1 and Col-0 in response to apoplastic ROS are divided into several subcategories discussed in the results. C, D) Transcript levels of 4544 genes responsive to O₃-treatment (log₂ ratio ±1, q<0.05) in one or both of the genotypes were compared. Genes were divided into O₃-induced (C) and O₃-repressed (D) according to the O₃-response specifically at each time point (0, 1, 2, 4, 8 and 24 h). For each gene the difference between O₃-treated rcd1 and Col-0 was calculated (log₂ ratio) and the number of genes in each range of differential expression (depicted in color) is shown on the y-axis. The percentage of genes with higher (rcd1>Col-0) and lower (rcd1<Col-0) expression in rcd1 compared to Col-0 was calculated. E) Expression of selected marker genes was studied in Col-0 and rcd1 plants with qPCR. Bars represent means of three biological repeats, error bars show standard deviation. Statistically significant difference between genotypes is depicted with asterisk (P<0.05:*; P<0.01:** and P<0.001:***).

Figure 2. Cluster analysis of genes differentially regulated in rcd1 compared to Col-0.

Bootstrapped Bayesian hierarchical clustering of 423 genes with at least two-fold changed expression (log₂ ratio ±1, q<0.05) in clean air or O₃-treated rcd1 is shown. Data sets used were rcd1 mutant grown in control conditions, O₃-treated Col-0, O₃-treated rcd1 and several other available experiments related to stress signaling and cell death (see “Materials and Methods” for the complete set of experiments). Six main clusters (I to VI) with subclusters (marked with a or b) were identified. GO and promoter element enrichment results are provided in Table S1. Magenta and green indicate increased and decreased expression as log₂ ratio compared with untreated or wild type plants, respectively.

Figure 3. Expression of selected marker genes in lesion mimic mutants.

Samples from three lesion mimic genotypes (acd2, acd5 and lsd1) 3 d after the transfer to LD were classified as samples from plants with no lesions (labelled 0); leaves with no lesions from plants with lesions (labelled −); leaves with lesions (labelled +). Averages of qPCR results (arbitrary units) from three biological replicates are shown; error bars depict standard deviation. Asterisks depict statistical significance (P<0.05) between to Col-0 (within 0 samples) or to 0 samples within the respective genotype (− and + samples).

Figure 4. Cluster analysis of gene expression in clean air and O₃-treated plants.

Three-week-old plants were treated with 6 h of O₃ (350 nL L⁻¹) and samples harvested at 2 and 8 h after the start of the O₃ exposure. Expression of selected marker genes in each genotype was studied with qPCR and bootstrapped Bayesian hierarchical clustering was applied to log₂-transformed expression values in arbitrary units (A). Gene expression in mutant genotypes was compared to the respective Col-0 sample at each time point (2 h, 8 h) and log₂-transformed fold changes were calculated for O₃ treatment (B) and control plants (C). Asterisks mark statistical significance according to the linear model (P<0.10:“.”; P<0.05:“*”; P<0.01:“**”; P<0.001:“***”.).

Figure 5. Quantification of O₃ -induced cell death in Col-0, rcd1 and various double mutants.

Plants were exposed to 6₃ (400 nL L⁻¹) and after 2 h recovery in the clean air, they were harvested for ion leakage measurements. Samples of untreated plants grown in clean air were simultaneously collected. Samples are ranked according to their ion leakage in O₃. Ion leakage percentages of O₃ samples were compared to O₃-treated Col-0 and rcd1-1 with linear models. Ion leakages of samples belonging to groups A and C did not differ from Col-0 and rcd1, respectively. Group B and C samples showed elevated O₃-damage compared to Col-0 (P<0.01), whereas samples in groups A and B had decreased O₃-damage in comparison to rcd1 (P<0.01). In clean air samples, only vtc2 differed from Col-0 (P<0.001). Bars represent means of two to seven biological repeats with standard error.

Figure 6. Response to apoplastic ROS in rcd1 is not influenced by AOX1 or UPOX.

Genotypes studied were Col-0, AOX1a OE (overexpression), AOX1a OE-CA (overexpression of constitutively active AOX1a), the corresponding vector control (VC) and the mutants aox1a, upox1, rcd1-1, rcd1-4, rcd1-1 upox1, rcd1-1 aox1a and rcd1-4 aox1a. Plants were exposed to 6 h of O₃ (350 nL L⁻¹) and after 2 hour recovery in the clean air they were harvested for ion leakage measurements. Samples of untreated plants grown in clean air were simultaneously collected. Clean air and O₃ samples were compared to wild type Col-0 with linear models (P<0.05:*; P<0.01:** and P<0.001:***). Double mutants (rcd1-1 aox1a, rcd1-4 aox1a and rcd1-1 upox1) were also compared to rcd1-1 and rcd1-4. Bars represent means of four to six biological repeats with standard deviation.

Figure 7. Constitutive SIMR in the rcd1 mutant is enhanced by ascorbate deficiency.

Plants were grown for four weeks and photos taken. Scale bar = 1 cm.