Transcriptional Dynamics Driving MAMP-Triggered Immunity and Pathogen Effector-Mediated Immunosuppression in Arabidopsis Leaves Following Infection with Pseudomonas syringae pv tomato DC3000

Abstract

Transcriptional reprogramming is integral to effective plant defense. Pathogen effectors act transcriptionally and posttranscriptionally to suppress defense responses. A major challenge to understanding disease and defense responses is discriminating between transcriptional reprogramming associated with microbial-associated molecular pattern (MAMP)-triggered immunity (MTI) and that orchestrated by effectors. A high-resolution time course of genome-wide expression changes following challenge with Pseudomonas syringae pv tomato DC3000 and the nonpathogenic mutant strain DC3000hrpA- allowed us to establish causal links between the activities of pathogen effectors and suppression of MTI and infer with high confidence a range of processes specifically targeted by effectors. Analysis of this information-rich data set with a range of computational tools provided insights into the earliest transcriptional events triggered by effector delivery, regulatory mechanisms recruited, and biological processes targeted. We show that the majority of genes contributing to disease or defense are induced within 6 h postinfection, significantly before pathogen multiplication. Suppression of chloroplast-associated genes is a rapid MAMP-triggered defense response, and suppression of genes involved in chromatin assembly and induction of ubiquitin-related genes coincide with pathogen-induced abscisic acid accumulation. Specific combinations of promoter motifs are engaged in fine-tuning the MTI response and active transcriptional suppression at specific promoter configurations by P. syringae.

Figure 1.

Dynamics of Differentially Expressed Genes during MTI and Disease Development. (A) Infection dynamics reveal suppression of MAMP-triggered defense responses by DC3000 using a reporter line expressing FRK1 (At2g19190) fused to luciferase. Rapid suppression of FRK1 expression is evident between 3 and 6 hpi following DC3000 challenge, whereas FRK1 expression decreases more slowly in the DC3000hrpA-challenged leaf. (B) Dynamics of expression in Arabidopsis leaves after challenge with either DC3000 or the DC3000hrpA- mutant, representing disease and defense responses respectively. Gene expression is represented graphically at each time point by a scatterplot. The y axis indicates log₂ fold change between DC3000 infection and mock, while the x axis indicates log₂ fold change between DC3000hrpA- and mock. In these plots, green represents DEGs changing in the same direction in both virulent DC3000 and mutant DC3000hrpA- challenges compared with mock (MgCl₂) inoculation. Therefore, genes categorized as green represent MAMP response genes not modified by effectors. Red represents DEGs between DC3000 and mock challenge but not between DC3000hrpA- and mock challenge. Thus, red represents genes actively influenced (induced or repressed) by effectors relative to mock and DC3000hrpA- infection. Conversely, blue represents MAMP-responsive genes whose response is attenuated by effectors. In summary, for DEGs in one treatment relative to mock, red represents effector-driven changes relative to DC3000hrpA- treatment compared with mock; blue represents MAMP responses modified by effectors; green represents persistent MAMP responses. Violet indicates DEGs between all three treatments, and these appear late in the time course. Gene expression analysis was performed using the LIMMA package in Bioconductor using a P value cutoff of 0.05 and FDR applied using the Benjamini-Hochberg method.

Figure 2.

Time at Which Gradients of DEGs Begin to Significantly Differ between Treatments. The histograms show the times at which the gradient profile of log expression ratios of DEGs between treatment pairs first diverges from zero as determined by the gradient tool (Breeze et al., 2011). Threshold for up/downregulation is three standard deviations of the gradient being significantly non-zero (P_non-zero < 0.05). (A) Mock-subtracted expression during DC3000hrpA- infection. (B) Mock-subtracted expression during DC3000 infection. (C) DC3000hrpA-subtracted expression during DC3000 infection.

Figure 3.

Growth Curves of DC3000 and DC3000hrpA-, with Selected GO Terms Enriched by Genes Changing Expression at Indicated Time Points. Bacterial growth in log₁₀(CFU/unit leaf area) of disarmed DC3000hrpA- (A) and virulent DC3000 (B) following syringe challenge of bacteria at 10⁸ cells/mL. Asterisk represents significance growth differences between treatments as determined by Students t test (P < 0.5, n = 5; means ± sd). Growth curves are annotated with overrepresented gene ontologies of up- (red) or downregulated (blue) genes separated by the time at which the gradients of DEG profiles begin to deviate (Figure 2). Ontologies of DEGs between DC3000hrpA- to MgCl₂ treatments (A); ontologies of DEGs between DC3000 and DC3000hrpA- challenges (B). GO enrichment was determined using BiNGO (Maere et al., 2005). Growth of YFP-expressing DC3000 within Arabidopsis leaves at 4, 8, and 22 hpi corroborates growth curve data (C).

Figure 4.

Response Categories of DEGs Capturing Different MTI and ETS Profiles. Categories derived from the Venn diagram of DEGs (Supplemental Figure 4) showing direction of change. Numbers of genes falling into each category with accompanying expression plots (y axis, log relative gene expression; x axis, hpi; bars indicate se) for a representative example are shown. GO enrichments of each subcategory were established using BiNGO (Maere et al., 2005).

Figure 5.

Analysis of Effector-Specific DEG Profiles Reveals Dynamic Changes in Functionally Related Genes and Possible ETI Responses. Heat maps were generated for chromatin and ubiquitin related genes identified as differentially regulated in Category F (Supplemental Figure 4 and Supplemental Data Set 5). Genes were scaled on a per-gene basis and expression represented in blue for genes induced in DC3000hrpA- relative to DC3000 and yellow for genes that were significantly higher in DC3000 relative to DC3000hrpA-. (A) Ubiquitin-associated genes differentially regulated in DC3000-challenged leaves. Annotated genes are color coded as follows: red = E2 ligases, green = E3 ligases, and black = other related genes (COPs, SUMOs, etc.). (B) Chromatin-associated genes were strongly suppressed in DC3000-challenged leaves. Annotated histone genes are color coded as follows: green = H2A, yellow = H2B, blue = H3, red = H4, and black = H1 linker genes. (C) Very rapid and strong induction of a gene encoding a predicted TIR plant disease resistance protein, At1g72940, suggests possible early ETI responses to effectors. (D) Dynamics of expression were validated using transient expression of an At1g72940 promoter luciferase reporter fusion in N. benthamiana.

Figure 6.

Revealing Links between TF Binding Motifs and Temporal Expression Patterns. Overrepresentation of known TF binding motifs in promoters of gene clusters in three sets of expression clusters. Genes were clustered by expression in DC3000hrpA- (A), expression in DC3000 (B), and expression in DC3000hrpA- subtracted from DC3000 (C). Clusters were ordered by profile similarity. Cluster numbers are given on the horizontal axis, colored symbols indicate clusters with similar profiles, and a selected cluster expression profile of each type is plotted below. Names and sequence logo representations of TF binding motifs (where character size indicates nucleotide frequency) are shown on the vertical axis. Colored boxes correspond to P values. P values are comparable across rows and columns, i.e., not affected by cluster sizes (see Methods). Rows/columns where at least one cluster-motif pairing shows significant enrichment (P ≤ 1e⁻⁴) are shown (for full results, see Supplemental Data Set 8).​

Figure 7.

Genes Containing the Same Transcription Factor Binding Site(s) in Their Upstream Promoter Sequences Are Coexpressed across Multiple Conditions. Wigwams modules containing genes showing statistically significant coexpression across at least two of the three conditions were tested for enrichment of TF binding motifs in gene promoter sequences. Genes containing enriched motifs in their promoters were identified. In all cases, the mean expression profile of representative genes is shown (green, mock; purple, DC3000hrpA-; red, DC3000) with shading indicating sd. The binding motifs, relevant TF family, and names of key genes are provided. (A) Genes coexpressed during DC3000hrpA- and DC3000 infection and containing a MYB TF binding motif (PLACE: S-000355) in their upstream 500-bp sequences. (B) Genes coexpressed during DC3000hrpA- and DC3000 infection and containing a WRKY TF binding motif (PLACE: S-000390) in their promoters. (C) and (D) Examples of genes from Wigwams modules enriched for motifs bound by different families of TFs suggesting combinatorial TF activity regulates expression of these genes. (C) Genes coexpressed during DC3000hrpA- and DC3000 infection and containing a bZIP binding motif (M00441) and an ABI3VP1 binding motif (S-000145) in their promoters. (D) Genes coexpressed during DC3000hrpA- and DC3000 infection containing bHLH (M00435), bZIP (M00442), and TCP (S-000474) binding motifs in their upstream promoter sequences.

Figure 8.

The Inferred Transcription Factor Network Model, Jointly Obtained for Mock, DC3000hrpA-, and DC3000. The model is limited to genes deemed differentially expressed in at least two of the three pairwise differential expression comparisons and showing an early response. The visualization shows the expression levels of the genes in the network at 8 hpi in a comparison between DC3000hrpA- (A) and mock, DC3000 and mock (B), and DC3000 and DC3000hrpA- (C). Higher expression levels in the former condition always correspond to red colored nodes, while green nodes represent higher expression in the latter condition. For ease of viewing, the genes in the network were grouped based on their expression trend in the DC3000 and DC3000hrpA- delta profile, as shown in (C).

Figure 9.

The Expression Profiles of Three Genes Present in the Inferred Transcription Factor Network Model. Three genes were identified as having a high number of downstream targets among genes upregulated in DC3000 response while being downregulated in DC3000hrpA- response. The identified genes are XND1 (AT5G64530), FBH3 (AT1G51140), and AT2G40620. XND1 is thought to negatively regulate cell death, FBH3 contributes to early flowering, and AT2G40620’s function is currently unknown.