Temporally coordinated expression of nuclear genes encoding chloroplast proteins in wheat promotes Puccinia striiformis f. sp. tritici infection |…

RNA-seq read mapping in wheat and Pst reflects the susceptibility of the interaction

We selected three bread wheat varieties (Oakley, Solstice and Santiago) previously demonstrated to display different susceptibility levels to our two selected Pst isolates (F22 and 13/14)14. We quantified visible phenotypes of pathogen infection and infection types (ITs) at 12 days post-inoculation (dpi) following the 04 scale16 (Supplementary TableS1). Oakley was fully susceptible to both Pst isolates, while Solstice was moderately susceptible to Pst isolate F22 and almost fully susceptible to Pst isolate 13/14; Santiago was resistant to Pst isolate F22 and showed moderate resistance to Pst isolate 13/14. These results confirmed the range of susceptibility/resistance exhibited by the selected wheat varieties for this study. We infected each of the three wheat varieties with each of the two Pst isolates individually (Fig.1a) and collected samples at 1, 3, 7 and 11 dpi for RNA-seq analysis, alongside mock-inoculated samples from each variety collected at 12h post-inoculation (hpi). Following quality filtering, we aligned clean reads from each of the 81 generated samples to the wheat reference genome (Refseq v1.1)17 and Pst reference genome (isolate Pst-104E18).

a Diagram of the stages of Pst development during plant infection. The time points selected for RNA-seq analyses (1, 3, 7 and 11 days post-inoculation [dpi]) are highlighted. S uredinospore, SV substomatal vesicle, IH invasive hyphae, HM haustorial mother cell, H haustorium, P pustule, G guard cell. Inspired by a schematic illustration from61. b Percentage of reads mapping to the wheat or Pst reference genomes across wheat varieties and pathogen isolates. Following quality filtering, RNA-seq reads were mapped to the Pst reference genome (isolate Pst-104E18) and wheat reference genome Refseq v1.117. Values represent an average of three independent biological replicates (independent infected plants) for each Pstvariety pair. c Principal component analysis (PCA) of wheat gene expression profiles shows that samples from all Pstvariety pairs cluster into two well-defined groups: 1) 1 dpi; and 2) all remaining time points. d Independent PCA on 1 dpi samples only (left) or remaining time points (right) illustrating the clustering of 1 dpi samples by host variety, for infection by Pst isolate F22. e Differentially expressed genes (DEGs) are more numerous at 1 dpi, with samples infected with Pst isolate 13/14 showing more isolate-specific DEGs than those infected with Pst isolate F22. The number of DEGs was defined at each time point by comparing normalised transcript abundance for each Pst-wheat interaction against the corresponding mock-inoculated control using a negative binomial regression (Wald test) in DESeq2. Genes were considered differentially expressed when q-value<0.05.

We detected similar proportions of reads mapping to the wheat and Pst reference genomes across samples collected at 1 and 3 dpi (average of 85.51.5% for wheat and<1% for Pst, Fig.1b). By 7 dpi, the percentage of reads mapping to the wheat and Pst genomes varied and reflected the degree of susceptibility between the respective varietypathogen pairs. We observed the largest differences between varieties at 11 dpi upon infection with Pst isolate F22. Indeed, while we obtained an average of 45.724.0% reads mapping to wheat and 18.711.5% to Pst for the most susceptible interaction (Pst isolate F22Oakley), the fraction of reads mapping to Pst decreased with higher host resistance. The moderately susceptible interaction (Pst isolate F22Solstice) returned 73.620% of reads mapping to wheat and 5.768.02% to Pst, compared to 87.00.92% of reads mapping to the wheat genome and 0.050.02% to Pst in the context of the most resistant interaction (Pst isolate F22Santiago). Notably, the percentages of reads mapping to the wheat genome were comparable for the SantiagoPst isolate F22 pair between early and later time points, as well as with mock-inoculated samples (87.31.8%), in agreement with the high resistance of the host to the pathogen (Fig.1b). By contrast, infection of all three varieties with Pst isolate 13/14 resulted in similar percentages of reads mapping to each reference genome (host and pathogen) at 7 and 11 dpi, although samples collected from the highly susceptible variety Solstice showed the largest percentage of reads mapping to Pst at 11 dpi relative to the other two varieties (Fig.1b). This analysis illustrates that the percentage of reads mapping to the wheat and Pst genomes at later time points reflect the degree of susceptibility of each Pstvariety interaction.

To assess the host response to Pst infection under different levels of susceptibility, we determined wheat transcript abundances at each time point for each Pstvariety interaction. We normalised our data to account for library size and samples with low read counts before conducting a principal component analysis (PCA). We generated scatterplots of the first two principal components for each Pst isolate, which identified two well-defined groups across all Pst-infected samples: (1) samples collected at 1 dpi and (2) samples collected at all remaining time points (Fig.1c). As samples collected at 1 dpi clustered separately from all others and might obscure later transcriptome patterns, we repeated the PCA by separating the 1 dpi samples from the others (Fig.1d and Supplementary Fig.S1). The scatterplot of the first two principal components for all 1 dpi samples demonstrated a clear separation by Pst isolate and wheat variety. We also noticed that separation between wheat varieties tends to follow their genetic relatedness, with Santiago grouping closely with its parent variety Oakley, whereas the unrelated Solstice variety clustered separately (Fig.1d and Supplementary Fig.S1). Analyses of the remaining time points showed a similar distribution for both Pst isolates, with mock-inoculated control samples clustering together and away from the remaining time points (3, 7 and 11 dpi). These results suggest that host transcript abundance is largely similar at 3 dpi onward irrespective of the Pst isolate or the level of susceptibility of the wheat variety used for infection.

We identified differentially expressed genes (DEGs) at the different time points by comparing normalised transcript abundance for each Pstvariety interaction against their respective mock-inoculated controls. Overall, we observed substantial overlap between DEGs from different Pstvariety pairs, ranging from 68.713.0% (standard deviation) to 59.514.2% shared between Pst F22- and Pst 13/14-infected samples. In agreement with the PCA, we detected far more DEGs at 1 dpi (q-value<0.05), with an average number of 27,9735453 DEGs across all Pstvariety interactions (Fig.1e), compared to 91251193 at 3 dpi, 13,3573305 at 7 dpi, and 13,9285222 at 11 dpi. Looking at Pst isolate-specific transcriptional responses, we determined that all wheat varieties exhibit more DEGs specific for Pst 13/14 infection than with Pst F22 (Fig.1e and Supplementary Data1 and 2). This pattern was particularly evident at 1 dpi, with 30,7331886 DEGs across the three varieties infected with Pst 13/14, of which 10,0355825 were unique to Pst 13/14. Conversely, across the three varieties infected with Pst F22 a total of 25,2136923 DEGs were identified, of which 4516 (range 9339987) were specific to Pst F22 at 1 dpi. Notably, 96.6% of all DEGs at 1 dpi in Santiago plants infected with Pst F22 were also differentially expressed in Santiago infected with Pst 13/14, despite the difference in susceptibility (resistance for Pst F22, moderately resistant for Pst 13/14).

To identify biological processes associated with variety-specific expression profiles in response to Pst infection, we generated functional enrichment networks for each Pstvariety pair (Fig.2a and Supplementary Figs.S2 and S3). Accordingly, we assigned gene ontology (GO) terms to all DEGs where possible and identified those significantly enriched in each condition (q-value>0.0005). We detected enrichment for second-level GO terms across all conditions and time points that reflected general responses to Pst infection and included GO:0009536 (plastid), GO:0009507 (chloroplast) and GO:0003824 (catalytic activity) (Fig.2a and Figs.S2 and S3). Focusing on DEGs at 1 dpi, all Pstvariety pairs showed enrichment in functions related to response to biotic stimulus, chloroplast and photosynthesis, metal binding (ironsulfur cluster binding), cell redox homoeostasis and cell metabolism, including transferase activity, hydrolase activity and phosphatase activity (Fig.2a). Looking across all wheat varieties, we identified 1494 DEGs specifically in response to infection with Pst F22 and another 8627 DEGs specific to inoculation with Pst 13/14 (Fig.2b). Functional annotation of each set of DEGs highlighted functions related to protein transport and protein localisation for those specific to Pst F22 infection (Fig. S4), while those specific to Pst 13/14 infection were related to part of the chloroplast, the chloroplast membrane and photosystems (Fig.2c).

a Functional enrichment network for each Pstvariety pair identified in samples taken at 1 dpi. Gene ontology (GO) terms were assigned to all DEGs where possible and those identified as significantly enriched (q-value<0.0005) in at least one Pst-varietal pair are represented by a node, with node sizes proportional to the number of genes annotated with the GO term. Edges indicate overlapping member genes and conservation of GO term enrichment is highlighted by node border colour. Highly similar gene sets formed clusters, which were annotated and labelled with appropriate summarising terms. b Venn diagram illustrating the extent of overlap between the number of DEGs conserved for the three wheat varieties at 1 dpi upon inoculation with Pst isolate 13/14 or Pst isolate F22. c Functional GO term enrichment analysis results for the 8627 Pst 13/14-specific DEGs. GO terms were annotated when Log(q-value)>20 (first panel) or Log(q-value)>15 (second panel). Circle size represents the number of genes annotated within the particular enriched function; circle colour represents the GO term classification: molecular function (MF, blue), biological process (BP, pink) and cellular component (CC, green).

We hypothesised that the greater number of DEGs shared across wheat varieties infected with Pst 13/14 reflects either the more homogeneous susceptible phenotypes or the stronger transcriptional reprogramming induced by this isolate. To explore this question in more detail, we built co-expression clusters for each Pstvariety pair by using the 8,627 DEGs identified at 1 dpi (Figs. S5S10). We classified the clusters into two classes based on expression profiles: (1) early upregulated clusters whose constituent genes were highly expressed at 1 dpi but returned to mock-inoculated levels by 3 dpi and (2) early downregulated clusters whose genes were expressed at lower levels than the controls at 1 dpi but returned to mock-inoculated levels by 3 dpi. For example, during the fully susceptible interaction between Santiago and Pst 13/14, we classified 2127 genes across two co-expression clusters as early upregulated and 2318 genes from two co-expression clusters as early downregulated. Using the same method in the context of the resistant interaction between Santiago and Pst F22, we identified 1826 genes across three co-expression clusters as early upregulated and 2069 genes from one co-expression cluster as early downregulated (Fig.3a).

a Example of co-expression clusters classified as containing early upregulated (red) or early downregulated (blue) genes following infection of Santiago with Pst isolates 13/14 and F22. Co-expression clusters were generated using the 8627 Pst 13/14-specific DEGs. The coloured line represents the average normalised expression of all genes in a given co-expression cluster. b, c GO terms for functionally enriched biological processes across the co-expression clusters from 8627 Pst 13/14-specific DEGs, assessed for each Pstvariety pair, and classified as early upregulated (b) or early downregulated (c) genes. Significant Log(q-value) values are represented using a 0100 scale and GO terms with Log(q-value)>5 are shown.

GO term enrichment analysis indicated that early upregulated DEGs are associated with a diverse array of cellular processes (Figs.3b and S11). All co-expression clusters for each of the three varieties infected with Pst 13/14 contained genes mainly involved in the myosin complex and peroxisomes. The resistant interaction (Pst isolate F22Santiago) was the only one associated with the NatA acetyltransferase complex, which also contained genes involved in protein deubiquitination. In terms of biological processes, early upregulated genes in the context of resistant and moderately susceptible interactions included mRNA metabolism and protein modification by small protein conjugation or removal. Susceptible interactions comprised genes involved in organelle organisation, protein transport, RNA processing, protein modification and pyridine nucleotide salvage. By contrast, we observed shared functions across all conditions for genes classified as early downregulated (Figs.3c and S12). In terms of cellular components, these co-expression clusters included genes annotated as part of the chloroplast. In agreement with this observation, photosynthesis was the main biological process enriched in all clusters, with other enriched processes such as organonitrogen compound biosynthesis, peptide metabolism and translation. Notably, the specific early downregulated genes associated with the chloroplast and involved in photosynthesis differed between each Pstvariety pair.

Among the DEGs at 1 dpi, we observed an enrichment for functions associated with defence-related responses. We selected genes participating in programmed cell death (48 genes), response to salicylic acid (SA; 59 genes), the innate immune response (179 genes), defence response to fungi (151 genes) and those predicted to encode nucleotide-binding site leucine-rich repeat (NLR)-type R proteins (9078 genes) for further analysis. We normalised their expression and determined the median value (Fig.4a, b). Most varieties exhibited a consistent upregulation of transcript levels across all categories at 1 dpi, followed by a drop in expression at 3 dpi and a later increase at 7 and 11 dpi. Importantly, the expression of genes belonging to all four defence-related response processes reaches a higher peak at later stages of infection (711 dpi) in the resistant interaction (Pst isolate F22Santiago) relative to its susceptible counterpart (Pst isolate 13/14Santiago) (Fig.4a). Turning to genes annotated as encoding potential NLRs, we detected most DEGs from this class at 1 dpi. At this time point, we identified the greatest numbers of NLR DEGs for Oakley infected with Pst 13/14 (most susceptible interaction), followed by Solstice infected with Pst 13/14 (fully susceptible) and Pst F22 (moderate susceptibility). The lowest numbers of NLR DEGs were for Santiago infected with Pst F22 (resistant interaction) and Oakley infected with Pst F22 (fully susceptible) (Fig.4b). However, we noted that at 1 dpi in Oakley infected with Pst F22, many genes involved in defence-related responses lacked the expression peak seen in other Pstvariety pairs, likely due to the peak occurring outside of the sampling timepoint in this case. Overall, our results suggest that the outcome of the hostpathogen interaction may be decided early during initial fungal colonisation.

a Median expression of normalised transcripts per million (tpm) values obtained for genes annotated as being involved in response to salicylic acid (GO:0009751), defence response to fungus (GO:0050832), innate immune response (GO:0045087) and cell death (GO:0012501). The peak in gene expression at later stages of infection (711 dpi) is more pronounced in resistant interactions (Pst isolate F22Santiago) when compared to its susceptible counterpart (Pst isolate 13/14Santiago). b The number of DEGs encoding proteins with typical NLR domains is greatest at 1 dpi, with the most DEGs at this time point identified in samples from Oakley infected with Pst 13/14 (most susceptible interaction). Typical NLR domains were defined as IPR001611:Leu-rich_rpt, IPR032675:LRR_dom_sf, IPR002182:NB-ARC, IPR027417:P-loop_NTPase. Genes were considered differentially expressed compared to the control when q-value<0.05.

Among the early downregulated genes, we noticed the presence of many genes encoding proteins with GO terms associated with the chloroplast (Fig.3c). We identified components of photosystem I (Psah2) and II (PsbQ proteins and PsbO2), enzymes from the CalvinBensonBassham cycle (pyruvate kinase [PRK], Ribose-5-phosphate isomerase [RPI], Rubisco, Fructose-bisphosphate aldolase [FBA1]), chloroplast calcium signalling components (CAS), proteins involved in chloroplast RNA metabolism (CSP41a and CSP41b) and isochorismate synthase 1 (ICS1) that synthesises SA in the chloroplasts from chorismic acid (Fig.5a). In each case, their gene expression was downregulated at 1 dpi, followed by a sharp peak in expression at 3 dpi and a second rapid decline by 7 dpi. The most resistant interaction (Pst isolate F22Santiago) was the notable sole exception across Pstvariety pairs, as the expression of many of these genes, failed to decline or declined to a lesser extent after 3 dpi than with more susceptible interactions (Supplementary Fig.S13).

a Schematic illustration of the chloroplast. The genes encoding the proteins marked with a star were identified as differentially expressed at 1 dpi across wheat varieties upon infection with Pst isolate 13/14. b Many genes are annotated with chloroplast-related functions among the 8627 Pst 13/14-specific DEGs, as 1038 DEGs belong to eight second-level GO terms with chloroplast-related functions. c Chloroplast-related DEGs show a conserved, temporally regulated expression profile during Pst infection. Normalised transcripts per million (tpm) values were used to determine the median expression levels for genes assigned to each of the eight chloroplast-related GO terms.

We explored the expression patterns of these nuclear genes encoding chloroplast-localised proteins (NGCPs) during a susceptible Pst-wheat interaction by re-examining the enriched GO terms among the 8627 Pst 13/14-specific DEGs. We obtained 1038 DEGs that belong to eight second-level GO terms with chloroplast-related functions. For each of the eight categories, we determined the genome-wide number of genes associated with each GO term, which illustrated the high proportion of chloroplast-related genes among the 8627 DEGs (26.685.7% for each GO term) (Fig.5b and Supplementary Data3). In addition, all chloroplast-related genes followed the same pattern of expression observed above, with a sharp increase in expression at 3 dpi, followed by a rapid decline by 7 dpi, except in the highly resistant interaction (Pst isolate F22Santiago; Fig.5c). This conserved gene expression profile likely reflects a well-coordinated transcriptional modulation of genes encoding chloroplast-targeted proteins upon pathogen recognition.

We selected the putative chloroplast-localised stem-loop RNA binding protein TaCSP41a among NGCPs for detailed analyses. TaCSP41a was selected due to the availability of tetraploid Kronos TILLING mutants and as CSP41 abundance has previously been linked to abiotic stress in Arabidopsis and tomato (Solanum lycopersicum)19,20. To investigate the expression pattern of CSP41 in more detail in response to biotic stress, we performed an RT-qPCR analysis of TaCSP41a transcript levels at 12 hpi, 2, 5, 9 and 11 dpi following infection of the wheat varieties Oakley, Santiago and Solstice with Pst F22. We designed primers to amplify all three TaCSP41a homoeologues simultaneously and compared expression levels between infected and mock-inoculated plants (Fig.6a). TaCSP41a was substantially more highly expressed at 12 hpi in the highly susceptible variety Oakley and expressed significantly lower levels in the highly resistant variety Santiago upon infection (Fig.6a). In all susceptible interactions, TaCSP41a was initially more highly expressed before decreasing substantially, reaching its lowest levels by 5 dpi for infected Oakley and 2 dpi for infected Solstice. These observations confirmed a link between TaCSP41a expression early during infection and the extent of susceptibility to Pst infection as shown in the RNA-seq analyses.

a TaCSP41a expression during a controlled infection time course of the wheat varieties Oakley, Solstice and Santiago with Pst isolate F22. Relative TaCSP41a expression was measured by RT-qPCR from all three homoeologous copies simultaneously and compared to mock-inoculated control plants, with the UBC4 gene used as a reference53. Two independent leaves from the same plant were pooled and three independent plants were analysed for TaCSP41a expression at each time point. Asterisks denote statistically significant differences (***p<0.005, **p<0.01, *p<0.05; 2-tailed t-test). b TaCSP41a-A co-localises with chlorophyll autofluorescence. TaCSP41a-A-GFP was transiently expressed in N. benthamiana and images were captured after 2 days. Images are representative of >10 images captured, all displaying co-localisation of TaCSP41a-A-GFP and chlorophyll autofluorescence. Left, individual TaCSP41a-A-GFP (top) and chlorophyll autofluorescence (bottom) patterns; right, merged image of TaCSP41a-A-GFP and chlorophyll autofluorescence illustrating co-localisation. Scale bars, 10m.

To test the subcellular location of TaCSP41a, we scanned the predicted protein sequence of the three homoeologues TaCSP41a-A, TaCSP41a-B, TaCSP41a-D for potential targeting signals. We detected a chloroplast targeting peptide with a high probability (>99%) in all three homoeologues (Supplementary TableS2). Encouraged by this result, we generated a fusion construct by cloning the TaCSP41a-A coding sequence in-frame and upstream of that of the green fluorescent protein (GFP) and transiently infiltrated the resulting TaCSP41a-A-GFP construct in Nicotiana benthamiana leaves. We observed GFP fluorescence in foci that co-localise with chlorophyll autofluorescence, as determined by confocal microscopy, supporting the notion that TaCSP41a is a chloroplast-resident protein (Fig.6b).

To assess the contribution of TaCSP41a to Pst-induced disease progression, we looked for tetraploid Kronos TILLING mutants21. We identified two mutant lines (Kronos3238 and Kronos3239) introducing early stop codons in the TaCSP41a-A sequence at amino acids 218 and 174 (Supplementary Fig.S14). We obtained homozygous TILLING mutant lines by self-pollination. We infected F2 homozygous progeny (TaCSP41a-AF218* and TaCSP41a-AQ174*) with Pst 13/14 and compared their disease phenotypes to the wild type (WT, cv. Kronos) and a Kronos3238 sibling carrying the wild-type allele at TaCSP41a-A (Fig.7a). Both mutant lines displayed limited sporulation and higher Pst resistance at 20 dpi, with a substantial reduction in the extent of leaf area infected by Pst, compared to both the Kronos WT and the wild-type Kronos3238 sibling (Fig.7b). Leaves of the TaCSP41a-AF218* and TaCSP41a-AQ174* mutant lines remained largely green outside of a few necrotic spots consistent with localised programmed cell death. By contrast, both WT lines were uniformly chlorotic, with low or no necrotic lesions (Fig.7a). The TaCSP41a-AQ174* mutant line displayed a stronger phenotype, with no chlorosis and only small necrotic regions in all plants tested. Together, these results demonstrate that disrupting TaCSP41a-A function promotes tolerance to Pst 13/14, indicating a role for TaCSP41a in supporting Pst disease progression.

a TaCSP41a-AF218* and TaCSP41a-AQ174* disruption mutants are more resistant to infection by Pst isolate 13/14 compared to the Kronos wild type (WT) or the Kronos ethyl methanesulfonate (EMS) mutant Kronos3238 carrying a WT allele at TaCSP41a. Images were captured at 20 dpi. b Lower rates of leaf infection in the TaCSP41a-A disruption mutants at 20 dpi, represented as box and whiskers plots. Lowercase letters denote statistically significant differences by Duncans multi-range test (p<0.05). Horizontal bars, median values; boxes, upper (Q3) and lower (Q1) quartiles; whiskers, 1.5the inter-quartile range.

Read the original:
Temporally coordinated expression of nuclear genes encoding chloroplast proteins in wheat promotes Puccinia striiformis f. sp. tritici infection |...