The Biotrophic Development of Ustilago maydis Studied by RNA-Seq Analysis

Abstract

The maize smut fungus Ustilago maydis is a model organism for elucidating host colonization strategies of biotrophic fungi. Here, we performed an in depth transcriptional profiling of the entire plant-associated development of U. maydis wild-type strains. In our analysis, we focused on fungal metabolism, nutritional strategies, secreted effectors, and regulatory networks. Secreted proteins were enriched in three distinct expression modules corresponding to stages on the plant surface, establishment of biotrophy, and induction of tumors. These modules are likely the key determinants for U. maydis virulence. With respect to nutrient utilization, we observed that expression of several nutrient transporters was tied to these virulence modules rather than being controlled by nutrient availability. We show that oligopeptide transporters likely involved in nitrogen assimilation are important virulence factors. By measuring the intramodular connectivity of transcription factors, we identified the potential drivers for the virulence modules. While known components of the b-mating type cascade emerged as inducers for the plant surface and biotrophy module, we identified a set of yet uncharacterized transcription factors as likely responsible for expression of the tumor module. We demonstrate a crucial role for leaf tumor formation and effector gene expression for one of these transcription factors.

Figure 1.

Changes in the Amount of Fungal Transcripts and Fungal Biomass during Infection. (A) Schematic view of cross sections of U. maydis infected maize leaves illustrating the stages of fungal development as well as plant tumor formation at the time points analyzed by RNA-seq-based transcriptional profiling. U. maydis infection is not synchronized, and each sample thus contains fungal transcripts from different developmental stages. Green, plant leaf tissue; brown, vascular tissue; orange, fungal cytoplasm; gray, empty fungal hyphae separated by septa; beige, plant tumor cells; rose, matrix; ornamented in black, fungal spores. (B) Amount of fungal transcripts based on the RNA-seq analysis. For each time point (0.5, 1, 2, 4, 6, 8, and 12 dpi) the ratio of reads uniquely mapped to the U. maydis genome relative to the total number of uniquely mapped reads (U. maydis and maize) was determined. Error bars denote sd of three biological replicates. (C) Fungal biomass determination based on the amount of genomic DNA. A qPCR with plant-specific (GAPDH) and fungus-specific (ppi) primers was performed using the same infected plant material that was used for the RNA-seq analysis. Data points give mean ratios of fungal DNA to plant DNA (2^−ΔCt). Error bars denote sd of three biological replicates.

Figure 2.

Assessment of the RNA-Seq Data Set of U. maydis during Infection. (A) Principal component analysis of RNA-seq data. The replicates of each developmental stage of U. maydis (axenic, 0.5, 1, 2, 4, 6, 8, and 12 dpi) form distinct clusters. (B) The expression data of the eight analyzed developmental stages of U. maydis (axenic, 0.5, 1, 2, 4, 6, 8, and 12 dpi) was the basis to extract all 28 possible contrasts. Genes with a log₂ fold change > 0.5 and adjusted P value < 0.01 were considered differentially expressed. Gray triangles depict the number of genes expressed at higher levels at the stages denoted by the horizontal labels, and yellow triangles depict the number of genes expressed at higher levels at the vertically labeled stages. In total, 5759 of the 6766 U. maydis genes were differentially expressed. (C) Modules of coexpressed genes during pathogenic development of U. maydis. The RNA-seq expression data set was subjected to WGCNA to detect modules of coexpressed genes. Each graph shows the expression of the module eigengene, which can be considered as the representative gene of the respective coexpression module. The vertical axes indicate log₂ expression values relative to the mean expression across all stages. The horizontal axes indicate the stages, i.e., axenic (ax), 0.5, 1, 2, 4, 6, 8, and 12 dpi. Error bars indicate sd of three biological replicates. The modules are named according to their color, and the number of genes residing in each module is given in parentheses.

Figure 3.

Biological Processes Enriched in Selected Coexpression Modules. GO enrichment analysis for the yellow (A), light-cyan (B), dark-green (C), red (D), magenta (E), cyan (F), light-green (G), and blue (H) modules. Only biological process terms were considered in the analysis. Each significantly enriched gene set (hypergeometric P value < 0.005) is represented by a node. Node sizes are proportional to the number of genes within the respective gene set, and the edges indicate overlapping member genes. Highly similar gene sets tend to form clusters, which were manually circled and labeled with appropriate summarizing terms. Gene sets that have no overlap with other enriched GO terms are shown in the rightmost corner and are not labeled despite two exceptions, one in (D) (secondary metabolism) and one in (F) (N-linked glycosylation). See Supplemental Data Set 8 to retrieve all GO terms of the enriched gene sets.

Figure 4.

Expression Pattern and Virulence Function of Nitrogen Compound Transporters. (A) The heat map shows the expression profiles of the U. maydis urea transporters; log₂ expression values are visualized relative to the mean expression across all stages. (B) U. maydis dur3 transporters are important for nitrogen utilization from urea. Serial dilutions of FB1 and FB2 wild-type strains and the respective dur3-1,2,3 triple mutants in FB1 (PH72 and PH109) and FB2 (PH110 and PH112) were spotted on minimal medium with ammonium or urea as sole nitrogen source in the indicated final concentrations. (C) The indicated mixtures of compatible strains were injected into maize seedlings and symptoms were scored 12 d after infection according to severity; the color code for each category is given below. Three independent experiments were performed and the average values are expressed as percentage of the total number of infected plants (n), which is given above each column. (D) The heat map shows the expression profile of the U. maydis peptide transporters; log₂ expression values are shown relative to the mean expression across all stages. (E) The indicated mixtures of compatible haploid strains were spotted on charcoal-containing agar plates. FB1 and FB2 are wild-type strains, DL755 and PH158 are compatible opt2,3,4 mutants, and PH89 and PH167 are opt2,3,4 mutants simultaneously complemented with wild-type opt2,3,4 genes. The occurrence of white mycelium indicates the formation of dikaryotic hyphae. (F) The indicated mixtures of compatible strains were injected into maize seedlings and symptoms were scored 12 d after infection according to severity as described in (C). For (C) and (F), the gene name followed by “em” indicates that the respective gene was inactivated by CRISPR-Cas9. The gene name followed by “-C” indicates that a single copy of the respective gene was introduced into the indicated strains to test for complementation. Please note that dead plants, which represent the most severe symptom category, are a result of the artificial virulence assay that is based on young maize seedlings and a high inoculum. U. maydis is a strict biotroph and does not kill plants under natural conditions.

Figure 5.

Expression of the U. maydis Secretome. The heat map shows expression of differentially expressed genes encoding putative secreted proteins. For each module indicated on the left, genes were hierarchically clustered, and log₂ expression values are visualized relative to the mean expression across all stages. Modules are colored according to Figure 2C. Black bars on the right indicate for each gene the presence of known signatures (based on InterPro scan), the predicted hydrolytic capabilities, and more specifically the predicted ability to degrade the plant cell wall (PCWDE).

Figure 6.

Heterogenous Effector Gene Expression in Fungal Aggregates. Maize seedlings were infected with mixtures of FB1 carrying the indicated reporter constructs (top) and FB2 wild-type strains. Seven days postinoculation, fungal aggregates within the tumor tissue were visualized by confocal microscopy. The fluorescence of GFP and mCherry was monitored and merged with the respective bright-field (BF) image. All images are projections of multiple z-stacks. GFP fluorescence indicative of effector gene expression was mainly detected at the surface of the aggregates, while mCherry control fluorescence was rather evenly distributed throughout the aggregates.

Figure 7.

Association of Transcription Factors with Modules Containing Secreted Proteins. (A) and (B) Network of gene expression profiles of differentially regulated genes encoding secreted proteins (A) and differentially regulated transcription factors (B) belonging to the respective modules in which the secreted proteins reside. The weighted network is based on the topological overlap matrix of the expression data, edges were included when the pairwise overlap was greater than 0.2, and genes are colored according to their modular membership. Selected transcription factors are labeled with their respective names. (C) Connectivity of transcription factors to the red (left panel), magenta (middle panel), and cyan module (right panel). Depicted are all transcription factors having an intramodular connectivity of greater than 0.9 to any of the three modules. Color intensity indicates connectivity strength.

Figure 8.

Role of U. maydis nlt1 in Virulence. (A) The heat map shows the expression profile of the U. maydis nlt1 gene; log₂ expression values are visualized relative to the mean expression across all stages. (B) The indicated mixtures of strains were injected into maize seedlings and symptoms were scored 12 d after infection according to severity; the color code for each category is given on the right. The nlt1 mutants were unable to induce tumors in leaves. nlt1em indicates that the nlt1 gene was inactivated by CRISPR-Cas9. nlt1-C indicates that a single copy of nlt1 was introduced into the indicated strains to test for complementation. Three independent experiments were performed and the average values are expressed as a percentage of the total number of infected plants (n), which is given above each column. Please note that dead plants, which represent the most severe symptom category, are the result of the virulence assay that is based on young maize seedlings and high cell density of the inoculum. U. maydis is a strict biotroph and does not kill plant tissue under natural conditions. (C) Representative leaves of infections with wild-type strains, nlt1 mutants, and the complemented strains at 4 and 8 dpi are shown. Examples of stem tumors and dead leaves observed in infections with nlt1 mutants are depicted in Supplemental Figure 6.

Figure 9.

Regulation of Effector Gene Expression by nlt1. The expression of selected effector genes during pathogenic development in crosses of FB1 and FB2 wild-type strains (circles) as well as in crosses of compatible nlt1 mutants (triangles) was measured via RT-qPCR. Six effector genes of the magenta module (leftmost two panels) and six potential effectors of the cyan module (rightmost two panels) were tested. In each graph, the expression of the respective gene in the wild-type 2 dpi samples was set to 1 and relative expression is depicted. Significant expression differences (P value < 0.01, Student’s t test) between nlt1 mutants and wild-type strains are indicated with an asterisk if applicable. Error bars denote sd of three biological replicates. Four of six genes of the cyan module required nlt1 for expression while none of the six effector genes of the magenta module was regulated by nlt1.