Genomic clustering and co-regulation of transcriptional networks in the pathogenic fungus Fusarium graminearum

Abstract

Background: Genes for the production of a broad range of fungal secondary metabolites are frequently colinear. The prevalence of such gene clusters was systematically examined across the genome of the cereal pathogen Fusarium graminearum. The topological structure of transcriptional networks was also examined to investigate control mechanisms for mycotoxin biosynthesis and other processes.

Results: The genes associated with transcriptional processes were identified, and the genomic location of transcription-associated proteins (TAPs) analyzed in conjunction with the locations of genes exhibiting similar expression patterns. Highly conserved TAPs reside in regions of chromosomes with very low or no recombination, contrasting with putative regulator genes. Co-expression group profiles were used to define positionally clustered genes and a number of members of these clusters encode proteins participating in secondary metabolism. Gene expression profiles suggest there is an abundance of condition-specific transcriptional regulation. Analysis of the promoter regions of co-expressed genes showed enrichment for conserved DNA-sequence motifs. Potential global transcription factors recognising these motifs contain distinct sets of DNA-binding domains (DBDs) from those present in local regulators.

Conclusions: Proteins associated with basal transcriptional functions are encoded by genes enriched in regions of the genome with low recombination. Systematic searches revealed dispersed and compact clusters of co-expressed genes, often containing a transcription factor, and typically containing genes involved in biosynthetic pathways. Transcriptional networks exhibit a layered structure in which the position in the hierarchy of a regulator is closely linked to the DBD structural class.

Figure 1

Hierarchical clustering of the F. graminearum TAPs and their orthologues. Rows represent genomes and columns homologous TAPs. A cell in the heatmap is filled if a homologue is detected and coloured according to the TAP class (D, O, C, B, P – defined in Table 1) of the matching Fusarium sequence. “[−]” indicates no homologue was detected.

Figure 2

Chromosomal distribution of F. graminearum TAPs. (A) The genome was divided into four blocks representing high (R ≥ 8) to low (R < 1) recombination rate (cM/27kb). Chromosomal locations of the B, C, P TAPs and D, O TAPs are shown above and below, respectively, of the recombination map. (B) The observed (Obs) and expected (Exp) TAP gene counts are shown per recombination block for each TAP class, compared with a uniform distribution of TAPs over F. graminearum gene positions (χ²-test; *: p < 0.05). (C) Graph of the percentage of clade-specific DNA-binding TAPs lying in regions of high recombination (R ≥ 3). The percentage of all DNA-binding TAPs (short dotted line), and of all genes (long dotted line), lying in regions with high recombination rate are indicated.

Figure 3

Genes expressed in transcriptomics experiments. (A) Gene expression data sets used in the study, listed by PlexDB identifier and description. (B) Number of genes expressed (Detected – ‘# genes’) and the number which are TAPs (‘# TAPs’ ) within each F. graminearum transcriptomics data set, and number that are also differentially expressed (‘Detected and differentially expressed’). (C) Venn diagram [21] summarizing the distribution of differentially expressed genes across the four sets of conditions. (D) Number of TAPs in each TAP class which were detected, or detected and differentially expressed, within each experiment (B, C, D, O, P; Table 1). (E) Venn diagram showing the overlap of DNA-binding TAPs differentially expressed in the microarray experiments.

Figure 4

Co-expression groups derived from transcriptomics experiments. Column one shows the expression pattern observed in each co-expression group for a given data set, and columns two and three show the number of genes (‘# genes’) and TAPs (‘# TAPs’), respectively, with this expression profile. ‘% D-TAPs’ indicates the percentage of DNA-binding TAPs in the co-expression group. The final column shows the heatmaps corresponding to the various co-expression groups; increasing (red) and decreasing (green) expression is shown relative to the initial time point for the developmental and infection studies, and relative to complete media for the nutrient-limited conditions.

Figure 5

Methods testing for chromosomal clustering of co-expressed genes. (A) Genomic clustering of co-expressed genes. Within a given co-expression group (dark red), for each gene the distance to the nearest member was recorded. The total number (N_gprox) of genes lying within g_prox genes of a co-expressed gene was counted and compared to uniform sampling from the genome. (B) TAP-centred clusters of co-expressed genes. For each TAP (red), the genes lying within a window spanning t adjacent genes around the TAP were tested for enrichment for co-expressed genes (dark red). The window size t was increased incrementally (t = 1 to t = 20). (C) Localized clustering of co-expressed genes. Each genomic region was tested for enrichment of co-expressed genes using PGE. (D) Chromosomal locations of the seven TAP-centred (TC) and seventeen localized (LC) gene clusters detected by the TAP-window and PGE methods, respectively.

Figure 6

TAP-centred clusters. The eight TAPs (underlined) located within the seven regions significantly enriched for co-expressed genes are shown. Each cluster is displayed as three columns with the left showing the genes on the chromosome, the middle displaying genes of interest (red), and the right the members of the TC (non-DNA-binding TAPs shown in green, with D-TAPs in brown). The colour of the FG gene IDs (next to the chromosomal locations of the TC members) indicates taxonomic specificity, i.e. in which phylogenetic groups homologues are found. Alongside this identifier is given a functional annotation derived from the sequence searches and protein family detection. Abbreviations: C6-zf DBD-protein, Zn(II)₂Cys₆ DNA-binding protein.

Figure 7

Localized clusters identified by PGE. The FgraMap images for the LCs have an additional column (solid light green) indicating the extent of the LC regions. A key in the upper left hand corner indicates which LC and TC clusters have the same chromosomal locations.

Figure 8

Distribution of DNA-binding domains among F. graminearum transcriptional regulators. (A) Barplots showing the distribution of DNA-binding domains present in D-TAPs (‘All D-TAPs’), and top- and second-tier regulators. ‘#DBDs’: number of DNA-binding domains detected; p-value (displayed as ‘–log10(p)’) obtained from Fisher’s exact test when comparing the top and second-tier distributions with the background. (B) Distributions of the pTF DBDs. (C) Heatmap visualizing the Pearson correlation coefficients of the DBD abundances – the pairwise correlation between the fraction of background, top and second-tier regulator and pTF DBDs.

Figure 9

Examples of potential transcriptional networks. (A) A top-tier/’broad’-domain regulator (FGSG_12970) is shown with its putative binding site (sequence logo) in the promoters of genes comprising the FG6↓ co-expression group together with its similarity to the S. cerevisiae RIM101/PacC DNA-binding site. An arrow indicates possible interactions between the regulator and target promoters, with circles indicating second-tier regulators and their fill colours representing DBDs as described in Figure 8A. One second-tier regulator, aurR1, regulates the TC1 biosynthetic cluster (indicated by dotted line). (B) The sequence logo for a ‘narrow’-domain regulator (FGSG_08010) is displayed with its homologous UME6 binding site in budding yeast and putative target genes in the FG2.10 co-expression group and lying in the LC12 cluster.