A phenome-based functional analysis of transcription factors in the cereal head blight fungus, Fusarium graminearum

Abstract

Fusarium graminearum is an important plant pathogen that causes head blight of major cereal crops. The fungus produces mycotoxins that are harmful to animal and human. In this study, a systematic analysis of 17 phenotypes of the mutants in 657 Fusarium graminearum genes encoding putative transcription factors (TFs) resulted in a database of over 11,000 phenotypes (phenome). This database provides comprehensive insights into how this cereal pathogen of global significance regulates traits important for growth, development, stress response, pathogenesis, and toxin production and how transcriptional regulations of these traits are interconnected. In-depth analysis of TFs involved in sexual development revealed that mutations causing defects in perithecia development frequently affect multiple other phenotypes, and the TFs associated with sexual development tend to be highly conserved in the fungal kingdom. Besides providing many new insights into understanding the function of F. graminearum TFs, this mutant library and phenome will be a valuable resource for characterizing the gene expression network in this fungus and serve as a reference for studying how different fungi have evolved to control various cellular processes at the transcriptional level.

Figure 1. Classification and deletion strategy of putative transcription factors (TFs) in Fusarium graminearum.

(A) Total TFs were classified based on nucleic acid binding domains. More than half of the total TFs are Zn₆Cys₆ zinc finger and C2H2 zinc finger proteins. (B) Each target TF gene was deleted using the split marker method based on triple homologous recombination.

Figure 2. Analysis of transcription factor (TF) mutant phenotypes.

(A) Number of TFs showing multiple mutant phenotypes. TFs without mutant phenotype were omitted. (B) Chromosome distribution of 709 putative TFs. TFs that exhibit mutant phenotypes (orange bars) and those that do not (blue bars) are randomly distributed throughout the genome. Black bars represent the TF genes that could not be disrupted. (C) Number of mutants showing mutant phenotypes in each TF subfamily. Orange and blue bars represent the number of TFs whose deletion results in a mutant phenotype or no phenotype change, respectively. Data representing the Zn₂Cys₆ subfamily are not shown in this graph. (D) Phenotype network based on PCC (Table S6). PCC values were calculated from the results of tested phenotypes. Orange edges indicate positive correlations (PCC>0.60).

Figure 3. Representative images of wild-type or transcription factor (TF) mutants.

TF mutants are defects in sexual development 10 days after sexual induction using dissecting or differential interference contrast (DIC) microscopy. Scale bar = 0.5 mm (left panel in wild-type, for dissecting microscopic pictures), 40 µm (right panel in wild-type, for DIC images).

Figure 4. Relationship between multiple phenotypes, gene conservation in fungal kingdom, and transcription factor (TF) expression levels.

(A) Relationship between BLAST score sum and number of phenotype changes caused by a single TF deletion. Statistical analysis was performed using the Fisher LSD test. *p<0.05. (B) Correlation between TF conservation and various phenotypes with and without perithecia formation. (C) Comparison of expression level and mutant phenotype TFs with and without perithecia formation. Statistical analysis for (B) and (C) was performed using the Duncan multiple range test. Values with different superscript letters are significantly different (p<0.01). (D) Relationship between BLAST score sum and the no perithecia formation phenotype. Statistical analysis was performed using the Duncan multiple range test. Values with different superscript letters are significantly different (p<0.05).

Figure 5. Transcription factors (TFs) with phenotype changes are interconnected either by co-expression or predicted protein-protein interaction (PPI).

(A) Co-expression analysis (PCC>0.75 cut-off) of Fusarium graminearum TFs using public microarray data. PCC values were calculated using the 121 F. graminearum Affymetrix array data (Table S11). (B) Predicted PPI of F. graminearum TFs using the interolog map of the S. cerevisiae PPI dataset. (C) Phenome of F. graminearum TFs was generated. (D) Interconnection of mutant phenotype TFs through integration of co-expression (orange lines) and predicted PPI (blue lines). The nodes represent TFs whose knock-out displays a mutant phenotype.