RNA-Seq analysis reveals new gene models and alternative splicing in the fungal pathogen Fusarium graminearum

Abstract

Background: The genome of Fusarium graminearum has been sequenced and annotated previously, but correct gene annotation remains a challenge. In addition, posttranscriptional regulations, such as alternative splicing and RNA editing, are poorly understood in F. graminearum. Here we took advantage of RNA-Seq to improve gene annotations and to identify alternative splicing and RNA editing in F. graminearum.

Results: We identified and revised 655 incorrectly predicted gene models, including revisions of intron predictions, intron splice sites and prediction of novel introns. 231 genes were identified with two or more alternative splice variants, mostly due to intron retention. Interestingly, the expression ratios between different transcript isoforms appeared to be developmentally regulated. Surprisingly, no RNA editing was identified in F. graminearum. Moreover, 2459 novel transcriptionally active regions (nTARs) were identified and our analysis indicates that many of these could be missed genes. Finally, we identified the 5' UTR and/or 3' UTR sequences of 7666 genes. A number of representative novel gene models and alternatively spliced genes were validated by reverse transcription polymerase chain reaction and sequencing of the generated amplicons.

Conclusions: We have developed novel and efficient strategies to identify alternatively spliced genes and incorrect gene models based on RNA-Seq data. Our study identified hundreds of alternatively spliced genes in F. graminearum and for the first time indicated that alternative splicing is developmentally regulated in filamentous fungi. In addition, hundreds of incorrect predicted gene models were identified and revised and thousands of nTARs were discovered in our study, which will be helpful for the future genomic and transcriptomic studies in F. graminearum.

Figure 1

Statistical and quality control analysis of RNA-Seq data. A. Read distribution over exons, introns, untranslated regions (UTRs) and intergenic regions. B. 84% of all reads matched to unique locations. The remaining 16% of reads matched to multiple locations in the genome of F. graminearum (2% matched to 2–10 sites and 14% to more than 10 different locations). C. The distribution of all reads matching to multiple locations in intergenic regions, UTRs and coding regions. D. The total read coverage along the gene body from 5′ to 3′ end in wt PH-1 and mutant ebr1. The genes were divided into 100 equal windows. E. Scatter plot analysis of two technical replicates from both wt PH-1 and mutant ebr1. Log2 transformed reads number of all predicted genes was used for comparison.

Figure 2

Visualization of RNA-Seq data in the CLC genomic workbench software. A. Reads from the RNA-Seq data of wt PH-1 and mutant ebr1 were mapped to the gene FGSG_04412. 168 reads in PH-1 and 216 reads in ebr1 uniquely matched to the exon of FGSG_04412. In contrast, no reads matched to the intronic regions. Y-axis represents the coverage number of reads of each nucleotide. B. Analysis of the reads matching to the EBR1 gene in wt PH-1 and mutant ebr1, respectively. 75 reads from wt PH-1 RNA-Seq data matched to EBR1, whereas no reads from mutant ebr1 RNA-Seq data matched to EBR1.

Figure 3

Identification of incorrect gene models in the F. graminearum database. A. Statistical analysis of the incorrect gene models identified from RNA-Seq data, including incorrect predictions of introns, incorrect intron splice sites, novel introns and other incorrect annotations. B. One example of a gene with incorrect intron predictions is shown. Two introns were annotated in gene FGSG_01636 in the Broad F. graminearum database, but RNA-Seq data clearly showed that the second intron is absent. C. Three genes with incorrect intron predictions were selected for confirmation by RT-PCR. For each gene, one genomic DNA template and two cDNA templates (one is from wt PH-1, one is from mutant ebr1) were tested. Primers were designed flanking the intronic regions, as shown in B. Bands with identical size were amplified from both genomic DNA and cDNA template for all three genes, indicating that predicted introns do not exist. The correctly annotated gene FGSG_10264 was used as a control.

Figure 4

Identification of alternative splicing. A. Statistical analysis of alternative splicing identified from RNA-Seq data. Four different types of alternative splicing were identified. Blue bars represent the number of genes that are alternatively spliced. Orange bars represent the total number of alternative splicing events. B. FGSG_05122 is an example of a gene showing intron retention. There are four reads representing the intron splice sites (green letters). Meanwhile, several reads uniquely match to the intronic region. Black letters represent exonic region; orange letters represent intronic region. C. Three genes with intron retention were selected for confirmation by RT-PCR. Primers were designed flanking the intronic regions. Gene FGSG_10264 was included as a control.

Figure 5

Example of alternative 3′ splicing. A. Three different 3′ intron splice sites were identified in the intron of gene FGSG_06760. There are seven reads showing the reference 3′ intron splice site, 16 reads showing alternative 3′ intron splice site 1, and one read showing alternative 3′ intron splice site 2. Black letters represent the exonic region; orange letters represent the intronic region; green letters represent different 3′ intron splice sites. B. RT-PCR confirmed the alternative 3′ splicing events in the intron of FGSG_06760. C. Protein alignment shows that 16 or 17 amino acids located between coiled coil and HMG domain are lacking in the proteins encoded by the two alternatively spliced transcripts.

Figure 6

Alternative splicing is developmentally regulated. RNA samples were isolated from wt PH-1 isolate at five different time points (0 h, 2 h, 8 h, 24 h, and 36 h after incubation of conidia in liquid CM medium-containing shake cultures). RT-PCRs were performed at the different time points for four alternatively spliced genes (FGSG_00303, FGSG_06760, FGSG_05122 and FGSG_04141). Genomic DNA template was used as control. Capital letter R represents reference transcript isoform. Capital letter A represents alternatively spliced transcript isoform. The experiment was independently performed twice with similar results.

Figure 7

Non-canonical splice sites identified in F. graminearum. A. Genes with non-canonical splice sites are shown. Capital letters represent exonic sequences, small letters represent intronic sequences. B. Analysis of nucleotides preference flanking GC donor splice site and AG acceptor splice site by using the MEME program [42]. Y-axis indicates sequence conservation at each position. The height of symbols in each position represents the relative frequency of each nucleic acid. Arrows indicate splice sites.