Developmental Dynamics of Long Noncoding RNA Expression during Sexual Fruiting Body Formation in Fusarium graminearum

Abstract

Long noncoding RNA (lncRNA) plays important roles in sexual development in eukaryotes. In filamentous fungi, however, little is known about the expression and roles of lncRNAs during fruiting body formation. By profiling developmental transcriptomes during the life cycle of the plant-pathogenic fungus Fusarium graminearum, we identified 547 lncRNAs whose expression was highly dynamic, with about 40% peaking at the meiotic stage. Many lncRNAs were found to be antisense to mRNAs, forming 300 sense-antisense pairs. Although small RNAs were produced from these overlapping loci, antisense lncRNAs appeared not to be involved in gene silencing pathways. Genome-wide analysis of small RNA clusters identified many silenced loci at the meiotic stage. However, we found transcriptionally active small RNA clusters, many of which were associated with lncRNAs. Also, we observed that many antisense lncRNAs and their respective sense transcripts were induced in parallel as the fruiting bodies matured. The nonsense-mediated decay (NMD) pathway is known to determine the fates of lncRNAs as well as mRNAs. Thus, we analyzed mutants defective in NMD and identified a subset of lncRNAs that were induced during sexual development but suppressed by NMD during vegetative growth. These results highlight the developmental stage-specific nature and functional potential of lncRNA expression in shaping the fungal fruiting bodies and provide fundamental resources for studying sexual stage-induced lncRNAs.IMPORTANCEFusarium graminearum is the causal agent of the head blight on our major staple crops, wheat and corn. The fruiting body formation on the host plants is indispensable for the disease cycle and epidemics. Long noncoding RNA (lncRNA) molecules are emerging as key regulatory components for sexual development in animals and plants. To date, however, there is a paucity of information on the roles of lncRNAs in fungal fruiting body formation. Here we characterized hundreds of lncRNAs that exhibited developmental stage-specific expression patterns during fruiting body formation. Also, we discovered that many lncRNAs were induced in parallel with their overlapping transcripts on the opposite DNA strand during sexual development. Finally, we found a subset of lncRNAs that were regulated by an RNA surveillance system during vegetative growth. This research provides fundamental genomic resources that will spur further investigations on lncRNAs that may play important roles in shaping fungal fruiting bodies.

Keywords: Fusarium graminearum; Xrn1 exonuclease; fruiting body; long noncoding RNAs; perithecia; sexual development.

FIG 1

Transcriptome of F. graminearum perithecia. (A) Emergence of new tissues at the defined developmental stages during perithecial formation (S1 to S5 [not drawn to scale]). (B) Venn diagram showing the number of differentially expressed (DE) genes between two successive developmental stages (>4-fold; 5% FDR). Note that most of the DE genes were unique in each comparison. (C) Functional enrichment analyses for DE genes between two successive developmental stages. Fifty-three GO terms—which can be broadly categorized into 7 biological processes—were assessed for degree of functional enrichment and were projected to two-dimensional semantic spaces. Only GO terms with a P value of <0.05 are depicted in each panel.

FIG 2

Genomic features of F. graminearum lncRNA. (A) Distribution of mRNA (blue bars) and lncRNA (red bars) across the four chromosomes. (B) Distribution of exon numbers per transcript. (C) Transcript length distribution. (D) A/U content of mRNA coding regions (CDS), 5′ untranslated regions (UTRs), 3' UTRs, intergenic regions, long intergenic lncRNA (lincRNA), and antisense lncRNA (ancRNA). Box and whisker plots indicate the median, interquartile range between the 25th and 75th percentiles (box), and 1.5 interquartile range (whisker). (E) Distribution of developmental stages at which mRNA and lncRNA showed the highest expression level. (F) Cumulative distributions of ratios of maximum and mean expression values across the developmental stages. Blue lines indicate mRNA, and red lines indicate lncRNA.

FIG 3

Sexual stage-induced lncRNA in F. graminearum. (A) Coexpressed clusters of lncRNAs. Trend plots of Z-score normalized expression values for lncRNAs (numbers in parentheses) in a given cluster are presented. (B) Expression distribution of mRNA (upper panel) and lncRNA (lower panel) for the sexual development transcriptome (gray boxes) and the vegetative growth transcriptome (white boxes). Box and whisker plots indicate the median, interquartile range between the 25th and 75th percentiles (box), and 1.5 interquartile range (whisker).

FIG 4

Examples of lncRNA expression across the sexual development stages. Per-base coverage of transcripts was plotted for both DNA strands in a 5-kb window. For the perithecial transcriptome data sets (S0 to S5), mapped reads of 3 biological replicates were pooled and then subsampled to 60 million reads for visual comparison of expression levels across the stages. For the degradome-seq data sets (DG), mapped reads of 2 replicates were pooled and displayed. The positions of lncRNAs (red arrows) and their neighboring genes (white arrows) are shown in the annotation track with the genome coordinate at the bottom of each panel. The genes overlapping lncRNAs on the opposite strand are labeled with abbreviated gene names in boldface: CENP-T, centromere protein T (FGRRES_16954), HIR1, histone regulatory protein 1 (FGRRES_05344); NSE4, nonstructural maintenance of chromosome element 4 (FGRRES_17018); ORC1, origin recognition complex subunit 1 (FGRRES_01336); RMD1, required for meiotic division 1 (FGRRES_06759). In relation to lncRNA position, neighboring genes are also labeled as follows: div., divergently transcribed gene on the opposite strand; conv., convergently transcribed gene on the opposite strand; up., upstream gene in tandem on the same strand.

FIG 5

lncRNA associated with small RNA-enriched loci. (A) sRNA reads mapped to mRNAs without antisense transcripts (10,928 loci), sense mRNAs for ancRNAs (295 loci), ancRNAs (276 loci), and lincRNA (235 loci) are represented as RPKM. The number of sRNA reads aligned to ancRNA loci was more than those of the other classes of transcripts (Benjamini-Hochberg adjusted P value of <0.0001; Dunn’s pairwise multiple comparisons). (B) Correlation analysis between transcript abundance and degradome tag count at the meiotic stage. Lines depict regressions for different classes of transcripts. (C) sRNA reads mapped to the top 80 sRNA clusters in different genotypes are represented as counts per million (CPM). The numbers of sRNA reads mapped to the clusters were significantly reduced in the Δdicer1/2 mutant, with double deletion of Dicer genes (FGRRES_09025 and FGRRES_04408), and the Δago1/2 mutant, with double deletion of Argonaute genes (FGRRES_16976 and FGRRES_00348) (47). (D) Fractions of sRNA reads mapped to intergenic regions (white), coding genes (blue), overlapped regions (purple), and noncoding genes (red), including lncRNAs marked with asterisks in the top 80 sRNA clusters. Overlapped regions were defined if coding or noncoding genes in the region were present on the both sides of DNA in the de novo annotations. Expression values (RPKM) for the closest coding gene (mRNA) to the center of each sRNA cluster are shown as heat maps, along with expression values for ncRNAs, if any, that were present in the same cluster.

FIG 6

Parallel induction of sense mRNA and antisense lncRNA pairs during sexual development. Expression data of sense mRNA and antisense lncRNA pairs in 18 samples for the perithecia transcriptome (S0 to S5) were reordered by the BLIND program. The RPKM values of sense mRNA and antisense lncRNA pairs with absolute Pearson’s correlation greater than 0.7 (P < 0.05; Fisher’s exact test) were clustered by Euclidean distance, and heat maps of Z-score normalized RPKM values are presented.

FIG 7

Identification of lncRNAs regulated by the NMD pathway. (A) Asexual spore (macroconidia) morphology (upper panels) and sexual spore (ascospore) morphology (lower panels). Note the size variation of ascospores in the Δxrn1 mutant. Scale bars = 20 µm. (B) Principal-component analysis of the perithecial transcriptome data from the wild type (S0 to S5 [blue circles]) and Δxrn1 transcriptome data (S0 and S4 [yellow circles]). (C) Venn diagram showing the overlap of Xrn1-sensitive unstable transcripts (XUTs) at S0 and S4 (D) Two-dimensional plots of different classes of transcripts between the wild-type (abscissa) and Δxrn1 mutant (ordinate) expression data at S0 and S4 (diagonal [gray line]). The yellow dashed line shows regression for the XUTs. (E) Expression distribution of 25 lncRNAs that were identified as XUTs at S0 and S4. Box and whisker plots indicate the median, interquartile range between the 25th and 75th percentiles (box), and 1.5 interquartile range (whisker).