Genome-wide identification of long non-coding RNAs suggests a potential association with effector gene transcription in Phytophthora sojae

Abstract

Numerous long non-coding RNAs (lncRNAs) identified and characterized in mammals, plants and fungi have been found to play critical regulatory roles in biological processes. However, little is known about the role of lncRNAs in oomycete plant pathogens, which cause devastating damage to the economy and ecosystems. We used strand-specific RNA sequencing (RNA-seq) to generate a computational pipeline to identify lncRNAs in Phytophthora sojae, a model oomycete plant pathogen. In total, 940 lncRNAs with 1010 isoforms were identified from RNA-seq data obtained from four representative stages of P. sojae. The lncRNAs had shorter transcript lengths, longer exon lengths, fewer numbers of exons, lower GC content and higher minimum free energy values compared with protein-coding genes. lncRNAs in P. sojae exhibited low sequence conservation amongst oomycetes and P. sojae isolates. Transcriptional data indicated that P. sojae lncRNAs tended to be transcribed in a stage-specific manner; representative lncRNAs were validated by semi-quantitative reverse transcription-polymerase chain reaction. Phytophthora sojae lncRNAs were concentrated in gene-sparse regions, and lncRNAs were associated with secreted protein and effector coding genes. The neighbouring genes of lncRNAs encoded various effector family members, and RNA-seq data revealed a correlation between the transcription level of lncRNAs and their neighbouring genes. Our results provide the first comprehensive identification of lncRNAs in oomycetes and suggest a potential association between lncRNAs and effector genes.

Keywords: Phytophthora; RNA-seq; effector; long non-coding RNAs; transcriptional regulation.

Figure 1

Identification pipeline and characteristics of Phytophthora sojae long non‐coding RNAs (lncRNAs). (A) The computational pipeline for the identification of lncRNAs in P. sojae from strand‐specific RNA sequencing (RNA‐seq) data. aa, amino acid; CPC, Coding Potential Calculator; FPKM, fragments per kilobase of transcript per million mapped reads; nt, nucleotide; ORF, open reading frame; PE, paired‐end. (B) Diagram and abundance of transcripts in each Cufflink class code. (C) Number of P. sojae lncRNA candidates in various classes. (D) The transcript length, exon length and intron length of lncRNAs and protein‐coding genes. Values inside each box correspond to the middle 50% of the data [between the 25th (Q1) and 75th (Q3) percentiles] and the line within the box represents the median. The whisker ends reach Q3 + 1.5 × interquartile range (IQR) and Q1 – 1.5 × IQR (IQR = Q3 – Q1), respectively (namely 1.5 × IQR rule). (E) The number of exons per transcript of lncRNAs and protein‐coding genes. (F) The GC content of gene exons, intergenic regions, gene introns and lncRNAs in P. sojae.

Figure 2

Polymorphism of long non‐coding RNAs (lncRNAs) in various oomycete species and Phytophthora sojae isolates. (A) Relative conservation levels of P. sojae lncRNAs in various oomycete species. blastn was used with an e‐value cut‐off of <1e–5 to compare the lncRNA sequences with the genome sequences of other species. The blast score ratio (the query score/self‐comparison score for each lncRNA) of the best hit was calculated for each lncRNA to represent its conservation amongst species. The oomycete species included in this analysis were as follows: Phytophthora fragariae, Phytophthora rubi, Phytophthora alni, Phytophthora cambivora, Phytophthora infestans, Phytophthora capsici, Phytophthora parasitica, Phytophthora ramorum, Phytophthora litchii, Hyaloperonospora arabidopsidis, Peronospora hyoscyami f. sp. tabacina, Plasmopara halstedii, Pythium ultimum and Saprolegnia parasitica. (B) The average identity of sequences aligned with the Phytophthora genomes for the gene exons, intergenic regions, gene introns and lncRNAs. (C) The average sequence identities of non‐gene regions, various gene groups and lncRNAs in three P. sojae isolates, using P6497 as a reference.

Figure 3

Transcription level and pattern of the long non‐coding RNAs (lncRNAs). (A) Transcription level of lncRNAs and protein‐coding genes in four stages. The outliers in boxplots are defined using the 1.5 × interquartile range (IQR) rule. (B) Density plot of Shannon entropy of lncRNAs and protein‐coding genes. (C) Hierarchical clustering of transcription pattern for lncRNAs and protein‐coding genes. Genes represented in the black box were transcribed in a stage‐specific manner. Z‐score normalization was applied for the transcription level of genes in various stages. (D) RNA‐sequencing (RNA‐seq) read alignment of representative stage‐specific lncRNAs. The lncRNAs are depicted as a black bar and the arrow indicates the transcription direction. (E) Semi‐quantitative reverse transcription‐polymerase chain reaction (sqRT‐PCR) validation of representative stage‐specific and constitutively transcribed lncRNAs. The templates were cDNA synthesized from RNA obtained from mycelia (MY), zoospores (ZO), germinated cysts (GC) and 3 h after inoculation onto susceptible soybean leaves (IF3h). The primers used in first‐strand synthesis were RT Primer Mix (M), gene strand‐specific (S) or water (W). Genomic sequence (g) and no template (n) served as PCR controls. A DNA size marker (m) is shown.

Figure 4

Genomic features of long non‐coding RNAs (lncRNAs) and correlation of transcription between lncRNAs and their neighbouring genes. (A) Distribution of the intergenic region lengths in all Phytophthora sojae genes, core genes, effector genes and neighbouring genes of lncRNAs. (B) Boxplots showing the length of the 5′‐flanking intergenic region (5′‐FIR) and 3′‐FIR for all genes, effector‐coding genes and neighbouring genes of lncRNAs. The outliers are defined using the 1.5 × interquartile range (IQR) rule. (C) Proportion of the lncRNAs associated with the indicated gene families. (D) Number of lncRNA neighbour genes encoding the indicated effector family members. CRN, crinkler; NLP, necrosis‐ and ethylene‐inducing‐like protein; RxLR, N‐terminal RxLR (Arg‐Xaa‐Leu‐Arg) motif containing secreted protein; SCP, small cysteine‐rich protein; YxSL, N‐termial YxSL[RK] (Tyr‐Xaa‐Ser‐Leu [Arg/Lys]) motif containing secreted protein. (E) RNA‐sequencing (RNA‐seq) read alignment of representative lncRNAs and their neighbouring effector genes. (F) Pearson's correlation coefficient of the transcription levels in different categories of gene–gene pairs or lncRNA–gene pairs.