Whole genome discovery of regulatory genes responsible for the response of chicken to heat stress

Abstract

Long noncoding RNAs (lncRNAs) are functional bridges connecting the genome with phenotypes by interacting with DNA, mRNA, and proteins. Using publically available acute heat stress (AHS)-related RNA-seq data, we discovered novel lncRNAs and tested their association with AHS along with ~ 8800 known lncRNAs and ~ 28,000 mRNA transcripts. Our pipeline discovered a total of 145 potentially novel-lncRNAs. One of them (Fishcomb_p-value = 0.06) along with another novel transcript (annotated as protein-coding; Fishcomb_p-value = 0.03) were identified as significantly associated with AHS. We found five known-lncRNAs and 134 mRNAs transcripts that were significantly associated with AHS. Four novel lncRNAs interact cis-regulated with 12 mRNA transcripts and are targeted by 11 miRNAs. Also six meta-lncRNAs associate with 134 meta-mRNAs through trans-acting co-expression, each targeted by 15 and 216 miRNAs, respectively. Three of the known-lncRNAs significantly co-expressed with almost 97 of the significant mRNAs (Pearson correlation p-value < 0.05). We report the mentioned three known-lncRNAs (ENSGALT00000099876, ENSGALT00000107573, and ENSGALT00000106323) as the most, significantly regulatory elements of AHS in chicken. It can be concluded that in order to alleviate the adverse effects of AHS on chicken, the manipulation of the three regulatory lncRNAs could lead to a more desirable result than the manipulation of the most significant mRNAs.

Keywords: Acute heat stress; Long-noncoding-RNAs; Meta-analysis; RNA-seq.

Figure 1

The analytical workflow of the four RNA-seq datasets.

Figure 2

Genome wide distribution of the discovered novel-lncRNAs.

Figure 3

Basic features of the identified 145 novel-lncRNAs in comparison with those of the known-lncRNAs and mRNAs. (A) Expression values (log10-FPKMs) of the transcripts in the liver of the heat stressed and un-challenged control chickens. (B) Length distribution of the transcripts (base pair) C. Density plot of the number of exons per transcript.

Figure 4

Protein–protein interaction (PPI) network analysis of the protein coding genes targeted by the novel-lncRNAs.

Figure 5

Protein–protein interaction (PPI) network analysis of the protein-coding genes targeted by the novel-lncRNAs and miRNAs.

Figure 6

Comparison of conservation score of novel-lncRNAs, known-lncRNAs, and protein coding transcripts. The comparison was made based on the average conservation score of exonic locations of ten transcripts within each biotype.

Figure 7

Exon–intron structures of the two identified novel transcripts.

Figure 8

The volcano plot of the expression of the 140 meta-DETs within the four datasets. From the left to right, the datasets were, dataset1, dataset2, dataset3, dataset4, respectively. Here, for simplifying the understanding of the power of meta-analysis in detecting the significant but less variable genes related to the trait of interest, the expression pattern of the differentially expressed transcripts within each of the four datasets were not shown.

Figure 9

Protein–protein interaction (PPI) network analysis of 134 meta-mRNAs. The significant differential expression of these 134 mRNA transcripts were identified following the meta-analysis of four distinct but similar RNA-seq datasets (Fishcomb p-value < 0.05). All datasets were originated from the liver transcriptome of the chickens that were challenged with acute heat stress (half of the samples) for 3–4 h in order to compare with those of un-challenged control ones.

Figure 10

The association of the identified significant gene ontology terms (Bonferroni adjusted p-value < 0.05) enriched by the 134 meta-mRNAs.

Figure 11

The regulatory networks between the ENSGALT00000099876 meta-lncRNA and its targeted 91 meta-mRNAs (the red edges). In addition, the mutual regulation of 35 meta-mRNAs by ENSGALT00000099876 meta-lncRNA and four miRNAs is illustrated with the blue, dashed edges. Some of the meta-mRNAs were under the regulation of more than one miRNAs, however in this figure only one of them were shown to make the understanding of the network easy. All of the transcripts were significantly differentially expressed between the liver tissue of chickens under acute heat stress and their control counterparts, which identified following the meta-analysis of four distinct but similar RNA-seq datasets (Fishcomb p-value < 0.05).