Comprehensive circRNA expression profile and hub genes screening during human liver development

Abstract

Background: Understanding the expression of non-coding RNA in the liver during embryonic development provides important insights into liver diseases. Therefore, we investigated circular RNA (circRNA) roles in human liver development, an unexplored research domain.

Methods: Using high-throughput sequencing and bioinformatics, we analysed foetal liver samples across developmental stages (7-20 weeks post-conception). Differentially expressed (DE) genes were identified and subjected to enrichment analysis using Gene Ontology (GO), Kyoto Encyclopaedia of Genes and Genomes (KEGG), and Disease Ontology (DO). Modular analysis was performed using the Search Tool for Retrieval of Interacting Genes (STRING), followed by construction of a protein-protein interaction (PPI) network using Cytoscape software. The key genes were screened using Molecular Complex Detection (MCODE). The mRNA levels of hub genes were validated using quantitative reverse transcription polymerase chain reaction (qRT-PCR).

Results: There were 645 DE circRNAs and 5,145 DE mRNAs between human livers at the three growth stages (HB, EH, and LH). It was found that the activity of circRNAs was boosted remarkably in the hepatoblastic stage. Enrichment analysis found they mainly involved in nervous system regulation of liver function, embryonic organ development and digestive system development. In addition, DE circRNAs were primarily involved in the PI3K-AKT, MAPK and calcium pathways, potentially contributing to adult liver diseases. Notably, only hsa_circ_001471 and novel_circ_017382 were simultaneously identified at all stages and were persistently downregulated. A co-expression regulatory network involving these circRNAs was established. Three hub genes (LGR5, FOXL1 and RSPO3) were identified from the PPI network of 167 genes and may play key roles in human liver development. The RT-qPCR validation results were in agreement with the sequencing data.

Conclusions: Our findings provide the first insights into the roles and regulatory networks of circRNAs in human liver development, laying the groundwork for further investigations of molecular and signalling networks.

Keywords: Human liver; circular RNAs; embryonic development; high-throughput RNA sequencing.

Figure 1.

Overview of high-throughput sequencing results of human liver development. A: Length distribution of circular RNAs (circRNAs) in human embryonic liver tissue. B: Type distribution of circRNAs in the human embryonic liver tissue. C: Principal component analysis (PCA) of different samples at three time points in human embryonic liver tissue, based on normalised circRNA expression levels. D: PCA of different samples at three time points in human embryonic liver tissue, based on normalised mRNA expression levels. E: Violin plot showing the expression characteristics during human liver tissue development over time. F: Venn diagram illustrating the expression patterns of circRNAs and overlapping circRNAs in the three groups. G: Venn diagram illustrating the expression patterns of mRNAs and overlapping mRNAs in the groups. H: Hierarchical cluster analysis of circRNAs in the nine groups; colours from green to red indicate circRNA correlation expression levels from low to high. The colour bar represents Pearson’s correlation coefficient. I: Hierarchical cluster analysis of mRNAs in the nine groups. Colours from green to red indicate mRNA expression levels from low to high. The colour bar represents Pearson’s correlation coefficient. HB, hepatoblast stage; EH, early foetal hepatocyte stage; LH, late foetal hepatocyte stage

Figure 2.

Differential expression of circRNAs and mRNAs during human liver development. A: Venn diagram illustrating the overlapping differentially expressed circRNAs (DE circRNAs) detected in the three groups. hsa _circ_001471 and novel_circ_017382 were identified in all three growth stages. B: DE circRNAs among the three groups. C–E: Volcano plots showing upregulated and downregulated circRNAs in the three groups; red dots represent upregulated circRNAs and green dots represent downregulated circRNAs. F–H: Hierarchical cluster analysis and heat maps showing the top 10 DE circRNAs in the three groups; colours from yellow to orange indicate low to high circRNA expression levels. Each column in the heat map represents a sample and each row represents a circRNA. I: Differentially expressed mRNAs (DE mRNAs) in three groups. J: Venn diagram illustrating overlapping DE mRNAs detected in the three groups. HB, hepatoblast stage; EH, early foetal hepatocyte stage; LH, late foetal hepatocyte stage

Figure 3.

Co-Expression network construction and dynamic expression of human liver circRNAs. A: Hsa_circ_001471, novel_circ_017382, and co-expressed mRNA networks were constructed based on the RNA sequencing data. B: Overall circRNA expression pattern during human embryonic liver development. C: Overall mRNA expression patterns during human embryonic liver development. Coloured squares indicate significant enrichment (p < 0.05). The profile IDs are shown in the upper-left corner, and the number of genes in the profile is shown in the lower-left corner.

Figure 4.

GO and KEGG analyses. A: Gene Ontology (GO) categories of circRNA source genes during human liver development. B: Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways of circRNA source genes during human liver development. C: Disease ontology analysis of circRNA source genes during human liver development. D: GO term enrichment associated with embryo development for DE circRNAs based on their co-expressed mRNAs. E: KEGG pathways for DE circRNAs based on coexpressed mRNAs. F: GO and KEGG pathways for hsa _circ_001471 based on co-expressed mRNAs. G: GO and KEGG pathways for novel_circ_017382 based on co-expressed mRNAs.

Figure 5.

Protein–protein interaction network construction and hub gene identification. A: Protein–protein interaction network of 167 target genes of hsa _circ_001471 and novel_circ_017382 plays an indispensable role in human liver tissue development. This network comprised 77 nodes and 123 edges. The node size increased from the smallest to the largest according to the connectivity of each gene. B: PPI network with three hub genes isolated from A using the Molecular Complex Detection method. The network comprised three nodes and three edges.

Figure 6.

qRT-PCR verification and sanger sequencing. A: Validation of DE circRNA expression (RSPO3, FREM2, GPX8, LPAR1, and KIF5A) during liver development by qRT-PCR. *p < 0.05, n = 3. B: qRT-PCR products of hsa_circ_001471 and novel_circ_017382 confirmed by Sanger sequencing. The red arrow indicates the cyclisation site. HB, hepatoblast stage; EH, early foetal hepatocyte stage; LH, late foetal hepatocyte stage.

Figure 7.

Flow chart of this study.