Background

Diarrhea is the most common disease in pig industry, which seriously endangers the health of piglets and the stable development of pig industry. Enterotoxigenic Escherichia coli (ETEC) expressing the F4 fimbriae is a major cause of diarrhea in neonatal and pre-weaned piglets [1], which leads to considerable economical loss in the pig industry. Three antigenic variants of F4 have been described: F4ab, F4ac and F4ad, of which F4ac is the most prevalent [2]. Studies have shown that the resistance and susceptibility phenotype of piglets to E. coli diarrhea is determined by the presence or absence of F4 receptors on their small intestinal epithelial cells [3]. In our previous study, ITGB5 was found to be a key gene related to the adhesion phenotypes (Small intestinal epithelial cells with or without F4 receptors) [4]. ITGB5 genoty** can effectively distinguish piglets susceptible or resistant to E. coli diarrhea [5]. The small intestinal epithelial cells of piglets with CC genotype were non-adhesive and resistant to ETEC-F4ac diarrhea, whereas piglets with TT genotype were adhesive and susceptible to ETEC-F4ac diarrhea.

Circular RNA (circRNA), a novel class of non-coding RNA, is characterized by a closed-loop structure generated by pre-mRNA back splicing [6]. Different from the traditional linear RNA (including 5 'and 3' tail), circRNA molecules have a closed ring structure and are not affected by RNA exonuclease, thus their expression is more stable and not easy to degrade [7]. Using high-throughput RNA sequencing (RNA-seq) techniques, recent results have shown that a large number of circRNAs are endogenous, stable and widely expressed in mammalian cells, often exhibiting cell type-specific, tissue-specific or developmental stage-specific expression [8,9,10]. In terms of function, recent studies have shown that circRNA molecules are rich in microRNA (miRNA) binding sites and act as miRNA sponges in cells [11], thereby relieving the inhibition of miRNA on target genes and increasing the expression level of target genes. This mechanism of action is known as the competitive endogenous RNA (ceRNA) mechanism. Li et al. [12] found that a novel circ-PPARA could promote the formation of intramuscular fat in pigs through the adsorption of miR-429 and miR-200b. In addition, circRNA are involved in the development of various disease conditions, such as cardiovascular diseases [13], diabetes [14], neurological diseases [15] and cancer [16]. Emerging evidence suggests that circRNAs may be potential new clinical diagnostic markers or therapeutic approaches for many diseases [17]. Yan et al. [18] investigated circRNA expression profiles in spleen of piglets infected with Clostridium perfringens type C, identifying eight circRNAs associated with necrotizing enteritis caused by Clostridium perfringens type C. Chen et al. [19] analyzed the circRNA expression profile during porcine endemic diarrhea virus (PEDV) infection in IPEC-J2 cell line and identified 26,670 candidate circRNAs.

In this study, we comprehensively analyzed characteristics of circRNA in the adhesive and non-adhesive small intestinal epithelial cells of piglets using RNA-seq data and bioinformatics methods, and explored the role of DE circRNAs in susceptibility to ETEC-F4ac. We also constructed ceRNA network including circRNA, miRNA and mRNA to identify molecule markers involved in susceptibility and resistance to ETEC-F4ac Diarrhea. The results of this study enhance our understanding of the role of circRNAs in regulating ETEC diarrhea resistance, and reveal the great potential of circRNA as a therapeutic target to biological treatment for ETEC diarrhea in piglets.

Results

Characterization of circRNAs in porcine small intestinal epithelial cells

A summary of RNA-seq data from eight porcine small intestinal epithelial cell samples is shown in Table 1. Two kinds of software (find_circ and CIRI2) were used for identification of circRNAs based on back-spliced reads produced from high-throughput RNA sequencing data. We identified 13,199 circRNAs with at least two independent back-spliced junction reads via two kinds of software (Additional file 1: Table S1). The obtained 13,199 circRNAs were aligned with circNet [20] and circAtlas [21] databases, among which 12,020 circRNAs (91%) were known, and 1179 circRNAs were novel (Fig. 1a). Chromosome distribution of those circRNAs is concentrated on chromosome 1, and the least on chromosome Y. Some circRNAs come from mitochondria, and the rest of the chromosomes are roughly evenly distributed (Fig. 1b). Most of the circRNAs identified were exonic circRNA (12,079, accounting for about 91.5%), and 796 circRNAs were intronic (Fig. 1e). These circRNAs were generated from 4400 genes, among which 42.95% genes produced only one circRNA isoform (Fig. 1c). The remaining 324 circRNAs originated from intergenic regions. The most circRNAs were made up of 2–3 exons, and few are more than 10 exons (Fig. 1d). The counts of the back-spliced junction reads of 8 samples were normalized as the spliced reads per billion map** (SRPBM) (Additional file 1: Table S2). The circRNAs originated from exon, intron and intergenic region showed no significant changes in expression abundance (SRPBM) (Fig. 1f). Then we aligned the flanking intron pairs of exonic circRNAs to identify reverse complementary matches (RCMs) using Basic Local Alignment Search Tool (BLAST). Among the identified RCMs, Short interspersed nuclear element (SINE) and Simple_repeat accounted for 77.5% and 12.7%, respectively (Fig. 1g).

Table 1 Summary of RNA-seq data
Full size table
Fig. 1
figure 1

Profiling of circRNA in small intestinal epithelial cells of piglets. a CircRNAs identified in piglet intestinal epithelial cells (IEC) were overlapped with circRNAs annotated in circNet and circAtlas databases. b The distribution of identified circRNAs in different chromosomes. c The distribution of host genes encoding different number of circRNAs in piglet intestinal epithelial cells. d The distribution of number of exons that form the exonic circRNAs. e The proportion of different categories of circRNAs in intestinal epithelial cells of piglets in each sample. f Cumulative distribution of different categories of circRNA expression. g The proportion of different categories of RCMs in the flanking introns of exonic circRNAs. IEC: intestinal epithelial cells; SRPBM: spliced reads per billion map**; LINE: long interspersed nuclear elements; LTR: long terminal repeat; SINE: short interspersed nuclear element

Full size image

Differentially expression circRNAs in piglet small intestinal epithelial cells that susceptible/resistant to diarrhea

We further compared the circRNAs in porcine intestinal epithelial cells of the adhesive and non-adhesive groups to identify molecular markers associated with E. coli diarrhea. No significant difference was observed in the distribution density of circRNA expression determined based on the SRPBM value obtained from the RNA-seq data of the two groups (Fig. 2a). Next, 305 differentially expressed circRNAs (DEC) were obtained by using DEseq2 R package between adhesive and non-adhesive group (Additional file 1: Table S3), of which 259 were down-regulated and 46 were up-regulated in the non-adhesive group (Fig. 2b). The clustering heatmap comparison showed some circRNAs predominately expressed in adhesive group and some mainly expressed in non-adhesive group (Fig. 2c).

Fig. 2
figure 2

DE circRNAs in porcine small intestine epithelial cell with adhesive and non-adhesive group. a Density curves of circRNA expression levels in different samples. b Volcano plots depicting DE circRNAs in adhesive group and non-adhesive group. c Heatmap showing the expression of all DE circRNAs identified in this research

Full size image

Functional enrichment analysis of differentially expressed circRNAs revealed their regulatory role in the biogenesis of piglets E. coli diarrhea

The present study performed GO and KEGG analyses of the DE circRNA host genes to elucidate their roles in ETEC-F4ac diarrhea. Functional annotation identified 171 significantly enriched GO terms and 46 significantly enriched pathways (Additional file 1: Table S4). The significantly GO terms of DE circRNA host genes were preferentially enriched in functions related to cytoskeletal components, cell junction, enzyme binding and ion transport. In addition, two pathways related to bacterial adhesion were enriched, namely gap junction and adherens junction (Fig. 3b).

Fig. 3
figure 3

Functional analysis of the DE circRNA host genes. a The Venn plot of DE circRNAs’ host genes and DE genes. b The top 15 KEGG pathways enriched using host genes of DE circRNAs (p < 0.05). c Expression levels of circRNAs produced by genes enriched in the adherens junction pathway

Full size image

Notably, circRNAs produced by the parental genes involved in the adherens junction pathway were highly expressed in the adhesive group, indicating that these circRNAs may play an important role in regulating the biogenesis of E. coli diarrhea (Fig. 3c). Then we performed an overlap analysis of DE circRNA host genes and DE genes between the adhesive group and non-adhesive group, and obtained eight overlap** genes (Fig. 3a), among which sorbin and SH3 domain-containing protein 1 (SORBS1) attracted our attention due to it was involved in the adherens junction pathway.

Construction of a ceRNA network

It was widely accepted that circRNA can act as a miRNA sponge to directly bind to miRNA, and therefore affect the expression levels of the miRNA target genes [22]. Thus, the present study predicted the targeted miRNAs of the DE circRNAs using the miRanda software, so as to explore the potential molecular sponge function of circRNA. The analysis predicted DE circRNAs binding to 451 porcine mature miRNAs and 6485 circRNA − miRNA interactions (Additional file 1: Table S5). Among them, circ-SORBS1 tends to have a lot of miRNA binding sites, indicating its potential molecular sponge function. Figure 4a and b show part of circ-SORBS1’s miRNA binding sites (Fig. 4a, b). Subsequently, 10 miRNAs were selected to construct the circRNA-miRNA interaction network (Fig. 4c), among which 5 were targeted miRNAs of the key candidate circRNA found in this study, and the other 5 selected miRNAs were found to be related to bacterial diarrhea in previous studies [RNA isolation and quality assessment

The total RNA was extracted from small intestine tissues of eight (8) selected piglets by Trizol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instruction. The kaiaoK5500®Spectrophotometer (Kaiao, Bei**g, China) was adopted to monitor RNA purity. The concentrations of isolated RNA were determined using the Nano Drop spectrophotometer. RNA integrity and concentration was assessed using the RNA Nano 6000 Assay Kit of the Bioanalyzer 2100 system (Agilent Technologies, CA, USA). The quality of all the RNA samples were good enough (OD260/280 > 1.90, RNA integrity number > 8.7) to do the sequencing. Then 20 µL of the isolated total RNA from each sample were sent to company (Annoroad Gene Technology Corporation -Bei**g) for sequencing.

Library preparation for RNA sequencing

A total amount of 2 μg RNA per sample was used as an input material for the RNA library preparations. Sequencing libraries were generated using NEBNext® Ultra™ RNA Library Prep Kit for Illumina® (#E7530L, NEB, USA) following the manufacturer's recommendations and index codes were added to attribute sequences to each sample. Briefly, mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. More details about library preparation have been described in our previously published paper [61]. The libraries were sequenced on an Illumina platform (HiSeq Xten) and 150 bp paired-end reads were generated.

Quality control for raw reads and circRNA identification

Raw reads in fastq format were firstly processed through quality control. Briefly, reads containing adapter, polyN and low-quality bases at high proportion were removed to obtain clean reads. The above steps have already been done by sequencing company. All the downstream analyses were based on the clean reads with high quality. Reference genome and annotation files were downloaded from genome website (version: Sscrofa11.1; GCA_000003025.6).

The circRNAs were detected by two software of CIRI2 [62] and find_circ [63]. The intersecting results of the two tools were considered as candidate circRNAs and used for subsequent analysis. In this study, the spliced reads per billion map** (SRPBM) method was used to estimate the circRNA expression level. The calculation formula was as follows:

$$\mathrm{SRPBM }= (\mathrm{Back spliced junction reads }/\mathrm{ Total number of mapped reads})\times 109$$

Candidate circRNAs were annotated using circAnno software [64]. To identify reverse complementary matches (RCMs), we aligned two introns’ sequences flanking the same exonic circRNA using Basic Local Alignment Search Tool (BLAST) with ‘blastn’ task, ‘-word_size 7’ and ‘-evalue 20’. Briefly, the upstream and downstream introns of an exonic circRNA were input as query sequences and subjected sequences, respectively. For each intron pair, several alignments were obtained, and the alignment with lowest e-value that passed the threshold was regarded as reverse complementary match (RCM). Then we downloaded the repetitive sequences in pig from UCSC Table Browser (http://genome.ucsc.edu/cgi-bin/hgTables) and compared it with RCM genomic coordinates to discover significant overlaps of RCMs and repeat elements.

Differential expression and functional enrichment analysis of circRNA

Differential expression (DE) analysis of adhesive and non-adhesive groups was performed using the DESeq2 R package (1.34.0) based on the negative binomial distribution [65]. Candidate circRNAs subject to criteria of p value < 0.05 and |log2(FoldChange)|> 1 were assigned as DE circRNAs between non-adhesive group and adhesive group.

Gene ontology (GO) enrichment and KEGG [66] pathway analysis for parental genes of DE circRNAs were conducted by the KOBAS software [67]. GO terms and KEGG pathway with p < 0.05 were considered to be significantly enriched.

CircRNA-miRNA-mRNA network analysis

Combining our previous mRNA data (PRJNA562774, more details can be found in previously published paper [61].) with predicted DE circRNA targeted miRNA data, we constructed a regulatory network of circRNA-miRNA-mRNA to reveal the potential association of ETEC-F4ac adhesion in small intestinal epithelial cells of piglets. All porcine mature miRNAs sequences published in miRbase [68] (http://www.mirbase.org/) were downloaded for further analysis. In details, miRNA binding sites of all DE circRNAs were predicted using miRanda software with free energies of ≤ -20.0 kcal/mol and paring score ≥ 150. Subsequently, target genes of miRNA were predicted using TargetScan (https://www.targetscan.org/vert_80/) and miRnet [69] (https://www.mirnet.ca/) databases. The intersection results of the DE genes that obtained in our mRNA data and predicted target genes were taken as the candidate target genes. Potential circRNA-miRNA-mRNA interactions were established and visualized using Cytoscape 3.9.1 software [70] (http://cytoscape.org/). Gene ontology (GO) enrichment and KEGG pathway analysis for candidate target genes were conducted by the KOBAS software. GO terms and KEGG pathway with p < 0.05 were considered to be significantly enriched.