circRNA N6-methyladenosine methylation in preeclampsia and the potential role of N6-methyladenosine-modified circPAPPA2 in trophoblast invasion

Abstract

Here, we performed N6-methyladenosine (m6A) RNA sequencing to determine the circRNA m6A methylation changes in the placentas during the pathogenesis of preeclampsia (PE). We verified the expression of the circRNA circPAPPA2 using quantitative reverse transcription-PCR. An invasion assay was carried out to identify the role of circPAPPA2 in the development of PE. Mechanistically, we investigated the cause of the altered m6A modification of circPAPPA2 through overexpression and knockdown cell experiments, RNA immunoprecipitation, fluorescence in situ hybridization and RNA stability experiments. We found that increases in m6A-modified circRNAs are prevalent in PE placentas and that the main changes in methylation occur in the 3'UTR and near the start codon, implicating the involvement of these changes in PE development. We also found that the levels of circPAPPA2 are decreased but that m6A modification is augmented. Furthermore, we discovered that methyltransferase‑like 14 (METTL14) increases the level of circPAPPA2 m6A methylation and that insulin-like growth factor 2 mRNA-binding protein 3 (IGF2BP3) maintains circPAPPA2 stability. Decreases in IGF2BP3 levels lead to declines in circPAPPA2 levels. In summary, we provide a new vision and strategy for the study of PE pathology and report that placental circRNA m6A modification appears to be an important regulatory mechanism.

Figure 1

Results of m6A RNA sequencing data analysis with MACS and diffReps software. Venn diagram showing the overlap of m6A-modified sites on circRNAs among samples. The total number of m6A-modified sites on circRNAs in each group is shown (a P1–P3 represent three PE subgroups. b N1–N3 represent three control subgroups. MACS software was used). There were 136 upregulated m6A-methylated circRNAs and 86 downregulated m6A-methylated circRNAs in PE placentas compared to control placentas (c with diffReps Software, fold change > 2 and P < 10^–5). Hierarchical clustering analysis of the top twenty differentially expressed methylation sites (d).

Figure 2

Results of m6A RNA sequencing data analysis with the HOMER or Python annotation tool. Proportions of m6A peaks distributed in the 5′-untranslated region (5′UTR), start codon (startC), coding DNA sequence (CDS), stop codon (stopC) and 3′-untranslated region (3′UTR) across the entire set of circRNA-corresponding transcripts in PE and control samples (a, b). Density distribution of m6A peaks across mRNA transcripts (c). m6A peaks were abundant near the startC or stop codon (stopC) (black arrow). We found that the density of m6A at the 3′UTR in PE placentas was higher than that in control placentas (blue arrow).

Figure 3

Results of m6A RNA sequencing data analysis with DAVID 6.7 software. Gene Ontology (GO) enrichment (a–c) and Kyoto Encyclopedia of Genes and Genomes (KEGG) signalling pathway (d) analyses for the host genes of upregulated m6A-methylated circRNAs. (BP: biological process; CC: cellular component; MF: molecular function).

Figure 4

Volcano plots displaying distinct m6A-modified circRNAs and their statistical significance (fold change ≥ 5 and P < 0.05) (a). The main consensus motif (GAGGC) was identified in both the control and PE groups by sensitive, thorough, rapid enriched motif elicitation (STREME) of m6A sequencing data (b). m6A peaks were enriched in the CDS or 3′UTR of the PAPPA gene according to alignment visualization. The squares mark increases in m6A peaks in PE compared with control samples (IGV software, genome: Homo_sapiens_HG19) (c) (IP: m6A RNA immunoprecipitation).

Figure 5

Fluorescent in situ hybridization assays showed that circPAPPA2 is located in the placenta (a). The expression of circPAPPA2 was increased in PE placentas compared to normal control placentas, as determined by RT-PCR (PE (n): control (n) = 6:6; P < 0.001) (b). Transwell invasion assays revealed that invasive capacity was significantly lower in the circPAPPA2 shRNA (circPAPPA2-sh) group than in the normal control shRNA (NC-Sh) group; the opposite result was obtained when circPAPPA2 was overexpressed (circPAPPA2-OE) (P < 0.05). Collectively, the results indicate that circPAPPA2 may promote trophoblast invasion (c). Insulin-like growth factor 2 mRNA binding protein 3 (IGF2BP3) was downregulated in PE placentas in the immunohistochemistry assay (*P < 0.05) (d).

Figure 6

Through overexpression experiments, insulin-like growth factor 2 mRNA binding protein 3 (IGF2BP3) was found to increase circPAPPA2 expression but have no effect on the levels of m6A on circPAPPA2 (a). RNA immunoprecipitation (RIP) revealed that IGF2BP3 can bind to circPAPPA2 (b). Luciferase assay results showed that downregulation of IGF2BP3 can downregulate circPAPPA2 expression in an m6A-dependent manner (c). In an analysis of circPAPPA2 stability, after transcriptional inhibition using actinomycin D, overexpression of IGF2BP3 increased the circPAPPA2 half-life (d); knockdown of IGF2BP3 in TEV1 cells shortened the circPAPPA2 half-life (e) (all *P < 0.05).

Figure 7

METTL14 showed increased expression in PE placentas in the immunohistochemistry assay (*P < 0.05) (a). When METTL14 was overexpressed in TEV1 cells, the circPAPPA2 m6A methylation level increased. With si-METTL14, the level of circPAPPA2 m6A methylation decreased (*P < 0.05) (b). There were 4 upregulated and 7 downregulated m6A-methylated circRNAs with the potential to encode proteins according to m6A RNA sequencing data analysis using cORF software and IRESfinder software (c). circRNAs were selected based on an ORF length > 150, the presence of an IRES, and the presence of start codon regions with different methylation sites (GPR133, ALG3, GNAI1, PAPPA2, PLCE1, INVS, SLC2A11, PDE3A, FAT3, PTN, and ALPK2 represent circGPR133, circALG3, circGNAI1, circPAPPA2, circPLCE1, circINVS, circSLC2A11, circPDE3A, circFAT3, circPTN, and circALPK2, respectively).