Abstract

Background. Cellular senescence can lead to many diseases. However, the roles and regulation of circular RNAs (circRNAs) in senescence are poorly understood.

Objectives. To investigate the altered expression pattern and mechanism of circRNA during cellular senescence and find potential targets to prevent senescence.

Materials and methods. The Arraystar Human circRNA Array and bioinformatics were used to profile the differentially expressed circRNAs in human embryonic lung fibroblasts (IMR-90) between young cells and senescent cells and quantification in the clinical materials. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed. The miRNA targets were predicted using TargetScan and miRanda.

Results. A total of 113 differentially expressed circRNAs were identified, including 109 upregulated and 4 downregulated circRNAs (fold change >2 and p-value <0.05). Real-time qualitative polymerase chain reaction (qPCR) showed that the expression levels of 4 circRNA were significantly increased in senescent cells, and that of hsa_circ_0007113 was significantly decreased, consistent with the microarray. siRNA against hsa_circ_0007113 increased p21 and p53 expression levels and β-gal staining. The hsa_circ_0007113 has a binding site for miR-515-5p, which is involved in regulating the p53/p21 signaling pathway. The expression level of hsa_circ_0007113 was also decreased in aged people.

Conclusions. The study showed an altered circRNA expression pattern in cellular senescence, which might play important roles in senescence-related physiological processes. These findings provide a new direction for studying the molecular mechanism underlying senescence and a new possibility for the treatment of senescence by modulating circRNAs.

Key words: senescence, human circRNA array, hsa_circ_0007113, hsa-miR-515-5p, P53/P21 pathway

Background

The essence of aging is cellular senescence, in which cellular functions gradually decrease or are lost, leading to a loss of tissue function. Cell senescence can be accelerated or enhanced by external environmental factors such as radiation, oxidizing agents and therapeutic agents.^{1, 2} Numerous mechanisms participate in cellular senescence, including DNA damage, telomeres, oncogenes, activated MAPK cascade, and p53 and p16^lnk4a pathways.^{3, 4, 5, 6, 7}

Non-coding RNAs play a vital role in the regulation of all cellular pathways.⁸ Recently, several studies revealed that non-coding RNAs are linked to the control of cellular senescence.^{9, 10, 11} Circular RNA (circRNA) is a special class of non-coding RNA, which has a closed circular structure protecting them from exonuclease R. Circular RNAs are conserved in evolution, stable, abundant, and show specific tissues and developmental stage expression.^{14, 15} Moreover, they have been shown to regulate microRNA (miRNA) expression at the transcriptional or post-transcriptional levels.^{12, 13} Specific circRNAs play important roles in human diseases such as cancer, stroke, ischemia, neurodegenerative disorders, and heart disease.^{16, 17, 18, 19, 20, 21, 22} However, so far, few studies have evaluated circRNA changes in cell senescence.^{21, 23, 24}

Circular RNA can bind miRNA-induced silencing complexes via the miRNA response element and affect miRNA concentration to regulate the activity of the downstream gene.^{25, 26} However, the differential expression of circRNAs during senescence has only rarely been reported. Therefore, this study aimed to use the Arraystar Human circRNA Array to detect the changes in circRNA expression profiles during cellular senescence. Next, the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) pathway analyses were carried out to anticipate the potential functions of circRNAs during senescence. These findings could provide a better understanding of cellular senescence and eventually slow down senescence associated with pathological conditions.

Objectives

We aimed to find the relationship between senescence and circRNAs with high-throughput quantitative analysis of circRNA. Furthermore, we quantified circRNAs in clinical samples to clarify the potential biomarkers of human aging, which can provide a certain clinical reference value for clinicians.

Materials and methods

Cell culture

The human embryonic lung fibroblast cell line IMR-90 was seeded in a CO₂-incubator containing 5% CO₂ at 37ºC in Dulbecco’s modified Eagle’s medium (DMEM; Hyclone, Logan, USA) containing 10% fetal bovine serum (FBS; Hyclone) and 1% of penicillin and streptomycin (Gibco, Invitrogen, Waltham, USA). IMR-90 cells were used as young cells in population doublings (PDL) between 15 and 25, and senescent fibroblasts were utilized in PDL 55–65 following additional culture time. Total cellular RNA was isolated using TRIzol reagent (Invitrogen, Waltham, USA).

Labeling and hybridization

Array hybridization and specimen labeling were carried out in accordance with the manufacturer’s instructions (Arraystar, Rockville, USA). Total RNA was digested with RNase R (Epicentre, Madison, USA), and the enriched circRNA was amplified and converted into fluorescent cRNA utilizing the random method (Arraystar Super RNA Labeling Kit; Arraystar). The RNeasy Mini Kit was then used to purify the labeled cRNAs (Qiagen, Hilden, Germany). Using a NanoDrop ND-1000, the labeled cRNAs (pmol Cy3/μg) were measured (Thermo Fisher Scientific, Waltham, USA), and 1 μg of cRNA was cleaved by adding 5 μL of a ×10 blocking agent and 1 μL of a ×25 fragmentation buffer. To dilute the labeled cRNA, 25 μL of ×2 hybridization buffer was added to the mixture, and it was incubated for 30 min at 60°C. Finally, space slides containing 50 μL of the hybridization solution were distributed and assembled on the microarray slide for the circRNA expression (Agilent Technologies, Santa Clara, USA). The slides were incubated for 17 h at 65°C in a mixed oven, and then the hybridized arrays were cleaned, fixed, and scanned using the Agilent Scanner G2505C (Agilent Technologies).

Arraystar human circRNA array analysis

The gathered array images were examined using Agilent Feature Extraction software (v. 11.0.1.1; Agilent Technologies). Then, utilizing the R software package (R Foundation for Statistical Computing, Vienna, Austria), quantile normalization and data processing were carried out (Bioconductor, Github, CRAN; https://www.bioconductor.org/). Through volcano map screening, the statistically significant differential expression of circRNAs between the 2 groups was determined, which was displayed as hierarchical clustering. Fold changes ≥2 and p-values <0.05 indicated significant differences in the circRNA expression.

Comprehensive analysis of the circRNAs-miRNAs-mRNAs networks

The software StarBase (v. 2.0; http://starbase.sysu.edu.cn) was used to anticipate the preferred miRNAs of selected circRNAs. The target miRNAs of the identified circRNAs were anticipated using Mireap, Miranda (v. 3.3a; https://cloud.oebiotech.com/task/detail/array_miranda_plot) and TargetScan (v. 7.0; http://www.targetscan.org). The circRNA-miRNA-mRNA regulation networks were analyzed using miRTarBase (v. 6.1; https://miRTarBase.cuhk.edu.cn), and the culminating correlations were clarified with Cytoscape (https://cytoscape.org/).

Gene Ontology and KEGG pathway analyses

We used the gene function classification system, GO, to determine the characteristics and functions of our genes of interest.²⁷ In the GO database (http://www.geneontology.org), all source genes are mapped to GO terms, and the determination of whether a gene fits a term is calculated using an false discovery rate (FDR) threshold of 0.05. Kyoto Encyclopedia of Genes and Genomes pathway analysis identified the significantly enriched pathways in the source genes, as compared to the whole genome background.²⁸ The calculation equation is identical to that used in GO analysis, and the cascades with FDR ≤0.05 were deemed as significant enrichment.

siRNA transfection and SA-β-galactosidase activity

The small interfering RNAs (siRNAs) employed for cell transfection were obtained from RiboBio (Guangzhou, China) with the following sequences: circRNA_0007113 siRNA (5’-CAA GUG UUG CCA ACC CAU CUG AUG GA-3’) and Ctrl siRNA (5’-AAU UCU CCG AAC GUG UCA CGU-3’). The siRNA was transfected with Lipofectamine™ 2000 (Invitrogen) at a final concentration of 100 nM. Aging-related senescence-associated (SA) β-galactosidase activity was validated using a kit purchased from Cell Signaling Technology (CST; Danvers, USA).

Human blood sample collection and real-time quantitative polymerase chain reaction analysis

Total blood specimens were obtained from 40 healthy individuals, aged 30–39 or 60–69 years (male, body mass index (BMI) = 20–26 kg/m²) who visited the hospital for routine health examinations. All participants gave their informed consent for inclusion in the study prior to participation. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of the Hospital and Medical College of Yangzhou University (approval No. YXYLL-2020-02)

TRIzol reagent was used to extract the total cellular RNA, which was then transcribed into cDNA with a Reverse Transcription Kit (Takara, Shiga, Japan). Real-time qualitative polymerase chain reaction (qPCR) was conducted using a kit following the manufacturer’s instructions (Takara Bio SYBR Green; Takara). The reaction parameters were as follows: 95°C for 30 s, then 35 amplification cycles (5 s at 95°C, 30 s at 60°C). All specimens were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and the experiment was repeated 3 times. Finally, SDS v. 1.4 software (Applied Biosystems, Foster City, USA) was used to analyze the data based on the 2^–ΔΔCt method. Origin v. 9.0 software (OriginLab, Northampton, USA) was utilized to analyze the histogram.

Protein extraction and western blotting

Radioimmunoprecipitation assay (RIPA) buffer (Beyotime Biotechnology, Shanghai, China) was used to lyse IMR90 cells, and protein was measured using a Bicinchoninic (BCA) protein assay kit (Bio-Rad, Hercules, USA). Total protein (35 μg) was isolated using 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene difluoride (PVDF) membranes (MilliporeSigma, St. Louis, USA). The membranes were blocked with 5% skimmed milk in Tris-buffered saline with Tween (TBST) buffer, and then the primary anti-p53 and anti-p21 antibodies were introduced and incubated overnight at 4°C (Santa Cruz Biotechnology, Santa Cruz, USA). Membranes were washed thrice in TBST, and incubated with the secondary horseradish peroxidase (HRP)-conjugated antibodies (1:5,000 in TBST; Beijing Zhong-Shan Biotechnology, Beijing, China) for 1 h at room temperature. The protein was exposed using an improved chemiluminescence reagent (Millipore Sigma), and the reactive bands were analyzed for relative intensity using ImageJ software v. 1.46 (National Institutes of Health, Bethesda, USA).

Statistical analyses

All observations in this study were made in triplicate, and the results were analyzed using GraphPad Prism v. 8 software (GraphPad Software Inc., San Diego, USA). The one-sample nonparametric test was used to compare the difference between candidate circRNA and its corresponding control. The Shapiro–Wilk test and Bartlett test were used to determine the variables’ normality, whereas homogeneity and a t-test (if not specifically indicated) or Wilcoxon Mann–Whitney test were used to compare the differences to control samples. A value of p < 0.05 was considered statistically significant.

Results

Differential expression profile analysis of circRNA between young and senescent human embryonic lung fibroblasts

A total of 3,936 circRNAs were identified using the circRNA microarray and displayed as hierarchical clustering (Figure 1A) and box plot analysis (Figure 1B). Scatter (Figure 1C) and volcano plots (Figure 1D) were used to identify differences in circRNAs between young and senescent cells. Among them, as shown in Table 1, Table 2, 113 differentially expressed circular RNAs (circRNAs) were analyzed; 109 circRNAs were upregulated, and 4 circRNAs were downregulated, with p < 0.05 and |log2(fold change)| >1. More than 80% of the differentially expressed circRNAs belong to exonic circRNA, which is exclusively composed of exons.

Gene Ontology and Kyoto Encyclopedia of Genes and Genomes pathway analyses

The mRNAs produced from the parent genes of 113 changed circRNAs were examined using GO and KEGG pathway analysis to hypothesize the pathological and physiological significance of circRNAs throughout cellular senescence. The main supplemented and meaningful GO terminologies and biological process (BP) were ‘‘establishment or maintenance of cell polarity’’, ‘‘regulation of translational initiation” and ‘‘ATPase activity’’. In terms of molecular function (MF), it was found that most of the circRNA-associated mRNAs were in “ATP binding” and “protein binding” states. As for cellular components (CC), the most enriched CC terms were ‘‘cytosol’’ and ‘‘cytoplasm’’ (Figure 2A). Gene-enriched KEGG cascade analysis demonstrated that these pathways may be linked with the progression of aging. Most of these circRNAs are target genes linked with Parkinson’s disease, p53 signaling pathway and RNA transport (Figure 2B).

Differentially expressed circRNAs’ evaluation with qPCR

Nine significantly upregulated and 4 significantly downregulated circRNAs were chosen for qPCR confirmation. Primers are shown in Table 3. We found that hsa_circRNA_0083335, hsa_circRNA_0068464, hsa_circRNA_0009361 and hsa_circ_0001303 expression levels were significantly elevated in senescent cells, while the hsa_circ_0007113 and hsa_circ_0092374 expression levels were significantly decreased, which was in line with the microarray results (Figure 3).

Cellular senescence was exacerbated after siRNA treatment of hsa_circ_0007113

Following the administration of siRNA to silence hsa_circ_0007113, the expression levels of p21 and p53 were significantly elevated (Figure 4AB). Moreover, the signal was significantly increased after β-gal staining, indicating that cell senescence was exacerbated after siRNA treatment against hsa_circ_0007113 (Figure 5A–C).

hsa_circ_0007113 could alleviate cellular senescence via miR-515-5p

CircRNAs can serve as “sponges” for miRNAs to control target gene expression. Therefore, we established a regulation network of circRNAs-microRNAs-mRNAs. The hsa_circ_0007113 is anticipated to bind with 275 miRNAs, thereby controlling a wide array of genes. It was found that hsa_circ_0007113 has a binding site for miR-515-5p, which is involved in the regulation of p53/p21 signaling pathway (Figure 6). Five miRNAs, including hsa-miR-515-5p, hsa_miR-1301-3p, hsa-miR-181c-5p, hsa-miR-22-5p, and hsa-miR-141-5p, were selected and were shown to target cellular processes including inflammation (IL17A, IL11RA, IL17AC), energy metabolism (ATP2B1, ATP7B, ATP12A), cell apoptosis (BCL2, BCL2L13, CASP9, CASP7, CASP10), cell senescence (SIRT2, GPR17, GPR135), cell cycle and division (CDK11A, CDK6), and zinc finger protein (ZNF784, ZNF655, ZNF589).

The expression of hsa_circ_0007113 decreases with aging

The relative expression of hsa_circ_0007113 was measured in 40 healthy individuals, 30–39 and 60–69 years old, and was shown to be higher in the 30–39-year-old group compared to the 60–69-year-old group (Figure 7).

Discussion

Age-linked diseases like osteoarthritis, atherosclerosis, cancer, Parkinson’s disease, Alzheimer’s disease, and type 2 diabetes are all impacted by cellular senescence.^{16, 17, 18, 19, 20, 21, 22, 29, 30} Therefore, understanding the regulatory mechanism of cellular senescence may enable interventions in these aging-related diseases. The main signaling pathways controlling cellular senescence from a mechanical perspective are P53-P21^CIP1 and P16^INK4A-Rb. Thus, it is thought that P53 and P16^INK4A are crucial elements in the induction of cellular senescence.³¹

Circular RNAS are important molecules involved in many biological processes, play important roles in regulating many cellular functions, and are expected to be biomarkers or treatment targets for diseases. However, their role in cellular senescence and the mechanisms of how circRNAs regulate it have not been previously described.

Therefore, the present study focused on the examination of alterations in circRNA expression profiles in senescence using the Arraystar Human circRNA Array, showing they are altered during cellular senescence. There were 109 upregulated and 4 downregulated circRNAs, possibly playing important roles in senescence-related physiological processes. Most circRNAs related to aging showed a trend towards increased expression, and we speculate that this may be due to the tendency of circRNA to accumulate in aging tissues, which in turn leads to their upregulation. In the microarray results, the most upregulated and downregulated circRNA were hsa_circ_0052318 (fold = 5.78) and hsa_circ_00052730 (fold = –5.48), respectively. However, the differential expression of circRNA detected was not as obvious as it is in cancer. This may be because cancer is a pathological process and is affected by drug stimulation, while aging is an overall slower, chronic and progressive physiological process. As this study revealed differentially expressed circRNA in human embryonic lung fibroblasts, our results may offer a new avenue for studying the molecular mechanism underlying senescence, and novel opportunities for senescence medication through the modulation of circRNAs.

Among the differentially expressed circRNA, circRNA_0007113, a downregulated circRNA, was selected, and the pathological phenotype associated with cellular senescence was investigated. The parent gene of circRNA_0007113 is ubiquitin ligase E3 (HERC4). hsa_circ_0007113 was obtained from exons 19 to 23 of the HERC4 gene. E3 ubiquitin ligase, a novel protein only discovered in recent years, plays a vital role in the ubiquitin-protease system due to its substrate recognition specificity and is also inextricably linked to cellular aging.³² Currently, there are limited data on its related functions, such as participating in lung, cervical, breast and liver cancer, and participating in the incidence and development of other tumors as well,³³ but its role in the aging process has not been reported. To begin preliminary investigations on the potential effect of circRNA_0007113 in cellular senescence, loss-of-function experiments were conducted using siRNA silencing. The results demonstrated that reducing circRNA_0007113 expression significantly increased the p53 and p21 protein expression levels, which are well known to trigger cell senescence.³⁴ These results were also confirmed with β-gal staining.³⁵ The role of circRNA_0007113 in cellular senescence has not been reported yet, and no other studies are available for comparison. Still, circRNAs have been shown to be involved in all cellular processes, including senescence.^{23, 24, 25, 26} The present study also suggests that hsa_circRNA_0007113 is involved in total body senescence, as hsa_circRNA_0007113 levels were lower in older individuals than in younger participants. Therefore, additional studies are necessary to confirm and refine the potential mechanisms.

Competing endogenous RNAs analyses showed that circRNAs modulate miRNA target gene expression. Therefore, a bioinformatics analysis was performed, which suggested that hsa_circ_0007113 has a binding site for miR-515-5p. The miR-515-5p was also reported to be involved in the p53/p21 pathway,^{36, 37} supporting the association of hsa_circ_0007113 with cell senescence. Indeed, the levels of p53 and p21 were increased in senescent cells, and studies showed that p53 expression is necessary for the maintenance of senescence.^{38, 39, 40, 41, 42} However, it is noteworthy that decreasing p53 and p21 expression in senescent cells leads to the restoration of the cell cycle and immortalization.^{38, 39, 40, 41, 42} Therefore, silencing hsa_circ_0007113 would increase miR-515-5p levels, leading to higher levels of p53 and p21, supporting the hypothesis that a reduction in hsa_circ_0007113 is a hallmark of cellular senescence. Notably, KEGG pathway analysis showed p53 signaling to be enriched under these circumstances. Thus, hsa_circ_0007113 could bind miR-515-5p, modulating the p53/p21 pathway and regulating cell senescence. However, hsa_circ_0007113 is only 1 of many factors that regulate cellular senescence. In this study, the new function of one circRNA derived from the HERC4 gene was elaborated, the function was verified in human lung fibroblasts, and its relationship with aging was confirmed.

Limitations

The specific molecular mechanism requires further clarification. For example, over-expression of hsa_circ_0007113 both in vitro and in vivo would also show the function.

Conclusions

This study showed that altered circRNA expression patterns are present in cellular senescence, which may play important roles in senescence-related physiological processes. These findings offer a fresh approach to understanding the molecular mechanism underlying senescence, as well as a new way to potentially cure senescence by altering circRNAs. Additional investigations are necessary to recognize circRNA roles in cellular senescence.