This article is mentioned in:

Introduction

Ultraviolet A (UVA) radiation, which penetrates deep into the dermis and affects human dermal fibroblasts (HDFs), has an important role in inducing photoaging (1). The histopathological changes of photoaging are mainly manifested by the reduction of collagens and the denaturation and accumulation of elastic fibers (1,2). UVA not only affects the appearance of the skin, but also causes a variety of light-associated diseases, including actinic granuloma, squamous cell carcinoma and malignant melanoma (1,3-5).

Although the mechanism of photoaging remains to be fully elucidated, there have been numerous advances in recent years. For example, microRNA (miR)-377 have been reported to promote the senescence of human dermal fibroblasts by targeting DNA methyltransferase (6), whilst the long noncoding RNA (lncRNA) RP11-670E13.6 delays the senescence of UVB-irradiated dermal fibroblasts by sponging microRNA-663a (7). These previous studies have emphasized the essential roles of non-coding RNAs in photoaging.

Circular (circ)RNAs, in contrast to conventional linear RNAs, are a class of non-coding RNAs formed by reverse splicing, without a 5' cap and a 3' polyA tail (8). Compared with other biomolecules, circRNAs are durable and stable functional molecules that do not get degraded (9). circRNAs bind miRNAs with complementary sequences through base-pair binding, thus acting like a sponge to regulate the expression of miRNA target genes (10).

It has been previously reported that circRNAs are involved in the development and progression of various diseases, including heart failure, cancer and Alzheimer disease (11-13). However, the role of circRNAs in photoaging remains to be fully elucidated. Therefore, the present study explored the expression profiles of circRNAs in ultraviolet A (UVA)-irradiated human dermal fibroblasts compared with those in the non-irradiated control group.

Materials and methods

Cell culture and UVA irradiation

The present study was approved by the Medical Ethics Committee of the Third Affiliated Hospital of Sun Yat-sen University (Guangzhou, China). The foreskin dermis of six healthy children aged from 3 to 9 years (7.00±1.63 years) was used to isolate and culture fibroblasts using procedures previously described (14). Cells were cultured in Dulbecco's modified Eagle's medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 100 U/ml penicillin (Sigma-Aldrich; Merck KGaA;), 100 U/ml streptomycin (Sigma-Aldrich; Merck KGaA) and 10% fetal bovine serum (PAN-Biotech GmbH) in a 37˚C humidified incubator with 5% CO2. Cells between the 4 and 6th passages were utilized for the assays. When the fibroblasts were in the exponential growth phase and their density reached 70%, cells were exposed to UVA irradiation, and control cells were treated under the same conditions but without UVA exposure. The radiation device was a UVA lamp with a wavelength of 320-400 nm (Sigma-Aldrich; Merck KGaA), and the UVA dose was measured using a UVX digital radiometer (Shenzhen Sunkun Technology Co., Ltd.). The daily dose was 10 J/cm2 (15). After 14 days of UVA irradiation, cells were harvested for subsequent experiments.

Confirmation of establishment of UVA-irradiated HDF model of ageing

A Senescence-Associated β-Galactosidase (SA-β-Gal) Staining kit (cat. no. C0602), purchased from Beyotime Institute of Biotechnology, was used to detect the senescent cells using manufacturer's protocol.

Detection of cell cycle-associated protein expression

Protein was extracted from cells using a Protein Extraction kit (Nanjing Keygen Biotech Co., Ltd.). The BCA Protein Assay kit (Thermo Fisher Scientific, Inc.) was used to detect the protein concentration. The protein (50 µg/lane) was separated by 12% SDS-PAGE and electro-transferred onto a 0.45-µm nitrocellulose membrane (EMD Millipore) for 90 min. The membrane was blocked with 5% bovine serum albumin (NeoFROXX GmbH) and incubated overnight at 4˚C on a shaker with the following monoclonal primary antibodies: Rabbit P21 (1:1,000 dilution; cat. no. 2947; Cell Signaling Technology, Inc.), rabbit P16 (1:1,000 dilution; cat. no. ab51243; Abcam) and mouse P53 (1:1,000 dilution; cat. no. 9282; Cell Signaling Technology, Inc.). Subsequently, the membrane was washed with TBS-T (0.05% Tween-20 in TBS; pH 7.4) and incubated with horseradish peroxidase-conjugated secondary antibodies (1:5,000 dilution; cat. nos. 7074 and 7076; Cell Signaling Technology, Inc.) for 1 h at room temperature and protein bands were visualized by chemiluminescence (ECL Advanced Detection Kit; EMD Millipore). GAPDH (1:1,000 dilution; cat. no. 2118; Cell Signaling Technology, Inc.) was used as an internal control. The band densities were measured using ImageJ software (version 1.50; National Institutes of Health).

RNA extraction and quality control

Total RNA was extracted from fibroblasts by TRIzol® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol, and a NanoDrop ND-1000 (NanoDrop; Thermo Fisher Scientific, Inc.) was used to determine the purity and concentration of total RNA in the samples.

circRNA microarrays and bioinformatics analysis

Total RNAs from each sample were treated with RNase R (cat. no. M1228-500; BioVision, Inc.) to enrich circRNAs. Subsequently, referring to the Arraystar super RNA Labeling scheme (Arraystar, Inc.), the enriched circRNAs were amplified with random primers and transcribed into fluorescent complementary RNA (cRNA). The labeled cRNAs were hybridized onto Arraystar Human circRNA Arrays V2 (8x15 K; Arraystar, Inc.) and incubated for 17 h at 65˚C in an Agilent hybridization oven (Agilent Technologies, Inc.). After washing, the slides were scanned with the Agilent Scanner G2505C (Agilent Technologies, Inc.). Agilent Feature Extraction software (Version 11.0.1.1) was used to extract the data (Agilent Technologies, Inc.). Subsequently, the R software package ‘limma’ (16) was used to perform a series of data processing steps. After the original data was normalized, high-quality probes were screened. The probes were marked as present (P), marginal (M) or absent (A). In total, ≥ three of the six samples of circRNAs were flagged in ‘P’ or ‘M’ under ‘All Targets Value’, and these circRNAs were retained for improved differential analysis. A statistically significant difference (fold change >1.5) in expression of circRNAs was identified between two groups by multiple change cut-off or volcano plot filtration. The Arraystar proprietary miRNA target prediction software and TargetScan Human 7.2 (http://www.targetscan.org/vert_72/) were used to identify the potential interactions between circRNAs and their target microRNAs. TargetScan v7.2 and miRDB v5 were used to predict the target genes of these microRNAs. Using the overlapping data, three circRNAs were selected that were associated with COL1A1 and ELN for further qPCR verification. Gene Ontology (GO) annotation and KEGG pathway analysis were performed to provide evidence for the further functional prediction of the differentially expressed circRNAs. These circRNAs were input into the Database for Annotation, Visualization and Integrated Discovery (http://david.abcc.ncifcrf.gov/) and the KEGG database (http://www.genome.jp/kegg/). The P-values denote the significance of the GO term enrichment and the correlation of KEGG pathway (P<0.05 was considered to indicate a statistically significant result). The circRNA expression chip and bioinformatics data analyses were completed by Shanghai Kangcheng Biotechnology Co., Ltd.

RT-qPCR

Total RNA was isolated from the samples using the TRIzol® Plus RNA Purification kit (Thermo Fisher Scientific, Inc.), and its concentration and purity were determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Inc.). Complementary (c)DNA was synthesized by reverse-transcribing 2 µg total RNA using the PrimeScript™ RT Reagent kit (Takara Bio, Inc.). cDNA was amplified according to the instructions of the TB Green Fast qPCR Mix (Takara Bio Inc.). The volume of the PCR amplification mixture was 20 µl, including 10 µl 2X qPCR SYBR®-Green Master Mix, 0.5 µl forward primer (10 µM), 0.5 µl reverse primer (10 µM), 1 µl cDNA and 8 µl RNase-free H2O. The amplification conditions were as follows: Pre-denaturation at 95˚C for 5 min, followed by 40 cycles of denaturation at 95˚C for 10 sec and annealing at 60˚C for 34 sec. The relative expression was quantified using the 2-ΔΔCq method (17), and GAPDH was used as an internal reference gene. Sequences of the primer pairs were listed in Table I.

Table I Primer sequences of circular RNAs.

Table I

Primer sequences of circular RNAs.

circBase ID	Sequence (5'-3')	Size, bp
hsa_circ_0006766-F	TGTTGCCATTACAGGGGTAAGT	146
hsa_circ_0006766-R	CGGGAAGGAAAGTGATATTTGG
hsa_circ_0011129-F	CTTCCGGGCCCAGGTCCTTC	149
hsa_circ_0011129-R	AGTGCAGGCGCCAGAAGCTGG
hsa_circ_0017502-F	CCAGATCATGCCAGGTGACAT	135
hsa_circ_0017502-R	TACGGAACTGCACGCTGAAC
GAPDH-F	GACACCATGGGGAAGGTGAA	79
GAPDH-R	AGTTAAAAGCAGCCCTGGTG

[i] Circ, circular RNA; F, forward; R, reverse.

Statistical analysis

The results are representative of three independent experiments and are presented as the mean ± standard error of the mean. Statistical analysis was performed using SPSS 20.0 statistical software (IBM Corp.). The F-test was used to test the homogeneity of variance. The independent samples Student's t-test was used to compare the means between the chronic UVA irradiation group and the unexposed control group. P<0.05 was considered to indicate a statistically significant difference.

Results

Verification of the skin fibroblast model of photoaging

A number of changes including increased SA-β-Gal activity and cell cycle-associated protein expression have been reported to be induced by UVA irradiation (18,19). Compared with that in the normal control group, the senescence staining was increased in the UVA-treated group (Fig. 1A). Western blot analysis revealed that the expression of p16, p21 and p53 in the UVA-treated group was higher compared with that in the control group (Fig. 1B).

Figure 1 Verification of the UVA-induced fibroblast model of senescence. (A) The cells with blue stained cytoplasm were considered aged cells (magnification, x100). Cells were counted in six randomly chosen microscopic fields, and the percentage of cells positive for SA-β-Gal activity was calculated. (B) P16, P21 and P53 were used to detect the differences in the expression of senescence-associated proteins between the UVA-treated and control groups. GAPDH was used as an internal reference. Values were expressed as the mean ± standard error of the mean (n=3), and all experiments were performed in triplicate. *P<0.05, **P<0.01 vs. control. UVA, ultraviolet A; SA-β-Gal, senescence-associated β-galactosidase.

circRNA expression profiles in UVA-irradiated human dermal fibroblasts

A circRNA microarray was used to obtain the expression profiles of circRNAs in UVA-irradiated human dermal fibroblasts. The distribution of circRNA expression profiles exhibited no differences between the UVA-treated group and the non-irradiated control group (Fig. 2A). Differentially expressed circRNAs between two compared groups were identified through Fold Change filtering (Fig. 2B) or Volcano Plot filtering (Fig. 2C). In Fig. 2D, hierarchical clustering was utilized to show circRNA expression profiling among the all samples. The data showed that there was a distinctly distinguishable circRNAs expression profile between the UVA-treated group and non-irradiated group. Compared with the non-irradiated control group, 128 circRNAs were determined to be differentially expressed after UVA exposure based on the criteria FC>1.5 and P<0.05, including 39 upregulated and 89 downregulated circRNAs. The top 20 aberrant circRNAs are listed in Table II.

Figure 2 circRNA expression profile of fibroblasts after UVA irradiation. (A) The distribution of total circRNA expression between the UVA-treated and control groups was similar. (B) The scatter plot demonstrated the differences in circRNA expression between the UVA-treated and control groups. The green line is an FC line, and the difference in circRNA expression between the top green line and the bottom green line is >1.5-fold. (C) Volcano plot displaying circRNAs differentially expressed with statistical significance (FC>1.5; P<0.05). (D) Hierarchical clustering analysis of differentially expressed circRNAs between the UVA irradiation group and the control group. UVA, ultraviolet A; circRNA, circular RNA; FC, fold change; UV, UVA-treated group; C, control group.

Table II circRNAs differentially expressed following ultraviolet A treatment.

Table II

circRNAs differentially expressed following ultraviolet A treatment.

A, Upregulated circRNAs
circRNA	FC (abs)	Gene symbol	P-value
hsa_circRNA_103065	4.1715	MYBL2	0.0118
hsa_circRNA_100445	4.0847	TATDN3	0.0011
hsa_circRNA_103361	2.7804	SMARCC1	0.0362
hsa_circRNA_000401	2.1920	BRD9	0.0305
hsa_circRNA_048148	2.1574	CNN2	0.0356
hsa_circRNA_104484	2.0623	ZC3HC1	0.0494
hsa_circRNA_404870	2.0559	METTL15	0.0197
hsa_circRNA_005411	2.0248	EXTL3	0.0407
hsa_circRNA_101093	1.7798	NUP107	0.0299
hsa_circRNA_092418	1.7606	B2M	0.0434
B, Downregulated circRNAs
circRNA	FC (abs)	Gene symbol	P-value
hsa_circRNA_101621	2.7131449	CEMIP	0.032211923
hsa_circRNA_007624	2.4404224	BCAR3	0.037482416
hsa_circRNA_004585	2.4282438	CEMIP	0.032890625
hsa_circRNA_000675	2.4117771	NDUFA10	0.034232964
hsa_circRNA_006766	2.3634626	CCL24	0.026828638
hsa_circRNA_100191	2.2776152	PTPRF	0.048836243
hsa_circRNA_004594	2.2729704	MLLT1	0.002923375
hsa_circRNA_102817	2.219963	SAP130	0.003660652
hsa_circRNA_062539	2.1416076	BCR	0.00056699
hsa_circRNA_104647	2.1249749	ZFHX4	0.006894599

[i] circRNA, circular RNA; FC (abs), absolute fold-change.

GO and KEGG pathway analyses

To determine the roles of circRNAs in the physiology and pathology of fibroblasts after UVA irradiation, GO and KEGG pathway analyses of mRNAs transcribed from parental genes of the differentially expressed circRNAs were performed. The results demonstrated that the most significantly enriched relevant GO terms in the biological process category were ‘collagen catabolic process’ and ‘collagen metabolic process’, as well as ‘extracellular matrix’ (ECM; Fig. 3A). In the cellular component category, the parental genes of the differentially expressed circRNA were identified to be accumulated in the ‘complex of collagen trimers’ (Fig. 3B). In the molecular function category, the most significantly enriched terms were ‘extracellular matrix structural construction’, ‘metalloendopeptidase’ and ‘protein binding’ (Fig. 3C). KEGG pathway analysis revealed the top 10 pathways that may be involved in skin photoaging (Fig. 3D). Among these, the ‘Age-RAGE signaling pathway in diabetic complications’ had been reported to be involved in skin photoaging (20,21).

Figure 3 GO enrichment and pathway analysis derived of the host genes of the differentially expressed circRNAs. (A-C) GO terms in the categories (A) biological process, (B) cellular component and (C) molecular function. (D) Bar chart displaying the top 10 enriched pathways. GO, gene ontology.

Verification of differentially expressed circRNAs by RT-qPCR

The main pathological changes of photoaging are the accumulation of abnormal elastin and a severe loss of collagen fibers in dermis (1,2). To verify the data of the circRNA microarray, three circRNAs that were predicted to be associated with collagen type I (COL1A1) and elastin (ELN) were selected to be confirmed by RT-qPCR analysis. There were no significant differences in the expression levels of circ-0017502, but there were significant downregulations in the expression levels of circ-0006766 and circ-0011129 in the UVA-irradiated group compared with those in the control group (Fig. 4). The expression pattern was similar to that obtained from the circRNA microarray.

Figure 4 Reverse transcription-quantitative PCR verification results of three circRNAs. Quantitative PCR validation was performed on six samples for Circ-0006766, Circ-0011129 and Circ-0017502. Values are expressed as the mean ± standard error of the mean (n=3 per group). *P<0.05, **P<0.01 vs. control. Circ/circRNA, circular RNA.

Discussion

Accumulating evidence indicates UVA to be a major contributing factor in photoaging, but its mechanisms remain to be fully elucidated (1). Furthermore, UVA is associated with a variety of light-associated diseases, including skin cancer (4). Therefore, developing new markers for early diagnosis and the design of preventive strategies would be of benefit. Non-coding RNAs have been widely assessed in studies on gene function and disease treatment. For example, MRX34, a microRNA (miRNA) mimic, has reached phase 1 studies in patients with advanced solid tumors, including hepatocellular carcinoma, pancreatic cancer and cholangiocarcinoma (22). miR-146a may serve as a promising therapeutic target for dengue (23). Wheatley et al (24) previously developed a DNA vaccine vector co-expressing an miRNA designed in silico miRNA to inhibit PERK with a HIV-1 envelope. In the field of skin photoaging, non-coding RNA has also received increasing attention. For instance, downregulation of miR-34c-5p may inhibit the p53 pathway by increasing the level of the downstream target E2F transcription factor 3 and ultimately impair cell senescence (25). However, miRNAs and lncRNAs have a linear structure, due to which their stability is insufficient to withstand treatment with RNase (the 3' exonuclease of hydrolyzable linear RNAs) (26), limiting their clinical application. Compared with linear non-coding RNAs such as miRNAs and lncRNAs, circRNAs have the advantage of being more stable (9,27).

The role of circRNAs in photoaging remains to be fully elucidated. Previous studies by our group have indicated that circCOL3A1-859267, which targets miR-29c, is able to regulate the expression of type I collagen in fibroblasts (19,28). In the present study, the expression profile of circRNAs in UVA-irradiated fibroblasts was determined, and 128 differentially expressed circRNAs (FC>1.5 and P<0.05) were identified, including 39 upregulated and 89 downregulated circRNAs. GO and KEGG analyses indicated that the aberrant circRNAs were implicated in ECM organization and metabolism. Subsequently, RT-qPCR was used to verify three differentially expressed circRNAs and the results were similar to those of the circRNA microarray. Of note, the differentially expressed circRNAs identified in the present study require further assessment by bioinformatics analyses and in vivo and in vitro experiments. For instance, the aberrant accumulation of ELN and the extensive loss of collagen fibers in dermis are the fundamental pathological changes of skin photoaging (29). Through the Arraystar proprietary miRNA target prediction software and TargetScan, the present study identified that the miRNAs that circ-0011129 may sponge, including hsa-miR-484, hsa-miR-3619-5p and hsa-miR-6732-5p, share binding sites with photoaging-related proteins, such as collagen type I α1 (COL1A1), COL3A1 or elastin.

Zhang et al (30) previously revealed that miR-6732-5p was one of seven significantly upregulated miRNAs in atypical meningioma patients that are resistant to radiotherapy, where the differentially expressed miRNAs were enriched mostly in the transforming growth factor-β (TGF-β) signaling pathway. In addition, Quan et al (31) revealed that UV irradiation impaired the TGF-β/Smad pathway in human dermal fibroblasts, thereby downregulating the expression of TGF-β target genes such as type I procollagen. Type I collagen is one of the main components of the dermal extracellular matrix. Extensive degradation of type I collagen is a fundamental pathological change of skin photoaging (1). Therefore, circ-0011129 is hypothesized to regulate the expression of type I collagen by targeting the miR-6732-5p and TGF-β/Smad pathway.

Several limitations of the present study should be acknowledged. The effect of ultraviolet A (UVA) on fibroblasts was explored but not ultraviolet B (UVB). Although both UVA and UVB are involved in skin photoaging, UVA penetrates deeper than UVB and induces more profound alterations in the dermal connective tissue (31,32). Therefore, for fibroblasts in the dermis, UVA has a greater impact. In addition, regulatory mechanisms of the differentially expressed circRNAs on photoaging remain to be fully elucidated, which serve to be the next step in this research.

In conclusion, the present study provided a unique circRNA profile in UVA-irradiated fibroblasts. These altered circRNAs may provide an avenue to elucidate the mechanisms of skin photoaging, as well as facilitate the development of small molecules to prevent and treat photoaging and associated skin diseases.

Acknowledgements

Not applicable.

Funding

The present study was supported by a grant from the National Natural Science Foundation of China (grant no. 81673047).

Availability of data and materials

The datasets used and/or analyzed during the present study are available from the corresponding author on reasonable request.

Authors' contributions

WL proposed the experimental concept and designed the study. ML and YZ established the UVA irradiation model and were major contributors in writing the manuscript. ML, QL, YL, QX and YL performed the microarray data analysis and performed critical data analyses. All authors read and approved the final manuscript.

Ethics approval and consent to participate

Informed written consent of patients' parents was obtained prior to tissue resection. The present study and all procedures were approved by the Medical Ethics Committee of the Third Affiliated Hospital of Sun Yat-sen University [no. (2016)2-63].

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References