UVA influenced the SIRT1‐miR‐27a‐5p‐SMAD2‐MMP1/COL1/BCL2 axis in human skin primary fibroblasts

Abstract Both SIRT1 and UVA radiation are involved in cellular damage processes such as apoptosis, senescence and ageing. MicroRNAs (miRNAs) have been reported to be closely related to UV radiation, as well as to SIRT1. In this study, we investigated the connections among SIRT1, UVA and miRNA in human skin primary fibroblasts. Our results showed that UVA altered the protein level of SIRT1 in a time point–dependent manner. Using miRNA microarray, bioinformatics analysis, we found that knocking down SIRT1 could cause up‐regulation of miR‐27a‐5p and the latter could down‐regulate SMAD2, and these results were verified by qRT‐PCR or Western blot. Furthermore, UVA radiation (5 J/cm2), knocking down SIRT1 or overexpression of miR‐27a‐5p led to increased expression of MMP1, and decreased expressions of COL1 and BCL2. We also found additive impacts on MMP1, COL1 and BCL2 under the combination of UVA radiation + Sirtinol (SIRT1 inhibitor), or UVA radiation + miR‐27a‐5p mimic. SIRT1 activator resveratrol could reverse damage changes caused by UVA radiation. Besides, absent of SIRT1 or overexpression of miR‐27a‐5p increased cell apoptosis and induced cell arrest in G2/M phase. Taken together, these results demonstrated that UVA could influence a novel SIRT1‐miR‐27a‐5p‐SMAD2‐MMP1/COL1/BCL2 axis in skin primary fibroblasts, and may provide potential therapeutic targets for UVA‐induced skin damage.

Being a class of histone deacetylases, sirtuins are associated with mammalian metabolism and lifespan of lower organisms. 9,10 Silent information regulator 1 (SIRT1) is the homology of silent information regulator 2 (Sir2) that has been previously extensively studied. SIRT1 regulates expression and activity of proteins, such as PGC-1α, p53 and FOXO1 via deacetylating histone and non-histone proteins in a nicotinamide adenine dinucleotide + dependent way. 11,12 SIRT1 is involved in the regulation of numerous biological processes, such as metabolism, cell cycle, DNA repair, cell survival and ageing. 9,10 SIRT1 also participates in the process of resisting oxidative stress and eliminating ROS. Csiszar showed that caloric restriction (CR) diet could significantly decrease the production of TNF-α-induced ROS, inhibit the activity of NF-κB and reduce the oxidative stress responses, and these effects could be significantly weakened in SIRT1-knockdown rat models. 13 The overlapping between SIRT1 involved-protection mechanisms and UVA induced-damage processes indicated a close relationship between SIRT1 and UVA.
MicroRNAs (miRNAs) is a group of highly evolutionarily conserved non-coding small RNA molecules (19)(20)(21)(22)(23)(24) nucleotides) that regulate gene expression at the post-transcriptional level. 14 MiRNAs have been reported to be closely related to UV radiation. For example, UV-inducible miR-16 down-regulated CDC25a and regulated cell proliferation in human fibroblasts. 15 Increase in miR-22 may protect cells from UV-induced apoptosis by repressing PTEN in HEK293T cells. 16 Up-regulation of miR-125b induced by NF-kappa B promoted cell survival upon UV radiation. 17 There are also associations between miRNAs and SIRT1. Guan et al have reported that miR-30a inhibited the proliferation, invasion and apoptosis of lung cancer cells by inhibiting SIRT1 in vivo and in vitro. 18 Overexpression of miR-221 has been found to reduce the protein abundance of SIRT1, and cause inflammation and insulin resistance in differentiated 3T3-L1 cells. 19 As these evidences have suggested certain potential connections among SIRT1, UVA and miRNA, we hypothesized that SIRT1 regulates miRNA medicating downstream pathways/biomarkers, and thus, they play important roles in UVA-induced skin damage. Therefore, we silenced SIRT1 to identify potential target miRNA and its' downstream markers in human skin primary fibroblasts, under UVA irradiation circumstance or not. We were sought to find a novel SIRT1-miRNA regulating axis to provide potential therapeutic targets for UVA-induced skin damage. Cells were maintained with Dulbecco' s modified Eagle medium (HyClone) containing 10% foetal bovine serum (HyClone) and 1% penicillin-streptomycin solution (BI, ISR), and cultured at 37°C in 5% CO 2 humidified atmosphere.

| UVA radiation
The cells 1 × 10 5 were seeded per well in 6-well plates. Fibroblasts were grown to 60%-70% confluence in a 6-well plate before UVA exposure. Fibroblasts were washed with PBS twice and covered with 500 μL of PBS per well in 6-well plates before exposed to UVA radiation by UVA Therapy System (5 J/cm 2 UVA, UV 801 KL, Waldmann, GER). The distance between the cells and the lamp was 15 cm. Fresh medium was then used to replace PBS and maintain the cell culture for subsequent experiments.

| Immunofluorescence staining
Skin primary fibroblasts were fixed with 4% paraformaldehyde for 20 minutes, permeated with PBS containing 0.2% Triton X-100 and 1% BSA for 5 minutes, and blocked with 1% BSA in PBS for 1 hour Cells were then incubated with rabbit anti-vimentin (VIM) antibody (diluted 1:100, BOSTER) or mouse anti-alpha smooth muscle actin

| Additions of resveratrol, SRT 1720 or Sirtinol
The resveratrol (RSV) and SRT 1720 were SIRT1 activators, and Sirtinol was SIRT1 inhibitor. When primary fibroblasts were at a confluency of 70%, cells were washed with PBS for three times and then were incubated with 10 μM of RSV (MCE), 2 μM SRT 1720 HCL (Selleck) or 10 μM Sirtinol (MCE) for 24 hours. The cells were harvested for subsequent experiments.

| qRT-PCR assay
Total RNAs were extracted from skin primary fibroblasts with miRNeasy mini kit (Qiagen). For mRNA, cDNA synthesis was performed using GoScript™ Reverse Transcription System (Promega), and then, the cDNA was subjected to real-time PCR with GoScript™ qPCR Master Mix (Promega). Primers (BGI) used for PCR were as follows: The 2 -△△Ct was used to calculate relative gene expression. All experiments were tested at least three times.

| Flow cytometry assay
For apoptosis analysis, cells were harvested and stained with 5 μL

| MiRNA microarrays
Total RNAs were extracted from three SIRT1-silenced (si-SIRT1) skin primary fibroblasts and three control fibroblasts. The potential miRNAs were detected by using TaqMan Array Human MicroRNA A + B Cards (Thermo Scientific). This microarray could quantify 754 human miRNAs (Table S1). Three endogenous controls (U6, RNA44 and RNA48) and one non-human negative control were also included in this microarray. The experiment was performed following the manufacturer's instructions. In brief, each fill reservoir of the card was loaded with 100 μL of prepared PCR reaction mix, including 450μl PCR Master Mix, 6μl cDNA and 444μl Nuclease-fee

| Prediction of miRNA target genes
Eight bioinformatics tools, including miRanda, PicTar, PITA 6.0, RNA22 2.0, TargetScan 7.2, miRDB, miRWalk 2.0 and miRNAMap 1.0, were used to predict the potential target genes of differentially expressed miRNAs. Target genes were selected based on the criteria of being identified by more than four bioinformatics tools with statistical significance. The combined positions between miR-NAs and potential target genes were predicted by TargetScan 7.2.

| Bioinformatics analysis
Predicted target genes were analysed using Gene ontology (GO) enrichment analysis to identify three independent categories, that is biological process (BP), cellular component (CC) and molecular function (MF). Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway Database was used to assess enriched pathways of potential target genes. GO analysis and KEGG pathway analysis were conducted with DAVID Bioinformatics Resources 6.8 (https:// david.ncifc rf.gov). Visualization results were achieved using R 3.5.0.
The luciferase assay was carried out following the instructions of Promega Dual-Luciferase system. The assay was performed at least 3 times in independent experiments.

| Statistical analysis
All analyses were performed using Statistical Package for Social Sciences (SPSS) software (Version 20.0, Inc). Results were expressed as mean ± SEM. Multi-group comparisons of the means were carried out by one-way analysis of variance (ANOVA) test with post hoc contrasts by Student-Newman-Keuls test. A P-value < .05 was considered to indicate statistical significance.

| Human skin primary fibroblasts were obtained and interfered with SIRT1 siRNAs
Cells obtained from foreskin tissues were identified as human skin fibroblasts by results of vimentin (+) DAPI (+) α-SMA (-) using immunofluorescence staining ( Figure 1A).
Two different SIRT1 siRNAs (siRNA 1 and siRNA 2) were, respectively, transfected into cells, and the transfected efficiencies were verified by qPCR ( Figure 1B

| UVA caused MMP1 increase and SIRT1 decrease in skin primary fibroblasts
Our group has previously reported that 5 J/cm 2 of UVA radiation could significantly increase the mRNA expression of MMP1 in skin fibroblasts, and MMP1 presented sinusoidal increase from 6 hours to 48 hours (peaking at 24 hours). In addition, the mRNA level of MMP1 was inhibited from 6hours to 48 hours when treated with 0.01 mmol/L RSV. 20 The present study validated the previous data by showing UVA radiation (5 J/cm 2 ) could induce the protein levels of MMP1 at 12 hours compared with UVA-free group. In addition, increased expression of MMP1 could be decreased by 0.01 mmol/L RSV at 24 hours. (Figure 2A).
When we detected the protein level of SIRT1 at different time points, we observed that SIRT1 was inhibited as soon as 1 hr after UVA radiation, and restored 8 hours later ( Figure 2B). We have also found a UVA dose-dependent increase in SIRT1 expression at 24 hours ( Figure 2C) western-blotWestern-blot.

| Inhibition of SIRT1 increased the expression level of miR-27a-5p
Many studies have reported the close associations between miR-NAs and SIRT1. 18,19 In the present study, after successfully silencing SIRT1, a total of 176 miRNAs were detectable in all groups by the microarray. Among of these miRNAs, 15 miRNAs were identified to be differentially expressed between si-SIRT1 cells and normal cells (log 2 fold change (FC)<−1 or > 1 and P < .05). The heatmap analysis and volcano plot analysis displayed that 14 miRNAs were decreased and 1 miRNA (miR-27a-5p) was increased in si-SIRT1-treated cells ( Figure 3A,B). The details of differentially expressed microRNAs were listed in Table 1.
In the process of data mining, we chose a FC of 3 as the cut-off.
When predicted target genes of the 8 miRNAs were analysed using GO and KEGG pathway analysis, only miR-27a-5p and miR-432-5p got positive results. GO analysis showed that the predicted target genes of the two miRNAs mostly enriched in nucleus, acted as molecular binding and involved in transcription process. KEGG pathway analysis revealed that the predicated target genes of miR-27a-5p, however miR-432-5p not, were enriched in TGF-beta signalling pathway and MAPK signalling pathway which were closely associated with cell ageing. 21,22 The GO and KEGG analyses of predicted target genes of miR-27a-5p were shown in Figure 3C and Figure 3D, respectively.
Among the predicated target genes of both miR-27a-5p and miR-432-5p, SMAD2 was the known transcription factor of TGF-beta signalling pathway. 23 The luciferase reporting experiment was done to verify the correlation of miRNA and SMAD2, and positive result was found for miR-27a-5p and SMAD2, but negative for miR-432-5p and SMAD2 ( Figure 3E). So, miR-27a-5p was chosen for the subsequent experiments.
The entire screening process was shown in Figure 3F.
Furthermore, the characters of GO analysis (top 10) and KEGG pathway analysis (top 10) of the predicted target genes of miR-27a-5p were shown in Table 2. The predicted target genes of miR-27a-5p were presented in Table S2.

| SIRT1 up-regulated SMAD2 via downregulating miR-27a-5p
Having found SIRT1 negatively regulated miR-27a-5p by miRNA microarrays, we then validated these results by qRT-PCR assay ( Figure 4A). According to the facts that SIRT1 down-regulated miR-27a-5p and miR-27a-5p has the combined position on the 3'-UTR of SMAD2, we assumed that knockdown of SIRT1 could decrease SMADs through the increased miR-27a-5p. So we tested the protein levels of SMAD2 and association proteins including SMAD3/4, YAP and TAZ by Western blot, under different conditions. SMAD2 was obviously decreased by SIRT1 siRNA or by miR-27a-5p mimic (P < .05), although uninfluenced by RSV or SRT 1720. The expression patterns of SMAD3 and TAZ were similar to those of SMAD2 ( Figure 4B). No significant differences were seen for SMAD4 or YAP among groups. These results revealed SIRT1 could regulate SMAD2 via miR-27a-5p.
Previous studies supported that SMAD2 could regulate MMP1, Con, fibroblast cells without any disposal; si-SIRT1, fibroblast cells transfected with SIRT1 siRNA. RSV, Resveratrol. Asterisk (*) represents statistical significance (P < .05) compared to control group COL1 24 and BCL2, 25 and the latter markers were closely related to ageing and cell apoptosis. So we speculated SIRT1 could regulate MMP1, COL1 and BCL2. The results showed that absence of SIRT1 led to increased expression of MMP1 and decreased expression of COL1 and BCL2. When miR-27a-5p was overexpressed by mimic, we detected the similar changes in these downstream targets. In addition, compared to untreated cells, both RSV and SRT 1720 successfully up-regulated SIRT1, decreased MMP1, as well as increased COL1 and BCL2 ( Figure 4C). These results revealed a signalling axis of SIRT1-miR-27a-5p-SMAD2-MMP1/COL1/BCL2.

| SIRT1 decreased cell apoptosis and induces cell arrest in G2/M phase via miR-27a-5p
In the further investigation, we used flow cytometry assay to observe the impact of SIRT1 on cell apoptosis and cell cycle in skin primary fibroblasts. Apoptosis analysis showed that silencing of SIRT1 by siRNA strategy significantly increased the percentage of apoptotic cells (P < .05), which was significantly decreased when SIRT1 was up-regulated by RSV and SRT 1720 (P < .05). The percentage of apoptotic cells was also significantly increased in miR-27a-5p mimic-treated cells (P < .05), consistent with that in the si-SIRT1 group ( Figure 4D). These data displayed that SIRT1 could regulate cell apoptosis via miR-27a-5p. Cell cycle analysis showed activating SIRT1 by SRT 1720 decreased the percentage of G0/1 phase cells and increased the percentage of S and G2/M phase cells. When skin primary fibroblasts were pre-treated with SIRT1 siRNA or miR-27a-5p mimic, the percentage of G0/1 phase cells was decreased and the percentages of S and G2/M phase cells were increased ( Figure 4E).
We hypothesized that UVA may affect MMP1, COL1 and BCL2 via SIRT1. We initially treated the skin primary fibroblasts with UVA radiation or 10 uM Sirtinol. After 24 hours of treatment, we see that UVA or Sirtinol demonstrated contrary impacts on MMP1, COL1 or BCL2 against SIRT1 ( Figure 4F).
Since UVA radiation and Sirtinol both caused increased MMP1 and decreased COL1 and BCL2, it was not surprise to see additive impact on these three target proteins under the combination of UVA radiation and Sirtinol. In addition, when RSV was given to UVA and Sirtinol pre-treated skin primary fibroblasts, reversed changes in MMP1, COL1 and BCL2 were detected ( Figure 4F). These results further confirmed that UVA could regulate MMP1, COL1 and BCL2 via SIRT1 in skin primary fibroblasts.
As we have found that SIRT1 regulated MMP1, COL1 and BCL2 via miR-27a-5p, we came to the question that if miR-27a-5p was also involved in the UVA-SIRT1-MMP1/COL1/BCL2 axis. To address this question, skin primary fibroblasts were dealt with UVA radiation and miR-27a-5p was found to be up-regulated. Under the treatment of miR-27a-5p mimic, the protein levels of MMP1, COL1 and BCL2 were consistently altered with that of SIRT1 knocking down.
The schematic illustration of this study was shown in Figure 5.
UVA down-regulated SIRT1 and SIRT1 down-regulated miR-27a-5p, then regulated the damages of skin primary fibroblasts through influencing SMAD2-MMP1/COL1/BCL2 axis. activation, thus, impeded Mn-induced neurotoxicity. 32 In psoriatic fibroblasts, oxidative stress, mitochondrial damage and apoptosis, and reduced SIRT1 expression were simultaneously observed. 33 Our study also found the protective role of SIRT1 by showing decreased apoptotic cells when SIRT1 was overexpressed by RSV/SRT 1720, or increased apoptotic cells when SIRT1 was knocked down in human fibroblast cells.

| D ISCUSS I ON
The protective role of SIRT1 against apoptosis was likely to be executed via down-regulating miR-27a-5p, as shown by increasing of apoptotic cells when the cells were treated with miR-27a-5p mimic.
Some other miRNAs such as miR-3613-3p 34 and miR-125b 35 have also been reported to inhibit apoptosis in BE(2)C cells and myeloma cells, respectively. In contrary, certain miRNAs played roles in inducing apoptosis. For example, apoptosis was promoted by miR-99a in human granulosa cells 36 and miR-223 in hepatic carcinoma cells. 37 In the analysis of cell cycle, we observed significant decreased percentage of cells in G0/G1 phase, and increased percentage of cells in G2/M phase by SRT 1720, suggesting SIRT1 induced cell cycle arrest in G2/M phase. In consistency with our finding, Cao et al found the activation of SIRT1 could rescue the G0/G1 phase cell cycle arrest in TM4SF1-deficient BCa cells. 38 However, other interference factors such as SIRT1 siRNA or miR-27a-5p mimic failed to cause significant changes in fibroblasts.
We speculated the reason was that the original expression level of SIRT1 was not enough to induce cell cycle changes, and therefore, we could not detect cell cycle changes when down-regulated SIRT1 or up-regulated miR-27a-5p. The exact regulatory function of SIRT1 or miR-27a-5p on cell cycle should be achieved on more primary fibroblasts in future study.
In conclusion, the present study identified a novel SIRT1-miR-27a-5p-SMAD2-MMMP1/COL1/BCL2 axis and a mechanism of UVA influencing human primary fibroblasts through this axis. Our finding may provide potential therapeutic targets for UVA-induced skin damage. We suspected that increasing the expression of SIRT1 or decreasing the expression of miR-27a-5p may reduce UVA-related diseases, or even play a role in the treatment of UVA-induced skin cancer. However, limitations, such as the unclear relationships between MMP1 and BCL2, and the lack of animal work, should be addressed in future studies.

ACK N OWLED G EM ENTS
This work was supported by grants from the National Natural Science Fund (Grant number 81972940), Liaoning Province Natural Science Fund (Grant number 2019-ZD-0763).

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest. Validation (equal).

DATA AVA I L A B I L I T Y S TAT E M E N T
No data sets were generated for this study.