MicroRNA-760 Resists Ambient PM2.5-induced Lung Injury Through Upregulating Heme-oxygenase 1 Expression

Background: PM 2.5 (particles matter smaller aerodynamic diameter of 2.5 μm ) exposure, as one major environmental risk factor for the global burden of disease, is associated with high risks of respiratory diseases and lung cancer. Heme-oxygenase 1 (HMOX1) has been considered as one of the major molecular antioxidant defenses to mediate cytoprotective effects against diverse stressors, including PM 2.5 -induced toxicity and SARS-CoV-2 infection; however, the regulatory mechanism of HMOX1 expression still needs to be elucidated. In this study, using PM 2.5 as a typical stressor, we explored whether microRNAs (miRNAs) might modulate HMOX1 expression in lung cells. Results: Systematic bioinformatics analysis showed that seven miRNAs have the potential to target HMOX1 gene. Among these, hsa-miR-760 was identied as a response miRNA to PM 2.5 exposure. More importantly, we revealed a “non-conventional” miRNA function in hsa-miR-760 upregulating HMOX1 expression, by targeting the coding region and interacting with YBX1 protein. In addition, we observed that exogenous hsa-miR-760 effectively elevated HMOX1 expression, reduced the reactive oxygen agents (ROS) levels, and then rescued the lung cells from PM 2.5 -induced apoptosis. Conclusion: Our results revealed that hsa-miR-760 might play an important role in protecting lung cells against PM 2.5 -induced toxicity, by elevating HMOX1 expression, and offered new clues to elucidate the diverse function of miRNAs.

Enzymatic antioxidant defense has been considered as one of major molecular antioxidant defenses to eliminate ROS. The antioxidant enzymes are produced to restore cellular redox homeostasis in the presence of oxidative stress [13]. Heme oxygenase 1 (HMOX1) is a rate-limiting enzyme for heme metabolism and a key enzyme for endogenous carbon monoxide (CO) synthesis [14,15]. In addition, HMOX1 is one of the Nrf2-targeted genes, and has been reported to be involved in the protection of in ammation and apoptosis [16]. In most tissues/organs, except for the spleen, HMOX1 is usually expressed at low levels; however, it is highly inducible in response to a variety of stimuli to protect cells against oxidative and in ammatory injury. Several lines of evidence have indicated that PM 2.5 -induced oxidative stress might function as a signaling molecule to activate HMOX1 expression [5,13,17,18] ; however, the detailed mechanism remains unrevealed.
In the past two decades, non-coding RNAs have been identi ed as an essential epigenetic factor in modulating gene expression and therefore affecting biological functions. MicroRNAs (miRNAs) are a class of small non-coding RNAs with multiple function in diverse biological and pathological processes [19][20][21][22]. Due to their involvement in extensive biological functions, miRNAs are becoming clinical biomarkers for disease diagnosis and have been considered to have therapeutic potential in preclinical research [23]. MiRNAs typically suppress gene expression at post-transcriptional levels, but increasing evidences showed that speci c miRNAs might increase the target gene expression [24][25][26]. Till now, lots of dysregulated miRNAs have been identi ed after PM 2.5 exposure [27-30]; however, it remains largely elusive about how speci c miRNAs might contribute to HMOX1 upregulation under PM 2.5 exposure.
In this study, we explored the hypothesis that PM 2.5 might modulate speci c miRNA expression, which contributes to HMOX1 upregulation to resist PM 2.5 -induced lung injury. We observed a signi cant upregulation of hsa-miR-760 in the lung cells upon PM 2.5 exposure, and manifested that hsa-miR-760 might upregulate HMOX1 expression by binding to the coding region, i. e., an unconventional miRNA function. Further, we discovered that hsa-miR-760 up-regulates HMOX1 by elevating the mRNA stability in a YBX1dependent manner. We also found that upregulation of hsa-miR-760 has an anti-apoptosis effect in lung cells exposed to PM 2.5 .

Results
Hsa-miR-760 was predicted to modulate HMOX1 expression in response to PM 2.5 exposure Firstly, we retrieved the top ten genes interacting with PM 2.5 exposure from the CTD database. All genes were reported to interact with PM 2.5 over 50 times, suggesting their pivotal roles in response to PM 2.5 exposure (Fig. 1a). Among these, HMOX1 has been documented to induce anti-apoptotic effects after PM2.5 exposure in vivo and in vitro, but the underlying biological mechanism is not clear. Using HMOX1 as the target gene, a total of seven miRNAs, including has-miR-485-5p, hsa-miR-760, hsa-miR-671-5p, hsa-miR-520a-5p, hsa-miR-525-5p, hsa-miR-3126-5p, and hsa-miR-3142, were predicted to bind to the HMOX1 transcript ( Fig. 1b; Table 1). Among them, hsa-miR-760 showed high free energy to bind to both the 3'UTR and coding region of the HMOX1 transcripts, indicating its strong regulatory roles in modulating HMOX1 expression.
We then treated HBE cells with different concentrations of PM 2.5 , and measured the expression changes of hsa-miR-760 and HMOX1 at 24h after PM 2.5 exposure. As shown in Fig. 1c, PM 2.5 exposure signi cantly elevated hsa-miR-760 levels in HBE cells. Intriguingly, both RNA and protein levels of HMOX1 were observed to be upregulated in a dose-dependent fashion after PM 2.5 exposure (Fig. 1d), indicating a potential unconventional miRNA function of hsa-miR-760 in modulating HMOX1 expression.
Hsa-miR-760 binds to the coding region of HMOX1 As mentioned above, hsa-miR-760 was predicted to potentially target the 3'UTR and coding region of the HMOX1 transcript. We rst constructed the reporter gene plasmids containing the core 3'UTR of HMOX1, and observed that exogenous hsa-miR-760 failed to affect the luciferase activities (Fig. S1), suggesting that hsa-miR-760 was not able to bind to HMOX1 3'UTR. Subsequently, we constructed three modi ed reporter gene plasmids (Fig. 2a), including HMOX1 CDS-WT (containing two predicted binding sites of hsa-miR-760), and two mutant luciferase plasmids HMOX1 CDS MUT-1 and HMOX1 CDS MUT-2 (destroying the 1 st and 2 nd binding sites of hsa-miR-760, respectively). Fig. 2b showed that exogenous hsa-miR-760 transfection signi cantly increased the luciferase signal produced by the HMOX1 CDS-WT and HMOX1 CDS-MUT2 plasmids in both HEK293T and HBE cells (all P < 0.01), but failed to elevate the luciferase signal produced by HMOX1 CDS-MUT1 plasmid as compared with the ones transfected with miRNA NC, proving that hsa-miR-760 was able to bind to the 2 nd binding site, which with the free energy of -36.60 kcal/mol, in HMOX1 coding region.
The direct interaction between hsa-miR-760 and its cognate target in the HMOX1 coding region was then further validated by FREMSAs. As shown in Fig. 3a(lane 3), hsa-miR-760 formed distinct complexes with the 2 nd binding site in the HMOX1 coding region. In competition assays, excess unlabeled miRNA oligonucleotides attenuated the miRNA:mRNA complexes (Fig. 3a, lane 5), but excess unlabeled nonspeci c oligonucleotides failed to show any inhibitory effects (Fig. 3a, lane 4), exhibiting the sequence-speci c interaction between hsa-miR-760 and its cognate binding site. In addition, the miRNA pull-down assays were conducted to test the miRNA:mRNA complex in vivo. As shown in Fig. 3b, the HMOX1 transcript was signi cantly enriched (P < 0.01) in cells transfected with the 3′-biotinylated oligonucleotides of hsa-miR-760, compared with the ones transfected with miRNA NC.
Hsa-miR-760 upregulates HMOX1 levels in lung cells Our ndings described above revealed a direct interaction between has-miR-760 and HMOX1 transcript. To measure the regulatory signi cance of hsa-miR-760 in endogenous HMOX1 expression, were transfected different concentrations of hsa-miR-760 mimics and inhibitors into HBE and H226 cells.
Together with the elevation of hsa-miR-760 levels (Fig. S2), both mRNA and protein levels of HMOX1 were signi cantly increased in a dose-dependent fashion in HBE and H226 cells (all P < 0.05) (Fig. 3c, 3d), respectively. Further, the cells transfected with hsa-miR-760 inhibitors signi cantly decreased HMOX1 expression, at both RNA and protein levels, compared with the ones transfected with inhibitor negative control (all P < 0.05) (Fig. 3c, 3d).
Knockdown of YBX1 attenuates the upregulatory effect of hsa-miR-760 on HMOX1 expression Having observed that has-miR-760 upregulates HMOX1 production, we further investigated the underlying molecular mechanism via multiple functional experiments. The miRNA pull-down assays were conducted, and the proteins that could speci cally bind to the biotinylated has-miR-760 were identi ed by mass spectrometry (MS) analysis. Among the proteins speci cally enriched by biotinylated has-miR-760 (Supplementary Table 2), YBX1, a multifunctional RNA binding protein, was selected for further functional validation. Indeed, evidences obtained from both miRNA pull-down and RIP assays showed that YBX1 could bind to has-miR-760, directly ( Fig. 4a and 4b). Intriguingly, as indicated in Fig. 4c, YBX1 could also speci cally bind to HMOX1 transcript.
To check the biological role of YBX1 in the interaction between hsa-miR-760 and HMOX1 transcript, HBE cells were transfected with or without hsa-miR-760 mimics, together with either negative control siRNA or YBX1 siRNA, and the RNA and protein levels of HMOX1 gene were measured, respectively. As shown in Fig. 4d, treatment with siRNAs against YBX1 signi cantly eliminated the hsa-miR-760 activity (P< 0.01). Interestingly, has-miR-760 mimics was also able to upregulate the YBX1 production (P < 0.05) (Fig. 4e), which further indicates the interaction between hsa-miR-760 and YBX1.
It was known that YBX1 is involved in RNA biology, such as translational repression, RNA stabilization, and mRNA splicing. Our results showed that exogenous hsa-miR-760 transfection signi cantly inhibited the HMOX1 mRNA degradation, compared with the control group (Fig. 4f). However, once the cells were treated with siRNAs against YBX1, hsa-miR-760 failed to affect the stability of HMOX1 transcripts (Fig.  4g). All these evidences proved that YBX1 protein participates in the regulatory process of hsa-miR-760 on HMOX1 expression by increasing its mRNA stability.

Knockdown of HMOX1 upregulates of ROS and apoptosis levels in HBE cells
To understand the antioxidant mechanism of HMOX1, the HBE cells were transfected with HMOX1 siRNAs, and the ROS and apoptosis levels were measured. We observed that the knockdown of HMOX1 ( Fig. S3a) led to a signi cant ROS accumulation (both P < 0.05) (Fig. 5a). Moreover, the apoptosis levels were also signi cantly increased in the siRNA against HMOX1 groups (all P < 0.05), compared to the control group ( Fig. 5b and 5c). In line with this phenotype, the levels of BCl2/BAX, a molecular indicator of apoptosis, was decreased after the knockdown of HMOX1 (Fig. 5d). These results proved that HMOX1 is an antioxidant factor by inhibiting ROS accumulation and apoptosis.
Exogenous hsa-miR-760 expression reduced apoptosis levels in HBE cells exposed to PM 2. 5 We further investigated the upregulatory effects of has-miR-760 on HMOX1 expression under PM 2.5 exposure. As shown in Fig. S3b, transfection with hsa-miR-760 mimics signi cantly increased mRNA and protein levels of HMOX1 in PM 2.5 exposure group. Moreover, transfection with hsa-miR-760 inhibitor was able to reverse the increase of HMOX1 expression by PM 2.5 exposure, at both mRNA and protein levels ( Fig. S3c).
Though HMOX1 was also upregulated upon PM 2.5 exposure ( Fig.1d and Fig.S3b), HBE cells exhibited signi cant ROS accumulation and apoptosis (Fig. 6 a, b, c, and d), indicating the endogenous HMOX1 upregulation failed to protect the cell against apoptosis. However, exogenous transfection of hsa-miR-760 effectively reduced the ROS levels, and then rescued the cells from PM 2.5 -induced apoptosis (Fig. 6 a, b, c, and d).

Discussion
MicroRNAs are the members of small non-coding RNAs contains about 20-24 nucleotides, which act as important regulators in diverse pathogenesis processes. It has been reported that lots of miRNAs are dysregulated under PM 2.5 exposure, and these miRNAs might regulate the process of PM 2.5 -induced toxicity. For example, PM 2.5 exposure signi cantly dysregulated the miR-331 and miR-146-3p expression, which contributing to human airway epithelial cells in ammation following PM 2.5 exposure [25, 32].
Multiple epidemiologic studies also observed the negative correlation between PM 2.5 exposure and the expression of miRNAs, such as miR-146a, miR-21, miR-222, miR-21-5p, and miR-187-3p [31, 32]. However, our understanding of the biological signi cance of miRNAs in response to PM 2.5 exposure is still the tip of the iceberg. In this study, we reported hsa-miR-760 as a new response miRNA to PM 2.5 exposure for the rst time. More importantly, we revealed a "non-conventional" miRNA function in hsa-miR-760 modulating HMOX1 expression and then proved the exogenous hsa-miR-760 might serve as a potential drug to prevent PM 2.5 -induced ROS accumulation and apoptosis in lung cells.
HMOX1 is a heme oxygenase whose transcription is prone to be triggered by a wide variety of stressors, including heme, hypoxia, and oxidative stress [32]. Moreover, it was reported that the HMOX1 pathway could be a target for the treatment and prevention of SARS-CoV-2 of 2019 (COVID- 19) [33]. To reveal whether HMOX1 is epigenetically modulated in its response to PM 2.5 exposure, we rst predicted seven miRNAs targeting HMOX1 transcript using four databases. Among them, hsa-miR-760 exhibited the highest free energy to bind to the 3'UTR and coding region of HMOX1. Hsa-miR-760 has been reported as a potential anti-tumor miRNA in various cancers [23,34,35]. For example, Yan et al. reported that hsa-miR-760 suppressed NSCLC cell proliferation, cell cycle, and migration, by regulating the ROS1 pathway [36]. In addition, hsa-miR-760 was reported to mediate cell apoptosis and inhibit the phosphorylation of ERK and JNK in Rheumatoid arthritis [37].
Intriguingly, in the cells exposed to PM 2.5 , both hsa-miR-760 and HMOX1 were signi cantly elevated, indicating that hsa-miR-760 might act as a "non-conventional" miRNA, i.e., increasing but not suppressing target gene expression. Subsequent functional experiments, including dual-luciferase reporter gene assay, FREMSA, and miRNA pull-down assay, identi ed that hsa-miR-760 was able to bind to the 2nd binding site, which with the free energy of -36.60 kcal/mol, in the HMOX1 coding region. Moreover, exogenous hsa-miR-760 transfection resulted in a strong elevation of endogenous HMOX1 expression under our experimental conditions, validating the direct regulatory role of hsa-miR-760 in HMOX1 production.
It's well-known that most miRNAs typically bind to 3'UTR of the target gene, leading to mRNA degradation or translation inhibition [20,38]. More and more studies noticed that miRNAs could elevate target gene expression via binding to the 3'UTR or 5'UTR region [25, 39,40]. The functional signi cance of miRNAs targeting the coding region remains elusive, though a few studies proposed that such miRNAs might be involved in translation inhibition [41,42]. We provided solid evidences that hsa-miR-760 is able to upregulate HMOX1 expression by targeting the coding region, which supplies new clues to elucidate the functional signi cance of miRNAs.
Another highlight in this study is the identi cation of pivotal proteins involved in the interaction between miRNA and target mRNA. Usually, miRNAs are loaded into Argonaute (AGO) protein-containing complexes to exert canonical repression mechanisms [43]. In the present study, by using miRNA pull-down together with MS analysis and RIP assays, we found that hsa-miR-760 exerts unconventional regulatory function by interacting with YBX1, but not AGO family proteins. YBX1, a versatile RNA binding protein (RBP), has been implicated in multiple cellular processes, including transcription regulation, RNA stabilization, and translation process [44] [45,46]. RBP interacting with miRNA is not unprecedented. For example, Eiring et al. reported that miR-328 might interact with hnRNP E2 [47]. To our knowledge, this is the rst study to report the interaction between YBX1 and miRNA. In addition, we found that hsa-miR-760 mediated HMOX1 upregulation might partially attribute to the elevated HMOX1 mRNA stability.
HMOX1 is induced rapidly in response to a wide variety of stressors and it plays a key role in cellular protection [48]. Once large amounts of ROS were produced and released, cells will suffer from redox imbalance, which leads to series of phenomena such as DNA damage, destruction of cell membrane structure, and damage of mitochondria and other organelles, and nally induces cell apoptosis [49]. Multiple studies have indicated that PM 2.5 -induced ROS may function as signaling molecules to induce HMOX1 expression to protect cells against oxidative stress [5,13,17,18]. In the current study, we also found that HMOX1 knockdown could upregulate ROS and apoptosis levels in cells, which consistent with the notion that HMOX1 plays an important role in cellular defense by inhibiting ROS accumulation. However, although HMOX1 was upregulated upon PM 2.5 exposure, HBE cells still exhibited ROS accumulation and apoptosis, under our experimental conditions. One plausible explanation is that the amounts of HMOX1 induced by PM 2.5 exposure fail to protect the cell against oxidative stress. Indeed, we observed that exogenous hsa-miR-760 transfection effectively elevated HMOX1 expression, reduced the ROS levels, and then rescued the cells from PM 2.5 -induced apoptosis. Our study provided a new mechanism on the miRNA function under PM 2.5 exposure on lung injury.

Conclusions
In summary, our ndings revealed that PM 2.5 exposure upregulated has-miR-760 levels, and then activated HMOX1 expression that contributes to the protection against ROS accumulation and cell apoptosis. Therefore, has-miR-760 upregulating the HMOX1 pathway could prove to be preventive agents for PM 2.5 -induced lung injury. Most important of all, we revealed a "non-conventional" miRNA function, by which hsa-miR-760 upregulated HMOX1 expression via the YBX1-dependent manner.
Cell culture, transfection, and PM 2.5 treatment Human lung cancer cell line H226 and embryo kidney cell line HEK293T were obtained from American Tissue Type Culture Collection (ATCC, Manassas, VA). Human lung normal epithelial cell line (HBE) is a gift from professor Wen Chen (Sun Yat-Sen University, Guangzhou). H226 was cultured in RPMI-1640 medium, HEK293T was cultured in DMEM medium, and HBE was cultured in MEM medium plus 10% FBS, respectively. The cells were maintained at 37 • C in a humidi ed 5% CO 2 incubator.
Multiple commercially obtained miRNA mimics (Ruibo, Shanghai, China), miRNA inhibitor (IDT, Coraville, IA), and siRNAs targeting YBX1 and HMOX1 (Ruibo, Guangzhou, China) were transfected into cells using Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA), while constructed plasmids were transfected into cells by Lipofectamine 2000 (Invitrogen). Ambient PM 2.5 were collected from Shijiazhuang, China, and the details about the collection, sample extraction and components analyses were presented in our previous study [50]. PM 2.5 was dissolved in DMSO to produce 100mg/ml stock solutions, and then added to the cell culture at the nal concentration of 25μg/ml, 50μg/ml, and 100μg/ml, respectively. HBE cells treated with an equal amount of 0.1% DMSO were used as solvent control.
Fluorescent-based RNA electrophoretic mobility shift assay (FREMSA) All oligonucleotides and primers used in this study were purchased from Sangon Biotech (Shanghai, China). The dye-hsa-miR-760 oligonucleotide was labeled with IRDye ® 800 dye on its 5' end, while the cognate HMOX1 mRNA FREMSA was performed according to the protocol described in our previous study [51]. Brie y, Brie y, both miRNA and mRNA oligonucleotides were heated for 2 minutes at 95℃ to relax RNA secondary structures and immediately placed on ice. And then, 400 nM synthetic miRNA or/and 400nM cognate mRNA oligonucleotides were mixed with basic reaction buffer (1x Binding Buffer, 5% glycerol, 200 mM KCl, and 100 mM MgCl 2 ). All reaction mixtures were cultured at 25℃ for 20 min to form miRNA:mRNA duplex. The reaction mixtures were then separated on a 12% PAGE by electrophoresis at 4℃, and the resultant mobility shifts were scanned by Odyssey CLx Infrared Imaging System (LI-COR Biosciences, Lincoin, NE).

MicroRNA pull-down assays and MS analysis
The miRNA pull-down experiment was carried out as described in the previous study [52]. Brie y, a biotin molecule was covalently attached to the 3′ end of the mature hsa-miR-760 or negative control strand. Modi ed biotin-miR-760 mimics and negative control were transfected into the HBE cells. Cells were harvested at 24h after transfection, and lysed using the lysis buffer that containing 3% IGEPAL ® CA-630 (Sigma-Aldrich, St. Louis, MO), protease inhibitor cocktail (Roche, Basel, Switzerland), and RNaseOUT (Thermo scienti c, Tewksbury, MA). The cytoplasmic lysate was mixed with precoated beads (Life technologies) and incubated (with rotation) at 4 °C overnight to obtain miRNA-mRNA-protein complex. The harvested complex was then divided into two parts. One portion was used to isolate RNA using Trizol (Life Technologies, Carlsbad, CA), while the other portion was used to analyze proteins based on LC-MS/MS system by Applied Protein Technology (Shanghai, China).

Dual-luciferase reporter gene assay
Hsa-miR-760 was predicted to target both the 3'UTR and coding region of the HMOX1 gene. The core HMOX1 3'-UTR that harboring the hsa-miR-760 response element was chemically synthesized and subcloned into the psiCHECK2 reporter vector, to detect the potential interactions between hsa-miR-760 and HMOX1 3'UTR. We further modi ed the psiCHECK2 vector to express the luciferase-HMOX1 fusion protein, by deleting a "T" nucleotide from the stop codon "TAA" of Renilla luciferase gene based on the change mutation method. Then the coding region of HMOX1 that containing the response elements of hsa-miR-760 was subcloned into the modi ed psiCHECK2 vector. Speci c mutations in the target sites matching to hsa-miR-760 seed sequence were introduced by PCR-based site-speci c mutagenesis.
For the luciferase assays, HBE and 293T cells were seeded at 60% con uency in 96-well plates and cotransfected by constructed plasmids (100 ng/well) together with hsa-miR-760, or negative control (NC) mimics (100 nM) using lipo2000. Cells were harvested at 48h after transfection, and the luciferase activity was measured using the dual-luciferase kit (Promega, Madison, WI).

RNA isolation and quanti cation
All mRNA and miRNA primer sequences used in this study were listed in Supplementary Table 2. Both mRNAs and miRNAs were extracted from cells using TRIZOL, and cDNA was synthesized with a random primer or speci c miRNA or stem-loop reverse transcription primers, using Revert Aid First Strand cDNA Synthesis Kit (Thermo scienti c). Quantitative Real-Time PCR (qRT-PCR) was conducted using SYBR Green Kit (Qiagen) on the Light Cycler 480 Analyzer (Roche). All of the reactions, including the notemplate controls, were run in triplicate. The miRNA levels were normalized to U6 snRNA, while mRNA levels were normalized to GAPDH.

mRNA degradation assay
This assay is designed to measure the decay kinetics of mRNAs. HBE cells were transfected with hsa-miR-760 mimics or NC mimics, incubated for 24h, and exposed to actinomycin D ( nal concentration: 3μg/ml). The cells were then harvested at 0, 4, 8, and 16 h after actinomycin D treatment, respectively, and total RNAs were prepared to test the relative HMOX1 expression as mentioned above.

RNA immunoprecipitation (RIP)
RIP assay was conducted using the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, Burlington, MA), according to the manufacturer's protocols. Brie y, the cell lysate was incubated with antibodies against YBX1 and IgG control (Abcam, Cambridge, MA) overnight at 4 °C. The RNA-protein complex was recovered by Protein G Dynabeads and reverse cross-linked with proteinase K. Finally, the protein-binding RNAs were extracted using TRIZOL reagent and analyzed by qRT-PCR.

Western blot analysis
Total proteins were isolated from the cells using RIPA lysis buffer (Thermo Scienti c). Antibodies against HMOX1 and GAPDH were purchased from Abcam, and antibodies against BCL2 and BAX were obtained from Cell Signal Technology (Danvers, MA). Western blotting assays were carried out using the Odyssey™ Western Blotting Kit (LI-COR Biosciences), and analyzed based on the Odyssey CLx Infrared Imaging System.

TUNEL Apoptosis Assays
Apoptosis was detected using the riboAPO ™ One-Step TUNEL Apoptosis Kit (RiboBio, Guangzhou, China), according to the manufacturer's protocol. In brief, cells were xed in 4% paraformaldehyde, permeabilized with 0.5% Triton X-100, and incubated with TdT Enzyme and TAM-dUTP Mixture protected from light, serially. The reaction was stopped by SSC, and the apoptotic cells stained by TUNEL were observed under a uorescence microscope.

ROS Detection
Cells were treated with PM 2.5 as indicated and then exposed to membrane permeable uorescent probe DCFH-DA ( nal concentration: 10 mM). After incubation for 20min at 37°C, the cells were harvested and resuspended in PBS. Fluoresce signaling was detected by Beckman Coulter (USA).

Statistical analyses
The quantitative variables are exhibited as mean ± standard deviation (SD). Statistical analysis was conducted using SPSS Statistics Version 21.0 (SPSS Inc., Chicago, IL). Student's t test was used to compare the differences in luciferase reporter gene assays, while one-way ANOVA on ranks test was used to detect the differences between subgroups for protein or RNA levels. Differences were regarded as statistically signi cant when P < 0.05.        levels. *P < 0.05, **P < 0.01; mi-NC, miRNA mimic negative control.