1. Introduction
Environmental tobacco smoke (ETS) exposure is associated with an increased frequency of lower respiratory tract infections, increased incidence and severity of asthma episodes, overall decreased pulmonary function, and cancer development.
Extracellular vesicles (EVs) are membrane particles released by virtually all cells, shuttling active biological molecules such as proteins, lipids, and nucleic acids to neighboring cells, and to distant sites [
1,
2,
3]. EVs have been isolated from various biofluids such as blood, breast milk, bronchial lavage, saliva, urine, amniotic, and cerebrospinal fluids [
4,
5,
6,
7,
8,
9]. EVs represent a heterogeneous population and vary in size (30–2000 nm in diameter) and composition, based on the cellular origin and environmental stimuli [
10,
11,
12]. According to their diameter, EVs can be classified into two general subgroups: 1) small EVs (<200nm), and 2) medium/large EVs (>200 nm) [
10]. Recent studies have identified EVs as critical players in intercellular communication under various physiological and pathological conditions such as neurodegenerative diseases, cancer, preterm birth, angiogenesis, immune responses, and viral infections [
3,
13,
14,
15,
16,
17,
18,
19,
20].
Because EVs transfer a complex content of biological information (nucleic acids, lipids, and proteins), the manipulation of EV cargos represents a potential tool for drug and selective biomolecule delivery to target cells [
21,
22]. MicroRNAs (miRNAs) are a type of small noncoding RNA (sncRNA) which is able to modify the stability of messenger RNAs (mRNAs), usually silencing the gene expression in various cellular processes as cell apoptosis, angiogenesis, and inflammation [
23]. Recent works have reported the presence of miRNAs in EVs released from human bronchial epithelial (HBE) cells of patients with asthma [
24], and with chronic obstructive pulmonary disease (COPD) [
25]. However, very few studies have analyzed the RNA cargo of EVs generated after ETS. Recently, Stassen and his group showed that tissue factor procoagulants are released from bronchial epithelial cells in response to contact with cigarette smoke extract through an EVs mechanism, suggesting a possible function of EVs after smoke exposure [
26,
27].
In this study, we analyzed the RNA composition of EVs derived from control and CSC-treated human small airway epithelial (SAE) cells using next generation sequencing (NGS). We observed that CSC exposure led to diverse signatures of new RNA molecules in EVs, especially snRNA types, such as miRNAs. We identified the targets of a novel miRNA, hsa-miR-3913-5p, significantly enriched in EVs derived in response to CSC treatment. Analysis of hsa-miR-3913-5p potential target genes and their biological functions suggest that this miRNA plays a role in lipid transport/binding and the regulation of gene transcription. Our data add novel information about the cargo of EVs released from AECs in response to ETS, which could be used to develop biomarkers for the diagnosis of cigarette smoke-related diseases, as well as for the development of future therapeutic approaches.
2. Materials and Methods
2.1. Small Airway Epithelial Cultures and Stimulation with Cigarette Smoke Condensate
SAE cells (Lonza Inc., San Diego, CA, USA), derived from the terminal bronchioli of cadaveric donors, were grown in culture medium containing 7.5 mg/mL bovine pituitary extract (BPE), 0.5 mg/mL hydrocortisone, 0.5 µg/mL hEGF, 0.5 mg/mL epinephrine, 10 mg/mL transferrin, 5 mg/mL insulin, 0.1 µg/mL retinoic acid, 0.5 µg/mL triiodothyronine, 50 mg/mL gentamicin, and 50 mg/mL bovine serum albumin. CSC was prepared by smoking University of Kentucky’s standard research cigarettes on an Federal Trade Commission smoke machine, as previously described [
28]. The total particulate matter collected was extracted with dimethyl sulfoxide (DMSO) to generate a 4% solution. Cell monolayers were plated in basal media (no supplemented added) for 4–6 h prior to CSC (1 μg/mL) or DMSO vehicle (control) exposure for 48 h.
2.2. Extracellular Vesicles Purification
After CSC exposure, 100 mL of cell supernatant was centrifuged at 3000 g for 15 min at 4 °C for debris removal. The clear media was subjected to further cleaning by filtration through 0.22 μm sterile filter (Millipore, MA, USA) to remove any remaining debris. The filtered media was transferred to Amicon
® Ultra-15 centrifugation filters (Millipore, Billerica, MA, USA) and centrifuged at 2500 g for 35 min. ExoQuick-TC™ (System Biosciences, USA) reagent was added to the media, mixed thoroughly, and incubated overnight at 4 °C to precipitate the EVs. The following morning, the mixture was subjected to centrifugation at 1500 g for 30 min, and the EV pellet was washed and resuspended in filtered PBS. The resuspended EVs were passed through the Exo-spin
™ columns (Cell Guidance Systems, St. Louis, MO, USA), and 300 μL of purified EVs were eluted from the column and used for experimental procedures. Protein concentration was determined using a protein assay kit from Bio-Rad, USA. The purified EVs were further characterized using nanoparticle size tracking and the determination of protein markers by Western blot analysis (
Figure 1).
2.3. Nanoparticle Tracking Analysis with ZetaView®
EVs size distribution and number of particles were analyzed using the ZetaView PMX 110 (Particle Metrix GmbH, Meerbusch, Germany) and its corresponding software (Zeta-View® 8.04.02, Particle Metrix GmbH). Samples of control or CSC EVs solution were run according to the manufacturer’s instructions and measured three times to ensure reproducibility. The machine was cleaned between samples using filtered water.
2.4. Western Blot Analysis for EVs Markers
EVs samples were lysed in a buffer (50mM TrisNaCl, 0.5% Triton, 300 mM NaCl) supplemented with protease and phosphatase inhibitor cocktail. Equal amount of proteins, 15 µg in total, were processed as described previously [
20]. The primary antibodies for Western blot were rabbit anti-human CD63 (1:1000; System Biosciences), mouse anti-human Alix (1:800; Santa Cruz), mouse anti-human EpCAM (1:500; Santa Cruz), mouse anti-human Flotillin-1 (1:500; Santa Cruz), and mouse anti-human GM130 cis –Golgi (1:800; Santa Cruz). A densitometric analysis of band intensities was calculated using the UVP VisionWorks
® Life Science Software 8.0 RC 1.2 (UVP, Upland, CA, USA), verifying for nonsaturation and subtracting background.
2.5. Extraction of EVs RNA and Next Generation Sequencing (NGS)
RNA was extracted from control or CSC EVs by the phenol/chloroform method using all RNA-grade reagents and according to our published protocol [
20]. Small RNA libraries were made using the QIAseq
® miRNA Library Kit (QIAGEN) following the manufacturer’s protocol. After Agilent Bioanalyzer analysis, the sample libraries were pooled and sequenced by the UTMB Next Generation Sequencing Core on an Illumina NextSeq550 (single end 75 base) using TruSeq SBS kit v3 (Illumina) and protocols defined by the manufacturer. The miRDeep2 software package, version 2.0.0.8, was used to trim adapter sequences from the reads and quantify miRNA read counts using the miRBase database, release 22. piRNAs were counted by mapping the trimmed reads to piRNA sequences downloaded from the DASHR database [
29] using Bowtie version 1.2.2 with parameters -v2 -l18 -a -M10 -best -strata. For other small RNAs and protein coding genes, the reads were mapped to the hg38 reference with the same Bowtie parameters. Reads per gene were counted using the feature counts function of the subread programs [
30] and the GENCODE release 29 annotation file.
2.6. Reverse Transcription (RT)-PCR
To validate the up- and down- regulated miRNA and piRNA expressions, 1 microgram of isolated RNA was converted into cDNA using miScript II for miRNAs or the miScript Plant cDNA synthesis reagents (QIAGEN) for piRNAs species, following the manufacturer’s instructions. Since piRNAs have 2′-O-Me modifications on the 3′ terminal base and are refractory to polyadenylation, a ligation reaction was performed to overcome the polyadenylation step. Reverse transcription was performed following ligation. First, 1 µl cDNA served as a template for the PCR analysis using the miScript SYBR green PCR kit (QIAGEN) and custom miRNAs and piRNAs primers (QIAGEN).
2.7. Predicted Targets of Hsa-miR-3913-5p and Functional Analysis
TargetScanHuman version 7.2, with a context ++ score, was used to predict the biological targets of hsa-miR-3913-5p identified in EVs; 19475 unique genes, 28353 transcripts, were scanned. Candidate target genes for hsa-miR-3913-5p were called whenever the gene was paired with the miRNA or its variant in the default predictions table of TargetScan. We used Panther interface to the Gene Ontology (GO) database to identify the most prevalent functions of the significant predicted target genes for hsa-miR-3913-5p. We chose three GO categories: biological process, cellular component, and molecular function. Relative frequencies were calculated against the entire human genome as background. p value < 0.05 was designated to be statistically significant.
2.8. Statistical Analysis
EV size and concentration (n = 3), western blot (n = 4), NGS (n = 3), and RT-PCR (n = 4) are representative of independent experiments. The raw read counts of NGS analysis were normalized across all samples and then used for pairwise differential expression analysis using the R package DeSeq. Significant differentially-expressed RNAs were determined by p value with a threshold of 0.05. Log2 fold changes between samples were hierarchically clustered using Pearson correlation. Some miRNAs and piRNAs exhibited a large fold change on average, but the variance among samples was too high to call the difference significant. The fold change of RT-PCR experiments was calculated by the 2-ΔΔCT method and represented mean ± SEM using GraphPad Prism v4 (GraphPad Software). A p value < 0.05 was considered statistically significant using the student t-test statistics.
4. Discussion
The aim of this study was to characterize snRNA content using high-throughput technologies in EVs derived from normal AECs after exposure to CSC, as a proxy for ETS [
28,
41], since the role of EVs and exosomes in the pathogenesis of cigarette smoke-related diseases is still mostly unknown. A recent in vitro study reported that HBE cells exposed to cigarette smoke extract (CSE) released EVs, and EVs from both normal and CSE-treated cell conditions displayed the expression of CD63, the EV marker. Also, the size and number of EVs derived from control and CSE-treated HBE cells did not vary significantly between the two groups [
25]. The presence of the CD63 marker was also confirmed in EVs isolated from untreated and CSE-treated immortalized HBE cells (BEAS 2B), and while no significant differences of EVs size were detected in both conditions, CSE exposure increased the amount of EVs released from BEAS 2B cells [
42]. Similar to the published finding of Xu’s team, our data suggested that human SAE cells are able to secrete EVs under control and cigarette smoke exposure conditions in an in vitro model. Although the differential centrifugation technique is widely used for the EV purification, we opted for the two-step procedure, i.e., precipitation reagent solution and size exclusion chromatography (SEC), as it is associated with a higher purity of EVs than differential centrifugation alone [
43]. We found that CSC exposure did not cause significant differences in the concentration and size of SAE-derived EVs. CD63 and Alix, EVs markers, were present in both groups, with EVs isolated from CSC-SAE cells showing increased Alix content. A similar result was reported for the EV isolated from CSC-exposed human monocytes [
44].
Our RNA sequencing analysis revealed that EVs isolated from CSC-treated SAE cells displayed a different amount of various small noncoding RNAs (sncRNA) compared to the control EVs. We decided to explore two snRNA classes associated with the gene regulation processes: miRNAs and piRNAs. miRNAs are 22 nucleotides long. miRNAs usually functions in gene silencing via translational repression or target degradation [
45]. NGS analysis showed that five miRNAs were upregulated (e.g., miR-3913-5p, miR-574-5p, miR-656-5p, and miR-3180-5p), and three downregulated (miR-618, miR-222-5p, and miR-130b-5p) in EVs derived from CSC-exposed cells, compared to control cells. Real-time PCR confirmed that CSC treatment led to the upregulation and enrichment of novel miRNAs in EVs, namely miR-3913-5p, miR-574-5p, and miR-500a-5p, whose function is still unexplored. Of the downregulated miRNAs, it has been previously reported that miR-222 was downregulated in the lungs of rats exposed to cigarette smoke [
46]. Changes in small RNA (including miRNA) composition have been reported in EVs derived from HBE cells after CSE treatment. Recent studies demonstrated that two miRNAs, miR-210 and miR-21, were upregulated in EVs after cell exposure to CSE [
25,
47]. When using BEAS 2B cells, He and colleagues found that a low percentage of CSE reduced the amount of miR-21 while a higher CSE percentage increased the miR-21 content of EVs [
48]. In our experiments, miR-21 was detectable in both CSC and control EVs, with a marginal increase in the CSC ones, which, however, was not statistically significant. The different type of airway epithelia cells and cigarette smoke sources could be responsible for the differences in our results. For example, Baskoro demonstrated that SAE cells exhibit a different proinflammatory response to cigarette smoke exposure compared to HBEs [
49].
Since hsa-miR-3913-5p was the highest upregulated miRNA in CSC EVs, we focused on the identification of potential genes regulated by this miRNA using the TargetScan software, and classified them into functional groups. PLEKHS1 was the potential target with the highest score, and lipid transport, lipid binding, and regulation of mRNA stability were the most enriched biological and molecular function groups. Lipids are critical components of airway mucosa, and cigarette smoke exposure has been shown to alter lipid homeostasis in mouse alveolar macrophages [
50]. It is known that PLEKH domains have a specific affinity for binding lipids product as phospholipid molecules [
51]. APOA4, NME4, and S100-A10 protein (S100A10) represented the most significant groups of targets of hsa-miR-3913-5p involved in lipid binding (
p <0.05). Although ApoA4 has not been investigated in the context of smoke exposure, ApoA-I, another apopoliprotein, exhibits multiple protective features in normal and pathological lung conditions (asthma, emphysema, influenza, and lung cancer). In particular, ApoA-I displayed anti-inflammatory and antioxidant properties in a mouse model of cigarette smoke-induced emphysema [
52]. Future studies will focus on identifying the actual targets regulated by hsa-miR-3913-5p and their biological function.
In conclusion, a better understanding of EV generation and composition in response to ETS exposure, as well as the identification of the biological targets of their cargo, could lead to their potential utilization as diagnostic biomarkers, and to the identification of novel therapeutic targets of smoke and ETS-dependent diseases.