Increasing evidence reveals that dysregulation of exosomal miRNAs in the TME plays a critical role in cancer initiation, proliferation, vascularisation and metastasis, as well as other biological characteristics [17, 18]. Exosomal miRNAs may be derived from cancer cells themselves or from stromal cells like CAFs. To date, most studies have focused on exosomal miRNAs derived from cancer cells [19, 20]. CAF-derived exosomal miRNAs are now receiving increasing attention, but no studies have yet been conducted to explore the aberrant expression profiles of exosomal miRNAs derived from CAFs in patients with SLSCC.
The present study is the first to describe dysregulation of the expression profile of miRNAs encapsulated in the CAF-derived exosomes in different patients with SLSCC and analysis of the potential involvement of their signalling cascade in cancer progression. Previous investigations have focused on only a single pair of CAFs and NFs, whereas we initially collected ten pairs of SLSCC specimens and paired normal connective tissues for primary culture. Three pairs were successfully cultured and used for subsequent investigations. Laryngoscopy revealed that all three SLSCC tumours originated from the epiglottis, with obvious enhancement in CT scans. Pathological examination showed that they were all moderately differentiated squamous cell carcinomas (Fig. 1a-i).
Both the CAFs and NFs showed long, fusiform morphology, with several cell prominences. However, the cell prominences were fewer on the NFs than on the CAFs, which probably reflected the fact that they were in a relatively inactive state (Fig. 2a-h). After purification, the CAFs were distinguished from the NFs by positive immunofluorescence staining for FAP and α-SMA, two CAFs biomarkers, indicating their activated state (Fig. 2i-r).
Exosomes, as intraluminal vesicles, range in size from 30 to 150 nm and can be isolated by various methods, such as ultracentrifugation, chromatography, density gradient separation and immunoaffinity capture techniques [21, 22]. In the present study, exosomes were isolated from conditioned media surrounding the CAFs and NFs using Ribo™ Exosome Isolation Reagent, as previously described [13].
The collected exosomes were verified as such by TEM, western blotting and NTA. TEM showed the typical size and morphology of membrane‑bound exosome particles (Fig. 2s). Previous studies have reported that exosomes are generally enriched with certain specific marker proteins, such as CD9, CD63, CD81, TSG101 and HSP70 [23, 24], with lower expression of some cellular proteins, such as cis-Golgi matrix protein GM130 [25] and endoplasmic reticulum (ER) calnexin [14, 15]. Exosomes can therefore be identified by western blotting or flow cytometry. In the present study, we selected CD63 and TSG101 (two of the well-established surface markers for exosomes) and calnexin (a negative marker of extracellular vesicles) to confirm the isolation of exosomes by western blotting. The blots revealed the presence of CD63 and TSG101and the absence of calnexin, consistent with the previously reported characteristics of exosomes markers (Fig. 2t). In addition, NTA analysis showed the exosome particles to have a size of 109 ± 44.1 nm, which was consistent with the reported size range of exosomes (30–150 nm) [26] (Fig. 2u). Our findings confirmed that these isolated particles qualified as exosomes.
We also analysed the chromosomal locations and sequence lengths of the target small RNAs extracted from the exosomes derived from all three pairs of CAFs and NFs. We found that all the mapped small RNAs mainly originated from chromosomes 3, 5, 6, 7, 13 and 14. No significant difference was detected for the distribution of exosomal small RNAs locations in the genome between CAFs and NFs (Fig. 3a). Most sequenced small RNAs had a size of approximately 20 nt and 31 nt (clean reads), which was consistent with small RNA populations (Fig. 3d).
The aberrant expression profiles of exosomal miRNAs derived from the three pairs of purified CAFs and NFs from patients with SLSCC were then analysed by next-generation sequencing (Table 1–3, Fig. 4). Most of the aberrantly expressed miRNAs were downregulated, but a few were upregulated. Eleven miRNAs (miR-452-5p, miR-651-5p, miR-16-5p, miR-424-5p, miR-378d, miR-32-5p, let-7i-3p, miR-16-2-3p, miR-136-3p, miR-221-5p and miR-34c-5p) were downregulated simultaneously in two patients. Two miRNAs (miR-184 and miR-92a-1-5p) were upregulated simultaneously in two patients. Interestingly, four exosomal miRNAs (miR-656-3p, miR-655-3p, miR-337-5p and miR-29a-3p) were downregulated simultaneously in the purified CAFs from all three patients (Fig. 5a, pie chart).
A few reports have appeared about miR-656-3p and miR-337-5p and a few about miR-655-3p in different types of cancers. For example, miR-656-3p was found to be dysregulated in colon cancer tissues [27] and downregulated in non-small cell lung cancer tissues [28]. Overexpression of miR-337-5p in colorectal cancer tissues was significantly associated with improved overall survival [29]. Downregulation of miR-655-3p was reported in ovarian cancer tissues [30] and in hepatocellular carcinoma tissues and cell lines [31]. However, all three miRNAs in these reports were extracted either from tumour tissues or from cancer cell lines. None of the miRNAs were extracted from exosomes, let alone from the CAF-derived exosomes from different patients with SLSCC.
MiR-29a-3p is a potential tumour-suppressive miRNA and is downregulated in several cancer tissues, such as papillary thyroid carcinoma tissues and cell lines [32],hepatocellular carcinoma tissues and cell lines [33], gastric cancer tissues [34] and high-grade glioma tissues [35]. More studies have been conducted on miR-29a-3p than on miR-656-3p, miR-337-5p or miR-655-3p, and studies on exosome-derived miR-29a-3p are very rare. We found only one report that examined exosomal miR-29a-3p from oral squamous cell carcinoma cell lines [36]. No studies have reported CAF-derived exosomal miR-29a-3p.
Eleven miRNAs (miR-452-5p, miR-651-5p, miR-16-5p, miR-424-5p, miR-378d, miR-32-5p, let-7i-3p, miR-16-2-3p, miR-136-3p, miR-221-5p and miR-34c-5p) were downregulated simultaneously in two patients. However, the results in previous reports were controversial. The expression of miR-452-5p was significantly increased in 387 clinical lung squamous cell carcinoma specimens [37] and renal cell carcinoma tissues [38], whereas it was downregulated in 1007 prostate cancer samples [39]. MiR-16-5p has been proposed to function as a tumour suppressor and it was downregulated in non-small cell lung cancer cell lines [40], breast cancer samples [41], hepatocellular carcinoma tissues [42], chordoma tissues [43] and glioma tissues [44]. It was also downregulated in the plasma of gastric cancer patients [45]. However, exosomal miR-16-5p was upregulated in malignant mesothelioma cell lines [46]. The expression of miR-424-5p was significantly decreased in liver cancer tissues [47], intrahepatic cholangiocarcinoma tissues [48] and basal-like breast cancer tissues and cell lines [49]. However, it was upregulated in laryngeal squamous cell carcinoma tissues [50], thyroid cancer tissues [51], gastric cancer tissues and cells [52] and oesophageal squamous cell carcinoma tissues and cell lines [53]. A significant downregulation of miR-32-5p was reported in cervical cancer tissues and cells [54], but a significantly upregulation was detected in colorectal cancer tissues [55] and prostate cancer tissues and cells [56]. MiR-34c-5p was downregulated in leukaemia stem cells [57], osteosarcoma tissues and cells [58] and aryngeal squamous cell carcinoma [59]. Let-7i-3p was upregulated in the serum-derived exosomes from osteosarcoma patients [60], but downregulated in the sera of lung cancer patients [61] and in hepatoblastoma tumours [62]. MiR-16-2-3p was downregulated in locally advanced gastric cancer samples [63] and upregulated in sera of patients with pancreatic ductal adenocarcinoma [64].MiR-221-5p was significantly upregulated in renal cell carcinoma tissues and cell lines [65], but downregulated in prostate cancer samples [66]. No studies have reported the regulation of miR-651-5p, miR-378d and miR-136-3p in cancer research.
The two miRNAs (miR-184, miR-92a-1-5p) that were upregulated simultaneously in two patients in our study have also had controversial findings. For example, miR-184 expression was upregulated in tongue squamous cell carcinoma tissues and cell lines [67, 68], hepatocellular carcinoma tissues [69] and renal carcinoma tissues [70], but it was downregulated in glioblastoma tissues and cell lines [71], nasopharyngeal carcinoma cell lines [72] and endometrial carcinoma tissues [73]. MiR-92a-1-5p was upregulated in tumours from the luminal A to the basal-like prostate cancer subtypes [74] but downregulated in urinary exosomes from prostate cancer patients [75]. These results suggest that these miRNAs play different regulatory roles in different types of tumours.
However, most of the miRNAs reported in different studies were extracted from either tumour tissues or cancer cell lines. Most of the exosmal miRNAs reported in different studies were conducted using miRNAs extracted from cancer cell-derived exosomes. Although increasing studies focused on the exosmal miRNAs extracted from CAFs, almost all of them used only one pair of CAFs and NFs. And almost no of them were conducted using miRNAs extracted from exosomes derived from CAF primary cultures from different patients. We believe that the miRNAs with the most significant statistical difference or showing the same expression trends in more than one patient sample are the miRNAs that jointly constitute a carcinogenic TME and play decisive roles in the initiation and progression of SLSCC.
The bioinformatics analysis using four online databases (TargetScan, miRDB, miRWalk and miRTarBase) predicted the potential target genes of the exosome miRNAs that were differentially expressed between CAFs and NFs. Our exploration of the potential functions of these target genes using GO analysis revealed an involvement of three GO categories (biological process, cellular component and molecular function) that had transcription and regulation of RNA as important functions (Fig. 5b-d, Fig. 6a). The KEGG pathway enrichment analysis identified the top thirty pathways involved in many types of cancers, in critical signalling pathways in cancer initiation and progression and in regulation of the cell cycle. Our findings indicated that the differentially expressed exosomal miRNAs play a critical role in the processes of tumourigenesis and in the regulation of cancer-associated signalling pathways, the cell cycle and other aspects of cancer biology (Fig. 6b).
We also constructed an interaction network of selected exosomal miRNAs and target genes. This network contained 12 aberrantly expressed exosomal miRNAs that formed a wide range of connections with the corresponding target genes (Fig. 7a). Of these, the top five miRNAs were miR-16-5p, miR-29a-3p, miR-34c-5p, miR-32-5p and miR-490-5p. MiR-16-5p was reportedly downregulated in breast cancer [76], hepatocellular carcinoma [77] and glioma [78], in agreement with our findings. Overall, miR-29a-3p has been reported to function as a tumour suppressor and is downregulated in many types of cancers, such as thyroid carcinoma [32], colorectal carcinoma [79] and hepatocellular carcinoma [80].The downregulation of miR-34c-5p has been reported in several types of cancer, such as leukaemia [57] and osteosarcoma [58], and it has been closely associated with poor prognosis. Similarly, miR-490-5p is markedly downregulated in hepatocellular carcinoma tissues [81], in renal cell carcinoma tissues and cells [82] and in childhood neuroblastomas and cell lines [83]. MiR-32-5p was significantly downregulated in cervical cancer tissues and cells [54] but significantly upregulated in colorectal cancer tissues [55]. In our study, the expression of exosomal miR-32-5p was significantly repressed in CAFs compared to NFs, indicating that miR-32-5p might play different regulatory roles in different cancers.
We also found the top five most common overlay target genes (CCND1, CDKN1B, CDK6, PTEN and FOS) (Fig. 7b, Table 4). Of these, CCND1, CDKN1B and CDK6 are critical regulatory subunits required for the cell cycle G1/S transition. Cancer cells frequently overcome pRb-dependent growth inhibition by phosphorylation thereby inactivating pRb function by cyclin-dependent kinase (CDK) partnered with cyclin D, termed cyclin-CDK complexes. The CDKN1B gene that encodes for the p27 protein, acts as a cyclin dependent kinase (CDK) inhibitor. Dysfunction of these three genes may therefore lead to uncontrolled cell cycle progression and ultimately to tumourigenesis [84, 85].
PTEN acts as an inactivated tumour suppressor gene in cancer. It is a critical negative regulator of the PI3K signalling pathway, which is one of the most significant cell growth and survival signalling pathways in cancer. Variation in or loss of the PTEN gene is commonly observed in many types of human cancers. PTEN inhibits tumourigenesis through different mechanisms, such as phosphatase-dependent and independent activities that modulate a variety of cellular functions including DNA repair, growth, proliferation and cell motility [86, 87]. The Fos gene family (c-Fos, FosB, Fra-1 and Fra-2) dimerise with Jun proteins to form the transcription factor complex AP-1. Genes regulated by AP-1 include important regulators of proliferation, differentiation, invasion, metastasis, hypoxia and angiogenesis [88] and may play important roles in the initiation and progression of cancer.