Abstract

   Transforming growth factor-beta 1 (TGF-β1), a pro-fibrotic
   tumour-derived factor promotes fibroblast differentiation in the tumour
   microenvironment and is thought to contribute to the development of
   pro-tumourigenic cancer-associated fibroblasts (CAFs) by promoting
   myofibroblast differentiation. miRNA dysregulation has been
   demonstrated in myofibroblast transdifferentiation and CAF activation,
   however, their expression varies among cell types and with the method
   of fibroblast induction. Here, the expression profile of miRNA in human
   primary oral fibroblasts treated with TGF-β1, to derive a
   myofibroblastic, CAF-like phenotype, was determined compared to
   untreated fibroblasts. Myofibroblast transdifferentiation was
   determined by the expression of alpha-smooth muscle actin (α-SMA) and
   fibronectin-1 extra domain A (FN-EDA1) using quantitative real-time PCR
   (qRT-PCR) and western blot. The formation of stress fibres was assessed
   by fluorescence microscopy, and associated changes in contractility
   were assessed using collagen contraction assays. Extracellular vesicles
   (EVs) were purified by using size exclusion chromatography and
   ultracentrifugation and their size and concentration were determined by
   nanoparticle tracking analysis. miRNA expression profiling in oral
   fibroblasts treated with TGF-β1 and their extracellular vesicles was
   carried out using tiling low-density array cards. The Database for
   Annotation, Visualization, and Integrated Discovery (DAVID) was used to
   perform functional and pathway enrichment analysis of target genes. In
   this study, TGF-β1 induced a myofibroblastic phenotype in normal oral
   fibroblasts as assessed by expression of molecular markers, the
   formation of stress fibres and increased contractility. TaqMan
   Low-Density Array (TLDA) analysis demonstrated that miR-503 and miR-708
   were significantly upregulated, while miR-1276 was significantly
   downregulated in TGF-β1-treated oral fibroblasts (henceforth termed
   experimentally-derived CAF, eCAF). The gene functional enrichment
   analysis showed that the candidate miRNAs have the potential to
   modulate various pathways; including the Ras associated protein 1
   (Rap1), PI3K-Akt, and tumour necrosis factor (TNF) signalling pathways.
   In addition, altered levels of several miRNAs were detected in eCAF EV,
   including miR-142 and miR-222. No differences in size or abundance of
   EV were detected between eCAF and normal oral fibroblast (NOF). Little
   overlap was observed between changes in cellular and EV miRNA profiles,
   suggesting the possibility of selective loading of EV miRNA. The study
   reveals miRNA expression signature could be involved in myofibroblast
   transdifferentiation and the miRNA cargo of their EV, providing novel
   insight into the involvement of miRNA in CAF development and function.

Introduction

   Several lines of evidence have demonstrated that the altered stroma
   microenvironment makes a significant contribution to the malignant
   progression of cancers, including oral squamous cell carcinomas (OSCC)
   [[34]1]. The tumour microenvironment is a dynamic milieu consisting of
   a mixture of cancerous cells together with stromal cell populations.
   The majority of tumour stromal cells are activated fibroblasts which
   commonly express α-SMA [[35]2] and FN-EDA1 [[36]3]. TGF-β1 is a
   well-established factor capable of inducing myofibroblast phenotype in
   vitro [[37]4]. Activin A, a member of the TGF-β family of proteins,
   provokes myofibroblast differentiation, similar to that of fibroblasts
   stimulated with TGF-β1 [[38]5]. Other growth factors and cytokines
   mediate myofibroblast transdifferentiation including interleukin-6
   (IL-6), platelet-derived growth factor (PDGF) and connective tissue
   growth factor (CTGF) [[39]6]. Taken together, various cytokines and
   growth factors are capable of modulating the myofibroblast
   transdifferentiation, although the precise signals, and underlying
   molecular mechanisms, promoting the formation of myofibroblastic CAF in
   vivo remain to be fully elucidated.

   miRNAs are small non-coding RNA molecules that regulate target gene
   expression and modulate various biological processes [[40]7].
   Dysregulation of miRNA expression has been implicated in myofibroblast
   transdifferentiation and the CAF phenotype [[41]3, [42]8–[43]10]. Shen
   et al [[44]11] reported the upregulation of miR-7 in CAFs of oral
   cancer compared with their paired normal fibroblasts, and inhibition of
   miR-7 induces a functional change of CAFs into normal fibroblasts.
   Thus, differently expressed miRNAs in myofibroblasts opens the
   possibility of a treatment approach targeting the tumour stroma with
   miRNA modulators (mimic or inhibitor). In addition to their
   cell-autonomous effects, miRNAs are increasingly recognised to play a
   role in intercellular signalling via their presence in EV. EV,
   encompassing exosomes, microvesicles and other nano-scale lipid
   encapsulated vesicles, are released by every cell type and present in
   every body fluid studied to date. They carry 2cargo including RNA, DNA,
   protein and lipid, and have been implicated by a range of evidence to
   induce responses in recipient cells, at least in part by the transfer
   of functional miRNA. Although the miRNA cargo of EV derived from cancer
   cells has been widely studied, much less is known about their presence
   or functional roles in fibroblast-derived EV.

   While most studies focused on the effects of miRNA-derived CAFs on
   cancer progression, only limited information is available regarding the
   functional role of miRNA in regulating myofibroblast
   transdifferentiation. In this study, we have successfully characterised
   a myofibroblastic, CAFs-like phenotype in oral fibroblasts. We have
   further determined the differential miRNA expression; miR-503 and
   miR-708 were upregulated, while miR-1276 was downregulated in eCAFs.
   Furthermore, we identified pathways potentially regulated by altered
   miRNA that may influence CAF phenotype and function. In addition, we
   demonstrate the presence of a wide array of miRNA in fibroblast and
   eCAF-derived EV and identify differences in EV cargo that may
   potentially contribute to their pro-tumorigenic functions.

Materials and methods

Cell culture

   Fibroblast cultures (NOF804 and NOF822) were kindly provided by Dr.
   Helen Colley (University of Sheffield, UK; University Research Ethics
   Approval Reference Number 003463). Fibroblasts were maintained in
   Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% Fetal
   Bovine Serum, L-glutamine and 100 U/ml penicillin and 100 μg/ml
   streptomycin. All materials were purchased from Sigma-Aldrich, USA.
   Fibroblasts were incubated in a humidified atmosphere containing 5%
   CO[2] at 37 ^0C. Cell passages from 1 to 7 were used in this study. For
   TGF-β1 treatment, fibroblasts were incubated with ranging
   concentrations of TGF-β1 (0.05–5 ng/ml) for 24 h and/or 48 h.

Quantitative real-time PCR

   Total RNA was extracted from fibroblasts using the RNeasy mini kit
   (Qiagen, Germany) according to the manufacturer’s instructions. cDNA
   conversion of 100 ng RNA was done using the High-Capacity cDNA reverse
   transcription kit (Life Technologies, UK) according to the
   manufacturer’s instructions. Total cDNA was amplified using the Power
   SYBR green PCR master mix (Life Technologies, UK). Primers sequence
   were as follows; U6 Forward (F) 5′CTCGCTTCGGCAGCACA3′, U6 Reverse (R)
   5′AACGTTCACGAATTTGCGT3′, αSMA F 5′GAAGAAGAGGACAGCACTG3′, αSMA R
   5′TCCCATTCCCACCATCAC3′, fibronectin-1 with extra domain A (FN-EDA1)
   5′TGGAACCCAGTCCACAGCTATT3′, FN-EDA1 R 5′GTCTTCTCCTTGGGGGTCACC3′.
   Relative quantitation of targets in different samples was calculated by
   the ΔΔCt method, normalised to endogenous control U6.

Immunocytochemistry

   Fibroblasts were fixed with methanol (Fisher-Scientific, UK) for 20 min
   and permeabilised by using 4 mM sodium deoxycholate (Fisher-Scientific,
   UK) for 10 min. Cells were blocked for non-specific binding sites in
   2.5% (w/v) bovine serum albumin in phosphate buffered saline (PBS) for
   30 min, followed by incubating with a FITC-conjugated anti-human α-SMA
   mouse antibody (clone 1A.4; 1:100; Sigma-Aldrich,USA) for overnight at
   4°C. Coverslips were washed twice with PBS, mounted using mounting
   media containing 4′,6-diamidino-2-phenylindole (DAPI; Vector
   Laboratories Inc, USA), viewed using a Ziess Axioplan 2 fluorescence
   light microscope at 40x magnification and photographed using Proplus
   7.0.1 image software.

Gel contraction assay

   Fibroblasts (2.5 x 10^5 cells/well) were mixed with 4 mg/ml of rat tail
   collagen (Roche,UK) in DMEM and pH adjusted to 7 using 0.1 mM NaOH
   (Sigma-Aldrich,UK). The cell:collagen mixture was added to 24 well
   plates and incubated for 45 min for gel to polymerize. The gels were
   incubated overnight at 37 ^0C in serum-free DMEM. The gels were then
   loosened from the edges of the well by a scalpel and were incubated
   with serum free medium containing 5 ng/ml of TGF-β1 for 48 h. Collagen
   lattices were photographed and the surface area for each lattice was
   calculated using the Image J.

Immunoblotting

   Total protein lysates were prepared in radioimmunoprecipitation assay
   buffer (Sigma-Aldrich, USA) supplemented with complete mini protease
   inhibitor cocktail (Roche, UK). Protein quantification was determined
   by BCA protein assay kit (Thermo Scientific, UK). Protein lysates (20
   μg) were resolved by 12% (v/v) sodium dodecyl sulphate-polyacrylamide
   gel electrophoresis (SDS-PAGE) and transferred onto nitrocellulose
   membrane (Milipore, USA) by iBlot dry transfer (Life technologies, UK).
   Following blocking of non-specific protein binding, membranes were
   incubated with mouse monoclonal anti-human αSMA (1:1000; Sigma–Aldrich
   USA) or mouse monoclonal β-actin (1:4000, Sigma-Aldrich, USA) at 4°C
   overnight. After a few washing steps, the membrane was incubated with a
   horseradish peroxidase-conjugated secondary antibody (1:3000, Cell
   Signalling Technology, USA) for 1 h at 37°C. All antibodies were
   diluted in 5% (w/v) skimmed milk powder and 3% (w/v) bovine serum
   albumin in Tris-buffered saline (TBS) with 0.05% (v/v) Tween 20
   (TBS-T). The membrane was visualised by enhanced Pierce
   chemiluminescence (Thermo Scientific, UK). Densitometry was performed
   using Image J.

Taqman Tiling Low-Density Array (TLDA)

   miRNA expression profiling was performed using commercially available
   Tiling Low-Density Array; TLDA (Life Technologies, UK), according to
   the manufacturer’s protocol. Briefly, total RNA (3 ng) was used as an
   input for retrotranscription using a TaqMan MicroRNA Reverse
   Transcription Kit (Life Technologies, UK) and the TaqMan Megaplex pool
   kit (Life Technologies, UK), which are customised to run the TLDAs.
   Preamplification was performed using TaqMan PreAmp Master Mix (Life
   technologies, UK) according to the manufacturer’s protocol. RT-PCR was
   carried out using a 7900HT thermocycler (Applied Biosystems, USA). The
   raw data were analyzed by Data Assist Software version 3.0 (Life
   Technologies, UK). The ΔCt values were normalized to U6 endogenous
   control at a Ct cut-off value of 34, according to technical
   recommendation. EV-miRNA abundance was normalised to mean Ct value
   across all detected miRNA in each sample with a detection cut-off of Ct
   32. Fold change of the differentially expressed miRNAs were calculated
   from ΔΔCt values relative to the untreated sample.

Extracellular vesicle isolation and characterisation

   EV were isolated as described in Peacock et al [[45]12]. Briefly, NOF
   or eCAF were rinsed with PBS and incubated in serum-free DMEM for 24–72
   h. Conditioned medium collected and centrifuged at 300 x g for 10 min,
   2000 x g for 10 min and 10,000 x g for 30 min. The supernatant was
   reduced to 0.5 ml using a Vivaspin-20 (100 kDa molecular weight
   cut-off) column (GE Healthcare, Buckinghamshire, UK). EV were isolated
   by size exclusion chromatography using Sepharose CL-2B (GE Healthcare,
   Uppsala, Sweden) stacked in disposable Econo-Pac columns (Biorad,
   Watford, UK) and eluted in PBS. Where required, EVs were pelleted by
   ultracentrifugation at 100,000 x g for 1 h. Size profile and
   quantification of EV was performed by nanoparticle tracking analysis
   using a Zetaview instrument (Particle-Metrix) according to the
   manufacturer’s instructions.

Bioinformatics analyses

   Target genes of candidate miRNAs were predicted using online
   bioinformatic tools; TargetScan
   ([46]http://www.targetscan.org/vert_71/), miRDB
   ([47]http://www.mirdb.org/), and miRWalk
   ([48]http://zmf.umm.uniheidelberg.de/apps/zmf/mirwalk2/index.html).
   Venn diagrams ([49]http://bioinformatics.psb.ugent.be/webtools/Venn/)
   were used to obtain overlapping target genes from the three
   bioinformatic tools. GO enrichment analyses of biological process (BP),
   cellular compartment (CC), molecular function (MF) and Kyoto
   Encyclopedia of Genes and Genomes (KEGG) pathway were obtained using
   the DAVID ([50]https://david.ncifcrf.gov/) bioinformatics tool
   [[51]13]. For EV-associated KEGG pathway analysis, DIANA miR-Path v3
   was used [[52]14]. A false discovery rate (FDR) <0.05 was used as the
   cut-off criteria.

Statistical analyses

   Data are expressed as the mean ± standard error of the mean (sem) from
   three independent experiments, unless otherwise stated. Statistical
   analyses were made between two groups using a paired two-tailed
   Student’s t-test (Graph Pad Prism 7), as appropriate and indicated in
   figure legends. A value of p < 0.05 was considered statistically
   significant.

Results

TGF-β1 induces myofibroblast transdifferentiation

   In order to establish a model of CAF-like (eCAF) myofibroblast
   differentiation, we exposed NOF to TGF-β1. In keeping with previous
   reports, TGF-β1 induced expression of α-SMA and FN-EDA1 transcripts in
   eCAF with maximum α-SMA and FN-EDA1 expressions occurring at 48 h
   post-treatment at a concentration of 5 ng/ml ([53]Fig 1A and 1B). Thus,
   this concentration and time point were used for subsequent experiments.
   Similarly, TGF-β1 caused increases in α-SMA protein abundance and the
   appearance of α- SMA-positive fibroblasts in eCAF ([54]Fig 1C and 1D).
   The percentage of α- SMA positive cells in eCAF was significantly
   higher than that in NOF. These data are concordant with greater
   contractility of collagen 1 lattice in eCAF compared to NOF ([55]Fig
   1E).

Fig 1. TGF-β1 induces myofibroblastic phenotype in human oral fibroblasts.

   [56]Fig 1
   [57]Open in a new tab

   Fibroblasts were treated with TGF-β1 (0.05–5 ng/ml) or serum-free
   medium (untreated), for 24 h and 48 h (A-B). Fibroblasts were subjected
   to RNA extraction, cDNA preparation and the expression of α-SMA (A) and
   FN-EDA1 (B) was determined using qPCR. Fibroblasts were treated with
   TGF-β1 (5 ng/ml) or serum-free medium (untreated) 48 h (C-E). Cell
   lysates were immunoblotted against α-SMA and β-actin (C). Fibroblasts
   were fixed in 100% methanol and were observed under a fluorescence
   microscope after being stained with the α-SMA-FITC antibody (green) and
   DAPI (blue) (D). The collagen gel contractility was assessed by
   measuring the gel surface area (E). Each data represents the mean ± SEM
   from three independent experiments. Statistical analysis was determined
   using two-tailed Student T-test with *p < 0.05, **p < 0.005, ***p <
   0.001, ****p < 0.0001 compared to untreated.

miRNAs are differentially expressed in eCAF

   It has been reported that several miRNAs are involved in myofibroblast
   transdifferentiation and alterations in their expression provoke or
   attenuate the CAF-like myofibroblastic phenotype. We next examined the
   expression of cellular miRNA in eCAF and NOF to gain insight to the
   potential role of miRNA in regulating the eCAF phenotype. We observed
   that only three miRNAs were differentially expressed in eCAF compared
   to NOF. Two miRNAs (miR-503; 28.6-fold and miR-708; 2.8-fold) were
   significantly upregulated, while miR-1276 was significantly
   downregulated (0.1-fold) in eCAFs ([58]Fig 2A). In order to gain
   understanding of the potential biological significance of these
   changes, we used an in silico approach to identify potential target
   genes of miR-503, miR-708 and miR-1276 using three publicly available
   bioinformatics tools: TargetScan, miRDB, and miRWalk. As shown in
   [59]Fig 2B, we identified a total number of 745 overlapping predicted
   target genes for these miRNAs from these algorithms. To understand the
   functional roles and mechanisms of identified target genes, GO and KEGG
   analyses were performed using DAVID. The results demonstrated that
   genes were enriched in five biological process (BP): synapse
   organization, phosphatidylinositol phosphorylation,
   phosphatidylinositol biosynthetic process, actin cytoskeleton
   reorganization and phosphatidylinositol-mediated signalling ([60]Fig
   2C). While for cellular component (CC), the genes were enriched in
   nucleoplasm, cell-cell junction, neuron projection, cytoplasm and
   plasma membrane ([61]Fig 2D). Additionally, molecular function (MF)
   analysis showed that the genes were enriched in protein binding,
   protein kinase activity, 1-phosphatidylinositol-3-kinase activity,
   kinase activity, and protein serine/threonine kinase activity ([62]Fig
   2E). The KEGG pathway analysis indicated that the genes were mainly
   involved in pathways in cancer, Rap1 signalling pathway, PI3K-Akt
   signalling pathway, and TNF signalling pathway ([63]Fig 2F).

Fig 2. Differentially expressed miRNAs in myofibroblast transdifferentiation
and their gene ontology and signaling pathway enrichment analysis.

   [64]Fig 2
   [65]Open in a new tab

   Fibroblasts were treated with or without TGF-β1 (5 ng/ml) for 48 h. RNA
   was extracted, cDNA was amplified and used for TLDA analysis. Only
   miRNAs above the horizontal line are significant (*p<0.05), the green
   dots denote miRNAs that are downregulated, red dots represent miRNAs
   that are upregulated, and the black dots are either miRNAs that are not
   significant or below the fold change cut off (A). Venn diagrams
   ([66]http://bioinformatics.psb.ugent.be/webtools/Venn/) of target genes
   of each miRNAs using three bioinformatics tools (B). GO enrichment
   analysis demonstrated the top 5 genes enriched in biological process
   (BP) (C), cellular component (CC) (D), molecular function (MF) (E), and
   the KEGG pathway analysis (F).

eCAF-derived extracellular vesicles display altered miRNA cargo

   In recent years an understanding has emerged of the contribution of
   miRNA to intercellular signalling mediated by small membrane-enclosed
   moieties collectively termed EV [[67]15]. The miRNA cargo of EV is
   reported to mediate numerous effects of cancer cell-derived EV in the
   tumour microenvironment, but considerably less is understood of the
   role EV-associated miRNA might play in CAF functionality. Having
   established changes in the cellular expression of miRNA in eCAF
   formation, we next therefore assessed the miRNA cargo of eCAF-derived
   EV compared to their normal fibroblast counterparts. We first isolated
   EV from NOF and eCAF and assessed their physical characteristics by
   nanoparticle tracking. No significant differences were observed in the
   size profile of NOF-EV compared to eCAF-EV, nor was there a significant
   difference between the number of EVs released by the different
   phenotypes ([68]Fig 3A and 3B). miRNA profiling however revealed a
   number of differentially represented miRNA species ([69]Fig 3C).
   Substantial differences in the abundance of a number of miRNAs were
   observed for EV isolated from eCAF compared to normal fibroblasts; this
   included a greater than ten-fold elevation in the abundance of miR-92a,
   miR-222 and miR-186, and a greater than ten-fold reduction in miR-223
   and miR-142-3p ([70]Fig 3C). Downregulation of the small nuclear RNA U6
   negated its use as a reference gene for normalisation, in keeping with
   previous reports [[71]16]. KEGG analysis of putative targets of the
   three most elevated miRNA in eCAF-EV (miR-222, miR-92a and miR-186)
   indicated that if delivered to recipient cells, the miRNA cargo has the
   potential to modulate pathways including ‘adherens junctions’,
   ‘extracellular matrix (ECM) receptor interaction’ and ‘proteoglycans in
   cancer’, key processes in the tumour microenvironment ([72]Fig 3D).

Fig 3. Characterisation of eCAF-derived EV and their miRNA cargo.

   [73]Fig 3
   [74]Open in a new tab

   EV were isolated from the conditioned media of fibroblasts treated with
   TGF-β1 (5 ng/ml) or their untreated counterparts using size-exclusion
   chromatography. The concentration (A) and size (B) of the isolated
   particles was analysed using nanoparticle tracking (Zetaview, Particle
   Metrix). Total RNA was extracted from isolated EV and their miRNA cargo
   was assessed using tiling low- density array quantitative PCR (C). KEGG
   pathway analysis on the three most elevated miRNA (miR-222, miR-92a and
   miR-186 combined) was conducted using DIANA (D) [[75]14].

Discussion

   Emerging evidence has demonstrated that miRNAs are involved in CAF
   activation from resident stromal fibroblasts. However, the
   heterogeneity of miRNA expression varies due to patient-to-patient
   variability. It is well-known that cancer cells secrete high levels of
   TGF-β, and this paracrine secretion is believed to trigger a transition
   of resident fibroblasts to CAF phenotype with myofibroblastic features
   in various cancers [[76]17]. Therefore, with this in mind, the model
   for TGF-β-induced myofibroblast transdifferentiation was used to
   confirm that the phenotype assessed in this study were
   CAF-like-myofibroblastic cell. Myofibroblasts are commonly
   characterised by the expression of activation markers, α-SMA and de
   novo expression of fibronectin [[77]18, [78]19]. Increased α-SMA and
   fibronectin expression is consistent with the study of Melling et al
   [[79]3] and Dally et al [[80]20]. α-SMA is rapidly incorporated into
   actin stress fibres, thereby increasing the contractility of
   fibroblasts [[81]21]. The appearance of actin stress fibres is closely
   related to the generation of contractile force [[82]22, [83]23]. In
   keeping with this, TGF-β1 provoked the development of α-SMA positive
   stress fibres and promoted the contractile phenotype of NOF. Herein, we
   show that TGF-β1 profoundly induces CAF-like-myofibroblastic phenotype
   in NOF.

   miRNAs are likely to play a role in TGF-β1 signalling; most members of
   the TGF-β1 pathway are known or predicted to be targeted by one or more
   miRNAs [[84]24]. Moreover, TGF-β1 directly regulates the biogenesis of
   miRNA through Smads [[85]25]. We thus think that miRNA plays a role in
   myofibroblast transdifferentiation. We show that miR-503 and miR-708
   were highly upregulated in eCAF, compared to NOF. Several studies
   demonstrated that miR-503 modulates myofibroblast transdifferentiation
   and fibrosis. Wu et al [[86]26] demonstrated that miR-503 expression
   was decreased in TGF-β1 stimulated lung fibroblasts and upregulation of
   miR-503 attenuated CAF-like myofibroblastic phenotype in
   myofibroblasts. Several miRNAs are involved in cardiac fibrosis
   (reviewed in [[87]27]), and another study has proposed that miR-503
   modulates fibrosis by enhancing the ECM deposition in cardiac
   fibroblasts [[88]28]. miR-708 is one of the more recently discovered
   miRNAs to play a role in fibrosis. A recent study demonstrated that
   miR-708 was upregulated in fibrotic liver tissues and ectopic
   expression of miR-708 inhibits hepatic stellate cell activation and
   reduces the ECM accumulation [[89]29]. We also showed that miR-1276 was
   significantly downregulated in eCAF, compared to NOF. Several studies
   have shown that miR-1276 participates in cancer progression, however,
   to date, the functional role of miR-1276 in myofibroblast
   transdifferentiation or fibrosis has not been investigated. Taken
   together, these miRNAs might be involved in myofibroblast
   transdifferentiation. Further functional analyses using miRNA
   inhibitors or miRNA mimics will give a more definite understanding of
   the involvement of these miRNAs in myofibroblast transdifferentiation
   in the future.

   GO analysis demonstrated that the deregulated genes were mainly
   involved in phosphatidylinositol-mediated signalling (ontology: BP),
   nucleoplasm (ontology: CC), and protein binding (ontology: MF).
   Previous study has revealed that phosphatidylinositol-mediated
   signalling regulates pro-inflammatory interleukin 32α (IL-32α)
   expression in human pancreatic periacinar myofibroblasts [[90]30]. At
   the level of cellular component, the target genes were associated with
   miRNA translocation. Some miRNAs are present at much higher levels in
   the nucleolus and occur at very low levels in the nucleoplasm and/or
   cytoplasm [[91]31]. As for the molecular function, numerous proteins
   including Smad7 have been reported to bind to TGF-β receptors [[92]32].
   The most enriched pathways from KEGG analysis were the TNF-signalling
   pathway and PI3K-Akt signalling pathway. TNF mediates various
   biological processes such as cell differentiation, cell proliferation,
   inflammation and apoptosis [[93]33]. TGF-β can induce activation of
   PI3K-Akt indirectly via smad-dependent or–independent pathways [[94]34,
   [95]35]. Additionally, TGF- β indirectly activates PI3K-Akt signalling
   by inducing the expression of several miRNAs. Previous studies have
   revealed that TGF- β activates the PI3K-Akt pathway by inducing the
   expression of miR-216a/217 and miR-21 in kidney cancer cells and
   hepatoma cells [[96]36, [97]37].

   Given the well-described roles for EV in modulating paracrine
   signalling in cancer and other contexts by the action of their miRNA
   cargo on target genes in recipient cells, it is potentially significant
   that the data presented here indicate the differential abundance of
   specific miRNA species in EV isolated from eCAF compared to NOF.
   Although requiring further validation, several of these miRNAs have
   known their putative roles in regulating the expression of a number of
   genes and pathways relevant to cancer. miR-92a, for example, the
   abundance of which was found to be increased by more than 10-fold in
   eCAF-derived EV, is widely reported to have pro-tumourigenic effects,
   with elevated abundance in EV isolated from cancer patients associated
   with poor survival [[98]38]. The proposed mechanism of action of
   miR-92a is the promotion of epithelial-mesenchymal transition (EMT), a
   process favouring metastasis also known to be promoted by CAF-derived
   EV [[99]38, [100]39]. Another miRNA identified here as elevated in
   eCAF-derived EV, miR-222, is also reported to promote EMT [[101]40] and
   that elevated EV-associated miR-222 promotes tumourigenesis and
   correlates with poor survival in pancreatic cancer patients [[102]41].
   By contrast, miR-142-3p and miR-223, both found to be decreased in
   abundance in eCAF-derived EV, are widely ascribed tumour-suppressive
   functions [[103]42, [104]43], although this may be context dependent.
   None of the miRNA found to be significantly up or down-regulated in the
   cellular transcriptome were altered in EV, which could imply selective
   loading of miRNA into EV, but it should be noted here that these
   samples weren’t obtained from the same culture used for cellular miRNA
   profiling and therefore differences between cultures cannot be excluded
   from accounting for this observation. Taken together, this provides
   preliminary evidence that the miRNA cargo of eCAF-derived EV has the
   potential to enhance tumour growth compared to that of EV isolated from
   NOF.

   In summary, the findings reported here build on an emerging
   understanding of the role miRNA may play in the ability of TGF-β1 to
   induce a myofibroblastic CAF-like phenotype in vitro. The study also
   represents, to our knowledge, the first assessment of changes in EV
   miRNA cargo occurring during the transition of oral fibroblasts to a
   CAF-like myofibroblastic phenotype. Further experimental studies are
   needed to determine the functional role and the underlying mechanism
   (s) of these miRNAs.

Supporting information

   S1 Fig. Uncropped and original Western blot against α-SMA from
   untreated and TGF-β1-treated NOF cells.

   Proteins were collected from 3 independent batches.

   (TIF)
   [105]Click here for additional data file.^ (369.2KB, tif)

Acknowledgments