miRNA Regulation of the Hyperproliferative Phenotype of Vascular Smooth Muscle Cells in Diabetes

Phenotypic switch of vascular smooth muscle cells (VSMCs) is a crucial event in the etiopathogenesis of vasculoproliferative diseases such as atherosclerosis, restenosis, and asthma (1–6). Diabetes mellitus (DM) exacerbates the synthetic/proliferative phenotype of VSMCs, which underlies the very high rate of vascular complications in patients with diabetes (7–9). Several key pathological factors associated with DM contribute to a particularly aggressive phenotypic switch of VSMCs in response to injury (10–12).

miRNAs have been implicated in several pathophysiological conditions including vasculoproliferative diseases (13–16). However, there is scant current knowledge about the role of miRNAs in VSMC dysfunction in diabetic vascular disease. Recently, it has been reported that DM misregulates miR-504, which promotes VSMC dysfunction in db/db mice with type 2 DM (T2DM) (17). However, it is likely that the detrimental diabetic phenotype of VSMCs in vasculoproliferative disease is mediated by multiple miRNAs whose contemporary dysregulation act together to alter the multifactorial vascular response in diabetes.

Using RNA sequencing (RNA-Seq) technology, here we identified a cluster of six differentially expressed miRNAs after vascular injury in rats with type 1 DM (T1DM) or T2DM. The latter were named vascular “Diab-miRs.” Among these Diab-miRs, contemporary dysregulation of miR-29c and miR-204 was shown to be necessary and sufficient to determine the exaggerated proliferative phenotype of VSMCs in DM. Indeed, local release of a bicistronic construct to simultaneously upregulate miR-29c and inhibit miR-204 prevented neointima formation after vascular injury in diabetic animals.

All animal experiments were performed under directive 2010/63/EU of the European Parliament and were approved by the Italian Ministry of Health. All rat strains were purchased from Charles River Laboratories. The study design is summarized in Supplementary Fig. 1A. To induce experimental T1DM, streptozotocin was injected intraperitoneally in male adult Wistar rats (18). As a model of T2DM, male obese inbred Zucker diabetic Fatty (ZDF) rats were used. Balloon injury of the right carotid artery was performed in rats, as previously described (13,18). Arterial stenosis (neointimal formation) was assessed at 14 days after balloon injury, as previously described (5,13,18).

At 2 days after injury, total RNA was extracted from carotid arteries in each group and processed for RNA-Seq as previously described (19–23). Indexed libraries were prepared from 600 ng/each purified RNA pool with TruSeq Stranded Total RNA Sample Prep Kit (Illumina). Libraries were sequenced at a concentration of 10 pmol/L per lane on HiSeq1500 platform (Illumina). The raw sequence files generated underwent quality control analysis using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/), and the quality-checked reads were then aligned to the rat genome (rn4 assembly) using TopHat version 2.0.10.

Gene annotation was obtained for all known genes in the rat genome, as provided by Ensembl genome browser.

We calculated the number of reads mapping to each transcript with HTSeq-count. Raw read counts were then used as input to DESeq for the calculation of normalized signal for each transcript in the samples, and differential expression was reported as the fold change along with associated adjusted (Benjamini-Hochberg) P values. For small RNA-Seq, indexed libraries were prepared from 1 μg/each purified RNA pool. Sized indexed miRNA libraries were gel purified and sequenced in triplicate on a HiSeq1500 system (Illumina). The sequence data obtained were analyzed using the iSmart pipeline, using as reference miRBase version 18. Samples were technical quadruplicates. However, the single biological sample for each group refers to the pooling of RNA obtained from six carotid arteries from a corresponding number of animals. Functional and pathway analyses were performed using Ingenuity Pathway Analysis (IPA; Qiagen [www.ingenuity.com]).

For in vivo VSMC proliferation analysis, all animals received BrdU before sacrifice (13). At 2 days after injury, carotid arteries were embedded in paraffin and 5-μm cross sections were stained as previously reported (5,13). For the immunofluorescence detection of Epithelial Membrane Protein 2 (EMP2) and Caveolin 1 (CAV1), arterial sections were incubated with a rabbit monoclonal antibody against Cav1 (Cell Signaling Technology) and a rabbit polyclonal antibody against EMP2 (Bioss) and were revealed with appropriate anti-rabbit secondary antibody as per manufacturer instructions (5,13). VSMCs were detected using an antibody against α-smooth muscle actin (1:100; Sigma-Aldrich). The nuclei were counterstained with DAPI (Sigma-Aldrich). Fluorescence in situ hybridization of miR-29c and miR-204 was performed as previously described (13,24). Specific probes for mir-29c and miR-204 were from Qiagen (miRCURY LNA miRNA Detection Probe) and were revealed according to manufacturer instructions. All stainings were acquired and analyzed using confocal microscopy (TCS SP5 and SP8 microscopes; Leica).

Adenovirus vectors encoding an siRNA for Cav1 (Ad-siCav1), miR-29c (Ad-miR-29c), and miR-204 inhibitor (Ad-miR-204-inhibitor) and their respective negative controls (NCs) were purchased from Applied Biological Materials. A bicistronic construct harboring miR-29c-mimic and miR-204-inhibitor was cloned in a standard adenoviral vector (Vector Biolabs). The adenovirus vector was named Adeno-Combo-miR-29c-m/miR-204-i (or Ad-29cM/204I). Adeno-Combo-negative mimic/inhibitor NC (for simplicity Ad-NC) was used as a control. Immediately after vascular injury, the balloon-dilated common carotid arteries were distally clumped to stop perfusion and were infused through the external carotid artery with the specific adenoviral constructs for 30 min (4 × 10¹⁰ plaque-forming units/mL). At the end, adenoviruses were retrieved from the external carotid arteries, the common carotid arteries were washed to ensure complete adenovirus removal, and then the distal clump was removed to allow the recovery of normal carotid perfusion (13). Animals were then sacrificed at 2 and 14 days. In separate experiments, miR-204-mimic and miR-29c-inhibitor were mixed with pluronic gel solution and were locally applied over the injured carotid artery (18).

Human atherosclerotic plaques were derived from patients affected by symptomatic significant carotid artery lesions undergoing surgical carotid endarterectomy. Carotid endarterectomy was performed using standard techniques, with minor variations depending on surgeon and patient. After surgical removal, the specimen containing the plaque lesion was immediately flash frozen in liquid nitrogen and subsequently stored in liquid nitrogen until RNA extraction. Patients with T2DM and without diabetes (ND) were included in the study and did not differ significantly for any clinical parameter other than T2DM (Supplementary Table 1).

Total RNA was extracted with the TRIzol reagent or mirVana miRNA Isolation Kit (Invitrogen) (13). miRNA expression was evaluated using TaqMan miRNA assays (Applied Biosystems). TaqMan Gene Expression Assay (Applied Biosystems) was used for quantification of Emp2, Cav1, Tagln, and Myh1 transcripts (15).

Primary rat aortic smooth muscle cells (SMCs) and human coronary artery SMCs were from Life Technologies. For gain-of-function and loss-of-function experiments, we used mimic and inhibitor RNA molecules purchased from Life Technologies. Cells were transfected using Lipofectamine RNAiMAX (Life Technologies). Specific ACCELL siRNAs against Emp2 (siRNA/Emp2; Dharmacon) were used to inhibit Emp2 mRNA expression in VSMCs. Cells were also transfected with Emp2 Plasmid Vector [lacking Emp2 3′-untranslated region (UTR)] by Lipofectamine 2000 (Invitrogen).

Proliferating VSMCs were identified using Click-it EdU (5-ethynyl-2′-deoxyuridine) kit (Invitrogen) for the detection of EdU incorporation into DNA. For cell cycle analysis, VSMCs were stained with propidium iodide (PI) using an FxCycle PI/RNAase Staining Solution Kit (Life Technologies). Flow cytometric analyses of the cell cycle were performed with FACSAria III (BD Biosciences).

The wild type 3′ UTR of Emp2 was cloned downstream of the Firefly/Renilla Duo-Luciferase reporter vector (GeneCopoeia Inc). HEK293 cells were transfected with Emp2 3′ UTR plasmid alone or cotransfected with miR-29c mimic. Luciferase activity was measured by Luc-Pair Luciferase Assay Kit (GeneCopoeia).

Western blotting and immunoprecipitation were performed by standard protocol as previously reported (5,13,18). The antibodies used were the following: mouse anti-CAV1 (sc-70516; Santa Cruz Biotechnology), goat anti-EMP2 (sc-132670; Santa Cruz Biotechnology), and mouse anti-GAPDH. Experiments were performed in biological quadruplicate, and the data are shown as the mean ± SD.

Statistical analysis was performed with GraphPad Prism version 6.00 for Macintosh (GraphPad Software). Quantitative data are reported as the mean ± SD and binary data by counts. Significance between two groups was determined by Student t test or paired t test, as appropriate. For comparison between multiple groups, ANOVA was used. A P value <0.05 was considered significant. The Bonferroni post hoc method was used to locate the differences. In these cases, the type I error (α = 0.05) was corrected by the number of statistical comparisons performed. For the in vitro cell and molecular biology experiments with an n > 4 sample size, giving the low number of the sample, the Kruskal-Wallis test (for multiple‐group comparison), and the Mann-Whitney U test (for comparison between two groups) were performed.

Balloon injury of the right common carotid artery was performed in T1DM male adult Wistar rats, T2DM ZDF rats (25), and in respective euglycemic controls (Supplementary Fig. 1A). At 2 days after injury, carotid arteries from both T1DM and T2DM animals showed a significantly increased VSMC proliferation rate when compared with controls (Fig. 1A and B). The increased VSMC proliferation rate was followed by a significant increase in neointima formation in injured carotid arteries from T1DM and T2DM rats (Fig. 1C and D).

Thus, both DM models reproduced previous findings (7,8,18) and confirmed that DM exaggerates the proliferative response of VSMCs in the media layer, increasing arterial narrowing at the site of vascular injury.

Total RNA was extracted from injured and uninjured carotid arteries of each group and processed for small RNA-Seq as previously described (14). The bioinformatics analysis demonstrated the differential expression of >300 known miRNAs in each group. A read count cutoff of ≥100 in at least one replicate was established. After this correction, the bioinformatics analysis covered >70 miRNAs for each group (Supplementary Tables 2–4). A heat map of global differentially expressed miRNAs is shown in Fig. 2A. Heat map visualization of miRNA expression values between uninjured and injured vessels from each group is provided in Supplementary Fig. 1B. Different miRNAs already proven to be involved with the proliferative phenotype of VSMCs were accordingly upregulated or downregulated after balloon injury in euglycemic control rats. The latter confirms previous reports (13,26,27) and shows that the approach used can reliably assess the question posed with this experimental design.

The Venn diagram in Fig. 2 summarizes the number and overlap of upregulated (Fig. 2B) and downregulated (Fig. 2C) miRNAs in the injured arteries of the three groups. The analysis revealed significant differences in the dysregulated miRNAs among the experimental groups. We then filtered the list of miRNAs showing a differential expression in at least one group with an arbitrary fold change ≥1.5 and an adjusted P value false discovery rate (FDR) <0.05. Based on these parameters, we observed the abundance of significantly deregulated miRNAs in each group (Fig. 2D–F). Thirty-six miRNAs were identified as being robustly differentially expressed in injured arteries in euglycemic rats, 47 miRNAs in T1DM rats, and 38 miRNAs in T2DM rats.

The Volcano plots representing the dysregulated miRNAs in injured versus uninjured arteries revealed that a few miRNAs were specifically upregulated or downregulated in both T1DM and T2DM rats but not in euglycemic rats (Fig. 2G–I). In particular, miR-195, miR-29c, miR-140-3p, and miR-199a-5p were downregulated in both T1DM and T2DM, while miR-25 and miR-204 were, on the other hand, specifically upregulated in both DM types. Thus, these six miRNAs (Diab-miRs), potentially constitute a diabetic signature underlying/regulating the proliferative response of VSMCs in DM.

Interestingly, the fold changes of all selected miRNAs as determined by quantitative RT-PCR (qRT-PCR) strongly correlated with the values from the RNA-Seq (Fig. 3A).

Among the four downregulated Diab-miRs, mir-29c downregulation by a miR-29c-inhibitor shows the more pronounced increase in VSMC proliferation in low-serum conditions. Concurrently, miR-204 upregulation increased VSMC proliferation (Fig. 3B).

We then compared expression levels of the six identified miRNAs in human atherosclerotic plaque specimens from age- and sex-matched patients with diabetes and nonpatients with diabetes undergoing surgical carotid endarterectomy for symptomatic carotid artery stenosis. We found that miR-195, miR-29c, and miR-199a-5p were all downregulated while miR-204 was upregulated in diabetic versus ND samples (Fig. 3C).

Among the dysregulated miRNAs, considering their functional preliminary evaluation in vitro and their expression profile in human vascular tissues, mir-29c and miR-204 appeared to have the highest potential to specifically regulate diabetes-dependent VSMC hyperplastic phenotype. The latter is further reinforced by the absence of miR-29c and miR-204 dysregulation in peripheral blood monocytes and T cells from T1DM, T2DM, and euglycemic control rats 2 days after injury (Supplementary Fig. 2A–C). This further suggests that diabetes-induced dysregulation of miR-29c and miR-204 primarily involves VSMCs without targeting other cell types known to contribute to pathological arterial narrowing.

Rat aortic VSMCs were treated with high glucose (HG; 25 mmol/L) and high insulin (HI; 200 nmol/L) as DM conditions in vitro. Control cells were cultured in normal (low) glucose (5 mmol/L) and insulin (2 nmol/L). miR-29c was downregulated, while miR-204 was upregulated in HI/HG conditions (Supplementary Fig. 3A). Similar effects were also detected in human coronary artery SMCs (Supplementary Fig. 3B).

A standard model of the VSMC phenotypic switch in vitro involves the treatment of VSMCs with platelet-derived growth factor (PDGF)-BB (28,29). VSMCs were transfected with miR-29c and miR-204 mimics or inhibitors to test whether PDGF-BB–dependent VSMC proliferation in HG conditions is modulated by these miRNAs. miR-29c overexpression or miR-204 inhibition in PDGF-BB+HG–treated VSMCs caused a significant decrease in proliferating cells when compared with the relative specific controls, whereas the opposite effect was obtained with the inhibition of miR-29c or miR-204 overexpression, respectively (Fig. 4A).

The VSMC phenotypic switch commences with cell dedifferentiation through the downregulation of VSMC contractile genes (13). We then analyzed the effects of miR-29c and miR-204 gain-of-function or loss-of-function on the expression of Transgelin (Tagln) and Myosin Heavy Chain 11 (Myh11) in PDGF-BB+HG–treated VSMCs. These two genes are indeed specifically expressed by contractile VSMCs, and their downregulation is classically related to the phenotypic switch of VSMCs underlying vasculoproliferative disorders (13,15). miR-29c overexpression partially inhibited while miR-29c inhibition exacerbated PDGF-BB–induced Tagln and Myh11 downregulation in HG conditions (Supplementary Fig. 3C). A significant decrease in Tagln and Myh11 was also observed in the VSMCs transfected with miR-204 mimic compared with the control group (Supplementary Fig. 3D). The opposite effect was elicited by miR-204 downregulation.

Furthermore, miR-29c-mimic or miR-204 inhibitor reduced HI/HG-stimulated VSMCs in the S phase of the cell cycle compared with cells transfected with relative controls (Fig. 4B). On the contrary, miR-204 overexpression or miR-29c inhibition increased the number of VSMCs in S phase (Fig. 4B).

miR-29c overexpression or miR-204 inhibition causes a significant decrease in EdU⁺ proliferating VSMCs in response to HI/HG conditions (Fig. 4C). Accordingly, miR-29c inhibition or miR-204 mimic fostered a significant increase in EdU incorporation in HI+HG-stimulated VSMCs. Notably, concomitant miR-29c overexpression and miR-204 inhibition had an additive effect when compared with the single miRNA modulation in reducing HI/HG-induced VSMC proliferation (Fig. 4D). Finally, simultaneous miR-29c upregulation and miR-204 downregulation increased the cell cycle inhibition of PDGF-BB+HG–stimulated VSMCs when compared with unstimulated VSMCs and VSMCs treated with either miR-29c-mimic or miR-204 inhibitor (Fig. 4E).

Thus, these data collectively show that miR-29c and miR-204 additively modulate the hyperproliferative phenotype of VSMCs in diabetic conditions in vitro.

Whole transcriptome analysis was performed on the same RNA extracted for miRNome analysis from uninjured and injured carotid arteries of T1DM, T2DM, and euglycemic rats 2 days after injury (23). Several mRNAs were differentially expressed (read count cutoff >10 in at least one replicate) using the normalized reads showing a >1.5-fold change in injured carotid arteries compared with the uninjured control group. We found 2,210 upregulated and 2,469 downregulated genes in injured versus uninjured carotid arteries from T1DM rats and 1,494 upregulated and 2,246 downregulated genes in injured versus uninjured carotid arteries from T2DM rats. Functional analysis of differentially expressed mRNAs through IPA indicated that several pathways related to VSMC phenotypic switch were activated in both DM models (Fig. 5A). Disease and function analyses identified by IPA also indicated that the highest fraction of dysregulated genes in both T1DM and T2DM cluster in gene ontologies involved with cell proliferation, RNA processing, cell apoptosis, cell migration, and metabolism (Fig. 5B).

Dysregulated mRNAs in injured versus uninjured carotid arteries were searched to identify all of the putative miR-29c and miR-204 mRNA targets whose changes were specular to the relative miRNA modulation. miRanda analysis shows that of the 477 mRNA targets of miR-29c, 88 were upregulated in T1DM rats and 58 in T2DM rats. Concurrently, of the 597 putative mRNA targets of miR-204, 123 were downregulated in T1DM rats and 108 in T2DM rats. Then, we restricted the analysis on dysregulated genes that had a fold change of less than or equal to −1.5 or ≥1.5 (|1.5|) and that were present in both T1DM and T2DM rats but not in the euglycemic controls or that were significantly more upregulated or downregulated in both T1DM and T2DM rats versus euglycemic controls. The heat map of differentially expressed mRNAs in each group is shown in Fig. 5C and D.

Of note, among the small set of putative miR-29c and miR-204 targets, IPA analysis revealed enrichment in specific signaling and metabolic pathways involved in VSMC phenotypic switch and vascular disorders (Supplementary Fig. 4A and B). The miRNA-mRNA target deregulated in both DM models included molecules implicated in cell cycle regulation, cell death, morphology, and migration (Supplementary Fig. 5A and B).

Overall, the bioinformatics analysis suggested that miR-29c and miR-204 coherently target relevant mRNAs underlying VSMC hyperplastic phenotype.

Putative targets of miR-29c were further screened for their role on various biological processes potentially involved in the VSMC phenotypic switch regulation. We identified Dclk1 (Doublecortin-Like Kinase 1), Tnfrsf12a (TNF Receptor Superfamily Member 12A), and Emp2 as being among the top nine candidate transcripts identified. RT-PCR analysis revealed no significant differences in Dclk1 and Tnfrsf12a expression in miR-29c-mimic–transfected or miR29c-inhibitor–transfected VSMCs (Supplementary Fig. 6A). Importantly, miR-29c-mimic significantly inhibited the expression of Emp2 mRNA, whereas the opposite effect was obtained with the miR-29c-inhibitor (Fig. 6A). Since Emp2 is a key element for different cell functions such as proliferation, migration, and cell adhesion (30,31), we next aimed to assess whether Emp2 plays a role in the miR-29c regulation of VSMCs proliferation.

Consistent with the in vivo data, we first observed that EMP2 protein expression was increased in response to HI/HG conditions compared with VSMCs cultured in normal glucose and insulin (Fig. 6B). EMP2 protein decreased in VSMCs transfected with miR-29c-mimic and its specular increase in miR-29c-inhibitor–treated VSMCs (Fig. 6B). Ago proteins-RNA immunoprecipitation assay revealed that Emp2 mRNA was significantly enriched in the Ago-immunoprecipitated RNAs from SMCs transfected with miR-29c-mimic compared with those treated with mimic NC (Supplementary Fig. 6B). On the contrary, Emp2 mRNA significantly decreased in the Ago-immunoprecipitated RNAs from miR-29c-inhibitor–treated SMCs compared with cells transfected with inhibitor NC (Supplementary Fig. 6B). Accordingly, miR-29c-mimic reduced luciferase activity of a Emp2 3′ UTR cloned into a luciferase reporter construct transfected in HEK293 cells, showing that this miRNA directly targets Emp2 3′ UTR (Fig. 6C).

Finally, siRNA/Emp2 significantly decreased VSMC proliferation in response to HI/HG conditions in vitro (Fig. 6D and Supplementary Fig. 6C). EdU incorporation assay was also performed in VSMCs cotransfected with plasmids encoding for Emp2 (pEMP2) and miR-29c-mimic. pEmp2 overexpression increased proliferating VSMCs despite miR-29c upregulation when compared with cells cotransfected with a NC plasmid and miR-29c-mimic (Fig. 6E).

Hence, overall these data show that miR-29c directly represses Emp2, which is a main target modulating miR-29c regulation of VSMC proliferation.

We searched cellular and molecular functions of all putative miR-204 targets identified by RNA-Seq and in silico analysis. However, we did not identify any targets potentially involved with the hyperproliferative VSMC phenotype in the annotated mRNAs in injured arteries.

We thus evaluated the expression of CAV1 protein, a known target of miR-204 for its potential role in modulating VSMC proliferation (32,33). Accordingly, Emp2 was previously shown to negatively regulate Cav1 expression (34,35). Importantly, miR-204 directly affects CAV1 protein, but shows very limited effects on its mRNA expression (36). First, Ago-RNA immunoprecipitation assays show that Cav1 mRNA was significantly enriched in the Ago-immunoprecipitated RNAs from the miR-204–overexpressing SMCs compared with the relative control, confirming that Cav1 is a direct target of miR-204 (Supplementary Fig. 6D). Second, Cav1 protein was significantly decreased in HI+HG–treated versus control VSMCs (Supplementary Fig. 6E). Importantly, miR-204 overexpression significantly inhibited CAV1 protein expression, whereas the opposite effect was obtained through functional inhibition of miR-204 (Supplementary Fig. 6E).

Furthermore, CAV1 protein expression was reduced by Emp2 plasmid overexpression (Supplementary Fig. 6F), whereas siRNA-Emp2 significantly increased CAV1 protein levels in VSMCs (Supplementary Fig. 6G). Finally, we assessed whether CAV1 protein levels could be modulated by concomitant miR-29c overexpression and miR-204 inhibition. VSMCs transfected with miR-29c-mimic or miR-204-inhibitor showed significantly increased CAV1 levels when compared with relative control cells (Fig. 6F). Interestingly, CAV1 protein expression was further induced by the combination of miR-204 downregulation and miR-29c upregulation (Fig. 6F).

These findings show that miR-204 fine-tunes CAV1 protein expression. This regulation links miR-204 and miR-29c activities, as they both converge to target Cav1 in the molecular network activating VSMC proliferation.

We assessed arterial stenosis formation (neointimal formation) in balloon-injured arteries from T1DM and T2DM rats infected with Ad-miR-29c or Ad-miR-204-inhibitor. Balloon-injured vessels showed reduced miR-204 levels 48 h after infection with Ad-miR-204-inhibitor, whereas Ad-miR-29c infection increased vascular miR-29c levels (Supplementary Fig. 7A). Concurrently, Ad-miR-29c significantly inhibited vascular Emp2 mRNA expression (Supplementary Fig. 7B), whereas CAV1 protein levels were upregulated by Ad-miR-204-inhibitor (Supplementary Fig. 7C). Neointimal formation (measured as Neointima-media areas ratio) in injured vessels infected with Ad-miR-29c or Ad-miR-204-inhibitor were both decreased at 14 days compared with the control Ad-NC–treated injured vessels in T1DM and T2DM rats (Supplementary Fig. 7D). However, these two miRNA modulators separately did not decrease neointima formation beyond the vessel response to injury in euglycemic rats (Supplementary Fig. 7D).

We therefore designed a bicistronic construct to contemporarily overexpress miR-29c and antagonize miR-204. This bicistronic construct was cloned in an adenoviral vector for in vivo administration. The Adeno-Combo-miR-29c-m/miR204-i (or Ad-29cM/204I) was preliminarily tested in vitro, showing its effectiveness in inhibiting VSMC proliferation (Supplementary Fig. 7E). Ad-29cM/204I was locally released in injured carotid arteries from T1DM and T2DM rats where it increased vascular miR-29c levels and reduced miR-204 levels 2 days after balloon injury both in T1DM and T2DM rats when compared with the respective controls (Supplementary Fig. 8A). Fluorescence in situ hybridization showed that at 2 days after injury, miR-29c was downregulated, whereas EMP2 was upregulated specifically in VSMCs of injured carotid arteries from T2DM rats when compared with uninjured controls (Fig. 7A). miR-204 in situ levels were barely detectable in VSMCs of the media layer of uninjured arteries; however, miR-204 levels significantly increased in the VSMCs of balloon-injured carotid arteries, whereas CAV1 showed a specular decrease (Fig. 7A). Finally, Ad-29cM/204I infection respectively increased miR-29c while it decreased miR-204 levels specifically in VSMCs of balloon-injured carotid arteries (Fig. 7A). On the contrary, Ad-29cM/204I transduction reduced VSMC EMP2 levels while it increased CAV1 (Fig. 7A). Western blot analysis confirmed that vascular EMP2 protein levels were downregulated while CAV1 was upregulated in injured carotid arteries from T1DM and T2DM rats treated with Ad-29cM/204I compared with relative controls (Fig. 7B). Fourteen days after injury, miR-29c and miR-204 levels were, respectively, persistently upregulated and downregulated in Ad-29cM/204I–treated vessels when compared with injured controls (Supplementary Fig. 8B). Specular yet consistent data were obtained evaluating EMP2 and CAV1 levels in the injured carotid arteries at 14 days (Supplementary Fig. 8C).

Ad-29cM/204I local vascular release almost abolished neointimal formation (neointima/media areas ratio [N/M]: T1DM rats 0.46 ± 0.07; T2DM rats 0.56 ± 0.09) 14 days after injury when compared with Ad-NC–treated diabetic rats (N/M: T1DM 1.81 ± 0.192; T2DM 1.73 ± 0.168) (Fig. 7C and D). Furthermore, Ad-29cM/204I in T1DM and T2DM significantly reduced neointimal formation even when compared with Ad-NC–treated euglycemic controls (N/M 1.21 ± 0.2) (Fig. 7C and D). On the contrary, contemporary miR-29c-inhibitor and miR-204-mimic cotransfection in injured vessels of euglycemic rats increased neointima formation when compared with control-transfected vessels (Supplementary Fig. 8D).

Finally, in order to further explore the role of the CAV1 modulation in the injured diabetic artery, we first transduced VSMCs with a recombinant Ad-siCav1 infection in vitro. Ad-siCav1 reduced CAV1 protein levels despite contemporary miR-29c upregulation/miR-204 inhibition when compared with VSMCs transfected with appropriate controls (Supplementary Fig. 8E). Ad-29cM/204I and Ad-siCav1 cotransduction fostered a significant increase of EdU incorporation in VSMCs compared with the cells infected with Ad-29cM/204I (Supplementary Fig. 9A). We eventually assessed the effect of Ad-siCav1 and Ad-29cM/204I coinfection on neointimal formation in injured arteries from T1DM and T2DM rats. As shown in Supplementary Fig. 9B, concomitant Ad-29cM/204I and Ad-siCav1 infection reduced CAV1 levels 48 h after balloon injury both in T1DM and T2DM carotid arteries when compared with the injured artery infected with Ad-29cM/204I. Importantly, Cav1 inhibition by Ad-siCav1 increased neointima formation of both T1DM and T2DM carotid arteries transfected with Ad-29cM/204I when compared with injured arteries infected with Ad-29cM/204I + Ad-siNC (Supplementary Fig. 9C).

Therefore, miR-29c and miR-204 are central Diab-miRs forming with their targets, EMP2 and CAV1, a molecular network that has a critical role in regulating the exaggerated detrimental VSMC phenotypic switch in DM.

The main findings emanating from this study are as follows: 1) the diabetic detrimental hyperplastic phenotype of VSMCs is characterized by an miRNA signature that diverges significantly from the miRNA network activated in ND VSMCs; 2) this miRNA signature is integrated in an miRNA-mRNA network that regulates gene cascades involved in various aspects of VSMC phenotypic switch upon vascular injury; and 3) miR-29c and miR-204 comodulation, and their molecular targets Emp2 and CAV1, are necessary in regulating diabetic VSMC proliferation in vitro and in vivo.

miRNAs are key controllers of several pathophysiological conditions and clinically specific biomarkers for various diseases (37). Patients with diabetes display a significant deregulation of miRNAs involved in angiogenesis, vascular repair, and endothelial homeostasis (38). VSMC phenotypic switch is a key cellular event in vasculoproliferative disorders (39,40), and VSMCs display a different and more aggressive phenotype in patients with diabetes (9–12). miRNAs play key roles in VSMC functions, but their role under diabetic conditions is unclear. Recently, small RNA-Seq identified several differentially expressed miRNAs in VSMCs from T2DM db/db mice versus ND db/+ mice (17). However, the study was not designed to specifically address VSMC phenotypic switch, and it did not assess miRNA expression upon vascular injury. Here, assessing the whole miRNome in diabetic injured versus uninjured vascular tissues, we identified a group of six miRNAs consistently showing robust deregulation in diabetic conditions. These Diab-miRs were similarly deregulated in human diabetic atherosclerotic plaques and coronary SMCs. miR-29c and miR-204 were studied in detail, because of their more pronounced dysregulation and robust effects on VSMC proliferation in vitro.

miR-29c is part of the miR-29 gene family, which has been extensively investigated in many types of cancers (41,42). Also, miR-29 is involved in the pathogenesis of kidney disease (43), where renoprotective drugs act through an increase of miR-29 expression (44). Importantly, individual members of the miR-29 gene family also function as regulators of vascular cells. In particular, miR-29b promotes VSMC differentiation from embryonic stem cells by targeting YY1 (Yin Yang 1), a transcription factor that inhibits muscle cell differentiation and muscle-specific gene expression (45). In addition, other examples of the roles of the miR-29 gene family in mediating phenotypic modulation of vascular cells have been reported (46,47). miR-204 plays a significant role in the development of eyes (48) and adipogenesis (49). It is believed that miR-204 acts mainly as a tumor suppressor; however, it has been shown that its function can also evolve to that of an oncomiR in certain instances (50). Of note, miR-204 overexpression plays a critical role in the development of diabetes (51).

Here we show that miR-29c upregulation and miR-204 downregulation inhibit VSMC proliferation in DM in vitro and in vivo. Their effect is additive because their simultaneous modulation to increase miR-29c and inhibit 204 abolishes the detrimental and exaggerated hyperplastic response of VSMCs in DM after vascular injury. miR-29c directly targets for repression Emp2 while miR-204 targets Cav1. A single miRNA has multiple mRNA targets, and often these multiple targets converge to regulate the same molecular and cellular process (52), whereas multiple miRNAs can converge on the same pathway, fine-tuning its activity via one or more mRNAs (53). The latter appears to be functioning in the combined effects of miR-29c and miR-204. miR-29c targets for Emp2 repression, which in turn regulates Cav1, which is a target of miR-204. In particular, miR-204 fine-tunes Cav1 protein expression without degrading its mRNA. The combined gene repression and fine-tuning effects of these miRNAs are necessary to promote the detrimental VSMC diabetic phenotype. Thus, we discovered an miRNA-mRNA-protein molecular network that is an essential to prevent vasculoproliferative response in DM in vivo.

Treatment of atherosclerotic lesions in coronary and peripheral artery diseases is still a challenge in patients with DM. It highly tempting to speculate that miR-29c and miR-204 comodulation may represent a more tailored and specific therapy for a local release approach to prevent vascular sequels in the treatment of vasculoproliferative diseases in diabetes.

Funding. This work was supported by grants PRIN2015 2015ZTT5KB_004 and FIRB-Futuro-in-Ricerca (RBFR12I3KA) Flagship Project InterOmics from the Italian Ministry of Education, University and Research; Finalized Research2010 (GR-2010-2318945) from the Italian Ministry of Health; and Investigator Grant (IG-17426) from Italian Association for Cancer Research.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. D.T. designed the research, supervised the experiments, and wrote the manuscript. C.Ia., F.M., C.Ve., I.A., M.S., T.M., E.C., C.Va., P.M., L.T., and C.Vi. conducted the experiments and analyzed the data. R.T., G.G., and A.W. conducted the RNA-Seq experiments. F.C.S., D.C., and M.T. contributed reagents. C.In. contributed to discussion and reviewed the manuscript. D.T. and C.In. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.