Involvement of miR-30a-5p and miR-30d in Endothelial to Mesenchymal Transition and Early Osteogenic Commitment under Inflammatory Stress in HUVEC

The endothelial to mesenchymal transition (End–MT) can be associated with vascular calcification, by providing mesengenic progenitors. In this study, we investigated a link between End–MT and the osteogenic process and explored the involvement of miR-30a-5p and miR-30d as potential regulators of these processes. End–MT was induced in Human Umbilical Vein Endothelial Cells (HUVEC) through transforming growth factor-β1 (TGF-β1), TGFβ-3 and tumor necrosis factor-α (TNF-α), for 24 h and 6 days. End–MT mediators, mesenchymal and osteo/chondrogenic markers were analyzed through Real-Time PCR, immunofluorescence, flow cytometry and Western Blot. miR-30a-5p and miR-30d over-expression was carried out in HUVEC to explore their effects on End–MT and osteogenic differentiation. HUVEC at 24 h and 6 days gained mesenchymal morphology markers, including matrix metalloproteinase 9 (MMP-9), SLUG, VIMENTIN and α-smooth muscle actin (α-SMA), and a significant migratory potential, notably with TNF-α. After 6 days, the osteo/chondrogenic markers runt-related transcription factor 2 (RUNX-2) and SRY box transcription factor 9 (SOX-9) were upregulated. At this time point, miR-30a-5p and miR-30d decreased. Over-expression of miR-30a-5p and miR-30d affected End–MT mediators and the osteogenic potency in HUVEC, by reducing SLUG, VIMENTIN and RUNX-2. Our data suggest that End–MT represents a key link between inflammation and vascular calcification. Further, miR-30a-5p and miR-30d can regulate both the End–MT and the osteogenic processes, prompting future studies for exploring their potential use as therapeutic targets or biomarkers in vascular diseases.


Introduction
Endothelial cells (ECs) constitute the inner cell layer within the blood vessels and are primarily involved in vascular homeostasis, regulating blood flow, hemostasis, blood cells' luminal adherence and vascular permeability [1]. The endothelial dysfunction is considered the triggering event in the initiation of atherosclerotic disease [2], and it is associated with the loss of the EC barrier, allowing the recruitment of monocytes and the tic approaches require an accurate evaluation of miRNA signaling mechanisms; indeed, miRNAs are broadly expressed among tissues and their manipulation can lead to side effects [19]. For this reason, unveiling the biology of miRNAs and their role in disease models would be crucial to their utilization for diagnostic and therapeutic purposes in the field of cardiovascular diseases.
In order to investigate the link between End-MT and plaque calcification, we firstly established an in vitro model of End-MT driven by inflammatory soluble factors: in addition to the well-known TGF-β1, we included TGF-β3, belonging to the same family, and Tumor Necrosis Factor-α (TNF-α), as an inflammatory cytokine mimicking the atherosclerotic milieu. Then, we explored the miRNA biology during End-MT and osteogenic differentiation, by inducing the ectopic expression of miR-30a-5p and miR-30d in this cell model.

Cell Viability Assay
The effect of miRNA transfection on cell viability was determined by 3-(4,5dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Vybrant MTT Cell Proliferation Assay Kit, Thermo Fisher Scientific, Waltham, MA, USA) following the manufacturer's instructions. To this aim, cells were seeded in 96-well plates at 10 4 cells/well in 100 µL growth medium. The MTT assay was performed on HUVECs after 24 h and 6 days exposure to TGF-β1, TGF-β3 and TNF-α, and after 24 and 48 h from miRNA over-expression. Briefly, cell medium was replaced with fresh growth medium after treatment, and 10 µL of 12 mM MTT component A (3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide) were added to each well and left in an incubator for 4 h. Then, 100 µL of MTT component B sodium dodecyl sulfate (SDS)-hydrochloride acid (HCl) were added for a further 18 h at 37 • C. Absorbance was read at optical density (O.D.) 570 nm by a micro-plate reader.

In Vitro Migration Assay
In order to evaluate the migration property of HUVECs following End-MT induction and miRNA transfection, we performed a scratch assay. Briefly, 1 × 10 5 cells were seeded in a 24-well plate. For the End-MT study, the cell monolayer was wounded with a sterile p200 pipette tip, washed with PBS, treated with TGF-β1, TGF-β3 and TNF-α (10 ng/mL) in serum-free DMEM and incubated for additional 24-48-72 h. Cells were fixed with formalin at room temperature (rt), washed with Phosphate Buffer Saline (PBS), stained with 0.1% Crystal Violet in 25% methanol for 25 min, and air-dried. The migration property was analyzed under the phase-contrast light microscope and images were taken with a digital camera (Nikon, Tokyo, Japan). The quantification of the wounded area was performed on three fields of each treatment group and experimental time point. The wounded area was measured using the image analysis software ImageJ [20], and expressed as percentage relative to the control group at time 0.

Osteogenic Differentiation Assay
In order to test the effects of miRNA overexpression on the osteogenic differentiation potential, HUVECs were seeded at a density of 4 × 10 4 cells/well on 24-well plates and left to adhere. After miRNA transfection, Optimem medium was changed with the Stem Pro Osteogenic Differentiation Kit (Thermo Fisher Scientific, Waltham, MA, USA). The miRNA transfection was repeated every three days to keep miRNA levels in cell culture. HUVECs cultured in DMEM with 10% FBS were used as controls. After 14 days, HUVECs were processed for Alizarin Red staining to detect the mineralization process, immunofluorescence, RNA and protein extraction to investigate osteogenic marker expression. For mineralization activity, cells were fixed with formalin, washed twice with PBS and stained with Alizarin Red for 30 min at rt. Images were taken under a phasecontrast light microscope and pictures were taken with a digital camera (Nikon, Tokyo, Japan). Then, Alizarin Red was dissolved by adding 10% cetylpiridinium chloride (in Na 2 HPO 4 10 mM, pH 7) to wells for dissolving Alizarin Red, for 15 min in the dark at rt. Calcium-bound Alizarin Red was measured by reading absorbance O.D. 570 nm by spectrophotometer.

RNA Extraction and Quantitative Real-Time Analysis
Total RNA was extracted from cell cultures through TRIreagent (Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer's instructions. Reverse Transcription was performed from one µg of total RNA in 20 µL reaction volume using the High-Capacity Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, MA, USA). Real-Time PCR was carried out in a CFX Connect Real-Time PCR Detection System (BioRad Laboratories, Hercules, CA, USA) using the SYBR green mix (BioRad Laboratories) and specific couples of primers were designed using the Basic Local Alignment Tool (BLAST) from National Center for Biotechnology Information (NCBI) (purchased from Sigma-Aldrich, St Louis, MO, USA; Table 1). Each assay was performed in triplicate and target gene expression was normalized to glyceraldehyde 3-phospate dehydrogenase (GAPDH). Final results were determined by the comparative 2ˆ− ∆∆Ct method and expressed as fold changes relative to untreated controls, and to control without lipofectamine in transfection experiments.

MicroRNA Expression Analysis
For the analysis of miRNA expression levels during End-MT, osteogenic differentiation and for the validation of miR over-expression, cDNA reverse-transcription was performed with miRNA-specific primers in the same reaction using TaqMan microRNA Assays and TaqMan microRNA Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, MA, USA), following the manufacturer's instructions. miRNA expression was performed through Real Time PCR, using TaqMan MicroRNA Assay-specific probes for each target and TaqMan Universal Master Mix II No AmpErase UNG. The reaction was carried out in CFX Connect Real-Time PCR (BioRad Laboratories, CA, USA). Each assay was executed in triplicate and target miRNA expression was normalized to non-coding small nuclear (snRNA) U6 gene. Final results were determined by the comparative 2ˆ− ∆∆Ct method and expressed as fold changes relative to controls.
For the luciferase assay, HUVECs were plated at a density of 1 × 10 5 in 96-well plates and, after 24 h, co-transfected with pmiR-GLO SLUG 3 -UTR and 10 pmol of miR-30a-5p mimics, or miR-30d mimic, or miR-negative control. Transfection was assessed with Lipofectamine RNAiMAX (Thermo Fisher Scientific, Waltham, MA, USA). After 24 h, the Firefly and Renilla luciferase activity were measured though the Dual-Glo Luciferase Assay System (Promega Corporation, Madison, WI, USA), according to the manufacturer's instructions.

Immunofluorescence
For immunofluorescence, cells were fixed with cold absolute methanol for 10 min at rt, or 2% paraformaldehyde for 4 min at rt for the analysis of filamentous (F)-actin. Then, cells were permeabilized with Triton X-100 1% in PBS for an additional 10 min at rt. The blocking of non-specific binding sites was performed with bovine serum albumin (BSA) 1% in PBS for 30 min at rt. After blocking, cells were incubated with primary antibodies. For staining of F-actin, Alexa fluor 488 Phalloidin was added (1:500, Thermo Fisher Scientific, Waltham, MA, USA) for 20 min at rt, washed with PBS and counterstained with Pro Long anti-fade reagent with 4 -6-diamidino-2-phenylindole (DAPI) (Thermo Fisher Scientific, Waltham, MA, USA). For VIMENTIN (1:500; Cell Signaling), SLUG (1:100, A7, Santa Cruz Biotechnology, Dallas, TX, USA), RUNX-2 (1:100; NBP1-77461SS, Novus Biologicals, Littleton, CO, USA), SOX-9 (1:500, Abcam, Cambridge, UK) and α-SMA (1:100, Sigma-Aldrich, St Louis, MO, USA) detection, the primary antibody was incubated for 1 h at 37 • C. Samples were then washed with PBS and incubated with anti-mouse Alexa Fluor 488 and anti-rabbit Alexa Fluor 546 (Thermo Fisher Scientific, Waltham, MA, USA) secondary antibodies in 1% BSA/PBS for 1 h at 37 • C in the dark. After washing with PBS, nuclei were counterstained with DAPI. Images were acquired by a Leica DMI4000 B inverted fluorescence microscope (Leica Microsystems, Wetzlar, Germany). Quantification of SLUG, RUNX-2 and SOX-9 positive cells was performed on digitalized images randomly acquired at 40× magnification, and a minimum of 5 fields were examined for each sample. Results were expressed as percentage of nuclei positive to the target protein and expressed as percentage of positive cells/total cells.

Immunohistochemistry
SLUG detection by immunohistochemistry was performed on pre-existing tissue blocks obtained from archival tissues of healthy multiorgan donors and patients affected by atherosclerosis aneurysm (APP-13-01, Di.Ce. 3868-2015; St. Orsola-Malpighi Ethic Committee). All samples were anonymous and treated according to the ethical guidelines of the 1975 Declaration of Helsinki and following revisions. Immunohistochemistry was assessed using a non-biotin-amplified method (Novolink, Leica Biosystems, Wetzlar, Germany). Briefly, sections (4 µm thick) of formalin-fixed and paraffin-embedded tissues were deparaffinized and rehydrated through a series of graded ethanol and rinsed in distilled water. Endogenous peroxidase activity was blocked in 3% H 2 O 2 in absolute methanol for 10 min at rt, antigen retrieval was performed using citrate buffer (pH 6) in autoclave (120 • C) for 20 min and, after cooling, slides were washed with Tris Buffered Saline (TBS). Sections were subsequently incubated with SLUG primary antibody (1:500; Santa Cruz Biotechnology, Dallas, TX, USA) in a moist chamber at 4 • C o/n, incubated with NovoLink Polymer for 30 min at rt and then exposed to the substrate/chromogen 3,3diaminobenzidine (DAB) prepared from Novocastra DAB Chromogen and NovoLink DAB buffer. Nuclei were counterstained with Mayer's hematoxylin. Samples were dehydrated, cover slipped and observed under a light microscope using the Image Pro Plus program. Quantification of SLUG-positive cells was performed on digitalized images randomly acquired at 25× magnification, and a minimum of 5 histological sections was examined for each sample, by using the Image Pro Plus measurement tool. Results were expressed as mean of SLUG-positive areas.

Statistical Analysis
Each experiment was executed at least in triplicate, and all data were expressed as mean ± standard deviation (SD). The graphs and the statistical analyses were performed by GraphPad Prism 6 statistical software. The significance of differences between two experimental conditions was evaluated using the unpaired Student's t test, whereas ordinary one-way analysis of variance (ANOVAs) followed by Bonferroni and Dunnett's test was applied for multiple comparisons. Results were considered statistically significant at the 95% confidence level (p < 0.05).

TGF-β1, TGF-β3 and TNF-α Induce End-MT in HUVECs
In order to study the End-MT occurring under pathological conditions in endothelial cells, we explored the effects of different soluble factors reproducing the atherosclerotic milieu in HUVECs for 24 h and 6 days. In addition to TGF-β1, the canonical inducer of End-MT, we tested TGF-β3 that is associated with the mesenchymal osteo-chondrogenic commitment, and the inflammatory cytokine TNF-α. The analysis of cell viability after these experimental conditions was measured through MTT and did not reveal cytotoxic effects both at 24 h and 6 days (Supplementary Figure S1A). After 24 h, HUVECs started an early morphological transition process from the typical cobblestone shape to a fibroblastlike morphology, mostly detectable in the TNF-α group ( Figure 1A). The morphological change was reflected on the cytoskeletal rearrangements, as shown by phalloidin staining in Figure 1B. Here, the intensity stain of F-actin was broadly conserved among the treatment conditions, however the acquisition of cytoplasmic extension typical of mesenchymal shape with cytoskeletal rearrangements was highlighted, especially with TGF-β1 and TNF-α ( Figure 1B; Supplementary Figure S1B). In parallel, we analyzed the expression of the endothelial gene VE-cadherin that was reduced only in cells treated with TGF-β1 (0.28 ± 0.07, p = 0.0003) and TGF-β3 (0.63 ± 0.3) ( Figure 1C). A marked upregulation of MMP-9 mRNA ( Figure 1C), associated with cell migration and matrix remodeling, and snail family transcriptional repressor 2 (SLUG, or SNAI2) mRNA ( Figure 1B), was detected in HUVECs exposed to TGF-β1 (MMP-9 2.25 ± 0.21; SLUG 2.46 ± 0.07, p = 0.0005), TGF-β3 (MMP-9 4.14 ± 1.15, p = 0.048; SLUG 3.03 ± 0.6, p = 0.0137) and, particularly, TNF-α (MMP-9 9.98 ± 3.7, with p = 0.019; SLUG 8.8 ± 0.94, p = 0.0002). SLUG increase was further demonstrated by immunofluorescence, which revealed higher nuclear staining in all experimental conditions, especially in the presence of TNF-α ( Figure 1D,E). SLUG, is a well-known marker of End-MT, broadly expressed in mesenchymal stromal cells, like VIMENTIN, whose upregulation was detected by immunofluorescence ( Figure 2A known marker of End-MT, broadly expressed in mesenchymal stromal cells, like VI-MENTIN, whose upregulation was detected by immunofluorescence ( Figure 2A   known marker of End-MT, broadly expressed in mesenchymal stromal cells, like VI-MENTIN, whose upregulation was detected by immunofluorescence ( Figure 2A) and Western blot ( Figure 2B,C).  All data are expressed as mean ± SD of 3 independent experiments, and statistical analysis was performed by unpaired Student's t-test, ** p < 0.005.

Persistent Inflammatory Stimulation Induces Mesenchymal Features in HUVECs
The End-MT process was mature at longer exposure, as shown by HUVEC morphology ( Figure 3A) and by immunofluorescence of VIMENTIN and α-SMA protein, emphasizing the cytoplasmic elongations typical of mesenchymal cells, especially in HUVECs exposed to TNF-α ( Figure 3B,C).
Biomolecules 2021, 11, x FOR PEER REVIEW 9 of 20 VIMENTIN detection in whole cell lysates assessed by Western Blot and (C) relative quantification to β-actin, through densitometric analysis of band intensities performed by Image J software. Results are normalized to the untreated controls. All data are expressed as mean ± SD of 3 independent experiments, and statistical analysis was performed by unpaired Student's t-test, ** p < 0.005.

Persistent Inflammatory Stimulation Induces Mesenchymal Features in HUVECs
The End-MT process was mature at longer exposure, as shown by HUVEC morphology ( Figure 3A) and by immunofluorescence of VIMENTIN and α-SMA protein, emphasizing the cytoplasmic elongations typical of mesenchymal cells, especially in HUVECs exposed to TNF-α ( Figure 3B,C). Accordingly, we analyzed the immunophenotype of HUVECs after 6 days, confirming the significant decrease of the endothelial markers CD31 and CD146 expression under End-MT stimulation ( Figure 4A). Here, we also observed a significant increase of the mesenchymal stem cell (MSC) marker CD105, whereas CD44 was comparable among the different conditions ( Figure 4A). Results are summarized in the graphs and table in Figure  4B. The gain of a mesenchymal-like phenotype was further supported by the migratory ability, as shown by the scratch assay ( Figure 4C,D). HUVECs were scratched after being exposed to TGF-β1, TGF-β3 and TNF-α, and acquired a pronounced migration to the wounded area, notably in the TNF-α condition, where the wound was completely closed at 72 h. These data confirm the phenotype plasticity of HUVECs under specific stimulation, indicating that the inflammatory response first promotes the End-MT and, later, the acquisition of a mesenchymal phenotype. Accordingly, we analyzed the immunophenotype of HUVECs after 6 days, confirming the significant decrease of the endothelial markers CD31 and CD146 expression under End-MT stimulation ( Figure 4A). Here, we also observed a significant increase of the mesenchymal stem cell (MSC) marker CD105, whereas CD44 was comparable among the different conditions ( Figure 4A). Results are summarized in the graphs and table in Figure 4B. The gain of a mesenchymal-like phenotype was further supported by the migratory ability, as shown by the scratch assay ( Figure 4C,D). HUVECs were scratched after being exposed to TGF-β1, TGF-β3 and TNF-α, and acquired a pronounced migration to the wounded area, notably in the TNF-α condition, where the wound was completely closed at 72 h. These data confirm the phenotype plasticity of HUVECs under specific stimulation, indicating that the inflammatory response first promotes the End-MT and, later, the acquisition of a mesenchymal phenotype. (C) Scratch assay was performed to test the migratory potential of HUVECs after treatment with TGF-β1, TGF-β3 and TNF-α. Cells were fixed and stained with Crystal after 24-48-72 h. Images are representative of three independent experiments (4× magnification). (D) Quantification of the wounded area, measured using ImageJ software, and expressed as percentage relative to control group at time 0. All data are expressed as mean ± SD of at least 3 independent experiments and statistical analysis was performed by unpaired Student's t-test. * p < 0.005; ** p < 0.001; **** p< 0.0001. CD, cluster of differentiation.
Based on these data, we asked whether the End-MT process could trigger the gain of mineralization features in HUVECs. Thus, we performed an osteogenic differentiation assay, by culturing HUVECs in the osteogenic induction medium in combination with End-MT mediators until 14 days. As observed by the Alizarin Red stain, and relative quantification data in Figure 5F,G, mineralization occurred when HUVECs were cultured with (D) Quantification of the wounded area, measured using ImageJ software, and expressed as percentage relative to control group at time 0. All data are expressed as mean ± SD of at least 3 independent experiments and statistical analysis was performed by unpaired Student's t-test. * p < 0.005; ** p < 0.001; **** p< 0.0001. CD, cluster of differentiation.
the osteogenic medium alone. However, the addition of TGF-β3 (1.35-fold higher compared to induced sample, p < 0.0001) and TNF-α (1.10-fold compared to induced sample, p < 0.0001) to the induction medium significantly augmented the synthesis of bone matrix ( Figure 5F,G). Figure 5. End-MT stimulates the origin of osteogenic progenitors. (A) Transcriptional levels of RUNX-2 and SOX-9, analyzed by Real-Time PCR in HUVECs exposed for 6 days to TGF-β1, TGF-β3 and TNF-α. Results are reported as fold changes relative to the untreated controls. Data are expressed as mean ± SD of at least 3 independent experiments, and statistical analysis was performed by unpaired Student's t test. * p < 0.005; ** p < 0.001; **** p< 0.0001. Immunofluorescence and relative quantification of (B,C) RUNX-2 (red spots within nuclei) and (D,E) SOX-9 (green nuclei) in HUVECs exposed to TGF-β1, TGF-β3 and TNF-α for 6 days. Scale bars = 25 μm. (F) Alizarin Red stain performed in HUVECs exposed to TGF-β1, TGF-β3 and TNF-α for 14 days in presence of osteogenic medium (OM). (G) Quantification of calcium-bound Alizarin Red by absorbance measured at O.D. 570 nm by spectrophotometer. Data are expressed as mean ± SD of at least 3 independent experiments, and statistical analysis was performed with respect to HUVECs induced with osteogenic medium (OM) by ordinary one-way ANOVA followed by Dunnett's multiple comparison test; * p < 0.005; ** p < 0.001; **** p< 0.0001. RUNX-2, runt-related transcription factor -2; SOX-9, SRY box transcription factor-9.

SLUG Is Expressed in Human Atheroma and During In Vitro Osteogenesis
In order to detect the occurrence of End-MT in vivo, we studied the expression of SLUG in tissue specimens from human carotid plaques and abdominal aortic aneurysms, compared to normal, healthy arterial tissue sections. In Figure 6A,B, it is possible to see focal and weak positivity in medial SMCs, and moderate to strong positivity in endothelial cells lining vasa vasorum. Conversely, in carotid plaques ( Figure 6C,E) and aneurysmal aortas ( Figure 6D,F), the SLUG positivity was intense and diffuse, especially in cells lining the lumen of neo-vessels and among inflammatory cell infiltrates. The sum of areas analyzed by Real-Time PCR in HUVECs exposed for 6 days to TGF-β1, TGF-β3 and TNF-α. Results are reported as fold changes relative to the untreated controls. Data are expressed as mean ± SD of at least 3 independent experiments, and statistical analysis was performed by unpaired Student's t test. * p < 0.005; ** p < 0.001; **** p < 0.0001. Immunofluorescence and relative quantification of (B,C) RUNX-2 (red spots within nuclei) and (D,E) SOX-9 (green nuclei) in HUVECs exposed to TGF-β1, TGF-β3 and TNF-α for 6 days. Scale bars = 25 µm. (F) Alizarin Red stain performed in HUVECs exposed to TGF-β1, TGF-β3 and TNF-α for 14 days in presence of osteogenic medium (OM). (G) Quantification of calcium-bound Alizarin Red by absorbance measured at O.D. 570 nm by spectrophotometer. Data are expressed as mean ± SD of at least 3 independent experiments, and statistical analysis was performed with respect to HUVECs induced with osteogenic medium (OM) by ordinary one-way ANOVA followed by Dunnett's multiple comparison test; * p < 0.005; ** p < 0.001; **** p < 0.0001. RUNX-2, runt-related transcription factor -2; SOX-9, SRY box transcription factor-9.
Based on these data, we asked whether the End-MT process could trigger the gain of mineralization features in HUVECs. Thus, we performed an osteogenic differentiation assay, by culturing HUVECs in the osteogenic induction medium in combination with End-MT mediators until 14 days. As observed by the Alizarin Red stain, and relative quantification data in Figure 5F,G, mineralization occurred when HUVECs were cultured with the osteogenic medium alone. However, the addition of TGF-β3 (1.35-fold higher compared to induced sample, p < 0.0001) and TNF-α (1.10-fold compared to induced sample, p < 0.0001) to the induction medium significantly augmented the synthesis of bone matrix ( Figure 5F,G).

SLUG Is Expressed in Human Atheroma and During In Vitro Osteogenesis
In order to detect the occurrence of End-MT in vivo, we studied the expression of SLUG in tissue specimens from human carotid plaques and abdominal aortic aneurysms, compared to normal, healthy arterial tissue sections. In Figure 6A,B, it is possible to see focal and weak positivity in medial SMCs, and moderate to strong positivity in endothelial cells lining vasa vasorum. Conversely, in carotid plaques ( Figure 6C,E) and aneurysmal aortas ( Figure 6D,F), the SLUG positivity was intense and diffuse, especially in cells lining the lumen of neo-vessels and among inflammatory cell infiltrates. The sum of areas of SLUG cell positivity was measured and reported as mean ± standard deviation in Figure 6G, evidencing a significant increase of SLUG protein in carotid plaques (CP) and abdominal aortic aneurysms (AAA) in comparison to the respective healthy controls. These data suggest the involvement of SLUG in both the physiological and pathological remodeling process, and its possible participation to atherosclerosis progression. In order to study the contribution of SLUG to the mesenchymal differentiation, HUVECs were exposed to osteogenic induction medium for 14 days (Figure 6H,I). After treatment, SLUG expression resulted increased in HUVECs primed to osteogenesis in comparison to controls, as shown by its increased transcript levels in Figure 6J (1.8-fold increase, p = 0.018). These results might support an association between End-MT and vascular calcification. of SLUG cell positivity was measured and reported as mean ± standard deviation in Figure 6G, evidencing a significant increase of SLUG protein in carotid plaques (CP) and abdominal aortic aneurysms (AAA) in comparison to the respective healthy controls. These data suggest the involvement of SLUG in both the physiological and pathological remodeling process, and its possible participation to atherosclerosis progression. In order to study the contribution of SLUG to the mesenchymal differentiation, HUVECs were exposed to osteogenic induction medium for 14 days (Figure 6H,I). After treatment, SLUG expression resulted increased in HUVECs primed to osteogenesis in comparison to controls, as shown by its increased transcript levels in Figure 6J (1.8-fold increase, p = 0.018). These results might support an association between End-MT and vascular calcification.

miR-30a-5p and miR-30d Are Downregulated during End-MT
We investigated the involvement of miR-30a-5p and miR-30d during End-MT and osteogenic differentiation, by analyzing their expression levels in HUVECs exposed to End-MT inducers for 6 days. Expression of miR-30a-5p and miR-30d was lower in HU- Results are reported as fold changes relative to controls in DMEM 10% FBS. All data are expressed as mean ± SD of at least 3 independent experiments, and statistical analysis was performed by unpaired Student's t test. *, p < 0.05; **, p < 0.001; ****, p < 0.0001.

Effects of miR-30a-5p and miR-30d Over-Expression in HUVECs
To investigate whether miR-30a-5p and miR-30d regulate the End-MT, we overexpressed these miRNAs in HUVECs. We preliminarily verified the effects of the transfection, observing the upregulation of miR-30a-5p and miR-30d of 239 ± 103.9, and 3289 ± 1757 -folds respectively, in transfected cells compared to controls after 48 h ( Figure 8A,B). Then, we explored possible effects of miRNA transfection on cell viability, which did not result affected, as shown by the MTT assay ( Figure 8C). The expression levels of the main End-MT markers were analyzed after 48 h over-expression. SLUG mRNA displayed a decreasing trend (Figure 8D), whereas a significant downregulation of its protein was observed after miR-30d over-expression (0.38 ± 0.16, p = 0.032) ( Figure 8E). VIMENTIN showed a significant downregulation of mRNA levels with both miRNAs (0.42 ± 0.15, p = 0.03 in miR-30a-5p over-expressing HUVECs; 0.58 ± 0.11, p = 0.028 in miR-30d over-expressing HUVECs) (Figure 8F), and a reduced protein expression mainly with miR-30d ( Figure 8G). Figure 7. miR-30a-5p and miR-30d are downregulated during End-MT. Relative expression of miR-30a-5p and miR-30d in HUVECs after 6 days of exposure to TGF-β1, TGF-β3 and TNF-α. Results are reported as fold changes relative to the untreated group. All data are expressed as mean ± SD of at least 3 independent experiments, and statistical analysis was performed by unpaired Student's t test. * p < 0.005; ** p < 0.001. miR-NC, miRNA negative control.

Effects of miR-30a-5p and miR-30d Over-Expression in HUVECs
To investigate whether miR-30a-5p and miR-30d regulate the End-MT, we overexpressed these miRNAs in HUVECs. We preliminarily verified the effects of the transfection, observing the upregulation of miR-30a-5p and miR-30d of 239 ± 103.9, and 3289 ± 1757 -folds respectively, in transfected cells compared to controls after 48 h ( Figure 8A,B). Then, we explored possible effects of miRNA transfection on cell viability, which did not result affected, as shown by the MTT assay ( Figure 8C). The expression levels of the main End-MT markers were analyzed after 48 h over-expression. SLUG mRNA displayed a decreasing trend ( Figure 8D), whereas a significant downregulation of its protein was observed after miR-30d over-expression (0.38 ± 0.16, p = 0.032) ( Figure 8E). VIMENTIN showed a significant downregulation of mRNA levels with both miRNAs (0.42 ± 0.15, p = 0.03 in miR-30a-5p over-expressing HUVECs; 0.58 ± 0.11, p = 0.028 in miR-30d over-expressing HUVECs) (Figure 8F), and a reduced protein expression mainly with miR-30d ( Figure 8G). The bioinformatic analysis carried out by the TargetScan algorithm identified SLUG among predicted targets of miR-30a-5p and miR-30d, whose binding sequence was found in the 3′-UTR ( Figure 9A). To demonstrate whether miR-30a-5p and miR-30d directly bind the 3′-UTR of SLUG, we assessed a luciferase reporter assay. To this aim, a portion of SLUG 3′UTR, containing two binding sites for miR-30a-5p and miR-30d, was cloned in the pmiR-GLO vector downstream of a firefly luciferase reporter gene. Then, HUVECs were co-transfected with this vector and miR-30a-5p, miR-30d, or miR-NC. After 24 h, we found that co-transfection of pmiR-GLO SLUG 3′UTR with miR-30a-5p in HUVECs reduced the luciferase activity by 39% (p = 0.014) in comparison to HUVECs co-transfected with miR-NC. Similarly, co-transfection of pmiR-GLO SLUG 3′UTR with miR-30d in HUVECs reduced the luciferase activity by 43% (p = 0.0073) in comparison to HUVECs co-transfected Figure 8. miR-30a-5p and miR-30d over-expression in HUVEC. Over-expression of (A) miR-30a-5p and (B) miR-30d in HUVECs transfected through Lipofectamine RNAiMAX. Results are reported as fold changes relative to control. (C) Effects of miR-30a-5p and miR-30d mimics on HUVEC viability at 24 and 48 h, evaluated through the MTT assay. Analysis of SLUG (D) mRNA and (E) protein, analysis of VIMENTIN (F) mRNA and (G) protein. Western Blot bands were normalized to total protein amount through gel stain free using the ImageLab Software. Real-Time results are reported as fold changes to control. All data are expressed as mean ± SD of 4 independent experiments, and statistical analysis was performed by unpaired t test. * p < 0.05; ** p < 0.01. miR-NC, miRNA negative control; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide.

Discussion
In the present study, we investigated the involvement of miR-30a-5p and miR-30d during End-MT and vascular calcification. To this aim, we firstly established a cell model for studying End-MT and its potential association with the in vitro osteogenic process. We therefore explored the endothelial cell plasticity and phenotype transition in the presence of growth factors and cytokines mimicking the inflammatory microenvironment expected to be present during the development of the atheromatous plaque. During End-MT, endothelial cells lose their typical cobblestone shape and markers, switching to a MSC phenotype, including markers, cytoskeleton rearrangements and properties, i.e., migration, multilineage differentiation. End-MT occurs under vascular injury, representing an intermediate step in atherosclerosis and calcification [21]. We demonstrated that TGF-β1, TGF-β3 and TNF-α reproduce the inflammatory setting in vitro and induce End-MT in HUVECs, which acquire morphology (evidenced by F-actin staining), markers (MMP-9, SLUG, VIMENTIN, CD105) and motility, typical of MSCs. In parallel, we detected the loss of endothelial markers in all experimental conditions (CD31, CD146). VE-CADHERIN expression resulted affected only in TGF-β1 and TGF-β3, whereas it was increased in cells treated with TNF-α, indicating a possible compensatory response to keep the endothelial barrier functionality.
Interestingly, we also detected the upregulation of the transcriptional factors RUNX-2 and SOX-9, commonly associated with the osteo/chondrogenic lineage. We further demonstrated that the combination of TGF-β3 and TNF-α with osteogenic induction medium significantly increased the mineralization activity in HUVECs. These results suggest the ability of transitional endothelial cells to become a significant contributor of special-

Discussion
In the present study, we investigated the involvement of miR-30a-5p and miR-30d during End-MT and vascular calcification. To this aim, we firstly established a cell model for studying End-MT and its potential association with the in vitro osteogenic process. We therefore explored the endothelial cell plasticity and phenotype transition in the presence of growth factors and cytokines mimicking the inflammatory microenvironment expected to be present during the development of the atheromatous plaque. During End-MT, endothelial cells lose their typical cobblestone shape and markers, switching to a MSC phenotype, including markers, cytoskeleton rearrangements and properties, i.e., migration, multilineage differentiation. End-MT occurs under vascular injury, representing an intermediate step in atherosclerosis and calcification [21]. We demonstrated that TGF-β1, TGF-β3 and TNF-α reproduce the inflammatory setting in vitro and induce End-MT in HUVECs, which acquire morphology (evidenced by F-actin staining), markers (MMP-9, SLUG, VIMENTIN, CD105) and motility, typical of MSCs. In parallel, we detected the loss of endothelial markers in all experimental conditions (CD31, CD146). VE-CADHERIN expression resulted affected only in TGF-β1 and TGF-β3, whereas it was increased in cells treated with TNF-α, indicating a possible compensatory response to keep the endothelial barrier functionality.
Interestingly, we also detected the upregulation of the transcriptional factors RUNX-2 and SOX-9, commonly associated with the osteo/chondrogenic lineage. We further demonstrated that the combination of TGF-β3 and TNF-α with osteogenic induction medium significantly increased the mineralization activity in HUVECs. These results suggest the ability of transitional endothelial cells to become a significant contributor of specialized cells able to release pro-calcific matrix. Since vascular calcification is currently considered a stem cell-driven process, endothelial cells seem to acquire a novel and dynamic role during vascular remodeling and disease progression. Thus, End-MT represents a novel and promising target for preventing the pathological bone formation. Comparing all data, we noticed that TNF-α and TGF-β3 were more effective than TGF-β1 at triggering endothelial cell plasticity versus a mesenchymal/osteogenic phenotype. These data confirm the pivotal role of inflammation to the vascular remodeling and calcification process, and also suggest the contribution of the TGF-β3 signaling to the atherosclerosis complications.
Once established this in vitro model of End-MT, we investigated the role of miR-30a-5p and miR-30d as possible modulators of endothelial cell behavior. Indeed, miRNAs recently emerged as novel and intriguing regulators of many biological processes, including End-MT, through their ability to target genes that are critical to endothelial cell trans-differentiation [22]. As a first step, we analyzed the expression levels of miR-30a-5p and miR-30d, members of the miR-30 family that controls the development and biology of bone, adipose tissue and blood vessels [23]. We found that these miRNAs were both downregulated in End-MT, especially in TGF-β3 and TNF-α conditions. This result strengthens the hypothesis on a direct link between the osteo/chondrogenic commitment, inflammation and End-MT. The analysis performed through TargetScan identified that VIMENTIN and SNAI1 are predicted targets of miR-30a-5p and miR-30d [15], thus we explored the interactions between these miRNAs and some End-MT mediators by performing gain-of-function experiments in HUVECs. The over-expression of miR-30a-5p and miR-30d in HUVECs was associated with the downregulation of SLUG, notably with miR-30d. Interestingly, we observed a significant reduction of the relative luciferase activity in HUVECs co-transfected with the pmiR-Glo 3 UTR SLUG and miR-30a-5p, and miR-30, after 24 and 48 h. These data demonstrated that a direct interaction exists between SLUG and both miRNAs, therefore SLUG is a direct target of miR-30a-5p and miR-30d. We also observed a significant decrease of VIMENTIN following miRNA over-expression, and this result was in accordance with literature studies that have already investigated that VIMENTIN is a direct target of miR-30a-5p [24,25]. These data imply that miR-30a-5p and miR-30d may target the pathological differentiation of endothelial cells. The role of the miR-30 family in End-MT has been poorly investigated. More data are available with regard to EMT. Chung et al. demonstrated the inhibitory role of miR-30a-5p on EMT by increasing the tight-junction claudin-5 in human upper tract urothelial carcinoma cells [16]. Further, miR-30a-5p targets the ROR1, the orphan like receptor tyrosine kinase (RTK), reducing EMT and metastasis formation in triple-negative breast cancer [26]. However, data on endothelial cells are few. Jiang et al. observed that miR-30a stimulates arteriolar branching [27], whereas another work on miR-30b showed its inhibitory role on capillary morphogenesis acting on TGF-β2 signaling [28]. Thus, members of the same miRNA family may exploit distinct functions.
Considering that End-MT is strongly connected with the osteogenic differentiation process and atherosclerotic calcification, we further explored this issue. The members of the miR-30 group can regulate the osteogenic process, indeed their over-expression regulated the BMP-2-induced osteogenic process by targeting RUNX-2 and Smad1 in mouse osteoblasts and bone marrow mesenchymal stem cells [29]. Moreover, it was shown that Bone Morphogenetic Protein 2 (BMP-2) was able to downregulate miR-30b and miR-30c, consequently increasing RUNX-2 expression in a model of human coronary artery SMCs [30]. A recent work also demonstrated that miR-30b is able to reduce vascular calcification in vivo [31]. Another study demonstrated that miR-30a-5p levels were decreased in serum of atherosclerotic patients, and the same research explored the role of miR-30a-5p in a monocyte-endothelial cell co-culture system, showing that miRNA overexpression in the human monocyte cell line THP-1 cells protects endothelial cells from apoptosis [32]. Our recently published study reported that miR-30a-5p and miR-30d are differentially expressed in carotid plaques exhibiting different calcification patterns. In addition, the expression levels of miR-30a-5p and miR-30d were reduced in endothelial cells after osteogenic differentiation [14]. These preliminary observations and previous studies, which elucidated the involvement of the miR-30 family in MSC osteogenic differentiation, prompted us to study miR-30a-5p and miR-30d in endothelial cells under osteogenic conditions. To this purpose, we transfected HUVECs with miR-30a-5p-mimic and miR-30d-mimic during osteogenic induction, for 14 days. At the end of treatment, we noticed a significant lowering of the mineralization ability in transfected HUVECs, in accordance with the reduced expression of RUNX-2 and BMP-2 genes.
In this study, we also highlight the relevance of SLUG to atheroma calcification, whose involvement during arterial calcification has not been elucidated yet. Sànchez-Duffhues et al. demonstrated that SLUG is involved in endothelial cell-mediated calcification [33]. Here, we report that SLUG is highly expressed in human atherosclerosis, both at the aortic and carotid level, as well as in vitro when HUVECs differentiated into osteoprogenitors. These data implicate the involvement of SLUG during the pathological remodeling occurring in atherosclerosis and calcification processes, providing novel cues for investigating the atherosclerotic calcification and the search for new therapeutic approaches.
To the best of our knowledge, this is the first study addressing the role of miR-30a-5p and miR-30d in endothelial cells with respect to the End-MT and calcification process. We showed that End-MT transformation is tightly associated with the vascular bone formation under the inflammatory milieu and represents a potential clinically relevant target. Our results, in accordance with the literature data, support a negative correlation between the investigated miRNA and the atherosclerotic calcification. miRNAs are attractive candidates for promising therapeutic approaches aimed at controlling the disease progression, by regulating the main upstream mechanisms, as well as possible stage-specific markers. The present study can therefore be preliminary to future investigations for elucidating the role of miR-30a-5p and miR-30d, novel signaling pathways connected with atherosclerosis and calcification and the possible delivery approach in the light of their utilization in the clinical practice.

Conclusions
In the present study, we demonstrated that two members of the miR-30 family, miR-30a-5p and miR-30d, are deregulated during End-MT and possess a key role during this process. We established an in vitro End-MT model in the presence of soluble factors belonging to the TGF family and to the inflammatory cytokines. We observed that TGF-β1, TGF-β3 and TNF-α stimulate the occurrence of End-MT, a phenotype switch that triggers different pathological vascular changes, and the increase of mineralization activity. Under these conditions, miR-30a-5p and miR-30d were downregulated; conversely, their over-expression was associated with a decreased expression of SLUG, VIMENTIN and with a less prominent osteogenic differentiation in HUVECs. Altogether, our data support the involvement of miR-30a-5p and miR-30d in the endothelial cell plasticity toward the osteogenic phenotype, but further investigations are needed to unveil this regulatory axis.

Conflicts of Interest:
The authors declare no conflict of interest.