Involvement of Fibulin-5 in endothelial to mesenchymal transition leading to cardiac fibrosis during metabolic syndrome

Abstract

Background

Cardiac fibrosis is a hallmark of metabolic syndrome, a condition linked to Western lifestyles and high cardiovascular risk. We previously demonstrated that dietary sodium restriction prevents cardiac fibrosis and remodeling in a rat model of metabolic syndrome through reduced macrophage infiltration. Here, we investigate genes involved in endothelial-to-mesenchymal transition (EndoMT), a key process in cardiac fibrosis.

Methods

Metabolic syndrome was induced in rats by high-fructose feeding combined with angiotensin II infusion. EndoMT was assessed in left ventricles and in vitro using TGF-β2-treated human aortic (HAEC) and umbilical vein endothelial cells (HUVEC) via immunofluorescence, western blotting, and RT-qPCR. Lentiviral shRNA was used to knock down target genes.

Results

Sodium restriction reduced vascular EndoMT in rat left ventricles and downregulated several candidate genes. In vitro, Fibulin-5, a matricellular protein, was markedly upregulated during EndoMT. Silencing Fibulin-5 prevented TGF-β2-induced EndoMT in HAEC and HUVEC, indicating its essential role. Mechanistically, Fibulin-5 modulated SMAD2/3, ERK1/2, and p38 MAPK pathways. In vivo, Fibulin-5 expression was significantly reduced in aortic intima and plasma of sodium-restricted rats.

Conclusion

Fibulin-5 emerges as a potential mediator of vascular EndoMT and cardiac fibrosis through TGF-β signaling modulation. Dietary sodium restriction mitigates this process, highlighting Fibulin-5 as a potentially important contributor in metabolic syndrome-related cardiac remodeling.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12967-026-08589-x.

Background

Metabolic syndrome (MetS), typically associated with Western lifestyle [1], comprises at least 3 of the following disorders: arterial hypertension, insulin-resistance, dyslipidemia and abdominal obesity, which are recognized as leading risk factors for metabolic and cardiovascular diseases [2]. With an estimated prevalence of 25% in the general population, individuals affected by MetS have a 70% increased risk of sudden cardiac death, and a cardiovascular risk increased by more than 2-fold [1]. Cardiac fibrosis is one of the silent features of cardiovascular disease (CVD) implicated in idiopathic dilated cardiomyopathy, diabetic cardiomyopathy, or hypertensive heart disease [3]. Fibrosis is the result of a continuous and excessive accumulation of proteins, such as type 1 collagen, in the extracellular space, produced by cells of mesenchymal origin, such as fibroblasts, progressively impairing cardiac function [4, 5].

Interstitial fibrosis occurs in the absence of the massive loss of cardiomyocytes and is caused by prolonged activation of fibrogenic stimuli. It typically occurs under pathological conditions, such as elevated pressure or volume overload [6], diabetes mellitus [7], and is generally associated with systolic ventricular dysfunction [8]. In interstitial fibrosis, resident fibroblasts become activated and undergo conversion to myofibroblasts with the expression of contractile proteins (i.e. α-SMA (Actin, alpha Skeletal Muscle), SM22α (Smooth Muscle Protein 22-Alpha)) and the de novo synthesis of matricellular proteins (i.e. type 1 collagen). Importantly, endothelial cells may also contribute to cardiac fibrosis through endothelial to mesenchymal transition (EndoMT) [6, 9]. EndoMT is a physiological process implicated in valve formation and heart septation. This transdifferentiation process involves the loss of endothelial identity and the acquisition of mesenchymal characteristics. EndoMT has been recognized as a key process in cardiac fibrosis and has been implicated in various cardiovascular diseases, including pulmonary arterial hypertension [10–14], atherosclerosis [15], diabetes mellitus [16–19] and chronic kidney disease [20]. Endothelial cells are indeed able to acquire mesenchymal properties under some physical conditions such as laminar shear stress [21], and hypoxia [22] or by various biochemical stimuli, especially Transforming Growth Factor-β (TGF-β) [23] and inflammation. Indeed, in the context of metabolic diseases, we and others [24, 25] demonstrated macrophages infiltrate the remodeling heart, influencing fibroblast activation by secreting a wide range of bioactive mediators, including the fibrotic factors belonging to the TGF-β family. However, the precise role of EndoMT in cardiac fibrosis and cardiovascular disease remains unclear, underscoring the need for further investigation. Establishing appropriate cellular and animal models is therefore essential to uncover the underlying mechanisms of this phenomenon [9].

We and others have reported that in MetS, low salt diet confers significant health benefits in humans [26] and in animal models [25, 27]. This non pharmacological intervention allows the prevention of cardiac remodeling in various models of hypertension [28–30] as well as in hypertensive patients [31]. Furthermore, sodium reduction shows cardiac benefits in insulin resistance and in the Dahl S.Z-Lepr^fa/Lepr^fa rat [27], another MetS model, with effects linked to a decrease in oxidative stress [28–30]. We have also previously demonstrated in a rat model of MetS that dietary sodium restriction prevents cardiac damage, especially interstitial fibrosis and cardiac hypertrophy despite persistent hypertension [25]. This strongly suggests that the beneficial effects of sodium restriction on cardiac remodeling cannot be explained solely by hemodynamic changes, indicating the involvement of pressure-independent mechanisms. In that study, we identified several groups of genes differentially expressed in the cardiac left ventricle of MetS rats fed a low sodium diet, which correlated with a significant reduction of cardiac fibrosis and cardiomyocyte hypertrophy [25]. Therefore, the present study pursued two main objectives: (1) to demonstrate the involvement of EndoMT in fibrosis associated with MetS, and (2) to assess the potential contribution of novel candidate genes to this process. In particular, we identified for the first time Fibulin-5, an extracellular matrix protein, as being involved in vascular endothelial to mesenchymal transition associated with cardiac fibrosis in the context of metabolic syndrome.

Methods

Animal model

The experiments were performed in accordance with the European and French laws (permit numbers B-3417226 and 34179, agreement D34-172-25) and conformed to the “Guide for the Care and Use of Laboratory Animals” published by the National Institute of Health (National Academies Press US, 8th edition, 2011). Rats were housed in climate-controlled conditions with a 12 h light/dark cycle in a temperature-controlled room (22 ± 1 °C).

Two groups of 10 male Sprague-Dawley rats (Charles River, 180–200 g) were fed for 8 weeks a 60% fructose diet with either a normal sodium (NSF: 0.64% NaCl) or a low sodium content (LSF: <0.01% NaCl). After four weeks, all the rats were infused subcutaneously with angiotensin II (AngII, 200 ng.kg^− 1.min^− 1) using Alzet mini-pumps for the remaining four weeks. A summary of the metabolic parameters used to assess the establishment of metabolic syndrome in Sprague Dawley rats is indicated in Table 1.

Table 1.

Summary of key metabolic parameters used to assess the establishment of metabolic syndrome in the fructose + angiotensin II rat model under normal-salt (NSF) and low-salt (LSF) dietary conditions. Both groups were treated for 8 weeks, and standard markers of metabolic dysfunction were measured. Data are presented as mean ± SEM (N = 10 rats per group). Differences between the two groups were evaluated using unpaired t-tests

Parameter	NSF	LSF
Final body weight (g)	418 ± 8	342 ± 7*
BW change from week 4 (g)	60.0 ± 4.4	32.6 ± 3.3*
Fasting blood glucose (mg/dL)	125 ± 4	120 ± 4
Systolic tail-cuff pressure (mmHg)	196.7 ± 16.4	174.5 ± 18.6*
Heart weight (g)	1.39 ± 0.04	1.02 ± 0.07*
Heart weight index (mg/gBW)	3.30 ± 0.06	3.00 ± 0.07*
LV weight (g)	1.03 ± 0.03	0.75 ± 0.03*
LV weight index (mg/gBW)	2.48 ± 0.05	2.17 ± 0.05*
Cardiac fibrosis (% Sirius Red–positive area)	6,5 ± 1.2	4,97 ± 0,8*

* : p < 0.05 vs. the NSF group

BW: Body Weight; LV: Left Ventricle

Cardiac sirius red staining

The left ventricle was paraffin embedded, and 3–5 μm slices were cut for histological analysis. Cardiac fibrosis was quantified by measuring the collagen-stained area fraction (percentage of Sirius red-positive area relative to total tissue section area, as previously described [32]).

RNA extraction, microarray and statistical data analysis

Methods were previously described in [25]. Sample treatment was performed as described in the “one color microarray-based gene expression analysis” protocol (version 6.7, part number G4140-90040, Agilent). Briefly, 200ng of each total RNA samples (n = 12) were used for amplification and labelling with the Agilent Low Input Quick amp Labelling kit, one color. Yields of cRNA and dye incorporation rate were measured using ND-1000 spectrometer (NanoDrop technologies, Peqlab biotechnologies GmbH, Erlangen, Germany). Hybridization procedure was performed using Agilent Gene Expression Hybridization kit. Fluorescence signals were detected using Agilent’s Microarray Scanner System and the Agilent Feature Extraction Software (version 11.0.1.1) to read out and process the microarray image files.

The microarray data have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO series accession number GSE84524.

Statistical analysis was performed using R software version 3.3.0⁴ along with the Limma package. The method used for background correction was based on the normal-exponential convolution model with the saddle-point approximation to maximum likelihood. Normalization was performed using cyclic loess method. Only probes whose signal was considered as higher than background in at least four out of six replicates in at least one condition were selected for further analysis. Within-array replicate probes were replaced with their average. The assessment of differentially expressed mRNAs between NS diet and LS diet was performed using the limma GLM (Generalized Linear Model) method followed by Benjamini Hochberg correction for multiple testing. Genes with a corrected p-value lower than 0.1 were selected for further investigation. Candidate genes identified through transcriptomic screening were first confirmed using qPCR analysis. Genes showing validated differential expression were then selected based on their biological relevance to EndoMT and cardiac fibrosis, assessed through careful literature review, and subsequently grouped into functional categories for further investigation.

Cell culture and reagents

Endothelial primary cells

Human Aortic Endothelial Cells (HAEC, #CC-2535, Lonza) and Human Umbilical Vein Endothelial Cells (HUVEC, Promocell) were used between the 5th and the 8th passage. Cells were maintained in endothelial growth medium consisting of EBM^®-2 (Lonza) supplemented with EGM-™2 Bullet Kit (Lonza) containing 2% FBS (Lonza). Cells were maintained at 37 °C in a humidified 5% CO2 atmosphere.

To induce EndoMT in vitro, cells were uniformly plated (104 cell/cm2) on 0.1% gelatin-coated dishes (attachment factor 1X, Gibco). At 70% confluence, cells were treated with human recombinant TGF-β2 (10ng/ml, Cell Signaling) for 5, 7 or 10 days. This concentration of TGF-β2 was previously shown to be effective in promoting EndoMT [33].

Lentiviral particles expressing FBLN5 shRNA or control particles were purchased from Origene. Cells were transduced at passage 5 in the presence of polybrene (10 µg/ml, Abcam) at a multiplicity of infection (MOI) of 30 O/N. The medium was replaced the following day and cells were treated for 7 days with TGF-β2 (10ng/ml) to induce EndoMT. Description of the validation experiments performed to confirm the efficacy of the lentiviral shRNA construct used for FBLN5 knockdown, including the knockdown efficiency at both mRNA and protein levels are shown in additional file 1 (additional Fig. 1).

Fig. 1.

Expression validation of selected targets potentially involved in endothelial to mesenchymal transition in MetS rats. A, Protein-Protein Interaction (PPI) network of genes downregulated by sodium restriction in the left ventricle of rats with metabolic syndrome (STRING database). B, mRNA expression levels performed on total RNA extracts from left ventricles, n = 10 animals per group. Statistical analyses were performed using unpaired t-test. The p-values are as follows: Anxa2 (p = 0.0002), Lphn1 (p = 0.0002), Cilp1 (p = 0.0011), and Fbln5 (p = 0.0001). C, Western blot on left ventricles total protein extracts. n = 6 rats from each group are shown. For WB quantification, band density was measured using Image J. Statistical analysis were performed using unpaired t-test. The p-values are as follows: Anxa2 (p = 0.002), Lphn1 (p = 0.200), Cilp1 (p = 0.008), and Fbln5 (p = 0.001). One representative experiment out of 3 is shown. *p < 0,05, **p < 0.01, ***p < 0.001

RNA extraction, reverse transcription, and real-time quantitative PCR analysis

Tissues

Immediately after anaesthesia, the heart was collected, cleaned and weighed, and left ventricle samples were frozen in RNAlater^® solution (Life Technologies). Tissues were then dissociated in RLT buffer using a Fast-Prep^® homogeneizer (MP Biomedicals). Total RNA was extracted using RNeasy^® Fibrous Tissue mini-kit (Qiagen).

Cells

Cells were scraped in RLT + β-mercaptoethanol buffer, and total RNA was extracted using RNeasy^® mini-kit (Qiagen), following manufacturer’s protocol.

RNA yield and purity were determined using the NanoDrop™ Spectrophotometer (Thermo Scientific, USA). One µg total RNA was reverse-transcribed using the Omniscript^® RT Kit (Qiagen) according to the manufacturer’s instructions. Real-time quantitative PCR was carried out using the LC96 real time PCR instrument (Roche). Primer pairs (IDT) were generated using NCBI primer design software and were designed to span an exon-exon junction. Forward and reverse primers sequences are listed in additional file 1 and were used at a 60 °C annealing temperature. The relative amounts of PCR products were quantified using the relative threshold cycle (ΔΔCt) method. The relative gene quantities were normalized against the mean of expression of the housekeeping genes Gapdh, HPRT1 and ACTB respectively for rat left ventricles, HAEC and HUVEC.

Protein analysis

Protein extraction

Snap frozen left ventricles from NSF (N = 6) and LSF (N = 6) rats, or cells (N = 3 independent cultures) were dissociated in RIPA buffer (Abcam) supplemented with 1% protease and phosphatase inhibitor cocktail (Cell signalling) using a Fast-Prep^® homogeneizer (MP Biomedicals). Total proteins were quantified using BCA assay (Pierce).

Western Blot

Protein samples were diluted in Laemmli buffer, boiled for 5 min at 95˚C. Twenty microgram of protein were loaded in 10% SDS-PAGE gel (3 h at 100 V) and after migration were transferred to 0.45 μm PVDF membranes (Immobilon^®-P, Millipore). Nonspecific sites were blocked with PBS, 5% dry skimmed milk for 2 h at room temperature (RT) under moderate shaking. The proteins of interest were then labelled with appropriate primary antibodies (detailed in additional file 1) diluted in PBS, 5% dry skimmed milk, 0.1% Tween-20, and incubated overnight at 4 °C. GAPDH and α-TUBULIN were used as housekeeping genes. The secondary HRP antibodies (detailed in additional file 1) were incubated for 1 h at RT. Blots were visualized using the chemiluminescence kit detection solution (Immobilon™, Millipore) and imaged using the ChemiCapt Imaging System (Vilber Lourmet). Band intensities were quantified using ImageJ software. All uncropped western blots are shown in additional file 2.

ELISA

After 8 weeks of diet, rats were anaesthetized, and 1,5 ml of carotid blood was collected from all study animals in heparin tubes. Blood samples were centrifuged at 3000 rpm for 10 min at 4 °C. Then, plasma was aliquoted and stored at − 80 °C for further experiments. Fibulin-5 concentration was determined using a rat Sandwich ELISA kit (OKEH06327, Aviva systems biology, California, USA) according to the manufacturer’s protocol. Plasma samples were diluted at 1:50. Absorbance was measured at 450 nm using a microplate reader (Tecan, USA).

Immunofluorescence and image analysis

After sacrifice, left ventricles and thoracic aorta were immediately formalin fixed and included in paraffin. Blocks were cut (3 μm) on SuperFrost slides, and dried for 1 h at 37 °C. Slides were deparaffinized, dehydrated and treated in Tris-EDTA pH9 buffer (left ventricles) or citrate pH6 buffer (thoracic aorta) for 40 min. After washing, blocking was performed in PBS BSA 2% normal donkey serum (Abcam) 5% for 1 h at RT. Primary antibodies were incubated 2 h at RT. After washings, secondary antibodies were incubated for 2 h at RT. The list of antibodies used is detailed in additional file 1. Slides were mounted in Vectashield Vibrance containing DAPI (Vector). Entire aortic rings were imaged using the mosaic acquisition mode of a Leica DM6B Thunder fluorescence microscope at 40× magnification. Image analysis was performed on entire aortic ring sections using a standardized segmentation protocol. The inner band (tunica intima) was determined as a 50 pixel wide ring extending inward from the luminal edge, and the external band (tunica adventitia) as a 75 pixel wide ring extending outward from the outer edge of the aorta. The entire aortic wall was segmented as the area between these two contours, calculated as the XOR combination of the outer and inner boundaries. For each region, the integrated intensity (sum of pixel intensities within the defined area) was quantified.

Cells were plated into 24-well dishes on gelatin-coated glass coverslips, with 5.10⁴ cell per well. At defined times, cells were fixed for 15 min in PFA 4%, and permeabilized in 0.1% Triton X-100 for 30 min. Then, nonspecific sites were blocked for 45 min with 5% normal donkey serum. Endothelial and mesenchymal markers were stained using specific antibodies detailed in additional file 1. After 3 × 10 min of PBS washing, secondary antibodies were incubated for 1 h at RT. Glass slides were mounted using Vectashield Vibrance containing DAPI (Vector). To estimate changes in Fibulin-5 protein expression on the immunofluorescence images, we calculated the Corrected Total Cell Fluorescence (CTCF) using Image J software and the method described at https://theolb.readthedocs.io/en/latest/imaging/measuring-cell-fluorescence-usingimagej.html.

Confocal microscopy

For evaluation of EndoMT in left ventricles, images were collected on a LSM780 laser scanning confocal microscope (Carl Zeiss, Germany) using a 40x oil immersion objective (N.A. 1.4). Dapi, Alexa Fluor-488 and Alexa Fluor 647 were sequentially imaged at 405, 448 and 633 nm respectively, and respective emitted fluorescence were collected as follows: 415–485 nm for DAPI, 500–651 for Alexa Fluor-488, 645–701 nm for Alexa Fluor 647. z-stacks of 10 slices encompassing entire vessels found in left ventricles were collected. Maximum intensity projection images on single optical slices were processed using Omero software. Coexpression was determined using a line scan (shown in white) traced over the entire vessel. Data of relative staining intensity versus distance was exported and processed using Microsoft Excel.

Statistical analysis

All data are expressed as the mean ± SEM of at least three independent experiments. Quantitative PCR results are expressed as the mean ± SEM. The statistical evaluation was performed using Student’s t-test or Mann-Whitney U non-parametric test depending on the normality of the distribution. The normality of the distribution was assessed with a Kolmogorov Smirnov test. Differences were considered significant when p < 0.05. P-values are indicated in the figure legends or are indicated in additional file 1. Statistical analyses were performed with Excel and GraphPad Prism softwares.

For analysis of EndoMT using confocal microscopy, Student’s t-test was used to determine whether the proportions of vessels with EndoMT in each animal were different between the NSF and DSF groups, with p < 0.05 considered statistically significant.

Results

Downregulation of a set of 17 genes involved in fibrosis in rats with MetS and fed a low sodium diet

Our previous transcriptomic data allowed us to identify a new set of 17 genes involved in the process of fibrosis and all significantly down regulated in MetS rats fed the low sodium fructose diet (LSF), compared to the normal sodium fructose diet (NSF) [25]. The significant downregulation of all these genes was confirmed by RT-qPCR (Table 2). We then studied the gene network involved using STRING database and observe seven edges being present, with a PPI (Protein-Protein Interaction) enrichment p-value of 3.24e-07, meaning this network has significantly more interactions than expected by chance (Fig. 1A). Among those 17 genes, 3 functional modules could be identified: (1) genes implicated in extracellular matrix and fibrosis, such as Col1a2, Col4a1, Eln, Fbln5 and Loxl2, (2) genes known to be implicated in endothelial to mesenchymal transition, such as Vim and Nes, which are also often associated with epithelial to mesenchymal transition, and could suggest a role in remodeling processes affecting the heart; and (3) genes implicated in oxidative stress and inflammation: Pi16, Serpinf1, Aif1. In this network, we were specifically interested in target genes that have not yet been described for their involvement in endothelial to mesenchymal transition (EndoMT) leading to fibrosis. We therefore focused on Anxa2, Lphn1, Cilp1 and Fbln5 and were able to validate that the low sodium fructose diet downregulated their mRNA (Fig. 1B) and protein levels (Fig. 1C) in rat left ventricles. To investigate whether the fructose-angiotensin II condition can impact the expression of these markers, we included a healthy control group fed a standard diet and undergoing sham surgery. Fibulin-5, Cilp1, and mesenchymal markers (α-SMA, Vimentin) were significantly upregulated in NSF rats compared to healthy controls, confirming their pathological induction by the high-fructose diet combined with angiotensin II infusion. Notably, their expression in LSF rats was significantly lower than in both NSF rats and healthy controls, suggesting that sodium restriction actively suppresses pro-fibrotic signaling beyond simple normalization (Additional Fig. 2).

Table 2.

Fibrose related transcripts that are differentially expressed between NSF and LSF rats. Microarray and qPCR validation data are shown

Gene symbol	Microarray			qPCR		NCBI description
	Fold Change	p value	Corrected p	Fold Change	p value
Nppb	-2.43	9,65E-05	8,81E-02	-2.42	2,00E-04	Natriuretic peptide B
Cilp	-2.06	2,86E-04	9,43E-02	-2.22	< 1,00E-04	Cartilage intermediate layer protein
Nes	-1.92	3,70E-04	9,43E-02	-2.99	5,00E-04	Nestin
Ace2	-1.64	2,56E-04	9,43E-02	-2.00	3,00E-04	Angiotensin I converting enzyme 2
Loxl2	-1.62	3,91E-05	8,81E-02	-1.86	5,00E-04	Lysyl oxidase like 2
Fbln5	-1.55	1,43E-04	8,93E-02	-1.51	< 1,00E-04	Fibulin 5
Eln	-1.53	3,59E-04	9,43E-02	-1.71	1.50E-03	Elastin
Col4a1	-1.53	2,52E-05	8,81E-02	-1.47	5,00E-04	Collagen, type IV, alpha 1
Col1a2	-1,50	4,13E-04	9,43E-02	-1,53	2,79E-05	Collagen, type I, alpha 2
Aif1	-1,50	1,22E-04	8,81E-02	-1,25	1,75E-04	Allograft inflammatory factor 1
Sox18	-1.51	3,99E-04	9,43E-02	-1.58	< 1,00E-04	SRY box 18
Vim	-1.46	2,77E-04	9,43E-02	-1.62	5,00E-04	Vimentin
Pi16	-1,44	1,54E-04	8,93E-02	-1,24	1,44E-02	Peptidase inhibitor 16 defense immunity protein
Lphn1	-1.38	3,86E-04	9,43E-02	-1.65	< 1,00E-04	Adhesion G protein-coupled receptor L1 latrophilin 1
Serpinf1	-1.50	3,28E-04	9,43E-02	-1.21	5,00E-04	Serpin family F member 1
Anxa2	-1.35	4,10E-04	9,43E-02	-1.32	5,00E-04	Annexin a2
Gng12	-1.28	3,09E-04	9,43E-02	-1.38	5,00E-04	G protein subunit gamma 12

Fig. 2.

Endothelial to mesenchymal transition is significantly reduced in the heart of rats with metabolic syndrome fed a low sodium diet. A, Pecam1 (pink) and α-SMA (green) were detected by immunofluorescence in left ventricular cardiac tissue sections of the NSF and LSF groups using confocal microscopy. Colocalization was measured using the Omero line scan function. Images are from a single optical slice. B, Percentage of cardiac vessels exhibiting EndoMT, defined by co-expression of Pecam-1 and α-SMA per animal in the NSF and LSF groups. Data are expressed as mean ± SEM (N = 7 animals per group). Statistical comparison was performed using a Student’s t-test (p = 0.0018). C, Western blot performed on total protein extracts from left ventricles (N = 6 of each group). Statistical significance between NSF and LSF groups was assessed using an unpaired t-test. P-values were as follows: Pecam1 (p = 0.0074), Nos3 (p = 0.020), α-Sma (p = 0.0424), and Vimentin (p = 0.009). One representative experiment out of 3 is shown. Band density was measured using Image J. D, Representative Sirius Red staining images with quantification of collagen deposition (% stained area in left ventricular tissue). Mean ± SEM; unpaired t-test (NSF vs. LSF), (N = 6 animals per group p = 0.015). Magnification: ×20. *p < 0,05, **p < 0.01, ***p < 0.001

Sodium restriction significantly decreases EndoMT in the left ventricle of rats with MetS

To demonstrate the presence of cells undergoing EndoMT, we performed immunofluorescence and confocal microscopy on left ventricular cardiac tissue sections to visualize cells co-expressing the endothelial marker Pecam-1, together with the mesenchymal marker α-SMA (Fig. 2A). Blood vessels containing transitioning endothelial cells were counted revealing that Pecam-1 was coexpressed with α-SMA in 67,3 ± 4,9% blood vessels per animal in the NSF group, vs. 42,3 ± 3,8% blood vessels per animal in the LSF group (Fig. 2B, p < 0,05). By western blotting, Pecam-1 and Nos3 expression were significantly increased in LSF rats, while α-SMA and Vimentin expression decreased (Fig. 2C). These results were associated with a significant decrease of left ventricle tissue fibrosis in LSF vs. NSF rats (Fig. 2D). Regarding the healthy control group fed a standard diet, mesenchymal markers (α-SMA, Vimentin) were significantly lower than in NSF rats, confirming that the high-fructose diet combined with angiotensin II infusion induces pathological EndoMT (Additional Fig. 2).

Development of chronic EndoMT cellular models

To explore the mechanisms underlying EndoMT-associated cardiac fibrosis in MetS, we developed in vitro models of primary human endothelial cells treated with TGF-β2. As endothelial cells differentially respond to TGF-β2 according to their origin [33], we used two cell models: human aortic endothelial cells (HAEC), which are highly responsive to TGF-β2, and human umbilical vein endothelial cells (HUVEC) which show lower responsiveness. We investigated the progressive loss of endothelial markers and expression of mesenchymal markers under TGF-β2 treatment by RT-qPCR. Endothelial markers PECAM1 and CADH5 (VE-Cadherin) mRNAs were not found significantly modified at the different times of treatment, in both types of endothelial cells (Fig. 3A). These results were confirmed at the protein level (Fig. 3C). However, NOS3 mRNA was found significantly less expressed in TGF-β2 treated cells in HAEC at all time points, in HUVECs at day 5 and a tendency to decrease at days 7 and 10, which was confirmed at the protein level (Fig. 3C). Therefore, endothelial markers were either unchanged or downregulated under TGF-β2 treatment.

Fig. 3.

Endothelial to mesenchymal transition in primary endothelial cells treated with TGF-β2. A, Expression of endothelial markers and B, Expression of mesenchymal markers in HAEC and HUVEC by RTqPCR. Experiments were performed in n = 3 independent biological replicates. One representative experiment is shown. (*p < 0,05 **;p < 0,01 ***;p < 0,001) C, Western blot showing expression of endothelial markers and D, mesenchymal markers. Experiments were performed in n = 3 independent biological replicates. One representative experiment is shown. (*p < 0,05 **;p < 0,01 ***;p < 0,001) E, Immunofluorescence showing coexpression of endothelial and mesenchymal markers (Blue: DAPI). Experiments were performed in n = 4 independent biological replicates, magnification x100. One representative experiment is shown. For A and B, exact p-values are indicated in additional file 1

We also assessed the expression of several mesenchymal markers such as SM22α, SNAIL, SLUG and COL1A1 (Collagen type 1 alpha 1 chain), which are target genes regulated by the canonical TGF-β pathway and implicated in EndoMT [9]. These markers were significantly upregulated by TGF-β2 at all time points in both HAEC and HUVEC (Fig. 3B), except for COL1A1, which was significantly increased in HAEC at days 7 and 10, but not expressed in HUVEC at all time points (Cq > 35), consistent with the lower responsiveness of this cell type. SM22α and SNAIL were also overexpressed at the protein level, confirming the data observed at the mRNA level (Fig. 3D). Moreover, phosphorylated SMAD2/3 proteins were increased in TGF-β2 treated cells, more markedly in HAEC, while total SMAD2/3 proteins levels remained unchanged (Fig. 3D), indicating activation of the canonical TGF-β pathway. Finally, immunofluorescence co-staining of PECAM1 and COL1A1 or SM22α was then performed to validate the transdifferentiation process in both cell types. In HAEC treated with TGF-β2, we observed coexpression of PECAM1 and COL1A1 or SM22α (Fig. 3E), while both mesenchymal markers were absent in untreated cells. Moreover, enlarged cells were observed, containing SM22α in the form of cytoskeleton fibers characteristic of fibroblasts [34]. In HUVEC, we observed coexpression of PECAM1 and SM22α but not COL1A1, showing partial EndoMT profile (Fig. 3E). Therefore, in human primary endothelial cells treated with TGFβ-2, features of endothelial to mesenchymal transition are present, with a greater degree in HAEC. Taken together, these results indicate different magnitudes of EndoMT between the two cell types.

Regulation of candidate genes in EndoMT models

We evaluated the expression of 4 candidate genes (ANXA2, LPHN1, CILP1, and FBLN5) in the 2 previous cellular models, at mRNA and protein levels in the presence or absence of TGF-β2. ANXA2 was not significantly modulated by TGF-β2 (mRNA, protein) during the course of EndoMT in the 2 models (Fig. 4A, B). LPHN1 was not modulated at the mRNA level and seemed induced at day 5 at the protein level but not after. CILP1 was found not expressed in the 2 cellular models (data not shown). FBLN5 was found significantly overexpressed under TGF-β2 treatment in both cellular models with fold changes of 10 to 20 in HAEC and 6 to 10 in HUVEC (Fig. 4A), throughout the experiment. This was confirmed at the protein level (Fig. 4B). By immunofluorescence, we showed that FBLN5 was coexpressed with PECAM1 in both cell types (Figs. 4C) and was also coexpressed with SM22α. FBLN5 was observed in the cytoplasm of fibroblast-shaped enlarged cells, surrounding the nucleus, while PECAM1 was expressed at the cell membrane. Taken together, these results show that FBLN5 is upregulated in endothelial cells undergoing EndoMT, with an expression pattern resembling that of mesenchymal markers. As FBLN5 was the only modulated candidate with relevance (mRNA and protein levels), we further explored its regulation in the process of EndoMT.

Fig. 4.

Expression of candidate genes during endothelial to mesenchymal transition. A, Expression in HUVEC and HAEC by RT-qPCR. Exact p-values are indicated in additional file 1. B, Expression in HUVEC and HAEC by western blot. Experiments were performed in n = 3 independent biological replicates. One representative experiment is shown (*p < 0,05, **p < 0,01, ***p < 0,001). C, Immunofluorescence showing coexpression of Fibulin-5 along with Pecam1 (endothelial) or SM22α (mesenchymal) marker after 7 days of TGF-β2 treatment (Blue: DAPI). Experiments were reproduced n = 4, magnification x100. One representative experiment is shown

Regulation of a candidate gene during the EndoMT process

To further explore the role of FBLN5 in EndoMT, we transduced HAEC and HUVEC with lentiviral vectors expressing FBLN5 or scramble shRNA, where EndoMT is induced by TGF-β2 treatment. In both models, we observed that cells transduced with FBLN5 shRNA showed an abolished FBLN5 expression (mRNA and protein, Fig. 5A, B) compared to control. Regarding endothelial markers, PECAM1 expression, which was found unchanged with TGF-β2 treatment, was significantly reduced at the transcriptional level in cells transduced with FBLN5 shRNA (Fig. 5C) but not at the protein level (Fig. 5D). CDH5 remained unchanged at the transcriptional level under all conditions tested (Fig. 5C). However, the significant NOS3 downregulation observed with TGF-β2 treatment was prevented by FBLN5 shRNA in both models (Fig. 5C, D). Thus, FBLN5 silencing selectively modulated endothelial marker expression.

Fig. 5.

Transduction with lentiviral particles expressing FBLN5 shRNA abolish Fibulin-5 expression and inhibit endothelial to mesenchymal transition in primary endothelial cells. A, RTqPCR and B, western blot showing downregulation of Fibulin-5 expression after lentiviral transduction and after 7 days of TGF-β2 treatment in HUVEC and HAEC. C, Consequence of Fibulin-5 abolition on EndoMT markers by RT qPCR, D, by western blot (one representative experiment is shown) and E, by immunofluorescence, after 7 days of TGF-β2 treatment. Magnification x60. Corrected Total Cell Fluorescence (CTCF) was calculated using Image J software and the method described at https://theolb.readthedocs.io/en/latest/imaging/measuring-cell-fluorescence-usingimagej.html. It was calculated for N = 30 cells per condition from a single representative experiment. Data are presented descriptively without statistical analysis. For A and C, exact p-values are indicated in additional file 1

Regarding mesenchymal markers, SM22α, SNAIL and SLUG overexpression under TGF-β2 treatment was prevented by FBLN5 shRNA in HUVECs (Fig. 5C). In HAEC, increased expression was prevented for SM22α, SLUG, but not for SNAIL (Fig. 5C). Moreover, COL1A1 expression was lost in the presence of FBLN5 shRNA in those cells. At the protein level, the TGF-β2-induced SM22α and SNAIL was abolished in both cell types (Fig. 5D). We also explored the impact of FBLN5 shRNA on both canonical and non-canonical TGF-β signalling pathways. FBLN5 knockdown reduced the phosphorylation of SMAD2/3, ERK1/2 and p38 MAPK phosphorylation compared to controls.

We confirmed that Fibulin-5 expression is strongly decreased in the presence of the shRNA compared to the control in TGF-β2 treated cells (Fig. 5E). The experiments also showed a drastic decrease in the number of SM22α positive cells displaying a fibroblastic morphology, in both cell types, when cells were treated with FBLN5 shRNA compared to control (Fig. 5E). The results were similar for PECAM1/SM22α labelled cells. Altogether, these results demonstrate that Fibulin-5 silencing prevented EndoMT features in both cell models at the mRNA, protein and morphological levels.

Regulation of Fbln5 in MetS rats

Finally, we explored Fibulin-5 expression by immunofluorescence in MetS rat aortas, and revealed the protein is localized in the intima and in the adventitia of NSF rats, while its expression was significantly reduced in LSF rats, particularly in the intima layer (Fig. 6A, B), in line with our previous results. Fbln5 was also expressed within the cardiac muscle, both in the extracellular matrix and in the vascular compartment, of left ventricles from all rats but diffuse labelling prevented signal quantification (data not shown). At last, we measured plasma Fibulin-5 by ELISA and showed that the protein is significantly less present in rats fed a low sodium diet (Fig. 6C). Together these results show a significant downregulation of aortic and plasma Fibulin5 in MetS rats fed the low sodium diet.

Fig. 6.

Expression of Fibulin-5 in MetS rats aortas. A, IF showing localization of the protein in the intima and adventitia (magnification x40) B, Quantification of Fibulin-5 integrated intensity within the intimal layer. Data are presented as mean ± SEM; statistical significance was assessed using an unpaired t-test (NSF vs. LSF) (N = 7 rats per group; p = 0.0115). C, Rat Fibulin 5 ELISA was performed on plasma samples. Data are presented as mean ± SEM; statistical significance was assessed using an unpaired t-test (NSF vs. LSF) (N = 8 rats per group; p < 0.0001)

Discussion

The aim of this study was to identify the mechanisms underlying the cardioprotective effects of a salt-restricted diet in the context of metabolic syndrome [25, 32]. Importantly, we show here that sodium restriction modulates a previously unrecognized mechanism, namely endothelial to mesenchymal transition contributing to cardiac fibrosis.

In our previous studies, we sought to develop an animal model representative of the abnormalities observed during the development of a complex and multifactorial disorder such as metabolic syndrome, closely mimicking human pathology. In this syndrome there is a clear activation of the renin-angiotensin-aldosterone system, leading to sodium sensitivity, vasoconstriction, and increased cardiac output [35]. We therefore combined a 60% fructose diet with angiotensin II infusion that mimics the activation of the renin-angiotensin-aldosterone system [36]. Therefore, the supraphysiological levels of circulating angiotensin II achieved by continuous infusion may amplify some of the observed effects induced by metabolic disorders. Importantly, in this model, sodium restriction only partially reversed the rise in blood pressure induced by angiotensin II (approximately 50%), and blood pressure in the low sodium group remained significantly elevated compared to normotensive controls [25]. Despite this persistent hypertension, cardiac hypertrophy and fibrosis were prevented in the low sodium group. Therefore, although sodium restriction only partially attenuated angiotensin II pathway activation, its beneficial effects strongly rely on additional mechanisms related to improvement of abnormalities associated with metabolic syndrome. This assumption is confirmed in rats fed the fructose diet alone where cardiac mass and fibrosis are completely normalized by sodium restriction in the absence of exogenous angiotensin II [30, 32].

Here, we show for the first time that sodium restriction attenuates cardiac fibrosis by reducing vascular EndoMT in a MetS model. In the literature, accumulating evidence suggests that dietary sodium contributes directly to tissue fibrosis by promoting epithelial to mesenchymal transition (EMT) or EndoMT. In animal models, high salt intake induces peritoneal fibrosis with EMT features, accompanied by TGF-β and IL-6 upregulation [37]. Similarly, in the kidney, sodium excess activates endothelial TGF-β signaling and drives glomerular and tubulointerstitial fibrosis through EMT mechanisms [38]. More recently, high-salt diets were shown to aggravate EndoMT-mediated glomerulosclerosis in Dahl salt-sensitive rats, with loss of endothelial markers and increased α-SMA expression [39]. In vitro, sodium excess also triggers inflammatory and profibrotic responses in mesothelial and endothelial cells, including decreased epithelial markers and induction of mesenchymal markers such as ACTA2 and SNAIL1 [40]. In parallel, pharmacological studies highlight the central role of sodium-related transporters in regulating EndoMT. Sodium-glucose cotransporter 2 (SGLT2) inhibitors such as dapagliflozin attenuate cardiac fibrosis in diabetic and non-diabetic models by suppressing EndoMT and fibroblast activation via AMPKα/TGF-β/Smad or SIRT1/Notch1 pathways [41, 42]. Similarly, SGLT2 blockade reduces EndoMT markers and impairs fibroblast proliferation and migration [43]. Moreover, SGLT1 has been implicated in atrial fibrosis through its impact on EndoMT in atrial fibrillation [44]. Together, these studies converge to demonstrate that sodium, either through dietary overload or via sodium-dependent transporters, plays a key role in promoting fibrosis through mesenchymal transition processes.

By confocal microscopy, we detected co-expression of Pecam-1 and α-SMA, a marker of activated fibroblasts. Consistent with our findings, Plawecki et al. reported increased Pecam-1 and α-SMA costaining in left ventricles of obese hypertensive and insulin-resistant rats, correlating with cardiac fibrosis [45]. However, the precise functions and contributions of these cells to cardiac fibrosis remain to be determined in the future, especially in the context of metabolic diseases where interstitial fibrosis causes cardiac dysfunction. Indeed, whether EndoMT-derived fibroblast-like cells contribute similarly to replacement or reactive fibrosis remains unknown. To go further, in vivo lineage tracing models is a useful approach to investigate EndoMT in acute cardiac diseases such as pressure overload or myocardial infarction [9], but chronic conditions such as metabolic syndrome remain largely unexplored. Since EndoMT contribution may differ according to the type and localization of fibrosis [1], as suggested by its significant involvement in atherosclerosis [15, 46] it would be particularly valuable to apply inducible lineage tracing models to chronic cardiac fibrosis in the context of metabolic syndrome.

Among the candidate genes identified in our transcriptomic analysis, we focused on four genes never described in the context of EndoMT: Cilp1, Annexin A2, Latrophilin-1, and Fibulin-5. Cilp1 is a cardiac matricellular protein produced by cardiac fibroblasts, upregulated in pressure overload-induced fibrotic remodeling and showing anti-fibrotic properties via TGF-β1 signaling interference [47–51]. Annexin A2 acts as a biphasic modulator of cardiac fibrosis, with protective effects in the acute post-MI phase and profibrotic contributions in chronic heart failure [52–55]. Latrophilin-1, an adhesion GPCR mainly studied in synaptic activity, has been linked to obesity-induced insulin resistance and hepatic steatosis [56], two known risk factors for secondary fibrotic heart disease. In contrast, Fibulin-5 is significantly upregulated in our cell models, suggesting a possible involvement in cardiac fibrosis. Fibulin 5 [57, 58] is a secreted extracellular matrix glycoprotein, predominantly expressed during embryogenesis in great vessels and cardiac valves. In adults, it is mainly found in tissues that contain abundant elastic fibers, including heart, lung, ovary and skin [57]. Fibulin-5 has previously been involved in other types of fibrosis [59–61] and in cancer [62]. In our model of MetS with cardiac fibrosis and hypertrophy, we showed downregulation of Cilp-1, Anxa2, Lphn1 and Fbln5 at the mRNA and/or protein levels in cardiac left ventricles, in correlation with a significant reduction of vascular EndoMT, supporting a role in this process.

To confirm the presence of EndoMT in our in vitro human primary cellular models, we explored the expression of endothelial markers (PECAM1, CDH5, NOS3) and mesenchymal markers (α-SMA, SM22-α, SNAIL, SLUG, COL1A1). Myofibroblasts are characterized by the expression of contractile proteins (i.e. α-SMA, SM22-α), and synthesis of MEC proteins. Moreover, diverse fibroblast populations involved in heart injury and disease have been identified by genetic lineage tracing [63] and single-cell RNA sequencing [64]. Different fibroblast types appear over time after myocardial infarction. A classification was proposed based on previous reports [34], associating different markers. Collagen type 1 is the common marker expressed by all 4 identified states: resting, expansion, activated, resolution. We identify production of type I collagen in vitro in transdifferentiating HAEC from day 5 to day 10. In transdifferentiating HUVEC, all mesenchymal markers cited earlier but type I collagen were expressed at all time points. Indeed, in the literature, cells of aortic origin (like HAEC) show the best response to induction by TGF-β2 treatment [33], whereas HUVEC cells show a delay or resistance to enter transdifferentiation [65]. These results confirm that our culture conditions support transdifferentiation towards a mesenchymal phenotype. These findings further confirm that endothelial cell origin influences the ability to undergo EndoMT [9]. While co-expression of endothelial and mesenchymal markers is widely used in the field to identify cells undergoing EndoMT, particularly in the context of partial or intermediate transition states, future studies should incorporate functional assays to fully confirm the transition process [66]. It should be noted however that our in vitro models recapitulate a partial or hybrid EndoMT phenotype, characterized by the acquisition of mesenchymal markers without complete loss of endothelial identity, consistent with an intermediate transition state as described in the literature [9]. Among the candidate genes evaluated in our EndoMT models, Cilp-1, Anxa2, and Lphn1 did not show consistent modulation, precluding further investigation of their role in this process.

In contrast, we report here for the first time that Fibulin-5 is implicated in vascular EndoMT both in a MetS animal model and in primary endothelial cells treated with TGF-β2. Importantly, silencing FBLN5 expression strongly inhibited EndoMT in vitro, reducing activation of canonical and non-canonical TGF-β pathways, fibrosis-related proteins, and morphological changes associated with transition. Specifically, FBLN5 knockdown reduced activation of Smad2/3, ERK1/2 and p38 MAPK in response to TGF-β, consistent with its ability to stimulate these pathways in other cell types [67]. Fibulin-5 is a TGF-β–inducible protein [67, 68] essential for elastogenesis, ensuring elastin deposition and fiber organization. Consistently, FBLN5 KO mice develop severe elastinopathy with loose skin, arterial tortuosity and emphysema [69, 70], while human cutis laxa mutations impair FBLN5 incorporation and elastic fiber assembly [71, 72]. Beyond its structural role, Fibulin-5 is a key regulator of ECM mechanical properties. Loss of Fibulin-5 has been shown to decrease tissue stiffness and to abrogate the fibrotic phenotype in a mouse model of cutaneous fibrosis, suggesting that it contributes to a profibrotic feedback loop driven by matrix stiffness, independently of TGF-β [73]. This is particularly relevant in the context of cardiac fibrosis, where ECM stiffening is known to amplify TGF-β signaling. Indeed, it promotes the release and activation of latent TGF-β, which synergizes with integrin signaling to amplify Smad-dependent transcriptional programs [74]. Fibulin-5 may therefore contribute to EndoMT not only through direct signaling, but also indirectly by modulating the mechanical properties of the pericellular matrix, thereby creating a permissive environment for TGF-β pathway activation. In this study, we implemented an original MetS animal model, i.e. 60% fructose combined with an angiotensin II infusion [25]. As angiotensin II is known to stimulate TGF-β expression and to activate both canonical Smad2/3 and non-canonical MAPK pathways, it is therefore possible that part of the Fibulin-5 upregulation observed in the NSF group reflects a direct response to angiotensin II stimulation rather than solely a consequence of metabolic syndrome-driven endothelial dysfunction. Importantly however, Fibulin-5 is itself a TGF-β-inducible gene [67], and its upregulation in response to angiotensin II would therefore be consistent with, and mechanistically integrated into, the TGF-β/fibrosis axis that we describe. The fact that low-sodium diet simultaneously blunts angiotensin II-induced blood pressure rise, TGF-β signaling, and Fibulin-5 expression suggests that these effects are interconnected within the same pathological pathway, regardless of whether the upstream trigger is exogenous angiotensin II or endogenous metabolic dysfunction. The present data therefore support an association and a functional requirement of Fibulin-5 in EndoMT-associated TGF-β signaling; future studies will be needed to establish the precise molecular mechanisms by which Fibulin-5 regulates this pathway.

Fibulin-5 also promotes endothelial adhesion through β1-integrins [57], and its expression is increased during EndoMT [65]. Integrins are cell surface receptors which act in both cellular adhesion and signaling. They have been implicated in the development of fibrosis in several organs, including the heart [75]. In addition to their direct effects on cellular proliferation, migration and survival, mediated by their binding to ECM proteins, integrins can potentiate signals from soluble growth factors such as TGF-β, and act as receptors for matricellular proteins [76]. Fibulin-5 serves as a ligand for cell surface integrins αvβ3, αvβ5, α9β1 on endothelial cells surface [69] and α5β1 and α4β1 on smooth muscle cell [77]. In HUVEC, it has been shown that ECM proteins (Fibronectin and Vitronectin) activate αvβ3 and α5β1, inducing different signaling pathways (Akt, ERK, JNK) [78]. Furthermore, αv integrins trigger spatially restricted TGF-β activation in fibrotic tissues through protease-dependent or independent actions involving conformational changes of the latent TGF-β complex [79]. Through this integrin-binding capacity, Fibulin-5 could therefore directly modulate integrin-mediated TGF-β activation and its downstream signaling cascades, including Smad2/3 and MAPK pathways.

Finally, previous work has linked Fibulin-5 to other forms of fibrosis [59–61] and to EMT in cancer [62], reinforcing its connection to profibrotic remodeling. Fibulin-5 may therefore influence common signaling pathways of epithelial- and EndoMT through at least two non-mutually exclusive mechanisms: a direct route via integrin-mediated signaling, and an indirect route via regulation of ECM stiffness and consequent amplification of TGF-β availability. As TGF-β is a major regulator of endothelial activities and a stimulator of Fibulin-5 expression, our data suggest the existence of a profibrotic autocrine loop in which TGF-β induces Fibulin-5, which in turn amplifies TGF-β signaling through integrin engagement and matrix stiffening, ultimately driving EndoMT and cardiac fibrosis. Future studies incorporating gain-of-function approaches and mechanical perturbation experiments will be necessary to fully delineate these mechanisms. Altogether, our findings identify Fibulin-5 as a potential mediator of vascular EndoMT contributing to cardiac fibrosis, linking its functions in elastogenesis, integrin binding, and TGF-β–dependent signaling.

While these findings provide novel insights into the role of Fibulin-5 in EndoMT-associated cardiac fibrosis, several aspects warrant further investigation. First, although plasma Fibulin-5 levels statistically correlated with cardiac tissue changes in our model, it cannot be excluded that circulating Fibulin-5 partly originates from other organs undergoing fibrotic remodeling in the context of metabolic syndrome, such as the liver, kidney, or vasculature. Future studies using organ-specific approaches, such as conditional cardiac knockout models will be necessary to formally establish the cardiac origin of plasma Fibulin-5 in this context. Moreover, although Fibulin-5 expression was quantified at both the mRNA and protein levels in cardiac tissue, immunofluorescence analysis in the left ventricle could not be reliably quantified due to diffuse labeling. In addition, measurements obtained from the thoracic aorta and plasma do not definitively demonstrate its localized upregulation or functional involvement within the fibrotic cardiac interstitium. Second, while our results support the occurrence of EndoMT based on marker co-expression and phenotypic changes, lineage tracing approaches could be another strategy to follow endothelial cell fate. Such experiments would allow to confirm the transition at the single-cell level and to better quantify the contribution of EndoMT-derived cells [9]. However, current lineage tracing models also present technical limitations, including promoter specificity and incomplete labeling. Future studies using refined lineage tracing strategies would therefore allow to extend our findings in vivo. Another limitation of the present study is the use of 2D monocultures of primary human endothelial cells, which lack the mechanical cues and cellular crosstalk, notably with macrophages and cardiomyocytes, present in the cardiac microenvironment in vivo. Future studies incorporating co-culture systems or three-dimensional cardiac models will be necessary to further validate the role of Fibulin-5 in EndoMT-associated cardiac fibrosis. Finally, the translational relevance of our findings needs to be addressed in relevant patients cohorts, to evaluate whether plasma Fibulin-5 levels could be correlated to cardiac function and fibrosis.

Conclusions

Our data show that vascular EndoMT may contribute to cardiac fibrosis associated with metabolic syndrome’s manifestations, through endothelial cell transdifferentiation, leading to loss of endothelial homeostasis and acquisition of a profibrotic, matricellular protein-secreting phenotype. Importantly, we demonstrate for the first time that low-salt diet reduces EndoMT in cardiac vessels in the context of metabolic syndrome. At the molecular level, we uncovered that EndoMT is modulated by Fibulin-5 in a TGF-β-dependent manner. These findings provide new insights into the protective effects of salt restriction against cardiac fibrosis through modulation of EndoMT and identify Fibulin-5 as a novel potential contributor to this process.

Supplementary Information

Below is the link to the electronic supplementary material.

Authors’ contributions

CD and ADL designed and supervised the research study. MM, SD and LJ performed immunofluorescence, immunoblotting, RT-qPCR experiments. CD and MM performed data analysis. CD, MM, SD, NG and BJ performed animal experimentation. MPB performed confocal microscopy images. The manuscript was written by CD with input from all authors. All authors have reviewed, edited, and approved the final version of the manuscript.

Funding

This research was supported by The French Society of Nutrition (Société Française de Nutrition), by the Fondation de l’Avenir (Grant number AP-RM-22-012), by the University of Montpellier (Programme d’excellence I-SITE, soutien à la recherche), by the French TTO SATT AxLR, by the french ministry of higher education and research (Ministère de l’Enseignement Supérieur et de la Recherche) and by the Algerian Ministry of Higher Education and Scientific Research (MM Ph.D. grant).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon request.

Declarations

Ethics approval and consent to participate

The experiments on rats were performed in accordance with the European and French laws (permit numbers B-3417226 and 34179, agreement D34-172-25) and conformed to the “Guide for the Care and Use of Laboratory Animals” published by the National Institute of Health (National Academies Press US, 8th edition, 2011).

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.