Objective Systemic sclerosis (SSc) features multiorgan fibrosis orchestrated predominantly by activated myofibroblasts. Endothelial-to-mesenchymal transition (EndoMT) is a transdifferentiation by which endothelial cells (ECs) lose their specific morphology/markers and acquire myofibroblast-like features. Here, we determined the possible contribution of EndoMT to the pathogenesis of dermal fibrosis in SSc and two mouse models.
Methods Skin sections were immunostained for endothelial CD31 or vascular endothelial (VE)-cadherin in combination with α-smooth muscle actin (α-SMA) myofibroblast marker. Dermal microvascular ECs (dMVECs) were prepared from SSc and healthy skin (SSc-dMVECs and H-dMVECs). H-dMVECs were treated with transforming growth factor-β1 (TGFβ1) or SSc and healthy sera. Endothelial/mesenchymal markers were assessed by real-time PCR, immunoblotting and immunofluorescence. Cell contractile phenotype was assayed by collagen gel contraction.
Results Cells in intermediate stages of EndoMT were identified in dermal vessels of either patients with SSc or bleomycin-induced and urokinase-type plasminogen activator receptor (uPAR)-deficient mouse models. At variance with H-dMVECs, SSc-dMVECs exhibited a spindle-shaped appearance, co-expression of lower levels of CD31 and VE-cadherin with myofibroblast markers (α-SMA+ stress fibres, S100A4 and type I collagen), constitutive nuclear localisation of the EndoMT driver Snail1 and an ability to effectively contract collagen gels. Treatment of H-dMVECs either with SSc sera or TGFβ1 resulted in the acquisition of a myofibroblast-like morphology and contractile phenotype and downregulation of endothelial markers in parallel with the induction of mesenchymal markers. Matrix metalloproteinase-12-dependent uPAR cleavage was implicated in the induction of EndoMT by SSc sera.
Conclusions In SSc, EndoMT may be a crucial event linking endothelial dysfunction and development of dermal fibrosis.
- Systemic Sclerosis
- Qualitative research
Statistics from Altmetric.com
Systemic sclerosis (SSc) is a complex connective tissue disease of unknown aetiology characterised by widespread peripheral microvascular injury evolving into progressive fibrosis of skin and multiple internal organs.1–3 In SSc, fibrosis results from an unrestrained tissue repair process orchestrated predominantly by activated myofibroblasts that are a population of mesenchymal cells displaying unique biological functions. These include an increased synthesis of fibrillar type I and III collagens, a reduction in the expression of genes encoding extracellular matrix (ECM)-degrading enzymes and α-smooth muscle actin (α-SMA) expression and incorporation into stress fibres, which provides an increased contractile force that is crucial for their tissue remodelling properties.4–6 Indeed, myofibroblast contraction contributes to a large extent to a progressive increase in connective tissue stiffness, a recently recognised potent profibrotic stimulus.7–10
Given the crucial role of myofibroblasts in the pathogenesis of organ fibrosis in a variety of disorders, considerable attention has been paid to the identification of their putative cellular origins. Hence, extensive investigations have revealed that profibrotic myofibroblasts may arise from different sources including expansion and activation of resident tissue fibroblasts and perivascular pericytes, recruitment of bone marrow-derived circulating precursors, transformation of white adipocytes and transdifferentiation of epithelial cells into mesenchymal cells.4 ,11–13 More recently, it has been reported with increasing frequency that vascular endothelial cells (ECs) may also exhibit substantial plasticity by undergoing endothelial-to-mesenchymal transition (EndoMT), a transdifferentiation by which ECs disaggregate, lose polarity and acquire ECM-producing myofibroblast features.14–16 EndoMT is a phenotypical conversion in which ECs downregulate the expression of their specific markers, such as CD31/platelet-EC adhesion molecule-1, von Willebrand factor (vWF) and vascular endothelial (VE)-cadherin, and acquire mesenchymal cell products including α-SMA, S100A4/fibroblast-specific protein-1 (FSP1) and type I collagen, together with stabilisation and nuclear translocation of the transcriptional regulator Snail1, a crucial trigger of mesenchymal transition.14–16
To date, EndoMT has emerged as a player in the pathogenesis of tissue fibrosis and fibroproliferative vasculopathy in various diseases, including diabetic nephropathy, cardiac fibrosis, inflammatory bowel disease-related intestinal fibrosis, portal hypertension and primary pulmonary arterial hypertension (PAH).14 ,16–21 Of note, extensive research studies have shown that multiple pathways implicated in SSc pathogenesis, such as transforming growth factor-β (TGFβ), endothelin-1 (ET-1), notch, sonic hedgehog and Wnt pathways, as well as other putative pathways such as oxidative stress and hypoxia, may participate in the molecular mechanisms of the EndoMT process.16 For instance, EndoMT can be fully induced by TGFβ in cultured ECs from different tissues.20 ,22–24
Although recent studies support the notion that EndoMT may participate in the development of SSc-associated interstitial lung disease (ILD) and PAH,25 ,26 the occurrence of such a phenotypical change from ECs to activated myofibroblasts has never been demonstrated in the affected skin of patients with SSc. Therefore, in the present study we combined ex vivo, in vitro and in vivo approaches to investigate the possible contribution of EndoMT to the pathogenesis of dermal fibrosis in SSc and two mouse models of the disease.
Materials and methods
An extended methods section is provided in the online supplementary material.
Cell culture and reagents
Primary cultures of dermal microvascular ECs (dMVECs) were established by explantation from biopsies of the lesional forearm skin from six patients with early diffuse cutaneous SSc (dcSSc; disease duration <2 years from first non-Raynaud symptom)27 and from six healthy adult subjects under protocols approved by the local institutional review board at the Azienda Ospedaliero-Universitaria Careggi (AOUC), Florence, Italy. Skin biopsies were processed as previously described.28 ,29 Patient characteristics are summarised in online supplementary table S1. Adherent cells were detached and subjected to CD31 immunomagnetic isolation by incubation with anti-CD31 conjugated-microbeads.28 ,29 Isolated cells were further identified as ECs by labelling with anti-factor VIII-related antigen (vWF) and anti-CD105, followed by reprobing with anti-CD31 antibodies (see online supplementary figure S1). dMVECs from healthy subjects (H-dMVECs) and patients with SSc (SSc-dMVECs) were maintained as detailed in the online supplementary material. In selected experiments, H-dMVECs were treated with recombinant human TGFβ1 (10 ng/mL; PeproTech, Rocky Hill, New Jersey, USA) or 10% serum from patients with early dcSSc (n=6) and healthy subjects (n=6) for 24, 48 and 72 hours. In some experimental points, sera were preincubated with the matrix metalloproteinase-12 (MMP-12) specific inhibitor MMP408 (10 nM; Sigma-Aldrich, St. Louis, Missouri, USA) before cell stimulation.
At the end of the experiments, cells were fixed with 3.7% buffered paraformaldehyde and immunofluorescence with antibodies against CD31, VE-cadherin, α-SMA, S100A4/FSP1, type I collagen and Snail1 (all from Abcam, Cambridge, UK) was performed as detailed in the online supplementary material. In some specimens, Alexa 488-labelled phalloidin (Invitrogen, Carlsbad, California, USA) was applied to the cells to visualise the arrangement of the F-actin cytoskeleton. For primary and secondary antibodies, refer to the online supplementary material.
RNA isolation and quantitative real-time PCR
At the end of the experiments, cultures were harvested and total RNA was isolated using the RNeasy Micro Kit (Qiagen, Milan, Italy). First-strand cDNA synthesis and mRNA quantification by SYBR Green real-time PCR were performed as reported elsewhere.29 For predesigned oligonucleotide primer pairs, refer to the online supplementary material.
Whole-cell protein lysates from dMVECs were subjected to immunoblot analysis as described elsewhere.29 For details on primary antibodies against CD31, VE-cadherin, α-SMA, S100A4/FSP1, type I collagen, Snail1, Friend leukaemia integration-1 (Fli1), urokinase-type plasminogen activator receptor (uPAR) domain 1 (D1) and domain 2 and α-tubulin, refer to the online supplementary material.
Collagen gel contraction assay
Collagen gel contraction assays were performed as described in the online supplementary material.
Levels of MMP-12 in serum samples were measured by quantitative ELISA as described in the online supplementary material.
Fluorescence immunohistochemistry on human and mouse skin
Paraffin-embedded sections of lesional forearm skin biopsies were obtained from 12 patients with SSc (n=4 with limited cutaneous SSc and n=8 with dcSSc) and 10 age-matched and gender-matched healthy donors, as described elsewhere.28–30 Skin sections from two mouse models of dermal fibrosis were also used. First, 6-week-old male C57BL/6 mice (Charles River Laboratories, Calco, Lecco, Italy) received subcutaneous injections of 100 μL of bleomycin dissolved in 0.9% NaCl (saline solution) at a concentration of 0.5 mg/mL every other day for 4 weeks in well-defined areas of the upper back. Subcutaneous injections of 0.9% NaCl served as controls.31 The second model consisted of 12-week-old male uPAR-deficient mice and wild-type littermates as described elsewhere.32 ,33 All animal protocols were performed in accordance with DL 116/92 and approved by the Institutional Animal Care and Use Committee of the University of Florence. Each experimental group consisted of at least six mice. Double-label immunofluorescence using antibodies against α-SMA and CD31 or VE-cadherin was carried out as detailed in the online supplementary material. The percentage of dermal vessels displaying CD31/α-SMA and VE-cadherin/α-SMA co-localisation was determined in five randomly selected high-power fields of the dermis from each of three sections per sample.
Transmission electron microscopy
Ultrathin skin sections from five patients with dcSSc and five healthy controls were processed and examined according to previously published protocols34 as detailed in the online supplementary material.
Statistical analyses were performed using the Statistical Package for Social Sciences software for Windows, V.20.0 (SPSS, Chicago, Illinois, USA). Data are expressed as means and SEM. The Student's t-test was used for statistical evaluation of the differences between two independent groups. A p value of <0.05 according to a two-tailed distribution was considered statistically significant.
EndoMT in dermal vessels of patients with SSc and experimental models of SSc
In order to determine ex vivo the presence of transitional EndoMT cells, skin sections from patients with SSc and healthy donors were subjected to double immunofluorescence staining for the EC markers CD31 or VE-cadherin and the myofibroblast marker α-SMA. In the healthy dermal microvasculature, α-SMA expression was mostly restricted to pericytes and vascular smooth muscle cells surrounding the endothelial layer (figure 1A). On the contrary, we observed co-localised CD31/α-SMA and VE-cadherin/α-SMA in the endothelium of numerous dermal capillary vessels and arterioles from patients with SSc, suggestive for cells in intermediate stages of EndoMT (figure 1A). Indeed, the percentage of vessels displaying CD31/α-SMA and VE-cadherin/α-SMA co-localisation was significantly increased in skin biopsies from patients with SSc compared with healthy skin (p<0.001 for both) (figure 1B). No difference in the frequency of transitional EndoMT cells was observed between SSc cutaneous subsets (data not shown). Furthermore, transmission electron microscopy analysis revealed that the presence of vWF-storing Weibel-Palade bodies was clearly reduced in SSc dermal endothelium (figure 1C).
Next we investigated in vivo the presence of transitional EndoMT cells in the skin of two mouse models of SSc, namely mice with bleomycin-induced dermal fibrosis and uPAR-deficient mice.31–33 The frequency of transitional EndoMT cells in murine skin was assessed by co-localisation of either CD31 or VE-cadherin and α-SMA. As displayed in figure 2, using both marker combinations we observed transitional EndoMT cells to be present at very low levels in saline-treated control mice, with significantly higher levels in the bleomycin treatment group (p<0.001 for both). Similarly, a significantly higher percentage of vessels with CD31/α-SMA and VE-cadherin/α-SMA double-positive cells was detected in the dermis of uPAR-deficient mice compared with wild-type littermates (p<0.001 for both) (figure 2A, B).
Cultured SSc-dMVECs co-express endothelial and mesenchymal cell markers and exhibit a myofibroblast-like functional phenotype
The expression of endothelial and mesenchymal cell markers in dMVECs isolated from forearm skin biopsies was investigated by immunofluorescence and immunoblotting. In agreement with previous reports,28 ,35 H-dMVECs exhibited a typical endothelial morphology with a polygonal shape, whereas the majority of SSc-dMVECs had an elongated shape often characterised by branches (figure 3A). Both H-dMVECs and SSc-dMVECs were immunopositive for the pan-EC marker CD31 (figure 3A). However, the expression of CD31 and VE-cadherin was markedly decreased in SSc-dMVECs compared with H-dMVECs (figure 3A). SSc-dMVECs also expressed α-SMA, which often was incorporated into stress fibres, as well as S100A4/FSP1 and type I collagen (figure 3A). On the contrary, as expected, in H-dMVECs there was no evidence of α-SMA and type I collagen expression, and S100A4/FSP1 was almost undetectable (figure 3A). Double immunofluorescence staining clearly revealed the unique presence of numerous CD31+ cells displaying α-SMA+ stress fibres in SSc-dMVEC cultures compared with H-dMVECs (p<0.001) (figure 3B). Phalloidin staining further revealed that while H-dMVECs showed a weak and disorganised expression of F-actin fibres, SSc-dMVECs exhibited a marked increase in stress fibres mainly organised longitudinally (figure 3A). Furthermore, we investigated the expression of Snail1, a zinc-finger transcription factor that induces numerous transcriptional events leading to the acquisition of a mesenchymal cell-specific phenotype such as stimulation of α-SMA expression.16 ,24 As displayed in figure 3A, strong expression and nuclear localisation of Snail1 were constitutively detected in SSc-dMVECs, while Snail1 expression was negligible in H-dMVECs. Immunoblot analyses confirmed either a significantly lower protein expression of CD31 and VE-cadherin or a significantly higher expression of α-SMA, S100A4/FSP1, type I collagen and Snail1 in SSc-dMVECs compared with H-dMVECs (p<0.001 for all comparisons) (figure 3C). According to the immunofluorescence data, both α-SMA and type I collagen were undetectable in protein lysates from H-dMVECs (figure 3C). Moreover, SSc-dMVECs exhibited a significant reduction in protein expression of Fli1 (p<0.001 vs H-dMVECs) (figure 3C), a transcription factor that plays a pivotal role in the maintenance of EC homeostasis and whose deficiency may be implicated in EndoMT.36–38 The occurrence of EndoMT was confirmed functionally by the evidence that SSc-dMVECs were able to effectively contract collagen gels (figure 3D).
Treatment with SSc sera induces a myofibroblast-like phenotype in H-dMVECs
Previous studies have demonstrated that treatment with sera from patients with SSc impairs the angiogenic performance of H-dMVECs in vitro.29 ,39 ,40 Nevertheless, whether these antiangiogenic effects may be in part related to the induction of the EndoMT process has never been investigated. To address this issue, H-dMVECs were challenged with sera from patients with early dcSSc and healthy subjects and subsequently assayed for changes in cell morphology and the expression of endothelial and mesenchymal cell markers. According to the literature,20 ,22–24 stimulation with recombinant human TGFβ1 was performed in parallel as a positive control of EndoMT. After 48-hour treatment with SSc sera, H-dMVECs started to disaggregate losing their characteristic polygonal cobblestone-like morphology (figure 4A). These changes progressed rapidly with the appearance of numerous cells exhibiting a spindle-shaped morphology in H-dMVEC cultures treated with SSc sera for 72 hours (figure 4A). As expected, similar findings were observed when H-dMVECs were challenged with TGFβ1, whereas H-dMVEC morphology did not change over time in cultures treated with healthy sera (figure 4A). Indeed, 72-hour treatment either with SSc sera or TGFβ1 induced a significant increase in the percentage of spindle-shaped cells (both p<0.001 vs basal H-dMVECs) (figure 4A), which were able to effectively contract collagen gels (figure 4B).
As displayed in figure 5, real-time PCR analysis revealed a significant reduction in mRNA levels of CD31, CDH5 and FLI1 genes in H-dMVECs treated either with SSc sera or TGFβ1 for 48 hours (all p<0.001 vs basal H-dMVECs). This happened in parallel with the induction of ACTA2, S100A4, SNAI1, COL1A1 and COL1A2 mRNA expression (all p<0.001 vs basal H-dMVECs) (figure 5). On the contrary, 48-hour treatment of H-dMVECs with healthy sera did not affect mRNA expression levels of the aforementioned markers (figure 5). These results were confirmed by immunoblot and immunofluorescence assessment of endothelial and mesenchymal protein expression levels in cells treated for 72 hours (figure 6A–G). In particular, both untreated cells and those treated with healthy sera showed no expression of α-SMA and type I collagen along with very low levels of Snail1, whereas treatment either with SSc sera or TGFβ1 induced the appearance of α-SMA+ stress fibres, de novo synthesis of type I collagen and strong expression and nuclear localisation of Snail1 (figure 6B–G).
MMP-12-dependent cleavage of uPAR is implicated in the induction of EndoMT by SSc sera
We previously demonstrated that in SSc-dMVECs uPAR undergoes a MMP-12-dependent cleavage of domain D1 resulting in impaired angiogenesis.35 ,41 Interestingly, the cleavage of uPAR-D1 was shown to be a crucial step in fibroblast-to-myofibroblast transition.42 Therefore, we herein investigated whether MMP-12-dependent uPAR-D1 cleavage could be implicated in the induction of EndoMT by SSc sera. Consistent with previous reports,40 ,43 MMP-12 levels were raised in SSc sera (see online supplementary figure S2A). Treatment of H-dMVECs with SSc sera resulted in uPAR-D1 cleavage already after 24 hours (see online supplementary figure S2B). Such a cleavage was instead prevented when SSc sera were preincubated with the MMP-12-specific inhibitor MMP408 (see online supplementary figure S2B). As shown in online supplementary figure S3, preincubation with MMP408 significantly blunted the effects of 48-hour treatment with SSc sera on gene expression of endothelial and mesenchymal cell markers.
Our data provide the first direct evidence that EndoMT may take place in the skin of patients with SSc and may have therefore a role in the pathogenesis of dermal fibrosis. The ex vivo immunohistological data clearly demonstrate the presence of transitional EndoMT cells simultaneously expressing EC and myofibroblast markers in SSc dermal microvasculature. In contrast, EndoMT was only observed at negligible levels in control skin. These results are substantially in agreement with similar findings recently described in the pulmonary vessels of patients with SSc-associated ILD and PAH.25 ,26 We have further characterised in vitro the phenotype of dMVECs isolated from SSc skin and found that these cells are in an intermediate state between an EC and a myofibroblast-like contractile phenotype, combining markers of both cell types. The results also show that H-dMVECs can undergo EndoMT in response to treatment with SSc sera, thus supporting the hypothesis that such cellular transdifferentiation may be operative in SSc. In fact, after a prolonged challenge with SSc sera, H-dMVECs lost their typical endothelial cobblestone appearance and acquired myofibroblast-like structural and functional features. Consistent with these morphofunctional changes, SSc serum-treated H-dMVECs exhibited a reduction in the expression of EC markers CD31 and VE-cadherin and an upregulation of mesenchymal markers, including α-SMA+ stress fibres, S100A4/FSP1, type I collagen and nuclear Snail1. Furthermore, the presence of transitional EndoMT cells in dermal vessels of two murine models of SSc is a matter of interest. Indeed, previous studies have demonstrated the occurrence of EndoMT in animal models of cardiac, pulmonary and renal fibrosis, as well as in models of PAH.16 ,23 ,26 ,44 ,45 Although our experimental data support the notion that EndoMT may contribute to the accumulation of myofibroblasts and the development of dermal fibrosis in vivo, this needs to be further confirmed by using lineage tracing in different preclinical models of SSc.
Besides the increase in the number of profibrotic myofibroblasts, EndoMT may favour microvascular derangement and loss of ECs contributing to capillary rarefaction, impaired angiogenesis and chronic tissue ischaemia in SSc skin. Indeed, endothelial dysfunction is considered a pivotal factor contributing to peripheral vessel remodelling in SSc.3 ,15 ,41 A defective response to proangiogenic stimuli and several functional defects, such as an impaired ability to organise into capillary-like tubes in vitro, have been extensively reported in SSc-dMVECs.28 ,29 ,35 ,41 ,46 Moreover, transcriptome profiling studies have revealed profound differences in the expression of genes encoding a variety of angiogenic regulators between SSc-dMVECs and H-dMVECs.41 ,47 In this context, our present findings shed light on EndoMT as a pathogenic mechanism that in SSc may directly link EC dysfunction to the development of dermal fibrosis. The intrinsic propensity of SSc-dMVECs to transition towards a profibrotic myofibroblast-like phenotype might in effect largely explain their well-known defective angiogenic behaviour. In addition, here we clearly demonstrate that a prolonged treatment with sera from patients with SSc is capable of sustaining the EndoMT process in H-dMVECs. Of note, shorter time treatments with SSc sera have previously been shown to impair angiogenesis and survival of H-dMVECs.29 ,39 ,40 Mechanistically, our present findings show that MMP-12-dependent cleavage of uPAR, a process that has been deeply implicated either in the impaired angiogenic performance of SSc-dMVECs or in fibroblast-to-myofibroblast differentiation,35 ,41 ,42 takes part in the pro-EndoMT effect exerted by SSc sera. Besides MMP-12, additional as yet unidentified circulating factors might trigger EndoMT and the loss of microvascular integrity in SSc dermis. Though further in-depth studies will be required, potential candidates include a large array of mediators that are elevated in SSc and have been demonstrated to induce EndoMT in vitro, such as TGFβ1, ET-1, tumour necrosis factor-α, asymmetric dimethylarginine and endostatin.16 ,19 ,26 ,48 ,49 Consistent with our in vitro observations, a recent study reported that sera from patients with chronic kidney disease induced EndoMT, decreased proliferation and increased apoptosis of human coronary artery ECs.49 These effects were mainly attributable to increased concentrations of circulating angiogenesis and nitric oxide inhibitors.49 Finally, when considering the autoimmune background of SSc, we cannot exclude the possible implication of functional (agonistic) autoantibodies against cell surface receptors in the EndoMT process. Indeed, a high proportion of patients with SSc display agonistic autoantibodies against the angiotensin II type 1 receptor and the ET-1 type A receptor, which can induce a variety of cellular responses such as production of TGFβ by dMVECs and synthesis of type I collagen by skin fibroblasts.50 Further mechanistic studies aimed at identifying key initiators of EndoMT in SSc are warranted.
In summary, our data collectively support the notion that EndoMT is a process occurring in the dermal endothelium of patients with SSc, where it may represent a crucial link between EC dysfunction and development of fibrosis. Hence, preventing or blocking EndoMT might be a novel and useful approach to treat peripheral microvasculopathy and prevent, at least in part, skin fibrosis in patients with SSc.
Handling editor Tore K Kvien
Contributors Study conception and design: MM, ER, LI-M and MM-C. Acquisition of data: MM, ER, IR, SG, SB-R, ADP, LI-M and MM-C. Interpretation of data: MM, ER, IR, LI-M and MM-C. Manuscript preparation: MM and MM-C.
Funding Supported by grants from the University of Florence (Progetti di Ricerca di Ateneo to LI-M and MM-C).
Competing interests None declared.
Patient consent Obtained.
Ethics approval The study was approved by the local institutional review board at the Azienda Ospedaliero-Universitaria Careggi (AOUC), Florence, Italy.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.