EndMT role in EC biology

3.1 EndMT process overview

EndMT is a complex biological process characterized by profound morphological, molecular, and functional changes in EC phenotype.^46–48 EndMT plays an integral role in embryonic development^14,49,50 and was first described as a locally confined process involving subsets of ECs during cardiogenesis and vasculogenesis.^12,13,49 These seminal studies showed that cardiac cushion cells, which are the precursors to cardiac valves, originate from a distinct subset of endocardial ECs that transitioned to mesenchymal cells via EndMT.^50,51 EndMT has also been shown to be involved in the maturation of pulmonary arteries.¹⁶

A growing body of evidence indicates that EndMT also plays a central role in tissue dysfunction in human adult pathologies, including cancer^{15,17,52–54} and a variety of cardiovascular diseases.⁵⁵ However, it has been reported that EndMT might be beneficial to some extent, being involved also in neovascularization, angiogenesis, tissue repair, and regeneration.^4,56–59 Overall, post-natal reactivation of EndMT can be considered as a potential mechanism for adaptation to new (patho)-physiological settings, where different roles are possible in virtue of the highly dynamic nature of EndMT.

Despite intense research, many aspects of EndMT biology remain poorly understood, as most of the mechanistic insight on this phenomenon originates from the study of epithelial-to-mesenchymal transition (EMT).^11,50,60 EMT is a well-studied process, known to be evolutionary conserved and induced by a broad spectrum of stressors, including cytokines, mechanical forces, and metabolic factors. While EMT is required for normal embryonic development, it can be hijacked in pathological conditions to facilitate tissue fibrosis and cancer metastasis.^46,60 Although our understanding of EndMT is more limited than for EMT, the endothelium is a specialized sub-type of epithelium, thus affording the possibility to extend some of the prior knowledge regarding EMT to EndMT.¹¹ In EndMT biology, one of the main limitations is the lack of a consensus of an exact molecular and functional definition. As the expression of EndMT-associated markers varies with time and environment, EndMT assessment and data cross-comparability can be extremely challenging.^20,43,61 Despite these challenges, EndMT is determined by an increased expression of mesenchymal markers and the concurrent decrease in endothelial ones at transcript/protein level.^8,10 EndMT hallmarks consist of the myogenic proteins α-smooth muscle actin (α-SMA) and transgelin (TAGLN/SM22α), non-myogenic fibroblast markers such as fibroblast-specific protein-1 (FSP-1), and fibrillary collagens type I and type III (COL1A1 and COL3A1), with many others being reported in context- or stimulus-specific responses. On the other hand, EC lineage-specific markers negatively modulated during EndMT include platelet–endothelial cell adhesion molecule-1 (PECAM1/CD31), von Willebrand factor (VWF), and vascular–endothelial cadherin (VE-cadherin/CDH5). As already appreciated from the study of EMT, the EndMT programme is also orchestrated by TFs such as twist-related protein-1 (TWIST), the Mothers Against Decapentaplegic Homolog (SMAD) family member 3 (SMAD3), the zinc finger E-box-binding homeobox 2 (ZEB2), and the Snail family transcriptional repressors 1 and 2 (SNAI1, SNAI2).^52,62–66 It is worth mentioning that studies reporting EndMT tend to focus on a customized and restricted subset of markers, making it challenging to validate data and signatures across studies, thus underscoring the issue of a consensual molecular definition.

EC morphology changes such as cytoskeletal remodelling, loss of cell–cell junctions, and loss of cellular polarity are broadly reported in the literature.⁵³ Evidence shows that during early stages of EndMT there is a decrease in intercellular adhesion forces accompanied by increases in cellular stiffness.⁶⁷ These changes cause ECs to lose their characteristic cobblestone-like morphology to acquire an elongated shape.⁶⁸ Following cytoskeletal re-arrangements, ECs gain proliferative, migratory, and invasive properties.^15,69,70  Figure 2 summarizes the major changes associated with EndMT.

3.2 EndMT dynamics

While initial descriptions of EndMT in the context of embryonic development supported the idea that the conversion of ECs to mesenchymal cells was permanent, recent evidence suggests that this process of cellular trans-differentiation is flexible. Indeed, it is now widely accepted that cells undergoing EndMT evolve through various intermediate stages, indicating that this process exists in a continuum and that ECs can remain in an intermediate stage and only undergo partial EndMT.^33,71,72

In partial/intermediate EndMT, both endothelial and mesenchymal features may be present, whereas in a more complete transition, cells are suggested to reach a mesenchymal state. In PAH, lineage tracing enabled the identification of both partial and complete EndMT states with distinct marker profiles.⁷³ In this study, the partial EndMT cells expressed endothelial progenitor cell markers such as prominin-1 (PROM1/CD133) and CD34, while fully transitioned cells coexpressed mesenchymal stem cell markers such as the stem cell antigen 1 (SCA1) and endoglin (ENG/CD105).⁷³ Of note, ENG is expressed in vascular ECs but it further increases in mesenchymal-like cells. As such, it is used as one of a number of mesenchymal markers when defining endothelial or EndMT identity, and it is evaluated in terms of relative expression changes as opposed to its presence or absence. In the same PAH study, cells that had undergone complete EndMT also showed enhanced proliferative and migratory capacity; however, there was a reliance on only a limited set of markers to identify such properties. A recent lineage tracing study in a myocardial infarction (MI) model reported that EndMT genes were enriched in regions of EC clonal expansion.⁵⁶ The authors described this as partial EndMT because clonally expanded ECs did not lose expression of endothelial genes. In other studies, single-cell transcriptomic and epigenomic data show a spectrum of intermediate phenotypes.^74,75 Cell subpopulations that coexpress endothelial and mesenchymal markers have been identified also in cardiac,⁷⁶ pulmonary,⁷⁷ and dermal⁷⁸ fibrosis, as well as in embryonic and adult valve ECs.^71,79,80 Importantly, it is not understood if the path from partial to complete EndMT is linear or if a group of cells undergoes partial EndMT while others have the capacity to undergo a complete transition. It is important here to also account for EC origins and their significant heterogeneity, which could predispose a certain group of ECs to be more or less likely to undergo EndMT. It is believed, however, that partial EndMT is not a distinct process but rather an incomplete activation and differentiation towards EndMT, where certain signals (or lack thereof) prohibit ECs from fully transitioning towards a mesenchymal phenotype.⁸¹

While in contexts such as those of embryonic development, EndMT appears to be complete likely due to high cell plasticity that makes ECs more responsive to the prevailing molecular cues. Alternatively, complete EndMT may result from chronic activation of the EndMT signalling network which drives cells through their intermediate states and towards a robust transition to mesenchymal cells. An example of such plasticity are EndMT-derived mesenchymal cells of the endocardial cushions in the forming embryonic heart.^50,58,82 In angiogenesis, however, it is proposed that EndMT is partial and that there are regulatory cues in place that inhibit progression of the EndMT programme, which supress a complete transition towards a mesenchymal phenotype.^83,84 Whether this transient state leads to a complete acquisition of mesenchymal cell fates or is resolved seems to be dictated by specific intracellular milieus and pathological stressors. Furthermore, a recent study from Tombor and colleagues⁵⁷ reported an acute transient mesenchymal activation of ECs post-MI, where ECs adopt a mesenchymal signature within 7 days of MI and return to baseline endothelial identity by 14 days of MI, as opposed to fully committing to a mesenchymal cell fate. This transiency, defined as endothelial-to-mesenchymal activation (EndMA), may facilitate regeneration and does not appear to play a causal role in later disease stages. Notably, additional in vitro work showed that the removal of conditioning stimuli resulted in reversal from a mesenchymal-like state in cultured ECs. Consistently, independent research showed that decreased expression of receptors involved in transforming growth factor-β (TGF-β)-mediated responses (e.g. ALK2 and ALK5), as well as EndMT-associated TFs such as SNAI1, results in a comparable repression of EndMT in EC cultures.⁸⁵ These observations open new possibilities for the investigation of EndMT reversibility in relevant pathologies. Further studies are warranted to determine the precise mechanisms at play, especially in situations where EndMT acts chronically such as in atherosclerosis.

3.3 Integrated signalling pathways and contribution to EndMT

The EndMT programme is orchestrated by a variety of biochemical, biomechanical, and environmental signals and then executed by various signalling cascades together with TFs. In disease, it is advanced by acute/chronic environmental changes, including but not limited to inflammation, hypoxia, and alteration in biomechanical forces as reviewed elsewhere.^85,86 In particular, inflammation is an intrinsic component of EndMT wherein the inflammatory response is mediated by two key cytokines: interleukin-1 beta (IL-1β) and tumour necrosis factor alpha (TNF-α).

The activated signalling cascades up-regulate a set of TFs promoting acquisition of mesenchymal cell states and EC identity loss. While TGF-β signalling is considered a central aspect of the EndMT programme,⁸ additional pathways also contribute. As EndMT consists of multi-step fate changes, it is finely regulated in differential and sequential manners via inter-connected molecular pathways.^87,88 In addition, EndMT-associated TFs (e.g. SNAI1 and 2, ZEB1 and 2, and TWIST) can contribute to EndMT being at the intersection of different signalling cascades. However, their action is postulated to be non-redundant and context-dependent as in EMT-associated disorders.⁸⁹ Therefore, distinct ECs may display different TF modulation in terms of preference and timing. Although the precise molecular mechanisms and determinants governing EndMT, at a higher level, remain largely unknown, herein, we discuss the central role of the TGF-β pathway and its independent or synergistic action with other genetic and epigenetic mechanisms implicated in EndMT, with a focus on cardiovascular biology (respectively Figures 3 and 4). Causality of these biological processes in EndMT and cardiovascular pathologies is not discussed in detail as current knowledge about their precise contribution is still limited.

The TGF-β superfamily is the major modulator of EndMT processes and comprises an array of responses from direct TGF-β-mediated ones to those depending upon other family members including bone morphogenetic proteins (BMPs) and activins, among others. The canonical TGF-β signalling pathway is known to be a key driver of EndMT.^{9,33,40,42,90–92} While it is essential for correct embryo development, in the adult, EndMT is mostly initiated under pathophysiological cues that lead to aberrant hyperactivation of the TGF-β pathway.⁹³ Briefly, TGF-β ligands bind to type I and type II receptors (TGF-βRI and TGF-βRII), which upon heterodimerization and phosphorylation, activate regulatory SMAD2/3 and their intracellular cascades. SMAD2/3 then translocate into the nucleus to form transcriptional complexes that in turn mediate targeted gene activation/repression. Ultimately, the activation of this intracellular cascade regulates several processes including cell differentiation, adhesion, proliferation, migration, and apoptosis that altogether participate in EC activation and transition towards a mesenchymal-like state.¹⁹ The use of genetically modified mouse models with knockout of TGF-βRs markedly reduces EndMT and kidney fibrosis.⁹⁴ Additionally, in a hyperlipidaemic mouse model of atherosclerosis, TGF-β receptor knockout attenuated EndMT and reduced vessel wall inflammation and permeability with a subsequent arrest of disease progression and regression of the established lesions.⁴² TGF-β also promotes tissue fibrosis by ECM deposition and scavenging, and via the positive modulation of fibrogenic genes such as fibronectin and some collagen species (e.g. COL1A1 and COL3A1).⁹⁵ Finally, MEKK3 is a known mediator involved in EndMT promotion in cavernous cerebral malformations as a result of CNN impairment.²³ It has also been shown to play an unexpected role in cardiovascular disease by acting downstream of the TGF-β pathway. In more detail, MEKK3 disruption in the endothelium is involved in inward remodelling and promotes EndMT by supporting TGF-β and SMAD2/3 signalling, also correlating with more unstable and vulnerable atherosclerotic plaques.⁹⁶ As an alternative, TGF-β non-canonical pathways act in a SMAD2/3-independent manner, signalling via mediators like the mitogen-activated protein kinase (MAPK) family of serine/threonine-specific protein kinases, phosphatidylinositol 3-kinase (PI3K), p38 MAPK, Jun amino-terminal kinase (JNK), and ubiquitin ligase TNF receptor–associated factor 6 (TRAF6), among others. Transduction of non-canonical signalling cascades occurs in a context-dependent manner to fine-tune specific biological processes.⁵⁸ While there is no compelling evidence to date, it is conceivable that non-canonical and canonical responses may alter SMAD-dependent signalling, thereby contributing to shape the final impact of TGF-β on the system.

BMP-ligand interactions and receptor binding mediate signal transduction via SMAD-dependent and SMAD-independent pathways. This is important, as balance in the BMP/TGF-β axis seems to help maintain cell homeostasis by preventing excessive TGF-β sensitization.^93,97 BMP signalling has been implicated in osteogenic differentiation and mineralization,^98,99 which are consistent with EndMT. For instance, the down-regulation of the BMP type II receptor (BMPR2) is a key event in EndMT activation and vascular calcification.⁹⁹

Interestingly, similar to the TGF-β canonical pathway, activins induce a TGF-βRI activation thus triggering SMAD2/3-dependent responses. In PAH, activin A has been shown to promote EndMT by acting as a ligand for BMPR2 and targeting it for degradation.¹⁰⁰

Lastly, the TGF-β pathway is tightly interconnected to other EndMT signalling cascades that sustain or halt its action, as described below.

In embryo development, canonical wingless-related integration site (WNT) signalling is required for EndMT in the endocardial cushions of the developing heart.⁴⁷ In contrast, in adult life, pathway modulation has also been linked to EndMT in a variety of aberrant tissue remodelling scenarios.^92,101 In healthy endothelium, the WNT canonical pathway is inactive and expression levels of its primary effector, namely β-catenin, are negatively regulated by the destruction complex (APC-GSK3-Axin-CK1). Upon activation, frizzled (FZD) receptors bind to their coreceptor, low-density lipoprotein (LDL) receptor–related protein 5/6 (LRP5/6), to trigger β-catenin intracellular changes. After translocation to the nucleus, β-catenin promotes LEF/TCF-mediated transcriptional regulation. A post-MI study demonstrated that WNT pathway activity contributes to cardiac fibrosis by expanding αSMA-positive myofibroblasts.¹⁰¹ This is in line with increased expression of markers like SNAI2.¹⁰² Additionally, an independent study showed that inhibition of WNT signalling via the secreted frizzled-related protein 3 (sFRP3) blocks EndMT in mitral valves post-MI. sFRP3 plasma levels were negatively correlated with the size of MI and positively correlated with ejection fraction in sheep with MI.¹⁰³ When promoting EndMT, the WNT and TGF-β signalling cascades can synergize via β-catenin that upon translocation interacts with LEF/TCF and forms SMAD-associated transcriptional complexes to transcriptionally regulate shared target genes. However, WNT3a and WNT7a have an opposite effect on EndMT.⁴⁸

NOTCH signalling contributes to EndMT initiation in developmental and pathological conditions,^83,84 acting independently or in concert with the TGF-β pathway.¹⁰⁴ NOTCH signalling is initiated by ligand binding, which induces proteolytic cleavage of the transmembrane receptor and release of the NOTCH intracellular domain (NICD). NICD then translocates into the nucleus and associates with CBF1/Suppressor of Hairless/Lag1 (CSL), thereby recruiting coactivators to regulate transcription. NOTCH can directly activate TWIST1 expression and facilitates the recruitment of SMAD3 to SMAD-binding elements in the promoter of SNAI2 and other mesenchymal genes.^87,105 Additionally, Jagged1 (JAG1)-induced activation of NOTCH signalling results in the nuclear accumulation of the mesenchymal transcription factor RUNX3 that induces the expression of several mesenchymal genes.^104,106

Ligands of the fibroblast growth factor (FGF) family bind to FGF receptors (FGFRs), promoting their dimerization and trans-phosphorylation of specific tyrosine residues in cytoplasmic kinase domains, thus triggering their activation.⁴⁸ The additional constitutively docked adaptor protein FGF receptor substrate 2 alpha (FRS2α) generates docking sites for numerous cytoplasmic proteins. As a result, it can trigger an array of mediators and downstream signalling cascades, including phospholipase C gamma (PLCγ), PI3K/AKT, and Ras/MAPK responses.^47,91 FGF2 (or basic FGF, bFGF) participates in the maintenance of EC functionality by hindering TGF-β-mediated responses.^107,108 Accordingly, further research has also demonstrated that impaired signalling activity either at the receptor level (FGFR1) or of intracellular transduction (FRS2α) can lead to induction of TGF-β signalling and EndMT as assessed by increased neointima formation in transplant arteriopathy and atherosclerosis models.^90,91 Mechanistically, active FGFR1 recruits FRS2α, which then induces the expression of microRNA (miRNA, miR) let-7 and suppresses TGF-βRI expression.⁹⁰ Interestingly, FGFR1 expression levels in ECs are negatively affected by pro-inflammatory stimuli, including TNF-α and IL-1β and interferon-γ (IFN-γ), which further impair FGF.¹⁹ Reportedly, a similar miRNA/mRNA axis has been described for miR-20a.¹⁰⁷ FGFR1/2 impairment was also shown to increase TGF-β signalling and EndMT in response to hypoxic stimulation both in cultured ECs and in a murine model of pulmonary hypertension.¹⁰⁹ Finally, FGF has been postulated to play an important role in the maintenance of the vascular endothelial growth factor (VEGF) pathway by supporting VEGF receptor 2 (VEGFR2) expression, which is an important determinant of EC identity through FGF-Ras-MAPK signalling.⁴⁸

3.4 Epigenetic orchestration of EndMT

Epigenetic mechanisms can guide gene expression by regulating transcription independently of the genetic code.^22,110 The term epigenetics refers to those heritable changes that, rather than relying on alteration of the DNA sequence, influence chromatin structure and thereby gene expression. Epigenetic changes are principally based on chemical alterations of DNA (e.g. methylation), post-transcriptional modification of histone proteins (e.g. methylation, acetylation, and phosphorylation), or specific modifications driven by non–protein-coding RNAs (ncRNAs).^110–112 An increasing number of reports have implicated epigenetics in EndMT, thus adding another layer of regulation to the mechanisms facilitating or inhibiting EndMT biology. As shown in Figure 4, here, we aim to provide an outlook of such epigenetic mechanisms; however, this topic has been dissected in depth by recent work.^113–115

3.5 Epigenetic marks

The class I histone deacetylase (HDAC) member termed HDAC3 and the enhancer of zeste homolog 2 (EZH2) play an important role in EC function and in the EndMT programme, both during development and post-natally.^48,116,117 Briefly, EZH2 is recruited by HDAC3 to promote deposition of tri-methylation marks on the lysine 27 residue of histone 3 (H3K27me3), thereby mediating negative transcriptional regulation of EndMT by repressing TGF-β1 during cardiac development.¹¹⁶ Additionally, loss of EZH2 and impaired H3K27me3 deposition at the promoter of TAGLN results in EndMT promotion.¹¹⁷ Further exploring the link between HDACs modulation and EndMT, Lecce and colleagues³⁴ showed that HDAC9, a class IIa HDAC, positively regulates EndMT both under cytokine-evoked stimulation in cultured ECs and in a murine model of atherosclerosis. HDAC9 exerts its role in EndMT by repressing H3K9 and H3K27 acetylation and preventing the increase in H3K27 methylation. Accordingly, the histone demethylase jumonji domain-containing protein 2B (JMJD2B) is induced by EndMT in response to pro-inflammatory or hypoxic stimuli, advancing EndMT programme activation. While its disruption affects EndMT in an MI model, further studies are needed to define its potential benefits in heart disease.¹¹⁸ As another example, the epigenetic reader bromodomain-containing protein 4 (BRD4) has also been described as an EndMT inducer. Mechanistically, it supports EndMT-associated TFs by promoting enhancer occupancy and activation of SMAD-dependent TGF-β responses.^119,120

3.6 Non-coding RNAs

A wealth of research has validated the importance of ncRNAs in different pathologies, including those affecting the vasculature,¹¹⁰ the endothelium, and EndMT.^22,114,121 Based on their size, ncRNAs can be classified as miRNAs and long-ncRNAs (lncRNAs), ranging from ∼22 nucleotides in length to >200 nucleotides, respectively. At present, knowledge on lncRNA contributions to EndMT remains limited. However, among the TGF-β-responsive miRNAs studied in the context of EndMT, many have been implicated including miR-21, miR-27b, miR-155, miR-20a, and let-7.^{90,107,114,115,122}

The metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) is a lncRNA that activates EndMT via the canonical WNT pathway and regulates the miR-145/TGFβR2/SMAD3 axis. Another lncRNA of interest is H19, as its role has been previously reported in other cardiovascular conditions, including endothelial ageing, mineralization of aortic valves, ischaemia/reperfusion-evoked apoptosis, and cardiac hypertrophy.¹²³ Mechanistically, H19 is activated in ECs by TNF-α to promote TGF-β signalling transduction via H19/TET-1/let-7 axis modulation.¹²⁴ An additional study has reported a role for H19 in oxidized LDL-induced EndMT via modulation of the H19/mir-148b-3p/ELF5 axis.¹²⁵ The lncRNA GATA binding protein 6-AS (GATA6-AS) is induced by oxidized LDL and suppresses TGF-β2-induced EndMT in vitro by targeting lysyl oxidase-like 2 (LOXL2) and regulating its impact on angiogenesis.¹¹⁵ Concordantly, recent work from Monteiro and collaborators³⁵ has shown that lncRNA MIR503HG is crucial in maintaining EC identity and function. In more detail, MiR503HG interacts with polypyrimidine tract binding protein 1 (PTBP1) to broadly prevent the execution of the EndMT programme, both in vitro and in vivo. Interestingly, its mechanism of action is independent from miR-424 and miR-503, which are located in the same genomic locus and previously described in EMT.^126,127

Collectively, a better understanding of the genetic and epigenetic determinants that govern EndMT at a higher level is essential as it may permit a crystallizing of knowledge around key steps in the initiation, advancement, and execution of this programme. This may promote a shift in the current state-of-the-art, creating the necessary knowledge for the development of novel therapeutic approaches.

Source: https://rb.gy/or371e