Endothelial-to-mesenchymal transition: advances and controversies

Abstract

Endothelial-to-mesenchymal transition (EndMT) is a physiological process that is equally important during development and under certain pathological conditions in adult tissues. The last decade has witnessed a remarkable explosion of information about EndMT from molecular mechanisms responsible for its development to its role in various disease processes. The emerging picture is that of a complex set of interactions that underly pathophysiological basis of some of the most deadly and intractable diseases. This mini review brings together recent advances and attempts to present a unified view of this complex field.

The last ten years have witnessed an explosion of work in the endothelial-to-mesenchymal transition (EndMT) field. Once considered a research backwater with dubious claims to being a legitimate biological phenomenon, EndMT has gone to be seen as a common underlying pathophysiological basis, and, indeed, the driver of a large number of diseases characterized by ongoing chronic inflammation[1]. This brief review will focus on recent insights into the biology of EndMT and its rapidly evolving role in a number of disease states.

EndMT refers to the process by which normal endothelial cells completely or partially lose their endothelial identity, characterized by expression of certain “endothelial” fate markers such as vascular endothelial growth factor receptor-2 (VEGFR2), VE-cadherin and the like and acquire expression of “mesenchymal” markers such as α-smooth muscle actin (αSMA), SM22α, calponin, and collagens among many others [2–4]. The extent of these changes defines whether EndMT is classified as a complete (no expression of endothelial fate gene, high levels of expression of mesenchymal genes) or partial (decreased, but not totally absent endothelial fate gene expression). There are important caveats here that need to be born in mind in evaluating EndMT studies. First, EndMT cannot be effectively evaluated in vitro. It’s a slow process that takes a considerable amount of time and while cell culture studies are useful for evaluating certain mechanistic aspects of EndMT, in vivo studies are required for verification. Second, because EndMT can be complete (i.e. full disappearance of endothelial fate genes) any evaluation of its contribution requires endothelial fate mapping.

Molecular drivers of EndMT

Whilst there is a long list of stimuli and conditions that can induce EndMT, activation of endothelial TGFβ signaling is required for the cell fate transition to occur. This, in turn, requires an increase in endothelial TGFβ receptors (TGFβR1 and R2) expression as both are essentially not expressed by normal endothelial cells. [5,6] The key factor responsible for low TGFβR1/2 expression is a constitutive fibroblast growth factor (FGF) signaling that results in a high level of expression of let-7 miRNAs that inhibits TGFβR1 expression by inducing degradation of its mRNA [7]. This direct relationship between FGF and TGFβ signaling pathways was recently demonstrated in a hypoxic pulmonary arterial hypertension model. While hypoxic mice with endothelial-specific deletion of FGF receptors 1 and 2 activated endothelial TGFβ signaling and developed pulmonary hypertension, mice overexpressing a constitutively active FGFR in endothelial cells were protected both from activation of TGFβ signaling and pulmonary hypertension. [8] The importance of let-7 role in regulation of the FGF/TGF cross-talk was further confirmed by demonstration that the long non-coding RNA H19 that can directly control let-7 levels and, by extension, EndMT. [9] In addition to let-7, miRNA-20a has also been reported as an FGF signaling-dependent regulator of TGFβ signaling.[10]

One of key the factors regulating endothelial FGF signaling input and, hence, TGFβ signaling, is the presence of inflammation. A number of inflammatory cytokines, including TNFα, IL-1β and IFN-γ, are able to reduce endothelial FGFR1 expression with the extent of this suppression significantly magnified if two or more of these cytokines are present together [11]. It follows that a combined presence of inflammatory cytokines with TGFβ dramatically accelerates EndMT[12,13]. These in vitro observations have been confirmed in vivo both in animal models and in human disease samples [2,13,14]. In addition to inflammatory cytokines, disturbed shear stress also negatively affects FGFR1 expression thereby promoting TGFβ signaling while high shear stress inhibits cytokine-driven activation of this signaling pathway [15].

A combination of inflammation and abnormal shear is a potent EndMT-inducing milieu. Indeed, this combination is the key driver of atherosclerosis, as lipid accumulation in the arterial vessel all and disturbed shear combine to reduce FGFR1 expression in the endothelium lining high risk vascular areas (bifurcations and branch points, lesser curve of the aorta, etc.) leading to activation of TGFβ signaling. EndMT-dependent increase in expression of leukocyte adhesion molecules such as ICAM-1 and VCAM-1, leads to accumulation of leukocytes at EndMT sites thereby potentiating inflammation. This leads to a further decline in endothelial FGFRs expression, higher levels of TGFβR1 expression and, hence, more EndMT. This vicious circle of inflammation begetting EndMT begetting inflammation [1] likely accounts for the relentless progression of atherosclerosis even when some of the drivers (e.g. high cholesterol levels, hypertension) are controlled by medications.

It is important to point out that disturbed shear alone can induce EndMT even in the (initial) absence of inflammation. Indeed, disturbed shear-induced EndMT likely accounts for the development of cavernous cerebral malformations (CCMs) [16,17]. Disturbed (low) shear has been shown to directly activated Smad2/3 signaling [18] in a MEKK3-dependent fashion. [19] In addition, disturbed shear has been reported to induce expression of the transcription factor Snail [20] that mediates some of TGFβ signaling effects. Finally, disturbed shear can activate Alk5/Shc mechanotransduction pathway [21] and result in partial downregulation of Tet2, thereby leading to activation of Wnt/β-catenin signaling [22].

Another important point to consider is that EndMT can be partial or complete [3]. While a complete EndMT results in total loss of endothelial fate markers and the appearance of mesenchymal marker, a partial EndMT results in the of cells with intermediate endothelial/mesenchymal character. This partial EndoMT can represent a transient state on the road to the complete EndMT or maybe a transient process of its own. In particular, partial EndMT plays an important role in angiogenesis, whether developmental or adult, where a temporary and incomplete of “terminal” endothelial differentiation is necessary to activate endothelial proliferation. Mugraation and stalk/tip cell transitions [23].

EndMT and atherosclerosis

One of central features of atherosclerosis is the inexorable progression of disease once it sets it. A large number of clinical trials have demonstrated that while risk factor modification approaches such blood pressure- or cholesterol-lowering therapies can reduce the rate of progression, and perhaps somewhat favorable alter the composition of the atherosclerotic plaque, they do not induce meaningful regression of established lesions and reverse the disease. This resistance to regression has been attributed to the ongoing inflammation in the vessel wall. Critically, studies in human atherosclerotic tissues have shown a direct relationship between the extent of inflammation in the vessel wall, the extent of EndMT and the extent of activation of endothelial TGFβ signaling. [13] Yet, this link between vessel wall inflammation and EndMT and the nature of EndMT has not been fully appreciated and has not informed clinical trials.

Indeed, a number of clinical trials have attempted to affect the vessel wall inflammation by either broad inflammation inhibitors such as methotrexate and colchicine or individual cytokine inhibitors such as Il-β neutralizing antibody (CANTOS trail) or TNFα inhibitors [24–27]. Not surprisingly to connoisseurs of EndMT, all these attempts predictably failed. Since EndMT is initiated by physical factors such as disturbed shear that cannot be controlled and by a variety of inflammatory cytokines, attempts to block action of a single cytokine such Il-1β are doomed to fail as a number of other cytokines are still available and unfavorable shear dynamics are still operable. A systemic anti-inflammatory approach is equally unattractive since the amount of immunosuppression required would almost certainly result in significant side effects.

Recent advances in immunobiology of atherosclerosis point to a potentially critical roles played by immune checkpoint in regulation of vascular inflammation [28–30] and inhibition of certain checkpoints, such as CD40-CD40L dyad, has been proposed as a therapy for atherosclerosis [31]. Yet increased cardiovascular toxicity observed in cancer patients treated with checkpoint immunomodulators raises concerns about the safety of this approach [32,33]. Importantly, there is no information whether modulation of check points activity affects EndMT and whether EC undergoing EndMT promote deleterious check point activity. Nevertheless, this is potentially a far more promising approach then attempts at inhibiting any single inflammatory cytokine.

These considerations suggest that EndMT itself should be a target for anti-atherosclerotic therapies that seek to halt or reverse the disease. Sine TGFβ signaling is absent in normal endothelial cells and science its initiation is the primary driver of EndMT, it makes sense to target this process. The challenge lies in the fact that while TGFβ signaling is pathogenic in the endothelium, it is required for normal function of smooth muscle cells and, likely, other cell types. Indeed, a smooth muscle-specific knockout of TGFβ receptor(s) results in accelerated atherosclerosis and development of aortic aneurysms [34,35]. These observations strongly mitigate against any systemic approaches to inhibition of TGFβ signaling such as neutralizing antibodies or receptor tyrosine kinase inhibitors. On the other hand, endothelial-specific inhibition of TGFβ signaling looks promising. In mice models, targeted delivery of TGFβR1 siRNA using nanoparticles prevented development of atherosclerosis and induced a very significant (~60%) regression of established lesions. [36]

Non-atherosclerosis EndMT-related disorders.

While EndMT in atherosclerosis has received the lion’s share of attention, its presence underlies a number of other important disease states. Broadly speaking, any disease process that involves chronic vascular inflammation will have an element of EndMT and many, but not all, of fibrotic states are likely EndMT-related. One controversial area has been the role of EndMT in fibrosis associated with acute myocardial infarction. While initial studies suggested that a number of cardiac fibroblasts evident in the healing heart are of endothelial origin [37,38], subsequent studies cast doubt on that assertion [39,40]. Unlike the situation in the heart, however, it seems firmly established that EndMT plays an important role in generation of cancer-associated fibroblasts (CAFs) [41] and in development of systemic sclerosis (scleroderma) [42]. Scleroderma vasculature is characterized by the presence of chronic inflammation, loss of endothelial fate markers and increased expression of mesenchymal markers, all the features consistent with EndMT [43,44]. Furthermore, transcriptional analysis of skin biopsies from SSc patients demonstrated the presence of TGFβ/Smad gene signatures [45]. Whether EndMT is the consequence of the autoimmune process that is responsible for scleroderma or an important contributor to its pathogenesis and whether EndMT therapies would be clinically meaningful in this disease, has not been established.

Pulmonary arterial hypertension (PAH) is another condition where the existence of EndMT has attracted considerable attention. While PAH itself is highly heterogeneous and the name refers to a conglomeration of syndromes, the common morphologic basis underlying PAH is the presence of pulmonary fibrosis combined with abnormal smooth muscle cell (SMC) coverage of the distal pulmonary arterial tree. Combined, this results in narrowing of the distal pulmonary arterial vasculature, increase in pulmonary vascular resistance and, ultimate, PAH. While some of the SMCs covering the distal arterial bed are derived from the more proximal arterial media where they proliferate before migrating distally [46,47], others appear to be endothelial-derived, a feature strongly suggestive of EndMT. Furthermore, the presence of EndMT has been documented both in patient tissues and in a mouse model of hypoxic PAH induced by [48,49]. PAH further illustrates the importance of protective FGF signaling in preventing EndMT. Mice with endothelial-specific deletions of FGF receptors 1 and 2 developed much more severe PAH when subjected to hypoxia while mice with endothelial overexpression of FGFR1 were resistant to PAH development [8]. These results strongly support the idea of a primary pathogenic role of EndMT in PAH development. Another confirmation of this idea comes from a recent study demonstrating that a knockout of endothelial TGFβR1 reduces PAH induced by endothelial MEKK3-deletion dependent activation of TGFβ signaling. [19] These results advocate for a comprehensive study of EndMT in various forms of PAH.

EndMT 2023; where we are and where we go.

The last decade of work has clearly established EndMT as an important pathophysiological process responsible for the initiation and maintenance of chronic vascular inflammation. Importantly, EndMT is self-perpetuating: once it starts, the process feeds on it itself creating a perfect feed-forward loop. Indeed, one can remove the initial EndMT-initiating impulse and it will still continue[1]. This feature makes it not only the key driver of several important diseases such as atherosclerosis (and its consequences) and pulmonary hypertension but is also the main reason for the relentless progression of these illnesses.

If correct, the only hope of arresting and reversing EndMT-related disease lies in targeting the EndMT itself. This, in turn, requires endothelial-specific disruption of activated TGFβ signaling as systemic anti-TGFβ therapy carries significant risks. Indeed, a proof of principle study has demonstrated that such an approach arrests and reverses atherosclerosis in mice[36]. The next decade of work should establish if the same approach works in patients.