Why is endothelial resilience key to maintain cardiac health?

Restriction of oxygen and nutrient supply after myocardial infarction represents a challenge for all cells in the damaged tissue, leading either to cell death or induction of survival genes. Cell death is induced in cardiomyocytes and interstitial cells within 30 min post infarction. Cells close to the endocardium are more prone to injury, leading to propagation of cell death toward outer regions at the epicardium [103]. Moreover, I/R with fluctuations in oxygen levels lead to accumulation of ROS. Notably, the levels of ROS in ECs appears to determine whether survival or apoptosis genes are activated [110].

Loss of oxygen not only induces cell death, but activates and stabilizes the transcription factor HIF-1A, which induces survival genes and controls metabolic adaptation. Induction of HIF-1A augments glycolytic metabolism and leads to higher expression of angiogenic factors, such as ANGPT1 and ANGPT2 or VEGF [67, 126]. This initiates neovascularization, which will be discussed in “Angiogenesis and proliferation”.

Inflammation starts when cells of the innate immune system recognize danger-associated patterns, which lead to activation of tissue-resident macrophages and neutrophils and production of pro-inflammatory cytokines and chemokines, such as TGF-β, IL-1β, IL-6, TNF-α. Thereby, damage-induced matrix degradation, production of reactive oxygen species (ROS) and complement activation are the three major drivers in the initiation of immune responses [62, 63]. Interestingly, onset of inflammation largely depends on activation of Toll-like receptor 2 and 4 in the heart [9, 94].

The endothelium responds to the pro-inflammatory environment by supporting immune cell infiltration specifically via degranulation of Weibel–Palade bodies containing P-Selectin [90] and by induced expression of ICAM1 and VCAM1 [26] which promotes leukocyte adherence. ECs additionally produce cytokines and chemokines within 24 h after infarction [124], which fuels the inflammatory activation. TNF-α as well as other paracrine or soluble factors, induce endothelial activation, apoptosis and ROS production, whereas NO biosynthesis is reduced, overall leading to a destabilization of the endothelial monolayer [54, 56, 57]. Moreover, endothelial activation promotes the local adhesion of thrombocytes by von Willebrandt factor [12, 130]. Pro-inflammatory activation and increased thrombocyte adhesion were also shown to occur in remote vessels, such as larger arteries after myocardial infarct. These remote vessel alterations were attributed to systemic activation of the endothelium, leading to higher risk for atherosclerotic after infarction [85].

Recently, ECs were discussed to facilitate immune modulation beyond these classical functions in innate immunity. Evidence of a plastic conversion toward “immune cell-like” phenotype (EndICLT) has been shown on single-cell level when comparing static and disturbed flow conditions in a mouse carotid ligation model [6]. This proatherogenic phenotype was associated with increased expression of typical markers of macrophages, including major histocompatibility complex II (MHCII). Specialized endothelium, like sinusoidal ECs in the liver, has antigen-presenting functions, and can act as sentinel cells, sensing bacterial infections and cross-prime T cells to activation [15, 112, 121]. In the heart, the presence of MHCII-positive ECs with antigen-presenting activities has been described already in the 1990s [106]. In addition, mitral valve ECs were shown to express the hematopoietic marker CD45 post infarction [11], which suggests that these EC may undergo a hematopoietic transition.

The endothelial and hematopoietic fate are tightly linked. In evolution, the emergence of an EC monolayer is hypothesized to occur from specialized circulating blood cells, so called amoebocytes, which epithelized the basal membrane and form a primitive blood tissue barrier [87]. Higher vertebrate ECs share a common progenitor with cells from the hematopoietic system, the hemangioblast, which likewise gives rise to endothelial and hematopoietic progenitors [43, 145]. This explains not only why endothelial-specific signatures, such as PECAM1, CD34 or VEGFR2 [28, 87], are still also found in cells of the hematopoietic system, but also might be the reason that pro-inflammatory cytokines, which have pleiotropic functions on immune cell activity also influence endothelial plasticity. Notably, the transcription factor RUNX1, known to maintain expression of identity genes in hematopoietic stem cells, has been shown to be upregulated in ECs after infarction [82]. While its role in angiogenesis and tissue repair remains controversial, there is evidence that RUNX1 is not only crucial in the formation of hemogenic ECs in development but also has functions in adult heart ECs [19]. Finally, emerging concepts suggest that ECs may fully convert into an immune phenotype after prolonged activation in chronic inflammation [63]. These findings may imply that cardiac ECs are privileged to share functions with hematopoietic cells, allowing to combine forces with immune cells in events of early acute damage but also on the long-run.

EC may not only partially transition to a hematopoietic fate, but myeloid cells may acquire endothelial cell features. Here, the presence of circulating endothelial cells progenitors (cEPCs) has been debated for long time and hint to interesting links between the endothelial and hematopoietic system [123, 127]. Original research dated back to the 1990s and early 2000s showed the existence of bone marrow derived progenitor cells, which were circulating in the blood stream and were directed to injured vasculature [10, 20]. Difficulties in correct identification of these ECs and controversial results on the long-term incorporation into the vasculature questioned the role of cEPCs as true physical building block contributing to new vessels and suggested that they mainly may act as paracrine source to promote vessel growth [144]. In a model of wire-injured carotid artery transplants, virtually all EC regeneration was mediated by migration of adjacent healthy ECs toward injury [32]. Today, the self-renewal and differentiation capacity of ECs upon injury are discussed to be mainly mediated by specialized ECs rather than invading cells. Whether increased endothelial plasticity can lead to shedding of ECs from damaged vasculature to the blood stream, or activation of migratory phenotypes (as it is known in metastasis), which may result in circulating populations of endothelial cells, is not fully clear. Taken together, the contribution of endothelial plasticity to healing processes, such as clonal expansion and migration, could largely be explained by overlapping similarities with cells of the hematopoietic system.

The heart requires high amounts of energy and oxygen for ATP production. In fact, no other organ consumes metabolically so much energy as the heart. In homeostasis, the majority of ATP is generated via mitochondrial oxidative phosphorylation in cardiomyocytes. While the heart utilizes nearly all kind of energy sources, fatty acids are by far the most frequent [69]. Cardiac ECs have uniquely adapted to the organ demands as they express specific signatures of fatty acid transport, such as CD36 and fatty acid-binding proteins (FABP4, FABP5) [49, 95]. CD36 is required for endothelial up-take and transport of fatty acids to the neighboring cardiomyocytes. Partly, fatty acids are also retained and stored in ECs or used for own energy production. ECs, compared to other cells in the heart, have a relatively low number of mitochondria and do not require mitochondrial oxidative phosphorylation for ATP generation. This makes them more resistant to conditions of tissue ischemia and damage compared to cardiomyocytes [16]. Moreover, at high energy demands, ECs can utilize fatty acids, for DNA synthesis and cell division [62]. Under conditions of ischemia, the induction of HIF and oxygen-sensing mechanisms, such as eNOS or NADPH oxidases activity, augments glycolysis to maintain endothelial energy consumption in the low oxygen environment [7], while fatty acid metabolism is reduced [124]. This adaptation is mediated by various changes in the gene expression program leading to higher abundance of GLUT1 or PFKB3 [77] and repression of transcription factors such as FOXO1, which restrict glycolysis in quiescent ECs [136]. However, sufficient ATP generation in low oxygen levels is not the only advantage of a glycolytic program. Byproducts, such as glucose-6-phosphate, are fueled into the pentose-phosphate pathway to generate ribose-5-phosphate, a rate limiting source for nucleotides, which are required for proliferation. Additionally, hypoxia shifts glutamine metabolism from oxidation to reductive carboxylation [115]. In addition to the hypoxia and HIF-dependent changes in EC glycolytic and amino acid metabolism, genes encoding for fatty acid transport are among the strongest differential regulated genes in cardiac endothelial cells at the early stages after injury with the first 24 h infarction [124]. Loss of FABP4 and CD36 and induction of glycolysis-related genes is associated with increased plasticity and proliferation. In addition, loss of oxygen supply and high energetic requirements for angiogenesis promote de novo biogenesis and fusion of mitochondria [8, 91]. Increased demands for energy lead to ATP exhaustion, causing accumulating levels of AMP activates endothelial AMP-activated protein kinase (AMPK), which promote catabolic pathways in ECs in hypoxia after infarction [24, 137]. AMPK-mediated mitochondrial biogenesis, lipid metabolism and fat mobilization were shown to depend largely on SIRT1, another sensor of energy deprivation [75, 97]. Notably, in cardiac ECs, SIRT1 is known to be upregulated in ischemia induced neovascularization [100]. Downstream, SIRT1 represses FOXO1 and activates PGC-1A, which is known to induce mitochondrial biogenesis [14, 51]. De novo formation of mitochondria was associated with an anti-inflammatory phenotype, and suppressed activity of TNF-α and NFκB [51]. However, it is unclear whether and how mitochondrial biogenesis causally contributes to transient EC adaptation and state changes.

Reactivation of developmental pathways can induce phenotypical changes and promote endothelial-to-mesenchymal transition (EndMT). This process has initially been described during valve formation, when ECs acquire mesenchymal characteristics by following TGF-β gradients in the cardiac jelly [82, 83]. Here, TGF-β induction leads to a complete and permanent lineage transition, as these cells continue to be fibroblasts. In adulthood, tissue damage following inflammation and hypoxia modulates the perivascular space favoring similar, yet rather incomplete processes. The degree of transition to mesenchymal lineage after myocardial infarction is highly debated. Initially, ECs undergoing EndMT were thought to contribute to myofibroblasts and hence to fibrosis [141]. Recently, however, single-cell sequencing revealed that mesenchymal gene identity at day 7 post infarction is present in ECs but remains rather insignificant [64]. Parallel lineage tracing of endothelial and mesenchymal identities revealed no EndMT after myocardial infarction depicted by α-SMA and ZEB1 expression [142]. Further studies using more detailed time course analysis raveled that ECs have distinct transcriptional response program to TGF-β in very early phases after injury. This transient mesenchymal activation (EndMA) was reversible at later stages when TGF-β level decline, indicative of a temporal and incomplete process [124]. In contrast to EndMT, cells undergoing EndMA showed only modest up-regulation of mesenchymal genes and no complete transition to fibroblasts or other mesenchymal cells [64, 124]

How is this phenotype initiated? Several cytokines in combination with hypoxia initiate EndMA. TGF-β, produced from cells of the innate immune system in response to tissue damage is one of the most potent inducers. Cardiac macrophages selectively express MMP14, which activates latent TGF-β and induce paracrine signaling in ECs post injury [5]. Additionally, many other drivers of EndMA have been identified, including turbulent flow, hypoxia, glycolysis, WNT signaling, NOTCH signaling or reduction in FGF signaling. Interestingly, pro-inflammatory NFκB activation in EC, as it occurs upon IL-1β stimulation augments transition into a mesenchymal phenotype [76, 143] and induces generation of TGF-β1 and TGF-β2 [86].

ECs undergoing mesenchymal activation after infarction were more glycolytic than naïve ECs, which maintain oxidative phosphorylation and fatty acid transport [124]. In fact, reduced fatty acid oxidation by decreased levels of CPT1A has been identified as inducer of EndMT transition [139]. Increased levels of cytoplasmatic acetyl-CoA suppressed induction of EndMT, indicating that plastic transition toward mesenchymal lineage is inhibited in ECs which maintain normal ATP levels and citric acid cycle activity. In addition, genes regulating fatty acid metabolism and mitochondrial biogenesis such as SIRT1 are counteracting TGF-β induction in ECs [68]. In contrast, increased levels of MYC, a main transcription factor which virtually drives expression of all glycolysis-related genes, promotes EndMT under hypoxic conditions [2].

Another mediator of EndMT is mitochondrial dysfunction and oxidative stress. Elevated levels of ROS promote not only the activation of NFκB, but also suppress KLF4 which deregulates endothelial identity and promotes EndMT [88, 107]. Conversely, hydrogen sulfide, a potent inhibitor of EndMT and scavenger of ROS species was shown to suppress TGF-β signaling [119]. Elevated levels of H₂S are well known to induce an anti-inflammatory phenotype in ECs. This suggests that ECs, which are close to the injury site, are characterized by high level of ROS, glycolysis and mitochondrial dysfunction and share an inflammatory phenotype, which favors activation of TGF-β and hence EndMT. It seems reasonable that prolonged inflammation could lead to continuous transition to a complete mesenchymal phenotype and eventually fibrosis. Likewise, the reversible nature of EndMA after myocardial infarction might be a consequence of high endothelial resilience against inflammation and an active conversion to a naïve state. However, how ECs can switch back to the endothelial phenotype is poorly understood.

Single cell sequencing studies have consistently reported the expression of cardiomyocyte-specific gene expression signatures in ECs [64, 124]. While it is still debated whether these detected subpopulations are existing or may be the consequence of mRNA contamination during the single cells isolation procedure [1] some hints point to a possible activation of tissue-specific genes in ECs. Interestingly, Ribo-Seq of brain ECs revealed specific expression of synaptic genes, such as pleiotrophin [47]. In the heart, ECs show open chromatin of certain cardiomyocyte signature loci, express myofibrillar genes and cardiac-specific transcription factor MEF2C [70, 140]. Co-culture of human-induced pluripotent stem cell-derived cardiomyocytes with ECs induced MYL7 and MYL4 expression as well as NOTCH and BMP signaling in endothelial cells [35]. The functions of cardiomyocyte gene expression in healthy or injured endothelium however are unclear.

Neovascularization requires either sprouting angiogenesis of existing vessels or de novo formation of new blood vessels in the damaged area. Several cell types and origins have been discussed to infiltrate and expand in the tissue upon injury and contribute to new blood vessel formation.