Introduction

Until recent decades, the long-standing view of development, whereby cells lose their differentiation potential throughout development and developed cells have very limited capabilities to rejuvenate or become other cell types, had discouraged attempts to generate pluripotent/multipotent stem cells from somatic cells for the purpose of replacing damaged tissues or failing organs, an ultimate goal of biomedical research [1]. Additionally, the idea that replenishing defective tissues or organs in vivo through transdifferentiation from other cell types with a relatively large abundance such as fibroblasts could be a testable option for replacement therapy had been largely denied. However, recent studies have revealed the possibility of changing already determined cellular identities to those of other cell types, in a targeted manner. Overexpression of transcription factors can induce pluripotency in somatic cells (i.e., induced pluripotent stem cells, iPSCs) [2] or directly change one cell type into another (direct reprogramming or direct conversion) [3, 4]. The latter has several advantages over iPSCs in that it is simple, fast and has very low risk of tumorigenesis [4].

Transcription factors are proteins regulating a wide array of target genes to govern certain developmental pathways (e.g., endothelial lineage specification) or biological events (e.g., inflammation). In particular, pioneer transcription factors such as OCT4, FOXA, ASCL1 and PU.1 can open closed chromatin and induce the expression of target genes in ectopic cells, forcing them to acquire another identity [5]. In endothelial cells (ECs), ETV2 (also known as ER71), a member of the E26 transformation-specific or E26-specific sequence (ETS) transcription factor family, is considered a pioneer factor [6]. Upon overexpression, ETV2 can initiate the de novo generation of ECs from mouse embryonic stem cells (mESCs) [7, 8], and ETV2 alone is sufficient to directly convert fully differentiated non-ECs to functional ECs [9, 10]. Despite its promising outcomes, the translation of the therapeutic potential of ETV2 into clinical use for vascular diseases faces several key hurdles. There is a lack of a useful mode of delivery of ETV2 for clinical use with non-viral, non-genetic, non-inflammatory methods, as well as efficient and specific in vivo delivery tools, and the heterogeneous nature of reprogrammed ECs hinders its application. In the first part of this review, we describe the functions of ETV2 in vascular system establishment in developing embryos. Next, we discuss direct cell reprogramming mediated by ETV2 and its potential therapeutic functions in vivo. In the last section, we focus on future directions for the use of ETV2 as a therapeutic vehicle considering its clinical applicability.

Critical role of ETV2 in establishing the cardiovascular system in developing embryos

The ETS transcription factors are known to play critical roles during EC specification and the establishment of the cardiovascular system [11]. We and other laboratories have identified Etv2 (also known as Er71 or ets1-related protein, etsrp) as a novel ETS member with an irreplaceable role in embryonic vessel formation [7, 8, 12,13,14]. The significance of Etv2 in vessel development was first reported in a zebrafish study using a cloche mutant, with no manifestations of blood, endothelial and endocardial cell formation [12]. Microarray analyses between normal zebrafish and cloche mutants revealed several differentially regulated endothelial-specific genes, among which etsrp was identified as a novel endothelial gene. The knockdown of etsrp via morpholino (MO) injection resulted in a complete loss of blood vessel formation and marked reduction in mRNA expression of endothelial markers. In contrast, the forced expression of etsrp mRNA in cloche mutants led to a restoration of critical endothelial genes (flk1, scl, fli1 and cdh5), indicating that etsrp is essential for establishing embryonic vasculature [13]. The critical function of Etv2 in early endothelial lineage was corroborated in mammals as demonstrated by Etv2-knockout mice and mouse embryonic stem cell (mESC) differentiation approaches [7, 8, 14, 15]. Similar to the findings observed with etsrp MO in zebrafish, Etv2-deficient mouse embryos have no detectable hematopoietic or ECs, resulting in early embryonic lethality between embryonic day (E) 9.5 and E10.5. Furthermore, Etv2 overexpression in differentiating mESCs led to an increase in fetal liver kinase-1 (FLK1, also known as VEGFR2)-expressing cells, early mesenchymal/endothelial progenitor cells that can differentiate into hematopoietic and ECs [16]. Interestingly, Etv2 overexpression induced the de novo generation of FLK1+ cells in serum-free differentiation conditions, where mESCs differentiate only up to the embryonic germ layer stage. Genome-wide transcriptome, proteomics analyses and chromatin immunoprecipitation sequencing have demonstrated that ETV2 alone or often together with other transcription factors such as OVOL2 [17], GATA2 [18] and TET2 [19] directly binds its numerous endothelial target genes including Flk1, Cdh5, Tek, Flt4 and Egfl7 [20, 21] and activates their expression, driving EC generation in developing embryos.

Other than its role in endothelial fate acquisition, a recent study suggests that ETV2 may play a critical role in endothelial progenitor migration for embryonic vessel formation [22]. ETV2 was found to directly binds the ETS consensus sequences within the promotor of RhoJ, a Rho-related GTP-binding protein. Etv2 overexpression led to a robust RhoJ expression, but its message was abolished in embryoid bodies (differentiated cell aggregates of ESCs) lacking Etv2. In mouse embryonic fibroblasts (MEFs), ETV2-mediated migration was inhibited by RhoJ knockdown, which was partially rescued by the introduction of Etv2 into MEFs [22]. Regarding the protein family of Rho GTPases, another study reported it as a potential direct target of ETV2 in FLK1+PDGFRα− endothelial/hematopoietic progenitors in mESCs [20]. Additionally, RhoJ was shown to be directly activated downstream of ERG [23], which is also an ETS transcription factor acting downstream of ETV2 [20]. These observations support the pro-angiogenic role of ETV2 through RhoJ in regulating cell migration/motility. Similarly, we have shown that ETV2 can regulate ECs motility through Myct1, a novel endothelial gene directly activated by ETV2 (see below for further information) [24]. Altogether, studies over the past decade have clearly indicated a potent and indispensable function of ETV2 in ECs. For more details on the role of ETV2 in early development and its underlying molecular mechanisms, readers are referred to our previous review [25, 26].

ETV2 is an emerging target for direct reprogramming of non-endothelial cells into endothelial cells

Studies thus far have posited ETV2 as a master regulator of EC development. One important feature of ETV2 is that its expression becomes undetectable beyond E11.5 during mouse embryogenesis [7, 8, 14, 15]. In postnatal life, Etv2 expression is markedly observed in the testis [27] and remains silent in most tissues including quiescent ECs in adults [28], suggesting that the sustained expression of Etv2 might have detrimental biological consequences in a steady state. Indeed, when ECs were forced to express Etv2 throughout development, vessels of the mice (Tie2Cre;RosaR26-Etv2) were dilated, resulting in lethality [29]. Furthermore, ectopic Etv2 expression induced endothelialization in non-endothelial compartments [29], accounting for its transient expression. However, this seemingly undesirable ETV2 function could be beneficial when generating ECs through the directed differentiation of stem cells or reprogramming cells for cell replacement therapy. In the following sections, we discuss the translational applicability of ETV2 with an emphasis on EC generation from diverse cell sources for vascular-related disease.

John Gurdon’s historical discoveries that unfertilized enucleated frog eggs transplanted with the nucleus of a mature intestinal epithelial cell can generate a frog [30] challenged the classical view that cells lose their potential upon differentiation due to irreversible genomic modifications, which is well represented by Waddington’s famous rolling ball theory [31]. These provocative findings had gained support from the generation of Dolly the sheep, whereby a mammary gland cell’s nucleus was transferred into an enucleated egg [32]. Taken together with the reports that somatic cells such as fibroblasts and thymocytes fused with human ESCs (hESCs) generate hybrid cells with an ES character [33, 34], these studies suggest not only that the genome of a fully mature cell has totipotency or pluripotency, but also that a fully differentiated cell can be converted into an embryonic stage through appropriate environments or stimuli. Indeed, Yamanaka and colleagues have shown that somatic cells (mouse embryonic and adult tail-tip fibroblasts) can be reprogrammed into ES-like cells, known as iPSCs, through several key transcription factor overexpression [35, 36]. Twenty-four initially selected genes were tested for their ability to induce pluripotency in somatic cells as demonstrated by cell morphology, gene expression, differentiation potential and DNA methylation profile (bisulfite genomic sequencing). They were finally narrowed down to four factors; Oct3/4, Sox2, c-Myc and Klf4. Upon transplantation into athymic nude mice, the cells reprogrammed by the four factors (also known as Yamanaka factors) developed teratomas containing derivatives of three embryonic germ layers, confirming the bona fide pluripotency of reprogrammed cells. Consistent results were obtained with human somatic cells and the similar combinations of these transcription factors [36,37,38], opening a new era of cell reprogramming. These findings are particularly important for clinical use as iPSCs can ideally avoid ethical issues and the immune compatibility problems of hESCs.

In parallel, sporadic observations that certain types of somatic cells can acquire another identity have been reported. Overexpression of the eyeless transcription factor, Paired Box 6 (Pax6), can lead to Drosophila ectopic eye development on sites including the wings, legs and antennae [39]. Myogenic Differentiation 1 (MyoD), a member of the basic helix–loop–helix family of transcription factors, plays important functions in muscle differentiation [40]. Interestingly, non-myogenic cells, such as fibroblasts, and melanoma and neuroblastoma cells transfected with MyoD were converted into cells expressing muscle-specific genes and displaying muscular cell morphology [41]. Other studies have also shown the conversion of cell identity within the same cell lineages such as that of B cells into macrophages by CCAAT Enhancer Binding Protein alpha (C/EBPα) and CCAAT Enhancer Binding Protein beta (C/EBPβ) [42], and that of myeloblasts into eosinophils, thromboblasts and erythroblasts by GATA1 (GATA Binding Protein 1) [43]. Similarly, conversion of pancreatic exocrine cells into insulin-secreting β-like cells occurred in vivo when transcription factors Ngn3, Pdx1 and Mafa were introduced into mice [44]. These results strongly suggest that fully differentiated somatic cells can be converted or reprogrammed into other cell types through the overexpression of lineage-specific transcription factors or master regulators of specific cell lineages.

Encouraged by studies on iPSCs and cell type conversion between somatic cells, systematic approaches have attempted to directly convert diverse somatic cell types into other cell types through key transcriptional regulators. This strategy enables bypassing the pluripotent stage, thereby substantially reducing the tumorigenic potential and shortening procedural times, which are the major advantages of the direct cell conversion or reprogramming method over iPSC technology [4]. For example, a combination of the nervous system-specific transcription factors (Ascl1, Brn2 and Myt1l) directly converted fibroblasts into neurons. Ascl1 alone was sufficient to induce immature neural function, but its combined expression with Brn2 and Myt1l generated mature neuronal cells with higher efficiency [45]. Induced hepatic cells can be directly generated from mouse fibroblasts when hepatic lineage-specific transcription factors (transduction of Gata4, Hnf1α, Foxa3 and p19 Arf inactivation) are overexpressed [46]. Another study has shown that the combination of Gata4, Mef2c and Tbx5 can directly reprogram postnatal mouse fibroblasts and tail-tip fibroblasts into functional cardiomyocytes [47]. When fibroblasts transduced with these factors were transplanted into a mouse heart, they differentiated into cardiomyocyte-like cells. In a similar study, GAMT transcription factors (Gata4, Hand2, Mef2c and Tbx5) were used to directly convert adult mouse tail-tip and cardiac fibroblasts into cells with a cardiac phenotype [48]. Furthermore, the authors showed that the delivery of the four factors into the mouse heart enables non-cardiomyocytes to become cardiomyocyte-like cells and improves cardiac functions after myocardial infarction (MI). Please refer other reviews for further reading on iPSCs [2] and direct cell reprogramming to non-ECs [3, 4].

Accordingly, extensive efforts have been made to directly generate functional ECs from non-ECs. The first successful attempt was reported by Rafii and colleagues [49]. Lineage-committed human amniotic fluid-derived cells (ACs), defined as EpCAM+TRA1-81−c-KIT− and EpCAM−TRA1-81−c-KIT−, were directly reprogrammed into ECs through a first wave of transient Etv2 overexpression, followed by constitutive Erg and Fli1 expression, and TGF-β pathway inhibition [49]. Additionally, the authors showed that the induced vascular ECs (iVECs) exhibit mature EC characteristics and form stable and functionally perfused vessels in vivo, as shown by matrigel plug assay and hepatectomy-induced regrowth of sinusoidal vessels. Shortly after, direct reprogramming of mouse adult dermal fibroblasts to ECs was achieved by overexpressing five transcription factors: Foxo1, Klf2, Tal1, Lmo2 and Etv2. Importantly, the reprogrammed ECs improved perfusion recovery when cells were transplanted into a mouse model of hindlimb ischemia [50]. Although promising, these approaches have limited therapeutic applicability. Access to the ACs is invasive, and iVECs are incompatible with autologous cell transplantation. Additionally, the reprogramming requires multiple transcription factors in the form of viral particles, which could complicate long-term outcome and elevate genetic burdens, leading to genomic instability. Given that the transcription factors used for direct reprogramming are ETV2 targets, two groups, including our team, have successfully demonstrated that the single factor, ETV2, can directly convert human dermal fibroblasts (HDFs) into cells with endothelial functionality. Morita et al. showed that transient ETV2 expression in HDFs is sufficient to induce stable endothelial gene expression [9]. Additionally, the reprogrammed ECs were incorporated into host blood vessels in the Matrigel plug assay and enhanced blood flow recovery in mice undergoing ischemic insult. Independently, we also directly converted HDFs into functional ECs using ETV2 only [10]. Interestingly, we found two successive stages of direct reprogramming: early and late-reprogrammed ECs. At the early stage, the reprogrammed ECs display immature EC characteristics but have vascular regeneration and tissue repair functions, as demonstrated through in vitro assays and an in vivo hindlimb ischemia model. The late stage reprogrammed ECs are generated when the early reprogrammed ECs are further cultivated with a boost of ETV2 expression together with valproic acid (VPA, a histone deacetylase inhibitor) treatment. Late-reprogrammed ECs has similar transcriptomes to those of mature ECs and produced nitric oxide and highly expressed CD31, which are representative mature EC markers. Importantly, ETV2 expression in late-reprogrammed ECs was very low, meeting the criteria of mature ECs [28].

The potent direct cell reprogramming ability of ETV2 is also evident in other somatic cell types. The short-term expression of ETV2 while inhibiting TGF-β signaling changed human adipose-derived stem cells (hADSC) into EC-like cells with durable vascular identity [51]. In this study, KDR/VEGFR2+ cells isolated from ETV2-transduced hADSCs in the presence of TGF-β inhibitor were cultured without the continued expression of ETV2. Transcriptome analyses revealed a shared transcriptome profile between the further cultured cells and mature ECs. The authors showed the therapeutic potential of the reprogrammed cells in promoting vascular repair. The same reprogramming scheme induced endothelial identity in human umbilical cord-derived mesenchymal stem cells [51]. Mouse adventitial SCA1+ progenitors transduced by ETV2 acquired endothelial characteristics of gene expression profile, phenotype and function [52]. When the reprogrammed SCA1+ cells were transplanted into wire-injured femoral arteries, the mice receiving the cells showed improved vascular remodeling due to reduced smooth muscle cell proliferation and enhanced reendothelialization [52]. Additionally, fast skeletal muscle cells in zebrafish injected with Etv2 became ECs and then were incorporated into the vascular network. However, such a change was possible only when Etv2 was injected into embryos in a limited developmental window (between 22 and 30 postfertilization) [53]. The C2C12 mouse myoblast cell line can also express endothelial genes in response to Etv2. Consistently, Etv2-deficient hematopoietic and vascular progenitors differentiated into skeletal muscle cells in zebrafish embryos [54]. Altogether, these results strongly suggest that ETV2 alone and often together with other signaling molecules can directly convert non-ECs into functional ECs, which could have therapeutic potential for treating cardiovascular disease. A summary of ETV2-mediated direct reprogramming to ECs is shown in Table 1.

Table 1 Summary of ETV2-mediated direct reprogramming to endothelial cells
Full size table

Studies have also reported that the use of ETV2 and hypoxia increases the reprogramming efficiency [55]. Fibroblasts transduced with ETV2 under hypoxic conditions formed CD31+VEGFR2+ cells, and the reprogramming efficacy to endothelial progenitor cell was increased, compared with those under normoxic conditions. Intriguingly, Mathison et al. found that ETV2-overexpressing rat cardiac fibroblasts show increased cardiomyocyte markers cTnT and Actc1 and higher expression of cardiac markers when additionally treated with GMT (Gata4, Mef2c and Tbx5) [56]. Human cardiac fibroblasts also expressed higher cTnT level when treated with ETV2 and GMT, compared with cells treated with GMT alone. This study suggests that the transendothelial state induced by ETV2 overexpression induces cell reprogramming with high efficiency and represents a desirable cellular target for cardiac differentiation. Thus, it is tempting to speculate that the generation of EC subtypes such as arterial, venous and lymphatic ECs from fibroblasts can be efficiently achieved by ETV2 transduction, followed by that of EC subtype-specific transcription factors such as COUP-TFII for venous ECs and Prox1 for lymphatic ECs [16].

Therapeutic potential of etv2 in pathophysiological angiogenesis

In postnatal tissues, angiogenesis occurs during the repair phase of injury after inflammation via EC proliferation, vessel sprouting and remodeling. Although quiescent in adults, ECs have the ability to rapidly proliferate upon various environmental cues such as hypoxia, inflammation or reactive oxygen species, all of which can be found in developmental angiogenesis [57]. The transient expression of Etv2 only in early embryogenesis puzzles its potential role in postnatal angiogenesis despite its potent and specific function in EC generation and functionality. We provided the first evidence on Etv2 function in adults, using a mouse model of hindlimb ischemia. Etv2 expression was reactivated in hindlimb ECs upon ischemic injury [58]. Importantly, mice deficient in endothelial Etv2 showed significantly compromised vessel formation in a series of acute injury models, including hindlimb ischemic injury, laser-induced eye injury and skin wounding. In contrast, the forced expression of Etv2 in the mouse hindlimb under ischemia insults promoted blood perfusion recovery and new vessel generation. Considered along with findings of the significantly reduced necrosis and fibrosis in Etv2 injected hindlimb muscles, these results indicate that Etv2 plays essential roles in new vessel generation and tissue repair in response to ischemic damages in postnatal life. The potential therapeutic function of Etv2 was also demonstrated in a murine model of MI. The delivery of lentiviral Etv2 led to a significant improvement of cardiac functions and substantial induction of vessel formation in hearts with MI [59]. Interestingly, adeno-associated virus (AAV)-mediated Etv2 delivery also reduced infarct size and enhanced cardiac functions, further supporting its therapeutic feasibility for clinical applications. It is noteworthy that the therapeutic role of Etv2 in adult vessel repair shown in these two independent rodent models of hindlimb ischemia and MI may not be solely due to Etv2’s direct pro-angiogenic function in ECs, since anti-inflammatory and antifibrotic effects in injured tissue were also observed. Further in-depth studies on Etv2 function in endothelial, parenchymal and other interstitial cells are warranted.

The tumor angiogenesis studies reinforce the therapeutic applicability of ETV2 in vascular-related disease. Conditional inhibition of etsrp in zebrafish receiving mouse B16 melanoma cells resulted in a marked decrease in tumor vessel growth, compared with that of control zebrafish [60]. Other investigations of various human malignant tumor tissues (lung, breast, prostate and colon) revealed that ETV2 is expressed in tumor-associated ECs (TAECs), but not in ECs from healthy controls [61]. Consistently, Etv2 expression was also evident in the TAECs of mice transplanted with Lewis lung carcinoma cells, while remaining silent in control mice ECs [61]. In agreement with these findings, the essential function of Etv2 during tumor formation was confirmed in endothelial Etv2-deficient and Etv2 siRNA-treated mice, as shown by a reduced tumor volume with a significant decrease in tumor vessel formation. The role of Etv2 in pathological angiogenesis was also demonstrated in glioblastoma multiforme (GBM), a malignant tumor with high recurrence. Zhao et al. showed that ETV2 expression positively correlated with GBM severity and that ETV2-positive tumor cells in high-grade GBM tissues expressed the endothelial marker CD31 [62]. Remarkably, ECs that coexpressed ETV2 and CD31 were more enriched in the core region of GBM tissue than in the superficial region, suggesting that TAECs in hypoxic locations have higher ETV2 expression. ETV2 could also transdifferentiate CD133+/Nestin+ GBM neural stem cells (NSCs) to EC-like cells, possibly by suppressing critical neural differentiation genes [62], which is consonant with ETV2’s role in converting non-ECs to cells with an endothelial functionality. These findings prompt the speculation that ETV2 could be used to turn devastating cancer or cancer stem cells into non-cancerous cells. A summary on the function of ETV2 in vivo is provided in Table 2.

Table 2 Experimental validation of ETV2 function in therapeutic angiogenesis
Full size table

To understand the detailed molecular mechanism by which ETV2 controls EC generation and function, many groups have attempted to identify the direct downstream targets of ETV2, revealing several key factors such as Flk1, Cdh5, Tie2, Tal1, Lmo2 and RhoJ [20, 22]. Recently, our team has identified a novel and endothelial-specific gene, Myct1 (Myc target 1), as a direct bona fide effector downstream of ETV2. Based on the results of gene expression profiling from various solid tumor tissues of patients, and subsequent comparisons with the target genes of ETV2, we hypothesized that Myct1 could be directly activated by ETV2 [24]. Single-cell sequencing analyses confirmed its specific expression in ECs, and subsequent assays showed that ETV2 can directly bind to the promoter region of Myct1, activating its expression. We further found that Myct1 is expressed in both plasma membrane and the Golgi complex and regulates motility and tight junctional integrity of ECs. Importantly, Myct1 inhibition (i.e., Myct1 global knockout and Myct1 endothelial knockout) led to a reduction in tumor growth and angiogenesis, reminiscent of Etv2-deficient tumor vessel [61]. Interestingly, the lack of Myct1 in TAECs augmented antitumor immunity, as shown by the enhanced transendothelial migration of cytotoxic T lymphocytes and M1 macrophage polarization in the absence of endothelial Myct1. The combined inhibition of Myct1, programmed death 1 (PD1) and VEGFR2 signaling completely blocked tumor growth, whereas cotreatment with anti-PD1 and anti-VEGFR2 blocking antibodies failed to induce tumor regression. Collectively, these findings suggest that the ETV2-MYCT1 axis plays essential roles in arborizing tumor vessels and regulating tumor immunity, indicating that in-depth analysis of ETV2 functions could open unforeseen opportunities in the treatment of vascular disease such as cancer, MI and critical limb loss.

ETV2 for directed pluripotent stem cell differentiation into hematoendothelial cells

Pluripotent stem cells (iPSCs and ESCs) are undifferentiated pluripotent cells that can provide scalable amount of various types of differentiated target cells for therapeutic and experimental purposes [63]. It has been shown that hematopoietic differentiation of PSCs by Etv2/Gata2 overexpression was preceded by transient endothelial phenotypes (CDH5+/CD73−/CD43− cells), and they became hematopoietic cells (CD43+ cells) within 2 days [64]. The transfection of Etv2/Gata2 modified mRNA (mmRNA) activated the hematoendothelial differentiation in hESCs, resulting in a similar number of colony-forming cells to that obtained by lentiviral ETV2/GATA2 [64]. A recent study also demonstrated that untranslated region (UTR)-ETV2 mmRNA alone was capable of inducing PSCs into CDH5+ cells within 1–2 days, after which they became myeloid progenitors and, subsequently, functional neutrophils [65]. Therefore, these two studies suggest that ETV2-directed hematoendothelial cells can be obtained from PSCs without pan-mesodermal stages, providing an efficient method for generating hematoendothelial lineages. More recently, Wang et al. have reported a two-step method for generating ECs from iPSCs; iPSCs were first converted into mesodermal progenitor cells (MPCs) by Wnt/Nodal signaling activation using a glycogen synthase kinase 3 (GSK3) inhibitor (CHIR99021). The resulting MPCs were then transfected with ETV2 mmRNA [66]. This strategy allowed the generation of CDH5+/CD31+ induced ECs (iECs) (approximately 95% positive) after 48 h of ETV2 mmRNA delivery. These iECs robustly expanded for 3 weeks yielding approximately 70-fold increase. In contrast, bypassing the MPC stage by transfection of ETV2 mmRNA into iPSCs (not into MPCs) resulted in a marked decrease in iEC proliferation, yielding only twofold increase in 2-week period. In vivo, the number of functional, perfusable vessels surrounded by α-smooth muscle actin+ perivascular layer was higher in mice injected with iECs produced from a two-step protocol compared with that of mice receiving those from a one-step protocol. The results of this study underpin the importance of timely ETV2 expression in hematoendothelial cell progenitors for proper EC development. Furthermore, these results suggest the feasibility of ETV2 mmRNA for target cell generation (directed differentiation, direct cell reprogramming in vitro and in vivo) for clinical application due to its minimal mutagenic potential, unlike that of viral gene delivery. A summary on the function of ETV2 is provided in Fig. 1.

Fig. 1
figure 1

Schematic diagram of ETV2 function. ETV2 has multiple functions in the regulation of the cardiovascular system. The development of hematopoietic and endothelial cells is critically dependent on ETV2 as evidenced by a complete lack of both lineages in Etv2 deficient embryos. In adults, endothelial ETV2 is required for new vessel formation in response to injury. Furthermore, ECs lacking Etv2 have impaired ability to support tumor growth. In bone marrow (BM), ETV2 regulates hematopoietic stem cell proliferation, contributing to BM hematopoiesis. Recent studies have shown that ETV2 can directly generate ECs from non-ECs and PSCs. Created in BioRender.com

Full size image

Future perspectives on etv2 as a therapeutic vehicle for vascular disease

Advantages of ETV2-mediated vessel recovery

Currently, medications for vessel diseases include (para)sympathomimetics or chemicals for vasodilation, blood pressure reduction, cholesterol level reduction, clot prevention, as well as other drugs used for treating diabetes and hypertension [67]. As discussed, ETV2 is an indispensable transcription factor that can directly up-regulate wide arrays of genes necessary for early vessel development, reprogramming and neovascularization in adult vessels under pathophysiological conditions. Thus, it can be expected that ETV2 alone can reactivate the expression of multiple essential genes which are required for vessel regrowth unlike certain angiogenic factors such as VEGF that can only activate limited sets of genes. Also, ETV2 promotes the mesodermal progenitor cells to express FLK1 (VEGFR2), from where various cell types comprising vessels, including endothelial cells and vascular smooth muscles cells are formed [68]. From these findings, it can be speculated that ETV2 may have high potency for generating vascular organoids or vascular graft comprised with various cell types originated from autologous origin. Combined with conventional interventional therapies, ETV2-based therapeutics will provide an innovative and effective therapeutic modality for intractable vessel diseases.

The importance of cost effectiveness should be considered when creating new treatments. Although it is too early to begin clinical applications, it is important to consider that ETV2-based strategy should be affordable. Compared with other novel drug biological products (e.g., advanced therapeutic medicinal products such as genes, cells or combined therapies), mmRNA-based ETV2 delivery based on chemical methods (lipid nanoparticles, liposomes, polymer) [69, 70] would be preferred for rebuilding damaged vessels. Indeed, the efficiency and safety of mmRNA with proper adjuvants have become a novel platform in industry, due to the versatility of swift optimization without causing immunogenicity. This suggests that synthetic/modified ETV2 mRNA can become a safe, profitable and practically affordable strategy for vessel diseases [71, 72].

RNA-based direct reprogramming

Despite promising outcomes from experimental findings, strategies for the generation and reprogramming of clinically compatible ECs by ETV2 needs further optimization. Transcription factors including ETV2 have been delivered to cells or mice via viral systems, which can cause genomic instability and potential unexpected biological consequences, such as tumorigenesis [73]. Thus, methods avoiding such risks must be utilized to permit the therapeutic function of ETV2 for clinical use. Introducing mRNA into target cells has gained scientific interest for cell fate changes or therapeutic purposes because this biomolecule shows virtually no risk of genome integration, and no further transcription in the cytoplasm is needed [74, 75]. However, the applicability of mRNA has been hampered because of its instability and potential immunogenicity. These limitations are now being overcome by mmRNAs, which have a wide range of modifications in nucleotide, untranslated regions (UTRs), poly-A tail and 5′-capping 7-methylguanine, all of which improve the stability of mRNAs while reducing immunogenicity [76]. Since the first report on the use of mmRNA for reprogramming fibroblast into iPSCs, and their subsequent differentiation into myogenic cells [77], various studies have demonstrated that this approach is non-immunogenic, non-integrating and efficient for cell fate change [74]. In particular, the versatility of mmRNA in direct cell fate conversion has already been demonstrated in several cell types [78]. Corritore et al. demonstrated that human pancreatic duct-derived cells which are epithelial-origin could transdifferentiate into insulin-secreting β-cells through transfection with an mmRNA of V-Maf musculoaponeurotic fibrosarcoma oncogene homolog A, a basic leucine zipper transcription factor that regulates insulin expression in mature β-cells [79]. Additionally, such β-cells ameliorated hyperglycemia upon transplantation into the subcapsular space of diabetic SCID-beige mice. Another study showed that neural precursors can be generated from non-neuronal cells (adult HDFs) through with mmRNAs of SOX2 and PAX6, which are key regulators in neural development [80]. Furthermore, these cells acquired cellular and molecular phenotypes of GABAergic or glutamatergic neurons, depending on the differentiation protocol. These studies would warrant the use of mmRNA of ETV2 in direct cell reprogramming in vitro and in vivo. However, no study so far has reported a success in the direct reprogramming of somatic non-ECs to ECs by mmRNAs, including ETV2 mmRNA.

Chemical-mediated reprogramming

Another way of producing desired cell types is using chemicals. Chemical reprogramming is advantageous over conventional protocols using genetic factors in that it can provide better temporal/dosage control and is less immunogenic and tumorigenic [81]. Additionally, the underlying mechanisms for the action of chemicals in animal cells are often well identified, thus enabling one to determine how the reprogrammed target cells can be biochemically obtained [82]. An early investigation showed that mouse fibroblasts were converted into NSCs by the inhibition of GSK3β, MEK and TGF-β signaling in the presence of epigenetic regulators (VPA, Bix01294 and RG108) and a cell death blocker (vitamin C). These NSCs differentiated into neural cells, including astrocytes, oligodendrocytes and neurons in vitro and in vivo [83]. Since this finding, success in converting non-parenchymal cells (e.g., fibroblasts) into various target cells (hepatocytes, cardiomyocytes and skeletal muscles) under a specific set of chemicals has been reported [84]. A recent study also demonstrated that chemical reprogramming can convert cell fate via an intermediate stage. Upon inhibition of GSK3 and ALK5 signaling with simultaneous c-AMP stimulation, fibroblasts underwent epigenetic changes reaching a plastic stage that may possess multilineage potential, after which they further differentiated into diverse cell types including neural or skeletal cells [85]. Similarly, the conversion of human fetal lung fibroblasts and foreskin fibroblasts into cardiomyocytes upon treatment with nine compounds, probably through a mesoderm stage, has been reported [86]. A distinct but partly overlapping chemical cocktail induced neurons from healthy human foreskin fibroblasts and skin fibroblasts from patients with familial Alzheimer’s disease [87], highlighting the clinical potential of chemical reprogramming. Although the generation of chemically reprogrammed parenchymal cells (e.g., cardiomyocytes and neurons) from fibroblasts has been successful [86,87,88], the efficiency of chemical reprogramming is low [89], and chemical approaches have not been applied to date in endothelial reprogramming. Considering that some small molecules such as ROCK inhibitors, that GSK3 inhibitors and forskolin are the usual suspects for cell reprogramming, and that the generation and maturation of reprogrammed ECs by ETV2 require TGFβ-inhibition or VPA treatment [10, 49], a cocktail of chemicals with such small molecules would be applicable for endothelial reprogramming. Alternatively, combining chemical treatment with ETV2 activation via either a Crispr/Cas9 system or ETV2 mmRNAs encapsulated in lipid nanoparticles (LNPs) (see next paragraph) would increase the efficiency of endothelial reprogramming for basic research and clinical studies.

Gene delivery

Gene delivery systems other than retroviral and lentiviral systems still hold promise and should be optimized to enhance the clinical feasibility of ETV2-mediated therapy. Currently, the AAV system is being tested for treating hereditary neuron, muscle, eye and liver genetic disorders [90]. In contrast, attempts to evaluate its clinical use for other major organs including the heart, kidneys and lungs are still lacking [90, 91]. The AAV system was found to be efficient in reprogramming somatic cells into iPSCs in vivo after their administration with reprogramming factors in mice, suggesting that the AAV system may provide efficient tool for in vivo reprogramming [92]. Indeed, the feasibility of AAV-Etv2 in vascular disease was also demonstrated. In preclinical study, Lee et al. found that the local injection of AAV-Etv2 into rat MI hearts led to increased vessel growth, reduced scar formation and enhanced cardiac functions [59]. Other parameters including the serotypes of capsid proteins and packaging capacity should be taken into consideration to increase its success in clinical use [90]. Another promising delivery vehicle is Sendai virus (SeV), an RNA virus that does not integrate into the host genome [93]. SeV-mediated gene deliveries have been conducted to treat diverse diseases and have a demonstrated ability to reprogram cell fate [94, 95]. For example, a SeV vector expressing the cardiac reprogramming factors (Gata4, Mef2c and Tbx5) directly reprogrammed mouse fibroblasts and human cardiac fibroblasts into cardiomyocytes. The injection of these factors through SeV into MI hearts converted cardiac fibroblasts into cardiomyocytes and improved cardiac functions and reduced fibrosis [95]. Therefore, these results support the likelihood that SeV-ETV2 can serve as an efficient and safe means for direct reprogramming of non-ECs to ECs.

Since the first report of liposome synthesis [96], lipid nanoparticles (LNPs) have been widely tested in many clinical trials to deliver therapeutic substances. As important nanocarriers, LNPs can transport hydrophobic drugs and hydrophilic molecules including small chemicals, nucleic acids and proteins. They have been used for delivering antitumor, anti-inflammatory, antifungal drugs and antibiotics [97]. Various types of LNP-based mRNA vaccines have been developed for clinical trials, including those for viral diseases (rabies, Zika virus, cytomegalovirus, influenza, COVID-19, etc.) and cancers (melanoma, ovarian cancer, glioblastoma, etc.) [76, 97]. Considering mRNA’s proven biosafety and ease of preparation, coupling ETV2 mmRNA with LNPs for controlled release into host tissues would be an ideal option to overcome current limitations of viral gene delivery systems.

Engineering vascular organoids/tissues using ETV2

Human organoids have become essential tools for understanding organ development and disease progression, and testing drug efficacy [98]. Despite the progress on establishing various organoids, current protocols are suboptimal partly due to the lack of vasculatures to support the growth and maturation of organoids [99]. Thus, vascularized organoids will provide more physiologically relevant models that mimic in vivo counterparts. A recent study has reported that ECs derived from hESCs overexpressing ETV2 successfully remodeled to become vascular structures in human cortical organoids (hCOs) [100]. Functionally, vascularized hCOs outperformed avascular hCOs in blood–brain barrier functions, tight junction integrity and the abundance of capillary network vessels capable of perfusion into host circulation in immunodeficient mice. The potential of ETV2 in generating vascular grafts was also demonstrated. Palikuqi et al. have shown that the transient induction of ETV2 by lentiviral particles resets mature ECs (HUVECs) to an embryonic-like “reset” state, which allows vasculogenesis more efficient. Importantly, these reset ECs formed perfusable and hemodynamic vessels upon being mixed with specific extracellular matrix and were able to establish a vascular network that integrates with tissue-specific parenchymal cells. Functionally, the number of vessels in colon cancer organoids and microfluid-based pancreatic islet organoids was increased [101]. Other than reprogramming strategies, attempts for generating transplantable organs with human endothelium were reported in an animal biotechnology study. Das et al. reported that, upon being complemented with human iPSCs, pig preimplantation embryos deficient in ETV2 developed human vasculature at E17-18 [102]. This study showed the non-redundant function of ETV2 in vessel formation in a large animal and suggests that chimeric pig organs with human vasculature may become a novel source of various donor biocompatible organs with human endothelium. However, further investigations are needed to enable these chimeric piglets to develop to full term or at least to a stage where their organs can be procured for transplantation studies.

Challenges and possible strategies for clinical use of ETV2

The non-redundant and essential function of ETV2 in vascular development, regeneration and endothelization of non-ECs has raised the possibility of its effective clinical use [103]. However, there are several roadblocks to its clinical application. As discussed above, the prominent concern is that dormant ECs in adult tissue may not readily express co-activators/repressors or epigenetic activators for proper function of ETV2. Thus, it is crucial to identify the detailed molecular mechanisms by which ETV2 functions. For example, the comprehensive profiling on upstream regulators of ETV2, epigenetic changes on ETV2, ETV2 binding proteins and ETV2-mediated epigenetic regulations would advance our understanding of ETV2 and increase its therapeutic potential.

In clinical procedures, the most favorable route for drug administration is via venous flow. However, this often causes rapid degradation by plasma enzymes or inactivation after interaction with plasma protein [104, 105]. Furthermore, intravenous administration is often a suboptimal route for many cardiovascular disorders [106, 107]; thus, methods that enable targeted delivery are often needed. Accordingly, engineered nanoparticles/LNPs/exosomes/AAV encapsulated with ETV2 and its partners (in the form of mRNA, mmRNA, siRNA) that can target vessel lesions without provoking immune or inflammatory responses should be considered for therapeutic purpose [108]. Another challenge for clinical usage of ETV2 in vessel disease may rise from the complexed microenvironment where ECs are located. The structure and physiology of vessels vary depending on its classification (e.g., arteries vs veins) and the tissue/organ types. Moreover, many vasculopathies present with comorbidities (tumor, diabetes, hypertension, obesity, etc.). Thus, it is difficult to create ETV2-based therapies that target certain vasculature in various contexts. Microneedle-based nanotechnologies for local, controlled release of these potential drug products would be another possible alternative for clinical use to be considered [109, 110]. However, it is also important to note that comprehensive understanding of the safety of the aforementioned gene delivery methods is mandated. Indeed, one preclinical study showed that the transgene loaded in recombinant AAV vector inserted into host genome and was clonally expanded [111], although this has not yet been seen in human. Currently, no information is available regarding therapeutic efficiency of ETV2 in treating vascular disease since studies have only been performed in animal models to determine the role of ETV2 in cardiovascular disease. Thus, more systematic and controlled experiments would be warranted.

Conclusion

ETV2 as a pioneer factor for EC development has a potent ability to generate ECs from diverse cell sources. Further investigations on how to design clinically compatible forms of ETV2 and deliver them to targeted cell types or tissues will be required to grasp the therapeutic potential of ETV2 in treating vascular diseases (Fig. 2). Additionally, deciphering the molecular mechanisms of epigenetic changes (histone modifications), DNA methylation and chromatin accessibility along with single-cell omics will significantly advance our understanding of ETV2-mediated cell reprogramming.

Fig. 2
figure 2

Strategies for vessel regeneration by ETV2-driven reendothelialization. ETV2 may be exogenously administrated into vascular legions in the form of AAV, SeV or mmRNA. Alternatively, endogenous ETV2 can be reactivated via Crispr-Cas9. Therapeutic ECs of autologous origin can be generated from non-ECs or iPSCs by expressing ETV2 mmRNA, non-integrating gene delivery systems (e.g., NEP or LNP) or transactivating ETV2 by Crispr/Cas9. For tissue replacement therapy, these reprogrammed ECs can be either directly administrated or used for generating biocompatible organs/organomimetics such as engineered vascular grafts or vascular organoids. AAV Adeno-associated virus; Crispr-Cas9 Clustered regularly interspaced short palindromic repeats (CRISPR)-CRISPR-associated protein 9; ECs Endothelial cells; ETV2 Ets variant 2/ets-related protein 71; iPSCs Induced pluripotent stem cells; LNP Lipid nanoparticle; mmRNA Modified messenger RNA; NEP Nanochannel electroporation; SeV Sendi virus. Created in BioRender.com

Full size image