**1. Introduction**

The main goal of tissue engineering (TE) is to replace tissues and, more ambitiously, organs damaged by a large variety of insults. To this aim, TE relies on the combination of biocompatible scaffolds, suitable cellular sources and correct sets of signaling molecules. The integration of these factors is required for a successful and long-lasting regeneration process. The field is continuously evolving and the number of both in vitro and in vivo studies has grown exponentially over the last two decades. Despite this substantial increase still a very small fraction of bioengineered products is currently used for clinical applications.

The reason behind this discrepancy is mainly related to factors that cause graft failure, thus influencing the clinical translatability. It has been widely demonstrated that graft failure is mostly caused by the inadequate onset of a functional vasculature within the implanted scaffold. The insufficient vascularization of the neoforming tissue leads to a lack of integration of the construct with the host tissue due to insufficient metabolic supply and waste disposal [1]. In this scenario, different strategies have been developed, relying on the use of bioactive molecules [2], specific architectures [3] and topographic signals [4,5]. Concerning the support to vascular growth with bioactive factors [6] it has been proven that, in some cases, the host vasculature itself is unable to extend into the core of scaffolds exceeding 200 μm in thickness [7].

A possible approach to overcome this drawback is based on the incorporation of vasculature forming cells, namely endothelial cells (Figure 1), into the scaffold, as it has been already successfully performed in the case of bioengineered tissues [8,9] and organs [10]. ECs for scaffold vascularization could be derived from multiple sources. Doubtless, in most studies, the cells used are human umbilical vein endothelial cells (HUVECs), which hold several features that make them an attractive source of primary human ECs. They are retrieved from the umbilical cord, a tissue which is usually discarded, and is thus relatively abundant and easy to isolate [11]. In addition, a large set of assays has been set-up and widely validated. This means that a broad range of standardized tools to study angiogenic and antiangiogenic factors is available. Furthermore, a developing understanding of the cascade of molecular and cellular mechanisms of angiogenesis is crucial [12]. On the other hand, HUVECs show high heterogeneity depending on the donor, beyond the rapid loss of endothelial phenotype that they show when they are kept in culture [13]. The latter issue is extremely limiting in the view of an autologous cell transplant. Therefore, alternative EC sources are urgently needed for tissue engineering applications.

In addition, adult tissues such as skin, adipose tissue and aorta or coronary arteries could also provide ECs [14]. From the beginning of the 2000s, several studies using mouse models have indicated that microvascular endothelial cells isolated from human dermal tissue (HDMECs) are able to generate a functional vascular network anastomosed with the host vasculature [15,16]. In the following years, several works using scaffold entrapped growth factors in combination with HDMECs further confirmed the ability of these cells to form a fully functional vascular network [17]. Thus, ECs derived from adult tissues represent a good alternative to HUVECs. However, these cells suffer some major limitations that impair their translatability into the clinic. In particular, tissue procurement requires a procedure that is invasive for the patient; in addition, the in vitro proliferative potential of the isolated cells is very low. These limitations demonstrate the necessity to find an alternative source of cells suitable to be used in regenerative medicine.

Advances in vascular biology shed light on putative HUVEC substitutes: Asahara et al. showed the presence of Endothelial progenitor cells in 1997 [18], and a few years later, in 2000, Lin et al. identified these cells in peripheral blood, indicating them as Endothelial colony forming cells (ECFCs) [19]. ECFCs show a full set of endothelial cell markers. Beyond the molecular similarity to adult endothelial cells, ECFCs also hold a functional competence specific to ECs. In fact, in vitro studies demonstrate that ECFCs are capable to form more efficient vascular networks when embedded in a collagen matrix in comparison to other EC sources [20]. Within the in vivo setting, ECFCs display the ability to integrate and form perfused blood vessels when injected into immunocompromised mice [21,22]. Furthermore, Fuchs reported that these cells are able to guide the vascularization of an engineered bone tissue equivalent [23]. Although, starting from their first identification, the use of ECFCs has constantly increased [24], this EC source also implies severe restrictions. Indeed, if on one hand ECFCs are an efficient source of autologous ECs, on the other their use is strongly limited by their amount in the peripheral blood, where only 0.05–0.2 cells/Ml can be retrieved [25]. In addition, Mund et al. indicated the absence of specific markers, which is a further significant hindrance to the widespread use of these cells [26]. Overall, these concerns strongly discourage the isolation of ECFCs for tissue engineering purposes [26]. In this context, the best source of ECs is probably represented by embryonic stem cells (ESCs) derived from the Inner Cell Mass (ICM) of the blastocysts. ESCs are able to remain undifferentiated and to indefinitely proliferate in vitro, while maintaining the potential to differentiate

into derivatives of all three embryonic germ layers [27]. The use of human ESCs is strongly hampered by ethical concerns since the withdrawal of ICM results in the disruption of a human embryo. In this respect, even though several in vivo studies demonstrate their validity in forming new vessels, ESCs do not represent the ideal source of endothelial cells suitable for biomedical applications [28,29]. A remarkable breakthrough in cellular biology research was the discovery of induced pluripotent stem cells (iPSC) made by Takahashi in 2006 [30]. In adult life, multipotent stem cells can differentiate and replace almost all damaged tissues. Multipotency confers the ability to differentiate into cell lines belonging to the same germ layer. Pluripotent cells show a wider differentiation range; the germ layer of origin makes them even more exploitable for TE purposes. Yamanaka et al. [30] set up a method to reprogram mouse fibroblasts into iPSC by retroviral delivery of four reprogramming factors (OSKM factors: OCT-3/4, Sox2, Klf4 and c-Myc) while Takahashi et al., in 2007, improved the reprogramming method in order to obtain iPSC from human somatic cells (hiPSCs) [31]. The advent of iPSC represented a real milestone in the field of vascular biology since 2009, when Taura et al. collected the first evidence of the possibility to generate endothelial cells starting from iPSC (iPSC-EC) [32]. IPSC-ECs could be a valuable source of cells in regenerative medicine for several reasons [33]. These cells display the same pluripotency of ESCs, based on their gene expression profile, overcoming all the limitations that hampered embryonic stem cell usage [34,35]. Further iPSCs can be easily generated from patients; therefore, they can provide an autologous source of cells for regenerative medicine applications able to bypass the issue of host immune rejection. Moreover, iPSCs, as they are derived from adult somatic cells, do not present strong ethical concerns as ESCs do.

Among the various advantages, the most promising one is to have a tissue specific EC source; in fact, iPSC-EC display the same plasticity of immature ECs [36]. Evidence collected in independent studies demonstrates that, when exposed to tissue-specific cues, iPSC-ECs generate mature ECs able to almost completely resemble the characteristics of resident ECs [37,38]. On the other hand, a well-defined selection of iPSC is required, since, once implanted, they can easily induce teratoma formation [39]. In this paper, we aimed to display the potential of iPSC-ECs in vascular biology and regenerative medicine by analyzing the behavior of these cells both in vitro and in vivo. Furthermore, we critically reviewed the advances concerning the protocols set-up to generate and select these cells in order to overcome the most important safety issues.

#### **2. Pluripotency Induction**

The first method used to confer pluripotency to a somatic cell has been the nuclear transfer into an oocyte [40]. Pluripotency can be alternatively acquired by fusing a somatic cell to an ESC [41,42]. These findings indicate that both oocytes and embryonic stem cells possess factors able to confer pluripotency. Takahashi et al. identified putative pluripotency-associate genes, and among them, selected a minimum set of four genes responsible of the pluripotent state: OCT-3/4, Sox2, Klf4 and c-Myc [31] (Figure 2). These transcription factors, re-expressed into somatic cells, promote pluripotency, also affecting self-renewal and cell cycle progression [43–45].

**Figure 1.** Sources of endothelial cells (ECs) used in scaffold-based approaches for tissue engineering (TE). HUVECs: Human Umbilical Vein Endothelial Cells, HDMECs: Human Dermal Microvascular Endothelial Cells, ECFCs: Endothelial Colony Forming Cells, ESCs: Embryonic Stem Cells, hIPSC-ECs: Endothelial Cells derived from Human Induced Pluripotent Stem Cells.

**Figure 2.** Schematic representation of human Induced Pluripotent Stem Cells-EndothelialCells (hiPSC-ECs) generation. Firstly, somatic cells are collected from the patient, then pluripotency is induced by the re-expression of four genes identified by Yamanaka et al. in 2006: OCT-3/4, Sox2, Klf4 and c-Myc (OSKM factors) which are normally inactive in somatic cells. Afterwards, induced Pluripotent Stem Cells (iPSCs) differentiation is induced towards mature Endothelial Cells (ECs).

To induce pluripotency the OSKM factors were introduced in somatic cells by means of viral transfection. In particular, Takahashi and Yamanaka groups, by means of Murine Leukemia Virus (MuLV) and a lentivirus delivery, were able to generate induced pluripotent cells [30]. The construct carried out by the retrovirus consisted in a single polycistronic unit under the control of an inducible promoter, while lentivirus was necessary to deliver the viral construct. This approach leads to the expression of the genetic material as soon as the inducible factor persists. Therefore, when the induction is complete, the polycistronic unit is switched off. The method described relies on the integration of the retrovirus into the genome. Genome integration is itself a limitation of this induction approach. In this context, Okita et al. demonstrated that genomic integration of reprogramming factors increases the rate of tumor formation in chimeric mice [46]. This reprogramming method is dangerous because it may cause mutations in the site of insertion and, in addition, it shows a low induction efficiency. All these aspects strongly limit the translation of the induced pluripotent cells into the clinic.

Other reprogramming approaches could represent a safer alternative for clinical applications (Figure 3). In the context of integrating methods, a non-viral approach such as the transfection of linear DNA introduced by liposomes or direct electroporation can be used [47]. An intriguing approach to overcome viral delivery was developed by using PiggyBac (PB) transposon [48]. PB delivery is based on a kind of "cut and paste" mechanism by which the PB is co-transfected together with PB trasposase, causing a transgene cut from the PB vector, as well as the integration into the genomic TTAA sites. After this, the cut and paste mechanism includes a second transfection of the PB trasposase to remove the transgene from the insertion site. PB was also shown to be able to successful reprogram human somatic cells into iPSCs [49]. Despite the low efficiency, this method can be enhanced by adding butyrate to cell culture by 15- to 51-fold [50]. This approach, by involving innocuous vectors, is undoubtedly preferred to viral ones. Although PB can be considered a step forward in the development of a safe delivery method, the integration of transgene into the host DNA can cause genomic interruptions with uncontrollable downstream consequences [51].

**Figure 3.** Diagram representing the different approaches to deliver reprogramming factors OSKM (Oct-3/4, Sox2, Klf4 and c-Myc) to somatic cells. The Integrative approaches: Linear DNA, MuLV (Murine Leukemia Virus), PiggyBac. The Non-Integrative approaches: Episomal Vectors, Adenovirus, SeV (Sendai Virus), Modified RNA and CRISPR-dCas9 Synergistic Activation Mediators (SAM).

However, non-integrative approaches are the only option suitable for the clinic in order to avoid side effects of the integrating delivery techniques.

Among others, the first choice of a non-integrative strategy is adenoviral delivery. The adenoviruses used to deliver factors are defective for replication machinery. Stadtfeld et al. indicated that such adenoviruses can reprogram somatic cells, and that no traces of integration are detected afterwards [52]. However, this approach suffers from a low infection efficiency [53]. An intriguing option in the field of non-integrative approaches was reported by Yu et al., who derived human iPS cells from fibroblasts completely free of vectors and transgene sequences by a single transfection with oriP/EBNA1 (Epstein-Barr nuclear antigen-1)-based episomal vectors [54].

Episomal vectors stem from the Epstein-Barr virus and are plasmids well suited for the introduction of reprogramming factors into human somatic cells, since they can be transfected without the need of viral packaging, and can be subsequently removed by culturing the cells without the need of drug selection [54]. The oriP/EBNA1 vectors replicate only once per cell cycle, and they can be recognized by drug selection as stable episomes in about 1% of the cells transfected [55]. The absence of drug selection causes episome loss in the ~5% of cells per cell generation due to defects in plasmid synthesis and partitioning which make the isolation of cells free of plasmids very easy [56]. Unfortunately, the efficiency of iPSC generation by episomal reprogramming remains low [57]. In 2011, Okita et al. considerably improved the efficiency (10–100 fold) of the procedure by suppressing p53 and by using non-transforming L-Myc instead of c-Myc, during the reprogramming process [58]. However, the use of the p53 short-harpin RNA (shRNA) is problematic for translational purposes, since the interference with p53 pathway may antagonize the antitumoral function of the gene [59]. In 2009, Fusaki used a Sendai virus (SeV) as a vector to generate transgene-free iPSCs in different conditions [60]. Sendai virus is a negative-strand RNA virus, differently from other RNA viruses it replicates into the cytoplasm of infected cells and does not integrate into the host genome [61]. This characteristic makes Sendai virus-based vectors the safest viral-based tool to generate iPSCs since they are considered "zero footprint" and are diluted from the infected cells with the physiological cell division [62]. To maximize reprogramming efficiency during several steps, the use of inactivated feeders, but also the use of animal-derived products, was required; however, exposure of human cells to products of animal origin increases the risks of non-human pathogen transmission and immune rejection [63]. Macarthur et al. made a step forward into a safe generation of iPSCs in 2012 [64]. These authors were able to generate, by SeV infection, transgene-free human iPSCs in feeder-free and xeno-free conditions, even though they noticed a decrease in reprogramming efficiency [64]. However, since Sendai virus vectors can reprogram with high efficiency, they were able to obtain enough colonies for further expansion.

In a recent comparison between non-integrative methods to generate iPSC, Schlaeger indicated that the SeV reprogramming approach is the most efficient and reliable, with a low workload and a complete absence of viral sequences in most lines at higher passages [65]. However, no clinical grade SeV reprogramming vectors are available. Thus, in the view of clinical applications, SeV still presents major concerns.

In this scenario the gold standard non-integrative approach is the one proposed by Warren in 2010 [66], who used modified RNA to deliver reprogramming factors. These modifications included the replacement of the 5 cap with a synthetic one. In this protocol, RNA is complexed with cationic vehicle to facilitate cell uptake by endocytosis. Moreover, to prevent host ribonucleasic degradation and improve constructs half-life, the common cytidine and uridine bases are replaced respectively with 5 methylcitidine and pseudouridine. By these means, the authors were able to produce iPSCs [67,68]. A reprogramming method recently proposed is the one based on CRISPR-Cas9 fused to a synergistic activator mediator (SAM) [69]. This system is based on an engineered Cas9 protein (dCas9) serving as RNA-guided-DNA binding domain fused to a transcriptional activator domain (VP64). This chimeric activator complex can be directed towards promoter regions guided by specific single-guide RNAs (sgRNAs) [70]. Based on this approach Weltner et al., in 2018, generated iPSCs by targeting the promoters of OSKM factors [71].

#### **3. Protocols to Induce Mature EC Phenotype**

It is well known that for clinical use, it is mandatory to generate cells able to show a high degree of commitment. This requirement fulfills not only functional issues, but also safety ones, in order to prevent teratoma formation after implantation. Thus, a prerequisite to exploit iPSC-ECs in the clinical setting is the development of defined protocols to guide their differentiation into functional endothelial cells. The ideal induction protocol should be reproducible, easy to perform and relatively quick in order to allow yielding an adequate quantity of homogeneous cells [72]. Current induction strategies include embryonic bodies (EB) generation [73,74], differentiation on monolayers [75] and co-culture with primary cells (Figure 4).

**Figure 4.** Illustration of the main strategies used to differentiate hiPSC into hiPSC-ECs. Embryoid Bodies (EBs): cultured in suspension hiPSCs tend to auto-aggregate in embryoid bodies. Co-culture: hiPSCs are co-cultured with cells able to guide their differentiation into the mature phenotype. Two-dimensional (2D) monolayer culture: hiPSCs are seeded on matrix-coated plates where they are induced to differentiate.

IPSCs tend to self-assemble into three-dimensional (3D) structures (EB) when grown in suspension. From EBs, cell aggregates encompassing all three germ layers develop, and afterwards, within the positive mesodermal, EB cells tend to form vascular structures [36]. This method is affected by low efficiency (1–5%) [76] and slow production rate [77]; however, differentiation can be improved by adding proper growth factors to the culture medium [78,79]. Another approach involves a co-culture with primary cell lines able to induce iPSC-EC differentiation toward mature ECs. In detail, Choi et al. directed hiPSCs into mature ECs in the presence of OP9, a mouse bone marrow stromal cell line [80]. The authors speculated that these cells regulate iPSC induction via a paracrine signaling.

Monolayer differentiation holds a significantly higher efficiency that depends on external factors, such as medium constituents, showing a final yield that is still too low in the view of clinical applications [81]. To date, the best protocols showing the highest EC yields were developed by culturing a monolayer of hiPSCs on a matrix-coated culture plate and by treating them with different molecules or growth factors in a timed fashion in order to guide the progressive differentiation of hiPSCs toward the EC lineage [82]. In this context, GSK3 inhibitors play an important role among the set of molecules necessary to induce the differentiation of pluripotent cells into mature ECs [83]. In particular, vascular progenitors derive during human development from latero-posterior mesoderm [84]. To specify mesoderm [85,86], Wnt signaling, which is activated by GSK3 inhibition, is required [87]. In view of this, several authors have exploited GSK3 inhibitors to differentiate hiPSCs into ECs. Patsch et al. [72] exposed a monolayer of hiPSC to GSK3 inhibitor CHIR-99021 (CHIR) [88] and to mesoderm inducer bone morphogenetic protein 4 (BMP4). The combination of these two molecules led to the production of mature ECs in a relatively short time (six days) with 80% efficiency.

However, mesoderm induction is only the first step of differentiation. The second part starts upon mesodermal commitment by exposing the cells to factors that further induce the mature vascular phenotype. Gu [89] set up a protocol through which, after 4 days of treatment with VEGF and bFGF, he produced mature ECs in only 8 days. These cells, when tested, were molecularly and functionally similar to native ECs.

Paik et al. [90] added VEGF, bFGF and BMP4 to an already established protocol to produce mature ECs from hiPSCs within 12 days. Although this protocol requires more time compared to other ones reported in literature, the aim of this study was different. In fact, these cells were used to draw an RNA signature at different stage of differentiation.

The last step included in the differentiation protocol is the purification of positive cells. This step is essential to fish out a homogenous subset of cells and to ensure the safety needed for the future cell implantation and engraftment.

Cell sorting is usually performed by using magnetic beads on which surface specific antibodies are adsorbed [91]. These antibodies are directed against mature endothelial markers such as CD31 or VE-cadherin (also known as CD144).
