Bristol Heart Institute, School of Clinical Sciences, University of Bristol, Bristol, United Kingdom
National Heart Lung Institute, Hammersmith Campus, Imperial College of London, London, United Kingdom
Correspondence: Costanza Emanueli, B.Sc., Ph.D., F.A.H.A., Laboratory of Vascular Pathology and Regeneration, Bristol Heart Institute, University of Bristol, Bristol Royal Infirmary-level 7, Upper Maudlin Street, Bristol, England BS2 8HW, U.K. Telephone: 44-0-117-3423512; e-mail: firstname.lastname@example.org.
MicroRNAs (miRs) are highly conserved, short noncoding RNA molecules that negatively regulate messenger RNA (mRNA) stability and/or translational efficiency. Since a given miR can control the expression of many mRNAs, their importance in governing gene expression in specific cell types including vascular cells and their progenitor cells has become increasingly clear. Understanding how the expression of miRs themselves is regulated and how miRs exert their influence on post-transcriptional gene control provides novel opportunities to dissect gene regulatory networks in clinically relevant cell types. A multitude of miRs have been identified with key roles in vascular development, homeostasis, function, disease, and regeneration. In this review, we will describe the impact of miRs on angiogenesis and their capacity to modulate the behavior of stem and progenitor cells which may be utilitarian for promoting vascular growth in ischemic tissue. Moreover, we summarize these strategies available for modulating miR expression and function and future therapeutic applications. Stem Cells2014;32:1059–1066
MicroRNAs (miRs) are transcribed as primary miRs (pri-miRs) from either miR genes or intronic sequences of protein coding genes. A large proportion of miRs are clustered in polycistronic units. Pri-miRs are cleaved to reach their mature state. The canonical miR maturation process starts in the nucleus and requires Drosha (ribonuclease III-, RNase III), which (together with DGCR8) cleaves the pri-miR to a shorter double-strand stem-loop precursor miR (pre-miR). Next, the pre-miR is translocated to the cytoplasm by Exportin 5 and cleaved by a complex containing Dicer (a RNase III) and the RNA binding protein TRBP to form a ≈22 nt long double strand miRs composed by two mature miR sequences (oriented as 5p and 3p). Each of them can be recruited into a RNA-protein complex called RNA-induced silencing complex and assembled through processes that are dependent on Dicer, other RNA binding domain proteins, and members of the Argonaute (Ago) protein family . A Dicer-independent, Ago-2-mediated miRNA biogenesis pathway has also been described . The canonical end result of a miR action is the repression of protein production from the several mRNAs which it targets. This is reached through miR-induced mRNA degradation, inhibition of translation, or sequestration of the mRNA in the processing body (P-body), where it is degraded or stored (reviewed in ). A miR recognizes its mRNA targets through its “seed sequence” (of eight nucleotides) which is (semi)complementary to one or more “miR binding sites” usually located in the mRNA 3′-untranslated region. In excess of 1,000 human miRs have already been identified . By regulating the expression of several target genes, each miR has the potential to modulate multiple pathways. Furthermore, individual mRNAs can be targeted by multiple miRs, allowing for high combinatorial complexity and regulatory potential. In fact, miRs are predicted to exert fine-tuning of post-transcriptional regulation to >60% of mammalian protein-coding mRNAs .
Angiogenesis is the term given to the process through which new blood vessels form from existing vasculature. It denotes the series of events which shapes the primitive vascular network during embryogenesis. Angiogenesis is also relevant postnatally, including during the uterine remodeling (associated with the female menstrual cycle and in pregnancy), placental development, and wound healing. Moreover, aberrant angiogenesis is associated with human diseases, including tumor growth and diabetic retinopathy. By contrast, reparative angiogenesis is important for postischemic vascular repair [6, 7]. In this review, we will focus on therapeutically induced reparative angiogenesis. Effective pharmacological drug interventions for improving tissue perfusion safely and locally, thus alleviating disease burden, are lacking. Consequently, considerable efforts are directed toward new molecular and cellular therapies.
miRs in Endothelial Cells and Angiogenesis
It is now well established that miRs are important for vascular development, physiology, and disease. Initial evidence for a functional role of miRs in vascular development was provided by the observation that mice carrying a Dicer hypomorphic allele die prenatally with severely disrupted blood vessel formation accompanied with retarded expression of early endothelial markers . Moreover, Dicer silencing impaired the proangiogenic capacities of endothelial cells (ECs) . Table 1 illustrates the principal miRs affecting the angiogenesis process.
Endothelial-expressed miR profiling has been performed mainly by PCR and array methods and using human umbilical vein ECs (HUVECs). These efforts have led to the identification of high expression of miR-221/222, miR-21, the let-7 family, the miR-17–92 cluster, the miR-23–24 cluster, and miR-126 in vascular ECs. Of all the above, miR-126 is, so far, the only miR considered as expressed specifically in the endothelial lineage and hematopoietic progenitor cells (PCs) (reviewed in ). Deep-sequencing techniques allow for comprehensively profiling the entire miR population expressed by ECs. Voellenkle et al. used deep sequencing to identify 400 annotated miRs and a few novel miR species in HUVECs cultured under standard conditions or exposed to hypoxia, which is a classic proangiogenic stimulus . They reported that, in HUVECs, miR-21 and miR-126 totaled almost 40% of all annotated miRs .
Studies focusing on individual miRs or miR clusters suggest the importance of miRs in EC function and angiogenesis (reviewed in [12-14]). The miRNA-17–92 cluster is one of the most important miR systems for the regulation of angiogenesis. It encodes six miRs (miR-17, miR-18a, miR-19a, miR-20a, miR-19b-1, and miR-92-1), which are tightly grouped within an 800 base-pair region of human chromosome  and is transcriptionally regulated by c-Myc . The antiangiogenic effect of Dicer silencing in cultured ECs could be corrected by transfection of miR-17–92 cluster . In vivo, the miR-17–92 cluster was positively associated with cancer angiogenesis in mice . On the other side, individual components of the miR-17–92 cluster can have opposite regulatory effects on angiogenesis. miR-92a was one of the first miRs to be linked with postischemic angiogenesis responses , which it represses (vide infra). Individual miR-17, −18a, −19a, and −20a inhibit the angiogenic potential of cultured ECs, but do not affect tumor angiogenesis in vivo . Notably, miRs are secreted by their producing cells and can act via both cell autonomous and nonautonomous mechanisms. Moreover, the same miR can be produced by different cellular types, including cancer cells and ECs. This is, for example, the case for the miR-17–92 cluster. Dews et al. demonstrated that enhanced transcription of the miR-17–92 cluster in cancer cells increased in vivo tumor angiogenesis in mice. Similarly, miR-378 (which represses the tumor suppressors, Sufu and Fus-1) [10, 19] is produced by cancer cells and can promote cancer angiogenesis . It is feasible to hypothesize that transcription of proangiogenic miRs is increased in the tumor endothelium. Moreover, ECs may uptake miRs from cancer cells. In line with the latter hypothesis, miR-296 (which indirectly enhances the expression of VEGFR2 and PDGFRβ ) was reportedly increased in primary tumor ECs isolated from human brain tumors . Furthermore, miR-296 levels were increased in noncancerous brain ECs by coculturing with glioma cells . Finally, miR-132 (which acts by inhibiting p120RasGAP, a molecular brake for Ras)  was reported as highly expressed in the endothelium of human tumors and hemangiomas but undetectable in normal endothelium .
miRs Modulating Postischemic Vascular Repair
Several lines of evidence suggest the importance of miRs for regulating postischemic angiogenesis. Mice with endothelial-selective Dicer inactivation have an impaired angiogenic response to limb ischemia (LI) . Moreover, mice with EC-restricted deletion of miR-126 present leaky vessels and partial embryonic lethality. The subset of mutant animals that survive demonstrated a defective cardiac neovascularization response to myocardial infarction (MI) . Later on, miR-92a was shown upregulated in ischemic tissue and to act as a break on the native angiogenesis response to ischemia in several animal models . This was the first study testing the in vivo effect of a miR targeting approach for improving postischemic angiogenesis. In fact, systemic administration of an “antagomiR” (in this case a 2′O-methyl antisense oligoribonucleotides) designed to inhibit miR-92a showed the potential to induce blood vessel growth and functional recovery . A systemic antagomiR was also used to inhibit the antiangiogenic miR-24 in the mouse ischemic myocardium, which resulted in improved myocardial angiogenesis and cardiac function . These data were reproduced and expanded by us using a different inhibitory approach, namely the intramyocardial delivery of an adenovirus vector carrying a decoy for the seed sequence of miR-24 . The antiangiogenic miR-100 also regulates the vascular response to LI. Differently from other antiangiogenic miRs, miR-100 expression is reduced in mouse ischemic muscles, where further antagomiR-mediated miR-100 inhibition showed a therapeutic potential . The miR-34 family members (miR-34a, miR34b, and miR-34c) share a common seed sequence and are associated with a cardiac senescence program and with impaired angiogenesis in the MI setting. Indeed, systemic inhibition of the miR-34 members by an 8-mer locked nucleic acid (LNA) (a LNA sequence targeting the miR seed sequence) promoted post-MI angiogenesis in mice . Diabetes mellitus contributes to the prevalence and severity of ischemic disease, through acceleration of atherosclerosis and induction of microangiopathy. Moreover, diabetes compromises the native neovascularization response which helps restore tissue perfusion following an ischemic event. We have reported increased miR-503 expression in ECs exposed to an environment mimicking diabetes mellitus, associated with ischemia . Overexpression of miR-503 in ECs under normal culture conditions resulted in impaired cell proliferation, migration, and ability to form networks. Moreover, local miR-503 inhibition (by adenovirus-mediated miR-503 decoy delivery) improved post-LI vascular repair and blood flow recovery in diabetic mice. Direct targets of miR-503 include cell division cycle 25 homolog A and cyclin E1 . This study was the first to show an association between miRs and diabetes mellitus-induced impairment of postischemic angiogenesis.
miRs Regulating PC Functions
As discussed above, miRs regulate angiogenesis acting on mature vascular cells. miRs can regulate vascular growth via additional mechanisms, including acting on vascular PCs derived from the bone marrow (BM) or other tissues. Table 2 summarizes the miRs which affect the behavior of stem and progenitor cells holding angiogenesis potential. Several lines of evidence have demonstrated that ECs and primitive hematopoietic stem cells (HSC) arise from a common progenitor, the CD34+ hemangioblast cell. Furthermore, definitive HSCs are generated from a specialized EC, known as a hemogenic endothelial cell (HE) [29, 30]. The efficient generation of unlimited supplies of HEs with EC and HSC capacity from pluripotent stem cells (SCs) may provide alternative therapeutic strategies for cell therapies with wider application (see Future Applications). In order to achieve this, it is crucial to understand the mechanisms governing the bifurcation of the HE to HSC and ECs. Additionally, if it is possible to determine what developmental cues direct HE bifurcation to HSC lineage, they can be suppressed in order to optimize and enhance EC production for angiogenic cell therapy. We expect miRs regulate this critical cell fate decision. A recent study in Xenopus embryos has provided preliminary evidence that miR production is crucial for the specification of the hemangioblast precursors of HSCs, and has identified miR-142-3p as master regulator of HSC lineage specification . These findings have subsequently been confirmed in zebrafish and mouse models .
miRs also govern the function of adult PCs. For example, diabetes mellitus reduced the abundance of CD34pos-PCs in the human BM, possibly through inhibition of miR-155, which promotes cell survival through inhibition of FOXO3a . Vice versa, increased miR-15 and −16 levels compromised the regenerative potential of BM-derived circulating proangiogenic cells (PACs, previously denoted “early” endothelial PCs) in patients with critical LI . Moreover, recent evidence suggests that augmented miR-126 expression contributes to the impairment of PAC regenerative capabilities in patients with diabetes mellitus [35, 36]. In addition, miR-126 and −130a were found reduced in PACs from heart failure patients, which was associated with a compromised cardiac repair capacity of the cells , while miR-34 was increased in PACs from MI patients . miR-10*, miR-21, and the miR-34 family have also been associated with PACs senescence [38-40].
Finally, the proangiogenic miR-132 is expressed in human vena saphena-derived pericyte PCs  (aka the “Bristol pericytes” ), where it is upregulated by hypoxia. miR-132 released by these cells is uptaken by ECs, making them undergo a stronger angiogenesis response on Matrigel . We proved the therapeutic potential of the Bristol pericytes in mouse LI and MI models [39, 41]. Of note, the ex vivo pretreatment of these cells with an anti-miR-132 reduced their capacity to improve reparative angiogenesis and cardiac remodeling and function after transplantation in mice with MI . This suggests the contribution of transplanted cell-released miR-132 in the therapeutic effect of the Bristol pericytes.
miRs in Endothelial Differentiation of Pluripotent Stem Cells
The efficient generation of unlimited supplies of vascular cells from pluripotent SCs would prove beneficial in regenerative medicine. In vitro differentiation of pluripotent SCs into vascular cells has been accomplished through use of different protocols. Moreover, transplanted embryonic SC (ESC)-derived ECs have been reported to be able to integrate into the host circulation , and to promote blood flow recovery in animal models of ischemia [43-45]. Human-induced pluripotent SCs (iPSCs) also proved able to differentiate into ECs  and to promote postischemic vascular repair in mice . However, moving this research toward clinical intervention poses great scientific challenges. Cell differentiation from pluripotent SCs during in vivo development is a complicated and poorly defined process governed by various molecular signal pathways. Understanding the underpinning mechanisms might help toward the optimization of protocols for in vitro SC vascular differentiation. The roles played by miRs in the regulation of vascular differentiation are partially understood. The implementation of miR manipulation strategies to alter miR expression and/or function could present unique potential for directing differentiation to or for maintaining cells in a given cell specification, in order to develop new therapeutic strategies. We and others have established in vitro protocols for human ESC (hESC) differentiation into endothelial PCs [48-50]. Using a microarray approach, we have profiled miR expression in hESCs at subsequent stages of differentiation from pluripotency to ECs. This identified several miRs with increased expression as the differentiation progressed and three of these, miR-99b, −181a, and −181b, were further investigated . Lentiviral-mediated overexpression of these miRs in pluripotent hESC enhanced the efficiency of the differentiation protocol, therefore suggesting the miRs as possible contributors to the EC lineage commitment . Moreover ECs derived from hESCs which overexpress each of these miRs showed an improved capacity to promote postischemic blood flow recovery and increase microvascular density in mouse ischemic muscles . Owing to the potential to use pluripotent SC-derived ECs for cell-based therapies there are several laboratories assessing the involvement of angiogenesis-associated miRs  in EC differentiation from SCs. However, there are only a few published studies to date. For example, miR-126 is abundantly expressed in mesodermal PCs derived from pluripotent SCs  and is intimately involved in regulating mature EC function via vascular endothelial growth factor (VEGF) signaling . By contrast, the lack of an embryonic lethal phenotype with miR-126 deletion indicates that miR-126 does not have an essential role in differentiation and maturation of ECs . This has been corroborated with reports that miR-126 overexpression in pluripotent SC impaired EC differentiation, indicating that miRNA-126 does not command early EC lineage commitment but is induced during vascular differentiation . The miR-17–92 cluster and its individual components are also expressionally regulated during the induction of EC differentiation of pluripotent SC. However, similarly to miR-126, the miRNA-17–92 cluster members did not show to functionally impact on the endothelial differentiation of pluripotent SCs . To date, there is no evidence for a single miR possessing a pivotal role in the governance of EC fate decisions despite the demonstration of differentially expressed miRs during in vitro endothelial differentiation of SCs . Another topic that is of high relevance for scientists wishing to use pluripotent SCs for derivation of therapeutic ECs is the endothelial specification to venous, arterial, or lymphatic cells. Several genes are involved in this important stage of vascular development process [57-59], but their expressional regulation is less understood. It is possible that cell specification during the later stages of development may be controlled by miRs. For example, miR-181a  and miR-31  inhibit the expression of Prox1, a homeobox transcription factor required for lymphatic endothelial specification and maintenance. Overexpression of miR-181a reprogrammed lymphatic ECs toward a blood vascular phenotype in vitro . We reported that miR-181a knockdown induces Prox1 expression, while miR-181a overexpression inhibits Prox1 and augments the expression of arterial-specific genes in hESCs directed toward an endothelial lineage . Moreover, in Xenopus and Zebrafish embryos, miR-31 overexpression impaired development of the lymphatic system via direct inhibition of Prox1 . In order to understand the functional role of a miR in endothelial fate determination, it would be necessary to identify and validate target genes involved in developmental specification, including genes that may facilitate differentiation to alternative germ layers or mesoderm cell lineages as well as indirectly induce appearance of endothelial genes.
In summary, although there are a plethora of studies reporting miR expression profiles in ECs, it is unknown if these findings also reflect the role of miRs in the differentiation of pluripotent SCs to ECs. With profound changes in gene expression required to derive ECs from ESCs and iPSCs, we speculate that additional miRs might be involved. A better understanding of the miRNA regulation and processes involved in deriving ECs from pluripotent SCs is expected to facilitate development of more efficient protocols to progress from the bench to clinical cell products. One caveat to studies using pluripotent SC-derived vascular cells is that they are impaired toward identifying miR targets as most computationally predicted target genes will not be expressed in pluripotent, mesoderm, or vascular cells and their precursors. It may prove necessary to screen miRs across several next generation sequencing (small RNA-seq) and array platforms, including cells from all stages of embryogenesis and vascular development. This will facilitate correct identification of miRs involved in the fate decision from pluripotency to ECs.
The rationale behind therapeutic angiogenesis is to deliver blood flow through newly developed vessels in order to overcome tissue ischemia induced by vascular obstruction . Although there are several proangiogenic growth factors in mid/late phase clinical trials, the initial results are unsatisfactory. This may be in part due to the lack of penetrance with administration of the single factor selected so far for therapy, in addition to the use of less than optimal delivery systems. Therapy based on enhancing the “right” miR or combined miRs may provide improvements by targeting multiple genes and pathways. The various modulation strategies are shown in Figure 1. Moreover, it is possible to intervene therapeutically to inhibit pathogenic and antiangiogenic miRs. Candidate miRs for inhibition may include miR-15a/b, −24, −34a/b/c, 92a, and −100 [17, 24, 26, 27, 62]. As aforementioned, systemic delivery of either specific antagomiR or LNA anti-miRs was sufficient to induce efficient expressional inhibition of these miRs in vivo and resulted in therapeutic benefit in mice with either MI or LI [17, 24, 26]. The capability for targeting multiple potential “antiangiogenic” miRs belonging to the same family (and hence sharing the common seed sequence), such as the miR-16 family , at once is one of the possible advantages of the 8-mer LNA technology. In fact the 8-mer LNA targets the miR seed sequence . Moreover, the LNA specificity for the miR is usually higher than in antagomiR, which means that there are less off-target effects and a lower administration dose is sufficient to elicit the same in vivo response . Moreover, the previously mentioned “miR decoys” (aka “miR sponges”) are competitive inhibitors which function by a single piece of RNA containing multiple, tandem-binding sites for a miR seed family of interest . Recently, we demonstrated the efficacy of this approach for inhibiting miR-503 and −24 [25, 28]. Moreover, from a clinical translational perspective, in order to improve safety of miR therapies by minimizing the risk for off-target effects, we believe that local miR inhibition is to be favored over a systemic delivery approach.
Few overexpression strategies have been reported, most likely owing to the caveat that miR mimic oligos have a transient effect, are unstable, and would require multiple administrations. This raises the issue of delivery technologies and methods of administration. For example, minicircle technology has successfully been used to overexpress miR-210 in order to improve cardiac neoangiogenesis in a mouse MI model . Adeno-associate viral vectors have also been used to overexpress miRs-590-3p and miR-199a-3p which are able to promote post-MI cardiac regeneration . Furthermore, we demonstrated that lentiviral-mediated overexpression of miR-99b or −181b could induce enhance the therapeutic efficacy of hESC-derived ECs [48, 66]. Alternative strategies may include vessel-targeted nanoparticles to systemically deliver the miRNA mimic (or antagomiR) directly to the site of disease . It should be noted that cells are able to shed biologically active miRs via micro- and nanovesicles, which can then be uptaken by neighboring or distant cells (reviewed in ). This should suggest caution in using miR mimic approaches which may release miRs with potential to elicit pathogenic side effects in other tissue. For example, proangiogenic miRs could be released in the circulation by the targeted ischemic tissue and promote unwanted angiogenic responses at different locations.
One potential avenue that has not been extensively explored is combining miR-modulation in cells which would be used for cell-therapy strategies. For example, the transplantation of endothelial PCs into sites of ischemia with modulated expression of miRs may promote healing, potentially with a reduced cell load. For example, we have demonstrated that either PACs with ex vivo inhibited miR-15a and −16  or hESC-EC overexpressing either miR 99b or miR181b  promotes stronger vascular repair responses when trialed in a immunodeficient mouse LI model. This miR-mediated increase in neoangiogenesis efficiency may be a combination of a direct cell effect and miRNA-mediated proangiogenic paracrine mechanisms. This presents a novel strategy for combining cell and miR-based therapies (Fig. 2). Moreover, it may be possible to combine several modulation strategies to harness the temporal expression of miRs depending on the modus operandi of the disease to be targeted, for example, the potential to transplant cells to disease sites which may require multiple cell types for clinical efficacy. This would mean that it may be possible to inject a progenitor type cell previously subject to a miR modulation strategy that contains either an inducible or suicide cassette. This would permit stringent governance of temporal expression (in order to further direct cell lineage specification of the PCs), in addition to constitutive miR expression to target the dysfunctional gene regulation.
It should be noted that it is important to proceed with some caution in implementing the exciting miR targeting technologies as we still lack a complete understanding of all the processes required, the mechanisms by which all the regulatory components are integrated, the spectrum of targets, and as yet unidentified compensatory mechanisms in vivo. These, however, are exciting challenges and if the community can develop specific and efficient delivery systems which can differentiate between physiological and pathological angiogenesis, we may be able to harness miRs to significantly advance clinical interventions for a plethora of diseases.
Table 1. Principal microRNAs (and their target genes) affecting the angiogenesis process
A.J. Thrasher is a GOSHCC Consultant in Pediatric Immunology, a Wellcome Trust Senior Clinical Fellow and a National Institute of Health Research (NIHR) Senior Investigator. G.D. Angelini is a British Heart Foundation (BHF) Professor in Cardiac Surgery and a NIHR Senior Investigator. C. Emanueli is a BHF Senior Research Fellow and she is member of the Leducq Transatlantic Network in vascular miR-based therapeutic strategies in vascular disease (MIRVAD). This work was funded by the BHF Fellowship of C. Emanueli and the BHF Centre of Vascular Regenerative Medicine (to C. Emanueli), and supported by researchers (G.D. Angelini and C. Emanueli) at the National Institute for Health Research Bristol Cardiovascular Biomedical Research Unit.
N.M.K.: conception and design and manuscript writing; A.J.T. and G.D.A.: manuscript editing and critical discussion; C.E.: conception and design, manuscript writing, financial support, and final approval of manuscript.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.