Coronary vessel development and insight towards neovascular therapy

Abstract

Formation of the coronary arteries consists of a precisely orchestrated series of morphogenetic and molecular events which can be divided into three distinct processes: vasculogenesis, angiogenesis and arteriogenesis (Risau 1997; Carmeliet 2000). Even subtle perturbations in this process may lead to congenital coronary artery anomalies, as occur in 0.2–1.2% of the general population (von Kodolitsch et al. 2004). Contrary to the previously held dogma, the process of vasculogenesis is not limited to prenatal development. Both vasculogenesis and angiogenesis are now known to actively occur within the adult heart. When the need for regeneration arises, for example in the setting of coronary artery disease, a reactivation of embryonic processes ensues, redeploying many of the same molecular regulators. Thus, an understanding of the mechanisms of embryonic coronary vasculogenesis and angiogenesis may prove invaluable in developing novel strategies for cardiovascular regeneration and therapeutic coronary angiogenesis.

Cellular mechanisms of coronary vasculature formation

Coronary vessel formation is an elaborate process involving three distinct yet overlapping steps, namely vasculogenesis, angiogenesis and arteriogenesis. Vasculogenesis is defined as the formation of primitive vascular structures via the differentiation of endothelial precursor cells. Angiogenesis is the generation of new microvessels by endothelial proliferation and migration, either by intussusception, to divide the vessel in two or, by sprouting, to form new branches. Arteriogenesis describes the remodelling that occurs to form mature arteries by migration of supporting smooth muscle cells (SMCs) and pericytes from the epicardium during development. Collateral growth, a specialized type of arteriogenesis, usually refers to the formation of mature arteries from pre-existing interconnecting arterioles following coronary artery occlusion.

Heart development begins with the formation of a primitive tube, an avascular structure with myocardial and endocardial layers. At this early embryonic stage the heart receives sufficient oxygen and nutrients by simple diffusion across the endocardial wall, however, as the myocardium increases in size and the heart transitions to a more complex multilayered organ, diffusion becomes impractical and the heart requires a dedicated vascular system.

Much of what we know about the origin of cells that make up the coronary vascular network derives from immunohistochemical analyses, quail-chick chimera experiments, retroviral labelling, gene targeting experiments and, more recently, lineage tracing studies. Most coronary vascular cells stem from the epicardium, the outermost layer of the heart, following epithelial to mesenchymal transition (EMT) to form epicardium derived cells (EPDCs). Epicardial precursors arise from a transient extracardiac mesothelial cell population, the proepicardium (PE) situated on the surface of the heart, between the sinus venosus and liver in chick, and on the surface of the septum transversum in mouse (Viragh & Challice 1981).

Proepicardium development and migration varies between species. In murine models, PE development proceeds in a bilaterally symmetric pattern (Manner et al. 2001; Schlueter et al. 2006) following the first appearance of the left and right PE anlage at embryonic day (E)8.5. As each anlage matures, they merge to form a single PE that is fully developed by E9.5. This is in contrast to the chick (Schulte et al. 2007) and Xenopus (Jahr et al. 2008), where PE development occurs in a bilaterally asymmetric pattern; in chick, the right PE anlage first appears at HH-stage 14 whereas the left PE does not appear until HH-stage 15/16. Only the right PE anlage matures while the left remains undifferentiated (Schulte et al. 2007).

Once formed, PE mesothelial cells migrate to the developing heart and envelop the surface to give rise to the primitive epicardium and a matrix-rich subepicardial space (Viragh & Challice 1981; Hiruma & Hirakow 1989; Viragh et al. 1993). The epicardium is not derived entirely from the PE; epicardium in the region of the outflow tract is derived from the coelomic pericardial epithelium near the aortic sac (Perez-Pomares et al. 2003). In the avian system, PE connection to the myocardium is via the attachment of proepicardial villi which form a transient bridge structure followed by cellular migration of PE cells (chick HH-stage 17/18) (Nahirney et al. 2003). This is in contrast to mammals where PE translocation (mouse E9.5) is proposed to occur via differential growth of proepicardial projections and the release of free-floating epicardial aggregates (Rodgers et al. 2008).

Epicardial development is intimately associated with the development of the coronary vascular plexus. As the epicardium expands and migrates over the heart, a subpopulation of epicardial cells delaminate from the primitive epicardial epithelium and undergo an EMT generating a population of migratory mesenchymal cells, EPDCs, that populate the subepicardial space and subsequently the myocardium (Mikawa & Fischman 1992) (Mikawa & Gourdie 1996; Perez-Pomares et al. 1997, 1998; Gittenberger-de Groot et al. 1998). Epicardial EMT begins at the base of the heart (E11.5 in mouse) and proceeds in a wave-like pattern towards the apex (Lavine & Ornitz 2007).

It is generally accepted that EPDCs give rise to coronary SMCs, pericytes, fibroblasts and cardiomyocytes (Mikawa & Fischman 1992; Mikawa & Gourdie 1996; Dettman et al. 1998; Gittenberger-de Groot et al. 1998; Vrancken Peeters et al. 1999; Wada et al. 2003). As well as providing coronary vascular cells and cardiomycoytes, EPDCs have been suggested to play a role in myocardial compaction, purkinje fibre development and inhibition of endocardial EMT (reviewed in (Winter & Gittenberger-de Groot 2007b). Whether EPDCs are a source of coronary endothelial cells (ECs) is highly contentious. While retroviral tagging, adenoviral cell lineage tracing and quail-chick chimera studies have provided evidence to suggest that coronary ECs are derived from the epicardium (Mikawa & Gourdie 1996; Munoz-Chapuli et al. 1999, 2002; Perez-Pomares et al. 2002), it is widely accepted that not all coronary vascular ECs are contributed by the epicardium (Morabito et al. 2002). Indeed, an increasing number of studies now argue that very few, if any, coronary ECs are epicardium-derived (Winter & Gittenberger-de Groot 2007a; Cai et al. 2008; Zhou et al. 2008a). Quail-Chick chimera and retroviral labelling studies identified quail endothelial-specific marker (QH-1) positive angioblasts in the PE and subepicardial space. However, it is not clear whether they were transplanted with the PE or whether they represent a migratory cell population from a tissue, such as the liver, that pass through the PE on their way to the subepicardial space (Poelmann et al. 1993, 2002; Dettman et al. 1998). The liver is suggested by some to be a primary source for proepicardial endothelial precursor cells (Lie-Venema et al. 2005). An alternative viewpoint proposed is that the coronary endothelium originates by invagination from the endocardium (Viragh & Challice 1981). The difficulty in interpreting these studies is that endocardial cells closely resemble endothelial precursor cells in terms of the molecular markers expressed, thus confounding lineage trace and chimera analyses (Lie-Venema et al. 2005).

Regardless of origin, endothelial precursors within the subepicardial space migrate over the heart in the same direction as epicardial growth (in an anterior and ventral progression). Once endothelial precursors have completed their migration they coalesce to form a primitive vascular plexus such that by E13 in mouse, vascular clusters can be observed over the myocardium (Morabito et al. 2002). At this time blood cell precursors are found within the EC vessels; the origin and recruitment of hematopoietic precursors are only partially understood and discussed elsewhere (Loose et al. 2007). The primitive vascular plexus undergoes extensive remodelling and patterning to form a network that begins to resemble the mature coronary artery tree. Vessel growth is directed towards the base of the heart eventually connecting with the aorta. Upon perfusion, the capillaries are remodelled into larger vessels via vascular wall matrix enrichment and recruitment of SMCs and pericytes (arteriogenesis or collateral growth). Coronary SMCs originate from the PE or neural crest depending on their location within the heart. Neural crest gives rise to proximal coronary arteries as well as large arteries in the head and neck region, whereas the rest of the coronary artery smooth muscle stems from the epicardium (Luttun & Carmeliet 2003). Apoptosis is critical for appropriate arteriogenesis and proper development of the coronary vasculature (Poelmann & Gittenberger-de Groot 2005).

Expansion of the coronary vasculature continues postnatally to accommodate the increased myocardial thickness that results from physiological cardiomyocyte hypertrophy. Murine capillary density increases from three- to fourfold in the first 3 weeks and the number of SMC-covered coronaries increases tenfold over the same period (Luttun & Carmeliet 2003). Vessels also undergo extensive remodelling as they acquire adult branching characteristics and begin to resemble adult coronary arteries and veins.

Molecular mechanisms of coronary vasculature development

As summarized above, coronary vasculature formation depends upon a series of carefully regulated steps: (i) the development of the PE; (ii) formation of the epicardium via PE migration; (iii) derivation of EPDCs from epicardium by EMT and their migration prior to development of the primitive vascular plexus (vasculogenesis); (iv) expansion of the primitive vascular plexus (angiogenesis) and remodelling to form a mature coronary vessel network (arteriogenesis). Gene targeting experiments have offered considerable insight into the molecular mechanisms that regulate each of these steps to ensure formation of a healthy, functional coronary vasculature. Because of the vast number of factors involved and the complexity of the network of interactions between them, a detailed description of all molecules that function to regulate coronary vasculogenesis is beyond the scope of this review. However, the principle molecules involved are depicted in Figure 1 and listed in Table 1, with references for further information.

Figure 1.

Mechanisms of coronary vessel development. Coronary vasculogenesis begins with formation of the proepicardium (PE), a process which requires Nkx2.5 and Gata4. Epicardial precursors migrate from the PE to overly the myocardium, controlled by factors including retinoic acid, angiopoietin 1, Tbx5, Sp3 and WT1. Epicardium-derived cells (EPDCs) are formed by an epithelial to mesenchyme transition (EMT) in response to FGF and TGFβ signalling from the myocardium. Epicardial EMT is further regulated at the gene expression level by FOG-2 and BAF180 and by subepicardial matrix molecules, notably podoplanin. EPDCs are induced to migrate into the myocardium by myocardially secreted Tβ4 where they respond to angio- (VEGF/FGF-2) and arteriogenic (PDGF/TGFβ) factors to form coronary vessels.

Table 1.

Factors involved in regulating coronary vessel development

Factor	Phenotype/Role	References
Formation of the proepicardium
Gata4	Gata4 nulls lack PE	Watt et al. 2004
Nkx2.5	Nkx2.5 nulls have abnormal PE	Zhou et al.2008b
Formation of the epicardium from the proepicardium
Tbx5	Mice overexpressing and null forTbx5 have failed PE migration. Die at E10.5	Hatcher et al. 2004
Alpha 4 integrin	Nulls lack epicardium and coronary vasculature. Die at E11	Yang et al. 1995
Wt1	Nulls lack epicardium, lack coronary vasculature precursors	Moore et al. 1999; Kreidberg et al. 1993
VCAM	Nulls lack epicardium	Kwee et al. 1995
RxR	Nulls have disrupted PE-epicardium transition	Jenkins et al. 2005; Merki et al. 2005
Bves	Bves antibodies inhibit epicardial migration of PE in vitro	Wada et al., 2001
Ang1	Nulls die at mid-gestation displaying defective or absent epicardium and lack of coronary vasculature	Ward et al. 2004b
Epicardial EMT and derivation of epicardium-derived cells
FOG2	Fog2 nulls have no coronary vasculature and fail to undergo EMT	Tevosian et al. 2000
α4 integrin	Antisense α 4 -accelerated EMT, deeper EPDC migration, SMC fate inhibited	Dettman et al. 2003
RxRα	Epicardially restricted RXRα mutant mice have a thin myocardium, a detached epicardium, reduced subepicardium and abnormal arterial branching	Merki et al. 2005
TgfβR2/Tgfβ1	Nulls display failed vasculogenesis and embryonic lethality	Oshima et al., 1996; Shull et al., 1992
TGFβR1 (ALK5)	Nulls die at mid-gestation with angiogenic defects because of impaired migration of mutant ECs	Larsson et al., 2001
ALK1	Nulls display embryonic lethality at mid-gestation, angiogenic defects	Oh et al., 2000
Endoglin	Nulls have angiogenic defects, poor vascular SMC development and arrested endothelial remodelling	Li et al., 1999
smad5	Nulls die at mid-gestation with angiogenic defects and delayed heart development	Yang et al., 1999; Chang et al., 1999
TGFβR3	Nulls have a defective epicardium, reduced coronary vessels, and die at E14.5	Compton et al. 2007
Sp3	Nulls display embryonic lethality between E14.5-E18.5, epicardial layer abnormally attached, fewer EPDCs in subepicardial space, E-cadherin upregulated, suggesting abnormal EMT	van Loo et al. 2007
Podoplanin	Nulls died at mid-gestation with smaller PE, detached epicardium and fewer EPDCs. E-cadherin up-regulated in PE and epicradium, indicating abnormal EMT.	Mahtab et al. 2008
FGF1,2,7	In vitro epicardial EMT stimulated by addition of FGF1,2,7	Morabito et al. 2001
β-catenin	Proepicardial β-caterin knockdowns lethal between E15 and birth, impaired artery formation, reduced/absent subepicardial space, reduced EPDCs, impaired EPDC differentiation into SMCs	Zamora et al. 2007
Vasculogenesis-EPDC migration
BAF180	Impaired EMT, arrested epicardial maturation at E11.5	Huang et al. 2008
Tβ4	Myocardial Tβ4 knockdowns have thin, non-compacted myocardium, detached epicardium, disrupted coronary vasculogenesis	Smart et al. 2007
wt-1	Nulls display reduced EPDCs, premature EPDC differentiation	Moore et al. 1999; Kreidberg et al. 1993
FGF	FGF pathway activation leads to HH signalling which induces VEGF and Ang2 expression leading to formation of coronary vascular plexus. Redundancy among FGFs	Murakami and Simons, 2008
Periostin	Nulls have delayed/inhibited differentiation of EPDCs into fibroblasts, increased differentiation into SMC and cardiomyocytes	Norris et al., 2008
Coronary artery remodelling – angiogenesis and arteriogenesis
FGF/VEGF	Tube formation inhibited in quail heart explants in vitro	Tomanek & Zheng 2002
PDGF	Nulls die between E8-E16 with severe cardiac malformations and defective coronary arteriogenesis	Lindahl & Betsholtz 1998; Lindahl et al. 1997; Van Den Akker et al. 2005
VEGF	Severe angiogeneic defects; nulls display early embryonic lethality because of severe vascular defects	Carmeliet et al. 1999; Giordano et al., 2001
Connexin43	Coronary artery anomalies, dysregulated coronary vasculogenesis and angiogenesis	Li et al., 2002
HEXIM1	Mutant mice display abnormal coronary patterning, thin ventricular walls, decreased vascularization of myocardium, decreased VEGF and FGF9 expression	Montano et al., 2008

A vast number of factors have been implicated in the different stages of coronary vasculogenesis and angiogenesis. The mechanism of action of a subset of these factors has been elucidated while many more are inferred through gene targeting studies and their precise roles in coronary vessel development remain to be determined. A summary of the defined role or mutant mouse phenotype is given for the principal factors involved.

Formation of the proepicardium

Proepicardium formation, the earliest stage in coronary vasculature development, is only superficially understood at the molecular level; a number of transcription factors have been implicated by virtue of their expression pattern and mutant phenotype but few, if any, downstream targets, the effectors of PE formation, have been identified. The most notable of these is the zinc finger transcription factor, Gata4, which is expressed in the PE and is essential for epicardial development (Watt et al. 2004). Mice deficient in Gata4 lack a PE and consequently fail to develop an epicardium (Watt et al. 2004).

Wilm’s Tumor 1 (Wt1) is expressed in the primitive PE, epicardium and EPDCs. Wt1-traced epicardial progenitors co-express Nkx2.5 and Isl1 and are proposed to be descended from the definitive multipotent precursor population (Laugwitz et al. 2008) that contribute the majority of cardiomyocytes and a subset of smooth muscle and ECs to the heart (Zhou et al. 2008b). Nkx2.5 and Wt1 are not actively co-expressed in the PE but are either sequentially expressed or only transiently co-expressed in a subset of PE precursors. However, there is a requirement for Nkx2.5 expression for normal development of the PE as loss of Nkx2.5 resulted in abnormal PE formation and decreased Wt1 expression (Zhou et al. 2008b).

Members of the T-box family of transcription factors, Tbx5 and Tbx18, along with epicardin (also known as capsulin, pod1, Tcf21) are expressed in the PE but their roles are either undefined or implicated in later stages of PE migration (Moore et al. 1999; Hatcher et al. 2004). T-box genes may, however, sit at the top of a transcriptional hierarchy as Gata4 is reported to be a target of Tbx5 and Tbx5 nulls have reduced Gata4 expression (Plageman & Yutzey 2005).

Formation of the epicardium from the proepicardium

Proepicardial cells can give rise to the epicardium, coronary vasculature, cardiac fibroblasts and myocardium. The induction of the epicardial lineage is governed by a balance between levels of FGF2 (basic FGF, bFGF) and BMP2 (Kruithof et al. 2006). Chick PE cells display multipotency in vitro and give rise to both the myocardial and epicardial lineage, thus offering a good model to study the factors that may favour differentiation into either lineage. BMP2 is expressed in the distal inflow tract myocardium and stimulates cardiomyocyte formation whereas FGF2 is expressed in the PE and stimulates differentiation into the epicardial lineage (Kruithof et al. 2006). Overall, these studies reveal that the extrinsic mechanism which determines the induction of myocardial or epicardial lineages is regulated by the position of cells in the pericardial mesoderm (Kruithof et al. 2006).

Mechanical or genetic ablation of the PE or the blocking of its migration results in the absence of epicardium and defective coronary vasculogenesis. Such studies have been useful in identifying genes that regulate PE migration and epicardial formation.

T-box genes

Tbx5 levels are critical in regulating PE cell migration as both knockdown and overexpression of Tbx5 were shown to prevent PE cell migration resulting in embryonic lethality at E10.5 (Hatcher et al. 2004). Tbx18 is also expressed in the PE and epicardium but the viability of Tbx18 null mice suggests that it is dispensable for epicardial development, possibly due to compensation by other T-box genes (Chen et al. 2002; Hatcher et al. 2004).

Retinoic acid signalling

Tbx5 expression has been shown to be dependant on retinoid signalling in both chick and mouse (Plageman & Yutzey 2005). Retinoic acid (RA) signalling is essential for myocardial development and in coronary vasculogenesis. RALDH2, the enzyme responsible for RA synthesis, and retinoid X receptor α (RXRα) are present in the epicardium. RXRα may function in the early stages of epicardial formation as it has been shown that RXRα knockouts have a disrupted PE-epicardium transition (Jenkins et al. 2005).

Ang-Tie system

Angiopoietins represent a family of endothelial growth factors expressed in the myocardium which act on the endothelial specific receptor, Tie-2, to direct coronary vessel development (Ward et al. 2004a,b). Overexpression of Angiopoietin 1 (Ang1) resulted in mid-gestational embryonic death in 90% of embryos suggesting epicardial and coronary vessel defects. Immunohistochemical analysis revealed myocardial thinning, atrial dilation, defective or absent epicardium and a lack of coronary arteries (Ward et al. 2004b).

Wilm’s tumor-1

Wt1-null mice die at mid-gestation with a defective epicardium that is incompletely formed containing a reduced number of EPDCs, the precursors of coronary vasculature (Kreidberg et al. 1993; Moore et al. 1999; Wagner et al. 2005). Some aspects of WT1 function in coronary vasculogenesis may be mediated by the neurotrophin receptor TrkB, one of its identified targets, to promote vessel formation (Wagner et al. 2005). Transcriptional assays showed that WT1 binds the promoter of NTRK2, the gene that encodes TrkB, and directs expression to the epicardium and myocardial vessels. TrkB is required for normal vascularization. NTRK2 nulls have no epicardial defects but lack a significant portion of their coronary vessels, particularly in the subepicardial space (Wagner et al. 2005).

Cell adhesion molecules

Integral membrane proteins and cell adhesion molecules are instrumental in mediating the cell-cell interactions required for attachment of the epicardium to the myocardium. For example, a critical interaction has been demonstrated between α4 integrin on the primitive epicardium and VCAM (Vascular Cell Adhesion Molecule) in the myocardium (Kwee et al. 1995; Yang et al. 1995). Targeted deletion of α4 integrin in mouse results in a lack of epicardium at E11 and consequently a lack of coronary vessels (Yang et al. 1995). Earlier analysis at E10 confirmed that PE migration and formation of epicardium was not hampered, but lack of α4 integrin caused a failure in the attachment of the epicardium to the myocardium (Manner et al. 2001). Wt1 activates the α4 integrin gene, possibly explaining why α4 integrin and Wt1 nulls share similar phenotypes relating to abnormal epicardium and coronary vessel formation (Kirschner et al. 2006).

Much remains to be understood of the molecular mechanism of epicardial formation as the prerequisite for coronary vessel development. In particular few, if any, of the ‘effector’ molecules that actively establish the epicardium from the pro-epicardium have been identified to date, but may include factors that regulate cell migration or extracellular matrix remodelling. In such a rapidly evolving research field, these pathways are likely to be elucidated in the very near future, perhaps by gene array analyses to identify downstream targets of key transcription factors such as TBX5, GATA4 and WT1 and growth factor/mitogenic signalling mediated by FGFs and RA.

Epicardial EMT and EPDC migration

Reciprocal paracrine signalling between the myocardium and epicardium is important throughout coronary vessel development, many examples of which have been implicated in the EMT that is required to produce the migratory EPDCs for subsequent invasion into the myocardium to differentiate into vascular progenitors.

Role of growth factor signalling in epicardial EMT

Fibroblast growth factors (FGFs), specifically FGF1, 2 and 7, and transforming growth factor-β (TGFβ) are secreted as paracrine signals from the myocardium to positively and negatively regulate epicardial EMT, respectively (Morabito et al. 2001). This myocardial secretion of FGFs and TGFβ has been hypothesized to control the regional restriction of coronary development. That is, the myocardium secretes FGFs in areas where coronary vessel development begins and secretes TGFβ in areas where coronary vessels are not required (Morabito et al. 2002).

TGF-β signalling to stimulate EMT, at least in PE explants, occurs via downstream phosphorylation of ALK2 and is dependent on Smad proteins which transduce signals directly to the nucleus (Morabito et al. 2001; Olivey et al. 2006) although the distinct roles of inhibitory and stimulatory Smads in epicardial EMT remain unexplored. In vitro studies have shown that TGFβ1 regulates EC migration and proliferation via endoglin signalling and ALK1 and ALK5 activation (Goumans et al. 2002, 2003; Lebrin et al. 2004). Experiments in explant culture demonstrate that TGFβ induces loss of epithelial cell character in epicardial cells via ALK5, p38 MAP kinase and p160 rho kinase to promote smooth muscle differentiation (Compton et al. 2006). Mice deficient in TGFβ1 or the type II or III TGFβ receptors die at E14.5 with failed coronary vasculogenesis and epicardial abnormalities, demonstrating the importance of this pathway in coronary vessel development (Compton et al. 2007). Furthermore, mice null for type 1 receptors, ALK5 or ALK1, as well as endoglin and Smad5 have angiogenic defects (Goumans et al. 2002, 2003).

Transcriptional regulation of epicardial EMT

Friend of GATA (FOG) proteins are cofactors for GATA family proteins and the GATA/FOG complex plays an essential role in coronary vascular development; accordingly, Fog-2 null embryos share numerous similarities with the Gata4 mutant phenotype (Smagulova et al. 2008). Fog-2 nulls die at mid-gestation with an intact epicardium and appropriate expression of epicardial-specific genes yet lack coronary vasculature, attributed to defective epicardial EMT (Tevosian et al. 2000). Aspects of this vascular phenotype were rescued by transgenic re-expression of Fog-2 in the myocardium, implicating FOG-2 in downstream paracrine signalling (Tevosian et al. 2000; Smagulova et al. 2008). One of the downstream mediators of GATA 4/FOG-2 function in the epicardium is the LIM homeodomain protein, Lhx9 (Smagulova et al. 2008). Truncated Lhx9 (Lhx9αβ) is expressed in the PE and septum transversum at E9.5, becomes restricted to the epicardial mesothelium by E11.5, and down-regulated thereafter. It has been proposed that Lhx9αβ expression is down-regulated by FOG-2 to allow epicardial cells to mature. Fog-2 nulls can not repress Lhx9αβ thus EPDC differentiation is impaired (Smagulova et al. 2008).

Subepicardial matrix molecules are critical for epicardial EMT

The subepicardial space, situated between the epicardium and myocardium, is rich in fibronectin, collagens (I, IV, V and VI) proteoglycans, laminin, vitronectin, fibrillin, elastin, tenascin-X and flectin (reviewed in Manner et al., 2001). Interactions between epicardial cells and components of the subepicardial matrix also provide signals that induce epicardial EMT (Dettman et al. 2003). The notion that adhesion molecules can regulate EMT is supported by a quail-chick chimera study in which epicardial cells infected with adenovirus expressing antisense α4 integrin showed accelerated EMT and deeper migration of EPDCs into the myocardium providing evidence that α4 integrin represses EMT invasion and migration. Infected epicardial cells were also shown to be inhibited from forming SMCs and restricted to an interstitial fibroblast phenotype, indicating that regulators of epicardial EMT, such as α4 integrin, can also influence cell fate (Dettman et al. 2003). The importance of the subepicardial space is further demonstrated in conditional mutant mice lacking proepicardial β-catenin, which display impaired coronary artery formation and embryonic lethality between E15 and birth. Analysis of mutants revealed normal PE and epicardial formation, however, the subepicardial space was absent and EPDC differentiation was impaired suggesting a role for the epicardial β-catenin pathway in maintenance of the subepicardial space and provision of matrix components to promote epicardial EMT (Zamora et al. 2007). RXRα signalling in the epicardium is critically required for induction of EMT via a FGF-2/Wnt9b/β-catenin pathway. Epicardially restricted RXRα mutant mice have a thin myocardium, a detached epicardium, a reduced subepicardial component and abnormal arterial branching (Merki et al. 2005).

Subepicardial matrix components change as coronary vasculogenesis proceeds, revealing its influence on vascular fate (Manner et al. 2001). The matrix is initially rich in vitronectin, then fibronectin is deposited which may facilitate migration of EPDCs and endothelial precursors (Adams & Watt 1993). Laminin deposition follows, which coincides temporally with endothelial tube formation, and subsequently collagens (type IV first, followed by types I and III) are deposited (Rongish et al. 1996). Fibulin is deposited by the epicardium as it migrates over the heart and is produced by ECs of budding coronary arteries (Morabito et al. 2002). These changes in matrix composition that occur during development are recapitulated in the adult heart in the extracellular matrix around the regions of active coronary collateral remodelling postmyocardial infarction (MI) (Tyagi 1997).

Podoplanin, a transmembrane glycoprotein is expressed in the PE, epicardium and in a subset of myocardial cells. Podoplanin knockout mouse embryos die at mid-gestation with multiple cardiac defects including a smaller PE, atypical epicardial spreading and its detachment from the myocardium, demonstrated by patchy WT1 staining (Mahtab et al. 2008). Podoplanin is proposed to stimulate EMT by negatively regulating E-cadherin. Abnormal EMT in podoplanin null mice, as evidenced by elevated E-cadherin, resulted in fewer EPDCs and coronary artery anomalies at E14.5 because of SMC deficiency (van Loo et al. 2007).

While epicardial EMT would appear to be one of the better defined steps of coronary vasculogenesis, some questions remain. The precise relationship between EMT, migration and differentiation of epicardial cells remains, at best, ambiguous. A number of studies infer normal or defective epicardial EMT based on migratory capacity of EPDCs, either in vitro or in vivo (Smart et al. 2007; Huang et al. 2008). Certainly, EMT is required to transform non-motile epicardial cells into highly invasive EPDCs, yet it is still important to discriminate a failure in EMT vs. a failure in EPDC migration. The identification of suitable mesenchymal markers may facilitate this. Similarly, it is generally assumed that epicardial cells are required to undergo EMT before they can differentiate, yet differentiation of epicardial cells into ECs and SMCs occurred in BAF180 mutant mice despite their reported failure in EMT (Huang et al. 2008). Clarification of the relationship between EMT, migration and differentiation in the epicardium is essential, not only for a complete understanding of the embryonic processes underlying coronary vasculogenesis but, moreover, in extrapolating to the adult and the mechanisms required for neovascularization.

EPDC migration

Following their delamination from the epicardium, EMT confers upon EPDCs a greatly enhanced capacity to migrate into the myocardium wherein they are stimulated to differentiate into precursors of the coronary vasculature. Although primed for migration, EPDCs require specific stimuli to promote cell motility. One of the few identified factors shown to promote EPDC migration during coronary vasculogenesis is the actin-binding peptide, Thymosin β4 (Tβ4) (Smart et al. 2007). Myocardial-specific knockdown of Tβ4 in the developing heart revealed its essential role in regulating all three key stages of coronary vessel development. Embryos with reduced myocardial Tβ4 levels displayed a number of striking cardiac defects at E14.5, including a thin non-compacted myocardium and a detached epicardium which was mottled with abnormal surface nodules, representing aberrant vessels. Disrupted coronary vasculogenesis was apparent after immunostaining of coronary ECs with the endothelial specific receptor, Tie2 (Ward & Dumont 2002) and SMCs with smooth muscle α-actin (SMαA). In contrast to the weak myocardial staining, the epicardial nodules of mutant embryos were intensely stained with Tie2 and the subepicardial space was abundant with SMαA positive cells, suggesting that EPDCs, fated to form endothelial and SMCs, failed to migrate into the myocardium to form coronary vessels and instead activated their respective differentiation programmes in situ within the epicardium. Tβ4 was shown to provide an essential paracrine signal from the myocardium to act on EPDCs to promote their migration into the myocardium for coronary vasculogenesis.

Similarly, ablation of BAF180, an ATP-dependent SWI/SNF chromatin remodelling component abundantly expressed in the PE and epicardium, leads to impaired EMT and arrested epicardial maturation at around E11.5 (Huang et al. 2008). Embryos had an underdeveloped subepicardial space and a reduced number of mesenchymal cells. In vitro analysis of BAF180 null epicardial explants revealed decreased EMT and migration potential. FGF, TGF and VEGF (vascular endothelial growth factor) pathways were down-regulated in these mutant hearts indicating a role for BAF180 as a regulator of a number of genes essential for coronary development.

Coronary artery remodelling – angiogenesis and arteriogenesis

The process of coronary vasculogenesis in the embryo is complete shortly after mid-gestation but the arterial remodelling that subsequently ensues, via angiogenesis and arteriogenesis, continues throughout development and for the first 3 weeks of postnatal life in mouse (Luttun & Carmeliet 2003). Angiogenesis consists of a number of defined steps: ECs degrade their basement membrane and detach from the underlying ECM and neighbouring cells, they invade the extravascular space, migrate and proliferate towards a stimulus, further invading and dissolving the ECM. The ECs then restructure to form a lumen and secrete new basement membrane.

The roles of VEGF and FGF in coronary angiogenesis

By far the best characterized molecular regulators of coronary angiogenesis are VEGF and FGF-2 (Tomanek & Zheng 2002) and, as will be discussed later, an understanding of the developmental role for these factors has led to their use, in clinical trials, for treating patients with coronary artery disease. FGF-2 and VEGF stimulate vascular growth in the embryonic heart and the rate of vascularization is related to the rate of myocardial growth (Tomanek & Zheng 2002). Addition of anti-FGF-2 or anti-VEGF antibodies inhibit tube formation in quail embryonic heart explants and, when these antibodies were administered to rat pups, reduced capillary density was observed while reduced arteriolar density was observed only with anti-FGF-2, but not anti-VEGF (Tomanek & Zheng 2002).

Deletion of a single allele of VEGF-A causes early embryonic lethality because of severe vascular deficiency which precludes its use to study the role of VEGF in later stages of coronary vessel development (Carmeliet et al. 1996; Ferrara et al. 1996; Morabito et al. 2001) VEGF-A exists as a number of alternatively spliced forms (predominantly VEGF₁₂₀, VEGF₁₆₄ and VEGF₁₈₈) in mouse. Mice expressing the VEGF₁₂₀ isoform alone survive to birth but die by postnatal day 14 as a result of a failure to increase capillary or coronary artery density after birth, indicating that VEGF₁₂₀ is sufficient for vasculogenesis but that other isoforms are additionally required for postnatal angiogenesis (Carmeliet et al. 1999). Decreased formation of SMCs surrounding the coronary arteries of these mice correlated with reduced levels of platelet-derived growth factor (PDGF)-B and PDGFRβ in VEGF_120/120 hearts (Carmeliet et al. 1999), revealing a possible regulation of PDGF signalling by VEGF.

Hypoxia and mechanical forces regulate coronary angiogenesis

It is widely understood that vessel growth is triggered in hypoxic environments. Hypoxia-induced up-regulation of hypoxia inducible factor (HIF) 1-α activates VEGF (Forsythe et al. 1996), specifically the VEGF₁₂₂, ₁₆₆ and ₁₉₀ splice variants, and increases vessel tube formation. An epicardial to endocardial gradient of tube formation, correlating with levels of VEGF and HIF1α colocalization, supports the hypothesis of hypoxic regulation of coronary vasculogenesis and angiogenesis (Tomanek et al. 1999). Myocardial mechanical forces, such as the shear stress associated with blood flow and stretch because of diastolic filling, are also potential stimulators of vasculogenesis, angiogenesis and arteriogenesis. In vitro culture of ECs in conditioned media from stretched cardiomyocytes, in which VEGF was up-regulated, resulted in increased EC proliferation, migration and tube formation, via paracrine signalling (Zheng et al. 2001). Stretched ECs also induced VEGFR-2 (FLK1) and Tie-2 up-regulation providing evidence that stretch triggers angiogenic growth factors and receptors in both ECs and cardiomyocytes.

Angiogenesis is associated with myocardial growth and remodelling

Myocardial growth and vasculogenesis are intimately linked. Cardiomyocyte proliferation is controlled by a number of epicardial mitogens such as RA, FGFs and TGF-β, which activate myocardial FGFRs (Brutsaert 2003). As the heart grows, it stretches and creates a hypoxic environment, both factors which, as described above, induce VEGF. A number of loss-of-function mutant embryos with epicardial defects also display myocardial thinning, including WT1, VCAM, RXRα, Τβ4 and α4 integrin. Some of these factors, such as WT1 have been proposed to play a direct role in myocardial compaction, whereas in the majority of mutants with defective coronary vasculature, failure in myocardial compaction is likely a manifestation of a disruption in the relationship between vasculogenesis and myocardial growth or the reciprocal epicardial-myocardial signalling that is equally required for both processes.

Coronary arteriogenesis

The final step in formation of a functional coronary vasculature is arteriogenesis, the coating of coronary capillaries with mural cells (pericytes and SMCs), which will only be briefly discussed and is reviewed elsewhere (Cai & Schaper 2008). The principal signalling pathways involved in coronary arteriogenesis are those involving PDGF and Notch signalling (Van Den Akker et al. 2008). Others, such as TGF-β and the Ang-Tie signalling pathways have not been implicated specifically in coronary arteriogenesis but their involvement would be expected based on their essential role in arteriogenesis of the systemic circulation and the fact that Ang1 has been used to stimulate arteriogenesis in the adult heart (Siddiqui et al. 2003).

There is much evidence to support a role for PDGFs, specifically PDGF-A and PDGF-B and their respective receptors, PDGFR-α and PDGFR-β, in the maturation and stabilization of the coronary vasculature. Mice deficient in PDGF-B, PDGFR-α or PDGFR-β are all embryonic lethal between E8-E16 with severe cardiac malformations (Battegay et al. 1994; Lindahl et al. 1997; Lindahl & Betsholtz 1998). More specifically, immunohistochemical analysis revealed defective coronary arteriogenesis in these mutants, to complement quail-chick chimeras which provides evidence that PDGF signalling plays an essential role in EPDC maturation (Van Den Akker et al. 2005).

Notch signalling has been implicated in establishing arterio-venous identity in the developing vasculature and may well play the same role during the development of coronaries (Armulik et al. 2005). Activation of EC Notch induces expression of arterial markers (ephrin B2, CD44 and neuropilin 1) and suppresses venous markers (Eph B4), whereas repression of Notch in the endothelium establishes venous endothelial identity. Notch signalling, therefore, at least in the systemic vasculature, is likely involved in regulating interactions between endothelial and mural cells (pericytes and vascular SMCs).

An evolving paradigm in regenerative medicine is that tissue repair in the adult is frequently underpinned by a re-activation of the embryonic programme that created the tissue in the first instance. As such there is much to gain from understanding the embryonic mechanisms of vasculogenesis and angiogenesis. Key to the success of this approach lies in developing appropriate strategies to redeploy these pathways in the adult for therapeutic benefit.

Coronary vasculogenesis and angiogenesis in the adult

Until recently, the prevailing dogma was that vessels in the embryo derive from endothelial progenitors by vasculogenesis, whereas new vessels in the adult are derived only by angiogenesis, through proliferation of differentiated ECs. A wealth of evidence has since emerged to indicate that vasculogenesis also occurs in the adult, in many cases using the same populations of cells that contribute to vessel formation in the developing embryo. Following severe coronary artery stenosis in patients, the gradual development of a collateral coronary circulation has long been observed, but attempts to therapeutically stimulate these processes have only recently been applied.

The mammalian myocardium responds to stresses by activating a multitude of adaptive mechanisms to limit cellular injury and to repair, as much as possible, the damaged tissue. Ischaemia, both acute and chronic, has been shown to stimulate vasculogenesis and, as in embryonic vasculogenesis, hypoxia is recognized to be the major factor driving neovascularization (new vessel growth), a manifestation of the body’s attempt to restore blood and oxygen flow. Mediated by HIF-1α, hypoxia leads to the up-regulation of the same cohort of pro-angiogenic growth factors that are implicated in developmental coronary vasculogenesis, including VEGF and its receptors, FLK-1 and FLT-1 (Li et al. 1996; Levy 1998), FGF-1 and -2 and TGF-β (Ahn et al. 2008). Unfortunately, such adaptive mechanisms are clearly insufficient given the number of patients that present with disabling angina. The insufficiency of the intrinsic vascular response is confounded by both decreased cytokine production and EC viability in aged (Rivard et al. 1999a), diabetic (Rivard et al. 1999b; Yoon et al. 2005; Emanueli et al. 2007) and hypercholesterolaemic (Couffinhal et al. 1999) patients and animals.

Angiogenic therapy – growth factors

The goal of coronary angiogenic therapy is to improve myocardial perfusion in patients with coronary artery disease. Many animal studies have demonstrated efficient induction of collateral vessels in the ischaemic areas of the myocardium by growth factors, most prominently VEGF and FGF-2 (Waltenberger 1997). As a result, therapeutic angiogenesis advanced rapidly into clinical trials for patients with inoperable ischaemic heart disease by administration of growth factor, either in the form of purified protein or gene therapy (Losordo et al. 1998; Morishita et al. 2001; Isner 2002). Both VEGF and FGF are heparin-binding EC mitogens that potently induce angiogenesis. VEGF selectively stimulates ECs while FGF, not only stimulates ECs more potently, but additionally stimulates SMCs and fibroblasts (Senger et al. 1993). FGF-1 and -2 are chemotactic for ECs, thereby inducing their migration into areas of compromised endothelium (Joseph-Silverstein & Rifkin 1987). In animal models, results with FGF-1 are conflicting and, in most cases, there is no beneficial effect (Banai et al. 1991; Unger et al. 1993). Studies with FGF-2 have yielded more consistent results, demonstrating its efficacy as a therapeutic agent in acute ischaemia (Yanagisawa-Miwa et al. 1992; Horrigan et al. 1999), although neovascularization could not account for the full extent of the observed improvement. To date it is not clear whether FGF-2 or VEGF is the more effective angiogenic growth factor, despite the large number of clinical trials. The promising preclinical results obtained in animal studies have yet to be translated into man (Yla-Herttuala et al. 2007). The largest phase II VEGF trial (VIVA), involving 178 patients with coronary artery disease (CAD), recorded no significant effect in the primary end point of treadmill exercise capacity at 4 months; however, patients that received the highest dose of recombinant VEGF-A₁₆₅ displayed a significant improvement in angina class compared with placebo (Henry et al. 2003). In a similar vein, a large double blind phase II trial for FGF (FIRST), which recruited 337 CAD patients, revealed no improvement in exercise tolerance or in myocardial perfusion after 90 days (Simons et al. 2002).

Angiogenic therapy – current limitations

One of the major shortcomings of therapeutic angiogenesis appears to be the lack of concomitant arteriogenesis; capillaries that form by vasculogenesis, without smooth muscle support, are immature and unstable and regress over time. Compared with vasculogenesis, arteriogenesis is more complex, involving smooth muscle as well as ECs, and has proven more recalcitrant to therapy. Arteriogenesis is not stimulated by the same signals, such as hypoxia, but rather by haemodynamic forces, principally shear stress (Yu et al. 1999), therefore, while angiogenesis is stimulated as part of the endogenous response to MI, extrinsic factors are generally required to induce arteriogenesis.

While single genes or proteins have proven successful to therapeutically induce coronary angiogenesis, the simultaneous activation of multiple growth factors is considerably more effective and this is particularly true for stimulating arteriogenesis (van Royen et al. 2001). During embryogenesis, as described above, vascular growth comprises a series of events which depend upon integrating numerous signals by a multitude of separate factors. This has led to the evolution of therapeutic methodologies which aim to concomitantly activate several growth factors, for example by providing a mechanical or metabolic stimulus, such as the thyroxine analogue 3,5-diiodothyropropionic acid (DITPA) (Zheng et al. 2004). DITPA administration to rats post-MI led to enhanced up-regulation of VEGF, FGF-2, Ang-1 and Tie-2 within 3 days and led to increased arteriolar density as well as improved left ventricular function. Given the need to induce multiple growth factors for maximum coronary collateral growth, future therapies may benefit from the identification of miRNAs, such as miR-126 (Wang et al. 2008), that activate multiple pro-angiogenic factors or gene therapy with an upstream activator such as HIF-1α.

Identification of other factors for angiogenic therapy

Thus, despite the progress made in understanding the mechanisms that underlie coronary vasculogenesis and angiogenesis, the ability to therapeutically implement these processes in the injured adult heart requires further identification of key controlling stimuli. In attempting to discover new agents to stimulate vasculogenesis, valuable insight may be derived from understanding the key molecules which perform these functions during embryogenesis. One example of a molecule that has been shown in the adult to recapitulate its embryonic vasculogenic role is Tβ4.

Thymosin β4 is essential for coronary vasculogenesis, angiogenesis and arteriogenesis in the embryo (Smart et al. 2007) and the first intimation that Tβ4 offers hope for cardiac repair came in 2004 after Bock-Marquette et al. published on the potential for Tβ4 to restore function to the ischaemic adult mouse heart (Bock-Marquette et al. 2004). Following coronary artery ligation in mice, Tβ4 treatment led to increased myocardial preservation and improved cardiac function. The benefits of Tβ4 treatment were attributed to improved cell survival via Akt activation; angiogenic processes and improved vascularization of the infarcted myocardium were not determined.

Based on an essential role for Tβ4 in epicardial derived coronary vessel development in the embryonic heart, Smart et al. subsequently investigated whether Tβ4 may be used in the adult to stimulate the epicardium to give rise to vascular precursors, as this may contribute towards its known cardioprotective benefits in the injured adult heart (Bock-Marquette et al. 2004). Translation of a vascular development role for Tβ4 to that of an angiogenic therapy relies on the release of the adult epicardium from a quiescent state and restoration of pluripotency. It was previously perceived that adult epicardium resides in a state of dormancy, having lost all potential for migration, differentiation and signalling by early postnatal stages (Chen et al. 2002), hence no angiogenic therapy has previously attempted to recapitulate the embryonic mechanisms of coronary vasculogenesis at the level of the epicardium. Indeed, untreated adult epicardial explants displayed virtually no detectable outgrowth. In contrast, treatment with Tβ4 stimulated extensive outgrowth of cells which, as they migrated away from the explant, differentiated into a variety of discernable cellular phenotypes. The emerging cells were of epicardial origin (epicardin positive) and differentiated, with migration, into fibroblasts, smooth muscle and a limited number of ECs. These data demonstrate that, under the control of Tβ4, vasculogenic and, moreover, arteriogenic potential remains within the adult epicardium, which may be harnessed for therapeutic use. More recently, prokineticin-2 was identified as another paracrine factor that acts on the adult epicardium to promote neovascularization (Urayama et al. 2008). In support of these observations, epicardial cells from human adult heart were shown to undergo EMT and obtain characteristics of SMCs in vitro (van Tuyn et al. 2006).

Valuable insight into the epicardium as a source of vascular progenitors may be derived from studies in zebrafish. Following ventricular resection of the adult fish heart, the epicardium exhibits a rapid and robust response to injury, which includes proliferation and expression of embryonic epicardial markers, Tbx18 and Raldh2, within 1–2 days of resection (Lepilina et al. 2006). The activated epicardium envelopes the cardiac chambers, including the injured apex and a subpopulation of cells invades the sub-epicardial space and myocardium to contribute endothelial and SMCs to form new coronary vessels; this is an exact recapitulation of the processes involving the embryonic epicardium in coronary vasculogenesis for which Tβ4 was required (Smart et al. 2007). It is highly significant, therefore, that in a related study, Tβ4 was found to be up-regulated in regenerating zebrafish hearts (Lien et al. 2006). Taken together, these data, along with the ability of Tβ4 to mobilize murine adult EPDCs, provide strong support for the potential of Tβ4 to induce neovascularization and possibly other aspects of myocardial regeneration, in the injured adult heart.

A number of gene expression changes have been reported following Tβ4 treatment raising speculation that it may, in some way, modify transcription, consistent with its translocation to the nucleus (Huff et al. 2004). The most notable of genes from an angiogenic perspective is VEGF which was up-regulated following over-expression of Tβ4 in B16-F10 lung tumour cells (Cha et al. 2003); conversely, a down-regulation of VEGF in situ was observed in Tβ4 knockdown hearts suggesting that appropriate VEGF expression may be regulated by Tβ4 (Smart et al. 2007). Furthermore, in a study investigating cardioprotective effects of Akt over-expressing bone marrow mesenchymal cells (BMSCs), both Tβ4 and VEGF were significantly up-regulated in the MSCs during hypoxia, as potential mediators of myocardial protection (Gnecchi et al. 2006).

Angiogenic therapy: cell-based strategies

Aside from growth factor administration and gene therapy, the principal alternative approach currently under investigation is cell therapy. BMSCs and Endothelial progenitor cells (EPCs) are the cell types most widely-researched for their ability to induce cardiac neovascularization.

Endothelial progenitor cells

Endothelial progenitor cells are a specialized subset of haematopoietic cells found in the bone marrow and peripheral circulation (Asahara et al. 1997) which constitute up to 20% of the CD34+ population of peripheral blood in the adult. They are phenotypically characterized by expression of antigens usually associated with haematopoietic stem cells (HSCs) including CD133, CD34, c-kit, VEGFR2, CD144 (vascular endothelial (VE)-cadherin) and Sca-1 (Asahara et al. 1997). The discovery of circulating EPCs permanently changed the view that vasculogenesis occurs exclusively in the developing embryo. Vasculogenic mechanisms are now known to be retained in the adult and redeployed, when required, for neovascular repair. EPCs are mobilized from bone marrow and recruited to foci of neovascularization where they form new blood vessels in situ. Striking parallels in the regulatory steps of embryonic and postnatal vasculogenesis suggest that the underlying initiating stimuli and regulatory pathways may be conserved. EPCs are incorporated into injured vessels and develop into mature ECs during the processes of re-endothelialization and neovascularisation (Miller-Kasprzak & Jagodzinski 2007). As well as directly contributing to new vessel formation, EPCs secrete paracrine factors, including VEGF-A, VEGF-B, SDF-1 and insulin-like growth factor-1 (Narmoneva et al. 2004) which may play an equal, if not greater, role in promoting angiogenesis (Rubart & Field 2006). The use of EPC populations for therapeutic purposes has rapidly progressed into clinical trials with promising preliminary results (Strauer et al. 2002; Schachinger et al. 2006a).

Research into stem cell therapy, particularly in the heart, has been fraught with ambiguities and controversy. However, higher EPC number in the peripheral circulation generally correlates with reduced risk of a cardiovascular event or death from cardiovascular causes (Werner et al. 2005). Various populations of bone marrow-derived cells, including EPCs, have been tested, following ex vivo expansion and varying degrees of improvement in myocardial function have been reported, most of which are at the limit of clinical detection In animal studies, an EPC-enriched fraction of CD34+ human cells were introduced, either by intravenous infusion (Kocher et al. 2001) or by intramyocardial transplantation (Iwasaki et al. 2006) into rats following MI. In both studies, improved left ventricular function, and a degree of neovascularization were reported. Furthermore, in the study of intramyocardial transplantation, a subpopulation of cardiomyocytes and SMCs were shown to be of human origin (Iwasaki et al. 2006). Nevertheless, the issue of transdifferentiation remains highly controversial and the overall conclusion from a large number of such studies is that any beneficial effect likely results from the secretion of cardioprotective or pro-angiogenic paracrine factors from these cells.

As with other populations of bone marrow-derived progenitor cells, the move into clinical trials has been extremely rapid. Multicentre placebo-controlled clinical trials have been performed using both cultured and freshly isolated EPCs. The Transplantation of Progenitor Cells and Regeneration Enhancement in Acute Myocardial Infarction (TOPCARE-AMI) trial, in which patients were given an intracoronary infusion of cultured human EPCs shortly after MI, reported improvements in both microvascular and global left ventricular function (Assmus et al. 2002). A further randomized trial (Injection of Autologous CD34-Positive Cells for Neovascularization and Symptom Relief in Patients with Myocardial Ischemia) demonstrated the safety, feasibility and bioactivity of freshly isolated, autologous CD34+ cells administered intramyocardially (Losordo et al. 2007) with enrolment for larger, phase II trials underway. More consistent, positive outcomes will need to be demonstrated at this level before any definitive (phase III) trials are established.

An inherent limitation in the use of EPCs for therapeutic angiogenesis lies in their scarcity in peripheral blood. Extrapolation from animal studies, in which 1.0 × 10⁵ human EPCs per kilogram of body weight were required for satisfactory improvement post-MI, suggests that 6–12 l of blood would be required to treat each patient via systemic infusion. This problem is confounded in cardiac patients since both the number and function of circulating EPCs is impaired as a result of diabetes (Vasa et al. 2001; Ii et al. 2006), ageing (Heiss et al. 2005), hypertension (Vasa et al. 2001; Imanishi et al. 2005), hypercholesterolaemia (Vasa et al. 2001) and smoking (Kondo et al. 2004; Michaud et al. 2006). Proposed approaches to overcome these limitations include: (i) transplantation directly into ischaemic tissue; (ii) cytokine administration to enhance mobilization of EPCs from the bone marrow prior to collection; (iii) concomitant gene therapy to enhance EPC efficacy; or (iv) ex vivo expansion of renewable primitive progenitor cells from bone marrow or other sources, such as embryonic stem cells.

Kupatt et al. investigated the ability of transplanted embryonic EPCs (eEPCs) to induce neovascularization and tissue rescue after MI in mice (Kupatt et al. 2005). Systemic injection of eEPCs led to a measurable increase in neovascularization and improved tissue recovery. To investigate the underlying mechanism, genome-wide expression profiling of eEPCs were performed. Alongside recognized angiogenic factors such as VEGF-A and VEGF-B, Tβ4, pro-thymosin α and thymosin β10 were found to be among the most abundant of secreted factors. The same group later demonstrated that the paracrine secretion of Tβ4 from eEPCs was essential for cardioprotection in a porcine MI model (Hinkel et al. 2008).

Smooth muscle progenitor cells

With regard to arteriogenic therapy, EPCs would appear to be inferior to those progenitor cells that possess a broader differentiation potential that also includes pericytes or SMCs, given the need for mural cell stabilization of vessels. Smooth muscle progenitor cells (SMPCs) were more recently identified in bone marrow, circulating blood and vascular adventitia (Shimizu et al. 2001; Sata et al. 2002). Like EPCs, SMPCs can be mobilized by SDF-1, a chemokine induced by tissue hypoxia. In a model of critical hindlimb ischaemia, co-administration of human EPCs and umbilical cord SMPCs afforded a marked increase in capillary and arteriolar densities, despite the fact that only EPCs, not SMPCs, were incorporated into newly formed capillaries (Foubert et al. 2008). In vitro assays revealed the potential mechanism of this non-cell autonomous effect; production of Ang-1 by SMPCs activates Tie-expressing EPCs, resulting in increased EPC survival and formation of a stable network. SMPC-based therapy may similarly prove successful for coronary neovascularization. The ischaemic hindlimb model has proven invaluable in enhancing our understanding of therapeutic angiogenesis and interventions shown to be successful in treating the peripheral vasculature have seamlessly translated into the myocardium to improve coronary angiogenesis.

Bone marrow mesenchymal stem cells

Bone marrow mesenchymal cells have been extensively studied with respect to cardiac repair and neovascularization, although their ability to transdifferentiate is equivocal, controversial and largely disputed (Orlic et al. 2001a,b; Askari et al. 2003; Murry et al. 2004). However, BMSCs may offer potential benefit through their secretion of paracrine factors that are cardioprotective or angiogenic. Significantly, Tβ4 levels were elevated in Akt over-expressing BMSCs (Gnecchi et al. 2006), particularly under hypoxic conditions; injection of Akt-MSCs or even their conditioned medium considerably reduced infarct size and improved cardiac function. That conditioned medium conferred a comparable degree of cardiac repair suggests that, as with EPCs and SMPCs, secretion of angiogenic factors contributes significantly towards neovascularization induced by BMSCs.

Despite the controversy concerning the efficacy of BMSCs, a number of large scale multicentre clinical trials have been initiated in which autologous progenitor cells derived from bone marrow were reintroduced, following in vitro expansion, via intracoronary infusion into patients with acute myocardial infarction (AMI) (Schachinger et al. 2006b). To date, the majority of clinical studies have used bone marrow mononuclear cells and have shown, at best, modest improvements in cardiac function (Abdel-Latif et al. 2007). The majority of trials conclude that any benefits conferred likely result from the secreted paracrine factors and their induction of angiogenesis and improved microvascular function, as demonstrated in the REPAIR-AMI study (Abdel-Latif et al. 2007).

BMSCs and EPCs: failure in homing and recruitment?

In animal studies, and reflected in clinical trials, insufficient numbers of progenitor cells are found to home to the heart. The use of growth factors to enhance the efficacy of cell-based therapies has been employed to positive effect. Mobilization of stem cells from bone marrow is achieved by the action of cytokines, such as granulocyte-colony stimulating factor (G-CSF) (Winter et al. 1996), the closely-related granulocyte-macrophage-colony stimulating factor (GM-CSF) (Gianni et al. 1989; Takahashi et al. 1999), Flt-3 ligand (Fukuda et al. 2005), erythropoietin (Heeschen et al. 2003), stem cell factor (SCF, the ligand for c-kit) (Bodine et al. 1993); growth factors such as VEGF (Asahara et al. 1999; Gill et al. 2001), angiopoietin-1 (Hattori et al. 2001a) and placental growth factor (Hattori et al. 2002) as well as several chemokines such as SDF-1 (Hattori et al. 2001b), Interleukin (IL)-8 (Laterveer et al. 1995), growth regulated oncogene-β (GRO-β) (Pelus et al. 2004) and macrophage inflammatory protein-2 (MIP-2) (Wang et al. 1997). Of these, G-CSF, GM-CSF, VEGF and SCF were shown to improve cardiac regeneration in mice (Orlic et al. 2001b). This rapidly led to the initiation of clinical trials for the use of G-CSF in patients with CAD (Powell et al. 2005) and AMI (Valgimigli et al. 2005). G(M)-CSF-mobilized blood from patients contained 5–100-fold higher levels of HSCs, MSCs and EPCs, compared with non-mobilized blood (Takahashi et al. 1999; Powell et al. 2005; Valgimigli et al. 2005), however, the ability of these cells to improve cardiac remodelling and function after AMI has been disappointing (Kurdi & Booz 2007). While recruited BMSCs were unable to differentiate into cardiomyocytes to any significant degree, G-CSF treatment was shown to induce angiogenesis in the infarcted heart and to confer a paracrine protective effect on cardiomyocytes (Kurdi & Booz 2007).

In the Front-Integrated Revascularization and Stem Cell Liberation in Evolving Acute Myocardial Infarction (FIRSTLINE-AMI trial) trial, EPC mobilization, induced by G-CSF administration, led to improved cardiac function and reduced ventricular remodelling over the 1 year follow-up period, with no reported acceleration of restenosis or adverse events (Ince et al. 2005). In contrast, enrolment in the MAGIC trial was prematurely terminated because of an unexpected rise in the rate of in-stent restenosis in patients treated with G-CSF for progenitor cell mobilization from bone marrow (Kang et al. 2004). The frequency and severity of in-stent restenosis as a side-effect of cell therapy in MI patients, resulted in the use of drug-eluting stents in subsequent trials, to limit restenosis (Kang et al. 2006) but this remains a critical concern which will require close monitoring in future trials of angiogenic therapies.

Once mobilized, various factors are required for recruitment of BMSCs and EPCs to the heart, including endogenous stimuli which are released following myocardial infarction and act to enhance local homing signals. Factors shown to enhance progenitor cell recruitment to the heart include SDF-1α (Askari et al. 2003), integrin β1 (Ip et al. 2007), inflammatory cytokines such as IL-6 (Malek et al. 2006) and, possibly adenosine (Ryzhov et al. 2008). Adenosine, which is cardioprotective and promotes capillary proliferation in the heart, was recently shown in vitro to markedly increase eEPC adhesion to cardiac microvascular ECs, indicating a potential role for adenosine in the homing of EPCs to coronary endothelium (Ryzhov et al. 2008).

Resident cardiovascular stem/progenitor cells

Failure of BMSCs and EPCs thus far to deliver in terms of neovascular and myocardial regeneration has led researchers to focus on the identification of resident cardiovascular stem or progenitor cells and the factors that stimulate their proliferation or migration. The first indication that the adult heart harboured such a population of stem cells possessing regenerative properties came from a study of AMI patients showing an elevated number of proliferative immature cardiomyocytes in the infarct border zone (Beltrami et al. 2001). With the subsequent identification of small populations of resident cardiac progenitor cells (CPCs; cardiosphere-forming with vasculogenic potential) (Smith et al. 2007) and multipotent Isl1+ cardiovascular progenitors (MICPs) (Laugwitz et al. 2005; Moretti et al. 2006), along with EPDCs (Smart et al. 2007), an increasing number of suitable resident populations are now known to exist in the postnatal heart and give rise to vascular precursors that may be exploited to effect neovascular repair in the ischaemic heart. These discoveries exposed new opportunities for cardiac repair (Beltrami et al. 2003), of which, two approaches are currently under investigation: the first is to explant cardiac stem cells from the heart, induce their proliferation and differentiation ex vivo and to engraft them back into the injured heart (Dawn et al. 2005) and the second, which to date has received less attention, is to stimulate CPCs in situ to proliferate, migrate and differentiate following infarction without ex vivo manipulation. To this end, a major therapeutic goal is the identification of factors that stimulate cardiac stem cells to form replacement vascular progenitors, as well as cardiomyocytes, for regeneration of the injured heart. MICPs are exceedingly rare in the neonatal heart (500–600 per rat heart) and may not even exist in the adult heart, thus the therapeutic potential of these cells will depend on the ability to isolate them from the embryonic or postnatal heart, and expand them ex vivo prior to their re-introduction into the adult (Laugwitz et al. 2008). Assuming sufficient numbers of cells can be isolated as the starting population, ex vivo expansion runs the risk of phenotypic drift. In contrast, resident EPDCs, when activated, migrate in sufficient number within the adult heart (N. Smart and P. Riley, unpublished observations) to make them a realistic therapeutic target for neovascularization without the need ex vivo for expansion.

Vascular tissue engineering

A natural progression from cell-based therapy has been the development of tissue engineering techniques. New vessel formation and improvements in neovascularization have been reported in hindlimb ischaemia using human SMC sheets (Hobo et al. 2008) or matrigel plugs seeded with a combination of human endothelial and cord blood-derived mesenchymal progenitor cells (Melero-Martin et al. 2008). For coronary angiogenesis, the development of cardiac patches is being actively pursued, with some success in preliminary studies. Purified and expanded EPCs and MSCs, engrafted in a collagen matrix, were deposited on to the epicardium of mice 30 days after cryogenic MI (Derval et al. 2008), resulting in improved angiogenesis and cell survival. Only the MSC population, not EPCs, were found to invade the scar but were not incorporated into new vessels or regenerated tissue, thus any beneficial effects were attributed to their secretion of paracrine factors. In vitro engineered 3-dimensional neonatal cardiomyocyte sheets containing preformed EC networks were transplanted onto infarcted rat hearts, which resulted in a significantly elevated capillary density with blood vessels originating from the cardiac patch bridging across and connecting with host capillaries in the infarcted myocardium (Sekine et al. 2008). In a similar manner, porous collagen scaffolds were shown to elicit a powerful angio- and arteriogenic response in both intact and cryo-injured rat hearts (Callegari et al. 2007).

Conclusions

To date, attempts to develop therapies for neovascularization of the adult heart have met with limited success. On the whole these approaches have involved either (i) the application of growth factors whose embryonic role is predominantly in angiogenesis rather than vasculogenesis or (ii) the use of progenitor cell populations that play no known role in coronary vasculogenesis. Perhaps a more fruitful approach depends upon more faithfully recapitulating the embryonic process in the adult heart. Applying this rationale to neovascularization of the ischaemic heart would require the derivation of vascular progenitors by reactivation of the adult epicardium and the identification of other small molecules, such as Tβ4, will be instrumental in stimulating this apparently tractable lineage.

Several distinct populations of progenitor cells have emerged with potential for use in vascular repair, as well as myocardial regeneration. Based on lessons learned from animal studies and clinical trials to date, it can be concluded that any practicable progenitor population for successful vascular or myocardial repair should ideally (i) have a defined embryonic counterpart; (ii) exist in sufficient number within the adult and (iii) be resident within the heart. Expanding our knowledge of coronary vessel development and assessing whether embryonic cell-based mechanisms may be recapitulated in the adult should pave the way towards neovascular repair of the ischaemic heart.