Retinoic Acid and VEGF Delay Smooth Muscle Relative to Endothelial Differentiation to Coordinate Inner and Outer Coronary Vessel Wall Morphogenesis
Rationale: Major coronary vessels derive from the proepicardium, the cellular progenitor of the epicardium, coronary endothelium, and coronary smooth muscle cells (CoSMCs). CoSMCs are delayed in their differentiation relative to coronary endothelial cells (CoEs), such that CoSMCs mature only after CoEs have assembled into tubes. The mechanisms underlying this sequential CoE/CoSMC differentiation are unknown. Retinoic acid (RA) is crucial for vascular development and the main RA-synthesizing enzyme is progressively lost from epicardially derived cells as they differentiate into blood vessel types. In parallel, myocardial vascular endothelial growth factor (VEGF) expression also decreases along coronary vessel muscularization.
Objective: We hypothesized that RA and VEGF act coordinately as physiological brakes to CoSMC differentiation.
Methods and Results: In vitro assays (proepicardial cultures, cocultures, and RALDH2 [retinaldehyde dehydrogenase-2]/VEGF adenoviral overexpression) and in vivo inhibition of RA synthesis show that RA and VEGF act as repressors of CoSMC differentiation, whereas VEGF biases epicardially derived cell differentiation toward the endothelial phenotype.
Conclusion: Experiments support a model in which early high levels of RA and VEGF prevent CoSMC differentiation from epicardially derived cells before RA and VEGF levels decline as an extensive endothelial network is established. We suggest this physiological delay guarantees the formation of a complex, hierarchical, tree of coronary vessels.
Coronary morphogenesis has been receiving intense scrutiny because of its usefulness as a model to understand vascular development and as a source of cues for regenerative therapies. Coronary vessel morphogenesis is linked to the development of epicardium and proepicardium, a transient embryonic structure located between the sinus venosus and the liver primordium.1,2 Avian embryos have been instrumental to our knowledge on coronary development. Lineage-tracing studies indicate that the proepicardium contains precursors of the epicardium, coronary endothelial cells (CoEs) and coronary smooth muscle cells (CoSMCs).2–5
Proepicardial and primitive epicardial cells undergo epithelial-to-mesenchymal transformation (EMT),2,6 supplying mesenchymal cells to the outer surface of the heart. These epicardially derived cells occupy positions along the subepicardial space, where they will differentiate into coronary vessel cell types.7 Around Hamburger–Hamilton stage (HH)26, highly migratory, epicardially derived cells leave the subepicardium to invade the myocardium, thus disseminating part of CoE progenitors throughout the cardiac muscle.8,9
Coronary vessels are formed from primary subepicardial and intramyocardial endothelial plexuses. Subepicardial/intramyocardial coronary vessel morphogenesis display one of the most striking and least understood features of coronary development, ie, the pronounced delay between CoE/CoSMC differentiation. Already at stage HH24, precursors in subepicardium differentiate into CoEs and form endothelial tubes under the influence of myocardially secreted cytokines, such as vascular endothelial growth factor (VEGF) and basic fibroblast growth factor.10,11 In contrast, full CoSMC differentiation is only apparent after HH31 to HH32, concomitant with connection of CoE network to the aorta, stabilization of directional blood flow, and the onset of shear stress on coronary endothelium.11 Curiously, CoSMC differentiation is remarkably efficient when proepicardial-cells are removed from embryos and cultured,12 suggesting that there are active embryonic/heart mechanisms preventing premature CoSMC differentiation before an endothelial vascular outline is formed.
The mechanisms underlying this characteristic physiological delay have not been addressed. Potential candidate inhibitors of precocious CoSMC differentiation are the signaling pathways operated by the Vitamin A/retinol-derived morphogen retinoic acid (RA) and by VEGF. In avian embryos proepicardium, epicardium and epicardially derived cells express RALDH2 (retinaldehyde dehydrogenase-2), the main RA aldehyde dehydrogenase catalyzing retinaldehyde conversion into RA during amniote cardiac development.8,9,13,14 Furthermore, epicardial derivatives activate RA reporters and downstream targets of RA signaling, indicating that early vasculogenic steps take place in an RA-rich environment.13 Along the migration into the heart, epicardially derived cells undergo progressive coronary differentiation and gradually lose RALDH2 expression, as documented by waning of RALDH2 immunoreactivity as epicardially derived cells migrate into myocardium,8 suggesting that these phenomena might be causally related. Moreover, it is well known that synthesis of VEGF by the developing myocardium peaks after epicardial formation and decreases after the onset of coronary circulation.15
Therefore, we hypothesize that RA and VEGF signaling coordinate differentiation of CoE and CoSMC precursors, acting as brakes to slow smooth muscle differentiation until a network of endothelial tubes is established. To test this hypothesis, we developed in vitro and in ovo approaches to study the effects of RA and VEGF signaling pathways at the level of whole proepicardial populations and of single proepicardial cells. Here, we show that RA plays an important role in the physiological delay of CoSMC differentiation and, additionally, that VEGF inhibits CoSMC differentiation.
An expanded Methods section is available in Online Data Supplement at http://circres.ahajournals.org.
Proepicardia were dissected from quail embryos. To promote cell attachment, proepicardia were grown on DMEM+15% FBS during 30 hours. After, the medium was replaced by DMEM+0.5% FBS for 48 hours. Then, explants were supplemented with RA (1 μmol/L) or VEGF (75 ng/mL), in the presence (5% FBS) or absence of FBS, during 4 days. Control proepicardial explants were treated with DMSO.
Transductions of proepicardial explants were performed by replacing the medium, after cell adhesion, by DMEM+0.5% FBS containing 7×1011 viral particles per well. Treatments started after 24 hours. Treatments included retinol (5 μmol/L), retinaldehyde (5 μmol/L), DEAB (4-diethylaminobenzaldehyde, 5 μmol/L), or retinol/DEAB (5 μmol/L each) in DMEM or DMEM+5% FBS.
Proepicardial cultures and quail-to-chick chimeras (constructed after modification from2) were submitted to immunohistochemistry.
Cultured Proepicardial Cells Express Markers of Coronary Cell Lineages
To investigate CoSMC differentiation in vitro, proepicardial explants were cultured in the presence of FBS. Cultured proepicardia spread, forming an epithelial monolayer. Around the third day of culture, free migratory cells appeared at the periphery of the culture (Online Figure I).7,16
Figure 1A and 1B shows that explants express the proepicardial markers cytokeratin and WT1. WT1 tends to disappear from proepicardium cells expressing significant levels of CoE (QH1) and CoSMC (smooth muscle α actin [SMαA]) differentiation markers, although small traces of WT1 expression can still be detected in some differentiating cells (Figure 1C and 1D). QH1 labeling shows that CoE progenitors concentrate in the culture center (Figure 1E), whereas SMαA labeling indicates that CoSMCs are abundant at the culture border (Figure 1F). The SMαA-positive ring of proepicardial cultures includes cells displaying typical EMT/migratory phenotypes (Online Figure I), whereas the monolayer core is poor in those cells. Two cellular patterns of SMαA expression are evident. The first is characterized by dense SMαA expression, prominent in actin stress-fibers, which is typical of differentiated CoSMCs and of cells that have undergone EMT.17 The “dense” pattern is most often observed at the margin of the culture. The second pattern is characterized by SMαA expression in subcortical actin bundles at the periphery of the cell (Figure 1G).18 The “peripheral” pattern is most frequently observed close to culture center, where CoSMC differentiation is weaker. These 2 patterns will be henceforth referred to as “dense” and “peripheral.” Because CoSMC differentiation is associated with EMT in cultured proepicardium,12 we looked for focal adhesion kinase (FAK) expression as a further marker for dense and peripheral types. We observed that peripheral SMαA expression associates with a diffuse FAK expression, whereas dense SMαA correlates to FAK expression at focal adhesion sites (Figure 1H through 1M). The FAK expression patterns in cultured proepicardial cell suggest that the peripheral phenotype represents an intermediate state between epithelial and mesenchymal (dense) phenotypes.
Proliferation, CoSMC Differentiation, and Apoptosis in Cultured Proepicardial Cells
To establish the topology of differentiation and proliferation in cultured proepicardia we determined 5-bromodeoxyuridine (BrdUrd) immunoreactivity after a 12-hour exposure to BrdUrd. Figure 1F shows that the outer ring of CoSMC differentiation is matched by an opposite pattern of proliferation, which is higher at the center of the cell. This suggests that proepicardial cultures grow from inside-out, with cells dividing at the core and moving toward the periphery. To test this hypothesis, we labeled proepicardial explants with CMPTX, a vital die diluted with cell division, but not shared among neighboring cells. At 30 hours, all cells are stained. However, die distribution gradually shifts to periphery, consistent with an inside-out growth (Figure 1N). To evaluate whether cultures are significantly affected by apoptosis we submitted proepicardium cultures to the annexin assay. Annexin-positive cells are not found in the presence of FBS (not shown) and are rare and limited to the culture core in its absence (see below), suggesting that apoptosis does not play significant roles in our cultures.
RA Inhibits Expression of Smooth Muscle Markers in Cultured Proepicardium
To test whether RA signaling is involved in the physiological delay of CoSMC differentiation, we treated proepicardial cultures with RA. Exogenous RA abrogated SMαA expres-sion, except in rare cells at the periphery of the culture (Figure 2A and 2B). To establish whether RA effects on the density and distribution of SMαA-positive cells are contingent on the absence of FBS, a strong smooth muscle inducer (compare Figure 2A through 2E), we treated proepicardial cultures with RA in the presence of 5% FBS. As shown in Figure 1F, RA treatment produced a marked decrease in the number of SMαA-positive cells. RA effects were intense at the center of the cell, but comparatively weaker at the margins. Such marginal cells could represent a subpopulation of older progenitors already committed to the CoSMC fate. In sum, FBS did not prevent RA from reducing the relative number of SMαA-positive cells, indicating that the RA effects on SMαA expression are also observed in conditions that favor CoSMC differentiation. Similar effects were found in mouse proepicardial cultures (Online Figure II).
To check whether the effects of RA affect smooth muscle markers other than SMαA, we performed immunohistochemistry to smooth muscle γ actin (SMγA) and used quantitative RT-PCR to examine expression of markers CRP2, GATA-6, SRF, and SM22A, which are sequentially activated during early smooth muscle differentiation.12 In Figure 2I through 2K, we show that SMγA immunoreactivity is reduced by RA and that the retinoid also inhibits expression of the other four smooth muscle markers. These results indicate that RA inhibits the expression of multiple markers of smooth muscle differentiation.
RA Inhibits Smooth Muscle Differentiation
To establish whether the decrease in smooth muscle marker expression after RA treatment is attributable to the effect of RA on CoSMC differentiation, proliferation, apoptosis, or EMT, we performed proliferation, apoptosis, and collagen gel transformation assays. RA treatment in the presence of 5% FBS changed neither the global rate of proliferation nor the specific rate of CoSMC proliferation (Figure 3C and 3D). Although RA treatment did not change global proliferation, analysis of proepicardial cultures indicate that it changed the topological distribution of BrdUrd immunoreactivity, converting proliferation to a more homogeneous pattern (Figure 3A and 3B). RA did not alter the distribution of apoptotic cells (Figure 3E through 3G) nor changed EMT rates in proepicardial cultures (Online Figure III). Thus, the reduction in SMαA-positive cells by RA is not attributable to a decreased proliferation, increased apoptosis, or EMT inhibition, but to a direct effect on CoSMC differentiation.
A Nonreplicating Adenoviral Vector Provides RALDH2 Overexpression in Proepicardial Cells
Prompted by our results on RA inhibition of CoSMC differentiation, we investigated whether autocrine RA signaling would, likewise, inhibit CoSMC differentiation. Accordingly, we developed a nonreplicative adenoviral mouse (m)RALDH2 overexpression system in which high levels of mRALDH2 expression are driven by a CMV promoter in a mRALDH2-IRES-eGFP vector (Online Figure IV). Overall, transduced cells accounted for 10,5% of the culture. Using mouse-specific RALDH2 RT-PCR we provide evidence that transduced cells produce mRALDH2 at a level comparable to that attained by endogenous quail (q)RALDH2 expression in the whole proepicardium, suggesting that the average level of exogenous mRALDH2 expression by each transduced cell exceeds by 1 order of magnitude the endogenous level of qRALDH2 expression. In sum, the AdRaldh2eGFP vector is an effective tool of mRALDH2 overexpression.
Activation of Autocrine RA Signaling by AdRALDH2eGFP Inhibits CoSMC Differentiation
Overexpression of mRALDH2 inhibited CoSMC differentiation in the presence of the RA precursor retinol (Figure 4A and 4B). mRALDH2 overexpression in retinol-supplemented proepicardium cultures reduced the relative number of enhanced green fluorescent protein (eGFP)/SMαA-positive cells approximately by half (Figure 4D). A slightly less intense effect of mRALDH2 overexpression plus retinol supplementation was observed with 5% FBS (Figure 4H through 4I and 4K). Similar results were obtained with retinaldehyde (Figure 4F and 4G).
Next, we tested whether the effects of mRALDH2 overexpression combined with retinol supplementation reflect a specific activation of autocrine RA signaling or a nonspecific consequence of adenoviral infection and/or retinol treatment. As indicated in Figure 4N through 4P, transduction with a control AdLacZ vector did not alter the relative number of LacZ/SMαA-positive cells with or without retinol. This indicates that mRALDH2 overexpression effects are not attributable to unspecific effects of adenoviral infection/retinol supplementation. These results also suggest that endogenous qRALDH2 expression in cultured proepicardia is not enough to drive significant levels of RA signaling.
To establish whether the retinol-induced CoSMC inhibition displayed by cells overexpressing mRALDH2 is dependent on increased aldehyde dehydrogenase activity, we performed experiments in the presence of DEAB, an aldehyde dehydrogenase inhibitor. As indicated in Figure 4D and 4K, DEAB abrogated the retinol-induced decrease in CoSMC differentiation with or without FBS. In contrast, treatment with DEAB in the absence of retinol did not change the relative number of eGFP/SMαA-positive cells, in line with the view that endogenous qRALDH2 expression is not sufficient to activate RA signaling in cultured proepicardia. This indicates that the effect of mRALDH2 overexpression combined with RA precursor supplementation on smooth muscle differentiation is mediated by an aldehyde dehydrogenase activity, consistent with an increased autocrine production of RA via RALDH2.
Activation of Autocrine RA Signaling Reduces the Number of Dense CoSMCs
To establish whether autocrine RA signaling reduces CoSMC differentiation acting on peripheral or dense CoSMC phenotypes, we assessed the effect of mRALDH2 overexpression/retinol supplementation on these cells. In FBS-free cultures (Figure 4E), the relative number of dense cells was significantly reduced by retinol treatment, whereas the relative number of peripheral CoSMCs was not significantly changed. Similar results were found with FBS (Figure 4L). Thus, activation of autocrine RA signaling in proepicardial cultures is primarily associated with a decrease in the number of dense CoSMCs.
Inhibition of RA Signaling Induces Premature SMαA Expression In Vivo
To establish whether RA controls CoSMC development in vivo, we tested whether systemic inhibition of RA signaling would induce premature SMαA activation. DEAB-treated quail embryos display multiple SMαA-positive cells in the atrioventricular junction (Figure 5C) that are not present in controls. This premature SMαA activation associates with reduction of immunoreactivity to cytokeratin, a marker of noncommitted proepicardial/epicardial and epicardially derived cells.
To test whether localized changes in RA signaling would also be consistent with an in vivo role for RA in CoSMC differentiation we combined DEAB-soaked bead grafting with quail-to-chick proepicardial chimeras, so that both the fate of bona fide proepicardial cells and of resident cardiac cells could be simultaneously studied. Figure 5H through 5J shows that quail (donor) proepicardial cells, as well as host subepicardial cells, express SMαA and SMγA under DEAB released from the carrier, whereas control beads did not (Figure 5E through 5G). Moreover, host chicken hearts display increases in SMαA immunoreactivity at distance from the DEAB bead, suggesting inhibition of endogenous RA signaling. These results indicate that in vivo inhibition of RA synthesis affects transplanted proepicardial cells and resident host subepicardial CoSMC precursors.
Coculture of Proepicardia With Cardiomyocytes Inhibits CoSMC Differentiation
We tested whether additional signaling mechanisms are also implicated in physiologically delayed CoSMC differentiation. Because the myocardium has been suggested to control coronary progenitor fate, we evaluated whether coculture of proepicardial explants onto myocardial feeders would reduce CoSMC differentiation. In Figure 6A, double cytokeratin (red) and SMαA (green) staining indicate that proepicardia cultured onto myocardial feeders (SMαA-positive) hardly ever coexpress both markers (yellow). This contrasts with results in fibroblast feeders, in which a distinct yellow external ring marks proepicardial cells SMαA/cytokeratin-positive, reminiscent of the marginal ring of SMαA expression displayed by proepicardial cultures under FBS supplementation (Figure 6D). Figure 6C and 6F shows that proepicardium cultured on myocardial feeders display more QH1-positive cells than proepicardium cultured on fibroblast feeders and that endothelial cells are more homogeneously distributed in the former than in the latter. This strongly suggests that cardiomyocyte feeders possess specific mechanisms that inhibit the CoSMC differentiation and support the endothelial one.
Because the myocardium is a source of VEGF,19 VEGF signaling could account for the decreased immunoreactivity to SMαA in our cocultures by shifting the differentiation of proepicardial cells toward the endothelial phenotype. To check whether VEGF was implicated in the differential effect of cardiomyocytes and fibroblasts on CoSMC differentiation, we measured VEGF expression in these monolayers. Using quantitative RT-PCR, we show that VEGF expression by cardiomyocytes is strong, although it is negligible in fibroblasts (Figure 6M and 6N). Moreover, by complementing fibroblast layers with strong human VEGF production induced by a nonreplicative adenoviral VEGF overexpression system in which a high level of human VEGF expression is driven by a hVEGF-IRES-eGFP vector, we show decreased SMαA expression from cocultured proepicardium (Figure 6G). This indicates that VEGF is one of the inhibitors of CoSMC differentiation produced and secreted by myocardial and VEGF-complemented monolayers, which was confirmed by detection of VEGF immunoreactivity (ELISA) in conditioned medium from these cultures (Figure 6O).
VEGF Inhibits Smooth Muscle Differentiation by Paracrine and Autocrine Modes
To test the role of VEGF in proepicardial differentiation, we treated our cultures with exogenous VEGF. VEGF decreases the number of cells expressing SMαA, while increasing the number of QH1-expressing cells displaying the typical cobblestone endothelial phenotype (Figure 7A through 7D). To establish whether autocrine VEGF signaling reduces CoSMC differentiation, we transduced proepicardial cells with AdVegfeGFP (Figure 7E). We determined that autocrine VEGF significantly reduced CoSMC differentiation with (Figure 7F and 7G) or without FBS (not shown), whereas AdLacZ vector did not. Although autocrine activation of VEGF signaling reduced CoSMC differentiation, only 25% of transduced cells expressed QH1 (Figure 7H and 7I), suggesting that autocrine VEGF does not trigger efficient endothelial differentiation.
To evaluate the effects of VEGF signaling in a more physiological context, in which local extracellular concentration gradients are formed by cytokine secretion, diffusion, and degradation, we scored the number of SMαA-positive or QH1-positive cells in the immediate vicinity of cells transduced with the AdVegfeGFP (Figure 7J). We determined that fewer SMαA-positive cells and more QH1-positive cells are found in a radius of 130 μm from cells transduced with AdVegfeGFP than in cells transduced with control AdLacZ/AdeGFP, consonant with a paracrine effect of VEGF (Figure 7K and 7L).
The differentiation and organization of blood vessel layers (endothelium/smooth muscle) is a sequential, pivotal event in early vascular development, especially during coronary vasculogenesis. CoSMCs highly expressing SMαA are only detected among the myocardial layers at HH32, whereas CoE tubes form from HH24 onwards.2,6,7 How multipotent coronary progenitor cells are able to differentiate following this strict chronological sequence remains unclear. We herein show that RA and VEGF signaling restrain CoSMC differentiation before a primary CoE plexus is formed by vasculogenesis. Therefore, delayed CoSMC differentiation allows the development of a well-defined hierarchy of coronary vessels in an optimal, flexible, provasculogenic environment before vessel structure is stabilized with recruitment and addition of extracellular matrix-producing CoSMCs.
RA Signaling Is Well Poised to Delay CoSMC Differentiation Until a CoE Network is Formed
The delay in CoSMC differentiation in respect to CoEs is especially puzzling, considering that coronary progenitors easily initiate their differentiation into CoSMCs in vitro12 and that some cells expressing markers of smooth muscle differentiation are already present in the late proepicardium.20 Moreover, cells in proepicardium, epicardium, subepicardium, and myocardium express growth factors and receptors such as SRF, TGFβR, and PDGFR that could induce smooth muscle differentiation well before physiological CoSMC differentiation actually starts.21,22 Thus, it is likely that premature CoSMC differentiation is continuously repressed by signaling mechanisms that must operate from very early on, during proepicardium formation, up until an endothelial outline of coronary circulation is organized. Few molecules with strong signaling properties fit this role better than RA. Retinoids are synthesized in the heart and are known modulators of cell differentiation, whereas dysregulation of retinoid signaling in early embryo produces vascular dysmorphogenesis.9,13,14,23–25 In avians, RA signaling is initiated in coronary precursors at stages HH16 to HH17, when proepicardial cells activate RALDH2 expression. In avians and mammals, RA is produced by proepicardial-derived cells during formation of the epicardium, which is a continuous source of RA for the subepicardium and myocardium.9,13 Furthermore, epicardial cells undergoing EMT gradually lose expression of RALDH2 and WT1 as they migrate into the myocardium.8,16 Because this sequence of events is inversely related to coronary marker acquisition, RA is an excellent candidate to keep epicardial progenitors in an undifferentiated state, controlling the balance between proliferation and differentiation in coronary precursors.
RA Inhibits CoSMC Differentiation
By using viral/cellular/systemic methods to activate RA signaling, we showed that RA delays CoSMC differentiation through autocrine, paracrine, and systemic modes without disturbing cell proliferation, apoptosis, or EMT. Our data offer evidence that epicardially derived cells expressing RALDH2 produce their own RA and activate RA signaling in an autocrine fashion, being biased against CoSMC differentiation. By analyzing cell fates in the vicinity of cells transduced with the AdRaldh2eGFP, we suggest that the RA diffuses and inhibits CoSMC differentiation in their neighbors. This paracrine mode is supported by in vivo experiments in which SMαA and SMγA expression was precociously activated following systemic or local inhibition of RA synthesis.
The inhibitory effects of RA signaling on CoSMC differentiation offer a plausible explanation for the ease with which cultured proepicardial cells initiate their differentiation into CoSMCs. As we have shown, proepicardial cultures display an inefficient RA signaling. In fact, we showed that endogenous expression of qRALDH2 by cultured proepicardia containing ≈4000 cells is roughly equivalent to the level of mRALDH2 expression achieved through transduction of ≈10 times less cells. This level of qRALDH2 expression is apparently below the threshold for effective RA production, as indicated by the failure of retinol supplementation, or of DEAB, to change the numbers of CoSMCs in cultured proepicardia.
The combined effects of autocrine and paracrine RA on CoSMC differentiation can be modeled according to two schemes, depending on whether one believes that smooth muscle cells display a continuous range of differentiated phenotypes, or that each smooth muscle phenotype derives directly from a common CoSMC progenitor.26 Given the fact that we have identified a number of intermediate phenotypes between a promesenchymal, peripheral CoSMC and a more differentiated dense CoSMC, we currently favor the idea that RA inhibition of CoSMC differentiation reflects a global arresting effect of retinoids on CoSMC differentiation.
The Physiological Delay of CoSMC Differentiation Also Involves VEGF Signaling
We show that autocrine and paracrine RA act preferentially on CoSMC precursors to retard their differentiation, whereas leaving the timing of CoE differentiation unaffected (not shown). Using cocultures with myocardium, or fibroblasts, we provided evidence that the myocardium (but not fibroblasts) produces factors at levels able to inhibit CoSMCs and promote endothelial differentiation. We have shown that fibroblast cocultures complemented with virally delivered VEGF display decreased CoSMC differentiation and increased CoE differentiation, suggesting that VEGF plays a complex role in regulation of CoSMCs and CoE fates. This was confirmed in vitro by showing that VEGF controls CoSMC differentiation through autocrine, paracrine, and systemic modes.
The simplest way through which VEGF could decrease CoSMC differentiation is indirect, by shifting the fate of epicardially derived cells toward the endothelial phenotype. Our results on the activation of autocrine VEGF signaling by viral transduction clearly support this possibility. This indirect and autocrine, mode, however, does not appear to be the most physiological, or important one, because only 25.9% of VEGF-expressing cells activates endothelial marker expression, whereas CoSMC differentiation is reduced from 96.2% to 49.7%. This suggests that VEGF also works by directly inhibiting CoSMC differentiation. Thus, it is possible that VEGF produced by transduced cells diffuses and directly represses CoSMC differentiation in a paracrine fashion. We confirmed this by showing that although paracrine VEGF reduces CoSMC differentiation by 35.4%, it only marginally increases endothelial differentiation by 12.3%.
A Model for Coordinated CoE and CoSMC Differentiation
Coronary progenitors that reach subepicardial and intramyocardial domains are either under autocrine RA signaling because of their RALDH2 expression or within range of paracrine RA produced by neighboring cells. Thus, in our model (Figure 8), CoSMC differentiation is first checked by epicardial/subepicardial RA and later on by the direct action of myocardially secreted VEGF, which also orients uncommitted epicardially derived cells toward the CoE phenotype. Importantly, our results provide clues about the poorly understood mechanisms that trigger differentiation of CoSMCs, recruit them for medial layer formation, and synchronize these processes with the establishment of directional blood flow and shear stress in the newly formed coronary endothelial network. In our model, CoSMC differentiation is derepressed by the well-documented downregulation of RALDH2 expression in cardiac development9,15 and by reduced VEGF production secondary to efficient myocardial perfusion.27 It remains to be established whether RA and VEGF signaling inhibit CoSMCs additively, or synergistically, or whether they are in the same, or in parallel pathways. Moreover, although changes in HIF-1α activity secondary to the reduction of cardiac hypoxia induced by coronary perfusion27 offer a likely molecular explanation accounting for progressively reduced VEGF expression, the molecular pathways leading to progressive decreases in RALDH2 expression remain unknown. In summary, we provide evidence that RA is a crucial signal involved in the spatiotemporal control of CoSMC differentiation, acting coordinately with VEGF to regulate the formation of multilayered coronary blood vessels, an idea supported by recent demonstration of reduced VEGF receptor 2 and PECAM-1 expression in RALDH2 knockout embryos.28 Finally, our results provide important information on the cellular and molecular regulation of blood vessel morphogenesis, raising prospects for the discovery of additional regulatory networks during normal and abnormal coronary development.
Sources of Funding
Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) 05/60637-6, 06/50843-0; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES/DGU)4096/09-6; Conselho Nacional de Desenvolvimento Tecnológico (CNPq) 477418/2004-0, 305260/2007-3; Spanish Ministry of Innovation and Science BFU2009-07929; Junta de Andalucía P06-CTS-01614; Spanish cooperative networks on research TERCEL and RECAVA (ISCIII); European Union Sixth Framework Programme contract (“HeartRepair”) LSHM-CT-2005-018630.
We thank Marcio Bajgelman for advice on adenovirus production, Francisco Laurindo for discussion and insights, Lisete Teixeira and Milena Acencio for help with VEGF ELISA, and to Kleber Franchini and Carolina Clemente for FAK antibodies and protocols.
Mikawa T, Fischman DA. Retroviral analysis of cardiac morphogenesis: discontinuous formation of coronary vessels. Proc Natl Acad Sci U S A. 1992; 89: 9504–9508.
Perez-Pomares JM, Phelps A, Sedmerova M, Carmona R, Gonzalez-Iriarte M, Munoz-Chapuli R, Wessels A. Experimental studies on the spatiotemporal expression of WT1 and RALDH2 in the embryonic avian heart: a model for the regulation of myocardial and valvuloseptal development by epicardially derived cells (EPDCs). Dev Biol. 2002; 247: 307–326.
Lavine KJ, Long F, Choi K, Smith C, Ornitz DM. Hedgehog signaling to distinct cell types differentially regulates coronary artery and vein development. Development. 2008; 135: 3161–3171.
Landerholm TE, Dong XR, Lu J, Belaguli NS, Schwartz RJ, Majesky MW. A role for serum response factor in coronary smooth muscle differentiation from proepicardial cells. Development. 1999; 126: 2053–2062.
Niederreither K, Vermot J, Messaddeq N, Schuhbaur B, Chambon P, Dolle P. Embryonic retinoic acid synthesis is essential for heart morphogenesis in the mouse. Development. 2001; 128: 1019–1031.
Yoshida T, Owens GK. Molecular determinants of vascular smooth muscle cell diversity. Circ Res. 2005; 96: 280–291.
Tomanek RJ, Sandra A, Zheng W, Brock T, Bjercke RJ, Holifield JS. Vascular endothelial growth factor and basic fibroblast growth factor differentially modulate early postnatal coronary angiogenesis. Circ Res. 2001; 88: 1135–1141.
Vrancken Peeters MP, Gittenberger-de Groot AC, Mentink MM, Hungerford JE, Little CD, Poelmann RE. Differences in development of coronary arteries and veins. Cardiovasc Res. 1997; 36: 101–110.
Molin DG, Bartram U, Van der Heiden K, Van Iperen L, Speer CP, Hierck BP, Poelmann RE, Gittenberger-de-Groot AC. Expression patterns of Tgfbeta1–3 associate with myocardialisation of the outflow tract and the development of the epicardium and the fibrous heart skeleton. Dev Dyn. 2003; 227: 431–444.
Lai L, Bohnsack BL, Niederreither K, Hirschi KK. Retinoic acid regulates endothelial cell proliferation during vasculogenesis. Development. 2003; 130: 6465–6474.
Lin SC, Dolle P, Ryckebusch L, Noseda M, Zaffran S, Schneider MD, Niederreither K. Endogenous retinoic acid regulates cardiac progenitor differentiation. Proc Natl Acad Sci U S A. 2010; 107: 9234–9239.
Novelty and Significance
What Is Known?
The proepicardium is a transient embryonic structure that gives rise to epicardium and to coronary precursors.
Epicardially derived coronary precursors ingress into the myocardium, where they promptly undergo endothelial differentiation under the influence of cytokines such as fibroblast growth factor and vascular endothelial growth factor (VEGF) secreted by the heart.
Smooth muscle differentiation of cells derived from the epicardium is delayed in relation to endothelial differentiation, occurring only after a primitive endothelial outline of the coronary circulation is formed.
Before the endothelial network is established, epicardially derived cells are exposed to VEGF produced by cardiomyocytes and to intense RA signals triggered by expression of RALDH2, the major cardiac retinoic acid (RA)-synthesis enzyme.
After the onset of coronary circulation, the strength of both RA and VEGF signals is markedly reduced.
What New Information Does This Article Contribute?
The pronounced physiological delay between coronary endothelial and smooth muscle differentiation from their multipotent epicardial progenitors is controlled by both RA and VEGF signals.
Activation of autocrine and paracrine RA signaling inhibits coronary smooth muscle differentiation.
Activation of VEGF signaling, chiefly through a paracrine pathway, inhibits smooth muscle differentiation and further biases the differentiation of epicardial coronary progenitors to an endothelial fate.
Some aspects of coronary development, such as the pronounced temporal gap that separates endothelial from smooth muscle differentiation, are puzzling. This gap is critical to ensure the capacity for extensive remodeling before a complex, hierarchical vessel network is formed by the recruitment of extracellular matrix–producing smooth muscle cells. We show that this physiological delay is controlled by RA and VEGF signals. Using adenoviral vectors for RALDH2 and VEGF, and in vivo inhibition of RA synthesis, we show that RA acts as a brake to slow smooth muscle differentiation from coronary precursors. We show that besides inducing endothelial differentiation, VEGF acts directly on coronary precursors to decrease the expression of smooth muscle markers. Our results support a model in which epicardial/subepicardial RA production by RALDH2 and myocardial VEGF expression create an environment that prevents early smooth muscle differentiation, thus allowing the formation of an extensive endothelial network. Our results suggest that normal/abnormal coronary development involves novel cross-regulatory networks and suggest new approaches to direct differentiation of precursors toward specific coronary fates. We suggest that revascularization/regenerative therapies can be improved by taking into account the multistep nature of coronary development that coordinates inner and outer coronary vessel wall morphogenesis.
In April 2010, the average time from submission to first decision for all original research papers submitted to Circulation Research was 15.2 days.
Original received January 21, 2010; revision received May 18, 2010; accepted May 24, 2010.