Keywords
INTRODUCTION
The epicardiumis the outermost layer of the heart in vertebrates. In the adultorganism, it is constituted by a cubical mesothelium that covers aspace occupied by connective and vascular tissue. In the earlyembryo, the epicardium adopts the form of a squamous cellepithelium that either rests directly on the surface of themyocardium, or covers a subepicardial space that is more or lessdensely populated by mesenchymal cells. In this article we willreview the processes related with the development of theepicardium, appearance of the subepicardial mesenchyme, and itsdifferentiation. We will also examine evidence about the essentialrole of the subepicardial mesenchyme in myocardialdifferentiation.
The epicardiumis, without doubt, the cardiac component that traditionally hasreceived the least amount of attention from embryologists anddevelopmental biologists. The epicardium was long considered asimple derivative of the outermost layer of the heart tube. Theterm «epimyocardium» or «myoepicardium,»which refers to this primordium common to both tissues,1appeared in text books up until lately.2 Nevertheless,in recent years a series of studies have appeared that havedemonstrated, in first place, that the epicardium has a developmentindependent of the myocardium and endocardium, a proposal that wasmade almost a century ago.3 In second place, most of theconnective and vascular tissues of the heart, including thecoronary vessels, are epicardial derivatives. The most recentcontribution to the subject has come from studies that suggest thatcells derived from the epicardium provide essential signals for theformation of the compact layer of the ventricular myocardium. Theepicardium, therefore, has evolved from an apparently passive roleto prominence as a key protagonist of episodes of cardiacmorphogenesis. Our aim in writing this article was to reviewcurrent knowledge about these new functions of the epicardium andepicardium-derived cells.
The epicardium derives from an extracardiac mesothelialprimordium called the proepicardium
The epicardiumis the last layer of the heart to appear and is the only layer thatoriginates outside the primitive heart tube. General aspects of itsontogenetic and phylogenetic development have been recentlyreviewed.4 The epicardium forms from a clump ofmesothelial tissue that appears ventrally, on the limit between theliver and sinus venosus. This rudiment, which is called theproepicardium,5 can be simple and located to the rightof the sagittal plane, as in the case of the chick embryo, orbilateral, as in the mouse embryo.
The structure ofthe proepicardium varies according to the species in which it isdescribed. There are descriptions of the proepicardium inrepresentatives of almost all the major vertebrate groups. The mostprimitive representative, phylogenetically speaking, is a smallshark, the dogfish (Scyliorhinus canicula).6-8 Inthe embryos of these sharks, two large clumps of mesothelial cellsdevelop on the ventral and anterior portion of the epitheliumcovering the liver, where they join the sinus venosus. Later, whenthe septum transversum develops in this limit, the proepicardialmesothelial cushions are displaced towards the pericardial surfaceof the septum. The proepicardium of the dogfish is formed by roundmesothelial cells, with little extracellular matrix between them.In teleosts, the proepicardium has not been described, although hasbeen reported that the epicardium of the plaice (Pleuronectesplatessa) had an extracardiac origin, although itscharacteristics were not specified.9 The presence of theproepicardium in amphibians is deduced from studies of the cardiacdevelopment of the axolotl (Ambystomamexicanum).10 These authors affirm that theepicardium develops from the cells of the septum transversum.Information on this point is also scarce in reptiles, although hasbeen reported that the epicardial development of the turtle issimilar to that described in chick embryos.11
The organizationof the proepicardium has been extensively studied in chick andquail embryos.5,12-15 Its development begins aroundstages HH13-14, with rapid proliferation of the mesothelium thatcovers the horns of the sinus venosus ventrally on its limits withthe hepatic rudiment (Figures 1 and 2). Unlike what occurs in otheranimal models, the proliferation of the left side ceases while thatof the right side continues until it reaches a considerable size instage HH17. The proepicardium in this stage is formed by multipledigitations or protrusions that give it a«cauliflower-like» appearance. The digitations arecovered by mesothelium and contain numerous mesenchymal cells in anabundant extracellular matrix.
Fig. 1.The proepicardium of aquail embryo (stage HH21) by scanning electron microscopy. Aand B. Two different aspects of the proepicardialprotrusions in the phase of adhesion to the ventricle (V). Hindicates hepatic surface.
Fig. 2.Histological sections ofthe quail proepicardium in cross-section. A. Stage HH21. Theproepicardium (p) is located on the limit between the sinus venosus(SV)/Liver (L). Observe how the epicardium (EP) extends over themyocardium of the atrioventricular canal (AV). B. StageHH22, immunolocalization of the cytokeratin. The immunoreactivityof the proepicardial and epicardial mesothelium, as well as most ofthe subepicardial mesenchyme (SEM) is evident. Note that some cellsof the hepatic sinusoids (s) are also immunoreactive. M indicatesmyocardium.
Theproepicardium is bilateral in mammals, as has been noted, and itdevelops on the pericardial surface of the septum transversum, nearthe sinus venosus.16-20 The mesothelial protrusions atfirst consist in cushions of round cells and later acquire anappearance similar to that described in the chick embryo,digitations covered by mesothelium and containing mesenchymal cellsand extracellular matrix. In mouse embryos, the proepicardium ispresent from 9.5 to 11 days post coitus.
The factors thatcontrol the development of the proepicardium are not known. Amechanical effect has been proposed, an effect of«aspiration» produced by the cardiac contractions onthe mesothelium of the septum transversum.18 However, itis much more likely that the mesothelium of the zones in which theproepicardium develops receives some type of proliferative signalof unknown origin and nature. In fact, the proepicardium showsstrong mitotic activity.5,18
The proepicardium is transferred to the heart and originatesthe epicardial mesothelium
Theproepicardial cells, which originate and proliferate outside thelimits of the heart, must migrate to the cardiac surface in orderto constitute the primitive epicardium. Two main mechanisms oftransfer of the proepicardium have been described, which sometimescoexist in the same species.
In certaincases, proepicardial cells are shed and float free in thepericardial cavity. These cells adhere to specific areas of themyocardium, mainly in the atrioventricular and conoventricularfurrows. From the moment in which they adhere, the cells flattenand fuse to form an epithelium that progressively develops over thecardiac surface. This mechanism of proepicardial transfer is theonly one present in the dogfish8 and it has beendescribed in amphibians and mammals.10,17,18
The othermechanism of transfer of the proepicardium involves the directadhesion of proepicardial villi to the cardiac surface. Given theposition of the proepicardium with respect to the heart, thisadhesion usually occurs in the posterior (dorsal) part of theventricles and atrioventricular furrow (Figure 2). Direct adhesionof the proepicardium seems to be the fundamental mechanism oftransfer in the case of chick embryos, and coexists in amphibiansand mammals with the adhesion of free clumps of proepicardial cellsdescribed before. It is possible that the large communicationbetween the pericardial cavity and general coeloma of the body inthe bird embryo (due to the delayed development of the septumtransversum) is related to thisparticularity.10
In all thecases, the epicardium progresses around the atrioventricularfurrow, then extends to the left ventricle, ventral surface of theatrium, right ventricle and, finally, the roof of the atrium andoutflow tract.21 It is interesting to note that thegrowth of the epicardium stops at the limit between the myocardialand mesenchymal regions of the outflow tract, which is covered bythe growth of pericardial mesothelium.22
The adhesion ofproepicardial cells to the surface of the myocardium indicates theexistence of a specific mechanism of recognition. The details ofthis mechanism are not known because, although candidate moleculeshave been proposed, no conclusive evidence exists in their favor.For example, NCAM (neural cell adhesion molecule) is expressed inboth the epicardium and the developing myocardium.23,24Nevertheless, mutant mice deficient for NCAM do not presentepicardial defects.25 It has also been reported that thenaked myocardium presents discrete areas covered byfibronectin,20 and that the adhesion of free cellsinvolves a strong increase in fibronectin expression in theepicardium-myocardium interphase.26 Nevertheless, it isagain observed that mice deficient for fibronectin do not haveanomalies in epicardial development.27
In spite of theabsence of significant information about the mechanism of adhesionof proepicardial cells to the myocardium, there is evidence thattwo molecules are essential for maintaining epicardial integrity inthe initial moments of development. Mice deficient in VCAM-1(vascular cell adhesion molecule)28 or in theα4subunit of the integrins29 show a similar defect inepicardial development. This coincidence is consistent with therole of the α4 integrins (α4β1 andα4β7) as receptors for VCAM. The mice deficientfor these two molecules show a normal phase of adhesion ofproepicardial cells between days 9.5 and 11 of development, but theepicardium is shed immediately after and disappears. Thesemutations are lethal due, probably, to massive pericardialhemorrhage, the causes of which will be discussed below. VCAM-1 isexpressed in the embryonal myocardium, whereas integrinα4is expressed in the epicardium and proepicardium.30Therefore, their interaction seems to be essential for maintainingepicardial integrity.
The epicardium generates a population of mesenchymal cells byan epithelium-to-mesenchyme transition process
As theepicardium covers the embryonal myocardium, a space developsbetween these two tissues. This space, which we will call thesubepicardium, appears first around the atrioventricular andconoventricular furrows then later extends over the surface of theventricles and, in the case of mammals, the interventricularfurrow. The subepicardium is little developed in the atrium,particularly on the roof, where the epicardium adheres directly tothe myocardium. The formation of the subepicardial space can bedetermined by a change in the expression of adhesion moleculesand/or by an increase in the production of extracellular matrix inthe epicardium/myocardium interphase. This extracellular matrix isan extraordinarily complex medium that is rich in fibronectin andcollagens I, IV, V and VI,20,31-34 proteoglycans andlaminin,20 GP68,35 vitronectin, fibrillin-2and elastin,33 tenascin-X,36 andflectin.37
Thesubepicardial space is quickly populated by mesenchymal cells offibroblastoid appearance. For a long time it was thought that thesecells reached the subepicardium by migration from the region of theseptum transversum.15 Another possibility that has beendescribed in mammalian embryos is the transfer of cells insideproepicardial vesicles that are released into the pericardialcavity and adhere to the surface of the heart.38 Inbirds it is evident that the transfer of the proepicardium bydirect adhesion drags along mesenchymal cells that are thenincorporated by the subepicardium. Nevertheless, it is currentlyaccepted that much of the subepicardial mesenchyme is constitutedby cells derived from epicardium (CDEP).39
The CDEPs aregenerated by an epicardium-to-mesenchyme transition phenomenon,which is a particular case of a family of cellular processes thatare very important for development, the epithelium-to-mesenchymetransitions.40 These processes involve the acquisitionby epithelial cells of mesenchymal characteristics, which allowthem to separate from their neighboring cells, reorganize theircytoskeleton, break down the basement membrane and underlyingextracellular matrix, and acquire the capacity to migrate throughthis membrane. Examples of epithelium-to-mesenchyme transitions arethe formation of the mesoderm on the primitive line,41neural crest differentiation,42 dermomyotomedisintegration, or the formation of valvuloseptal mesenchyme in theendocardial cushions.43
In the case inquestion, the epicardium-to-mesenchyme transition begins in theatrioventricular and conoventricular furrows and later extends toother areas of the ventricular epicardium. However, most of CDEPs,at least in mammalian embryos, seem to be generated in theatrioventricular furrow (Figure 3A).
Fig. 3.Evidences ofepicardium-to-mesenchyme transition in mammalian embryos. A.Mouse embryo, 11.5 days post coitus, scanning electron microscopy.Cells of the epicardium (EP) of the atrioventricular furrow emitlong basal prolongations toward the underlying extracellularmatrix. M: myocardium. B. Hamster embryo, 11 dayspost-coitus, immunolocalization of cytokeratin. The mesenchymalcells that are generated in the atrioventricular furrow arepositive (arrow), demonstrating a mesothelial origin. A indicatesatrium; V, ventricle. C. mouse embryo, 11.5 dayspost-coitus, immunolocalization of factor WT1. Observe thereactivity of the epicardium and cells of the atrioventricularfurrow.
What proportionof subepicardial mesenchyme is constituted by CDEPs? Beforeanswering this question, it is necessary to note that a large partof the proepicardial mesenchyme also originates byepithelium-to-mesenchyme transition from the mesothelium coveringproepicardial vellosities.44 This is suggested bymorphological data, as well as by the presence of markers of theepithelium-to-mesenchyme transition in the proepicardialmesothelium and mesenchyme, as we will see below. This means thatcells derived from the proepicardial mesothelium are incorporatedinto the population of CDEPs generated in situ, so that mostof the subepicardial mesenchyme derives from the coelomicmesothelium, whether proepicardial or epicardial.
Theepithelium-to-mesenchyme transition involves cytoskeletalreorganization, as mentioned before. The cells of the epicardium(and of the proepicardial mesothelium) have intermediate filamentsconstituted by cytokeratins.21 Durin g thetransformation into mesenchyme, these filaments are replaced byfilaments of vimentin (which are intrinsic to mesenchymalcells).45 Of course, this substitution is notinstantaneous. Vimentin expression begins even in premigratoryphases, when the cell in transition still conserves its epithelialphenotype.40 On the other hand, cytokeratin is degradedprogressively throughout the process, but persists for a timeduring in mesenchyme derived from epithelium. This implies that thelocation of cytokeratin in mesenchymal cells is a marker of recentepithelial origin.45 In fact, the great majority of theproepicardial and subepicardial mesenchymal cells areimmunoreactive for anticytokeratin antibodies39 (Figures2B and 3B). On the other hand, numerous proepicardial andepicardial mesothelial cells are vimentin-positive, possiblyindicating that they are in premigratorystates.39
Likewise, thetranscription factor associated with Wilms´ tumor (WT1) isexpressed in the coelomic mesothelium, epicardium, and CDEPs, amongother locations46,47 (Figures 3C and 5B). The massivepresence of WT1 protein in the proepicardial and subepicardialmesenchyme of the chick has been characterized as evidence of itsmesothelial origin.48 On the other hand, mice that arecarriers of a reporter WT1 gene, with expression ofβ-galactosidase controlled by the WT1 promoter, show expressionof this gene in practically all the subepicardial cells of themouse.46
The mechanismthat regulates the epicardium-to-mesenchyme transformation islittle known, although there have recently been important advancesin this sense. Transformation begins, and is more intense, at thelevel of the atrioventricular cushions and the outflow tract. Sinceanother epithelium-to-mesenchyme transition is taking placesimultaneously in these cushions, which generate the valvuloseptalmesenchyme,43 from the first it was suspected that thesame signal from the myocardium might initiate both processes. Thissignal was associated with the presence of «adherons,»which are complex particles composed by several proteins that havebeen found in the extracellular matrix of the cushions andsubepicardium.49 At present it is thought that thegrowth factors of the BMP (bone morphogenetic proteins) andTGF-β (transforming growth factors-β) families are essential forendothelium-mesenchyme transformation.50,51 However, thesituation is much less clear in the case of the epicardium. Theprobable role of the FGFs (fibroblast growth factors), specificallyFGF-1, 2 and 7, has been reported.52 These factorsstimulate epicardium-to-mesenchyme transformation in vitro,while TGF-β1, 2 and 3 inhibit it. Other authors insist,however, that stimulation by FGFs is possible only when theepicardial cells have been previously activated by BMP type signals(Markwald, personal communication). It is important to note thatBMP-2 and BMP-4 are specifically expressed in the myocardium of theatrioventricular canal and outflow tract.53
In any case,after the signal or signals that induce the onset of theepithelium-to-mesenchyme transition, the implication of zinc-fingertype transcription factors pertaining to the Snail familyseems clear. Apparently, the functions of Snail in mammalsare carried out in the avian embryo by the product of another gene,Slug.54 These factors are essential to theformation of mesoderm and participate in different embryonalprocesses of epithelium-to-mesenchyme transformation, including theformation of the neural crest.55 The expression of Slug has been shown to be essential for the transformation ofthe endocardium in endocardial cushions,51,56 whereasthe presence of Slug has been found in the epicardium andCDEPs of the chick embryo.57
The function ofthe Slug/Snail factors seems to be to repress the expressionof cellular adhesion molecules. Slug, for example, repressesthe expression of desmoplakins and desmogleins.58 On theother hand, it has been demonstrated that Snail repressesthe expression of E-cadherin in what could be a key event inepithelium-to-mesenchyme transformation.59,60
Othertranscription factors of the zinc-finger type that are probablyimplicated in the epicardium-to-mesenchyme transition are Ets-1 andWT1, which has already been mentioned. Ets-1 activates theexpression of proteolytic enzymes and seems to be a key factor inthe degradation of the extracellular matrix that is associated withthe migrator phenotype.61 The presence of Ets-1 has beencorrelated with the areas of epicardium-to-mesenchyme transition inchick embryo.62 WT1 seems to have a function of its own,if not in the epicardium-to-mesenchyme transition per se,then in the differentiation of the CDEPs. WT1 could repress thedifferentiation of these cells, keeping them in a mesenchymal andproliferative state.48 This would explain the smallernumber of CDEPs observed in the heart of mouse embryos deficient inWT1, as we will see below.
FOG-2 (friend ofGATA) is a transcription factor expressed in the myocardium thatalso seems to have an essential function inepicardium-to-mesenchyme transition. FOG-2 is a cofactor of thetranscriptional factors of the GATA family, three of whose members(GATA4, 5 and 6) are expressed in the embryonal heart.63FOG-2 deficiency produces a cardiac phenotype characterized by alow number of CDEPs, absence of coronary vessels, and hypoplasia ofthe ventricular myocardium.64 FOG-2 could be implicatedin the generation of the myocardial signal for beginningepicardium-to-mesenchyme transformation.
Next, othertranscription factors expressed in the epicardium are cited thatcould participate in the generation of CDEPs.Epicardin65 (also described as capsulin66,67and POD-168) pertains to the bHLH family (basicHelix-Loop-Helix). It is expressed in the embryonal epicardium andin the mesothelium and submesothelial mesenchyme of the lungs,digestive tract, kidneys, and spleen. Its absence producespulmonary and renal hypoplasia,69 as well as agenesis ofthe spleen,70 but cardiac disturbances have not beendescribed in this model. On the other hand, two genes of epicardialexpression, Tbx571 and Tbx18,72pertain to the family of the T-Box factors (related with Brachyury). Mutations in Tbx5 are associated with theHolt-Oram syndrome.73 Finally, the embryonal epicardialexpression of the Rb (retinoblastoma) tumor suppressor, a proteininvolved in the control of the cell cycle that acts as a substrateof cdks (cyclin-dependent kinases), has also been detected. Thepattern of expression of Rb suggests its involvement inendocardium-to-mesenchyme transformation,74 which is whya parallel function in epicardial transformation cannot beexcluded.
The epicardium-derived cells differentiate into connectiveand vascular tissue
Thefibroblastoid appearance of CDEPs was the reason why they werefirst thought to be components of the subepicardial connectivetissue. Nevertheless, evidence suggests a more active role in thedevelopment of the coronary vascular system.
For a long timecoronary vessels have been considered to derive from buds of theaortic root that grow and invade the entire heart. At the end ofthe 1980s, a hypothesis (denominated the ingrowth hypothesis) wasformulated that these vessels organize in the subepicardium to forma vascular plexus that connects with the right and left Valsalvasinuses in a given moment.75 The sudden increase inintravascular pressure that originates this connection induces thearterialization of specific segments of the plexus that constitutethe coronary arteries.
According tothis hypothesis, the primary subepicardial vascular plexusorganizes by the connection of vascular precursors, that is to say,by vasculogenesis.76 Once established, it grows byangiogenesis or proliferation of established vessels. Thevasculogenic origin of the coronary vessels raised the question ofthe origin and differentiation of their cellular precursors, whichnecessarily had to be found in the subepicardialmesenchyme.
Experimentalevidence has demonstrated that the smooth muscle precursors of themedia of coronary vessels, as well as the fibroblasts of theadventitia, are differentiated from CDEPs. A fundamental techniquefor reaching this conclusion were the «epicardialchimera,» which are developed by grafting quail proepicardiumin the pericardial cavity of chick embryos in the HH17 stage ofdevelopment.22 This generates chick embryos whose heartsare covered by quail epicardium. The destiny of CDEPs derived fromthis epicardium can be followed using antibodies that recognizequail cells, but not chicken cells (Figure 5).
Using epicardialchimera, it was demonstrated that CDEPs from the donor contributedto the formation of the coronary musculature, vascular adventitia,and fibrous skeleton of the heart.77,78 In this system,the coronary endothelium also derived from the quail, although theexplanation given by the authors of these studies did notcontemplate a CDEP origin of the coronary endothelium, as we willsee below. The epicardial chimera used in these experiments showedthat CDEPs invade the myocardium and even colonize theatrioventricular cushions, reaching subendocardial levels. However,the cushions of the outflow tract do not show the presence ofCDEP.
Otherexperiments have confirmed these findings. The culture ofproepicardial cells on collagen gels generates a monolayer ofepithelial cells identical to those of the epicardium. These cells,once isolated, marked, and injected into the pericardial cavity ofchick embryos, give rise to the epicardium, coronary smooth muscle,and fibroblasts.79 In another series of experiments ithas been shown that the epicardial cells generated on collagen gelsby the culture of proepicardial explants expressed markers ofsmooth muscle in vitro, such as caldesmon and specificisoforms of actin and myosin.80 This process isstimulated by the presence of PDGF-BB (platelet-derived growthfactor-BB) and is dependent on the expression of SRF (serumresponse factor), a transcription factor of the MADS boxsuperfamily, which is expressed by the epicardium inculture.81 In fact, dominant-negative geneticconstructions for SRF inhibited the differentiation of smoothmuscle from CDEPs.
The BVES (bloodvessel/epicardial substance) protein, which probably represents anew family of adhesion molecules, has been found throughout theentire proepicardium-epicardium-CDEP-vascular smooth muscle cellline, and constitutes further evidence of the epicardial origin ofthe coronary medial layer.82,83Thedifferentiation of smooth muscle cells and fibroblasts from CDEPsis a well established phenomenon. Nevertheless, the origin of thecoronary endothelium is still debated. The first phase of theprocesses of vasculogenesis is the assembly of angioblasts andmesodermal endothelial precursors. In a second phase, theendothelial tubes recruit mesenchymal cells and induce theirdifferentiation into perivascular cells (pericytes and smoothmuscle cells). The origin of the angioblasts that give rise to theprimitive subepicardial plexus is uncertain, although there are twopossibilities that are not mutually exclusive. The firstpossibility is that the angioblasts migrate to the subepicardiumfrom the region of the liver and septum transversum, either throughthe proepicardium, or directly, when the subepicardium hasconnected with the hepatic splachnopleura.15 Thesecond possibility is by the differentiation of CDEPs.44In this case, the epicardium would be the origin of both theendothelium and the smooth muscle of coronary vessels.
The firstpossibility is supported by the findings of certain experiments, inwhich the epicardial chimera only developed endothelium derivedfrom the donor if the proepicardial graft was accompanied by afragment of liver tissue.84 This is why other chimera,which developed coronary endothelium from the donor, were notconsidered to be proof of the epicardial origin of thisendothelium.77 Nevertheless, the hypothesis of a hepaticorigin of the coronary precursors encounters two difficulties. Inthe first place, epicardial chimera in which the donor epicardiumforms a mosaic with the host epicardium showed how the limitsbetween the respective mesenchymes were perfectly clear andcoincided with the limits between epicardia.44 Stated inother terms, beneath the quail epicardium only quail mesenchyme isfound, and beneath the chicken epicardium, only chicken mesenchyme.Logically, the coronary vessels that developed in this system werea mosaic of endothelial cells from the donor and receptor. Theabsence of horizontal migration of the subepicardial mesenchymedoes not fit the hypothesis of an invasion of the cardiac surfaceby extracardiac angioblasts.
The secondargument to consider regarding a hepatic origin of the coronaryangioblasts is based on evidence of hepatic vasculogenesis. It islikely that the endothelium of hepatic sinusoids is a coelomicderivative, an idea that was proposed a long time ago85and has recently received experimental support. For example,cytokeratin remains can be found in the early sinusoidal cells ofthe chick embryo39 (Figure 2B), whereas in transgenicmice with the WT1 reporter gene, the hepatic endothelium expressesthis mesothelial marker.46 Direct labeling of thehepatic mesothelium of chick embryos with a fluorescent markershows that endothelial cells in the sinusoids are marked after only24 h of reincubation.86
The possibilitythat CDEPs differentiate into angioblasts is supported byexperimental evidence. This evidence includes the combined presencein subepicardial cells of cytokeratin remains with VEGFR-2(vascular endothelial growth factor receptor-2), also known asFlk-1,87 the earliest known vascular marker, as well asa series of experiments with epicardial chimera,44 andthe findings of direct proepicardial mesothelial and epicardiallabeling with fluorescent tracers and retrovirus.88 Ifthese findings are confirmed, both the embryonal heart and liverreceive angioblastic cells from the coelomic mesothelium. This is aprocess that may be generalized in other organs in whichvasculogenesis occurs (vitelline sac, lungs, digestive tube,allantois), and has been situated in a conceptual frameworkexplaining the origin of the vertebrate circulatorysystem.89
Epicardium-derived cells could be pluripotential vascularprecursors
The possibilitythat CDEPs give rise to at least three cell types (fibroblasts,smooth muscle cells and endothelium) raises interesting questionsabout the mechanisms that regulate this pluripotentiality. In thiscontext, it is important to underline the recent discovery ofbipotential vascular precursors derived from embryonal stem cellsand selected by VEGFR-2 expression.90 These cells, whencultured in the presence of serum or PDGF-BB, differentiate intosmooth muscle cells. However, in the presence of VEGF (vascularendothelial growth factor) they give rise to endothelial cells. Theexposure of these bipotential precursors to both growth factorsinduces the formation of mixed cultures in which smooth musclecells surround endothelial cells.
CDEPs couldconstitute bipotential vascular precursors from the moment in whichthey both express high-affinity receptors for the two growthfactors previously mentioned, VEGFR-2 andPDGFRβ87-91 (unpublished data). If this hypothesisis confirmed, an attractive scenario for the development of cardiacvascularization would be the following (Figure 4): myocardialsignals (BMPs, FGFs) would induce the transformation of epicardiuminto pluripotential mesenchyme. The first populations of CDEPswould be induced to differentiate into endothelial cells by thehigh level of epicardial and myocardial production ofVEGF.92 These cells would organize into a primaryvascular plexus that would recruit successive CDEPs through theproduction of PDGF-BB, a growth factor produced by theendothelium.93
Fig. 4.Hypothetical model of theorigin of the coronary vessels from cells derived from theepicardium (CDEP). 1. In the first phase, growth factors ofthe BMP and FGF families induce the epicardium-to-mesenchymetransition. 2. The presence of VEGF secreted by themyocardium and epicardium induces the endothelial differentiationof the CEDPs. 3. In the second phase, the production ofPDGF-BB by the endothelium formed induces the recruitment of newCDEPs and their differentiation into pericytes and smoothmuscle.
Variousobservations seem to support this hypothetical model. Thevascularization of the heart, like that of other embryonal organs,depends on a precise dose of VEGF. The absence of VEGF is lethal,even in heterozygosis,94,95 and myocardialoverexpression of VEGF leads to def ectivevascularization96 and vascular dilation, very similar towhat takes place in models in which signaling byPDGF-BB/PDGFRβ is disturbed.97
The phenomenonof the pluripotentiality of CDEPs could extend further thansuggested here, since there are two points that have not beenexamined, but will have to be considered in the future. In thefirst place, in spite of numerous morphologic descriptions to thiseffect, the existence of hematopoiesis in the embryonal heart isstill uncertain. The development of the subepicardial vascularplexus has been described in mammalian embryos (including humans)as a process of the fusion of «bloodislets.»18,98 Evidence in favor of the existenceof a common precursor of blood and endothelial cells, known as thehemangioblast,99 suggests that the process ofdifferentiation of the coronary precursors could be more complexthan described and leaves open the possibility of an even greaterdegree of CDEP potentiality. The role of the surprising epicardialexpression of erythropoietin and its EPOr receptor is uncertain inthis context.100 On the other hand, on several occasionsit has been suggested that the subepicardial mesenchyme could bedifferent in myocardiocytes,101 although no evidence ofthis possibility has been found in epicardialchimera.22
Epicardium-derived cells modulate myocardialdevelopment
A series ofrecent experiments have suggested that, in addition to contributingto the development of the vascular and connective tissue of theheart, as has been described, CDEPs could have an essentialmodulating role in the development of the ventricular myocardiumand, in particular, of the compact layer. Evidence to this effectis described below.
Adult myocardialcells in primary culture suffer a process of dedifferentiation bywhich they change their structure, function and gene expressionprofile. Co-culture of epicardial and myocardial cells has beenshown to delay this process and maintain the contractile phenotypeof the myocardium for a longer time.102 This phenomenon,which requires contact between the epicardium and myocardium, hasbeen attributed to some type of interaction of a physiologicalnature.
In bothmammalian and avian embryos, the first phases of ventricularcompaction coincide with an invasion of the myocardium byCDEPs.48,77 It is interesting to observe that thisinvasion takes place in the ventricle, but not in the atrium(Figure 5). The cells that penetrate the ventricular myocardiummaintain the expression of the WT1 transcription factor, incontrast with the CDEPs that differentiate into vasculartissue.103 One of the first indications of themodulating role of the CDEPs on myocardial differentiation comesfrom the study of the cardiac phenotype in mice deficient inWT1.104 The loss of WT1 function produces severeanomalies in the development of gonads, kidneys, the spleen, andadrenal glands. Nevertheless, the lethal nature of this mutation isdue to heart failure induced by ventricular hypoplasia,characterized by a thin ventricular wall and non-formation of thecompact layer. In WT1-/- mice, the formation of theepicardium takes place normally and CDEPs are generated, but insmaller amounts than in normal mice. Since WT1 is expressedexclusively in the epicardium and CDEPs, and these cellsspecifically invade the ventricular myocardium, it seems thatmyocardial compaction is dependent on the CDEP invasion.
Fig. 5.Signaling function ofcells derived from epicardium (CDEP), manifested in chick/quailchimera (quail epicardium on chick myocardium). The cells of thequail are recognized in all the cases by the immunolocalization ofQCPN antigen (in green). A. Stage HH29. The ventricle (v)has been completely invaded by CDEPs. EP indicates epicardium. B. Stage HH32, atrioventricular furrow. The WT1 transcriptionfactor has been immunolocalized in red. Observe the co-localizationof QCPN and WT1 in most of the subepicardial mesenchyme (SE).Co-localization results in a yellow color. The presence ofWT1-positive cells is also observed in the ventricle (V), but notin the atrium (A). C. Stage HH29, conoventricular furrow.The retinaldehyde dehydrogenase enzyme has been immunolocalized inred. Observe the reactivity of the epicardium and CDEPs, as well asthe presence of CDEPs in the ventricle (V), but not in the outflowtract (OT).
It is necessaryto comment that this so-called thin myocardium syndrome thatcharacterizes WT1 mutation also takes place in association with theimpaired function of various genes involved in epicardialdevelopment, as has already been mentioned in this review, such asFOG-2, VCAM-1, α4 integrin, erythropoietin, and its EPOrreceptor.
The precisemechanism of epicardium/myocardium interaction is not known,although signaling by retinoic acid seems to be involved. In fact,the epicardium and CDEPs express retinaldehyde-dehydrogenase-2(RALDH2), a key enzyme in the synthesis of retinoicacid103,105 (Figure 5C). It has been confirmed that thecells that invade the myocardium also express this enzyme, althoughthe expression tends to decrease with time and disappears with thedifferentiation of the CDEPs.103 This observation isrelated with the fact that mice deficient in the RXR* receptor of retinoic acidpresent a ventricular hypoplasia identical to that ofWT1-/- mice.106,107 In the first case, itcould be thought that the retinoic acid produced by the epicardiumand CDEPs is the signal that induces the formation of theventricular compact layer. However, the process seems to be morecomplex, since conditioned mutations that specifically annul theexpression of RXRα in the myocardium areinnocuous.108,109 Since CDEPs also express thisreceptor, it is more likely that retinoic acid is essential to theformation of an autocrine loop that maintains the CDEPs in aundifferentiated state and generating signals to the myocardium.These signals are of unknown nature, but they could maintain theventricular myocardium in a proliferative state. The absence of WT1may induce the premature differentiation of the CDEPs, which ceaseto produce both retinoic acid and signals to the ventricularmyocardium.
The functions ofthe CDEPs in cardiac development seem to depend, therefore, on avery fine balance between signals that originate in both themyocardium and the CDEPs, a balance that allows specificsubpopulations of CDEPs to enter differentiation pathways andcontribute to the connective and vascular tissue, or to remainundifferentiated, migratory, and producing autocrine and paracrinesignals. Evidence that this balance is exquisitely regulated comesfrom genetic manipulation that affects certain growth factors. Forexample, mice in which VEGF-A overexpression occurs in themyocardium develop large epicardial vessels, as could be expected,but they also show ventricular hypoplasia.96 A possibleinterpretation of this phenomenon is that the rupture of thebalance between CDEP subpopulations reduces the number of cellsmodulating the development of the ventricular compactlayer.
A final topic toaddress in this chapter are the possible relations between theCDEPs and the differentiation of conduction tissue. Somesuggestions to this effect have been made, indicating the spatialcoincidence between the distribution of the CDEPs, the maincoronary vessels that derive from them. and the Purkinjefibers.77 It has been suggested that endothelin producedby the coronary endothelium could mediate the differentiation ofconduction tissue.110
CONCLUSIONS AND PERSPECTIVES
Throughout thisarticle we have shown how the importance of role of the epicardiumin cardiac development has become progressively more apparent inrecent years. This is due to the discovery that the epicardiumprovides an important population of mesenchymal cells and to thepluripotential nature of these cells, a characteristic whosesignificance will have to evaluated in the coming years.
Thepluripotentiality of the CDEPs brings us to a concept that isbeginning to circulate among researchers in cardiac development,the idea that the epicardium constitutes a population of potentialcardiac stem cells. It should not be overlooked that the myocardiumand endocardium derive from the precardiac mesoderm when thismesoderm is a true coelomic epithelium, which means that the samecell type continues ontogenetically in the proepicardium andepicardium. On the other hand, the potentiality of the coelomicepithelia or mesothelia is dramatically illustrated by the tissueheterogeneity of a group of malignant mesotheliomas that give riseto osseous, cartilaginous, muscular, or hemangioblastic elements.The term «mesodermoma» has been proposed for thesetumors, since it is considered that the coelomic mesotheliumrecovers truly ancestral mesodermal properties in the process oftumor formation.111,112
It is possiblethat the adult epicardium retains a capacity for responding tosignals that can induce their transdifferentiation into cell typesof interest for the treatment of various pathologies. Many clinicalpossibilities could derive from such speculations.
Correspondence: Dr. R. Muñoz-Chápuli.
Departamento de Biología Animal.
Facultad de Ciencias. Universidad de Málaga.
29071 Málaga. España.
E-mail: chapuli@uma.es