cKit+ cardiac progenitors of neural crest originKonstantinos E. Hatzistergosa, Lauro M. Takeuchia, Dieter Saurb, Barbara Seidlerb, Susan M. Dymeckic, Jia Jia Maic,Ian A. Whitea, Wayne Balkana, Rosemeire M. Kanashiro-Takeuchia,d, Andrew V. Schallye,1, and Joshua M. Harea,1

aInterdisciplinary Stem Cell Institute, Leonard M. Miller School of Medicine, Miami, FL 33136; bDepartment of Internal Medicine, Mediziniche Klinik undPoliclinik Der Technischen Universitat Munchen, Munich 81675, Germany; cDepartment of Genetics, Harvard Medical School, Boston, MA 02115;dDepartment of Molecular and Cellular Pharmacology, Leonard M. Miller School of Medicine, Miami, FL 33136; and eDepartment of Pathology andMedicine, University of Miami School of Medicine and Veterans Affairs Medical Center, Miami, FL 33125

Contributed by Andrew V. Schally, August 29, 2015 (sent for review April 27, 2015; reviewed by Roger Joseph Hajjar)

The degree to which cKit-expressing progenitors generate cardio-myocytes in the heart is controversial. Genetic fate-mapping studiessuggest minimal contribution; however, whether or not minimalcontribution reflects minimal cardiomyogenic capacity is unclear be-cause the embryonic origin and role in cardiogenesis of these pro-genitors remain elusive. Using high-resolution genetic fate-mappingapproaches with cKitCreERT2/+ and Wnt1::Flpe mouse lines, we showthat cKit delineates cardiac neural crest progenitors (CNCkit). CNCkit

possess full cardiomyogenic capacity and contribute to all CNC de-rivatives, including cardiac conduction system cells. Furthermore, bymodeling cardiogenesis in cKitCreERT2-induced pluripotent stem cells,we show that, paradoxically, the cardiogenic fate of CNCkit is regu-lated by bone morphogenetic protein antagonism, a signaling path-way activated transiently during establishment of the cardiaccrescent, and extinguished from the heart before CNC invasion. To-gether, these findings elucidate the origin of cKit+ cardiac progenitorsand suggest that a nonpermissive cardiac milieu, rather than minimalcardiomyogenic capacity, controls the degree of CNCkit contributionto myocardium.

cardiac stem cells | cardiac neural crest | cardiomyogenesis | BMP signalling

Heart development is a highly regulated process during whichcell lineage diversification and growth programs are dynam-

ically coordinated in temporal and spatial manners (1). Theseprograms are activated sequentially, in parallel, or intersect to giverise to distinct heart domains. For example, the myocardial lineageoriginally develops from cardiac progenitors (CPs) of mesodermalorigin (2–5), which form the first and second heart fields. How-ever, later during morphogenesis, the cardiomyogenic programdiverges and activates cardiomyocyte proliferation signals, alongwith CPs from the hemogenic endothelium, epicardial, cardio-pulmonary, and cardiac neural crest (CNC) lineages, to producenew cardiomyocytes (1, 6–11). Gauging the relative contributionof each lineage for scaling their cardiomyogenic—and consequentlytherapeutic—capacity is a challenge. For example, many of the CPlineages are heterogeneous and incompletely characterized, andtherefore cannot always be traced under a straightforward geneticfate-mapping experiment. Furthermore, it is unknown whether andhow changes in the cardiac milieu (i.e., morphogens, tissue com-position, and size) regulate the final proportions of heart musclederived from each lineage.cKit is a receptor tyrosine kinase that marks several cell lineages,

including neural crest (NC), hematopoietic, and germ-line stem cells(12–15). Following the seminal description by Beltrami et al. (16) ofclusters of cKit cells in the postnatal mammalian heart, severallaboratories, including ours, suggested that cKit marks CPs (16–19),a finding that led to the clinical testing of these cells for heart repair(20). Recently, a straightforward genetic fate-mapping study showedthat a relatively small proportion of murine myocardium is derivedfrom cKit+ CPs, leading to the conclusion that the cardiomyogeniccapacity of cKit+ CPs is functionally insignificant (21). However, theidentity of cKit+ CPs and the mechanisms controlling their differ-entiation into cardiomyocytes remain controversial (22). Here, byusing a high-resolution genetic lineage-tracing strategy, as well asinduced pluripotent stem cell (iPSC)-based models of cardiogenesis,

we demonstrate that cKit marks CNCs. Furthermore, we show thattheir relatively small contribution to myocardium during embryogen-esis is not related to poor cardiomyogenic capacity, but rather tochanges in the cardiac activity of the bone morphogenetic protein(BMP) pathway that prevent their differentiation into cardiomyocytes.

ResultsGenetic Lineage-Tracing of cKit+ CPs. We used a well-characterizedcKitCreERT2/+ mouse line to lineage-trace cKit+ CPs (23–25).cKitCreERT2/+ are healthy, fertile, and express the white spottingphenotype (12, 23, 24, 26) (Fig. 1A).We first investigated whether cKit marks mesodermal CPs (e.g.,

first- or second-heart field CPs; or primitive hemogenic lineage)(1), by administering pregnant mice carrying cKitCreERT2;IRG em-bryos with tamoxifen (TAM) from embryonic days (E)7.5 to E8.5(Fig. 1B and Table S1). At E18.5, EGFP expression was detected inmesodermal cells (13, 14, 21, 26), including gonads, blood, andlungs (Fig. 1 C and D). At this stage of labeling (21), EGFP wasrarely detected in the heart, and EGFP+ heart cells were non-cardiomyocytes with rare colocalization with the cardiac tran-scription factor Gata4 (Fig. 1 E and F).Next, to test whether cKit marks other cardiomyogenic line-

ages (e.g., proliferating cardiomyocytes; or CPs of the epicardial,CNC, and definitive hemogenic lineages) (1), we administeredTAM to pregnant mice at selected time points during E9.5–E12.5

Significance

A high-resolution genetic lineage-tracing study in mice revealsthat cKit identifies multipotent progenitors of cardiac neuralcrest (CNC) origin. Normally, the proportion of cardiomyocytesproduced from this lineage is limited, not because of poordifferentiation capacity as previously thought, but because ofstage-specific changes in the activity of the bone morphoge-netic protein pathway. Transient bone morphogenetic proteinantagonism efficiently directs mouse iPSCs toward the CNClineage and, consequently, the generation of cKit+ CNCs withfull capacity to form cardiomyocytes and other CNC derivativesin vitro. These findings resolve a long-standing controversyregarding the role of cKit in the heart, and are expected to leadto the development of novel stem cell-based therapies for theprevention and treatment of cardiovascular disease.

Author contributions: K.E.H., A.V.S., and J.M.H. designed research; K.E.H., L.M.T., andR.M.K.-T. performed research; K.E.H., D.S., B.S., S.M.D., J.J.M., I.A.W., and J.M.H. contrib-uted new reagents/analytic tools; K.E.H., L.M.T., R.M.K.-T., and J.M.H. analyzed data; andK.E.H., D.S., W.B., A.V.S., and J.M.H. wrote the paper.

Reviewers included: R.J.H., Mount Sinai School of Medicine.

Conflict of interest statement: K.E.H. and J.M.H. report having a patent for cardiac cell-based therapy. K.E.H. and J.M.H. own equity in Vestion Inc. and are members of thescientific advisory board and consultants of Vestion, Inc. J.M.H. is a board member ofVestion Inc. Vestion Inc. did not participate in funding this work. The other authors reportno conflicts.1To whom correspondence may be addressed. Email: jhare@miami.edu or andrew.schally@va.gov.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1517201112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1517201112 PNAS | October 20, 2015 | vol. 112 | no. 42 | 13051–13056

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0

(Table S1). Cre-mediated recombination resulted in EGFPexpression in embryonic melanoblasts, craniofacial cells (27),neural tube (NT), dorsal root ganglia (DRGs), blood, gastroin-testinal cells, gonads, and pulmonary cells (Fig. 1 G–J). Unlikethe fate-map of E7.5–E8.5 cKit-expressing cells, EGFP epifluor-escence is detected within the cardiac outflow tract (OFT), epi-cardium, and myocardium (Fig. 1 G–J).To rule out a limited transgene expression, we also performed

fate-mapping using the R26 promoter-driven R26RlacZ allele. Theresults were similar using this reporter compared with EGFP (Fig.1 K–N).To identify the original population of CPs, we administered

TAM at selected time points between E9.5 and E13.5, and col-lected embryos 24 h after the last injection (Fig. S1A). Using liveepifluorescence and immunofluorescence (IF) imaging, we de-tected Cre-recombined cells in the NT, skin, lungs, gut, cono-truncus, OFT, and epicardium, but not within the myocardium(Fig. S1). These findings suggest that the cKitCreERT2/+-labeled heartcells during the period of TAM-induced recombination do not arisewithin the myocardium [i.e., differentiated cardiomyocytes (28) ortransdifferentiating cardiac fibroblasts (29) and hemogenic pro-genitors (6, 7)], and derived from an extracardiac CP lineage.Finally, similar to cKitCreERT2/+ reporters, cKit IF in cKit+/+

embryos marked cells within the NT, skin, lung, OFT, and epi-cardium, but not differentiated cardiomyocytes (Fig. S2).Collectively, our findings suggest that cKit marks a CP lineage

that emerges at ∼E9.5 and contributes to the development of themouse heart.

Intersectional Genetic Fate-Mapping of cKit and Wnt1 Protooncogenes.Because our findings are consistent with a CNC origin of cKit+

CPs, we used a well-established CNC-specific mouse, the Wnt1-Cre;RC::tdTomato (30–32), to examine the expression of cKit in CNCCPs(8, 10, 11, 31–33). IF against cKit illustrated its colocalization with

tdTomato in various NC-derived tissues of E12.5 Wnt1-Cre;RC::tdTomato embryos, including the NT and the heart, (Fig. S3). How-ever, compared with the cKit−/tdTomato+ cells, cKit+/tdTomato+

population exhibited a weak expression of tdTomato (Fig. S3).We therefore performed NC lineage-restricted genetic fate-

mapping of cKit. We generated a novel mouse carrying tworecombinase systems (Cre-loxP and Flp-FRT), which enablesintersectional genetic fate-mapping of cKitCreERT2/+ in the Wnt1-expressing CNC lineage and its derivatives (34). Two previouslyestablished dual-recombinase responsive indicator alleles, theRC::Fela (35) and RC::Frepe (36), and a novel Wnt1-Flpe recom-binase driver line, the Wnt1::Flpe4351, were used (Fig. 2A).Timed-pregnant cKitCreERT2;Wnt1::Flpe;RC::Fela or cKitCreERT2;

Wnt1::Flpe;RC::Frepe mice were administered TAM during E8.5–E11.5 and embryo analysis was performed at selected time pointsbetween E10.5 and E18.5 (Table S1). Wnt1::Flpe-mediated re-combination resulted in extensive labeling with the flp indicators inthe craniofacial region, melanoblasts, gut, DRG, OFT, as well as incells within the epicardium and myocardium (Fig. 2 and Fig. S4).When both Wnt1::Flpe and cKitCreERT2/+ were expressed, Flpe in-dicators persisted in all NC-derived tissues, including the DRG,skin, heart, and OFT (Fig. 2 and Fig. S4 A–E). However, expres-sion of intersectional indicators was also detected in the craniofa-cial region, melanocytes, DRG, OFT, as well as in cells within theepicardium and myocardium (Fig. 2 and Fig. S4). Importantly,expression of intersectional indicators in the lung was not docu-mented (Fig. 2N), consistent with the hypothesis that the heart andlung cKit+ progenitors are of different origins. These studies il-lustrate a lineal relationship between the cKit+ CPs and Wnt1+

CNCs (CNCkit).

CNCkit Derivatives in the Heart. Expression of Cre-reporters wasdetected in all expected cardiac NC derivatives (37), includingthe OFT (Figs. 1J and 3 A and B, and Figs. S2B and S5), the

Fig. 1. cKitCreERT2/+ lineage-tracing. (A) Phenotypeof cKitCreERT2/+ mice. (B) Summary of the experi-mental design. (C–F) administration of TAM duringE7.5–E8.5 (n = 10) marks testicular (C, arrowheads),pulmonary (D) and, rarely, immature cells in themyocardium (E and F, arrowheads). (G–J) Live tissueimaging of cKitCreERT2/+ (G), IRG (H), and cKitCreERT2;IRG (I) E18.5 littermates subjected to TAM duringE9.5–E11.5 (n = 7). Widespread EGFP epifluorescencein ventricles and atria (I and J), lungs (J), OFT(J, arrow), epicardium (J, arrowheads). (K–N) Lineage-tracing in cKitCreERT2;R26RlacZ mice (n = 8). (O) Sum-mary of cKit genetic fate-mapping. Panels F and J areconfocal tile-scans. Panels K–N are photomerged im-age tiles. (Scale bars, 10 μm in D and F; 200 μm in J;500px in K–N.) (Magnification, 100× in C, E, and G–I.)

13052 | www.pnas.org/cgi/doi/10.1073/pnas.1517201112 Hatzistergos et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0

tunica media of the aortic arch (Fig. 3C), cardiac and aorticvalves (Fig. 3 D and E), atria (Fig. 1K), inflow tract, satellite glialprogenitors, and sensory cells (Fig. S5 A and B and Movie S1).Consistent with their CNC origin (38), CNCkit contributed toendothelium and smooth muscle layers of the OFT (Fig. 3 A andB), although coronary vascular cell differentiation was not ob-served (Fig. 3F).In agreement with previous reports in zebrafish (8, 11) and

mice (10, 21, 31–33), our analysis with the Wnt1-Cre (Fig. S6 andMovie S2) and cKitCreERT2/+ alleles suggests that CNCs contrib-ute to the myocardial lineage. Particularly, we documented con-tribution of CNCkit to atrial and ventricular cardiomyocytes(29.9% ± 3.1% of total EGFP+ derivatives) (Fig. 3 K and L), andpericardial, endocardial, and epicardial cells (Figs. 1 I, J, and N,and 3 G–K, and Movie S3). The majority of CNCkit-derivedcardiomyocytes was localized in the interventricular septum(Figs. 1J and 3 K and L), which, unlike the left and right ven-tricular myocardium, is partly derived from mesoderm posterior1 homolog nonexpressing (Mesp1−) CPs of undefined origin (9).

CNCkit Identity. To better characterize the identity of CNCKit, westudied the expression of micropthalmia-associated transcriptionfactor (Mitf), a direct target and transactivator of cKit signaling,expressed not only in cranial NC derivatives and mast cells butalso in cardiomyocytes (39). IF analysis demonstrated that Mitf isalso expressed in CNCkit and their cardiomyocytic derivatives(Fig. S7 A–C). However, EGFP+ cells in the heart did not ex-press the melanocyte-specific markers tyrosinase or trp1, sug-gesting that Mitf+ CNCkit derivatives in the heart are notmelanocytes (Fig. S7C).Next, we investigated the expression of Isl1, a homeobox

transcription factor that specifies the majority of the mammalianCP lineages, including CNCs (36). Accordingly, cKitCreERT2;IRG

mice were crossed with mice carrying an Isl1 nuclear lacZ(Isl1nLacZ) allele (40). When pregnant mice were administeredTAM from E9.5–E11.5, (Table S1), colocalization of EGFP andnLacZ was documented in cells of the NT, DRGs, and the OFT(Fig. S7 D–J) in E12.5 embryos.

Transient BMP Antagonism Induces Cardiomyogenesis in CNCkit. Todetermine the full cardiomyogenic capacity of cKit+ CPCs(Fig. 1), we established iPSCs from cKitCreERT2;IRG mice(iPSCkit). We induced cardiomyogenesis in iPSCs with eitherascorbic acid (AA), which drives cardiogenesis partly via an in-termediate differentiation stage into cKit+/Nkx2.5+ CPs (15, 41),or BMP antagonism, a signaling pathway that regulates the de-velopment of both mesodermal and NC lineages (Fig. 4A) (2, 5,31, 42–46).Both AA or BMP antagonists enhanced cardiac differentiation

into spontaneously contracting embryoid bodies (EBs) comparedwith controls (Fig. 4B) (P = 0.0073). When EBs were treatedwith 4-OH TAM during differentiation (Fig. 4A), we detectedthat 79.05% ± 4.9% of the beating EBs generated via transientBMP antagonism were EGFP+, compared with 39.03% ± 7.7%and 43.14% ± 6.9% EGFP+ beating EBs following treatmentwith AA or vehicle, respectively (Fig. 4 B–E and Movie S4) (P <0.0001), suggesting that endogenous BMP signaling may limitcardiomyogenesis from cKit+ CPs.Compared with control, AA, and Noggin (NOG)-treated iPSCkit,

Dorsomorphin (Dorso) enhanced cardiomyogenesis by signifi-cantly repressing Brachyury transcription (Fig. 4F and Fig. S8),while up-regulating ISL1, NKX2-5 (Fig. 4F) (43). In addition, wedetected a significant up-regulation in CNC-related genes, in-cluding PAX3 and WNT1 (Fig. 4G and Fig. S8 E–H), whereas theexpression of the proepicardial genesWT1, TCF21, and TBX18 (2),as well as the coronary endothelium marker KDR (Fig. 4G and

Fig. 2. Intersectional genetic fate-mapping of cKitand Wnt1. (A) schematic of the two different ap-proaches of the study. (B–D) Live epifluorescenceimaging (B), followed by salmon-gal histochemicaldetection of nLacZ (C and D), in an E10.5 cKitCreERT2;RC::Fela embryo exposed to TAM during E8.5–E9.5.The flp- and intersectional indicators are not expressedin the absence of Flpe and Flpe/Cre-mediated recom-bination, respectively. (E–H) A Wnt1::Flpe;cKitCreERT2;RC::Fela littermate exhibits widespread GFP epifluor-escence (E) and a few salmon-gal+ cells in the NT andheart. (I–L) Live embryo imaging of mCherry and GFPepifluorescence in a E17.5 Wnt1::Flpe;cKitCreERT2;RC::Frepe embryo. EGFP+ CNCkit in the craniofacial region(I), skin (J), OFT (K), and the epicardial wall of the heart(L). (M and T) X-gal+ CNCkit derivatives in the OFT(M and G), heart (N, O, R, and S), and epicardium(P and T) of E17.5 Wnt1::Flpe;cKitCreERT2;RC::Felaembryos. Arrows in M–P are depicted in highermagnification in panels Q–T, respectively. Panels B, Eare photomerged image tiles. OFT, outflow tract;Ht, heart; Lu, lung. (Scale bars, 50 μm in M–P and300px in I–L.) (Magnification, 100× in B–H and Q–T.)

Hatzistergos et al. PNAS | October 20, 2015 | vol. 112 | no. 42 | 13053

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0

Fig. S8), were significantly down-regulated compared with controls.Finally, expression of cKit increased significantly over time (Fig.S8D), although the level of expression was similar between thedifferent treatment groups (Fig. 4G). As previously shown (15, 41),IF analysis confirmed that myocardial specification of cKit+ CPscommenced via coexpression of NKX2.5 (Fig. 5 and Fig. S9 A–D).Remarkably, BMP antagonism enhanced the development ofEGFP+/NKX2.5+ progenitors by ∼sevenfold (Fig. 5, Fig. S9 A–D,and Movie S5) (P = 0.0057). Moreover, iPSCkit-derived CNCkit

gave rise to all CNC derivatives, including EGFP+ smooth musclecells (Fig. S9 E and F), Isl1+ (Fig. S9 G and H), and Pax3+ CPs(Fig. S9 I and J), while innervating the beating EBs with neuro-filament-M+ and Tuj1+ neurons (Fig. S9 K–N).

DiscussionThe major findings are that cKit marks CPs of CNC origin, whichenter the embryonic mouse heart at ∼E9.5 and contribute arelatively small proportion of myocardium and other derivativesof the CNC, but not coronary vascular cells. In addition, we showthat CNCkit CPs with cardiomyocyte differentiation capacity can bederived in vitro from mouse iPSCs following transient antagonismof the BMP pathway, which drives the stage-specific differentiationof iPSCs toward the cardiac mesodermal and CNC lineages.Our findings confirm previous developmental studies in mice

showing that, during gastrulation, cKit is expressed in extraem-bryonic mesoderm and embryonic ectoderm, but not mesodermalCPs (4, 13, 14, 42). Furthermore, the findings are in agreementwith previous reports supporting the existence of cKit+ CPs, whichdo not contribute to coronary endothelium (15), as well as withrecent endothelial lineage fate-mapping analyses suggesting thatthe coronary endothelium is unlikely to originate from cKit+ cells(6, 29).Notably, although much controversy exists over the contribu-

tion of CNCs to the myocardium (8, 10, 11, 30–33), our studieswith the cKitCreERT2/+, Wnt1-Cre, and Wnt1::Flpe alleles stronglysupport the hypothesis that the mammalian CNC holds full

cardiomyogenic capacity, which our findings now suggest isundermined by developmental changes in the activity of BMP andWnt pathways preceding their invasion in the heart (5, 31, 45). It isalso noteworthy that a pool of multipotent postmigratory NCprogenitors (47, 48), some of which express cKit (49), has beenrecorded in other tissues; hence, it would be interesting to ex-amine their relationship to CNCkit.Our study differs from a recent cardiac genetic fate-map of

cKit, using different cKit alleles (21). First, in contrast to findingspresented here and elsewhere (6, 15, 29), van Berlo et al. (21)reported that cKit CPs contribute extensively to coronary en-dothelium. However, it is noteworthy that mutations in themouse W/cKit locus have not been associated with tangible car-diovascular defects (12), as would be likely if cKit+ CPs com-prised a major source of coronary vascular cells. Second, vanBerlo et al. (21) concluded that the minimal cardiomyocytecontribution of cKit CPs reflects minimal differentiation capac-ity. However, although our study agrees that the in vivo car-diomyocyte contribution of cKit+ CPs is lower than expectedfrom previous reports (50, 51), we show that this is not a result ofminimal differentiation capacity, but rather, because of theirdevelopmental origin in the CNC, which comprises a minorcontributor of cardiomyocytes to the mammalian heart. Impor-tantly, using iPSC modeling we demonstrate that differentiationof CNCkit to cardiomyocytes requires the BMP signaling path-way, which also directs differentiation of mesodermal CPs to themyocardium. This finding suggests that, although CNCkit holdfull cardiomyocyte differentiation capacity, their in vivo contri-bution is repressed by spatiotemporal changes in BMP activity,which render the cardiac milieu nonconducive for cardiomyocytedifferentiation during CNC invasion to the heart (45).Our findings have several important implications. First, they

resolve the current controversy over the existence and car-diomyogenic capacity of cKit+ CPs (22). We show that cKit+ CPsinvested within the developing heart are fully capable of pro-ducing new cardiomyocytes, both in vivo and in vitro. Therefore,

Fig. 3. Heart derivatives of CNCkit. (A and B)Colocalization of X-gal with SM1 (arrowhead) andPecam1 (arrows) in the OFT. (C) X-gal+ cells (arrow-head) within the aortic tunica media. (D) EGFP+

derivatives within the mitral valve (arrowhead) andaortic valve (arrow). (E) EGFP+ derivatives in theaortic valve (arrowheads). (F) CNCkit are associatedwith, but do not contribute to, coronary vasculature.(G) EGFP+/cmlc2v+ ventricular cardiac myocytes.(H) EGFP+ cardiac derivatives coexpress Gata4+. (I and J)EGFP+ pericardium (arrows), epicardium (arrowheads),and endocardium (J, yellow arrow). (K) Cardiomyocyticversus noncardiomyocytic EGFP+ derivatives in theheart. (L) Distribution of EGFP+ cells in the heart [n = 3embryos; 11 sections (K and L)]. AoV, aortic valve; CM,cardiomyocytes; cmlc2v, cardiac muscle light chain 2v;LA, left atrium; LV, left ventricle; LCA, left coronaryartery; MV, mitral valve; SM1, smooth muscle myosinheavy chain. Values represent means ± SEM.

13054 | www.pnas.org/cgi/doi/10.1073/pnas.1517201112 Hatzistergos et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0

coupled with the findings from many laboratories that cKit+ CPsare present in the postnatal heart, they represent an importanttherapeutic target for heart regeneration (17–19, 52, 53). Forexample, the activity of BMP in the damaged myocardium couldbe modulated pharmacologically, or via transplantation of cellscapable of regulating BMP activity, to support production ofmyocardium from endogenous or exogenously supplied cKit+

CPs (18, 52).Second, the findings advance our understanding of the cellular

and molecular mechanisms underlying mammalian cardiomyo-genesis, by illustrating a previously unknown relationship be-tween the spatiotemporal modulation of the BMP pathway andthe generation of myocardium from mesodermal and CNC CPs.Third, the generation of CNCs from iPSCs provides a unique

opportunity to study and understand the biology and function ofCNCs, as well as to test their regenerative capacity in novel cell-based therapeutic strategies. Finally, considering the technical limi-tations often associated with conditional gene-targeting approaches,our findings do not exclude the possibility that, in addition to CNCkit,the adult heart contains other cKit+ cells with full cardiovasculardifferentiation capacity, as those reported by others (21, 28, 50, 51),which may have remained undetectable with our reagents.In conclusion, our findings support the hypothesis that the

mammalian heart is invested with a cKit+ CP lineage, with fullcapacity to generate cardiomyocytes in vivo and in vitro, andtherefore provide an important therapeutic target for the pre-vention and treatment of heart disease. Modulation of the activity

Fig. 4. Transient BMP antagonism in iPSCkit inducesCNCkit and suppresses the epicardium. (A) Schematic ofthe experimental approach. (B) Quantification of thepercentage of beating EBs, and the percentage ofbeating EBs that are EGFP+ following Cre-recombina-tion. (C) Live fluorescent imaging of a vehicle-treatedspontaneously beating EB, coexpressing EGFP andDsRed. (D and E) Confocal microscopy of EBs followingtreatment with AA (D) or Dorso (E), illustrates that theEGFP+ cKitCreERT2/+ derivatives within the EBs arecTnnT+ cardiomyocytes. (F) Gene-expression analysis ofBrachyury during the time-course of iPSCkit differenti-ation into cardiomyocytes following treatment withvehicle, Dorso, or NOG. (G) Comparison of the expres-sion profiles of cardiac mesoderm- and CNC-relatedgenes in day 11 EBs, in response to treatment withvehicle, AA, or Dorso. Compared with controls, AA andDorso enhance cardiomyogenesis via a significant in-duction in CNC-related genes while suppressing pro-epicardial and endothelial progenitor genes. In addition,Dorso significantly enhances the expression of ISL1and NKX2.5. cTnnt, cardiac troponin T; values repre-sent means ± SEM.

Fig. 5. Derivation of CNCkit from mouse iPSCs. (A–C) Representative confocal immunofluorescence images illustrating EGFP+/NKX2.5+ derivatives withincTnnT+ EBs of vehicle-treated (A), AA-treated (B), or Dorso-treated (C) iPSCkit. Note that several of the EGFP+ cells in the vehicle- and AA-treated groups areNKX2.5− (asterisks). (Insets) Higher magnification. (D) Quantitation of EGFP+/NKX2.5+ cells between groups (n = 9 per group). Values represent means ± SEM.

Hatzistergos et al. PNAS | October 20, 2015 | vol. 112 | no. 42 | 13055

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0

of the BMP pathway in the heart may enhance the therapeuticregeneration of damaged myocardium from CNCkit.

Materials and MethodsAll animals were maintained in an Association for Assessment and Accreditationof Laboratory Animal Care-approved animal facility at the University of Miami,Miller School of Medicine, and procedures were performed using InstitutionalAnimal Care and Use Committee-approved protocols according to NIH stan-dards. cKitCreERT2/+ mice were developed as previously described (24). TheWnt1-Cre, RC::tdTomato, IRG, and R26RLacZ mouse lines were purchased from JacksonLaboratories. The Isl1nLacZ mice have been described elsewhere (40).Wnt1::Flpe,

RC::Fela and RC::Frepe mice were developed as previously described (35, 36).iPSCkit were generated from adult cKitCreERT2/IRG tail-tip fibroblasts. Geno-typing, TAM injections, gene-expression analysis, lineage-tracing, andhistological analysis of mouse embryos was performed as previouslydescribed (24). See SI Materials and Methods for more detailed discussion.

ACKNOWLEDGMENTS. We thank Dr. Sylvia Evans for providing the Isl1nLacZ

mice. This study was funded by National Institutes of Health GrantsR01 HL107110, R01 HL094849, R01 HL110737, R01 HL084275, and 5UMHL113460 (to J.M.H.); grants from the Starr foundation and the Soffer FamilyFoundation (to J.M.H.); and Deutsche Forschungsgemienschaft Grant SA1374/1-3 (to D.S.).

1. Rana MS, Christoffels VM, Moorman AF (2013) A molecular and genetic outline ofcardiac morphogenesis. Acta Physiol (Oxf) 207(4):588–615.

2. Witty AD, et al. (2014) Generation of the epicardial lineage from human pluripotentstem cells. Nat Biotechnol 32(10):1026–1035.

3. Burridge PW, Keller G, Gold JD, Wu JC (2012) Production of de novo cardiomyocytes:Human pluripotent stem cell differentiation and direct reprogramming. Cell Stem Cell10(1):16–28.

4. Yang L, et al. (2008) Human cardiovascular progenitor cells develop from a KDR+embryonic-stem-cell-derived population. Nature 453(7194):524–528.

5. Jain R, et al. (2015) HEART DEVELOPMENT. Integration of Bmp and Wnt signaling byHopx specifies commitment of cardiomyoblasts. Science 348(6242):aaa6071.

6. Fioret BA, Heimfeld JD, Paik DT, Hatzopoulos AK (2014) Endothelial cells contribute togeneration of adult ventricular myocytes during cardiac homeostasis. Cell Reports8(1):229–241.

7. Van Handel B, et al. (2012) Scl represses cardiomyogenesis in prospective hemogenicendothelium and endocardium. Cell 150(3):590–605.

8. Li YX, et al. (2003) Cardiac neural crest in zebrafish embryos contributes to myocardialcell lineage and early heart function. Dev Dyn 226(3):540–550.

9. Kitajima S, Miyagawa-Tomita S, Inoue T, Kanno J, Saga Y (2006) Mesp1-nonexpressingcells contribute to the ventricular cardiac conduction system. Dev Dyn 235(2):395–402.

10. Tomita Y, et al. (2005) Cardiac neural crest cells contribute to the dormant multi-potent stem cell in the mammalian heart. J Cell Biol 170(7):1135–1146.

11. Sato M, Yost HJ (2003) Cardiac neural crest contributes to cardiomyogenesis in ze-brafish. Dev Biol 257(1):127–139.

12. Reith AD, et al. (1990) W mutant mice with mild or severe developmental defectscontain distinct point mutations in the kinase domain of the c-kit receptor. Genes Dev4(3):390–400.

13. Orr-Urtreger A, et al. (1990) Developmental expression of c-kit, a proto-oncogeneencoded by the W locus. Development 109(4):911–923.

14. Kataoka H, et al. (1997) Expressions of PDGF receptor alpha, c-Kit and Flk1 genesclustering in mouse chromosome 5 define distinct subsets of nascent mesodermalcells. Dev Growth Differ 39(6):729–740.

15. Wu SM, et al. (2006) Developmental origin of a bipotential myocardial and smoothmuscle cell precursor in the mammalian heart. Cell 127(6):1137–1150.

16. Beltrami AP, et al. (2003) Adult cardiac stem cells are multipotent and supportmyocardial regeneration. Cell 114(6):763–776.

17. Kanashiro-Takeuchi RM, et al. (2012) Activation of growth hormone releasing hor-mone (GHRH) receptor stimulates cardiac reverse remodeling after myocardial in-farction (MI). Proc Natl Acad Sci USA 109(2):559–563.

18. Hatzistergos KE, et al. (2010) Bone marrow mesenchymal stem cells stimulate cardiacstem cell proliferation and differentiation. Circ Res 107(7):913–922.

19. Florea V, et al. (2014) Agonists of growth hormone-releasing hormone stimulate self-renewal of cardiac stem cells and promote their survival. Proc Natl Acad Sci USA111(48):17260–17265.

20. Chugh AR, et al. (2012) Administration of cardiac stem cells in patients with ischemiccardiomyopathy: The SCIPIO trial: Surgical aspects and interim analysis of myocardialfunction and viability by magnetic resonance. Circulation 126(11, Suppl 1):S54–S64.

21. van Berlo JH, et al. (2014) c-kit+ cells minimally contribute cardiomyocytes to theheart. Nature 509(7500):337–341.

22. Mummery CL, Lee RT (2013) Is heart regeneration on the right track? Nat Med 19(4):412–413.

23. Heger K, et al. (2014) CreER(T2) expression from within the c-Kit gene locus allowsefficient inducible gene targeting in and ablation of mast cells. Eur J Immunol 44(1):296–306.

24. Klein S, et al. (2013) Interstitial cells of Cajal integrate excitatory and inhibitoryneurotransmission with intestinal slow-wave activity. Nat Commun 4:1630.

25. Schönhuber N, et al. (2014) A next-generation dual-recombinase system for time- andhost-specific targeting of pancreatic cancer. Nat Med 20(11):1340–1347.

26. Wilson YM, Richards KL, Ford-Perriss ML, Panthier JJ, Murphy M (2004) Neural crestcell lineage segregation in the mouse neural tube. Development 131(24):6153–6162.

27. Goldstein BJ, et al. (2014) Adult c-Kit(+) progenitor cells are necessary for mainte-nance and regeneration of olfactory neurons. J Comp Neurol 523(1):15–31.

28. Tallini YN, et al. (2009) c-kit expression identifies cardiovascular precursors in theneonatal heart. Proc Natl Acad Sci USA 106(6):1808–1813.

29. Ubil E, et al. (2014) Mesenchymal-endothelial transition contributes to cardiac neo-vascularization. Nature 514(7524):585–590.

30. Jiang X, Rowitch DH, Soriano P, McMahon AP, Sucov HM (2000) Fate of the mam-malian cardiac neural crest. Development 127(8):1607–1616.

31. Stottmann RW, Choi M, Mishina Y, Meyers EN, Klingensmith J (2004) BMP receptor IAis required in mammalian neural crest cells for development of the cardiac outflowtract and ventricular myocardium. Development 131(9):2205–2218.

32. Brown CB, et al. (2001) PlexinA2 and semaphorin signaling during cardiac neural crestdevelopment. Development 128(16):3071–3080.

33. Tamura Y, et al. (2011) Neural crest-derived stem cells migrate and differentiate intocardiomyocytes after myocardial infarction. Arterioscler Thromb Vasc Biol 31(3):582–589.

34. Dymecki SM, Ray RS, Kim JC (2010) Mapping cell fate and function using recombinase-based intersectional strategies. Methods Enzymol 477:183–213.

35. Jensen P, et al. (2008) Redefining the serotonergic system by genetic lineage. NatNeurosci 11(4):417–419.

36. Engleka KA, et al. (2012) Islet1 derivatives in the heart are of both neural crest andsecond heart field origin. Circ Res 110(7):922–926.

37. Hutson MR, Kirby ML (2007) Model systems for the study of heart development anddisease. Cardiac neural crest and conotruncal malformations. Semin Cell Dev Biol18(1):101–110.

38. Waldo KL, Kumiski DH, Kirby ML (1994) Association of the cardiac neural crest withdevelopment of the coronary arteries in the chick embryo. Anat Rec 239(3):315–331.

39. Tshori S, et al. (2006) Transcription factor MITF regulates cardiac growth and hyper-trophy. J Clin Invest 116(10):2673–2681.

40. Sun Y, et al. (2007) Islet 1 is expressed in distinct cardiovascular lineages, includingpacemaker and coronary vascular cells. Dev Biol 304(1):286–296.

41. Christoforou N, et al. (2008) Mouse ES cell-derived cardiac precursor cells are multipotentand facilitate identification of novel cardiac genes. J Clin Invest 118(3):894–903.

42. Kattman SJ, et al. (2011) Stage-specific optimization of activin/nodal and BMP sig-naling promotes cardiac differentiation of mouse and human pluripotent stem celllines. Cell Stem Cell 8(2):228–240.

43. Hao J, et al. (2008) Dorsomorphin, a selective small molecule inhibitor of BMP sig-naling, promotes cardiomyogenesis in embryonic stem cells. PLoS One 3(8):e2904.

44. Choi M, Stottmann RW, Yang YP, Meyers EN, Klingensmith J (2007) The bone mor-phogenetic protein antagonist Noggin regulates mammalian cardiac morphogenesis.Circ Res 100(2):220–228.

45. Yuasa S, et al. (2005) Transient inhibition of BMP signaling by Noggin induces car-diomyocyte differentiation of mouse embryonic stem cells. Nat Biotechnol 23(5):607–611.

46. Chambers SM, et al. (2009) Highly efficient neural conversion of human ES and iPScells by dual inhibition of SMAD signaling. Nat Biotechnol 27(3):275–280.

47. Baggiolini A, et al. (2015) Premigratory and migratory neural crest cells are multi-potent in vivo. Cell Stem Cell 16(3):314–322.

48. Buitrago-Delgado E, Nordin K, Rao A, Geary L, LaBonne C (2015) NEURODEVELOPMENT.Shared regulatory programs suggest retention of blastula-stage potential in neural crestcells. Science 348(6241):1332–1335.

49. Motohashi T, Kitagawa D, Watanabe N, Wakaoka T, Kunisada T (2014) Neural crest-derived cells sustain their multipotency even after entry into their target tissues. DevDyn 243(3):368–380.

50. Ellison GM, et al. (2013) Adult c-kit(pos) cardiac stem cells are necessary and sufficientfor functional cardiac regeneration and repair. Cell 154(4):827–842.

51. Ferreira-Martins J, et al. (2012) Cardiomyogenesis in the developing heart is regulatedby c-kit-positive cardiac stem cells. Circ Res 110(5):701–715.

52. Williams AR, et al. (2013) Enhanced effect of combining human cardiac stem cells andbone marrow mesenchymal stem cells to reduce infarct size and to restore cardiacfunction after myocardial infarction. Circulation 127(2):213–223.

53. Karantalis V, et al. (2015) Synergistic effects of combined cell therapy for chronic is-chemic cardiomyopathy. J Am Coll Cardiol, in press.

54. Echelard Y, Vassileva G, McMahon AP (1994) Cis-acting regulatory sequences governingWnt-1 expression in the developing mouse CNS. Development 120(8):2213–2224.

55. Sundararajan S, Wakamiya M, Behringer RR, Rivera-Pérez JA (2012) A fast and sen-sitive alternative for β-galactosidase detection in mouse embryos. Development139(23):4484–4490.

13056 | www.pnas.org/cgi/doi/10.1073/pnas.1517201112 Hatzistergos et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 4,

202

0