Bone marrow cells adopt the cardiomyogenicfate in vivoMarcello Rota*, Jan Kajstura*, Toru Hosoda*, Claudia Bearzi*, Serena Vitale*, Grazia Esposito*, Grazia Iaffaldano*,M. Elena Padin-Iruegas*, Arantxa Gonzalez*, Roberto Rizzi*, Narissa Small*, John Muraski†, Roberto Alvarez†,Xiongwen Chen‡, Konrad Urbanek*, Roberto Bolli§, Steven R. Houser‡, Annarosa Leri*, Mark A. Sussman†,and Piero Anversa*¶

*Cardiovascular Research Institute, Department of Medicine, New York Medical College, Valhalla, NY 10595; ‡Cardiovascular Research Center,Temple University, Philadelphia, PA 19140; §Institute of Molecular Cardiology, University of Louisville, Louisville, KY 40292; and†Heart Institute and Department of Biology, San Diego State University, San Diego, CA 92182

Edited by Andrew R. Marks, Columbia University College of Physicians and Surgeons, New York, NY, and approved September 7, 2007 (received for reviewJuly 9, 2007)

The possibility that adult bone marrow cells (BMCs) retain a remark-able degree of developmental plasticity and acquire the cardiomyo-cyte lineage after infarction has been challenged, and the notion ofBMC transdifferentiation has been questioned. The center of thecontroversy is the lack of unequivocal evidence in favor of myocardialregeneration by the injection of BMCs in the infarcted heart. Becauseof the interest in cell-based therapy for heart failure, several ap-proaches including gene reporter assay, genetic tagging, cell geno-typing, PCR-based detection of donor genes, and direct immunoflu-orescence with quantum dots were used to prove or disprove BMCtransdifferentiation. Our results indicate that BMCs engraft, survive,and grow within the spared myocardium after infarction by formingjunctional complexes with resident myocytes. BMCs and myocytesexpress at their interface connexin 43 and N-cadherin, and thisinteraction may be critical for BMCs to adopt the cardiomyogenic fate.With time, a large number of myocytes and coronary vessels aregenerated. Myocytes show a diploid DNA content and carry, at most,two sex chromosomes. Old and new myocytes show synchronicity incalcium transients, providing strong evidence in favor of the func-tional coupling of these two cell populations. Thus, BMCs transdif-ferentiate and acquire the cardiomyogenic and vascular phenotypesrestoring the infarcted heart. Together, our studies reveal that locallydelivered BMCs generate de novo myocardium composed of inte-grated cardiomyocytes and coronary vessels. This process occursindependently of cell fusion and ameliorates structurally and func-tionally the outcome of the heart after infarction.

myocardial infarction � myocardial regeneration � stem cells �transdifferentiation

To date, the hematopoietic stem cell appears to be the mostversatile stem cell in crossing lineage boundaries and the most

prone to break the law of tissue fidelity (1). Early studies onc-kit-positive bone marrow cell (BMC) differentiation into myo-cardium have generated great enthusiasm (2, 3), but other obser-vations have rejected the initial results (4–6) and promoted a waveof skepticism about the therapeutic potential of BMCs for theinjured heart. The major criticisms include: (i) lack of utilization ofgenetic markers for the recognition of donor BMCs and theirprogeny; (ii) inaccurate interpretation of the original data due toautofluorescence artifacts; and (iii) the possibility that myocyteregeneration is mediated by fusion of BMCs with resident myocytesrather than BMC transdifferentiation. To address these importantquestions and demonstrate reproducibility of results, four labora-tories with complementary expertise have undertaken a series ofjoined experiments to acquire information on the plasticity ofBMCs and their therapeutic potential for the infarcted heart.

In this effort, BMCs for myocardial regeneration were obtainedfrom three transgenic mice. In the first, EGFP was driven by theubiquitous �-actin promoter; in the second, EGFP was driven by thecardiac-specific �-myosin-heavy-chain (�-MHC) promoter; and in

the third, a c-myc-tagged nuclear-targeted-Akt transgene wasdriven by the �-MHC-promoter (7). With the first category of BMC(�-actin-EGFP), all cardiac cells formed by BMC differentiationwere expected to express EGFP; with the second category of BMCs(�-MHC-EGFP), only myocytes formed by BMC differentiationwere expected to express EGFP; and with the third category ofBMCs (�-MHC-c-myc-tagged-nuc-Akt), only myocytes formed byBMC differentiation were expected to express c-myc in their nuclei.In all cases, male BMCs were injected in wild-type female infarctedmice so that cell genotyping would allow the distinction betweenresident female cardiac cells and newly generated male cardiac cells.These strategies allowed us to determine the destiny of BMCswithin the recipient heart by genetic tagging with EGFP, cell fatetracking with �-MHC-EGFP and �-MHC-c-myc, and cell genotyp-ing by sex-chromosome identification.

ResultsEngraftment of BMCs. A premise of these studies was the use of amethodology in which immunolabeling of proteins was obtained inthe absence of autofluorescence inherent in tissue sections offormalin-fixed myocardium. This objective was achieved by imple-menting a technology by which primary antibodies are directlylabeled by quantum dots (QD). This procedure eliminates the needfor secondary antibody and avoids the interference of autofluores-cence in the specificity of the reaction [see supporting information(SI) Fig. 6].

BMC differentiation and acquisition of the cardiogenic fateinvolves the engraftment of the donor cells within the host myo-cardium. Within 12–48 h after infarction and cell implantation,donor BMCs were seeded within the myocardium of the borderzone and were, in part, integrated structurally with resident cells (SIFig. 7). Junctional and adhesion complexes were detected betweenBMCs and between BMCs and adjacent myocytes and fibroblasts(Fig. 1 A and B). Myocytes and fibroblasts function as supportingcells within the cardiac stem cell niches (8), suggesting that BMCsform niche-like structures within the recipient myocardium (9).

Author contributions: M.R., J.K., T.H., R.B., S.R.H., A.L., M.A.S., and P.A. designed research;M.R., J.K., T.H., C.B., S.V., G.E., G.I., M.E.P.-I., A.G., R.R., N.S., J.M., R.A., X.C., K.U., R.B., S.R.H.,A.L., and M.A.S. performed research; J.M., R.A., and M.A.S. contributed new reagents/analytic tools; M.R., J.K., T.H., C.B., S.V., G.E., G.I., M.E.P.-I., A.G., R.R., N.S., J.M., X.C., K.U.,R.B., S.R.H., A.L., M.A.S., and P.A. analyzed data; and M.R., J.K., R.B., S.R.H., A.L., M.A.S., andP.A. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Abbreviations: BMC, bone marrow cell; EC, endothelial cell; SMC, smooth muscle cell.

¶To whom correspondence should be addressed at: Cardiovascular Research Institute,Department of Medicine, Vosburgh Pavilion, Room 302, New York Medical College,Valhalla, NY 10595. E-mail: piero�anversa@nymc.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0706406104/DC1.

© 2007 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0706406104 PNAS � November 6, 2007 � vol. 104 � no. 45 � 17783–17788

MED

ICA

LSC

IEN

CES

A large proportion of engrafted BMCs was cycling as docu-mented by BrdU labeling and expression of Ki67. Phospho-H3 wasalso detected in dividing BMCs (Fig. 1 C–E). Conversely, in theabsence of engraftment, BMCs died by apoptosis (Fig. 1 F and G).Apoptosis was restricted to BMCs that failed to express connexin43 and N-cadherin. At 48 h after cell implantation, an average 63 �6% BrdU labeling of engrafted BMCs was measured. For eachBMC category, cell proliferation increased progressively from 12 to24–36 and 48 h, whereas cell apoptosis was gradually attenuated (SIFig. 7). In each case, �25% or 14,000 of the 60,000 injected BMCswere present in the viable tissue of the border zone at 2 days (SIFig. 7).

To confirm the ability of BMCs to home to the myocardium,these cells were injected intravenously shortly after coronary arteryligation, and their localization in the heart was examined 1–2 (n �7) and 4 (n � 4) days later. In all cases, clusters of BMCs were foundin the infarct border zone where a fraction of these cells acquiredthe myocyte lineage. These cells were integrated within the recip-ient myocardium because connexin 43 was expressed at the inter-face between new and old myocytes. BMC-derived myocytes car-ried the Y chromosome. Additionally, single and doublets of BMCswere scattered in the viable myocardium adjacent to the infarct (SI

Fig. 7). Thus, BMCs engraft, survive, and grow within the myo-cardium by forming junctional complexes among them and withresident myocytes and fibroblasts.

BMCs Acquire the Cardiogenic Fate. The high level of proliferation inengrafted BMCs distinguished these cell clusters from residentcardiac niches that are composed predominantly of quiescent cells(8). At 12 h, BMCs were mostly CD45-positive, but at 24–36 h, alarge subset of BMCs was CD45-negative, and the absence of CD45was even more evident at 48 h (Fig. 2 A–D). The presence of Ychromosome and/or EGFP was used to distinguish resident frominjected CD45-positive cells. The expression of connexin 43 ac-companied the phenotypic conversion of BMCs into cardiac cells(Fig. 2C). Markers of myocytes, endothelial cells (ECs), and smoothmuscle cells (SMCs) were detected in the engrafted BMCs (Fig. 2E–J), suggesting that the myocardial microenvironment changedthe fate of BMCs. Nonengrafted CD45-positive BMCs were hardlydetectable at 5 days, and the few CD45-positive cells were ofrecipient origin. Thus, BMCs home to the myocardium where theyrapidly lose the hematopoietic phenotype and acquire the cardio-myocyte lineage.

Gap Junctions and BMC Destiny. Translocation of calcium frommyocytes to BMCs via gap junctions may have profound effects ontheir acquisition of the myocyte lineage, growth, and differentia-tion. Cell coupling was analyzed in vitro by two-photon microscopyafter loading BMCs with the red fluorescent dye DiI, whichintegrates stably in the cell membrane. Labeled BMCs (red) wereplated with rat neonatal myocytes, which were loaded with thegreen fluorescent dye calcein that translocates to neighboring cellsthrough the generation of gap junctions (10). The appearance ofgreen fluorescence in BMCs indicated the transfer of calcein frommyocytes via junctional complexes (Fig. 3 A–C). For real-time dyetransfer, cascade blue, which also cannot diffuse spontaneously, wasinjected in BMCs or myocytes, and the passage of the dye from onecell to the other was recorded. This event occurred in �3 min (Fig.3 D–N). When BMCs were loaded with rhodamine-labeled dextran,this high-molecular-mass dye (70,000 Da) did not transfer tomyocytes (Fig. 3O).

With time, BMCs differentiated into myocytes that contractedspontaneously or after stimulation at 1 Hz (Fig. 3 P–U). Addition-ally, connexin 43 was identified at the interface between EGFP-negative neonatal myocytes and EGFP-positive-myocytes derivedfrom BMCs of �-MHC-EGFP mice (Fig. 3 V–X). A diploid DNAcontent was measured in new myocytes excluding fusion events.Thus, BMCs and myocytes become electrically coupled, and thismay be critical for BMCs to adopt the cardiomyogenic fate.

BMC Differentiation. Three classes of male BMCs were used toinduce myocardial regeneration in female infarcted mice (SI Fig. 8).One cell category was EGFP-positive, and the other two wereEGFP-negative (SI Fig. 9). Importantly, the presence of the Ychromosome distinguished all bone marrow-derived male cardiaccells from resident female cells (SI Fig. 9). Engraftment of BMCsshortly after their delivery was followed by myocardial regeneration(Fig. 4 A–F and SI Figs. 10 and 11), which expanded from 5 to 10and 30 days after infarction. The regenerated myocardium had twoconsequences on cardiac remodeling; it attenuated the inflamma-tory response acutely–subacutely and prevented largely scar for-mation (SI Fig. 12).

At the three time points, regenerated myocytes, ECs, and SMCsexpressed the different transgenes in the anticipated fashion. BMCsobtained from �-actin-EGFP mice formed a myocardium in whichcardiomyocytes, ECs, and SMCs were mostly EGFP-positive (Fig.4C and SI Fig. 11). When the expression of EGFP or c-myc tag wasregulated by the �-MHC-promoter, the transgenes were detectedexclusively in myocytes (Fig. 4 A, B, E, and F and SI Fig. 10).However, the recognition of the Y chromosome allowed us to

Fig. 1. BMCs engraft and divide. (A and B) Clusters of BMCs within the recipientmyocardium after infarction. BMCs are c-kit-positive (green) and carry the Ychromosome (Y-chr, white dots). Connexin 43 (A, Cnx-43, yellow dots) andN-cadherin (B, N-cadh, yellow dots) are present between male BMCs (arrows) andbetween male BMCs and female myocytes (�-sarcomeric-actin, �-SA, red; arrows)and fibroblasts (procollagen, col, magenta; arrows). (C–E) BrdU (C, yellow), Ki67(D, yellow), and phospho-H3 (E, yellow) are detected in dividing BMCs (c-kit,green; (Y-chr, red dots). Inset in E shows metaphase chromosomes. (F and G)Apoptosis (TdT, magenta) of BMCs (c-kit, F, green; G, white; Y-chr, white dots).EGFP (G, green). Cnx-43 (G, yellow) is absent in apoptotic BMCs.

17784 � www.pnas.org�cgi�doi�10.1073�pnas.0706406104 Rota et al.

document that vascular structures were the result of BMC differ-entiation (Fig. 4 E and F).

The number of myocytes formed by BMCs exceeded the numberof myocytes lost with infarction (SI Fig. 13). Although the postnatalmaturation of the myocardium in rodents is completed at 1 month,the myocytes derived from BMCs failed to reach the adult pheno-type. The proportion of binucleated myocytes increased at 1 month(SI Fig. 13) but was lower than that present in the adult myocar-dium. Similarly, the number of capillaries was markedly lower thanin the adult heart, but arteriolar density exceeded the values of thedeveloping and mature myocardium (SI Fig. 13). Thus, adult BMCsdifferentiate into the cardiac lineages and repair, in part, theinfarcted heart.

BMC Differentiation and Cell Fusion. To determine the presence ofcell fusion, DNA content per nucleus and the number of sexchromosomes were measured in newly formed myocytes, SMCs,and ECs. Whereas sex chromosomes were evaluated in the progenyof the three classes of BMCs, DNA content per nucleus wasdetermined in myocytes, ECs, and SMCs formed by BMCs from�-actin-EGFP mice; the localization of EGFP allowed the recog-nition of all BMC-derived cardiac cells.

EGFP-positive-mononucleated and binucleated myocytes andEGFP-positive ECs and SMCs had 2n DNA content per nucleus (SIFig. 14). Moreover, regenerated myocytes, ECs, and SMCs showedat most one Y and one X chromosome, documenting the malegenotype of the restored myocardium (SI Fig. 14). Conversely, themyocytes of the border zone possessed, at most, two X chromo-somes, demonstrating that these cells retained the female genotypeand did not participate in fusion events (SI Fig. 14). Fusion betweenadult myocytes, 25,000 �m3 in volume or larger, and BMCs wouldresult in the generation of a myocyte progeny of nearly the samesize, which was not the case. The new myocytes had volumes thatvaried from 100 to 2,500 �m3. Thus, c-kit-positive-BMCs arecapable of restoring infarcted myocardium independently from cellfusion.

Molecular Documentation of BMC Differentiation. To strengthen themorphological results, the presence of the reporter genes EGFP

and c-myc was demonstrated by PCR of genomic DNA extractedfrom infarcted hearts. DNA sequences for EGFP and c-myc tagwere detected in all treated animals at 5, 10, and 30 days afterinfarction and BMC implantation. DNA bands had the expectedmolecular mass and sequence (Fig. 5A and SI Fig. 15). RT-PCR wasused to identify transcripts for EGFP and c-myc tag. Again, EGFPand c-myc tag mRNAs were found in all samples containing newlyformed myocardium. RT-PCR products had the anticipated mo-lecular mass and sequence (Fig. 5B and SI Fig. 15). Importantly,protein levels for EGFP and c-myc-tag were identified by Westernblotting (Fig. 5C).

The detection of transcripts and proteins for EGFP andc-myc-tag in infarcted hearts treated with BMCs from mice inwhich these transgenes were driven by the �-MHC promoterprovided strong evidence in favor of BMC transdifferentiationand active cardiomyogenesis. The recognition of EGFPmRNA and protein in infarcted hearts injected with BMCsfrom mice in which EGFP was under the control of the�-actin-promoter confirmed active engraftment of the donorcells and was consistent with the histological results. Addi-tionally, EGFP-positive and c-myc-positive myocytes wereisolated from the infarcted region of treated hearts (SI Fig. 16).Collectively, these data document that BMCs retain a degreeof developmental plasticity and acquire the myogenic and vas-cular phenotypes, restoring, in part, the infarcted myocardium.

Physiological Documentation of BMC Differentiation. An importantquestion was whether formed myocytes were functionallycompetent electrically and mechanically. Regenerated myo-cytes within the infarct and surviving myocytes away from theinfarct were isolated and their physiological properties deter-mined. Shortly after injection, cells positive for c-kit, CD45,and EGFP were identified, and these cells did not showelectrical properties of developing myocytes or contracted inresponse to electrical stimulation (Fig. 5D). At 15–30 days,small myocytes were identified, and these BMC-derived myo-cytes exhibited electrical characteristics similar to sparedmyocytes but showed a prolongation of the action potentialand enhanced cell shortening (Fig. 5 E–H).

Fig. 2. BMCs acquire the cardiogenicfate. (A and B) BMCs are mostly CD45-positive at 12 h (white) and mostly CD45negative at 48 h. (C) At 48 h, BMCs (c-kit,green; Y-chr:, white dots) express Cnx-43(yellow) and are CD45 negative. (D) Val-ues are mean � SD. *, P � 0.05 vs. 12 h; **,P � 0.05 vs. 24 –36 h. (E and F) BMCs (EGFP,green; Y-chr, white dots) express Nkx2.5(magenta, arrowheads) and Cnx-43 (yel-low). (G–J) BMCs (G, EGFP, green) showvon Willebrand factor (H, vWF, white, ar-rows) and �-SA (I, red, arrowheads). Jshows a merge.

Rota et al. PNAS � November 6, 2007 � vol. 104 � no. 45 � 17785

MED

ICA

LSC

IEN

CES

A remaining issue concerned whether the new myocytes wereintegrated structurally and functionally with resident myocytes invivo and participated in ventricular performance. Connexin 43 wasdetected between the viable and repaired myocardium. Impor-tantly, connexin 43 was shared by preexisting and new myocytes,documenting the structural integration between resident femalemyocytes and new male myocytes (SI Fig. 16).

The functional integration of regenerated EGFP-positive-myocytes with the surrounding myocardium was documented by anex vivo preparation and two-photon microscopy in treated hearts at1 month. The heart was perfused retrogradely through the aorta

with an oxygenated Tyrode solution containing the calcium indi-cator Rhod-2. The heart was then stimulated at 1 Hz, and calciumtransient was clearly present in EGFP-positive BMC-derived myo-cytes and EGFP-negative mouse myocytes. The synchronicity incalcium transients between these two myocyte populations pro-vided strong evidence in favor of the functional coupling of old andnew myocytes (Fig. 5I).

The injection of BMCs and myocardial regeneration restored, inpart, the loss of contraction in the infarcted region of the wall,documented echocardiographically at 30 days (SI Fig. 17). Addi-tionally, cardiac repair reduced chamber volume and increased the

Fig. 3. BMCs interact with cardiomyocytes.(A–C) DiI-labeled BMCs (A, red) were cocul-tured with calcein-labeled myocytes (B,green). Calcein transferred to BMCs (C, yel-low-green, arrows). (D–O) EGFP-positiveBMCs (D and J, green) cocultured with myo-cytes (E and K, dotted lines) were injectedwith cascade blue (F and L, arrows). Cascadeblue transferred to myocytes adjacent toBMCs (H, I, and N). Rhodamine-labeled dex-tran (O, red) did not transfer from BMC to theadjacent myocyte. (P–U) Cocultured BMCs dif-ferentiated into myocytes that contractedspontaneously (R) or after stimulation (U). (P)Myocyte derived from a DiI-labeled BMC (Q,red). (S) BMC-derived mononucleated myo-cytes express connexin 43 (T, white dots). Neo-natal myocytes were labeled by DiI (T, red).(V–X) BMC from �-MHC-EGFP mouse differ-entiated into a myocyte that expressed EGFP(V, green), �-SA (W, red), and Cnx-43 (X,white). The new myocyte possesses 2n DNAcontent. (W) Values of DNA content.

Fig. 4. BMCsandmyocardial regeneration. (AandB)BMCsfrom�-MHC-EGFP(A)and�-MHC-c-myc-tagged-nuc-Akt (B)miceregeneratedmyocardial infarcts. Formedmyocytes (arrowheads) express �-SA (red), EGFP (A, green) and c-myc (B, green). (C and D) Infarcts treated with BMCs from �-actin-EGFP mice. EP, epicardium; EN,endocardium. (C) Left, Center, and Right show EGFP (green), regenerated myocytes (MHC, red), and their merge. Arrows, nonregenerated infarct. (D) Y-chr localizationacross the infarct. Left, Center, and Right illustrate regenerated myocytes (MHC, red), distribution of Y-chr (white dots), and their merge. Arrows, nonregeneratedinfarct. (E and F) Arterioles formed by BMCs from �-MHC-c-myc-tagged-nuc-Akt mice. SMCs (�-smooth muscle actin, �-SMA, red) and ECs (vWF, yellow) carry the Y-chr(white dots). Vessel wall is c-myc-negative; myocytes (Lower) are c-myc-positive (green), carry the Y-chr, and are �-SA-positive (magenta).

17786 � www.pnas.org�cgi�doi�10.1073�pnas.0706406104 Rota et al.

wall thickness-to-chamber radius ratio (SI Fig. 17). Hemodynam-ically, myocardial regeneration attenuated the increase in LVEDP,improved LVDP and positive and negative dP/dt and decreaseddiastolic wall stress (SI Fig. 17). Thus, BMCs adopt the cardiom-yogenic fate structurally, electrically, and mechanically, improvingthe function of the damaged heart.

Paracrine Effects of BMCs. The possibility that BMCs activategrowth and differentiation of resident progenitor cells or induceangiogenesis through the recruitment of circulating BMCs hasbeen suggested (3, 11, 12). Therefore, we measured the forma-

tion of myocytes and vessels in the border zone. Because BrdUwas given to the animals throughout, newly formed structuresgenerated by endogenous mechanisms had to be female in origin,positive for BrdU, and negative for EGFP and c-myc. In all cases,the extent of myocyte and vessel formation was comparable withthat in untreated infarcted mice at 30 days (SI Fig. 18). More-over, the degree of cell replication determined by Ki67 andMCM5 at sacrifice was comparable in cell-treated and untreatedmice (SI Fig. 18). Although our results do not support aparacrine effect of BMCs, factors released in the infarcted regioncould have contributed significantly to the mechanisms of car-diac repair. Our data do not exclude this likely possibility.

Fig. 5. Myocyte differentiation and functional competence. (A) DNA sequences of EGFP and c-myc-tag by PCR. DNA from the tail of donor transgenic(TG) and wild-type (WT) mice was used as positive and negative control. (B) Transcripts for EGFP and c-myc-tag by real-time RT-PCR in infarcted treatedhearts (�). Samples in the absence of RT reaction (�). RNA from hearts of TG and WT was used as positive and negative control. (C) EGFP and c-myc-tagprotein by Western blotting. Protein lysates from hearts of TG were used as positive control, and protein lysates from untreated infarcted hearts were usedas negative control. (D) At 2–3 days, EGFP-positive cells lacked electrical activity. (E) Electrical properties of BMC-derived and spared myocytes. (F) At 30days, newly formed EGFP-positive myocytes were electrically excitable. (G) Spared myocytes had depressed fractional shortening. Values are mean � SD.

*, P � 0.05 vs. new myocytes. (H) EGFP-myocytes derived from BMC differentiation used for the evaluation of cell mechanics. Cell volume is indicated (�m3).(I) Mouse heart at 30 days after coronary artery ligation and implantation of BMCs from �-MHC-EGFP mouse. Calcium transient was detected inEGFP-positive-myocytes and EGFP-negative-myocytes.

Rota et al. PNAS � November 6, 2007 � vol. 104 � no. 45 � 17787

MED

ICA

LSC

IEN

CES

DiscussionThe results of the current study indicate that a category of adult cellsfrom the bone marrow shares a degree of developmental plasticitycommonly seen in embryonic stem cells; c-kit-positive BMCs en-graft in proximity to the infarcted myocardium and differentiateinto cells of the cardiogenic lineage, forming functionally compe-tent cardiomyocytes and vascular structures. The regenerated myo-cardium positively interferes with the development of the myopathyafter infarction, attenuating the dramatic changes in ventricular sizeand shape and the progressive deterioration in cardiac function thatoccur chronically after infarction (13). Our data strongly suggestthat adult BMCs implanted in the infarcted heart lose the hema-topoietic fate and integrate with the host myocardium by estab-lishing a microenvironment necessary for the engrafted cells toadopt the cardiac destiny and form de novo myocardium. Theseobservations are important because small double-blind multicenterclinical trials in which mononuclear BMCs have been administeredto patients with acute and chronic ischemic heart failure haverecently been completed (14, 15). Despite positive results, themechanism by which mononuclear BMCs improve the outcome ofacute myocardial infarction and chronic ischemic cardiomyopathyin humans remains unclear. Currently, effort is being made toinitiate large clinical trials, although uncertainties persist about theactual effects of BMCs on the decompensated heart. Our resultssuggest that myocardial regeneration is a likely possibility.

The early studies on BMC transdifferentiation and cardiac repair(2, 3, 16, 17) have been followed by numerous negative reports thathave rejected the notion of stem cell plasticity and the relevance ofBMCs for the management of the human disease (4–6). Thetherapeutic potential of BMCs was questioned at multiple levels,which included the lack of implementation of genetic strategies forthe recognition of the destiny of BMCs in the infarcted hearttogether with the misinterpretation of immunocytochemical results(18, 19). These limitations have been overcome here by theutilization of three categories of BMCs with distinct genotypiccharacteristics and the use of a methodology that excludes autofluo-rescence artifacts. An important variable that has frequently beenneglected experimentally and may explain contrasting results con-cerns the preparation, viability, and modality of injection of BMCs(20). Recently, the clinical application of improperly collectedBMCs (21) has resulted in the absence of benefit in patients withacute myocardial infarction (22). Conversely, carefully acquired,stored, and administered BMCs not only improve ventricularfunction chronically after infarction but increase survival in patientswith heart failure after infarction (23).

Importantly, fusion of BMCs with resident cardiac cells has beenproposed as the only mechanism by which cells of bone marroworigin can form myocytes and promote tissue repair (4–6). Thedocumentation that BMCs fuse with resident cardiomyocytes istechnically demanding. This involves the identity of the fused cell,the functional competence of the progeny, the recognition of thefusion partner, and the demonstration that the converted cell is ahybrid (24). The fate of BMCs after their injection in the infarctedheart has been characterized with the Cre-Lox genetic system.Unfortunately, the Cre-Lox model is not perfect. The unmodifiedCre-recombinase present in progenitor cells can cross the mem-brane of the recipient cell (25), mimicking fusion events. Also, thisenzyme can be translocated via nanotubules (26) from BMCscarrying the Cre-recombinase to the resident cells, giving the falseimpression of cell fusion. Therefore, we have excluded cell fusion bydocumenting a diploid DNA content together with one set only ofsex chromosomes in nuclei of newly formed myocytes, SMCs, andECs. The regenerated myocytes are functionally competent andsignificantly smaller than the host cells, further excluding fusionevents. In this regard, no viable partner cells are available for fusionin transmural myocardial infarcts. Fusion history may be concealedby reductive mitosis (27) in which the fused cells would expel partof their chromosomal DNA and convert tetraploid cells into diploidcells, masking the formation of heterokaryons. However, there is nodemonstration of this unusual cell behavior either in vitro or in vivo.

The possibility that BMC therapy of the infarcted heart exerts itsbeneficial effects by activation of resident progenitor cells has beensuggested (2, 3, 11, 12). According to this hypothesis, BMCs maycontribute indirectly to cardiac regeneration by releasing a varietyof peptides that exert a paracrine action on the myocardium and itsresident cells. The most convincing example in favor of a paracrineeffect of BMCs was shown recently (3). The current work does notexclude this indirect role of BMCs in cardiac repair. Collectively,these findings point to the notion that BMCs adopt the cardiacphenotype and potentiate the growth reserve of the adult heart,emphasizing the therapeutic import of BMCs for heart failure inhumans.

Materials and MethodsMyocardial infarction was induced in wild-type mice, andBMCs were injected in the region bordering the infarct.Myocardial regeneration was evaluated by immunocytochem-istry, molecular biology, and FISH analysis. The newly formedmyocytes were characterized electrically and mechanically.For additional information, see SI Materials and Methods.

1. Tosh D, Slack JM (2002) Nat Rev Mol Cell Biol 3:187–194.2. Orlic D, Kajstura J, Chimenti S, Jakoniuk I, Anderson SM, Li B, Pickel J, McKay R,

Nadal-Ginard B, Bodine DM, et al. (2001) Nature 410:701–705.3. Yoon YS, Wecker A, Heyd L, Park JS, Tkebuchava T, Kusano K, Hanley A, Scadova H, Qin

G, Cha DH, et al. (2005) J Clin Invest 115:326–338.4. Murry CE, Soonpaa MH, Reinecke H, Nakajima H, Nakajima HO, Rubart M, Pasumarthi

KB, Virag JI, Bartelmez SH, Poppa V, et al. (2004) Nature 428:664–668.5. Balsam LB, Wagers AJ, Christensen JL, Kofidis T, Weissman IL, Robbins RC (2004) Nature

428:668–673.6. Nygren JM, Jovinge S, Breitbach M, Sawen P, Roll W, Hescheler J, Taneera J, Fleischmann

BK, Jacobsen SE (2004) Nat Med 10:494–501.7. Tsujita Y, Muraski J, Shiraishi I, Kato T, Kajstura J, Anversa P, Sussman MA (2006) Proc

Natl Acad Sci USA 103:11946–11951.8. Urbanek K, Cesselli D, Rota M, Nascimbene A, De Angelis A, Hosoda T, Bearzi C, Boni

A, Bolli R, Kajstura J, et al. (2006) Proc Natl Acad Sci USA 103:9226–9231.9. Scadden DT (2006) Nature 441:1075–1079.

10. Cancelas JA, Koevoet WL, de Koning AE, Mayen AE, Rombouts EJ, Ploemacher RE(2000) Blood 96:498–505.

11. Mangi AA, Noiseux N, Kong D, He H, Rezvani M, Ingwall JS, Dzau VJ (2003) Nat Med9:1195–1201.

12. Fazel S, Cimini M, Chen L, Li S, Angoulvant D, Fedak P, Verma S, Weisel RD, KeatingA, Li RK (2006) J Clin Invest 116:1865–1877.

13. Opie LH, Commerford PJ, Gersh BJ, Pfeffer MA (2006) Lancet 367:356–367.

14. Schachinger V, Erbs S, Elsasser A, Haberbosch W, Hambrecht R, Holschermann H, Yu J,Corti R, Mathey DG, Hamm CW, et al. (2006) N Engl J Med 355:1210–1221.

15. Assmus B, Honold J, Schachinger V, Britten MB, Fischer-Rasokat U, Lehmann R, TeupeC, Pistorius K, Martin H, Abolmaali ND, et al. (2006) N Engl J Med 355:1222–1232.

16. Iwasaki H, Kawamoto A, Ishikawa M, Oyamada A, Nakamori S, Nishimura H, SadamotoK, Horii M, Matsumoto T, Murasawa S, et al. (2006) Circulation 113:1275–1277.

17. Kawamoto A, Iwasaki H, Kusano K, Murayama T, Oyamada A, Silver M, Hulbert C, GavinM, Hanley A, Ma H, et al. (2006) Circulation 114:2163–2169.

18. Jackson KA, Snyder DS, Goodell MA (2004) Stem Cells 22:180–187.19. Laflamme MA, Murry CE (2005) Nat Biotechnol 23:845–856.20. Anversa P, Leri A, Rota M, Hosoda T, Bearzi C, Urbanek K, Kajstura J, Bolli R (2007) Stem

Cells 25:589–601.21. Seeger FH, Tonn T, Krzossok N, Zeiher AM, Dimmeler S (2007) Eur Heart J 28:766–772.22. Lunde K, Solheim S, Aakhus S, Arnesen H, Abdelnoor M, Egeland T, Endresen K, Ilebekk

A, Mangschau A, Fjeld JG, et al. (2006) N Engl J Med 355:1199–1209.23. Assmus B, Fischer-Rasokat U, Honold J, Seeger FH, Fichtlscherer S, Tonn T, Seifried E,

Schachinger V, Dimmeler S, Zeiher AM (2007) Circ Res 100:1234–1241.24. Leri A, Kajstura J, Anversa P (2005) Physiol Rev 85:1373–1416.25. Will E, Klump H, Heffner N, Schwieger M, Schiedlmeier B, Ostertag W, Baum C, Stocking

C (2002) Nucleic Acids Res 30:e59.26. Koyanagi M, Brandes RP, Haendeler J, Zeiher AM, Dimmeler S (2005) Circ Res

96:1039–1041.27. Medvinsky A, Smith A (2003) Nature 422:823–825.

17788 � www.pnas.org�cgi�doi�10.1073�pnas.0706406104 Rota et al.