Cells Expressing Early Cardiac Markers Reside in the Bone Marrow and Are Mobilized Into the Peripheral Blood After Myocardial Infarction

Abstract

The concept that bone marrow (BM)– derived cells participate in cardiac regeneration remains highly controversial and the identity of the specific cell type(s) involved remains unknown. In this study, we report that the postnatal BM contains a mobile pool of cells that express early cardiac lineage markers (Nkx2.5/Csx, GATA-4, and MEF2C). These cells are present in significant amounts in BM harvested from young mice but their abundance decreases with age; in addition, the responsiveness of these cells to gradients of motomorphogens SDF-1, HGF, and LIF changes with age. FACS analysis, combined with analysis of early cardiac markers at the mRNA and protein levels, revealed that cells expressing these markers reside in the nonadherent, nonhematopoietic CXCR4⁺/Sca-1⁺/lin⁻/CD45⁻ mononuclear cell (MNC) fraction in mice and in the CXCR4⁺/CD34⁺/AC133⁺/CD45⁻ BMMNC fraction in humans. These cells are mobilized into the peripheral blood after myocardial infarction and chemoattracted to the infarcted myocardium in an SDF-1-CXCR4 −, HGF-c-Met−, and LIF-LIF-R− dependent manner. To our knowledge, this is the first demonstration that the postnatal BM harbors a nonhematopoietic population of cells that express markers for cardiac differentiation. We propose that these potential cardiac progenitors may account for the myocardial regenerative effects of BM. The present findings provide a novel paradigm that could reconcile current controversies and a rationale for investigating the use of BM-derived cardiac progenitors for myocardial regeneration.

Several recent reports suggest that bone marrow (BM)–derived hematopoietic stem cells (HSCs),^1,2 mesenchymal stem cells (MSCs),^3,4 and mononuclear cells (MNCs)^5,6 exhibit a high degree of differentiation plasticity and thus could potentially be used to regenerate infarcted myocardium in humans. These reports, however, have been challenged vigorously and the concept of transdifferentiation of adult HSCs has recently been called into question.^7,8 Furthermore, the identity of the specific BM-derived cellular subset(s) capable of regenerating cardiac tissue remains elusive.

These uncertainties are part of an ongoing intense debate regarding the mechanism responsible for the functional benefits observed after cellular transplantation. Two basic possibilities have been proposed: cell fusion^9,10 and transdifferentiation. In this report, we propose a third possibility, which may reconcile the apparently discordant results obtained by the proponents of cell fusion and transdifferentiation. We have previously demonstrated that the adult murine BM harbors CXCR4⁺ tissue–committed stem cells (TCSCs) for various tissues, including skeletal muscle, liver, and neural tissue^11–13 and have proposed that these TCSCs contribute to the organ/tissue chimerism observed after transplantation of BM- or mobilized peripheral blood (PB)–derived cells in earlier reports. In this study, we provide evidence, both in mice and in humans, that the postnatal BM harbors a population of CXCR4⁺/lin⁻/CD45⁻ nonhematopoietic MNCs that express early markers for cardiac differentiation. These cells are mobilized into the PB following myocardial infarction (MI) and are capable of migrating toward SDF-1, HGF, and LIF gradients. We propose that these cells may potentially be expanded in vitro and used for therapeutic myocardial regeneration.

Materials and Methods

The animal studies were approved by the Animal Care and Use Committee of the University of Louisville. The studies of human BM were approved by the human ethics committee of the Jagiellonian University, Cracow, Poland.

Murine and Human Bone Marrow Cells

Murine MNCs were isolated from the BM of 4- to 6-week-old Balb/C mice. Human MNCs were isolated from four cadaveric BM donors (52 to 65 years old) and, if necessary, depleted of adherent cells and T lymphocytes (A⁻T⁻ MNC).^14,15 Details are provided in the online data supplement available at http://circres.ahajournals.org.

FACS Isolation

Human Cells

CXCR4⁺/CD45⁺, CXCR4⁺/CD45⁻, CXCR4⁻/CD45⁺, and CXCR4⁻/CD45⁻ BMMNCs were isolated as specified in the online data supplement.

Murine Cells

Lin⁻/Sca-1⁺/CD45⁻ and Lin⁻/Sca-1⁺/CD45⁻ cells were isolated from the suspension of murine BMMNCs as specified in the online data supplement.

Hematopoietic Assays

For cell proliferation assays, murine or human sorted BMMNCs were plated in serum-free methylcellulose cultures.^14,15 For the CFU-spleen (CFU-S) assays, colonies were counted on the surface of the spleen using a magnifying glass.¹⁴ Specific conditions and methodology are provided in the online data supplement.

Chemotactic Isolation

The method for the chemotactic isolation has been described.^11,12 Briefly, cells were resuspended in serum-free medium and equilibrated for 10 minutes at 37°C. The lower chambers of Costar Transwell 24-well plates were filled with 650 μL of serum-free medium and 0.5% BSA containing SDF-1 (200 ng/mL), HGF (10 ng/mL), or LIF (50 ng/mL) or with medium alone (control). Details are provided in the online data supplement. Cells from the lower chambers were collected and their numbers counted by flow cytometry.^14,15

Real-Time RT-PCR

Total mRNA was isolated from cells and mRNA was reverse-transcribed with TaqMan Reverse Transcription Reagents. Quantitative assessment of Nkx2.5/Csx, GATA-4, MEF2C, VE-cadherin, and β-actin mRNA levels was performed by real-time RT-PCR as detailed in the online data supplement. The relative quantitation value of target, normalized to the endogenous control β-actin (house-keeping) gene and relative to a calibrator, is expressed as 2^−ΔΔCt (-fold difference), where ΔCt=(Ct of target genes)−(Ct of endogenous control gene, β-actin), and ΔΔCt=(ΔCt of samples for target gene)−(ΔCt of calibrator for the target gene).

Studies of Myocardial Infarction

Two groups (n=24/group) of wild-type mice (B6,129 strain, 8 to 12 weeks old) were used. The experimental preparation has been described in detail.¹⁶ Group I underwent a 30-minute coronary occlusion followed by reperfusion, whereas group II underwent a sham operation (1-hour open-chest state).¹⁶ Mice (n=6/group at each time-point) were euthanized at 6, 24, 48, or 96 hours after reperfusion, and blood samples as well as myocardial tissue samples from the ischemic and nonischemic regions were harvested.

Immunocytochemistry

The expression of GATA-4 and Nkx2.5/Csx was quantitated by confocal microscopy following immunocytochemistry as detailed in online data supplement.

Statistical Analysis

Data are reported as mean±SD. All statistical analyses were performed using the Instat 1.14 software (GraphPad). Data were analyzed using the Student t test for unpaired samples. Statistical significance was defined as P<0.01.

Results

Early Cardiac Markers in Chemoattracted BM-Derived Cells

RT-PCR analysis was performed on murine BMMNCs attracted to an SDF-1, HGF/SF, or LIF gradient. Figure 1A demonstrates that chemoattracted cells express a significantly higher level of mRNA for Nkx2.5/Csx and GATA-4. These observations at the mRNA level were confirmed by immunocytochemical detection of Nkx2.5/Csx and GATA-4 in the chemoattracted cells (Figure 1B and 1C). The response to SDF-1 of BMMNCs expressing Nkx2.5/Csx and GATA-4 was striking (Figure 1A, 1B, and 1C). However, the response of these cells isolated from 1-month-old mice to HGF and LIF was relatively weak. Next, we performed a systematic analysis at five different ages to determine whether age-related differences, as observed for HSCs,^17,18 exist in the abundance/responsiveness of this mobile population of BM cells. We isolated BMMNCs from 2-week- and 1-, 2-, 6-, or 12-month-old mice, subjected them to SDF-1, HGF, or LIF gradients, and evaluated the expression of early cardiac markers using real-time RT-PCR. Figure 2 shows that mRNA for the cardiac markers Nkx2.5/Csx and GATA-4 was easily detectable in chemoattracted cells from mice older than 2 weeks. Cells expressing early cardiac markers from 1-month-old mice strongly responded to an SDF-1 gradient, which corresponds with the time of their rapid growth in body mass. The responsiveness of these cells to SDF-1 was markedly diminished after 1 month (Figure 2, top panel). However, with increasing age, BM-derived cells expressing mRNA for Nkx2.5/Csx and GATA-4 were increasingly attracted to HGF gradients with a peak in 6-month-old mice followed by a decline at 1 year of age (Figure 2, middle panel). Although BMMNCs expressing Nkx2.5/Csx and GATA-4 mRNA were not appreciably attracted to LIF for the first 2 months of life, their responsiveness increased markedly at 6 months and then declined at 1 year (Figure 2, bottom panel).

Figure 1.

A, Detection of cardiac markers in BMMNCs that migrate to SDF-1, HGF, and LIF gradients. BM cells from 1-month-old mice were isolated from the lower transwell-chambers after chemo-taxis to SDF-1, HGF, and LIF (chemotactic isolation), and the expression of mRNA for cardiac markers (Nkx2.5/Csx and GATA-4) was compared by real-time RT-PCR between the same number of cells in the input (unpurified BM cells) and in the lower chamber. Data represent the average of three independent experiments (the BM of 10 mice was pooled for each experiment). Data are mean±SD. *P<0.00001 vs input cells. B, Immunocytochemical detection of protein expression for cardiac markers Nkx2.5/Csx and GATA-4 in input (unpurified) and SDF-1, HGF, and LIF chemoattracted BMMNCs from 1-month-old mice (the BM of 15 mice was pooled). BMMNCs migrating to an SDF-1 gradient exhibited greater abundance of cells positive for the cardiac markers. *P<0.05 vs input cells. C, Confocal microscopic images of BMMNCs positive for cardiac markers Nkx2.5/Csx (B, F, and J; red fluorescence), GATA-4 (A, E, and I; green fluorescence), and both (D, H, and L) in input (unpurified) BMMNCs (A through D) and BMMNCs chemoattracted to SDF-1 (E through H). I and J, Higher magnification images of a nucleus positive for both markers. Nuclei are identified by DAPI (C, D, G, H, K, and L; blue fluorescence). Scale bar=20 μm (A through H) or 10 μm (I through L).

Figure 2.

Expression of mRNA for cardiac markers in murine BMMNCs isolated by chemotaxis to SDF-1, HGF, and LIF at different ages. BMMNCs chemoattracted to SDF-1 (top panel), HGF (middle panel), and LIF (bottom panel) gradients were isolated from the lower transwell-chambers and the expression of mRNA for cardiac markers (Nkx2.5/Csx and GATA-4) was compared by real-time RT-PCR between the same number of cells in the input and in the lower chamber. All of the assays were performed in triplicate (the BM of 12 mice was pooled at each time-point). Data are mean±SD. *P<0.01 vs input.

FACS Isolation of Cells Expressing Early Cardiac Markers

Because cells exhibiting early hematopoietic stem cell markers such as murine Sca-1 and human CD34 and AC133 antigens have been reported to contribute to tissue regeneration,^19–21 we sought to determine whether these cells correspond to the cells expressing early cardiac markers. We used real-time RT-PCR to compare the expression of mRNA for cardiac markers between different subsets of cells isolated from murine and human BM. We found that FACS-sorted (Figure 3A) murine Sca-1⁺/lin⁻/CD45⁻ cells were highly enriched (up to 1000×) in mRNA for early myocardial markers (Figure 3B). Immunohistochemical staining (Figure 3C) demonstrated that Sca-1⁺/lin⁻/CD45⁻ cells (top panel) were smaller than Sca-1⁺/lin⁻/CD45⁺ cells (bottom panel) and that ≈6% of these cells expressed GATA-4 protein. Importantly, in contrast to Sca-1⁺/lin⁻/CD45⁺ cells, these cells were not hematopoietic, because they neither grew hematopoietic colonies in in vitro cultures, nor formed day-11 CFU-S colonies in vivo in lethally irradiated littermates (data not shown). Thus, the expression of CD45 is helpful in distinguishing, within the Sca-1⁺/lin⁻ BMMNC population, HSCs (Sca-1⁺/lin⁻/CD45⁺) from potential cardiac progenitors (Sca-1⁺/lin⁻/CD45⁻).

Figure 3.

Expression of mRNA for cardiac and endothelial markers in murine Sca-1⁺/lin⁻/CD45⁻ cells (potential cardiac TCSCs). A, Dot-plot of murine BMMNCs (top left) and dot-plot of these cells from the lymphoid gate (R1) labeled for the expression of Sca-1 antigen and lineage negativity (top middle). Cells from the lymphoid gate that were Sca-1⁺/lin⁻ (R1, R2) were sorted by FACS for CD45 expression (top right). M1 shows CD45⁺ and M2 shows CD45⁻ population of Sca-1⁺/lin⁻ cells. Dot-plot of Sca-1⁺/lin⁻/CD45⁻ is shown in bottom left panel (R5) and the dot-plot of Sca-1⁺/lin⁻/CD45⁺ is shown in bottom right panel (R7). Four independent sorting experiments were performed (the BM of 12 mice was pooled for each sort). Representative sorts are shown. B, Real-time RT-PCR for the expression of mRNA for cardiac and endothelial markers. Markers were compared between the same number of sorted Sca-1⁺/lin⁻/CD45⁺ and Sca-1⁺/lin⁻/CD45⁻ cells. Data represent the average of four independent experiments (the BM of 12 mice was pooled for each experiment). Data are mean±SD. *P<0.00001 vs unpurified BMMNCs. C, Representative images of BM-derived GATA-4 −positive Sca-1⁺/lin⁻/CD45⁻ cell (a through e, top panels) and GATA-4 −negative Sca-1⁺/lin⁻/CD45⁺ cell (f through j, bottom panels) sorted by FACS. a and f, Phase-contrast image of cells; b and g, GATA-4 −positive (b, red fluorescence) and −negative (g) nuclei; c and h, Nuclei identified by DAPI (blue fluorescence); d and i, Overlay of b and c, and g and h, respectively; e and j, Overlay of the respective immunofluorescent and phase-contrast images. Scale bar=5 μm (all panels)

As demonstrated in Figure 4, purified CD34⁺ and AC133⁺ cells in human BMMNCs were highly enriched (up to 25×) with mRNA for cardiac markers (Nkx2.5/Csx, GATA-4, and MEF2C). As expected,²² these CD34⁺ and AC133⁺ cells were also enriched with mRNA for selected endothelial progenitor markers (VE-cadherin, VEGFR3, and vWF). Furthermore, mRNA for early cardiac markers was additionally more enriched in CD45⁻ fraction of purified human CXCR4⁺, CD34⁺, and AC133⁺ cells, and similar to what was observed in mice, appropriate in vitro clonogenic assays confirmed that these cells are nonhematopoietic (data not shown). Together, these data demonstrate that BMMNCs expressing cardiac markers are abundant within a population of nonhematopoietic CXCR4⁺/Sca-1⁺/lin⁻/CD45⁻ BMMNCs in mice and within a population of nonhematopoietic CXCR4⁺/CD34⁺/AC133⁺/CD45⁻ BMMNCs in humans.

Figure 4.

CD34⁺ and AC133⁺ human BMMNCs are highly enriched with mRNA for cardiac and endothelial markers. Human unpurified BMMNCs, adherent cell-depleted BMMNCs, and BMMNCs sorted by FACS for CD34− and AC133-positivity were compared for the expression of mRNA for early cardiac (Nkx2.5/Csx, GATA-4, and MEF2C) and endothelial (VE-cadherin, VEGFR3, and vWF) markers. mRNA was harvested separately from BMMNCs of 3 donors; for each donor, the assays were performed in triplicate. Data are mean±SD. *P<0.00001 vs control.

Release of Cells Expressing Early Cardiac Markers Into the PB After Acute MI

Real-time RT-PCR was used to detect the expression of mRNA for cardiac and endothelial markers in PBMNCs collected at specific time-points after MI. Figure 5 demonstrates a marked increase in mRNA content for cardiac (Nkx2.5/Csx, GATA-4, and MEF2C) and endothelial (VE-cadherin) markers in PBMNCs harvested after MI, which peaked at 48 hours after MI and essentially disappeared by 96 hours (Figure 5). Importantly, at 48 hours after MI, the expression of both GATA-4 and Nkx2.5/Csx was detected by immunocytochemistry in ≈0.1% of the PBMNCs. These cells were not detectable in sham-operated animals (data not shown). These results validate our RT-PCR data at the protein level. Together, these data indicate that cells expressing cardiac markers are released from the BM into the PB in response to myocardial ischemia/reperfusion injury in a manner analogous to what has been described for endothelial progenitor cells^23,24

Figure 5.

Changes in the expression of cardiac and endothelial markers in PB of mice after MI. PBMNCs were harvested from mice that underwent myocardial infarction or a sham operation (control) 6, 24, 48, and 96 hours earlier. Expression of mRNA for early cardiac (Nkx2.5/Csx, GATA-4, and MEF2C) and endothelial (VE-cadherin) markers in the same number cells was quantitated by real-time RT-PCR and compared between groups. All of the assays were performed in triplicate (PBMNCs from 8 to 10 mice per group were pooled at each time-point). Data are mean±SD. *P<0.01 vs control.

Expression of SDF-1, HGF, LIF, and VEGF After Myocardial Infarction

We evaluated the expression of mRNA for SDF-1, HGF, LIF, and VEGF in the LV anterior and posterior walls of mice at 6, 24, 48, and 96 hours after a reperfused MI or a sham operation, respectively. The mRNA for all of these motomorphogens was noted to be maximally upregulated at 48 hours after MI. As demonstrated in Figure 6, 48 hours after a 30-minute coronary occlusion/reperfusion, the expression of SDF-1, HGF, LIF, and VEGF was markedly upregulated in tissue homogenates from the infarcted (LV anterior wall) compared with control (LV posterior wall) myocardium.

Figure 6.

Myocardial expression of SDF-1, LIF, VEGF, and HGF at 6, 24, 48, and 96 hours after MI in mice. Expression of SDF-1, LIF, VEGF, and HGF mRNA in control (nonischemic zone, LV posterior wall) and infarcted (LV anterior wall) myocardium was quantitated by real-time RT-PCR. All of the assays were performed in triplicate (myocardial samples from 8 to 10 mice per group were examined at each time point). Data are mean±SD. *P<0.01 vs control.

BM Cells Expressing Early Cardiac Markers Are Chemoattracted to Supernatants From Infarcted Myocardium in an SDF-1−, HGF−, and LIF-Dependent Manner

Having found that the supernatant from infarcted myocardium is enriched with mRNA for chemoattractants, we examined whether this supernatant would indeed attract these cells. BMMNCs were allowed to migrate to supernatants from myocardial homogenates obtained from control (posterior wall) and infarcted (anterior wall) myocardial tissue harvested at different time-points after MI. Figure 7A demonstrates that cells expressing early cardiac markers were strongly chemoattracted to supernatants from infarcted myocardium harvested at 48 hours after MI. This chemotactic response could be inhibited by pretreatment of BMMNCs with T140, K-252a compound, or anti-gp190 Ab, known inhibitors of CXCR4, c-Met, and LIF receptors, respectively^11,25 (Figure 7B). When combined together, these inhibitors completely blocked the chemotaxis of BMMNCs to the supernatant of infarcted myocardium (Figure 7B). Furthermore, a striking upregulation of mRNA for Nkx2.5/Csx, GATA-4, MEF2C, and VE-cadherin was detected in BMMNCs attracted to supernatant from infarcted myocardium (Figure 7C), indicating that the chemoattracted cells express markers of cardiac (Nkx2.5/Csx⁺/MEF2C⁺) and endothelial (VE-cadherin⁺) progenitors. In contrast, these cells did not express markers for skeletal muscle (MyoD and Myf-5) or neural (nestin and GFAP) TCSCs (Figure 7C).

Figure 7.

A, BMMNCs are chemoattracted to the supernatant from infarcted myocardial tissue in an SDF-1−, HGF−, and LIF-dependent manner. BMMNCs were added to the upper transwell chambers and allowed to migrate to the conditioned media (CM, supernatants) from control (LV posterior wall) and infarcted (LV anterior wall) myocardial tissue. BM of 5 mice was pooled, and the chemotactic assays were performed in triplicate. Data are mean±SD. *P<0.00001 vs control. B, Chemotaxis of BMMNCs to the conditioned media from control and infarcted myocardial tissue in the absence or presence of inhibitors to CXCR4 (T140), c-Met (K252a), and LIF-R (anti-gp190 Ab). BM of 5 mice was pooled, and the chemotactic assays were performed in triplicate. Data are mean±SD. *P<0.00001 vs CM from the infarcted myocardium. DMSO was added as control (solvent for K252a). C, Expression of mRNA for cardiac (Nkx2.5/Csx, GATA-4, and MEF2C), endothelial (VE-cadherin), and skeletal muscle (Myf5, MyoD, and myogenin) markers in control (unpurified) BMMNCs and BMMNCs that migrated to supernatants from infarcted myocardium. BM of 5 mice was pooled, and the chemotactic assays were performed in triplicate. Data are mean±SD. *P<0.0001 vs control.

Discussion

Mounting evidence from numerous preclinical as well as clinical studies supports the notion that BM-derived cells are capable of inducing regeneration in various tissues/organs.^{1,2,4–6,26,27} Despite the demonstration of cardiac repair with Sca-1⁺/c-kit⁺/lin⁻ BM-derived cells,^1,26 the very ability of committed adult “hematopoietic” stem cells to transdifferentiate into cells of nonhematopoietic lineage remains highly controversial.^7,8,10,28 Resolution of this issue is of paramount importance not only from a conceptual standpoint, but also because large-scale regenerative therapies would necessitate precise identification and in vitro expansion of the specific BM cellular subset responsible for cardiac regeneration.

Salient Findings

The major findings of the present study can be summarized as follows: (1) MNCs expressing mRNA and protein for early cardiac markers are present in postnatal BM in significant amounts and their abundance declines with age; (2) the responsiveness of these cells to chemoattractants changes in an age-dependent fashion; (3) cells expressing early cardiac markers reside in the nonadherent, nonhematopoietic population of CXCR4⁺/Sca-1⁺/lin⁻/CD45⁻ BMMNCs in mice and in the population of CXCR4⁺/CD34⁺/AC133⁺/CD45⁻ BMMNCs in humans; (4) these cells are released into the peripheral circulation after MI; and (5) these cells are chemoattracted to homogenates of infarcted myocardium in vitro in an SDF-1-CXCR4 −, HGF-c-Met−, and LIF-LIF-R–dependent manner. To our knowledge, this is the first demonstration that the postnatal BM harbors a nonhematopoietic population of cells that express markers for cardiac differentiation. We propose that these potential cardiac progenitors may account for the myocardial regenerative effects of BM. The present observations have important conceptual implications for our understanding of BM-dependent myocardial regeneration and provide a rationale for further studies aimed at optimizing therapeutic cardiac regeneration by selective use of BM-derived cardiac progenitors.

Postnatal Murine and Human BM Harbors Cells Expressing Early Cardiac Markers

In mice, we found that cells expressing early myocardial markers reside within a population of CXCR4⁺/Sca-1⁺/lin⁻/CD45⁻ BMMNCs. The ability of these cells to migrate to HGF and LIF gradients indicates that in addition to CXCR4, these cells also express functional c-Met and LIF receptors. Similarly, in humans, we detected a population of cells expressing cardiac markers within the CXCR4⁺/CD34⁺/AC133⁺/CD45⁻ MNC population. Together, these results suggest that potential cardiac progenitors reside in postnatal BM and constitute a subset of cells that express conventional markers for HSCs (Sca-1 in mice and CD34 and AC133 in humans), an observation that may explain, at least in part, the myocardial regenerative capacity and the functional improvement observed after transplantation of BM- or mobilized PB-derived cells enriched with markers for HSCs (Sca-1⁺/c-kit⁺/lin⁻ cells).^1,26 Furthermore, because CD34 and AC133 in humans and Sca-1 in mice are also expressed in BM-derived endothelial progenitors,^29–31 our observations may explain the increased vascularization of infarcted myocardium after injection of these cells.^1,25 We also found that cells expressing early cardiac markers are enriched with CXCR4⁺/Sca-1⁺/CD45⁻ cells that neither grow hematopoietic colonies in vitro nor form CFU-S colonies in lethally irradiated littermates in vivo. In the aggregate, these data demonstrate, for the first time, the existence of cells expressing early cardiac markers in postnatal BM and identify them as a subpopulation of cells distinct from HSCs.

Based on phenotypical analysis, our cells resemble the population of cardiac progenitor cells that was recently isolated from the heart as a “side population” of cardiac cells.³² Similar to our cells, these cells expressed Sca-1 antigen, were CD45 negative, and were found to be distinct from HSCs.³² Interestingly, these cells could also be detected at a very low concentration in PB where 1 cell in 150 000 peripheral blood MNCs displays this phenotype. In this study, we observed a 150-fold enrichment in these cells (1 in 1000 cells) in PB 48 hours after MI. This suggests that potential cardiac progenitors may, as we postulate here, “shuttle” between the BM and the heart and that their number increases during mobilization caused by MI or during pharmacological mobilization by G-CSF.^11–13

Before concluding that the cells identified in this study are cardiac progenitor cells, it will be necessary to demonstrate that they can regenerate damaged myocardium in vivo. Accordingly, these cells are described in our article as potential cardiac progenitors or cells expressing early cardiac markers. However, we are encouraged by our recent preliminary data showing that these cells can indeed differentiate in vitro into cells that express cardiac-specific markers such as troponin I and cardiac-specific myosin heavy chain (online Figure 1).

BM-Derived Cells Expressing Early Cardiac Markers and Aging

We observed an age-dependent change in the abundance of cells expressing early cardiac markers in murine BM, resulting in barely detectable levels at 1 year of age. These striking differences between old and young animals suggest that these cells are deposited in BM early in ontogenesis during rapid body growth/expansion. We also found that the responsiveness of these cells to SDF-1, HGF, and LIF gradients varies with age: whereas the cells from younger mice responded better to an SDF-1 gradient, the cells from older animals were chemoattracted more efficiently by HGF and LIF. These observations suggest a possible age-dependent phenotypic (density and/or functionality) shift involving the expression patterns of receptors (CXCR4, c-Met, and LIF-R) in these cells. The striking differences in the chemotactic responses of potential cardiac progenitors to these factors have obvious practical implications for BM cell– based therapies.

MI Causes Release of Cells Expressing Early Cardiac Markers That Migrate to SDF-1, HGF, and LIF Gradients

Our data demonstrate for the first time that potential cardiac progenitors are mobilized into the PB after MI, with a peak around 48 hours after MI and a gradual decline to nearly undetectable levels by 96 hours. By immunohistochemical staining, 48 hours after MI ≈0.1% of PBMNCs were found to express both GATA-4 and Nkx2.5/Csx proteins. In addition, our data demonstrate that the expression of SDF-1, HGF, and LIF increases in infarcted myocardium and that the chemoattraction of BM-derived potential cardiac progenitors to supernatants from infarcted myocardium can be completely abolished by T140, K252a, and anti-gp190, specific inhibitors of CXCR4, c-Met, and LIF-R, respectively, implying that these cytokines play an important role in orchestrating the migration of these cells.

The concept, supported by our findings, that SDF-1 is one of the crucial chemoattractants for recruiting BM-derived cardiac progenitors in ischemic cardiomyopathy is consistent with previous reports.^33,34 Our data also point to a potential role of the HGF-c-Met and LIF-LIF-R axes in myocardial regeneration. HGF has been recently shown to increase both proliferation and survival of cardiomyocytes by directly affecting GATA-4 expression.³⁵ Similarly, LIF has been reported to enhance survival of cardiomyocytes and to induce regeneration of myocardium after MI.³⁶

Because these cells reside in the BM, are mobilized into the PB after MI, and can be potentially chemoattracted to injured myocardium in an SDF-1−, HGF−, and LIF-dependent manner, the question rises as to why the cardiac regenerative ability of BM-derived cells remains controversial. There are several potential explanations. First, the number of potential cardiac progenitors residing in the BM may be too low and thus insufficient to reconstitute myocardium after the extensive cardiomyocyte loss associated with MI. The primary role of these cells may be to regenerate/repair the relatively small number of cardiomyocytes that die of apoptosis at steady state. Second, circulating cardiac progenitors may have to compete with other CXCR4⁺ cells (eg, macrophages and lymphocytes) for “homing” into the myocardium and thus may not effectively reach the infarcted region. Finally, in the case of older individuals or individuals experiencing chronic ischemic heart disease, the pool of cardiac progenitors in the BM may become depleted with time, as supported by our present findings.

Conclusions

The present study provides a novel paradigm to explain the ability of BM-derived cells to induce myocardial repair. Our results demonstrate that the postnatal BM contains cells that express markers for cardiac differentiation commitment, and that the abundance of these cells declines with age, concomitant with changes in responsiveness to chemoattractants. These cells constitute a CXCR4⁺/lin⁻/CD45⁻ subset of BMMNCs that are CD34⁺/AC133⁺ in humans and Sca-1⁺ in mice and are distinctly nonhematopoietic. These potential cardiac progenitors are mobilized into the PB after MI and are capable of migrating to SDF-1, HGF, and LIF; chemoattractants that are expressed in infarcted myocardium. On the basis of these observations, we propose the novel idea that these potential cardiac progenitors contribute to BM-dependent cardiac regeneration by differentiating into cardiac myocytes and vascular cells, a paradigm that involves neither cell fusion nor transdifferentiation of HSCs. According to this scenario, the BM is a reservoir of stem cells that have the potential to differentiate into cardiac lineage and that may be continually released into the PB to reach the heart where they contribute to its physiological turnover. These same cells may account for the cardiac regenerative effects of BM in acute MI. Future studies will be necessary to demonstrate the efficacy of these cells in repairing cardiac tissue after MI.