SEARCH

SEARCH BY CITATION

Keywords:

  • Adult stem cells;
  • Cardiac;
  • Cell culture;
  • Differentiation;
  • Myogenesis;
  • Myofibroblast

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Recent remarkable studies have reported that clonogenic putative cardiac stem cells (CSCs) with cardiomyogenic potential migrate from heart tissue biopsies during ex vivo culture, and that these CSCs self-organize into spontaneously beating cardiospheres (CSs). Such data have provided clear promise that injured heart tissue may be repaired by stem cell therapy using autologous CS-derived cells. By further examining CSs from the original CS protocol using immunofluorescence, quantitative reverse transcription-polymerase chain reaction, and microscopic analysis, we here report a more mundane result: that spontaneously beating CSs from neonatal rats likely consist of contaminating myocardial tissue fragments. Thus, filtering away these tissue fragments resulted in CSs without cardiomyogenic potential. Similar data were obtained with CSs derived from neonatal mice as wells as adult rats/mice. Additionally, using in vitro culture, fluorescence-activated cell sorting, and immunofluorescence, we demonstrate that these CSs are generated by cellular aggregation of GATA-4+/collagen I+/α-smooth muscle actin (SMA)+/CD45 cells rather than by clonal cell growth. In contrast, we found that the previously proposed CS-forming cells, dubbed phase bright cells, were GATA-4/collagen I/α-SMA/CD45+ and unable to form CSs by themselves. Phenotypically, the CS cells largely resembled fibroblasts, and they lacked cardiomyogenic as well as endothelial differentiation potential. Our data imply that the murine CS model is unsuitable as a source of CSCs with cardiomyogenic potential, a result that is in contrast to previously published data. We therefore suggest, that human CSs should be further characterized with respect to phenotype and differentiation potential before initiating human trials. STEM CELLS 2009;27:1571–1581


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

In the U.S. alone, 4.9 million patients suffer from heart failure, and with 550,000 new cases diagnosed each year, there is a high demand for novel treatments that can increase patient survival. Regenerative stem cell therapy is a new promising approach to repair and improve the function of the injured heart, although the ideal donor stem cell remains to be determined [1]. Embryonic stem (ES) cells [2, 3] and induced pluripotent stem cells [4, 5] possess clear cardiomyogenic potential, but have biological and/or ethical limitations. Several adult cardiac stem cell (CSC) populations have been reported to hold similar potential, but they are poorly identified by use of a single proposed stem cell marker (c-Kit [6], Abcg2 (side population) [7], or Sca-1 [8]). However, Messina et al. [9] suggested that putative CSCs might be enriched during ex vivo expansion of heart explant-derived cells, by an intrinsic potential of CSCs to migrate from the explant and self-organize into spherical clusters, referred to as cardiospheres (CSs). One remarkable feature of this CS model is that putative CSCs can be rapidly expanded into clinically relevant cell numbers for transplantation purposes [10]. Phenotypically, it has been reported that CS cells (a) self-renew, (b) express stem cell markers, and (c) differentiate into fully functional cardiomyocytes, which collectively classify them as actual adult CSCs. In line with this, CSs beat spontaneously (embryonic murine CSs) or when co-cultured with rat myocytes (adult mice, dog, pig as well as human CSs) [9–11]. Finally, intracardiac delivery of human CS-derived cells (CDCs) resulted in engraftment and partial regeneration of injured myocardium in SCID-beige mice [10]. Thus, CDCs have been suggested to be potent therapeutic stem cells for patients with heart failure [10], and they are currently being tested in preclinical studies.

However, the cardiomyogenic potential of CSs has been questioned, and recently Shenje and colleagues demonstrated that explant migrating cells (EMCs) lack cardiomyogenic potential [12]. These EMC results thus clearly contradict the CS data [9, 12], although the EMC study [12] did not include analyses of the CS stage of EMCs, a step that has been suggested to be obligate to unmask the cardiomyogenic potential of EMCs [9, 10].

Herein, we clarify and evaluate the cardiomyogenic potential of EMCs and the CS model by thorough phenotypical characterization of EMCs and CSs.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Animals and Cells

Experiments were performed with neonatal (1-3 days) and adult (>10 weeks) Wistar rats as well as neonatal (2 days) and adult (8-10 weeks) C57/bl6 mice (all Taconic Europe, Skensved Denmark, http://www.taconic.com). Rat fibroblasts, endothelial cells (kind gifts from Hasse Brønnum Jensen, University of Southern Denmark), and C2C12 (kind gift from Charlotte Harken Jensen, University of Southern Denmark) were maintained as recommended by the suppliers. Neonatal rat myocytes were obtained and cultured as previously described [13], as were mouse ES cells (strain 663) [14].

Generation of CSs

In all experiments—neonatal rat (n = 26, n* = 5–10, where n* refers to the number of animals in each experiment), adult rat (n = 7, n* = 1–2), neonatal mouse (n = 2, n* = 10), adult mouse (n = 12, n* = 5)—ventricular (right and left) rat heart tissue was dissected, washed in preparation buffer (1.2 mM KH2PO4 (pH 7.4), 0.25 g/l Na2CO3, 6.44 g/l NaCl, 2.6 mM KCl, 1.2 mM Mg2SO4, and 11 mM glucose, incubated at 37°C in 5% CO2, 24 hours before use) with 70 IE/ml heparin, and minced into 1- to 2-mm3 cubes with scalpels. Initially, tissue chunks were treated exactly as described by Smith et al. [10] with brief enzymatic dissociation, though as stated in results, we omitted this digestion step in many experiments because no difference was observed with or without enzymatic digestion (supporting information Fig. S1). Processed tissue pieces were plated on fibronectin (only experiments corresponding to those in Figs. 1, 2A–2J, supporting information Fig. S1, supporting information Movies S5–S8) (BD Falcon; BD Biosciences, San Diego, http://www.bdbiosciences.com) or 8% fetal calf serum (FCS) (only experiments corresponding to those in Figs. 2K, 2L, 3–7) coated dishes (no apparent difference was observed between fibronectin and FCS coating, supporting information Fig. S1) and cultured (8 and 15-25 days for neonatal and adult animals, respectively) in explant medium (Iscove's modified Dulbecco's medium [IMDM] (#31980), 20% FBS, 1% penicillin-streptomycin [PS]; all products of Gibco, Grand Island, NY, http://www.invitrogen.com). Fresh medium was added every third day. EMCs were harvested as previously described [9, 10], though in experiments that included cell sorting, we prolonged the enzymatic detachment period to 5 minutes in order to assure that all EMCs were contained within the final cell suspension. Where indicated, we additionally filtered the EMC cell suspension serially through 100- and 40-μm cell strainers (BD Falcon). For CS formation, cells were plated at 4 × 104 cells/cm2 (1–10 × 104 cells/cm2 for the cell plating density study) in poly-d-lysine-coated multidishes (BD Biocoat; BD Biosciences) and cultured in CS medium [10]—35% IMDM, 65% Dulbecco's modified Eagle's medium–F12 (#11320; Gibco), 3.5% FBS, 1% PS, 0.1 mM β-mercaptoethanol (#31350-010; Gibco), 2% B27 (17504-044; Gibco), 80 ng/ml basic fibroblast growth factor (#01-106; Upstate, Charlottesville, VA, http://www.upstate.com), 25 ng/ml rat epidermal growth factor (#400-25; PeproTech, Rocky Hill, NJ, http://www.peprotech.com), 4 ng/ml cardiotrophin (#250-25; PeproTech), and 1 U/ml thrombin (#T9549; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com)—for 2–14 days with the addition of fresh medium every 4 days. Generated CSs were collected by gentle trituration, and two rounds of sedimentation (gravity) and washing of CSs were used to remove contaminating single cells if present. In some experiments, CSs were dissociated into single cells and cultured as described previously [9, 10], in a monolayer (termed CDCs).

thumbnail image

Figure 1. Neonatal rat cardiosphere (CS) formation. (A): At day 3, fibroblast-like cells have migrated from the heart tissue specimen, and 5 days later (B), phase bright cells appear on top of this fibroblast-like layer. (C): Immunocytochemical α-sarcomeric actinin staining in neonatal rat heart explant. (D–F): Explant-migrating cells that were replated under CS conditions. Clearly defined CSs are seen at day 2 and 7 of culture. Scale bars = 100 μm (C) and 250 μm (A, B, D–F).

Download figure to PowerPoint

thumbnail image

Figure 2. Neonatal rat CSs are noncardiogenic and formed by aggregation. (A): CSs were embedded in paraffin, sectioned, and H&E stained. (B–G): Immunohistochemical staining for α/β-MHC in day 2 CSs generated from nonfiltered explant-derived cells (B), neonatal rat heart (C), paraffin-embedded neonatal rat heart day 8 explant (D), day 2 and day 7 CSs generated from filtered explant-derived cells (E, F), and CDC monolayer culture of day 2 CDCs (G). (H): Immunohistochemical staining for tropomyosin and troponin T in neonatal rat hearts and day 2 CSs generated from filtered and nonfiltered explant-derived cells. (I): Immunocytochemical staining of early cardiac markers (Nkx2.5, Islet-1, Mef2c) in CDC monolayer cultures of day 2 CDCs; scale bars = 50 μm. Differentiated mouse embryonic stem cells were used as positive controls for recognition of these transcription factors; scale bars = 25 μm. (J): Relative quantitative RT-PCR of GATA-4 (black) and Nkx2.5 (white) in neonatal rat hearts and CSs (day 2 and day 7). (K): CS number and size versus cell plating density of explant-derived cells under CS conditions. (L): Confocal laser scanning microscopy of a day 2 CS that was generated by plating a mixture (1:1) of DiO-labeled (green) and DiI-labeled (red) explant-derived cells under CS conditions; scale bar = 200 μm. Abbreviations: CDC, cardiosphere-derived cell; CS, cardiosphere; H&E, hematoxylin and eosin; MHC, myosin heavy chain; RT-PCR, reverse transcription-polymerase chain reaction.

Download figure to PowerPoint

thumbnail image

Figure 3. Characterization of neonatal rat heart explants and cardiospheres (CSs) (day 2 and day 7). Triple immunohistochemical staining of paraffin-embedded sections for collagen I (green), α-smooth muscle actin (red), and desmin (purple). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (blue). Pictures were captured in each fluorescence channel and briefly processed to allow visualization of the merged picture (bottom panel). Scale bars = 100 μm.

Download figure to PowerPoint

thumbnail image

Figure 4. Characterization of neonatal rat heart explants and cardiospheres (CSs) (day 2 and day 7). Double immunohistochemical staining of paraffin-embedded sections for CD45 (green) and collagen I (red), GATA-4 (red), and Ki67 (red). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (blue). Pictures were captured in each fluorescence channel and briefly processed to allow visualization of the merged picture (bottom panel). Scale bars = 100 μm.

Download figure to PowerPoint

thumbnail image

Figure 5. Evaluating the cardiosphere (CS) potential of CD45+ and CD45 explant-migrating cells (EMCs). (A): Phase picture of a neonatal rat heart explant. Migrating fibroblast-like cells with phase bright cells on top (box) are magnified in (B), where CD45 (green) was visualized by immunocytochemical staining; scale bars = 100 μm. (C, D): Fluorescence-activated cell sorting (see supporting information Fig. S6) and analysis of CD45+ and CD45 EMCs. (E): Relative quantitative reverse transcription-polymerase chain reaction of Cd45 transcripts in freshly isolated and sorted EMCs. (F–K): Freshly sorted (CD45+ and CD45) and nonsorted (outspring) cells cultured for 12 or 48 hours under CS conditions. (F–H): scale bars = 100 μm; (I–K): scale bars = 500 μm and 100 μm (insets).

Download figure to PowerPoint

thumbnail image

Figure 6. Characterizing the fibroblast phenotype of CD45 explant-migrating cells (EMCs) and the CD45 cardiospheres (CSs) derived thereof. (A): Freshly sorted CD45+ and CD45 EMCs cultured for 12 hours on fibronectin were triple (collagen I, green; α-smooth muscle actin [α-SMA], red; desmin, purple) or double (CD45, green; GATA-4, red) stained by immunocytochemistry. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI, blue). (B, C): Immunohistochemical staining (triple: collagen I, green; α-SMA, red; desmin, purple or double: CD45, green; GATA-4, red) of paraffin-embedded sections of CSs (day 2 and day 7) generated from CD45 EMCs; scale bars = 100 μm (A) and 25 μm (B, C). (D): Primary isolated and cultured adult rat cardiac fibroblasts were stained for collagen I, α-SMA, desmin, GATA-4, DDR2, filamentous actin (phalloidin), vimentin, and sarcomeric actinin (E) and cultured under CS conditions; scale bars = 100 μm (D) and 250 μm (E).

Download figure to PowerPoint

thumbnail image

Figure 7. Examination of the angiogenetic potential of explant-migrating cells (EMCs) and cardiosphere (CS) cells. (A, B): Cell morphology (scale bars = 100 μm) and Ac-DiI-LDL uptake (flow cytometry) of cells that, except for fibroblasts (normal growth medium), were cultured 8 days under endothelial differentiation conditions. (C–E): Phase pictures of CD45 explant-derived (C, E) and CD45 CS-derived (D) cells cultured 2 days (C, D) or 8 days (E) on Matrigel in endothelial differentiation medium. (F): Relative quantitative reverse transcription-polymerase chain reaction of Cd31, Cdh5, and miR-126 in freshly sorted EMCs, CDCs cultured in endothelial differentiation conditions (derived from day 4 CSs), as well as whole heart tissue.

Download figure to PowerPoint

Cell Counting

Cell numbers were determined using a Beckman Coulter Counter Z2 (Ramcon, Birkeroed, Denmark, http://www.ramcon.dk) fitted with a 100 = μm aperture. The size range of particles counted was set at 10.5-26 μm, and counting was performed in triplicate.

Cocultures and Aggregation Studies Using DiI- and DiO-Labeled Cells

Freshly isolated explant-derived cells were split into two pools and labeled with DiI or DiO (Vybrant multicolor cell labeling kit, #V22889; Invitrogen, Carlsbad, CA, http://www.invitrogen.com). Labeled cells were washed extensively, counted, and plated under CS-forming conditions. For aggregation studies, we mixed the two cell populations in a ratio of 1:1 and visualized DiI+ and DiO+ cells in CSs by confocal laser-scanning fluorescence microscopy (Olympus FV1000; Olympus, Tokyo, http://www.olympus-global.com) using a ⋄ 20 (numerical aperture, 0.5) Olympus water immersion objective. For coculture studies, DiO-labeled cells were used to generate CSs that were collected and plated on top of a layer of DiI-labeled myocytes. Beating of DiO-labeled cells was assessed by microscopy (using a Leica DMI4000B Cool Fluo Package instrument equipped with a Leica DFC340 FX Digital Cam; Leica, Heerbrugg, Switzerland, http://www.leica.com).

Immunofluorescence, Endothelial Differentiation, FACS, and Quantitative Reverse Transcription-Polymerase Chain Reaction

Immunocytochemistry, immunohistochemistry, and cell sorting were performed as previously described [15, 16]. A comprehensive description including endothelial differentiation conditions and a list of antibodies used can be found in supporting information data. The detailed FACS strategy is also available in supporting information data (supporting information Fig. S2). Relative quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was performed as described in supporting information data, although the raw data were retrieved with MxPro QPCR Software (Stratagene, La Jolla, CA, http://www.stratagene.com) and normalized against the optimal number of the most stably expressed endogenous control genes (supporting information Fig. S3) as determined by the freely available GeNorm [17] and q-Base [18] platforms as previously described [15].

Statistical Analysis

Each analysis consisted of at least three independent experiments, designated n; where indicated, n* refers to the number of animals in each experiment. The statistical analysis was performed using Student's t-test (significance level, α = 0.05), one-way analysis of variance (repeated measures, α = 0.05), or simple column statistics (mean ± standard deviation [SD]). GraphPad Prism software (4.0a Mac version; GraphPad Software Inc., La Jolla, CA, http://www.graphpad.com) was used for all statistical calculations.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Establishment of Neonatal Rat Heart CSs

CSs have previously been described for mouse, human, and dog [9–11], but because the rat is the best-known model for obtaining cardiomyocytes, we hypothesized that rat hearts would be an ideal source of cardiomyogenic CSCs for CS formation. Initially, we addressed whether CSs could be derived from neonatal rat hearts. We used the protocol described by Smith et al. [10], and minced rat ventricular tissue into 1- to 2-mm3 small pieces that were plated on fibronectin-coated dishes. After a few days, small fibroblast-like cells migrated from the explants (Fig. 1A), and at day 6-8 numerous small round “phase bright cells” appeared on top of the fibroblast layer (Fig. 1B) in approximately 25% of the attached explants (supporting information Fig. S1). Beating of neonatal rat heart explants was commonly observed (supporting information Movie S5), but was absent in explants derived from adult rats (data not shown). Whereas the original heart tissue of the explant culture revealed striated α-sarcomeric actinin staining, none of the migrated cells (fibroblast-like cells and phase bright cells) expressed this protein (Fig. 1C). By gentle trypsinization of day 8 explant cultures, migrated cells were harvested, and cells were cultured on poly-d-lysine-coated dishes in low serum (3.5%) containing medium supplemented with growth factors [10]. As soon as 12-24 hours following plating, groups of cells began to approach each other in foci (Fig. 1D), and in agreement with others [9], floating CSs of cells had formed after 48 hours of culture (Fig. 1E, 1F).

Neonatal Rat CSs Do Not Possess Cardiomyogenic Features

Spontaneous beating of embryonic mouse CSs has previously been described [9], though in the present study beating was only observed in a few neonatal rat CSs (supporting information Movie S6). Because whole-mount immunocytochemical staining of CSs revealed high levels of nonspecific antibody reactivity (data not shown), we performed immunohistochemistry on sections of paraffin-embedded neonatal rat CSs (Fig. 2A). Accordingly, a few CSs stained positive for the late cardiac markers α/β-myosin heavy chain (MHC) (Fig. 2B), cardiac troponin T (cTnT), tropomyosin (Fig. 2H), and sarcomeric α-actinin (data not shown). Because migrating explant cells never stained positive for late cardiac markers (in contrast to hearts and explants; Fig. 2C, 2D), we speculated that the few beating CSs and the α/β-MHC-, cTnT-, tropomyosin-, and sarcomeric α-actinin-positive CSs could be a result of the presence of “contaminating” heart tissue fragments in the explant-derived cell suspension. We therefore included an intervening step of filtration in which the explant-harvested cells were serially passed through cell strainers prior to CS formation. Subsequent experiments (n = 18) never revealed beating (data not shown) nor α/β-MHC (Fig. 2E–2G), cTnT, tropomyosin (Fig. 2H), or α-actinin (data not shown) staining of neonatal rat CSs (day 2 or day 7) or monolayer cultures of CDCs. In contrast, we observed that tissue clumps retained within the cell strainers were indistinguishable from CSs (microscopic examination), and many of these started to beat spontaneously under CS culture conditions (supporting information Movie S7). Additionally, we obtained CS cultures from neonatal mice as well as adult rats and mice, and these were also devoid of beating when the filtration step was included in the protocol (supporting information Movie S8). Together these results strongly indicate that the filtration step is necessary to avoid contamination of heart tissue fragments in explant cell harvests destined for CS formation, and we therefore applied this step to the remainder of the experiments presented herein. We also observed that explant attachment, beating, cell migration, phase bright cell appearance, CS number, and CS size were independent of the brief enzymatic pretreatment (supporting information Fig. S1) and therefore omitted this step in further experiments. Recent evidence suggests that Islet-1 is found in early cardiac precursors [19] and Nkx2.5 [20] and Mef2c [21] are found in cells of the cardiomyocyte lineage, whereas GATA-4 is associated with both muscle- and nonmuscle cells of the heart [22]. Immunocytochemistry of CDCs verified that neonatal rat CSs lack cardiac precursors expressing Islet-1, Nkx2.5, or Mef2c (Fig. 2I), and relative qRT-PCR confirmed this (Nkx2.5) and further showed that neonatal rat CSs expressed GATA-4 (Fig. 2J). Altogether, we infer from these results that cardiomyogenesis is not intrinsic to CSs.

To improve the cardiomyogenic culture environment, we cocultured neonatal CSs (DiO+) with monolayers of beating rat cardiomyocytes (DiI+). In contrast to what has been described for human CSs [10], neonatal rat CSs quickly lost sphere integrity and intermingled with the cells on which they were plated (data not shown). Importantly, we were not able to detect any beating CDCs in these cultures. Similar results were obtained with CSs established from adult rats and mice as well as neonatal mice (data not shown), which supports the above assumption that CS cells are devoid of a cardiomyogenic potential.

Neonatal Rat CSs Are Formed by Aggregation

It has not been clearly established whether CSs are formed by clonal expansion or cell aggregation [9, 23]. We found that neonatal rat CSs already consisted of at least 50 cells (as estimated from pictures like that in Fig. 1E) after 48 hours of plating, and that the number and size of the CSs depended on the initial cell plating density (Fig. 2K). Hence, the CS number was approximately fourfold greater when the cell plating density was raised from ∼10,000 to ∼40,000 cells/cm2, but declined thereafter (Fig. 2K). In contrast, CS size increased proportionally with cell plating density (Fig. 2K), indicating that CSs formed by cellular aggregation. To confirm this, we labeled explant-derived cells with either DiO or DiI and mixed these two cell suspensions before coculturing them for CS formation. By confocal microscopy, individual CSs were observed to harbor both DiI- and DiO-labeled cells (Fig. 2L), strongly suggesting that neonatal rat CSs form mainly by cellular aggregation.

Hematopoietic as Well as Nonhematopoietic Cells Are Found Within Neonatal Rat CSs

Using immunohistochemistry, we next examined the phenotype of cells that remained within neonatal rat CSs (Figs. 3, 4). As stated above, both fibroblast-like cells and phase bright cells migrated out of the tissue explant (Fig. 1A, 1B) and could potentially be the origin of CS-forming cells, though others have proposed that phase bright cells are the sole source of CS cells [9, 11]. Herein, analyses of paraffin-embedded sections of neonatal rat heart explants showed that cells that had migrated out of explants were either GATA-4+/collagen I+/α-smooth muscle actin (SMA)+/CD45 or GATA-4/collagen I/CD45+ (Figs. 3, 4). At day 2 of CS formation, CSs seemed to be comprised of both of these cell phenotypes, but with time (day 7) CSs consisted of a core of GATA-4+/collagen I+/α-SMA+/CD45 (Figs. 3, 4) cells surrounded by GATA-4/collagen I/CD45+ cells (Fig. 4). The central part of CSs encompassed a space characterized by a very low density of nuclei (Figs. 3, 4). Desmin, a protein associated with smooth muscle cells, and also cardiac myocytes, was expressed in the cells located in or lining this interior space. Ki67, a marker of proliferation, was expressed in both CD45+ and CD45 cells found within explants as well in newly formed CSs (day 2 CSs), but in day 7 CSs it was mainly found within the fraction of CD45+ cells (Fig. 4). Moreover, we examined the expression of c-Kit, a marker associated with CSCs [6] and human CSs [10]. Compared with neonatal rat hearts, explant-derived cells exhibited 29.9-fold ± 4.7-fold (mean ± SD; n = 3) lower levels of c-Kit (supporting information Fig. S3), indicating that explant cultures are not enriched for c-Kit+ cells.

Neonatal Rat CS-Forming Cells Are Nonhematopoietic and Exhibit a Fibroblast-Like Phenotype

Based on the above observations, we hypothesized that the fibroblast-like cells in the CSs reflected the migratory fibroblast-like cells in the explants, whereas the CD45+ CS cells originated from the phase bright cell fraction. Indeed, immunocytochemical staining of explants, performed directly in the culture plate, revealed that phase bright cells were CD45+ (Fig. 5A, 5B). In agreement with our observations that CSs consisted of both CD45+ and CD45 cells, others have shown that heart-residing bone marrow-derived cells contribute to CSs [23]. However, this issue still seem elusive because another study suggested that CDCs lack expression of CD45 and other hematopoietic lineage markers [10]. In order to address this topic in more detail, we isolated the two, CD45+ and CD45, explant cell fractions by FACS (Fig. 5C, supporting information Fig. S2). The purities of the sorted CD45 and CD45+ cell samples were around 95% and 90%, respectively (supporting information Fig. S2), and the ratio of CD45 to CD45+ cells in the outspring explant-derived cell population was approximately 1:1 (Fig. 5D). Relative qRT-PCR of the Cd45 transcript in sorted samples verified the cell enrichment achieved by cell sorting (Fig. 5E). When culturing the isolated CD45+ and CD45 cell fractions and the outspring (nonsorted explant-derived cells) population under CS conditions (Fig. 5F–5K), no CSs were generated from the CD45+ cell population (Fig. 5K). In contrast, the majority of CD45 cells participated in CS formation (only a few cells remained attached as single cells to the culture dish) (Fig. 5J), and the border of CD45-derived CSs was more even and pronounced (Fig. 5J, inset) than that of outspring CSs (Fig. 5I, inset). Morphologically CD45 cells revealed a fibroblast-like phenotype (Fig. 5G), whereas CD45+ cells were smaller and more round with long thin “arms” (Fig. 5H). In addition, these CD45+ cells were trypsin resistant, and within cultures of nonsorted spheres they gave rise to remarkable large multinuclear cells (supporting information Fig. S4) that resembled osteoclasts. Immunocytochemical staining of freshly sorted and briefly (12 hours) cultured CD45+ and CD45 EMCs verified that CD45+ cells were collagen I/α-SMA/desmin/GATA-4 (Fig. 6A). In contrast, CD45 cells were positive for these proteins, although only a minor fraction expressed desmin (Fig. 6A). As expected, this immunophenotype persisted in the neonatal rat CD45 CSs that were generated from the CD45 EMCs (Fig. 6B, 6C). This indicated that neonatal rat CS-forming cells resembled fibroblasts, and to support this postulation, we examined the potential of adult rat cardiac fibroblasts (Fig. 6D) to generate spheres (Fig. 6E). As with the neonatal rat heart explant-derived cells, fibroblasts gave rise to numerous clearly defined spheres after 24–48 hours of culture (Fig. 6E). These results thus show that it is the explant-derived CD45 fibroblast-like cells that harbor the ability to form CSs—an ability shared with primary cardiac fibroblasts.

Neonatal Rat CS Cells Reveal Some Features of Vascular Cells But Lack Complete Endothelial Differentiation Potential

Finally, we examined whether the CD45 explant-derived cell population or its derivative CSs (CD45 CSs) possessed endothelial differentiation potential, as previously suggested for skeletal muscle-, fat-, and dermis-derived sphere cells [24]. Endothelial cells are known to take up low density lipoprotein (LDL), whereas fibroblasts in proliferative growth lack this ability. By FACS analysis and microscopic examination, we found that both CD45 cells and CD45 CS cells took up some LDL (Fig. 7B). Furthermore, CD45 EMC cells (Fig. 7C) but not CD45 CS cells (Fig. 7D) exhibited capillary formation when cultured for 2 days on Matrigel, and after 8 days of culture some of these tube-like structures revealed alkaline phosphatase staining (Fig. 7E), an ability used to characterize endothelial cells [24]. However, even though these results indicate that both CD45 EMC cells and their derivative CD45 CS cells exhibited some features of endothelial cells, we never observed an endothelial morphology (Fig. 7A). To achieve a more detailed analysis of the endothelial differentiation potential, we therefore performed relative qRT-PCR of Cd31, Cdh5, and miR-126, all factors associated with endothelial cells (Fig. 7F). The levels of all three transcripts were much higher in neonatal rat hearts than in any of the other samples. In summary, these results suggest that cells in neonatal rat CD45 CSs share some features of endothelial cells, but they do not have a genuine ability to enter the endothelial lineage.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

In this study, we confirmed that migrating cells from murine explanted heart tissue are able to form CSs during ex vivo culture. However, our results show that CSs may not hold cardiomyogenic potential, and that they represent aggregated fibroblasts and not clonally expanded putative CSCs, implying that the CSs do not serve as a model for producing CSCs.

EMCs (fibroblast-like and phase bright cells) as well as the CSs described here are fully comparable with those described in previous studies [9–11, 25], because we used identical culture conditions. In some experiments, we implemented a few modifications to the original CS method, but these were carefully examined and, importantly, no discrepancies were observed with these small changes. In agreement with others, we thus found that EMCs were able to form CSs, and some expressed late cardiac markers and started to beat spontaneously. However, unexpectedly, we found that filtration of the EMC population resulted in complete loss of cardiomygenic activity. This was a rather disappointing observation, implying that the observed cardiomyogenic CSs likely consisted of small contaminating myocardial tissue remnants. Of particular importance, we noted that small heart tissue explants in suspension were impossible to distinguish from CSs in terms of morphology, and because some of them beat, they could easily be mistaken as beating CSs. Disagreements between our study and work by others [9–11, 25] could therefore likely be explained by contaminating small heart tissue pieces in the CS culture. Corroborative with the lack of differentiated cardiomyocytes in the CSs, we also did not observe features of CSCs/progenitor cells, like c-Kit [6], Nkx 2.5 [20], and Islet-1 expression, indicating that putative CSCs were not enriched in the EMC population, as previously proposed by some studies [8, 26], but in agreement with others [12].

In the same context, our results also demonstrate that CSs mainly formed by cellular aggregation, rather than by clonal expansion, which could have been expected from “true” CSCs. This result contradicts a recent study suggesting that CSs were clonally derived from an undifferentiated EMC [9]. However, our results do not exclude that some CSs may have formed by clonal expansion from single elements. Interestingly, neurospheres (NSs), which resemble CSs in terms of the derivation method and culture conditions [27], have long been considered a quiescent and homogeneous population of neural stem/progenitor cells. However, recent data have demonstrated that NSs are formed by nonclonal growth of a heterogeneous mixture of cells at distinct maturation stages [28].

Furthermore, we found that CSs exhibited a signature analogous to that of the preceding layer of explant-migrating fibroblast-like cells, suggesting that these cells, and not phase bright cells, as previously proposed [9–11], were the source of CSs. Indeed, we found that CS-forming cells were GATA-4+/collagen I+/α-SMA+/CD45, a cell phenotype that corresponds to that of cardiac fibroblasts. Desmin, a protein expressed in cardiomyocytes as well as smooth muscle cells [29], was observed in a small fraction of the CD45 CS population, but because no other myogenic proteins colocalized to these cells, they probably represent smooth muscle cells that had also migrated from the heart explant. Interestingly, one study argued that plating on poly-D-lysine ensured negative selection of fibroblast-like cells by strong adherence to the culture dish, whereas undifferentiated phase bright cells remained in suspension and formed CSs [9]. Conversely, a follow-up study by the same authors demonstrated that human CDCs shared similarities with fibroblasts, in that both cell types expressed CD105 and CD90, two well-known markers of fibroblasts [30–32]. The authors suggested that this fibroblast-like population supported CSCs during CDC expansion [10]; however, because all CDCs in that study [10] were CD105+, proposed CSCs must have been CD105+ as well. Others have recently demonstrated that sphere-derived cells from various tissues differentiate into endothelial cells [24]; however, as with the lack of cardiomyogenic potential, CD45 CSs did not display endothelial differentiation, further supporting a fibroblasts-like cell type.

Surprisingly, we found the subpopulation of refractive phase bright cells to be positive for the panhematopoietic marker CD45, indicating that these might be bloodborne cells. A recent study suggested a similar origin for these cells because explants from hearts predeprived of blood were devoid of phase bright cells [12]. In the present study, we found that CSs comprised of CD45+ cells, but not CD45 CSs, gave rise to large multinuclear cells that resembled osteoclasts (supporting information Fig. 4), suggesting a CD45+ EMC origin of these giant cells. One possibility is that the CD45+ population is comprised of circulating CD14+ monocytes, which are known to differentiate into osteoclasts [33], a scenario that could explain the previously reported phagocytotic activity of phase bright cells [12]. Nevertheless, our results clearly show that the CD45+ sorted population, which included a subpopulation of phase bright cells, did not give rise to CSs, as previously proposed [9–11]. Instead, we propose a scenario in which CD45+ cells are codetached from the culture plate with the GATA-4+/collagen I+/α-SMA+/CD45 cells that actually display the ability to form the spheres, and that a mixture of these two cell types, therefore, forms the CSs.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

In conclusion, the present study demonstrates that phase bright cells are blood-derived cells and not cardiac progenitors, and that CS-forming cells likely represent fibroblasts. The inability of CSs to unmask cardiomyogenic potential may well be explained by the compilation of predifferentiated cardiac fibroblasts, and strongly argues that growth in suspension does not select for CSCs with cardiomyogenic activity. This study raises caution about CSs as a source of therapeutic CSCs and strongly suggests that human CSs and CDCs should be further characterized with respect to phenotype and differentiation potential before initiating human trials. However, although we have clearly provided evidence that the murine CS model does not give rise to cardiac lineage cells, we cannot exclude that EMCs, CSs, or cells derived thereof will have beneficial effects when transplanted into injured hearts. It is likely that fibroblasts, with their known supportive functions, will improve cardiac function even in the absence of cardiomyogenesis/angiogenesis.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

The authors thank Bettina Mentz (LMCC, OUH) for technical help, Anette Kliem and Charlotte Harken Jensen (Immunology and Microbiology, SDU) for reagents and preparation of histological specimens, G. Leslie and I. Andersen (FACS core facility, SDU) for help with FACS, Per Svenningen (Physiology and Pharmacology, SDU) for help with confocal microscopy, and Moustapha Kassem (KMEB, OUH) for help with alkaline phosphatase staining. This work was supported by Odense University Hospital, The John and Birthe Meyer Foundation, and The Danish Ministry of Science, Technology and Innovation.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Additional supporting information available online.

FilenameFormatSizeDescription
STEM_72_sm_SuppFig1.tif814KSupporting Information Figure S1. Comparison of CS protocols. Neonatal rat hearts were minced into small fragments and plated as explants (modified protocol) or subjected to trypsine/collagenase treatment (original protocol) before plating as explants. Explants were plated on either 8% FBS (modified protocol) or fibronectin (original protocol) coated plates. P values were calculated using one-way ANOVA (repeated measures) statistical analyses. No significant differences of the analyzed parameters were observed between the different protocols.
STEM_72_sm_SuppFig2.tif1487KSupporting Information Figure S2. FACS sorting gating strategy used to isolate explant migrating CD45+/- cells. First, debris was excluded by gating on cells using size (FSC-A) and granularity (SSC-A). This FSC/SSC gate was established (data not shown) by incubating a small aliquot of the sample (non-stained to avoid spectral overlap between PI and PE-Cy5) with propidium iodide (PI); The live cell fraction (PI-) was then visualized in the FSC/SSC plot by backgating. Secondly, cell doublets, the major contaminant of most sortings, were excluded in a two-step manner, and thirdly, singlet cells were distinguished by CD45 expression. A second round of sorting established the purities of FACS sorted CD45-/+ samples. 96.3% and 90.1% of the sorted CD45- and CD45+ cells fell into the sort gates, respectively. Only gated cells are shown in the next plot, which makes it easier to distinguish the effect of gating
STEM_72_sm_SuppFig3.tif623KSupporting Information Figure S3. Validation of quantitative RT-PCR data for (A) Cd45-; (B) C-kit-; (C) Gata4/ Nkx2.5-; and (D) Cd31, Cdh5, miR-126 qRT-PCR analyses as indicated in the main article. The number and choice of control genes used for normalization of qRT-PCR data have large impacts on variations in relative mRNA levels misleading to erroneous conclusions1. Accurate normalization of relative qRT-PCR data was therefore performed using the geNorm1 platform to calculate the most stably expressed and optimal number of control genes that should be used for normalization of qRT-PCR data. 4-6 endogenous control (EC) genes were examined in each setup (A-D), and ECs were ranked according to average expression stability (M), where M is defined as the average pairwise variation of a certain gene with all other control genes. Lower M values reflect higher expression stability. The optimal number of ECs to be used was determined as the pairwise variation (V) of normalization factors using stepwise inclusion of the most stable control gene. The quality of the reference genes used in each setup was determined by the qBASE2 program following normalization. (E) Relative quantitative RT-PCR of Gata4 in neonatal rat hearts and explant migrating cells. The mean of three independent experiments is shown.
STEM_72_sm_SuppFig4.tif851KSupporting Information Figure S4. Non-sorted neonatal rat CSs after 7 days of culture. Phase pictures of giant multi-nuclear (arrows) cells with a ring-like structure in the outermost cytoplasm were observed. Cells with a similar morphology did not appear in CSs generated from CD45- explant migrating cells indicating that they arise from CD45+ EMCs. (Scalebars represent 100μm).
STEM_72_sm_SuppMov5.mov916KSupporting Information Movie S5: Beating neonatal rat heart explants.
STEM_72_sm_SuppMov6.mov975KSupporting Information Movie S6: Beating neonatal CS from unfiltered explant migrating cells.
STEM_72_sm_SuppMov7.mov1113KSupporting Information Movie S7: Beating of neonatal rat heart tissue fragments (>40 μm) that were indistinguishable from CSs and found in the explant derived cell harvests.
STEM_72_sm_SuppMov8.mov884KSupporting Information Movie S8 Beating of neonatal mouse heart tissue fragments (>40 μm) that were indistinguishable from CSs and found in the explant derived cell harvests.
STEM_72_sm_SuppMat.doc63KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.