|Home | About | Journals | Submit | Contact Us | Français|
Adipose stromal cells and dissociated brown adipose tissue have been shown to generate cardiomyocyte-like cells. However, it is not clear whether white mature adipocytes have the same potential, even though a close relationship has been found between adipocytes and vascular endothelial cells, another cardiovascular cell type. The objective of this study was to examine if white adipocytes would be able to supply cardiomyocytes.
We prepared a highly purified population of lipid-filled adipocytes from mice, 6–7 weeks of age. When allowed to lose lipids, the adipocytes assumed a fibroblast-like morphology, so-called dedifferentiated fat (DFAT) cells. Subsequently, 10–15% of the DFAT cells spontaneously differentiated into cardiomyocyte-like cells, in which the cardiomyocyte phenotype was identified by morphological observations, expression of cardiomyocyte-specific markers, and immunocytochemical staining. In addition, electrophysiological studies revealed pacemaker activity in these cells, and functional studies showed that a β-adrenergic agonist stimulated the beating rate, whereas a β-antagonist reduced it. In vitro treatment of newly isolated adipocytes or DFAT cells with inhibitors of bone morphogenetic proteins (BMP) and Wnt signalling promoted the development of the cardiomyocyte phenotype as determined by the number or beating colonies of cardiomyocyte-like cells and expression of troponin I, a cardiomyocyte-specific marker. Inhibition of BMP was most effective in promoting the cardiomyocyte phenotype in adipocytes, whereas Wnt-inhibition was most effective in DFAT cells.
White mature adipocytes can differentiate into cardiomyocyte-like cells, suggesting a link between adipocyte and cardiomyocyte differentiation.
Previous studies have suggested a close relationship between white adipocytes and cardiovascular cells. Planat-Benard et al.1 provided evidence that adipocytes and endothelial cells have a common progenitor cells, and Tang et al.2 recently showed that adipocyte progenitor cells reside in the mural cell compartment of the adipose vasculature. Furthermore, adipogenic differentiation or transdifferentiation of cardiomyocytes have been suggested in order to explain the progressive fibrofatty replacement of the myocardium in arrhythmogenic right ventricular dysplasia.3,4 However, it is not clear whether white adipocytes might supply functional cardiomyocytes similar to what has been reported for dissociated brown adipose tissue and the adipose stromal fraction.5–7
Mature lipid-filled adipocytes have been shown to possess the ability to lose the fat and revert to a more primitive, proliferative phenotype, so-called dedifferentiated (DFAT) cells, when subjected to specific culture conditions in vitro including ceiling culture.8–10 Recent studies suggest that these DFAT cells have lost the expression of adipocyte-specific markers, but have gained multi-potent characteristics and are able to differentiate into multiple mesenchymal cell lineages under appropriate culture conditions.8,9 It is thus possible that such adipocyte-derived multi-potential cells could be a source of cardiomyocytes.
Bone morphogenetic proteins (BMP) and Wnt/β-catenin cell signalling are essential in both adipogenic and cardiomyogenic differentiation.11–14 Early exposure of adipocyte progenitors to BMP-4 is critical for correct differentiation,15,16 and activators of Wnt/β-catenin signalling modulate the balance between adipogenic and other cell differentiation.14 However, transient inhibition of BMP-signalling enhances cardiomyocyte differentiation in embryonic mouse stem cells,17,18 and the Wnt signalling pathway appears to have a biphasic role in cardiac specification.13,19
In this study, we show that white mature adipocytes and DFAT cells can act as sources of spontaneously contracting cardiomyocytes in vitro, and that this process is enhanced by inhibition of BMP and Wnt signalling. The results support a possible link between adipocyte and cardiomyocyte differentiation that might be of importance for pathology and cardiac regeneration.
Subcutaneous adipose tissue was removed from C57BL/6 mice aged 6–7 weeks, and lipid-filled mature adipocytes were prepared and cultured in ceiling cultures as previously described8,20 (see Supplementary material online for details). As before, these cells were referred to as dedifferentiated adipocytes (DFAT cells).8 Adipose stromal cells (ASC) were prepared and cultured as previously described.8 The investigation conformed to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publications No. 85-23, revised 1996), and had been approved by the Institutional Review Board of the University of California, Los Angeles.
For quantification of beating cardiomyocyte-like cells, adipocytes (1 × 105 cells/well) or DFAT cells (1 × 104 cells/well) were seeded in 24-well plates. The adipocytes initially floated on the medium, then sank, and attached to the bottom. Previous studies have shown that these cells are indistinguishable from cells grown in ceiling culture.21 BMP-4, Noggin, Dickkopf (Dkk)-1, or Wnt5a (all from R&D Systems, Minneapolis, MN, USA) were added when the cells were seeded, and the treatment media were changed after 5 days. Beating cell colonies were counted in all wells when untreated control colonies had started to beat.
The purity of the isolated adipocytes was assessed by fluorescence-activated cell sorting (FACS) using the lipophilic fluorescent dye Nile red as previously described.22 For characterization of the phenotypes of the DFAT cells and the ASC, FACS analysis was performed after the first passage as previously described8 using fluorescein isothiocyanate (FITC)-, phycoerythrin (PE)-, or Alexa Fluor 488 (AF-488)-conjugated monoclonal rat anti-mouse antibodies against c-Kit (CD117), Sca-1 (Ly-6A/E), CD34, CD45, or CD11a (all 1:200; BD Pharmingen and eBioscience, San Diego, CA, USA) (see Supplementary material online for details).
Action potentials (APs) were recorded in isolated contracting adipocyte-derived cardiomyocytes using the current-clamp mode of the whole cell patch-clamp technique as previously described23 (see Supplementary material online for details).
For Ca2+ imaging, the adipocyte-derived cardiomyocytes were pre-labelled with the Ca2+ indicator fluo-3/AM (10–30 µmol/L, Molecular Probes, Eugene, OR, USA), 0.5 mM probenecid, and 0.02% (wt/wt) Pluronic F-127 (Molecular Probes)23 prior to obtaining fluorescence images (see Supplementary material online for details). In some experiments, Ca2+ stores in the sarcoplasmic reticulum (SR) were depleted using thapsigargin (Tg) and ryanodine (Ry).
For pharmacological treatment of the adipocyte-derived cardiomyocytes, the baseline contraction rate was recorded before addition of adrenergic reagents. Chronotropic responses were recorded after addition of isoproterenol, a β-adrenergic agonist (Sigma-Aldrich), alone or preceded by propanolol, a non-selective β-antagonist (Sigma-Aldrich).
Cells grown in chamber slides were fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, blocked with 1% goat serum and 1% BSA in TBS, and incubated over night at 4°C with anti-sarcomeric actin (Sr) (1:500, Sigma-Aldrich), anti-GATA4 (1:200, Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-Nkx2.5 (1:300, R&D Systems), and anti-connexin 43 (1:500, Sigma-Aldrich). The following day, the cells were washed three times with PBS, incubated for 1 h at room temperature with secondary antibodies, washed three times again with PBS, and visualized by confocal or regular fluorescence microscopy.
Data were analysed for statistical significance by ANOVA with Tukey's multiple comparisons test. The analyses were performed using the InStat Instant Biostatistics program version 3.0 (GraphPad Software, San Diego, CA, USA). All experiments were repeated a minimum of three times.
The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.
To assess the homogeneity of the isolated adipocytes and to determine the fraction that represented lipid-filled adipocytes, we stained the cells with the Nile red and examined them by FACS. The results showed that the adipocyte isolate was highly homogenous with 99.9% Nile red-positive cells (Figure 1A). DFAT cells (one passage after loss of lipid) were also stained for comparison and were Nile red negative (Figure 1A). Previous studies have shown the occurrence of multi-nucleated tissue adipocytes,25–27 with about 3–5% bi- or multi-nucleated adipocytes.25,26 To determine the number of nuclei in the isolated adipocytes, we stained the nuclei with DAPI and counted the number of mono-, bi-, and multi-nucleated adipocytes using a fluorescence microscope. The results showed 94.1 ± 3.4% mononucleated, 3.8 ± 2.1% binucleated, and 2.0 ± 1.4% multi-nucleated adipocytes (total of five experiments, with 2000 adipocytes counted in each) (Figure 1B), consistent with the previously reported range. Variations in the percentage of bi- or multi-nucleated adipocytes had no detectable effect on the results in subsequent experiments. We monitored adipocytes in culture over 7 days using Nile red staining, which showed a gradual release of lipid (day 3–5, Figure 1C) followed by proliferation of spindle-shaped DFAT cells (days 5–7, Figure 1C). These changes may alter the susceptibility of the cells to various environmental factors.
Nile red and nuclear staining of isolated adipocytes. (A) Isolated adipocytes were stained with Nile red, analysed by FACS, and compared with control-stained adipocytes. DFAT cells (one passage after loss of lipid) were also stained and were Nile red-negative. ...
We examined the expression of cell surface antigens of DFAT cells (passage 1) by flow cytometry. The results showed that c-kit+, Sca-1+, and Sca-1+/c-kit+ cells represented 5.5 ± 0.2, 86.8 ± 3.3, and 2.4 ± 0.1%, respectively, of total cells counted (Figure 2A). The expression of Sca-1 was maintained by the DFAT cells for at least nine passages, whereas c-kit expression was undetectable after four passages (data not shown). The DFAT cells were largely negative for CD11a (leukocytes), CD34 (endothelial progenitors), CD45 (hematopoietic progenitors), and CD133 (neuroepithelial progenitors), 0.01, 0.02, 0.01, and 0.04%, respectively (Figure 2A). Similar to previous reports,8 the DFAT cells were also negative for CD31 (PECAM-1, endothelial cells), CD11b (monocytes), and α-smooth muscle actin (α-SMA, smooth muscle cells) (data not shown). As previously reported,8 expression of CD34, CD45, CD11b, and CD31 ranged between 5 and 15% in the ASC, and was higher than in DFAT cells (data not shown). However, 2.2 ± 0.1% of the DFAT cells were positive for CD90 (hematopoietic progenitors), whereas the ASC were negative for CD90. Together, the results suggested that the DFAT cells may be a more homogeneous cell population than ASC, with higher expression of the stem cell markers c-kit, Sca-1, and CD90.
Expression of cell surface antigens in DFAT cells. DFAT cells (A) and ASC (B) were harvested after passage 1, stained with PE-conjugated antibodies to Sca-1, CD45, or CD90, or FITC-conjugated antibodies to c-kit, CD34, CD11a, or CD133, and analysed by ...
After about 7 days in ceiling culture, the isolated adipocytes had adhered and lost visible fat droplets, and the cell layer consisted mostly of fibroblast-like cells (approximately 87%) (Figure 3). Other morphologies started to emerge after about 10 days, including groups of small tubular cells associated with rounded cells (approximately 10%) (Figure 3), and clusters of cells re-accumulating lipid droplets (approximately 3%) (data not shown). After 10–14 days, myotube-like structures appeared, which grew in size, but were still surrounded by some of the rounded cells. After 14 days, some myotube-like structures independently exhibited spontaneous contractile activity (Figure 3 and Supplementary material online, Videos 1 and 2). In addition, contractile activity was observed in the rounded cells (Supplementary material online, Video 3). After 15–21 days, cohesive groups of myotube-like structures were seen, with branching fibres and tight connections. After 21 days, entire networks of cells were beating, in part in a synchronous fashion, and were reminiscent of tissue-like structures (Figure 3 and Supplementary material online, Videos 4 and 5). About 10% of the cells, as estimated prior to the formation of networks, exhibited spontaneous contractions that could be maintained for up to 2 months.
Morphology and cardiomyocyte-specific lineage markers of cardiomyocyte-like cells derived from adipocytes. (Top panels) After about 7 days of culture, spindle-shaped cells were observed; after 10 days, myotube-like cells appeared with associated rounded ...
To define the molecular phenotype of the contracting cells, expression of several cardiac-specific genes was assessed using PCR, and compared with the expression in mouse adipocytes (AD, starting material), DFAT cells (after the first passage), and cardiomyocytes (CM, positive control). The results showed that the adipocytes and the DFAT cells expressed no or low levels of cardiac markers; only the transcription factors MEF-2C and GATA4 were detected in the DFAT cells (Figure 3, bottom). However, the contracting cell populations (adipocyte-derived cardiomyocytes, ADCM) expressed cardiac-specific mRNA including the transcription factors Nkx-2.5, GATA4 and MEF-2C, the structural proteins α- and β-MHC, MLC-2a and -2v, and cardiac α-actin (Figure 3, bottom). Skeletal muscle phenotype was excluded because of the lack of MyoD expression (Figure 3, bottom). β-Actin was used as RNA control. Together the data supported the cardiomyocyte-like phenotype of the contracting cells.
To further characterize protein expression in the beating areas, we performed immunocytochemistry. Positive staining was detected with antibodies to cardiac proteins including cardiac sarcomeric actin (Sr), GATA4, cardiac troponin I, Nkx2.5, and connexin 43 (Figure 4). Notably, staining of the contractile sarcomeric actin and troponin I showed striated patterns (Figure 4). As expected, the junctional protein Connexin 43 was detected around the edges of the cells (Figure 4), and Nkx2.5 was detected in the cell nuclei. In contrast, we were unable to detect staining for MyoD and α-SMA in the contracting areas (data not shown). Interestingly, staining for sarcomeric actin and troponin I was also seen in the rounded cells associated with the myotube-like structures (Figure 4), suggesting that these cells may be precursor cells. Together the data further supported a cardiomyocyte-like phenotype, with well-organized structures and electrical junctions.
Immunocytochemical analysis of cardiomyocyte-specific markers. (Left panels) Cells were stained for sarcomeric actin (Sr, green), or double-stained for sarcomeric actin and GATA4 (red). (Middle panels) Cells were stained for Nkx2.5 (green), or connexin ...
To estimate the number of cells with cardiomyocyte-like characteristics, we stained cell cultures with beating cells in 12-well plates for sarcomeric actin or troponin I (Figure 5). Alternatively, we double-stained for GATA4 together with sarcomeric actin or troponin I (Figure 5). The nuclei were stained with DAPI. The percentage of stained cells of total adherent cells (estimated to be 1 × 105 per well) was determined; a total of six experiments were performed. The results showed that approximately 10–15% of the cells expressed cardiomyocyte-specific proteins (Figure 5, right panel).
Immunocytochemical analysis of cardiomyocyte-specific markers. (Left panels) Cells were stained for sarcomeric actin (Sr, green) or troponin I (green). (Middle panels) Cells were double-stained for GATA4 (red) together with sarcomeric actin or troponin ...
To characterize the contractile activity in the cardiomyocyte-like cells, we first examined intracellular Ca2+ transients in the cells using confocal microscopy. The studies were performed after 14–20 days of culture, in single cells that had just started to beat. We obtained line-scan images and spatially averaged fluorescence transients from a single cardiomyocyte-like cell loaded with the Ca2+ indicator fluo-3/AM (Figure 6A). Transients were evoked by external pacing with platinum electrodes at a frequency of 1 Hz. The spatially synchronous onset of each transient and the rapid upstroke of spatially averaged fluorescence transients suggested triggered Ca2+ release from intracellular Ca2+ stores, most likely the SR. Spontaneous transients had the same appearance (data not shown). In cardiac cells, SR Ca2+ is released through a cluster of ryanodine receptors. To determine whether the same mechanism occurs in adipocyte-derived contracting cells, we applied thapsigargin (Tg, 0.2 µM) and ryanodine (Ry, 10 µM) to deplete the SR Ca2+ stores (Figure 6B, representative results from single cells). This resulted in loss of Ca2+ transients and contractions, suggesting that the organization of the SR is functionally developed in the adipocyte-derived cardiomyocytes.
Action potentials and calcium transients from spontaneously beating cardiomyocyte-like cells derived from adipocytes. (A) Line-scan image of fluorescence transients from a cardiomyocyte-like cell loaded with the Ca2+ indicator fluo-3/AM. Transients were ...
We then performed electrophysiological studies. APs were recorded using the whole cell configuration of the patch clamp technique in response to 5 ms depolarizing pulses in current-clamp mode. The cardiomyocyte-like cells exhibited both triggered and spontaneous APs (Figure 6C). Spontaneous electrical activity resembled cardiac pacemaker activity, with slow diastolic depolarization from −60 to −40 mV followed by rapid upstroke to a peak voltage of −5 mV (Figure 6C). The spontaneous depolarization rate was about 60 per minute.
In separate groups of cells, we also examined the response of contracting cells to β-adrenergic pharmacological reagents known to influence heart rate. The β-agonist isoproterenol (0.5–1.5 µM) induced a significant increase in contraction rate (Figure 6D). However, pre-incubation with propranolol (5–20 µM), a non-selective β-adrenergic antagonist, reversed the acceleration induced by isoproterenol (1.5 µM) (Figure 6D). These findings suggested that the contracting cardiomyocyte-like cells had the functional characteristics of cardiomyocytes in vitro.
Previous studies suggested that inhibition of BMP signalling induced cardiomyocyte differentiation in embryonic stem cells,17 whereas inhibition of Wnt/β-catenin signalling blocked cardiomyocyte differentiation.19 In addition, non-canonical Wnt signalling was shown to promote cardiomyocyte differentiation in ASC.7 To determine the effect of inhibition of BMP and Wnt signalling, and the effect of non-canonical Wnt signalling on cardiomyocyte differentiation in the adipocyte-derived cells, we performed two sets of experiments. In the first set, adipocytes were subjected to treatment starting right after isolation, and in the second set, DFAT cells were treated after passage 1.
Adipocytes or DFAT cells were seeded in 24-well plates in control medium, or with BMP-4 (100 ng/mL), Noggin, a BMP-inhibitor (100 ng/mL), Dkk-1, a Wnt-inhibitor (50 ng/mL), or the non-canonical Wnt5a (50 ng/mL). The cells were allowed to differentiate until the cells in control medium developed well-defined colonies of spontaneously beating cells. On the day that the control cells started to beat, the beating colonies were counted in all wells, and expression of troponin I and Nkx2.5 was determined by real-time PCR. In both adipocytes and DFAT cells, Noggin and Dkk-1 significantly enhanced the number of beating colonies (Figure 7A) and cardiomyocyte differentiation as determined by expression of troponin I and Nkx2.5 (Figure 7B), whereas BMP-4 and Wnt5a had no significant effect. However, the effect of Noggin was greater than that of Dkk-1 when added directly to the isolated adipocytes, but less than Dkk-1 when added to the DFAT cells (Figure 7A and B). To confirm the increase in cardiomyocyte-like cells after treatment with Noggin and Dkk1, we also performed immunofluorescence for sarcomeric actin. The results showed an increase in of sarcomeric actin staining in the Noggin- or Dkk1-treated cells when compared with control cells (Figure 7C, different secondary antibodies used for treated adipocytes and DFAT cells, respectively). Cells treated with BMP-4 or Wnt5a, however, were indistinguishable from control cells (data not shown). Together, these experiments suggested that the differentiation of the cardiomyocyte-like cells was susceptible to growth factors, but that the sensitivity changed as the adipocytes lost the lipid as suggested in Figure 1C.
Inhibition of BMP and Wnt signalling enhanced cardiomyocyte differentiation in adipocyte-derived cardiomyocyte-like cells. Newly isolated adipocytes or DFAT cells (passage 1) were seeded in 24-well plates, and treated with BMP-4 (100 ng/mL), Noggin (100 ...
In this study, we show, for the first time, that white mature adipocytes and DFAT cells from mice can serve as sources of cardiomyocyte-like cells, which demonstrate spontaneous contractile activity and expression of cardiomyocyte-specific markers. In addition, the emergence of the cardiomyocyte phenotype is promoted by inhibition of the BMP- and Wnt-signalling. The results suggest a potential link between adipocyte and cardiomyocyte differentiation, possibly allowing mature fat cells to be a new cell source for cardiomyocytes. Other multi-potent cell types that have also been shown to generate cardiomyocytes or to improve cardiac function in murine myocardial injury models include bone-marrow-derived stem cells,28 ASC from white and brown adipose tissue,5,6 and murine and human cardiac stem cells.29,30
There are several advantages in working with the DFAT cells when studying cardiomyocyte differentiation or as a potential cell source of multi-potent cells. Similar to the ACS, the DFAT cells are readily available in large quantities.8,31 The need of propagation to generate sufficient cell numbers is therefore less, and the risk of expanding contaminating cell populations is also less. Another advantage of the DFAT cells is the simple medium used for differentiation. The DFAT cells spontaneously underwent cardiomyogenic differentiation in 20% FBS, whereas the ACS required a semisolid methyl cellulose medium containing interleukin (IL)-3, IL-6, and stem cell factor for successful cardiomyocyte differentiation.6 Furthermore, we did not need to add reagents such as 5-azacytidine,32 reprogram with lentiviral-mediated transduction of transcription factors,33 or coculture the DFAT cells with other cell types.34 It is not clear what triggered the DFAT cells to undergo cardiomyocyte differentiation under our conditions, though it is possible that the origins of the FBS and the antibiotics were important. The change in differentiation is also likely to be promoted by the removal of matrix components during adipocyte isolation, and a clear understanding of what matrix components promote vs. suppress cardiomyocyte differentiation would be valuable when attempting to enhance cardiac function in e.g. various types of cardiomyopathy.35 Species and age are other important factors. We used relatively young mice, but we have seen cardiomyocyte-like cells also in cultures of adipocytes from mice aged more than 12 months (unpublished observations). Further studies are needed to determine if human adipocytes can act as a source of cardiomyocytes, and if so, how age affects that ability.
Our studies suggest that the cardiomyocyte-like cells are derived from the white, mature adipocytes after loss of lipid. However, we cannot rule out that there are stem cell-like cells associated with the adipocytes that may also contribute. Although bi- and multi-nucleated adipocytes have been described by others at similar or higher levels than ours in adipose tissue,25,26 it is possible that these cells in part represent adipocytes with associated stem cells.
The DFAT cell population appeared to be more homogenous than the ASC. The determination of surface markers showed that the DFAT cells had a higher percentage of cells expressing c-kit and Sca-1 than the ACS (5.5% and 86.8% vs. 1.6% and 79.6%, respectively), which have both been associated with cardiomyocyte differentiation in several multi-potent cell types.7,29,30 The ability of the DFAT cells to achieve the cardiomyocyte phenotype may be more associated with expression of c-kit than Sca-1, since cardiomyocyte differentiation was largely absent after four passages similar to c-kit expression (data not shown). Very few of the DFAT cells expressed CD11a, CD11b, CD34, CD 31, CD45, or α-SMA, whereas 5–15% of the ASC were positive for the listed markers.8 Although cardiomyocyte progenitors from brown adipose tissue were positive for CD133,5 only 0.04% of the DFAT cells stained for CD133, suggesting a difference in the respective progenitors. Altogether the DFAT cells may provide a more pure source of stem cells for cardiomyocyte generation.
About 10 days after placing the adipocytes in culture, we observed groups of small tubular cells associated with rounded cells (Figure 3), which both demonstrated independent beating and staining for cardiac markers. Although other investigators have reported that the DFAT cells express skeletal muscle markers and fuse to form multi-nucleated cells after treatment with 5-azacytidine,36 we were unable to detect myogenic markers in our non-induced DFAT cultures.
The adipocyte-derived cells demonstrated electrical activity and intracellular Ca2+ transients that typify cardiomyocytes. The APs that were recorded in this study resembled strongly mouse atrial APs, which have minimal plateau compared with humans, rabbits, and guinea pigs because of the very high transient outward current (ITO) carried by potassium ions.37,38 Calcium release is only dependent on the activation of Ca2+ channels and the rapid repolarization rate, which cannot be appreciated from AP recordings. Ca2+ transients are typically longer than the AP durations.37,38 The diastolic depolarization we observed was consistent with pacemaker activity. Thus, the electrophysiological phenotype is clearly a cardiac atrial cell with pacemaking activity. Consistent with the electrophysiological data, transients in response to electrical stimulation were also typical of mouse myocytes.39 Those functional studies strongly support the conclusion that these adipocyte-derived cells are cardiomyocytes.
Our study showed that inhibition of BMP and Wnt promote cardiomyocyte differentiation in the adipocyte-derived cells, consistent with findings in other cell types.17,19 We found that Noggin was more efficient in promoting cardiomyocyte differentiation when added to adipocytes, whereas Dkk-1 was more efficient in the DFAT cells. This appears to be consistent with studies that demonstrated that early adipogenesis depends on BMP-4 activity.15,16 Since the adipocyte-derived cells have demonstrated an ability to re-accumulate cytosolic lipid,8,20 it is possible that BMP-inhibition prevents this, instead promoting differentiation along other lineages. Once established, the DFAT cells may represent a differentiation stage further removed from the adipocytes. The loss of lipids (as monitored over 7 days, Figure 1C) was associated with an increase in proliferation, and presumably also alterations in the sensitivity to growth factors.
Since normal cardiac tissue includes fat, especially in antero-apical region,40 it is possible that intracardiac or epicardial fat may be a reservoir of multi-potent cells that could be recruited in case of tissue damage for regeneration or scar formation. The clinical relationship between cardiac function and obesity, however, remains unclear. Some studies have suggested that being overweight is associated with better outcome in heart failure,41 and that patients with congestive heart failure have less pericardial fat mass than healthy subjects,42 whereas other studies have shown that obesity is associated with heart failure,43 and increased lipid accumulation in cardiomyocytes may lead to apoptosis.44
In summary, we show that white mature adipocytes and DFAT cells can act as sources of spontaneously contracting cardiomyocytes in vitro, and that this process is enhanced by inhibition of BMP and Wnt signalling.
Conflict of interest: none declared.
This work was supported by National Institutes of Health (HL30568 and HL81397 to K.I.B., HL70828 to J.I.G.), and the American Heart Association (Western Affiliate) (0755031Y to K.I.B.).