Fusion of transplanted stem cells with recipient cardiomyocytes has been observed in murine [13
] and porcine model systems [49
]. But since these first observations, few have sought to unravel the mechanisms that govern stem cell fusion or to study the implications of cell fusion for stem cell programming. Lack of study reflects the overwhelming opinion that cell fusion occurs too infrequently to be of relevance for stem cell programming and, by corollary, for tissue repair. However, this opinion fails to appreciate the possibility that (1) detection methodologies may be insufficient to accurately gauge the contribution of cell fusion following stem cell transplantation and/or (2) that we might be able to control or increase the frequency of cell fusion to more effectively induce programming of stem cells following transplantation. We have begun to explore this second possibility by co-opting the well-described fusion machinery of viruses. We find that mesenchymal stem cells modified to express viral fusogen VSV-G are more apt to fuse with cardiomyocytes in a pH-dependent manner. vMSC-CM fusion products formed in this way are prone to adopt cardiomyocyte phenotype and morphology. In addition, vMSCs delivered to the myocardium of mice following infarction can fuse with resident cardiac cell types at rates much higher than previous reports of spontaneous fusion [13
] and are more apt to fuse at the site of infarction than in the healthy myocardium.
Increasing the frequency of MSC-CM cell fusion will aid in the study of cell fusion in vitro
and may improve the therapeutic benefit of MSCs in vivo.
One way that the therapeutic benefit may be improved is via induction of programming of MSCs to a cardiomyocyte fate. Differentiation of MSCs into CMs can be initiated in vitro
via soluble factors including 5-azacytidine [62
] or with exposure to insoluble factors including laminin [65
]. However, functional differentiation of MSCs to cardiomyocytes has only been accomplished to date via cell fusion with mature cardiomyocytes. This result has been demonstrated in vitro
] and in vivo
wherein MSC-CM fusion products take on a cardiomyocyte morphology, express cardiomyocyte markers, and couple to adjacent cardiomyocytes [60
]. Here we find that when MSC-CM fusion is induced with viral fusogens, the CM fusion partner is dominant in that the majority of fusion products (regardless of medium type) adopt a CM-like morphology and maintain expression of MF20 and lose CD105. These data further support the exciting possibility that induction of fusion with viral fusogens could enhance MSC programming to a CM fate in vivo
. Of note, the CMs utilized here are HL-1 CMs. This cell line was used to enable large-scale and long-term studies. However, the heterogeneity and immortal nature of these cells may account for the seeming dominance of the CM phenotype and future studies will utilize primary fetal cardiomyocytes or induced pluripotent stem cell-derived cardiomyocytes.
Our results suggest that the differentiation of MSCs to a CM fate can be promoted by cell-cell fusion. However, in certain circumstances in vitro
, MSC-CM fusion products can reenter the cell cycle and proliferate suggesting cell-cell fusion can also promote reprogramming of the CM [67
]. Proliferation of fusion products may be as advantageous for cardiac tissue repair as differentiation of functional cell types since more cells could be produced to replace lost cells. In addition, recent evidence has demonstrated that MSC-CM fusion includes mitochondrial exchange, which is essential for somatic reprogramming [69
]. Understanding cell-cell fusion in conjunction with mitochondrial preservation may provide alternate, simple, and direct mechanisms to rescue cells following ischemia-induced damage. There is evidence indicating that the fusion product's proliferative capacity is regulated by the stem cell while the developmental direction is dictated by the somatic cell [70
], and the combination of both outcomes presented herein are means to repopulate the myocardium for functional improvement.
While we have utilized vMSCs to both understand and exploit the physiological role of MSC-CM fusion, induction of fusion of another stem cell, progenitor, or even mature cell types may augment our ability to repopulate and repair the damaged myocardium [59
]. In the case of mature or progenitor cell transplantation, the induction of fusion may be less beneficial from a differentiation standpoint and more beneficial from an engraftment or retention standpoint. One of the primary challenges for stem cell delivery is the ~90% cell loss after transplantation [80
] that has prompted the development of new methods to deliver and maintain cells in the heart [48
]. This is particularly problematic for cardiac therapy as the heart is mechanically active, rapidly flushing cells from the intended target region. If stem cells transiently express a viral fusogen, they might rapidly adhere and so be maintained long term in the heart. The added advantage of pH sensitive fusogens, such as VSV-G, is the ability to control activity such that cells only fuse at pH lower than 6.5. This has major implications for inducing temporally (the window during ischemia) and spatially (the ischemic region) regulated fusion in vivo
. In fact, vMSCs delivered to the heart were found in the patch and in damaged myocardium fused with mouse cells. The ability for VSV-G to induce fusion in the patch may be due to close proximity to the ischemic region, causing the environment to be more acidic or by the remodeling of the collagen patch [48
]. Collagen remodeling has been shown to occur via MSC secretion of matrix metalloproteinases (matrixins), serine proteases, and cysteine proteases [85
]. While matrixins are active at neutral pH, serine and cysteine proteases are active at acidic pH, indirectly demonstrating cells are able to make the microenvironment acidic [86
]. Taken together, the induction of cell fusion in the heart could exert functional benefit via multiple mechanisms.
A primary limitation of this approach is introducing viral machinery to an already damaged recipient. The entire virion, VSV, is known to be immunogenic and, at high enough concentrations, is lethal to mice [87
]. Purified VSV-G or VSV-G reconstituted in lipid bilayers administered to in vitro
cell culture is mitogenic (>0.8μ
]. Interestingly, if the lipid concentrations were increased, while VSV-G concentration was held constant, the mitogenicity decreased, suggesting that the spacing of VSV-G in the membrane plays a role. Confirming the importance of VSV-G arrangement, Ochsenbein et al. demonstrated that 1,000 times more antibody is produced by C57BL/6 mice against highly organized VSV-G on the nucleocapsid of intact VSV versus poorly organized VSV-G in micelles [89
]. The amount of viral proteins we delivered (based on the mass of the protein [39
], the proteins expressed per cell combined with the number of cells delivered) is 7 orders of magnitude below the reported amount to elicit an immune response [88
] and we express only the fusogen and not the entire virion. Even if methods were developed to increase expression levels per cell and/or in combination with high numbers of cells, spacing could be evaluated to avoid immune responses. However, based on the reported concentration required to elicit a response, delivery of vMSCs as prepared in this study would not trigger a response.
While vMSCs may not be immunogenic, transfection itself may cause adverse genetic effects. For instance, stable transfection with most viral systems causes integration of the gene at a random site in the genome [90
]. When mutagenesis occurs, integration may occur at a site that interferes with cells ability to regulate itself, resulting in deregulation of proliferation and tumorigenesis [93
]. In addition to experimental evidence of malignancy, this has been seen clinically ([95
], reviewed in [97
]). Here, transfection is largely transient and only rarely integrates into the genome. Clinical use would require further safeguards, perhaps including liposomal delivery of the protein.