The origin of the cancer stem cell: current controversies and new insights

Abstract

Most tumours are derived from a single cell that is transformed into a cancer-initiating cell (cancer stem cell) that has the capacity to proliferate and form tumours in vivo. However, the origin of the cancer stem cell remains elusive. Interestingly, during development and tissue repair the fusion of genetic and cytoplasmic material between cells of different origins is an important physiological process. Such cell fusion and horizontal gene-transfer events have also been linked to several fundamental features of cancer and could be important in the development of the cancer stem cell.

Main

For decades, tumour initiation and development has been regarded as a multistep process that is reflected by the progressive genetic alterations that drive the transformation of normal human cells into highly malignant derivates¹. Current evidence indicates that most cancers arise from a single cell that has undergone malignant transformation driven by frequent genetic mutations². Such events are thought to be followed by a clonal selection of variant cells that show increasingly aggressive behaviour. These cells are characterized by their ability to proliferate in defiance of normal growth-regulating mechanisms and by their ability to invade and destroy normal tissues^3,4. At present, there is an increasing body of evidence indicating that the cellular and molecular events leading to tumour initiation are orchestrated by cancer stem cell-like cells^5,6,7,8,9,10. This article focuses on current scientific controversies related to the establishment of the cancer stem cell — in particular, how cell fusion and horizontal gene-transfer phenomena might contribute to the evolution of cancer and the biological manifestations that occur as a result of such events.

Aneuploidy

In addition to specific gene mutations that occur in various types of cancer, cancer can also occur as a result of chromosomal gain, loss and/or rearrangement. The idea that cancer is a result of what is now defined as aneuploidy was suggested as early as 1890 (Ref. 11) and formulated in detail in 1914 by Boveri¹². In recent years, aneuploidy has taken a back seat to specific mutations in tumour-suppressor genes and proto-oncogenes. Indeed, it is still commonly thought that aneuploidy occurs as a late-stage effect rather than as a cause of cancer development. This might not always be true, as carcinogens — such as asbestos and arsenic — initially do not cause gene mutations, yet they produce aneuploid lesions^13,14. Furthermore, normal cells exposed to chemical carcinogens can become aneuploid long before they show signs of being cancerous^13,14. Therefore, it is possible that gains and losses of whole chromosomes can upset the checks and balances that regulate normal growth control. On the other hand, many normal cells, both in vitro and in vivo, can become cancerous after the right combination of oncogenes are introduced^15,16,17. However, only a fraction of these cells will give rise to cancer, implying that other as yet unidentified factors might also be involved in tumour initiation. One could conclude, therefore, that both mutations and chromosomal derangements are important in the initial stages of tumour development, and that both mechanisms might be involved in the establishment of the cancer stem cell.

Cancer stem cells

Although many tumours contain cells that display stem cell-like features, the identity of the normal cells that acquire the first genetic 'hits' that lead to tumour initiation has remained elusive¹⁸.

The term 'cancer stem cell' is an operational term defined as a cancer cell that has the ability to self-renew; dividing to give rise to another malignant stem cell and a cell that gives rise to the phenotypically diverse tumour cell population. Stem cells in different tissues vary with respect to their intrinsic abilities to self-renew and to differentiate into particular cell types^19,20.

Most cancers comprise a heterogeneous population of cells with differences in their proliferative potential as well as their ability to reconstitute the tumour on transplantation. Evidence for cancer stem cells was first documented in haematological malignancies where only a small subset of cancer cells were capable of forming new tumours. Using in vivo models where human acute myeloid leukaemia (AML) cells were transplanted into immunodeficient mice, a leukaemia-initiating cell was identified that showed strikingly immature features, expressing a CD34⁺/CD38⁻phenotype. By contrast, despite the fact that they showed a leukaemic blast phenotype, CD34⁺/CD38⁺ leukaemic cells were not able to initiate leukaemia in immunodeficient mice in most cases^21,22,23,24.

Tightly regulated self-renewal pathways — for example, expression of the B-lymphoma mouse Moloney leukaemia virus (Mo-MLV) insertion region 1 ( BMI1 ) gene or signalling through the Wnt–β-catenin, sonic hedgehog, Notch and PTEN pathways — indicate close links between cancer stem cells and normal stem cells^25,26. Cancer stem cells have been identified and enriched on the basis of their expression of cell-surface markers^22,27,28. On transplantation, these cells give rise to tumours comprising both new cancer stem cells as well as heterogeneous populations of non-tumorigenic cells, reminiscent of the developmental hierarchy in the tissues from which the tumours arose. To date, cancer stem cell-like cells have been identified in breast^19,20 and central nervous system tumours^6,7,8,9. It is still not clear if the cancer stem cells are derived from true tissue-derived stem cells, bone marrow stem cells or mature cells that have undergone a de-differentiation or a transdifferentiation process. It is also not entirely clear if the cancer stem cell represents one or multiple phenotypes^5,29.

Involvement of the microenvironment

Currently, there are a number of new studies indicating that the developmental limitations of tissue-specific stem cells are regulated by the microenvironment and that host cells under specific conditions, such as tissue injury or infection, might provide specific signals that counteract these restrictions^30,31. In a number of experiments, neural stem cells have been isolated from the central nervous system in mice and labelled with β-galactosidase or green fluorescent protein (GFP) to make them easy to trace together with other cell types. Interestingly, labelled stem cells that were cultured with myoblasts or embryoid bodies (produced from embryonic stem cells) differentiated into β-galactosidase-labelled or GFP-labelled muscle cells^32,33,34. These findings indicate the remarkable plasticity of tissue-specific stem cells that can be regulated by the tissue microenvironment.

In addition, bone marrow-derived stem cells show remarkable plasticity in vitro and in vivo as they can engraft and differentiate into a number of different, organ-specific cells^35,36,37,38. For instance, bone marrow cells have been shown to contribute to myofibroblast and fibroblast populations in tumour stroma in a mouse model of pancreatic insulinoma³⁹. Bone marrow-derived stem cells can also form an important part of the tumour endothelium in mouse tumours⁴⁰. Moreover, recent work has shown that inflammatory cells can also promote tumours through the production of growth-stimulating proteins and DNA-damaging chemicals that can trigger cancer-causing mutations³⁰. Working with mice, Houghton and co-workers found that Helicobacter felis infection caused an influx of bone marrow stem cells, probably reflecting a physiological process attempting to repair the epithelial lining. Interestingly, these visiting cells, and not the cells of the gastric epithelium, gave rise to stomach cancer⁴¹. No evidence of cell fusion was observed in these experiments, as no bi-nucleate cells were seen. Furthermore, flow cytometric DNA measurements showed no differences in DNA content, and female mice transplanted with male transgenic bone marrow showed a single X and Y chromosome⁴¹. However, cell fusion between a gastric epithelial cell and a bone marrow stem cell, which is probably a rare event, might have produced — after a reduction division — a synkaryon (a single nucleated cell) where one of the two X chromosomes might have been lost. Nevertheless, the findings by Houghton et al. indicate a strong environmental contribution of to the initiation of the cancer stem cell (Fig. 1).

Figure 1: Mutations in stem cells and/or progenitor cells might give rise to cancer stem cells.

The cancer stem cell might appear after mutations in specific stem cells or early stem cell progenitors. It is also possible that cancer stem cells can be derived from differentiated cells. There might be numerous factors in the host microenvironment that trigger the initial steps of tumour formation.

Experiments in vitro have shown that differentiated cells can also give rise to cancer. For example, a combination of epidermal growth factor receptor (EGFR) pathway activation and loss of both INK4A (also known as p16) and ARF (also known as p19) tumour-suppressor function provokes a common high-grade glioma phenotype in vivo, both from neural stem cells and differentiated astrocytes¹⁵. Therefore, it is possible that differentiated cells as well as immature precursor cells, under certain conditions, can de-differentiate into cancer stem cells (Fig. 1).

Cancer stem cells and cell fusion

Somatic stem cell plasticity has been challenged by work indicating that transdifferentiation might be caused by cellular fusion between stem cells and pre-existing differentiated cells. Circulating haematopoietic stem cells (HSCs) have been shown to fuse both in vitro and in vivo with several cell-types including hepatocytes, cardiac myocytes, oligodendrocytes and Purkinje cells^{42,43,44,45,46,47,48,49}. Physiological cell fusion is necessary for certain biological processes and is tightly regulated. Defects in such processes might lead to infertility (defects in sperm–egg fusion), certain muscle diseases (defects in myoblast fusion), pre-eclampsia and osteopetrosis (dense bones)^50,51,52,53. Cell fusion is essential for developmental and behavioural patterning in organisms and is therefore tightly regulated during development^54,55 (Box 1).

Cell fusion might lead to both multinucleated and mononucleated cells. The formation of multinucleated giant cells (syncytia) can occur during development, as in the case of bone, muscle and placenta⁵⁵ (Fig. 2a). Fusion of cells of the same or different types can also occur, one example being the fusion between stem cells and differentiated cells to form heterokaryons. The first demonstration of the distinct properties of heterokaryons was shown when Sendai virus was used to fuse murine Erlich ascites cells and human HeLa cells in vitro. The heterokaryons that were formed remained stable over time and had functions and characteristics that were specifically found in each fusion partner⁵⁶. At present it seems likely that heterokaryons have an important function in liver regeneration and in tissue repair⁴⁵ (Fig. 2b).

Figure 2: Fusion of cells under physiological conditions.

a | Syncytia are multinuclear cells formed by cell fusion. The syncytium can have an altered phenotype and might have a role in barrier formation. The fusion partners might be of the same or different cellular lineages. Syncytia can form as a result of viral infections and this process is not tightly controlled, whereas syncytia formation during myogenesis and pregnancy is tightly regulated. b | Cells of the same or different lineages can fuse to form a homokaryon or a heterokaryon, respectively. The heterokaryon might undergo transdifferentiation, eventually displaying another phenotype. c | Cells of the same or different lineage might fuse to form a cell with a single nucleus (synkaryon). The formation of synkaryons is characterized by loss or re-sorting of chromosomes. This process involves a heterokaryon or homokaryon stage. A number of as yet unknown environmental factors might influence the formation of syncytia, heterokaryons and synkaryons.

Synkaryons represent single nucleated cells where formation of the heterokayon is an intermediate step. The formation of a synkaryon is usually characterized by chromosome loss. The classic example of synkaryon formation is the fusion between murine myeloma cells and B-cells from an immunized mouse to form hybridomas (Fig. 2c). Synkaryons are also thought to be important in cellular transdifferentiation⁴⁵.

Increased incidences of cell–cell fusion might be closely related to cancer initiation. Many tumour cells are fusogenic, and fusion between tumour cells and normal somatic cells generates hybrid cells that are often more malignant than the parental cells^57,58. For instance, human stem cells originating from a grafted kidney might migrate to the skin, differentiate or fuse to adopt a keratinocyte phenotype and undergo transformation⁵⁹.

The generation of hybrid cells might also be a consequence of a transient transmembrane exchange of proteins and organelles between cells through the formation of intercellular nanotubular connections, which usually have a diameter of 50–800 nm (Refs 60,61). Such connections have been observed between somatic cells in vitro, but it is still unclear to what extent nanotubular connections contribute to transdifferentiation phenomena in vivo.

In the stem cell model for cancer, disruption of genes involved in the regulation of stem cell self-renewal seems to be important. Several models of cellular heterogeneity in solid tumours have been proposed. One model proposes that different cells in a tumour can proliferate and form new tumours¹. This model implies that the proliferating daughter cells do not necessarily show the same phenotype as the parental cells. Another model implies that only the cancer stem cell has the ability to proliferate extensively and form new tumours. This process indicates self-renewal of the cancer stem cell population⁵.

Some of the cancer cells in a tumour have the ability to differentiate, as do normal stem cells. The fact that such tumour cells might display many normal features points to a limited number of mutations in such cells. An assumption might therefore be that only a few mutations are needed to initiate a cancer. Currently, there is considerable evidence showing that only a subset of cells in tumours possess self-renewal properties^19,20. If only a few mutations are required to initiate a cancer, why would a cell show extensive chromosomal disorder, as is frequently seen in early cancers?

Chromosomal derangement has been observed in tumour-initiating cells isolated from both human medulloblastomas⁶ and glioblastomas⁹. Interestingly, both of these reports describe aneuploid cancer stem cell populations, yet the medulloblastoma stem cells and the glioblastoma stem cells have different karyotypes^6,9, indicating the presence of different aneuploid cancer stem cell populations. This leads to the obvious question: how does tumour aneuploidy arise? The early appearance of cellular aneuploidy in many tumours might imply that acquired aneuploidy occurs quickly. The general view is that a set of crucial mutations affect molecules that control genome stability, which in turn leads to aneuploidy. However, there are numerous reports showing that cells can adopt the phenotype of other cells by spontaneous fusion. For example, mouse bone marrow cells can fuse with embryonic stem cells in culture and will spontaneously adopt the phenotype of the recipient cells that, without detailed genetic analysis, might be interpreted as de-diffentiation or transdifferentiation⁶². Furthermore, neural stem cells have been fused with embryonic stem cells in vitro and the hybrid cells injected into murine blastocysts. The fused cells acquired the tissue-specific properties and differentiation potential of the host⁶³. Many fused cells are aneuploid. However, it cannot be ruled out that a fused cell can undergo a reduction division resulting in a diploid cell. So, a diploid cell might also represent a fused cell.

As cancer stem cell populations often show striking similarities to normal stem cells, the perceived idea is that normal stem cells represent the target for malignant transformation. An alternative model of tumour initiation might be the fusion of stem cells with cells that have undergone a set of mutational events related to cancer development. Such fused cells would probably express stem cell features and might show large chromosomal aberrations and aneuploidy. They could also harbour unique cell-survival programmes that are shared by normal stem cells and that might drive tumour progression. Therefore, a stem cell fused to a somatic cell that has received a number of defined mutational hits might explain the presence of the chromosomal derangements that can occur during early tumour development (Fig. 3).

Figure 3: Cell fusion, a potential mechanism of cellular transdifferentiation.

a | Transdifferentiation might involve a reduction division of fused cells. Fused cells can undergo a reduction division leading to new somatic cells. Therefore, diploidy in donor-derived cells cannot be used to rule out a potential cell-fusion event as the underlying mechanism of transdifferentiation. Transdifferentiation events are also probably triggered by environmental factors. b | Cell fusion as the underlying mechanism for the establishment of the cancer stem cell. Cell fusion between somatic cells and stem cells might create genetic instability. Alternatively, fusions between mutated stem cells or somatic cells might give rise to tumour stem cells. The mutations could occur in the stem cells, the somatic cells or the fused cells.

Given that cancer stem cells exist, it is a distinct possibility that such cells have the capacity to fuse with somatic cells and give rise to aneuploid, genetically unstable cancer cells. In conclusion, a transdifferentiation event of a normal cell into a cancer cell might occur during early tumour development and during progressive tumour growth. Fusion events of cancer stem cells with somatic cells of different phenotypes can, to a certain extent, also explain the cellular aneuploidy and heterogeneity seen in cancers. However, at present it is not clear to what extent stem cells contribute to tumour initiation and progression.

Fusion might also occur between different tumour cells as well as between tumour cells and normal somatic cells. This is probably a rare event, but it might still be important for tumour progression (Fig. 4).

Figure 4: Cell fusion in the context of tumour progression.

Cell fusion could generate cellular diversity. Cellular and environmental factors might induce cell fusion of both tumour cells and normal cells⁷⁸. Most of the fused cells will either die or enter quiescence, but a small fraction might be able to proliferate, giving rise to new tumour cells as well as cells that contribute to the tumour vasculature. It has been shown that endothelial cells in solid tumours can be cytogenetically abnormal. These cells are aneuploid with multiple chromosomes and multiple centrosomes. This implies that cell fusion between tumour cells and endothelial cells might have occurred⁷⁹.

Given the current focus on stem cell therapies, there is an urgent need to delineate the potential risk factors associated with stem-cell-fusion events. In this context, lessons from developmental biology will certainly shed new light on the initial steps of tumour formation.

Fusogenic factors

Studies of cell fusion processes have identified numerous proteins required for specific fusion events, several of which are both species-specific and cell-type-specific. Recent studies indicate that many proteins involved in cell–cell fusion also are required for seemingly unrelated processes, such as cell migration, axon growth, phagocytosis and synaptogenesis⁶⁴. In Caenorhabditis elegans, fusogenic factors have been studied in detail and there are a large number of molecules that both promote and inhibit cell fusion⁵⁵. In mammalian cells, fusogenic factors have not been appropriately characterized. In fact, almost nothing is known about the mechanisms of stem cell fusion. Whether this is a regulated or random process is also unclear. Fusogenic proteins known to be involved in mammalian cell fusion include CD44, CD47 and the macrophage fusion receptor PTPNS1 (Ref. 65). In addition to known fusogens, certain cytokines and chemokines have also been shown to facilitate fusion. For example, interleukin-4 (IL-4), functioning through its receptor, promotes myoblast fusion with myotubes in vitro. Such fusion events represent crucial steps in muscle growth^66,67. Many tumours, including gliomas, express high levels of the IL-4 receptor, which raises the possibility that IL-4 might promote cell–cell fusion in gliomas. Similarly, the CXCR4 receptor and its ligand, stromal cell-derived factor 1 (SDF1, also known as CXCL12), have a pivotal function in cancer metastasis as well as in the trafficking of stem cells⁶⁸. Recent evidence indicates that SDF1 also promotes monocyte fusion in vitro⁶⁹. These findings imply that the presence of certain cytokines and chemokines in tumours might also lead to an increased incidence of cell–cell fusion events.

Horizontal gene-transfer

Horizontal gene-transfer can occur as a result of phagocytotic or endocytotic processes, including the phagocytosis of apoptotic bodies. Apoptosis involves cell shrinkage, membrane blebbing, chromatin condensation and nuclear fragmentation^70,71. In vivo, these processes are not easily recognized owing to a rapid clearance of the apoptotic bodies by neighbouring cells⁷².

Horizontal gene-transfer has an integral role in the evolution of bacterial genomes and in the diversification of specific bacterial strains. Such processes are known to be important for the establishment of antibiotic resistance in bacteria^73,74. The horizontal gene-transfer process is characterized by three steps: first, the donor DNA has to be delivered to recipient cells; second, the acquired sequences must be incorporated into the recipient's genome; and third, the incorporated genes must be expressed in a manner that is beneficial for the recipient organism. The first two steps might occur through mechanisms such as transformation, transduction and conjugation. Transformation involves the uptake of naked DNA from the environment and has the potential of transmitting DNA between distantly related organisms. Transduction involves the introduction of new genetic material into the host by a bacteriophage that has replicated within the donor cell. Conjugation involves a physical contact between donor and recipient cells and can mediate the transfer of chromosomal sequences by plasmids that integrate into the recipient chromosomes.

In eukaryotic cells in vitro, DNA can be transferred from apoptotic cells to recipient cells by phagocytosis⁷⁵. Co-cultivation of cell lines containing integrated copies of Epstein–Barr virus (EBV) resulted in uptake and transfer of EBV-DNA as well as genomic DNA into the nucleus of the phagocytosing cell. Furthermore, expression of EBV-encoded genes was detected both at mRNA and protein levels, indicating a functional integration of the acquired genetic material⁷⁵. Moreover, apoptotic bodies derived from tumour cells can induce p53-deficient fibroblasts to form colonies in vitro and tumours in vivo, and whole chromosomes or chromosomal fragments can be transferred through the phagocytic pathway to recipient cells⁷⁶. This indicates that horizontal gene-transfer might have an important function in tumour initiation and progression. In this context, we and others have shown that tumour cells can have extensive phagocytic capacites⁷⁷. Such experiments indicate that the transfer of genetic material between tumour cells and normal cells does occur, and that this can contribute to tumour initiation and progression (Fig. 5).

Figure 5: Cancer initiation and progression mediated by horizontal gene-transfer phenomena.

Mutations in somatic cells can lead to apoptosis and nuclear fragmentation. Fragmented DNA might be taken up by other somatic cells through phagocytosis. This might lead to nuclear reprogramming of the phagocytes leading to new tumour cells. Fragmented DNA might also be taken up by other tumour cells (not shown).

Conclusions

Cancer stem cells might be derived from tissue-specific stem cells and bone marrow stem cells. They might also be derived from somatic cells that undergo transdifferentiation processes. Furthermore, cancer stem cells might be initiated as a result of cell fusion or horizontal gene-transfer processes. Therefore, the term 'cancer stem cell' probably relates to a broad group of cells that share some common properties, such as self-renewal and the ability to initiate a tumour. The challenge will be to identify and characterize this group of cells. What common properties do they share and how do they differ? Another challenge will be to pinpoint the differences between cancer stem cells and normal stem cells. Identification of these unique differences will provide novel targets for future cancer therapies.

Box 1 | Cell–cell fusion during development, tissue regeneration and cancer

Cell–cell fusion is fundamental to the development and physiology of organisms ranging from yeast to humans. The importance of cell–cell fusion during development and disease is highlighted by its involvement in numerous biological processes, including fertilization (fusion of sperm and egg), muscle development (fusion of mononucleated myoblasts to form multinucleated muscle fibres), bone (differentiation and fusion of macrophages to form osteoclasts), placenta (fusion of trophoblasts to form syncytiotrophoblasts at the maternal–fetal interface), and in the immune response (fusion of macrophages to form giant cells)⁵⁵.

Tissues such as liver, muscle, bone and cartilage are capable of regeneration after injury. Regeneration might involve the proliferation and differentiation of stem cells that reside in or migrate to injured tissue in response to injury-induced stress signals. However, cell–cell fusion has emerged as an unexpected mechanism in tissue regeneration by stem cells. Circulating haematopoietic stem cells have been shown to fuse with various target cells, including hepatocytes, cardiac myocytes, Purkinje cells and oligodendrocytes, and as a consequence have a modified gene-expression profile⁴².

Increasing evidence indicates that inappropriate cell–cell fusion might contribute to cancer progression. Numerous tumour cell-types are fusogenic, and promiscuous fusion between tumour cells or between tumorigenic cells and normal cells can endow hybrids with new properties, such as higher proliferation rates, the ability to metastasize and resistance to apoptotic signals, including chemotherapeutic drugs⁵⁷. Further evidence suggests that tumour-associated macrophages, through their high fusogenic potential, might have an important function in tumour cell–cell fusion events⁶⁴. It is tempting to speculate that stem cells attracted by mutated cells in a highly fusogenic environment might themselves become partners in cell–cell fusion events, which might lead to genetic reprogramming and the generation of cancer stem cells.