Cancer Stem Cells
Concise Review: Transmissible Animal Tumors as Models of the Cancer Stem-Cell Process†‡§
Author contributions: I.D.O: conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS September 28, 2011.
Tasmanian devil facial tumor disease (DFTD) and canine transmissible venereal tumor (CTVT) are highly unusual cancers capable of being transmitted between animals as an allograft. The concept that these tumors represent a cancer stem-cell process has never been formally evaluated. For each, evidence of self-renewal is found in the natural history of these tumors in the wild, tumor initiation in recipient animals, and serial transplantation studies. Additional data for stem-cell-specific genes and markers in DFTD also exist. Although both tumor types manifest as undifferentiated cancers, immunocytohistochemistry supports a histiocytic phenotype for CTVT and a neural crest origin, possibly a Schwann-cell phenotype, for DFTD. In these data, differential expression of lineage markers is seen which may suggest some capacity for differentiation toward a heterogeneous variety of cell types. It is proposed that DFTD and CTVT may represent and may serve as models of the cancer stem-cell process, but formal investigation is required to clarify this. Appreciation of any such role may act as a stimulus to ongoing research in the pathology of DFTD and CTVT, including further characterization of their origin and phenotype and possible therapeutic approaches. Additionally, they may provide valuable models for future studies of their analogous human cancers, including any putative CSC component. STEM Cells 2011;29:1909–1914.
Tasmanian devil facial tumor disease (DFTD) and canine transmissible venereal tumor (CTVT) are clonally transmissible tumors capable of being transmitted between animals as an allograft [1–3]. The concept that these cancers represent a cancer stem-cell (CSC) process has received little attention and not been formally examined. However, if one considers, broadly speaking, that CSCs are malignant cells with the ability to undergo self-renewal and the capacity to differentiate into a heterogeneous variety of cell types and produce multilineage cell progeny [4, 5], then from available data it could be argued that both DFTD and CTVT represent such a process. Although not formally investigated for CSC attributes, both tumors have been the subject of extensive genetic/pathological analysis and xenograft studies [2, 3, 6-12], and these provide some support for a stem cell element. In this article, analysis of existing data is presented, which may support the hypothesis that these unusual cancers represent models of the CSC process. With this, it is hoped that further analyses and formal investigations could be considered. These may provide additional insights and avenues of investigation of these tumors. Furthermore, if DFTD or CTVT genuinely represent CSC tumors, this may have translational implications for study of analogous tumors in man.
HISTORY AND NATURAL HISTORY OF CTVT AND DFTD
CTVT was first identified as a tumor capable of transmission between dogs by Nowinsky (1876) with subsequent confirmatory studies by Sticker, from whom the eponym, Sticker's sarcoma, is derived [13, 14], although it is believed to have originated far earlier. CTVT is generally recognized to represent a histiocytic sarcoma affecting primarily the external genitalia, being transmitted to both males and females via cell transfer during coitus [1, 3, 15-19]. Extragenital primary sites, as a result of licking or sniffing of lesional tissues include skin, nasal, and orofacial tissues. Tumors remain localized to the primary site of inoculation in the majority of cases but can demonstrate both widespread metastasis and tumor regression [1, 17-19]. “Allograft” acceptance is considered as a product of downregulation of tumor-cell major histocompatibility complex (MHC) antigen expression, with tumor regression promoted by subsequent re-expression [1, 3, 18]. CTVT distribution is worldwide, although reported to have a greater prevalence in tropical/subtropical regions. Although morbidity due to extensive local disease may be significant, overall mortality, while not formally reported is low [17, 18].
In contrast, DFTD is an aggressive cancer involving predominantly the facial tissues that present a serious extinction risk for the Tasmanian devil population, the largest living marsupial carnivore, found solely in the island of Tasmania, Australia [1, 19-22]. The earliest sighting, although quite probably not the index case was photographed in 1996, with the first pathologically documented case noted in 1997. Subsequently, additional cases were frequently reported, spreading geographically from northeastern Tasmania to other regions of the island [20–22]. Epidemiological data shows near 100% mortality within 6 months . Histologically, DFTD is an undifferentiated cancer, with tumor phenotype and cell of origin being the subjects of several studies. All available immunocytochemical data suggests a neural crest derivation , with more recent data supporting a Schwann cell origin [2, 9]. If this is the case, then DFTD should be considered a malignant peripheral nerve sheath tumor (MPNST). Transmission occurs as a product of devil social behavior, in particular biting during sexual contact and competition for food . Inoculation via cell transfer into a wound afflicted by another devil with an ulcerated intraoral tumor has been proposed . In 65% of DFTD cases, metastatic disease is seen, usually involving lymphatics and lungs but with multiorgan disease in some cases [20, 25]. As the index case for DFTD is undocumented, one cannot determine whether the index tumor was primary or metastatic to the orofacial tissues . Individual mortality is assumed to be a result of both metastatic disease and extensive local disease impairing feeding. The more significant extinction threat is a product of individual mortality and reduced breeding/reproductive success, each aided by the rapid spread through the resident devil population [6, 20-22]. Although the mechanism for such spread has been attributed to impaired tumor surveillance as a result of reduced MHC diversity within the species [6, 20-22], recent data demonstrating allogeneic responses, including mixed lymphocyte reactions and, importantly, allogeneic graft rejection between unrelated devils, indicate that other mechanisms exist to facilitate such spread . Further investigation is necessary, but this suggests that DFTD tumor cells may have developed intrinsic adaptive mechanisms to escape immune recognition.
EVIDENCE FOR CLONALITY OF CTVT AND DFTD
There is robust evidence from cytogenetic and genetic analyses that CTVT and DFTD are clonally transmissible cancers [1, 19]. For CTVT, cytogenetic analysis of tumors from various geographical locations all show a similar tumor karyotype (2n = 57–59) distinct from the constitutional canine karyotype (2n = 78) [1, 7]. Furthermore, in all cases of CTVT analyzed, the tumor cells harbor a unique LINE element insertion, near the c-myc locus, absent in the natural canine germline genome [3, 27]. This is a strong diagnostic marker in pathology practice . Microarray-based genome analysis has found remarkable uniformity of tumor-specific copy number and comparative genomic hybridization differences in tumors from diverse geographical origins [7, 28]. However, there exist certain differences in some aspects of the CTVT genome. MHC polymorphism analysis show two distinct clusters, although these are not geographically distinct, and may have arisen soon after this cancers origin, or perhaps represent deep branches diverging from one lineage of the CTVT clone [3, 7]. In addition, recent data demonstrating horizontal transfer of mitochondrial DNA between host somatic cells and CTVT cells also suggests two clades . Of some interest are the attempts to establish this date of origin based on variations identified within molecular data. From such analyses, there are estimates that a common ancestor for these two clusters arose between 47 and 470 years ago  or between 250 and 2,500 years ago . Based on an assumption that CTVT arose in wolves , a date of origin for the original CTVT cell line has been estimated to be between 7,800 and 78,000 years ago .
For DFTD, the proposal of transmission an allograft to explain its rapid spread through the devil population was based on initial cytogenetic studies which supported a clonal nature . The Tasmanian devils' constitutional karyotype is 2n = 14, comprising six pairs of autosomes, sex chromosomes, and, as a polymorphism in some animals, a pericentric inversion of chromosome 5. Pearse and Swift  demonstrated that DFTD tumor cells possess a distinctive unique tumor karyotype, characterized by a complex chromosomal rearrangement, including the absence of both copies of chromosome 2, absence of both sex chromosomes, only one copy of chromosome 6, and, in those animals who carried this rearrangement, the absence of inv5. Subsequent genotyping of microsatellite and MHC polymorphisms in individual tumor samples found identical genotypes at all loci examined, with the overall tumor microsatellite/MHC genotype for each tumor distinct from the host devil . Recently, as a part of a larger pivotal extensive study, genotypic data for 14 microsatellite loci showed similar results, with a similar tumor genotype across all loci distinct from the constitutional array for each animal .
EVIDENCE FOR SELF-RENEWAL IN DFTD AND CTVT
In general, investigation of putative CSCs involves isolation of a CSC subpopulation from a bulk tumor population and subsequent demonstration of tumor forming capacity via tumor transplantation studies. To show cell renewal, cells isolated from an initial primary xenograft must then be successfully transplanted into a second recipient. Such serial transplantation, although problematic, remains the gold standard assay for demonstrating putative CSCs for solid tumors [4, 5, 31-33]. Identification of and enrichment for CSCs in vitro prior to transplantation may be achieved via a number of techniques, including cell sorting via cell surface markers associated with both normal stem cells and CSCs, and nonadherent sphere assays which assess self-renewal capability [31, 33]. As neither CTVT nor DFTD have been formally investigated within the CSC context, no such in vitro data exists. However, for each, supportive in vivo data is available from the natural history of these cancers and formal xenograft studies. In addition, the expression of stem cell-specific genes and markers in DFTD may also indirectly support the possibility of a subpopulation of self-renewing CSCs.
Inductive observation of the natural history of both CTVT and DFTD suggests an obvious capability for self-renewal. Clonality and propagation across sequential animal hosts in the wild clearly indicates that a self-renewing cell population exists, although with certain caveats. It could be argued that inoculation/implantation of the tumor in the recipient animal is solely a product of reduced immunological response allowing acceptance of a tumor bulk population which may not possess CSC properties [32, 33]. However, while recognizing that metastatic potential is not a defining attribute of CSCs per se, both DFTD and CTVT demonstrate clinical metastasis [18, 25], suggesting a capacity for subsequent tumor initiating potential independent of primary implantation. Although no formal data or modeling has been reported, crude estimates of graft transmission in terms of secondary/tertiary and subsequent graft sequences can be made from epidemiological and demographic data so as to calculate numbers of “natural passages.” For example, for DFTD, if one assumes an average “tumor lifespan” of 9 months (i.e., from inoculation to transmission in any given devil, based on 3 months for the development of clinically evident tumor and observed mortality within 6 months) then one may deduce that for those live animals sampled by Loh et al. (from 2001 to 2004), the tumors may represent sequence number 5–9; for those sampled by Murchison et al. animals were sampled from 2006 to 2007 so may represent subsequent graft events [1, 8, 25]. Similar albeit more crude calculations can be made for CTVT, based on the estimates for origin discussed above. Using an average tumor lifespan of 8 years (based on no mortality; canine lifespan of 10 years; 8 years of sexual activity), it can be conservatively estimated that at least 975 sequential transmissions have occurred since the initial origin (7,800–78,000 years ago), with at least 5 and 30 transmissions respectively for the recognized clusters (47–470 or 250–2500 years ago) [3, 7].
While such natural allogeneic graft data is in some sense theoretical, more conventional evidence for self-renewal is available from within-species graft and murine xenograft studies for both CTVT and more recently DFTD [10–12]. In these, it should be noted that the protocols involved the use of a tumor cell population unenriched for any putative stem-cell component. CTVT can be successfully transplanted into other dogs and also sibling canid species (jackals, foxes, and coyotes), with failure in other species (cats and rodents) [17, 34]. Formal xenograft studies using immunodeficient (nonobese diabetic/severe combined immune deficiency [NOD/SCID] and athymic “nude”) mice show that 1 × 106 CTVT cells established from an initial murine xenograft are sufficient to propagate further sequential grafts [10, 11]. Metastatic disease was a feature in the NOD/SCID model, and at least seven sequential xenografts have been performed . Similar data now exist for DFTD. While, in part due to the extinction threat, intraspecies transplantation studies are limited, studies from two individual devils show that controlled transfer of DFTD does occur, with 2.5 × 104 DFTD cells capable of generating a tumor in a susceptible individual devil . More recently, formal xenograft studies in NOD/SCID mice have been performed. In these, 1 × 105 tumor cells from a primary tumor inoculated into mice generated confirmed DFTD tumor growths, with a similar cell number isolated from a primary xenograft capable of generating subsequent xenografts. However, in contrast to the natural state, no metastatic deposits were identified . It should be noted that xenografts with immunocompetent BALB/c mice yielded no viable grafts highlighting the importance of immune function in DFTD transmission. With respect to these cell numbers, it should be recognized that there are no data for the cell number transferred in the natural state (via biting/coitus) for either DFTD or CTVT, which may be lower than those used in xenograft studies. If so, it may be that that the number of cells that are transferred during a CTVT or DFTD transmission event is quite low, with perhaps a correspondingly low number of tumorigenic cells being transferred.
Expression of stem-cell-specific genes and markers in DFTD may also support a capacity for self-renewal and provide further, albeit, oblique support for the presence of a CSC element. Murchison et al.  performed transcriptome analysis of DFTD and quantified expression of selected genes relative to testis tissue. While their presentation of results focused on identification of increased expression of genes associated with Schwann cells and the myelination pathway, their data and extensive supplementary data could arguably support a (cancer) stem-cell element. For example, Sox2, nestin, and the low-affinity nerve growth factor receptor (NGFR)-p75, each recognized neural crest stem-cell markers, showed increased expression (fourfold) in tumor compared with testis. While nestin and NGFR expression is a feature of MPNST , and so may be a product of the proposed Schwann cell phenotype rather than a stem-cell marker per se, other markers also show increased expression. These include a twofold increase in POU3F1 (Oct 6) and a 14-fold increase in EphA2 . Clearly, one should not overinterpret such data, but it does suggest that increased gene expression of certain stem-cell markers is seen. In this respect, it is also worth noting that the tissue sampled was in effect the tumor bulk population. While this may be truly representative of this bulk population, suggesting that all DFTD cells are stem-like in nature (which in some way is contrary to the CSC model), it may be that only a subpopulation (the putative CSC component) actually expresses these markers and could be a relative underestimate of such expression. Finally, another aspect of Murchison et al.'s extensive data may be of interest. Although not increased, DFTD does show expression of PROM-1 (CD133), a classical stem-cell marker (and in some tissues of CSCs), as does the comparative devil testis tissue . This would seem to be the first report of expression in this species, and in marsupials, although phylogenetic studies suggest evolutionary conservation across animal kingdoms . CD44 gene expression was also seen in both tumor and testis tissues .
EVIDENCE FOR HETEROGENEITY OF CELL LINEAGE SPECIFICATION IN DFTD
A CSC hallmark is a capacity for differentiation toward a heterogeneous variety of cancer cell types forming the entire tumor population [4, 5, 31, 33]. In most models, heterogeneity has been demonstrated using cells isolated from a cancer of known phenotype based on the original pathology. In contrast, both CTVT and DFTD manifest as undifferentiated cancers, showing no particular pattern attributable to any specific phenotype, with our current knowledge based on immunophenotype data. As such, it is more challenging to ascribe such differentiation capacity within these tumors. Indeed, the strikingly uniform cell morphology on pathological examination for both DFTD and CTVT would seem at odds with the CSC model, although this may be countered, in part, by the observation that there is no noticeable sequential loss of heterogeneity across the tumors examined. Nevertheless, interpretation of existing data suggests that some degree of heterogeneity is present with respect to immunophenotype.
Initially one might assume, given their clonal nature, that all CTVT and DFTD tumors would show consistency in expression of certain markers associated with their proposed phenotype. This is not so. While variation between studies, in terms of laboratory technique and scoring criteria can explain certain differences in the percentage of tumors positive for any specific marker, in each study, for many markers, only a subset of tumors expressed the majority of key markers within the tumor population as a whole (Table 1). In CTVT, for example, Mozos et al. found that 40% tumors expressed lysozyme and 56% α-1-antitrypsin, while Marchal et al. found that 79% of tumors expressed the macrophage marker ACM1 [15, 16]. Similar findings are seen for DFTD. Murchison and colleagues found that certain Schwann-cell markers are expressed across all tumors, but clearly other markers are variably expressed: NGFR (65%), nestin (45%), neuron-specific enolase (15%), and myelin basic protein by only 10% of tumors. Such patterns are evident for both primary and metastatic disease [2, 9]. Similar, although less pronounced, observations can be made from an earlier study where 28% of DFTD cases expressed mel A . Furthermore, it is clear from these studies that for any individual CTVT and DFTD tumor, there exists significant heterogeneity of marker expression within the tumor cell population. Together, these data would suggest that in CTVT and DFTD, there is differential expression of lineage markers by individual tumors as a whole and by individual tumor cells within any given tumor.
Table 1. Reported antigen expression in CTVT and DFTD
DISCUSSION AND POSSIBLE UTILITY OF A CSC APPROACH
Although both CTVT and DFTD have been the focus of a number of elegant studies to define their clonality, origin, and biological nature, any possible CSC component has not been formally evaluated. While some previous mention has been made of these tumors in a stem-cell context, this is limited in nature. Rinkevich  discussed CTVT and DFTD cells acting as an independent unit of selection in evolutionary terms, rather than any formal evaluation of any CSC potential. More presciently, Tu  did briefly discuss DFTD as a potential CSC, within a larger monograph on CSCs, but no formal evaluation was presented.
The observations in this review relate the known biology of these tumors with particular reference to recognized key aspects of the CSC hypothesis. Certain features of these tumors do support a putative CSC component, including the natural history with transmission and subsequent tumor initiation/propagation in recipient animals in the wild, serial transplantation studies and, for DFTD, possibly the demonstration of stem cell markers. However, others are less convincing. While tumor cell immunophenotypic heterogeneity exists, the absence of any histological evidence of tumor heterogeneity for both tumor types may be significant. Other areas of question are uncertainty around the frequency of tumorigenic cells transferred in the natural state and the significance of the stem cell markers seen in DFTD. While speculative, they do provide some argument for more formal investigation of CTVT and DFTD as cancers with a potential CSC component. Such an approach could provide stimulus to the ongoing study of these animal tumors and perhaps for CSCs in their analogous human cancers. For both CTVT and DFTD, there remains limited data on in vitro behavior. Application of approaches in isolation of potential CSCs (“side-population” analysis, serial colony-forming units and sphere formation assays) could elicit selection of any putative CSC for further study [5, 31, 33]. If present, such a selected cell population may provide a source for additional studies, including in vitro/in vivo phenotypic analyses. For DFTD, the most recent data is supportive of a Schwann cell (MPNST) origin/phenotype [2, 9, 12], which contradicts earlier data indicating a neuroendocrine component [8, 25]. While the latter was based on expression of endocrine markers (chromogranin and synatophysin) and electron microscopy showing secretory granules, it should be noted that MPNSTs can show divergent differentiation that includes neuroendocrine features, although other MPNST-recognizable histology usually exists . Nevertheless, the plasticity of migratory neural crest cell fate and behavior is well recognized, and this diversity is recapitulated in the various neural crest-derived cancers . As such, for DFTD, while some uncertainty around the proposed cell of origin and subsequent phenotype remains, this is consistent with, and may also represent the acknowledged issues around CSC cell of origin  and subsequent phenotypic plasticity of tumors of neural crest origin [33, 41-43]. Additional analyses of any putative CSC population along with recent proposals for ongoing studies of DFTD may clarify this . Similar approaches for CTVT may also confirm its myeloid/macrophage lineage.
Perhaps of far more clinical importance is the identification of any putative CSC component as a therapeutic target. This may be of particular relevance for DFTD, where vaccination strategies have been proposed as a method of halting the devastating spread and extinction threat [2, 9], while similar approaches are being evaluated for CTVT . Although investigational, data suggest that for some tumors the CSC component represents a particular target for vaccine-based therapy [46–48]; as such, identification of a putative CSC for these tumors may be helpful.
While CTVT and DFTD can serve as models to examine immunological aspects of cancer spread and tumor surveillance [19, 20, 49], they can also act as models for other aspects of cancer biology. Indeed, there is a wealth of archival material available for both CTVT and DFTD that could be used to examine various aspects of the cancer (and CSC) process. Such material could represent an existing animal model archive as well as a source of cells for further study. For example, CSCs have been proposed as key to the metastatic process [50, 51], and it could be that any CSC population within DFTD and CTVT may exhibit certain properties considered key to metastasis. It would be of some interest if such cells, and in particular those derived from overt clinical metastasis compared with the primary tumor, showed enhancement of metastatic regulatory pathways, for example, CXCL12/CXCR4 axis. Analyses of collected material from tumors across chronologically sequential hosts can provide study of serial changes in tumor behavior (including epigenetic changes) and that of its supportive stroma.
Further studies of both CTVT and DFTD from a CSC perspective may also provide insight to their respective analogous human cancers (as currently understood), where the CSC model has been investigated. Some data, while equivocal and subject of some discussion, suggest that certain neural crest cancers (e.g., melanoma, neuroblastoma) may have a CSC component [33, 52-54], and recent data indicates that putative CSCs may exist in certain MPNST cell lines . The neural crest origin for DFTD implies that a similar evaluation for CSCs may be of interest to biologists investigating such cancers. Furthermore, if DFTD is indeed a MPNST then any potential CSC component could provide an important ancillary tool for further studies of nerve sheath tumors. In the case of CTVT, as a histiocytic sarcoma (and as such derived from a myeloid-macrophage lineage), a CSC component may seem even likelier given our understanding that hematopoetic cancers may have a far more common CSC pathogenesis [4, 5]. In humans, histiocytic sarcoma is rare and such an animal model would be welcome. Exploration of these relationships demands close liaison and cross-collaboration between scientists and veterinary/medical clinicians. An initial exploration of putative CSCs in CTVT and DFTD would seem one such potentially fruitful direction.
The author would like to thank the anonymous reviewers for their highly constructive comments and suggestions prior to the final publication of this article. The author declares that this study was not funded by any government agency or commercial sponsor.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The author indicates no potential conflicts of interest.