In the multistep process of cancer development, the concept that cancer stem cells are derived from normal stem cells that have gradually accumulated various genetic and epigenetic defects is gaining strong evidence. A number of investigations have identified molecular markers that, under normal conditions, are responsible for stem cell homeostasis but are also expressed in tumor “stem cell-like” subpopulations. In this regard, it was recently reported that a group of tumor-specific antigens known as cancer/testis antigens (CTAs) are expressed in human MSCs. It has long been stated that in normal tissue these antigens are exclusively expressed in germ cell precursors; however, based on these results, we suggest that CTAs are expressed at earlier stages during embryogenesis. The tumor-restricted expression of CTAs has led to several immunotherapeutic trials targeting some of these proteins. The clinical implications that these trials may have on the normal stem cell pools, as well as the immunologic properties of these cells, is to date poorly studied and should be considered.
It has been recently suggested that oncogenic mutations and other genetic and epigenetic defects could be inherited from transformed normal stem cells, giving rise to some of the cell populations observed in many tumors . This important feature of cancer was recognized by the fact that a small fraction of the tumor cell population keeps the stem cell properties of self-renewal and proliferation [2, 3, 4], giving rise to the concept of “cancer stem cells.”
Cancer stem cells were first identified in acute myeloid leukemia when surface markers were used to distinguish the stem cell population from the remaining cells with limited proliferative potential . In solid tumors, cancer stem cells have been identified in breast cancer [6, 7], glioblastomas , lung cancer , ovarian cancer , prostate cancer , and epithelial gastric cancer . Based on these observations, a cancer stem cell model has been proposed, and it is based on the concept that the great majority of the tumor cells have a limited proliferative potential, but a small cell population—the cancer stem cells—are able to self-renew and proliferate, maintaining the tumor cell mass. In this model, cancer is a disease of deregulated self-renewal of normal stem cells. Thus, in the cancer stem cell model, tumor recurrences and even metastases may occur due to residual cells—probably chemotherapy-resistant—that are able to expand to form secondary tumors .
In accordance with the clonal evolution of stem cell populations, a pretumor progression hypothesis was recently postulated in an attempt to explain the connection between stem cells and cancer [12, 13]. In this hypothetic model, which is based on the renewal of stem cells in human colon crypts, genetic and epigenetic defects might accumulate stochastically and sequentially in stem cells, and their “defective” progeny are responsible for the tumor progression. In this model, defective stem cells would mark the transition between pretumor to tumor progression .
A number of hypotheses, mostly based on new experimental evidences, have been proposed to explain the presence of cancer stem cells in cancer and the implications for future therapies. It has been suggested that certain cancer cells might acquire stem cell-like characteristics via genetic mutations and epigenetic defects or that cancer stem cells can occur when circulating stem cells are able to fuse with cancer cells. For example, in a recent article, Krivtsov and colleagues introduced the MLL-AF9 fusion protein into granulocyte-macrophage progenitors and were able to isolate leukemia stem cells from acute myeloid leukemia . In this case, the isolated leukemia stem cells had a gene expression profile similar to that of normal granulocyte-macrophage progenitors; however, a “self-renewal” transcription program was activated in the leukemic stem cells, indicating that genetic changes can activate the self-renewal program in premalignant cells . In another model, Riggi and colleagues generated tumors histologically similar to Ewing's sarcoma by introducing the EWS-Fli1 gene in bone marrow-derived mesenchymal progenitors . In regard to cell fusion events between circulating stem cells and cancer cells, it was already demonstrated that this could be the case in some types of tumors . Rizvi et al. provided in vivo evidence that supports the hypothesis that fusion between circulating hematopoietic stem cells and transformed intestine epithelium is likely to be involved in the generation of intestinal tumors . So, although the cancer stem cell model is the most discussed and accepted hypothesis at the moment, there is evidence that other events might explain the presence of a small cell population with self-renewal characteristics in tumors.
It is also clear that several pathways that are deregulated in tumorigenesis are important to stem cell homeostasis. For example, mutations in the Wingless signaling pathway have been well-documented in colorectal , prostate [18, 19], and ovary tumors . These defects are able to change stem cell survival in model systems . Furthermore, survival probability of stem cells in the niche can be changed by Catenin alterations , and these molecules have been strongly associated with tumor progression [23, 24, 25]. Several members of the Polycomb family that act as transcriptional repressors by regulating chromatin remodeling have been implicated in stem cell function and are defective in different types of cancer [26, 27]. Recently, it was proposed that some tumor-associated proteins known as cancer/testis antigens (CTAs) are also implicated in stem cell differentiation pathways  (Fig. 1). This new information suggests that the tumor-specific expression of CTAs may be linked to stem cell biology.
ESCs and MSCs
ESCs, the most pluripotent cells, are derived from the inner mass of the blastocyst-stage embryos  (Fig. 2). Early in mouse development, ESCs can proliferate and differentiate to generate a pluripotent population of cells that are known as the primitive ectoderm. Shortly after implantation, some of these cells form the epiblast (day 5, 6). Then, groups of epiblast cells undergo epithelial-to-mesenchymal transitions and develop the mesoderm and endoderm layers. Later, a temporally and spatially controlled process will generate different specialized cells . Accordingly, the epiblast generates both the mesenchymal stem cells and primordial germ cell progeny with an activated CTA program that later becomes inactivated as the result of cell differentiation. This has been proven in experimental systems in which mesenchymal stem cell differentiation into osteocytes and adipocytes is paralleled by a down-regulation of CTAs . Adding support, most CTAs have been detected in gametogonia but not in differentiated oocytes or spermatocytes . Thus, restricted expression of CTAs in undifferentiated somatic and germ cells raises the possibility that expression of some CTAs may be essential for embryonic development. This question could be better addressed by using ESCs or embryonic germ cell differentiation cultures. Both female  and male  germinal cells have already been generated in vitro from ESCs.
MSCs were first identified as adherent, fibroblast-like populations in the adult bone marrow that could regenerate rudiments of bone in vivo . MSCs have also been isolated from other tissues, including adipose tissue and several fetal tissues. MSCs proliferate extensively in culture, and because no definitive marker has been identified, MSCs generated ex vivo are defined as cells expressing a combination of phenotypic nonhematopoietic markers that retain the capacity to differentiate into adipose tissue, cartilage, and bone in vitro . The phenotype of the cells in vivo is not fully characterized, but the cells appear to be rare noncycling cells [36, 37]. Data on the function of MSCs in vivo are still very limited. Culture-derived MSCs engraft in multiple organs after i.v. infusion and show site-specific differentiation under normal circumstances. Besides their ability to “home” to sites of tissue injury, they are also part of the tumor stroma in experimental animal models. Some studies also suggest that MSCs may represent a precursor of certain tumorigenic cells [10, 38], but the data on whether they enhance or suppress tumor growth in animal models are not clear.
CTAs and Stem Cell Development
CTAs comprise a family of genes that share common characteristics: 1) they are expressed in a wide variety of malignant tumors, but expression in normal tissue is mostly restricted to germ cells of the testis, fetal ovary, and placenta and confined to immature cells such as spermatogonia, oogonia, and placental cells such as trophoblasts. Some CTAs can be expressed in nongametogenic tissues such as the pancreas, liver, and spleen at levels far below that observed in germ cells ; 2) their expression programs are strictly regulated by epigenetic mechanisms such as DNA methylation ; and 3) they are immunogenic. To date, nearly 40 distinct CTAs have been identified based on immunogenic properties , expression profiles , and by bioinformatic methods . However, little is known about their specific functions, and their functional connection with stem cell biology and cancer is widely unexplored. In this regard, it was recently reported that some CTAs such as N-RAGE, NY-ESO, MAGE-1, and SSX are expressed in human mesenchymal stem cells of the bone marrow, suggesting that CTA expression may not only be a hallmark of gametogenesis but also a stem cell marker  (Fig. 1). Importantly, several findings afford a role of the CTA SSX in stem cell biology. SSX is recruited to the nuclear domains occupied by the PcG protein Bmi-1 , a regulator of stem cell self-renewal . SSX is involved in mesenchymal-to-epithelial transition (MET) by promoting matrix metalloproteinase 2 expression and changing E-cadherin levels, consequently affecting cell migration . MET is a reversible process essential for the development of the different germ layers during embryogenesis. Thus, aberrant expression of SSX could trigger epithelial-to-mesenchymal transition and favor metastasis of tumor cells. This provides additional support for the role of SSX in self-renewal and differentiation of stem cells. Furthermore, the control mechanisms of SSX and Bmi-1 expression may be an intrinsic property of cancer stem cells.
Cancer/Testis Antigens in Gametogenesis and Cancer: New Definitions
A recent review article written by Simpson and colleagues  highlighted the link between CTAs with gametogenesis and cancer. The tightly specific definition of CTAs in this article refers to proteins expressed in several tumors but not in normal tissue. As described above, in normal tissue their expression is restricted to germ cells of the testis and placenta (i.e., spermatogonia and trophoblasts). However, we believe that CTAs are playing a role at earlier stages during embryonic development and in stem cell self-renewal. Because CTAs are not expressed or have very low expression in normal differentiated tissue, it is tempting to suggest that CTA expression in tumor tissues is restricted to cells that retain stem cell properties. In the tumor mass, a restricted cell population maintains stem cell properties, favoring tumor maintenance, proliferation, and metastasis.
The number of tumor-specific T cells can be increased by patient-specific vaccines generated to stimulate an immunological response against tumors. It can also be increased by adoptive cell transfer immunotherapy, based on ex vivo selection, numerical expansion, and activation of autologous tumor-reactive lymphocytes. Recent clinical reports have shown promising results in patients with metastatic melanoma using these approaches [47, –49].
The fact that CTAs are not expressed in normal but in malignant tissues, combined with the immunogenic properties of these antigens, has prompted researchers to generate T-cell-specific CTA responses against the tumors. The new observations discussed here suggest that these therapies should be evaluated taking into account the risk of side effects on host stem cell populations (especially MSCs).
It is possible that CTAs are true hallmarks of cancer stem cells and provide unique targets to treat recurrences and metastatic cancer. CTA expression is dependent on the maintenance of an undifferentiated phenotype in stem cells, and theoretically, cancer cells in which CTAs are expressed may have lost their ability to differentiate. Interestingly, CTAs are heterogeneously expressed in tumor tissues in only a few cells within the tumor mass. Assuming that cancer stem cells are progenitors of metastatic cells and/or recurrences, drugs developed to specifically target differentiation associated genes, such as CTAs, could improve the treatment of cancer.
Some cancer patients, particularly with advanced disease, generate CTA-specific humoral and T-cell immunity spontaneously. However, it is not clear whether circulating CTA-specific T cells are able to kill tumor cells in vivo. The fact that CTAs are expressed in some normal cells (such as MSCs) suggests that CTAs are not antigenic in vivo. Cytotoxic lymphocytes (CTLs) responsive against NY-ESO have been reported in a substantial number of patients, whereas T-cell responses against members of the MAGE family are rarely found. Some CTAs are highly expressed in MSCs as discussed before , and most of the CTAs are also expressed at very low levels in a few specific normal tissues . The fact that CTAs are expressed naturally in the body in a tissue or cell-type-specific manner could influence the fate of CTA-specific immunotherapy. Deletion of functional high-avidity self-reactive T cells in the thymus as well as peripheral deletion or anergy has been shown in various animal models. It is also possible that antigen processing of specific CTAs could influence its ability to be recognized by CTLs.
Natural peripheral T-cell responses against various CTAs exist in patients with melanoma , leukemias , and other carcinomas . Because it was shown that CTAs are also expressed in nonmalignant cells, the problems of immune tolerance and autoimmunity are important issues to consider when attempting to elicit T-cell responses against these antigens.
It is also possible that MSCs differ from cancer cells and escape recognition by therapeutically infused CTA-specific CTLs. This is supported by the fact that MSCs are not immunostimulatory in vitro when cultured with allogeneic lymphocytes [53, 54]. Furthermore, MSCs can escape lysis by allogeneic cytotoxic CD8+ T cells (CTLs). After transplantation of fully HLA-mismatched MSCs into an immunocompetent fetus, the cells persisted for a long term . The transplanted MSCs did not induce any immune response in the child, again indicating that MSCs have immunoevasive properties. It is also possible that MSCs, even though CTAs are expressed, would not be harmed by CTA-specific CTLs. However, the possibility that targeted therapies may have deleterious side effect on the host stem cells that populate different niches in the body will need to be considered in future studies.
The central role of CTAs should be reviewed in the context of stem cell biology. The growing concept of cancer stem cells suggests that stem cell-ness may be an intrinsic attribute of highly aggressive cancer cells and has stimulated the search for the molecular markers that are shared between cancer and stem cells. In addition, targeted therapies have been launched against these markers. However, we suggest that consideration should be taken in analyzing the possible side effects of these therapeutic approaches on the normal stem cell pools. We also stress that the immunological properties of stem cells should be properly studied before applying any kind of cancer therapy.
The authors indicate no potential conflicts of interest.
We thank the reviewers for their input and helpful suggestions. B.B. and K.L.B. also thank the support provided by the Swedish Cancer Society, the Children's Cancer Foundation, the Swedish Research Council, the Tobias Foundation, the Stockholm Cancer Society, the Swedish Society of Medicine, the Sven and Ebba-Christina Hagbergs Foundation, and Karolinska Institutet.