Article Text
PDF
Statistics from Altmetric.com
Colorectal cancer is the third most common form of cancer and the second leading cause of cancer-related death in the Western world, causing 655 000 deaths worldwide per year.1 The incidence of colon cancer is usually increased in developed countries, being seemingly linked with a sedentary lifestyle and a high caloric intake. The occurrence of colon cancer is slightly higher in men than in women and increases with age. Most cases occur between the sixth and the seventh decade of life, while cases before age 50 are uncommon, unless a family history of early colon cancer is present. Colorectal cancer originates from epithelial cells lining the gastrointestinal tract, which undergo sequential mutations in specific DNA sequences that disrupt normal mechanisms of proliferation and self-renewal. Such mutations, which can derive by inborn genetic aberrations, tobacco smoking, environmental carcinogens and chronic inflammatory states, drive the transition from healthy colonic epithelia to increasingly dysplastic adenoma and finally to colorectal cancer. Stem cells of the gastrointestinal tract represent the natural target of tumourigenic mutations, due to both their long life and their capacity for self-renewal. The hypothesis of stem cell-driven tumourigenesis in colorectal cancer has received substantial support from the recent identification and phenotypic characterisation of a subpopulation of colon cancer cells able to initiate tumour growth and to reproduce human colon carcinomas faithfully in mice. This review will discuss the aspects of stem cell biology that can contribute to explain tumour development in the particular context of colorectal cancer. First, we will consider the knowledge available on normal colon cancer stem cells and on the dynamics of colon development, which is crucial to understanding the process of tumourigenesis. Then, we will summarise the new discoveries about colon cancer stem cells and discuss their relevance to the diagnosis and treatment of colorectal carcinoma.
ADULT STEM CELLS: THE CURRENT PICTURE
Adult tissues undergoing massive turnover throughout life rely on a subset of self-renewing undifferentiated cells endowed with the capability of maintaining homeostasis.2 For example, in the human gut, a population of intestinal stem cells is responsible for the generation of >1010 differentiated cells every day. In the last decade, a number of studies demonstrated that stem cells reside even in mitotically inactive adult tissues such as the brain, where they participate in regeneration upon injury.3 Increasing interest in the field has led to the discovery of novel aspects of stem cell biology that fostered the perspectives of a therapeutic use of stem cells for tissue repair (see Korbling and Estrov4). The current paradigm for stem cell definition is the ability to reconstitute a given tissue autonomously in its fundamental morphological and functional aspects. This criterion has been successfully applied to define stem cells in many normal adult tissues,5 6 while for other tissues and organs the stem cell population remains elusive and can only be depicted on the basis of indirect evidence. Normal stem cells of the intestinal epithelium belong to the latter category, as the identification of stem cells in the gastrointestinal tract has not so far achieved a gold standard. We shall briefly revise the process of stem cell characterisation based on functional studies and consider how these criteria can contribute to the identification of normal and neoplastic stem cells of the intestine.
Functional criteria for stem cell identification: the case of haematopoietic stem cells
The milestone of stem cell identification is represented by studies on the haematopoietic system, which traditionally offered technical advantages in the processes of cell extraction, purification and transplantation. To identify the haematopoietic stem cell (HSC), the ultimate proof of stemness is lifelong reconstitution of haematopoiesis in sublethally irradiated mice.7 In this experimental system, human HSCs are purified from the site of definitive haematopoiesis of a donor on the basis of a combined immunophenotypical selection of positive (CD34, CD133, c-Kit) and negative (CD38, lineage-specific) markers. Cells are injected into recipient mice, which are then analysed for the presence and composition of a chimeric haematopoietic system. Subsequent long-term repopulation assays are then performed to confirm the enrichment of HSCs by limiting dilution and serial transplantation experiments. These selection criteria have allowed the definition of two populations of primitive haematopoietic cells: the first is composed of committed progenitor cells undergoing lineage differentiation that are capable only of short-term reconstitution, while the second includes proper HSCs capable of self-renewal, which can repopulate the haematopoietic system upon multiple rounds of serial transplantation. In this experimental framework, repopulation and immunophenotyping assays have reinforced each other in narrowing the definition of an HSC population (fig 1).
Non-functional parameters for stem cell identification: the search for intestinal stem cells
In contrast to what occurred for the haematopoietic system, the stem cells that normally maintain human colon crypts have been poorly characterised. Previous impediments to the identification of intestinal stem cells have been the lack of clonogenic and reconstitution assays and the difficulty of retrieving putative stem cells from their niche, where they are interspersed between more differentiated daughter cells. Therefore, although musashi-1 has been proposed as a putative stem cell marker in the gastrointestinal tract,8 an ultimate phenotypic marker that characterises intestinal stem cells is still missing. The colonic epithelium has a peculiar morphological and functional organisation, where cell proliferation and differentiation occur in an orderly fashion along the crypt–villus axis. Intestinal stem cells are thought to be responsible for producing the immature, transient amplifying progenitors which in turn generate the mature cell types, namely enterocytes, enteroendocrine cells and goblet cells (reviewed in Blanpain et al9). These are integrally turned over in a few days in a continuous upward flow of cells that lose proliferative capabilities and acquire their definitive differentiated phenotype before being shed in the lumen of the intestine. Based on the observation that differentiating cells migrate from the base of the crypt to the villi, intestinal stem cells have been inferred to reside between basal cells of the crypts, within a protective niche composed of proliferating epithelial cells and surrounding mesenchymal cells. Due to their pivotal role in homeostasis and carcinogenesis, colonic crypts have been the object of extensive studies that used methylation patterns to understand the biological features of resident stem cells. These studies have demonstrated that normal human crypts are relatively small in size (being composed of approximately 2000 cells), are quasi-clonal and contain multiple stem cells that undergo recurrent “bottlenecks” or reductions to a new single recently produced progenitor cell.10 Similarly, analyses of non-oncogenic mitochondrial DNA mutations in crypt cells have shown that normal human colonic crypts expand by fission, which represents the mechanism by which stem cells expand and form clonal patches in the colon.11
Several studies have attempted to pinpoint intestinal stem cells within colonic crypts by using indirect techniques based on biological features restricted to the stem cell compartment. Stem cells in adult tissues usually divide at a slow rate when compared with the progenitor population,12 the absolute value of these rates being dependent on tissue turnover and the relative ratio depending on whether the tissue is undergoing repair/regeneration. Based on this assumption, the rarely dividing stem cells are expected to retain increased amounts of compounds which label DNA as compared with more differentiated cells. By labelling the genetic material of proliferating cells in murine intestinal crypts with the DNA precursor [3H]TdR, it has been possible to identify low mitotic index cells that undergo only limited dilution of label over time and are located at the bottom of the crypts in the murine colon.13 Relative positioning in the crypt is presently the most valuable marker for putative stem cells of the colon. Similarly, in the small intestine, stem cells have been localised by the same procedure as a ring of cells occupying the region proximal to the bottom of the crypt, which in this case contains secretory Paneth cells (fig 2). The topographic localisation of intestinal stem cells has allowed their isolation by laser capture microdissection and the subsequent comparison of their molecular profiles with those of the surrounding differentiated progeny.14 As these studies proceed, they will provide new insights into the molecular features of intestinal stem cells and possibly uncover unique determinants that will allow their phenotypic characterisation.
Morphogenetic pathways in colon development
While the quest for a definitive marker of stemness in the intestine has so far been elusive, the molecular pathways that regulate normal and aberrant intestinal morphogenesis have been thoroughly investigated. The molecular description of morphogenetic pathways in the intestine received a great boost from studies aimed at defining the genetic background of familial syndromes associated with a high risk of colorectal carcinoma, such as familial adenomatous polyposis (FAP), juvenile polyposis syndrome (JPS) and Cowden syndrome. FAP results in the formation of multiple bowel adenomas in the second and third decades of life, which almost inevitably progress to colon carcinoma if patients are not subjected to prophylactic colectomy.15 In FAP, approximately 80% of patients display truncating mutations of the APC (adenomatous polyposis coli) gene, which encodes a multimodular protein that participates together with axin/axin2 in the formation of a scaffold that regulates the stability of β-catenin. When extracellular Wnt proteins stimulate their membrane receptors, cytosolic β-catenin levels are stabilised and the protein can then accumulate in the nucleus, where it serves as a coactivator for the Tcf family of transcription factors. Various Wnt ligands and modulators of Wnt signalling have been identified in the crypt–villus unit, where they are secreted in complex and partly overlapping patterns by both epithelial and mesenchymal cells.16 In the absence of Wnt ligands, β-catenin is recruited in a “destruction complex” with APC and axin, and becomes sequentially phosphorylated by casein kinase I and glycogen synthase kinase-3β to create docking sites for ubiquitin ligases that mediate its degradation by the proteasome (reviewed in Gregorieff and Clevers17 and Katoh18) (fig 3). Genetic studies have investigated the components of the so-called “canonical” Wnt pathway, demonstrating the absolute requirement for proper Wnt signalling in colonic development. Mice bearing loss-of-function mutations in key players of the transcriptional programme controlled by Wnt and β-catenin, such as Tcf-4, fail to develop colonic crypts, as do mice with gain-of-function mutations in inhibitory components of this pathway.19 20 The complex processes that contribute to intestinal development and are deregulated in tumourigenesis involve the contribution of other pathways acting at the stem cell level to control colon homeostasis. The Wnt cascade interplays with the Notch pathway to maintain undifferentiated, proliferative cells in normal crypts and adenomas.21 Another way through which Wnt proteins regulate intestinal cell proliferation is by promoting the expression of B subclass ephrins and their tyrosine kinase receptors. The transition from adenoma to colon carcinoma is associated with loss of ephrin-B receptor expression, and these receptors have recently been shown to coordinate migration and proliferation in the intestinal stem cell niche.22
The study of JPS has disclosed another crucial link between molecular genetics and developmental biology, revealing the importance of the SMAD/BMP (bone morphogenetic protein) pathway in shaping intestinal architecture. JPS is a condition that predisposes to hamartomatous gastrointestinal polyp formation, which can turn into malignant lesions in approximately 20% of cases.23 JPS is due to germline mutations in the SMAD4 gene in 15–20% of cases, and to mutations in the gene encoding BMP receptor 1A in 25–40% of cases. BMP family ligands are expressed by the villus mesenchyme, while epithelial cells display nuclear phosphorylated SMADs, implicating these cells as terminal recipients of the signal. By ectopic expression of Noggin, a negative modulator of BMP signalling, it is possible to direct the unregulated formation of new crypts in the intestine, leading to lesions reminiscent of the hamartomatous polyps seen in JPS. This has led to the definition of “landscaper” function for BMPs in the intestine24 and highlights the concept that alterations in the intestinal stem cell microenvironment might be as influential as cell-intrinsic events in normal development and tumourigenicity. The importance of cell–cell interactions in influencing stem cell survival and homeostasis is probably responsible for the lack of success in propagating adult stem cells in vitro, probably due to the inability to reproduce all the crucial aspects of the native stem cell niche in terms of both soluble factors and stem cell–stroma interactions. The Wnt pathway and the BMP pathway interact to control intestinal stem cell self-renewal, as they have been shown to converge mechanistically on the control of β-catenin through the action of PTEN (phosphatase and tensin homologue).25 PTEN mutations are responsible for 80% of cases of Cowden disease, which causes hamartomatous neoplasms of the skin and mucosa with common development of colon polyposis. Recently, the causal relationship between PTEN inactivation and intestinal polyposis has been elucidated by studies showing the widespread occurrence of hamartomatous intestinal polyps in PTEN-deficient mice. In these mice, intestinal stem cells undergo deregulated proliferation driven by the activation of Akt/PKB (that is normally kept under control by PTEN) and by the nuclear localisation of β-catenin, thus resulting in crypt overproduction.26
STEM CELLS AND CANCER: THE CANCER STEM CELL CONCEPT
It is generally accepted that neoplastic transformation is a multistep process caused by the accumulation of both mutations and epigenetic changes. Due to their longevity and self-renewing properties, stem cells seem to be the ideal target of the carcinogenic process, since they have a greater propensity to accumulate mutations compared with short-lived, more differentiated cells. The hypothesis that tumours are driven by cellular components that display stem cell properties is in line with the observation that, although monoclonal in origin, most cancers show a marked degree of morphological and functional cellular heterogeneity.27 According to this observation, a human tumour would not be a mere monoclonal expansion of a malignant cell but rather an “aberrant organ” sustained by a transformed cell able to generate progeny of cancer cells that become heterogeneous as a result of differentiation. At the functional level, the major difference between cells within a tumour mass is observed in their proliferative potential. Indeed, the concept that tumours contain cell populations with stem cell properties was first suggested by clonogenic assays showing that only a small portion of the cells isolated from tumour specimens have a high proliferative capacity, indicated by the number of colonies produced in soft agar.28 Furthermore, large numbers of primary human cancer cells must typically be injected into immunocompromised mice in order to obtain tumour formation, thus suggesting that only a minority of the cancer cells resident in the tumour are tumourigenic.
Models of tumour development
Two models have been proposed to explain tumour development. The “stochastic model” assumes that every cancerous cell has the capacity to proliferate extensively and regenerate the tumour. According to this hypothesis, discrete populations of cancer cells isolated on the basis of functional or phenotypic characteristics would not bear increased tumour-initiating capacity with respect to the bulk tumour population, and the failure in engraftment when tumour cells are injected into immunodeficient mice would have to be regarded solely as a technical flaw in the assay. In contrast, the “cancer stem cell” model assumes that only a small subset of cells within the tumour population has the capacity to initiate and sustain tumour growth. Therefore, when the number of tumour cells injected in recipient mice falls below a given threshold, the graft fails because no cells able to reconstitute the tumour are in fact injected. According to the cancer stem cell model, it is possible to isolate and purify a small population of tumour-initiating cells endowed with the ability to self-renew and to undergo differentiation. The cancer stem cell model has been supported by studies showing that the growth of several tumours depends on a small subset of stem cells displaying many features in common with their non-transformed counterparts. These cells, characterised by the capacity for self-renewal, the potential to differentiate and the ability to proliferate indefinitely, have been named “cancer stem cells” (CSCs). It is possible that CSCs arise from transformation of normal stem cells of the corresponding tissue from which they are isolated. However, CSCs may also originate from mutated transit amplifying cells that possess extensive proliferation abilities but lack the proper stem cell self-renewal capacity, as has been proposed to occur in chronic myeloid leukaemia.29 This possibility implies that a progenitor cell must acquire mutations associated with the re-establishment of the property to self-renew (fig 4). One should note that both cases might apply, especially in those tissues where the transit amplifying population is relatively long lived, or especially susceptible to the effect of mutagens that cause genomic rearrangements.30 Although a detailed discussion on the origin of CSCs is beyond the scope of this review, it is important to note that mutations that affect the pathways responsible for normal stem cell self-renewal are associated with a wide array of human cancers.31
Experimental evidence on CSCs
The identification of CSCs establishes the existence of a hierarchy within a tumour and has profound implications for our understanding of tumour biology. The assumption that a small subgroup of cancer cells is responsible for tumour growth was first proven in the context of acute myeloid leukaemia (AML) and since then has been confirmed in several cancers. Dominique Bonnet and John Dick first identified and isolated CD34+CD38− leukaemic stem cells (LSCs) from human AML and demonstrated that these cells were able to initiate leukaemia in non-obese diabetic/severe combined immunodeficiency (NOD/SCID) mice, whereas the more differentiated CD34+CD38+ and CD34− fractions failed to do so.32 Engrafted leukaemic cells could be serially transplanted into secondary recipients, thus providing functional evidence for self-renewal of the LSC population. This pioneer study demonstrated that tumours can be composed of a heterogeneous cell population, since LSCs possess an extensive proliferative and self-renewal capacity that is not found in the majority of the leukaemic blasts.
Subsequently to haematological tumours, CSCs have been isolated in solid cancers such as breast and brain tumours.33 34 Studies by Al-Hajj and colleagues demonstrated that only a small fraction of breast cancer cells expressing CD44 but little or no CD24 were able to establish tumour xenografts and to give rise to both tumourigenic (CD44+CD24−/low) and non-tumourigenic cells. Notably, as few as 100 CD44+CD24−/low cancer cells were able to form primary and secondary tumours, whereas the injection of tens of thousands of cancer cells with alternative phenotypes failed to do so.33
Similar results have been found for cancers of the central nervous system, where the isolation of CSCs has been achieved by exploiting in vitro culture conditions previously developed for normal neural stem cells.35 Normal neural stem cells express the surface marker CD133, which is commonly found on stem and progenitor cells of various tissues.36 Purification of CD133+ cells from primary human brain tumours allowed the isolation of brain tumour stem cells that exhibited self-renewal properties in vitro and, most importantly, in vivo.34 37 Injection of as few as 100 CD133+ cells into NOD-SCID mice produced a tumour that was a phenocopy of the patient’s original tumour and could be serially transplanted, whereas 105 CD133− cells were able to engraft but did not generate a tumour.37 The existence of CSCs has been demonstrated for other solid tumours besides those of the haematopoietic and nervous system. CSCs have been isolated from prostate cancer,38 melanoma,39 pancreatic cancer,40 hepatic cancer,41 head and neck carcinomas42 and colorectal cancer.43 44 Table 1 summarises the cell surface markers that have been used to characterise a tumour-initiating cell population according to the standards required to define CSCs.
STEM CELLS IN COLORECTAL CARCINOGENESIS
Studies on hereditary cancer syndromes, which account for 5–10% of colorectal carcinoma cases, contributed massively to the understanding of cancer biology.15 The observation that the accumulation of mutations in tumour suppressors and oncogenes seems to parallel the progression of the disease along the adenoma–carcinoma sequence led Fearon and Vogelstein in 1990 to propose a model of successive genetic changes leading to colon cancer45 (fig 5). In the original proposal, the authors stressed that mutations in key genes controlling cell proliferation and genome integrity, such as KRAS and p53, were essential for tumour development. During the last 15 years, the molecular alterations underlying the progressive changes in colon histology have been unravelled, as studies on relatively rare inherited cancer predisposition syndromes led to the identification of genes that play a major role in sporadic colon carcinomas.
Mutational spectrum of colon cancers
Studies on FAP and hereditary non-polyposis colon cancer (HNPCC, also known as the Lynch syndrome) allowed the identification of two alternative pathogenic mechanisms for colon tumourigenesis. Work on FAP identified the “gatekeeper” tumour suppressor gene APC as the initial mutation involved in the adenoma–carcinoma progression. The early occurrence of APC mutations during colorectal tumourigenesis was confirmed by the sequence analysis of 41 colorectal tumours, revealing that most adenomas (63%) and carcinomas (60%) contained a mutated APC gene.46 Moreover, other studies indicated that >80% of tumours had at least one mutation in the APC gene, of which >60% had two mutations.47 These observations strongly suggest that somatic mutations of the APC gene are associated with the development of the great majority of colorectal tumours.48 APC mutations typically affect the central domain of the protein containing the binding site for β-catenin and, consequently, determine the increase of nuclear β-catenin and the transcriptional activation of specific target genes, such as the oncogene c-myc.49 50 Furthermore, approximately 10% of colorectal cancers carry activating mutations in the highly conserved serine/threonine residues of β-catenin, which are required for recognition and degradation of the protein.51
HNPCC syndrome is the most common form of hereditary colorectal cancer characterised by the absence of prior polyp formation.52 Ninety percent of HNPCC patients carry mutations in the “caretaker” mismatch repair genes, hMSH2 and MLH1.53 54 Consequent inactivation of the mismatch repair system leads to the instability of microsatellites, genomic regions characterised by multiple repeats of short DNA sequences or single nucleotides, which represents the hallmark of tumours in HNPCC. The microsatellite instability typical of HNPCC creates a favourable state for accumulating mutations in genes that control cell survival, proliferation and self-renewal, probably fostering the generation of colon carcinomas. Not surprisingly, colorectal tumours with mismatch repair deficiency have been shown to bear an excess of frameshift mutations in the “gatekeeper” gene APC.55 However, it has been hypothesised that genomic instability alone is not sufficient for colon cancer development, but has to be combined with the selection of genetic alterations through the clonal expansion of an aberrant stem cell population. Taken together, the studies performed on FAP and HNPCC patients demonstrate the importance of both “gatekeeper” and “caretaker” gene function. FAP results from an increased rate of tumour initiation due to abrogation of the gatekeeper function of APC. In contrast, the mismatch repair defect in HNPCC results in an enhanced rate of mutation that greatly accelerates tumour initiation. HNPCC and FAP patients both develop cancer at a median age of 42 years, suggesting that both tumour initiation and tumour progression are rate limiting for colon cancer formation.
Models of adenoma histogenesis
The formation of adenomas during the neoplastic transformation of colonic epithelium has been proposed to occur according to two alternative models, named respectively the “top-down” and the “bottom-up” model.56 57 The top-down model is based on the observation that, in early neoplastic lesions, the dysplastic cells found at the luminal surface of the crypts are genetically unrelated to the cells at the base of the same crypt, which consist of epithelial cells indistinguishable from those found at the crypt bases in normal colonic mucosa. These observations imply that the development of adenomatous polyps proceeds through a top-down mechanism, since the dysplastic cells in the superficial portions of the mucosae spread laterally and downward to displace the normal epithelium of the adjacent crypts and eventually replace them. This model has profound implications for concepts of stem cell biology in the gut, placing the stem cell compartment in the intracryptal zone or defining the intracryptal zone as the site where stem cells that have acquired a second mutagenic hit clonally expand (fig 6). An alternative hypothesis, known as the bottom-up model, proposes that the malignant transformation process takes place among the stem cell population at the crypt base, and that the transformed stem cell then migrates to the apex of the crypt where it expands. This proposal is based on findings that the more frequent lesion in FAP is the unicryptal or monocryptal adenoma, where the dysplastic epithelium occupies an entire single crypt.58 According to studies in mice that analysed the stability of mutant crypts in the human gut, such a monocryptal lesion should be clonal.59 Similar crypt-restricted lesions presumably caused by the clonal expansion of mutated stem cells have been well documented in patients heterozygous for the gene encoding the enzyme O-acetyl transferase, responsible for the O-acetylation of sialic acid in goblet cell mucus. In these patients, initially half and then the whole crypt is colonised by the progeny of the mutant stem cell.60 However, the strongest evidence for the clonality of human colonic crypts came from the in situ hybridisation of Y chromosome studies performed on an XO/XY chimeric patient who underwent colectomy for FAP. All the 12 000 crypts analysed were composed of either Y chromosome-positive crypts or Y chromosome-negative crypts.57 Furthermore, the observation that the base of the crypt is the repository for stem cells that repopulate the colon after epithelial injury61 gives additional support to the bottom-down theory of colon tumourigenesis.
Identification and isolation of colon cancer-initiating cells
Despite the fact that colon cancer has been traditionally regarded as a disease of the intestinal stem cell, until recently this assumption had not been experimentally demonstrated. Lately, two research groups independently identified a colon cancer-initiating cell (CC-IC) population in human tumours.43 44 CC-ICs consist of undifferentiated cells characterised by the expression of CD133, a surface marker commonly found on stem/progenitor cells in various tissues.36 CD133+ cells express the epithelial antigen BerEp4 (also known as ESA and EpCAM), but not differentiation markers such as cytokeratin-20, an intermediate filament protein commonly present in mature cells of the gastric and intestinal epithelium. CD133+ cells are also found in normal colon tissues, although at a much lower frequency. This observation suggests that the increased number of CD133+ cells in cancer samples results from the oncogenic transformation of normal colonic stem cells. The subcutaneous injection into SCID mice of 3000 CD133+ colon cancer stem cells freshly isolated from tumour specimens readily reproduces the original human tumour.43 The number of CD133+ cells able to generate tumour xenografts further decreases in a NOD/SCID mice xenograft model, in which as few as 100 CD133+ cells injected into the subrenal capsule are able to reproduce the original colon tumour. This model allowed the authors to perform in vivo limiting dilution experiments to estimate the frequency of stem cells in colon tumours. These experiments indicated that, on average, there is one CC-IC every 5.7×104 unfractionated colon cancer cells and one CC-IC every 262 CD133+ cells.44 The long-term tumourigenic potential of CD133+ colon cancer cells was demonstrated by serial transplantation assays, which also showed that transplanted tumours maintained their original phenotypic appearance but displayed an increased aggressiveness.43 Conversely, the CD133− cell population extracted from colon tumour specimens was constantly unable to generate tumours.
Importantly, colonic cells obtained from dissociation of cancer specimens can be propagated and expanded in a serum-free medium containing epidermal growth factor and fibroblastic growth factor43 (fig 7). In these conditions, tumour stem cells and progenitor cells grow exponentially and give rise to floating CD133+ cell aggregates named colon spheres, which can engraft and generate tumours in SCID mice. Upon growth factor deprivation and serum addition, CD133+ tumour spheres gradually differentiate into adherent cells that express cytokeratin-20 and high levels of the colon carcinoma-specific marker CDX2, whereas CD133 expression is progressively downregulated. As expected, differentiated cells obtained from CD133+ tumour spheres are unable to transfer the tumour into SCID mice.43 The purification of colon cancer stem cells on the basis of CD133 expression has been subsequently achieved by other investigators, who additionally demonstrated the peculiar capacity of colon cancer spheroids cultured in vitro in the presence of serum and matrigel to differentiate into complex crypt-like structures.63
In addition to CD133-based identification of colon cancer stem cells, Dalerba and co-workers recently reported an alternative protocol for the isolation of human CC-ICs that exploited the surface phenotype EpCAMhigh/CD44+/CD166+.64 EpCAM and CD44 antigens were selected on the basis of their previously described expression on human breast cancer stem cells, whereas CD166 is known as a mesenchymal stem cell marker whose increased expression in colon cancer is associated with a poor clinical outcome.65 Experiments performed on xenograft cell lines obtained by the injection of primary tumour samples showed that the tumourigenic potential was restricted to CD44+/CD166+ cells within the EpCAM+ population. Based on the analysis of xenograft-derived cells, the authors suggested that EpCAM and CD44 may represent more robust markers than CD133, as CD44 expression can also be detected in tumour cell populations that score negative for CD133 expression. EpCAMhigh/CD44+ cells derived from a mouse xenograft of colon carcinomas were shown to express other stem cell-associated markers such as CD49f and aldehyde dehydrogenase enzymatic activity. The presence of additional phenotypic markers on CD133+ colon cancer stem cells will facilitate their isolation, but this still awaits confirmation by extensive analyses on surgical specimens.
THERAPEUTIC IMPLICATIONS OF THE CSC CONCEPT
According to the CSC hypothesis, colorectal cancer can be considered a disease in which mutations convert normal stem cells into aberrant counterparts responsible for tumour generation and propagation. The statement that only a rare subset of cells drives tumour formation has major implications for the development of new therapeutic strategies aimed at targeting and eradicating the tumour stem cell population. Traditional chemotherapeutic regimens indiscriminately kill proliferating cells, thus failing to account for potential differences in drug sensitivity or target expression between CSCs and the more numerous non-tumourigenic neoplastic cells. Consequently, CSCs may well survive treatments,66 and relapse may then occur as the result of CSC-driven expansion (fig 8). Drugs able to kill CSCs may be overlooked in screening methods that rely on rapid reduction of cell number/tumour size, while screenings specifically designed to target tumourigenic cells may yield more effective antitumour treatments. DNA and tissue microarray analyses of tumours have been currently used to identify subtypes of cancers in order to improve diagnosis and treatment by allowing prediction of the response to therapies. The extension of cluster analysis to enriched populations of tumourigenic cancer cells could generate new findings of rapid therapeutic application and lead to the identification of novel diagnostic markers. The possibility of obtaining a virtually unlimited expansion of colon cancer tumourigenic cells through combined in vitro and in vivo approaches has considerable implications for the evaluation of antineoplastic drug efficacy. The use of colon cancer spheres to generate animal models recapitulating the parental tumour might be applied to identifying agents that selectively kill CSCs. Studies performed on CD133+ colon cancer stem cells have recently demonstrated that this cell population produces interleukin 4 (IL4) as an autocrine growth factor that promotes tumour survival through the upregulation of antiapoptotic genes.63 In a mouse model of human colorectal cancer obtained by subcutaneous injection of CD133+ cells, treatment with the IL4Rα antagonist or an anti-IL4 neutralising antibody significantly enhanced the antitumour efficacy of conventional chemotherapeutic drugs by specifically targeting the tumour stem cell population. These observations demonstrate how a deeper understanding of the biological features of CSCs could translate into new therapeutic strategies.
Acknowledgments
We thank the Italian Ministry of Health, the Italian Ministry for University and Research (FIRB_RBIP06ZJ78) and the Italian Association for Cancer Research (AIRC) for supporting the colon cancer stem cell research.
REFERENCES
Footnotes
Competing interests: None.