There is a growing realization that many, if not all cancers contain a small population of cancer stem cells (CSCs). These cells exhibit stem cell characteristics, particularly the ability to self-renew, and give rise to a hierarchy of progenitor and differentiated cells that sustain the tumor and are responsible for much of the cellular heterogeneity seen in most cancers. Moreover, many tumors probably have their origins in normal stem cells. The identification of CSCs and the factors that regulate their behaviour is likely to have a large impact on the way neoplasia is treated in the future. Models of skin cancer in rodents carried out in the 1950s strongly suggested that cancers had their origins in long-lived epidermal stem cells, but the idea that cancers themselves might have CSCs is only just gaining widespread acceptance, despite the fact that Hamburger and Salmon, using colony formation in soft agar as a surrogate stem cell assay, found that for many human tumors, only 1 in a 1,000 to 1 in 5,000 cells were able to form macroscopic colonies.1

Markers of “normal” stem cells are being actively sought with the expectation that many of these molecules will also be found in their malignant counterparts. Collectively these molecules appear to be involved in maintaining “stemness”, ensuring adhesion to the niche and cytoprotection- ensuring survival of these key cells.2 For example, hematopoietic stem cells (HSCs) have high levels of aldehyde dehydrogenase (ALDH), a detoxifying enzyme that confers resistance to alkylating agents such as cyclophosphamide.3 In the bronchiolar epithelium, stem cell function appears to be the property of rare pollutant-resistant cells linked to a deficiency in the phase I drug metabolising enzyme CYP450 2F2.4 In the liver too, cells known as small hepatic progenitor cells (SHPCs) can proliferate despite exposure to pyrrolizidine alkaloids because of a similar deficiency in cytochrome P450.5

A common defence mechanism adopted by stem cells is their high expression of ABC membrane transporters. [An informative website compiled by Michael Muller's group at Wageningen University, The Netherlands ( currently lists 49 human ABC transporters organised into 7 subfamilies; A-G.] These proteins are characterized by expression of an ATP-binding cassette region that hydrolyses ATP to support energy-dependent substrate exportation against steep concentration gradients across membranes, principally from the intracellular cytoplasm to the extracellular space. This property was exploited by Goodell et al. who reported a method for the isolation of HSCs based on their ability to efflux a fluorescent dye.6 Similar to the activity of the P-glycoprotein (encoded by the mdr1 gene), this could be inhibited by verapamil. If cells are subjected to Hoechst 33342 dye staining and fluorescence activated cell sorting (FACS) analysis, then those that actively efflux the Hoechst dye appear as a distinct population of cells known as the “side population” (SP- see Fig. 1 and legend). In this issue of HEPATOLOGY Chiba and colleagues7 report that two of the four hepatocellular carcinoma (HCC) cell lines that they studied had SP cells comprising 0.25% and 0.8% of the cell population. These cells were highly proliferative and relatively resistant to apoptosis in vitro. Microarray analysis indicated that several genes implicated in ‘stemness’, e.g. Wnt pathway genes, were substantially upregulated in the SP cells in comparison to non-SP cells. The SP phenotype in rodent and human tissues often appears to be determined by the expression of a protein known as the ABCG2 transporter (ATP-binding cassette [ABC] subfamily G member 2, also known as breast cancer resistance protein [BCRP1]),8 however in Chiba's study7 the SP phenotype seemed not to be related to ABCG2, but rather to several other ABC transporters, notably ABCB1 (MDR1). Interestingly, hepatic progenitor cells (HPCs) that are found in the reactive ductules seen in human parenchymal liver disease strongly express MDR1, together with MRP3 (ABCC3) and MRP1 (ABCC1),9 while in the rat, oval cells activated in response to 2-acetylaminofluorene and partial hepatectomy can have high levels of ABCG210 and MRP3 and MRP1.11 This expression alone does not necessarily equate with “stemness”, as HPCs/oval cells are essentially a transit amplifying population derived from an as yet uncharacterized stem cell population, thought to be located in the canal of Hering.12 Chiba's study7 adds to the growing belief that the SP fraction more or less equates with the stem cell population in both normal tissues13 and tumors.14, 15 Moreover SP cells may contribute to cancers relapsing and developing drug-resistance due to their drug-effluxing properties.

thumbnail image

Figure 1. The experimental protocol: The cells are stained by the Hoechst 33342 dye and then separated by Flow Cytometry; those small proportion of cells that are able to efflux the dye have a characteristic fluorescent emission profile in dual wavelength analysis, with a low red fluorescence but a high blue:red ratio due to a low dye load — the SP fraction. In the present study7 this property was at least partly based on high expression of ABCB1 (MDR1). The subcutaneous transplantation of 103 SP cells consistently gave rise to tumors, while 106 non-SP had no tumor initiating ability.

Download figure to PowerPoint

Cancer may arise from the dedifferentiation of mature cells that have retained the ability to divide, or from the ‘maturation arrest’ of immature stem cells.16 Most observers now accept that the arrested differentiation of tissue-based stem cells or their immediate progenitors, so called “blocked ontogeny”, is linked to the development of not only teratocarcinomas and hematological malignancies, but also carcinomas. In the liver, the consensus is that a variety of intrahepatic stem cells can give rise to HCC and cholangiocarcinoma (CC).17 There may be at least four distinct cell lineages susceptible to neoplastic transformation, however many human liver tumors arise during cirrhosis when hepatocyte senescence triggers the activation of HPCs. Also some liver tumors have features of both HCC and CC, combined with the presence of numerous HPCs — consistent with an origin from HPCs or their antecedents. Indeed, gene expression profiling has identified a subset of HCCs with a poor prognosis that have a profile consistent with a HPC origin.18 The question arises, have the malignant SP cells in Chiba's HCC cell lines7 originated from HPCs or their precursors, and moreover can SP cells be identified in normal liver? SP cells have been isolated from normal liver, but their location in situ is uncertain, not least because of the widespread distribution of ABC transporters in the liver19: clearly the vast majority of cells in the liver expressing ABC proteins are not stem cells.

Long-term in vitro expansion of murine fetal liver cells has resulted in 5% to 20% of the population being CD34/CD45-negative SP cells,20 while from adult mouse liver Wulf and colleagues21 isolated an SP fraction of which 75% were CD45+. Nevertheless both the CD45+ and CD45 fractions showed hepatocyte differentiation both in vitro and in vivo when transplanted into periportally-damaged livers. Although a few of the murine SP cells expressed Thy-1 and c-Kit, none expressed biliary cytokeratins or the oval cell marker OV6; in contrast over half of the human SP cells isolated by Chiba et al.7 expressed both CK19 and alpha-fetoprotein — perhaps indicating an origin from a bipotential HPC or its predecessor. Transplanted murine hematopoietic SP cells can also contribute to liver regeneration, with a pattern of engraftment suggesting they give rise to the periductular null cells.21 The normal human liver also has an SP fraction of which most (90%) are CD45+, but again both CD45-positive and -negative populations can show hepatocyte differentiation in vitro.22 Thus, the CD45 status of liver SP cells seems to have little bearing on their potential for hepatocyte differentiation, though mouse liver CD45+ SP cells have the same hematopoietic potential as bone marrow SP cells23 — it clearly depends on the niche.

With the development of non-obese diabetic/severe combined immunodeficient (NOD/SCID) mice for the xenografting of human tumors, came the first good in vivo evidence for the existence of CSCs.24 Bonnet and Dick showed that only a small subset of AML cells was capable of producing leukemic progenitors and leukemic blasts upon transplantation into immunodeficient mice, resulting in a disease phenotype identical to the donor: the putative CSCs had the same phenotype, CD34+CD38, as described for primitive HSCs. CSCs have also been demonstrated in various solid tumors. For example in the breast, the existence of stem cells was inferred by the clonogenicity (ability to form mammospheres) and multipotentiality of a subpopulation of murine mammary cells.25, 26 Then in breast cancer, a subpopulation of cells with the surface antigen expression pattern of ESA+CD44+CD24−/low was shown to be capable of tumor formation when implanted in limiting dilution into immunodeficient mice and of self-renewal as shown by serial transplantation.26 Similarly, Chiba et al.7 using NOD/SCID mice, have found that transplanting 103 liver SP cells consistently yielded tumors, whereas transplantation of 106 non-SP cells failed to give rise to tumors. Likewise in brain cancer, clonogenicity and self-renewal is exclusive to a minor subpopulation of tumor cells expressing the human neural stem cell marker CD133, regardless of the tumor phenotype.27 These cells can differentiate into cells with neural and glial phenotypes in vitro in proportions resembling the original tumor, and their proliferative capacity is also proportional to the aggressiveness of the original tumor. As few as 100 CD133+ cells from medulloblastomas could produce tumors in NOD/SCID mice that were serially transplantable and identical to the original tumour.28 In the brain, CD133 appears to mark both normal stem cells and CSCs, but in the breast the relationship between the CSCs (expressing ESA+CD44+CD24−/low) and normal stem cells (being α6+ CK19+ ESA+ MUC1 CALLA) is less clear.2

As a word of caution, Hill29 has counselled against relying solely on SP sorts for stem cell identification, given that non-SP cells may suffer from Hoescht dye toxicity and thus any potential “stemness” may be masked. Nevertheless, further scrutiny of the SP cells in Chiba's study7 and of those recently isolated from gastrointestinal cell lines30 will further the understanding of liver and gut stem cells. It seems likely that CSCs do exist in most tumors, liver tumors have a subset of CSCs and that further characterization may ultimately allow specific therapeutic targeting.


  1. Top of page
  2. References