Identification and characterization of cancer stem cells | ПРЕЦИЗИОННАЯ ОНКОЛОГИЯ

Identification and characterization of CSCs

Cancer immunology. Bench to bedside immunotherapy of cancersNima Rezaei (Ed.). Springer-Verlag (2015)

Experimental observations on hematopoietical malignancies revealed that a small subset of cancer cells are capable of extensive proliferation [17], which led to the emergence of the CSCs hypothesis. It has been subsequently shown in solid cancers that tumor cells are phenotypically heterogeneous and only a small proportion of cells are clonogenic in culture and in vivo models [18]: these cells could be considered as CSCs as they are the pool of cells within the tumor which sustain malignant growth, similar to stem cells within normal tissues sustaining growth. Noteworthy, CSCs do not forcibly correspond to the “cell of origin” (the normal cell that undergoes initial transformation): In most cases, the “prime” cell presents a phenotype different from CSCs [19].

According to this classical concept, CSCs would share peculiar functional features with physiological stem cells:

  • Unlimited self-renewal: CSC can undergo indefinite division.
  • Differentiation potential: CSCs can generate a differentiated progeny that would constitute the bulk population of tumor cells and contribute to tumor heterogeneity.
  • Transition to a quiescent state: CSCs can switch from a dormant, slow-cycling condition to an activated state. This could account for resistance to standard therapies that target cycling cells and also explain relapses occurring years after completion of conventional therapies.

Nevertheless, this classical concept is still controversial. The landmark studies of Lapidot et al. [20] demonstrated that only a small fraction of human leukemia cells could generate leukemia in severe combined immune-deficient (SCID) mice, whereas the majority of tumor cells failed to engraft. However, these studies can be criticized because xenotransplantation assays could be selecting for cells more ?t to grow in a foreign environment [21, 22]. Although these arguments were partially invalidated by recent experiences of genetic tracing in mouse models which spontaneously developed tumors [23–25], there is a need to find a further correlation with evidences from experimental results, in order to establish solid criteria to isolate and identify CSCs.

Surface markers

Regarding solid tumors, a significant step forward has been the identification of immunophenotypical cell surface markers: Al-Hajj et al. [10] reported that CD44+/CD24-/low fractions from metastatic pleural effusions of primary invasive breast tumor had significantly higher tumorigenic potential compared to CD44+/CD24+ cell fractions in a NOD/SCID mouse model. Following this work, candidate CSC biomarkers were extensively described in a wide range of tumors: examples include the CD34+CD38-/low phenotype of many human acute myeloid leukemia stem cells and the CD133+ or CD15+ phenotype of human brain tumor CSCs [26, 27]. Combinations of different antigens in a panel of molecular surface markers have been tested, in order to improve their reliability, in each histological subtype.

Since the isolation of CD133+ stem cell-like cancer cells from brain tumors, CD133, also known as prominin-1, has been one of the most popular biomarkers in CSC-related research [11]. Its expression as CSC marker has been shown in several tumors, such as colon [5], lung [28], gallbladder [29], hepatocellular [30], and prostate cancer [31].

CD44, a multifunctional class I transmembrane glycoprotein, has been suggested as a candidate marker of stemness in a number of cancers including breast [10], colorectal [32], prostate [33, 34], head and neck [35], bladder [36], ovarian [37], cervix uteri [38], gastric [39], and gallbladder cancers [29] as well as hepatocellular carcinoma [30].

CD24 positivity has been reported, in association with CD44, as a potential phenotype for CSCs in breast [40] and pancreatic cancers [13]. As an exclusive marker, interest has been drawn to CD24 in ovarian [41] and colorectal cancer CSCs [42] and, more interestingly, in hepatocarcinoma CSCs [43].

CD90 has been described as a surface label for CSCs in hepatocarcinoma [9], high-grade gliomas [44], and lung cancer [45]. Other surface markers encompass EpCAM [46], CD166 [47], and CD117 [48].

Nonetheless, this list of markers for CSCs identification has several limitations. In fact, several CSC clones may coexist within primary tumors (intra-tumor heterogeneity) and different tumors might express different sets of CSCs surface markers (inter-tumor heterogeneity) [49]. Moreover, a marker valid for the identification of CSCs by one method may not be specific when using different approaches [50, 51]. In addition, the limited specificity of the markers used to identify CSCs is a major issue: none of the known markers solely mark CSCs and they may need to be used in combination with other markers [4, 5]. It has also been shown that antigen positivity is strongly dependent on the employed technique [52] and is widely Influenced by epigenetic phenomena implied in the regulation of their expression [53]. Surface markers are frequently expressed in a broad variety of malignant and nonmalignant cells at different steps of their differentiation, resulting in a lack of specificity [5]: implementation of a subset of antigenic variants which selectively targets CSCs such as CD44v6 splice variant is currently under evaluation [54].

In addition, correlation between the expression of biomarkers and content in CSCs was not found in melanoma, possibly representing an interesting exception to the CSC model. In this malignancy, a highly enriched CD271+ population was able to develop tumors in Rag22/2cc2/2 mice, while CD271 cells did not [55], suggesting their stem cell-like properties. Nevertheless, further experiences on nude mice invalidated this result, implying that melanomas could follow a stochastic model (where tumorigenicity is a random feature distributed among all tumor cells) rather than a hierarchical model (with a cancer stem cell compartment) of local tumor growth and distant spread [56]. Thereby, in order to overcome the poorly specific surface markers and to replace them with more direct functional markers, investigators focused on enzymes or signaling pathways, involved in the maintenance of CSC properties.

Side population

Early CSC isolation protocols relied on the peculiar ability of a subpopulation of cells, the socalled side population (SP), to exclude the fluorescent dye Hoechst 33342 [57] on fluorescence-activated ?ow cytometric analysis. When dye emission is analyzed on a ?ow cytometer equipped with a 405 nm laser in a twoparameter display of red and blue emission wavelengths, a tailing population (side population) exhibiting dim fluorescence is observed, as compared to the majority of cells with bright fluorescence [58]. This specific SP cell feature is related to the expression of the adenosine triphosphatebinding cassette (ABC) transporter of enzymes, especially ABCG2 [59, 60]. Moreover, depletion of ABCB5 in CSCs was related to reduced tumorigenic activity, suggesting a functional role for this transporter in CSCs [61].


ALDEFLUOR assay detects expression of cytosolic aldehyde dehydrogenase 1 (ALDH1) [62], and it has also been proposed as a method to identify CSCs. Cytosolic aldehyde dehydrogenases (ALDHs) are a group of enzymes involved in oxidation of aldehydes into carboxylic acids: the expression of ALDH1 has been proposed as a putative marker of stemness in normal mammary tissue as well as in breast cancer cells and seems to correlate with the outcome in breast cancer patients [63]. However, ALDH1 does not appear to be a reliable CSC marker in all tumor types [64]. Moreover, it has been suggested that the stem cell population identified by the ALDEFLUOR assay is heterogeneous and must be dissected using additional surface markers [2]. Recent studies have also shown that ALDH1 inhibition enhances expression of a stem cell-like phenotype, suggesting a possible role of ALDH1 in regulating differentiation – likely related to its involvement in retinoic acid synthesis [65] – rather than in maintaining stemness.

Sphere-forming assay

CSCs have the ability to generate nonadherent, three-dimensional (3D) tumor spheres under serum-free conditions in a clonogenicity assay called “sphere-forming” assay, which measures the frequency with which these prospectively isolated cells form colonies when plated at clonal density in nonadherent culture [66, 67]. Originally used for isolation of normal neural stem cells [68], the sphere-forming assay was then adapted to estimate the CSC fraction in various tumors [69–74].

However, this technique has several limitations, such as the possible formation of artifacts due to cell aggregation if cells are plated at a too high density or the likely selection of CSC phenotype made by the sphere assay culture conditions, that can alter the sphere counts and thus confound the interpretation of the obtained results [75].

Signaling pathways

Multiple regulatory networks are suggested to be involved in CSC self-renewal and differentiation (Fig. 8.1): the constitutional activity of these intracellular signaling pathways can be additionally enhanced by interaction with external stimuli from the cancer cell microenvironment, as discussed in the following section.

Cancer Immunology_ Bench to Bedside Immunotherapy of Cancers-Springer-Verlag Berlin Heidelberg (2015) 8.1

Fig. 8.1. Main signaling pathways involved in the molecular mechanisms underlying CSC control. The activation of these pathways following different stimuli results in the enhanced transcription of several genes (i.e., cyclins, c-Myc, EGF, VEGF) involved in physiological cell processes, including cell proliferation, growth, and survival. FZD Frizzled (Wnt receptor), γ-secretase enzyme responsible of the release of the Notch intracellular domain, SHh — Sonic Hedgehog Homolog, DHh — Desert Hedgehog Homolog, IHh — Indian Hedgehog Homolog, PTCH1 — Patched 1 (Hedgehog receptor), Smo — smoothened, Gli — Gli transcription factors, PI3K — phosphatidylinositol 3-kinase, Akt also known as Protein Kinase B (PKB), mTOR mammalian target of rapamycin, JAK Janus kinase, STAT signal transducer and activator of transcription, NF-kB nuclear factor-kappa-light-chain-enhancer of activated B cells, IkB inhibitor of kB, IKK IkB kinase (responsible of IkB degradation)


The Wnt/β-catenin signaling pathway is a pivotal developmental pathway [76], reported to control proliferation vs. differentiation in normal stem cell maintenance and growth [2, 77–80]. Aberrant Wnt activation is a key factor for the initiation and progression of various tumors [81–83]. Furthermore, increasing evidence suggests the involvement of Wnt signaling in the molecular mechanisms underlying CSC control. It has been reported that Wnt pathway triggers a response to DNA damage [84] and that genomic instability may drive the malignant transformation of nontumorigenic stem cells to CSCs [85–87].

Wnt pathway-related genes such as FZD6 and WNT7B are highly expressed in undifferentiated mouse mammary tumor cells that are grown in mammosphere to enrich for progenitor-like cells, compared with the differentiated population [88].

The importance of Wnt signaling pathway in CSC control has been strengthened by Vermeulen et al. [89], who demonstrated that Wnt signaling activation was a marker for colon CSCs. In addition, a role for Wnt pathway in cutaneous CSCs has been highlighted which appeared enriched for Wnt signaling: ablation of the β-catenin gene resulted in the loss of CSCs and complete tumor regression in a model of squamous cell carcinoma [90].

Connections between Wnt signaling and epithelial–mesenchymal transition (EMT)—the process by which cells acquire a mesenchymal identity, losing cell–cell adhesion properties and polarity [91]—have also been suggested in numerous studies, but the exact role of Wnt pathway in promotion or reversal of EMT is still unclear.


The Notch family regroups four single-pass transmembrane protein receptors involved in cell development [92]: aberrant expression and dysregulation of Notch proteins, ligands, and targets has been described in hematological malignancies [93] and in a multitude of solid tumors [93, 94].

Notch is activated via γγ-secretase-mediated cleavage, which releases the intracellular domain from the membrane, allowing it to translocate into the nucleus, where it forms a short-lived transcription complex.

Crosstalk between Notch and other oncogenic pathways has been described to exert a mutual regulation on TGF-α [95, 96], VEGF [97], Wnt [98], and PEA3 [99] pathways. In recent years, Notch activity has been reported to be implicated in the maintenance of CSCs in various cancers [100–108]. The Notch pathway is an important factor in the linkage between angiogenesis and CSC self-renewal; thus, Notch pathway targeting is increasingly considered a therapeutic strategy for cancer treatment, by eliminating CSCs [109]. Accordingly, selective blockage of Notch reduced self-replication and tumor formation capacity of leukemic CSCs [110] and impaired mammosphere formation in vitro [111]. Moreover, the inhibition of Notch pathway in association with trastuzumab has been proven effective in preventing tumor relapse in ErbB2-positive xenograft murine model of breast cancer [112]. Additionally, accumulated evidence has demonstrated that Notch signaling might contribute to cancer metastasis [113].


The hedgehog (Hh) signaling pathway coordinates development of tissue progenitors and expansion of stem cells [114]. Pathway activation is initiated by binding of one of the three ligands, Sonic (SHh), Desert (DHh), and Indian Hedgehog (IHh), to the patched receptor (PTCH1), disabling its constitutive repression of smoothened (Smo), that leads to activation of the Gli transcription factors [115] and, hence, to the enhanced transcription of several genes involved in cell proliferation, such as cyclins, c-Myc, epidermal growth factor (EGF), and vascular endothelial growth factor (VEGF) [116].

Aberrant Hh signaling has been described both in individuals affected by predisposing genetic syndromes [117] and in tumors displaying sporadic mutations involving PTCH loss [118]. Emerging data from many human tumors suggest that Hh is required to maintain selfrenewal, proliferation, and tumorigenic potential of CSCs in a complex fashion involving both intracellular signalization and interaction with differentiated tumor cells and with the microenvironment [119–121].

Hh signaling has been found to be preferentially activated in clinical specimen-derived colon [122] and breast [116] CSCs, as evidenced by the increased expression of GLI1, GLI2, and PTCH1.

Accordingly, inhibition of GLI1 reduced proliferation of breast CSCs [26], and loss of Smo led to depletion of chronic myeloid CSCs [2]. Furthermore, inhibition of Hh pathway has been proven effective in a pancreatic tumor xenograft model [2].


The phosphatidylinositol 3-kinase (PI3K)/Akt/ mammalian target of rapamycin (mTOR) signaling pathway is a key regulator of physiological cell processes, including growth, cell proliferation, and survival [123–125]. mTOR signaling aberrant activation is frequently observed in human cancers [126]. Recent studies showed that mTOR pathway may be involved in the regulation of CSC biology, notably cell cycle progression and survival [123, 127].

Furthermore, Akt1/2 proteins were more expressed in mammosphere cells than in more differentiated cells [116]. Akt downregulates glycogen synthase kinase-3β (GSK-3β), thus enhancing β-catenin-induced CSC self-renewal. Hence, inhibition of the Akt signaling pathway could be an effective tool to reduce CSCs [128].

STAT and NF-kB

Several studies have demonstrated that chronic Inflammation is a key factor in initiation and progression of various cancers [129–131].

Pivotal molecular links between Inflammation and cancer are the signaling pathways of signal transducer activator of transcription (STAT) and nuclear factor (NF)-kB.

STAT and NF-kB are crucial transcriptional regulators of activation of genes associated with cell proliferation, angiogenesis, metastasis, and suppression of apoptosis [132–134].

TNF-α activates NF-kB by phosphorylation of the inhibitor IkB by IKK. After the dissociation of the inhibitor IkB, free NF-kB migrates into the nucleus and activates the expression of downstream genes, some of which are antiapoptotic.

Recent studies demonstrated that IL-6/JAK2/ STAT3 pathway is required for the maintenance of CD44+CD24CSCs in breast cancer [135] and that ovarian CSCs are characterized by constitutive activation of NF-kB [136]. Therefore, these pathways have been suggested as potential therapeutic targets in cancer stromal cells, in tumor cells, and also in CSCs [137–139].

Accordingly, it has been reported that STAT3 inhibition disrupts proliferation and maintenance of glioblastoma stem cells [140, 141], reduces CD133+ALDH+ colonic CSCs—thus affecting colonosphere formation [142]—and reduces the frequency of ALDH+ CSCs in prostate cancer [143]. Similarly, inhibition of the NF-kB pathway results in CSC apoptosis and induces cell death in the chemoresistant ovarian CSCs [139].

Actually, the identification of signaling pathways is an active area of investigation, but at present, none of these pathways showed a CSCspecific activity.

Although combinations of these different markers have improved reliability to some extent, their limited specificity is a major obstacle to a definitive validation: advances in transcriptional and proteomic profiling could be helpful to provide more reliable tools to identify CSCs [144].