Increasing evidence indicates that effective anticancer drugs should target cancer-specific tumor-initiating cells, which are functionally and morphologically different from their bulk tumor counterparts. In our previous studies we found that in several analyzed clinically aggressive metastatic colon cancer specimens, as well as in the established colon cancer HCT116 cell line derived from a poorly-differentiated colonic adenocarcinoma, the majority of cells expressed low-to-moderated levels of CD133, CD44 and CD166 [
20]. However, FACS analysis clearly showed the presence of the minority cell population with highest levels of combined expression of the CD133 and CD44. In contrast to the major bulk cell population, CD133
high/CD44
high phenotype possessed high tumorigenic and clonogenic capacities after serial transplantations of the low cell numbers, and was able to produce cells of all the original phenotypes, showing characteristic stem cell plasticity [
54]. Original cell phenotypes were retained in both mice tumor xenografts and floating colonospheres induced by CD133
high/CD44
high cells, suggesting that they represent the tumor-initiating, or colon CSCs in a HCT116 cell line. Moreover, both mice tumors and spheroids induced by highly purified CSCs contain higher ratios of cells with original transplanted phenotypes compared to original sources, which is in line with other studies [
55]. Genome-wide gene expression analysis with the Affymetrix DNA microarray assay revealed that in a CD133
high/CD44
high cell population, many genes related to drug resistance and stemness, including
NOTCH, Shh, Wnt, Oct4, were significantly upregulated compared to their bulk counterparts [
20]. In the present study we further characterized this colon CSC phenotype using pathway-specific gene expression profiling with PCR Array assay (SABiosciences) in the three independent colon cancer cell lines, HCT116, HT29 and DLD-1. We have found that CD133
high/CD44
high cell phenotype was present in all these cell lines, possessed relatively high sphere-forming capacity, and induced spheroids which expressed upregulated levels of the majority of stem cell-related genes compared to their differentiated counterparts. However, HCT116 cell populations revealed significantly higher expression of the stemness genes and higher sphere-forming potential compared to HT29 and DLD-1.
There is accumulating knowledge that tumor-initiating cells in metastatic colon cancer may not be demarcated solely by the expression of CD133 [
18,
56]. Although both CD44 and CD133 were reported as putative markers for many cancer-specific CSCs, including colorectal cancer (described in Introduction), it remains to be resolved whether they are of equal functional importance and what are their interrelationships. A recent study has demonstrated the unique role of CD133 in the normal and malignant colon, showing that CD133
+ normal stem cells at the base of crypts in the adult intestine (a stem cell niche) not only generate the entire intestinal epithelium, but give rise to all the neoplastic cells in mice colon tumors [
57]. It was shown that the proportion of CD133
+ cells in colon cancer metastases is higher than in primary tumors [
58], which reflects the well known fact that metastatic lesions are more resistant to treatment. However, another study has shown that only a knockdown of CD44, but not CD133, strongly prevented clonal formation and inhibited tumorigenicity in mice xenograft model [
59]. Authors reported that CD44
+ did not colocalize with CD133
+ cells within colorectal cancer. Similar results reported by Horst and colleagues showed that the expression of CD133 correlates with that of CD166, while both do not correlate with CD44 [
60]. However, this data contradicts multiple reports which not only show colocalization of the CD133 and CD44 in several types of human cancer [
13,
17‐
20,
61], but also suggest their combined expression as the best CSC marker [
18,
61]. Such inconsistency may be due to the high heterogeneity of clinical specimens, diversity of the experimental approaches, and lack of the specific CSC markers.
Therefore, we have selected the CD133
high/CD44
high cell phenotype to study CSC-targeted activity of a new-generation taxoid, SB-T-1214 in three independent colon cancer cell lines. Given that this compound led to complete remission in
in vivo colon cancer xenograft model (Pgp+ DLD-1), we hypothesized that it could modulate some stemness genes/signaling pathways. The traditional
in vitro model, the monolayer of the bulk tumor cells, poorly represents normal physiological conditions and has limited relevance to the hierarchically organized
in vivo tumors. In particular, this model is not optimal for stem cell-based studies, because even highly purified CSCs can undergo relatively fast differentiation after being placed in standard adherent cultures with serum-containing medium, and therefore, observed drug responses will be related to the bulk tumor cells, not CSCs. However, in our previous studies we have found that short-term culturing of repeatedly sorted cells on type I collagen-coated surfaces in serum-free stem cell medium led not only to the retaining, but to significant increase of the ratios of the tumor-initiating cell phenotypes [
19,
20]. This data is in line with a recent study showing that human colorectal carcinoma cells grown on type I collagen in serum-free medium undergo an epithelial-mesenchymal-like transition and downregulation of E-cadherin and β-catenin at cell-cell junctions [
52]. Authors have found that collagen type I inhibited cell differentiation, increased clonogenicity and promoted expression of CD133 and Bmi1, indicating that it promoted expression of a stem cell-like phenotype in colon cancer cells. In this context, culturing the adherent to a type I collagen CSC-enriched cell population in a serum-free stem cell medium can provide a useful tool for the preliminary evaluation of the CSC-focused drug responses. However, altered cell-to-cell and cell-to-matrix contacts in monolayer cell cultures can affect their signal transduction pathways and response to treatment [
62‐
65]; therefore, we studied the SB-T-1214 cytotoxicity using an alternative model: free-floating colon cancer spheroids induced by the purified colon cancer tumorigenioc cell phenotypes. A three-dimensional model of cancer spheroids was established by Sutherland and colleagues [
66,
67] more than three decades before the discovery of CSCs in human cancers. It is now well documented that this model is more closely related to original tumors with respect to cell morphology, metabolic and proliferative gradients, oxygen and drug penetration and cell-cell junctions compared to the traditional cancer cell monolayers [
66‐
70]. In addition, cancer spheroids are organized hierarchically, similarly to the
in vivo tumors, containing a relatively small population of the tumorigenic cells and a large spectrum of their progenitors and differentiated cells, which comprise a bulk mass of the tumor or spheroid [
70‐
72]. Spheroid cultures favor the proliferation of undifferentiated cells and can be passaged for many generations, reflecting the fact that they contain a population of cells with profound self-renewal capacity. We have shown that both mice tumor xenografts and 3D spheroids induced by more purified CD133
high/CD44
high cells contained much higher ratios of cells with the original phenotype, and possessed higher tumorigenic and clonogenic potentials [
19,
20]. Similar findings were reported for metastatic breast cancer cells [
73] and prostate cancer cells lines [
16]. Many of the commonly used anti-cancer drugs induce about 20-fold lower cytotoxicity against 3D cancer spheroids compared to monolayer cultures [
74], and exhibit chemoresistance which recapitulates this resistant phenotype
in vivo[
39,
62,
74‐
78]. In this context, SB-T-1214 revealed promising cytotoxicity against CSC-induced floating colon spheroids. In contrast to earlier studies on 3D cell cultures, which were usually limited to the relatively short-term gross evaluation of the inhibition of spheroid growth and apoptosis, we have analyzed specific stem cell-related responses. Thus, using a stem cell-focused PCR array assay we have found that single treatment of the 3D spheroids induced by CD133
high/CD44
high cell populations with relatively low concentrations of the SB-T-1214 resulted not only in significant down-regulation of the majority of stem cell-related genes, but more importantly, led to a dramatic reduction of their sphere-forming capacity. In addition, the expression of several key regulators of pluripotency of the embryonic stem cells, Oct-4, Sox-2, Nanog, Lin-28 and c-Myc, was also inhibited after single treatment with 100 nM of SB-T-1214 for 48 hours. The synergistic action of these tissue-specific transcription factors is a pivotal mechanism for determining cellular phenotypes and self-renewal of embryonic stem cells. Thus, the introduction of four genes (Oct-3/4, SOX2, c-Myc, and Klf4) into adult fibroblasts can transform them into pluripotent stem cells [
50,
51]. It has been demonstrated later that Oct-3/4 and SOX2 are crucial transcriptional regulators whose absence makes induction of pluripotency impossible, and additional genes, including Klf4, c-Myc, Nanog and LIN-28 are important for the induction efficiency [
79]. There is growing data that the pluripotency markers are also expressed by CSCs. Thus, the CD133-positive cells isolated from lung cancer tissues possessed the characteristics of stem-like cells and malignant tumors and expressed Oct4 [
80], which was linked to chemo- and radioresistant properties in lung cancer-derived CD133
+ cells. Expression of Oct3/4 and Sox2 was also associated with an unfavorable clinical outcome [
81]. Of interest, down-regulation of Lin-28 can lead to the rapid production of the mature Let-7 g miRNAs, which are required for cell differentiation [
82]. Therefore, significantly decreased levels of expression of stem cell-related genes in general, and shut-down of several major players, including Oct-4, Sox-2, c-Myc and Lin-28 after treatment with SB-T-1214 is promising, because it most likely means that treated CSCs in colonospheres were promoted to a more differentiated state. This data is important in the light of growing evidence indicating that standard anti-cancer therapies often promote self-renewal of tumor-initiating cells and further resistance to treatment [
28‐
30]. CSC resistance to treatment may be due to several mechanisms, including overactivated ABC transporters, high capacity for DNA repair and activated anti-apoptotic machinery [
43‐
45]. It was recently demonstrated that treatment with 5-FU and oxaliplatin, a standard therapy for metastatic colon cancer, induced up to 30-fold enrichment of CD133+ and up to 2-fold enrichment of CD44+ cells in HT29 cell line [
31]. This data is in line with our observation that after a single treatment with 100 μM Paclitaxel for 24 hr, the clonogenic potential of the dissociated HT29 and DLD-1 spheres cells was significantly increased, so we can assume that post-treatment spheroids contained a higher proportion of putative colon CSCs compared to untreated spheroids. Since we have studied the SB-T-1214 induced alterations in the stemness gene expression profiles using total cell lysates (equal amounts of the total RNA for PCR arrays and total protein for western blot analyses), the significant inhibition of the stem cell-related genes induced by SB-T-1214 is promising.