Background
The tumors contain a sub-population of specific cells which are primarily involved in tumor formation and maintenance [
1]. Those cells are referred as tumor stem-like cells (TSLCs), and CD133 (also known as prominin-1 or AC133) has been considered as an important marker to enrich the stem-like population in tumors of various tissues, including those of the brain [
2], prostate [
3], pancreas [
4], liver [
5], colon [
6,
7], and skin for melanoma [
8]. CD133
+ tumor cells possess the ability to self-renew without limit and to generate the majority of differentiated progenies [
9]. They are also more resistant to chemo- and radiotherapy when compared to CD133
- tumor cells, resulting in tumor progression and recurrence, and thus are considered as a potential therapeutic target to eradicate tumors [
10‐
12]. However, the molecular mechanisms underlying this tumor stemness are still under investigation.
CD133 is a cellular surface glycoprotein containing five transmembrane regions and two glycosylated extracellular loops and has a molecular weight of 97-120 kDa. It was identified as a specific antigen for human hematopoietic stem cells [
13], and is currently used for the isolation of stem-like cells from numerous tissues [
13‐
20]. Although little is known about the biological function of CD133, recent studies have shown that prominin-1 null mice were born and aged normally [
21], but resulted in progressive degeneration of mature photoreceptors with complete loss of vision after postnatal day 15 [
22]. Transcription of
CD133 gene is known to be controlled by alternative five promoters (P1, P2, P3, P4, and P5), and exon 1 produces different spliced 5'-UTRs, which are expressed in a tissue-specific manner [
23]. We previously showed that the methylation status of CpG sites residing in P1 and P2 regions is inversely correlated with the expression levels of
CD133 mRNA in human glioma tissues [
24]; however, any molecules involved in the transcriptional regulation of
CD133 gene are still unknown.
The E26 transformation-specific (ETS) family consists of over 35
Ets genes that can be structurally categorized into 11 subfamilies in humans [
25]. Individual Ets genes share a highly conserved DNA-binding domain composed of about 85 amino acid residues referred to as the ETS domain, which recognizes purine-rich GGA(A/T) (
E TS
b inding
s ite; EBS). Several Ets factors have been shown to be nuclear targets for the activation of the Ras/Raf/MEK/ERK signaling pathway and are involved in various biological processes, including cell proliferation, apoptosis, differentiation, hematopoiesis, tissue remodeling, angiogenesis, metastasis, and oncogenic transformation [
25]. The altered expression levels of Ets or chromosomal amplification, deletion, and translocation are known to cause human leukemia or specific types of solid tumors [
26].
To investigate the molecular mechanism underlying stem-like features in human tumors, we characterized a promoter region of human CD133 gene by using human carcinoma and sarcoma cell lines. We found that the ERK pathway is involved in the expression of CD133, and its inhibition was also shown to decrease the frequency of side population (SP), another hallmark of stem-like cells. Finally, it was revealed that Ras-mediated transformed astrocytes have an ability to form greater colonies in the neural stem cell culture condition, together with the increased CD133 mRNA expression. Thus, our finding could provide important insights into the molecular basis of tumor stemness.
Discussion
Traditionally, therapeutic procedures for human cancer have been performed based on the implicit understanding that the tumor population is homogeneous. However, emerging evidence has suggested that tumors are hierarchically organized and the capacity of tumor propagation depends mainly on a sub-population of stem-like cells. The discovery of stem-like cells in solid tumors convincingly accounts for chemoresistance, and recurrence arose in a number of human cancers. Many studies have been carried out using stem-like population enriched by a stem cell marker CD133, and these have demonstrated an increased resistance of CD133+ stem-like tumor cells to treatment with chemotherapeutic agents compared with CD133- tumor progenies. In addition, the side population has been also used as one of the methods to enrich the stem-like tumor cells, as well as normal stem cells, and is defined by Hoechst dye exclusion property. Although it remains to be clarified whether the expression of CD133 and transporter molecules directly contribute to tumor progression, the regulatory mechanism of stem-related gene expression could help our understanding of tumor stemness and should be investigated further to improve the development of eradicative therapies against human malignancies.
Previously, we and other investigators reported that the expression levels of
CD133 mRNA are positively correlated with tumor stage and the poor prognosis of patients [
24,
41‐
47]. However, it is still controversial whether CD133 is just a concomitant marker for tumorigenic process or whether it directly leads to tumorigenesis. To examine the role of CD133 expression in normal cells, we established NHA/TSC cells and found that overexpression of CD133 is not sufficient to induce tumorigenic transformation
in vivo (Figure
5B and Additional file
1, Figure S7). Interestingly, a recent study using genetically engineered mice suggested that CD133 is just a concomitant marker of stem-like cells. Tumors had developed throughout the entire intestine when Wnt signaling was selectively activated in CD133
+ or Lgr5
+ adult small intestinal stem cells [
21,
48]. In contrast, carcinomas with lower malignancy were found in less than one in five mice when the same system was targeted to non-stem cells [
48]. Therefore, it is conceivable that malignant expansion of tumors depends on the latent potential of adult stem cells to proliferate or differentiate, and thus the increase of
CD133 mRNA in a higher stage of tumors might be a result of unregulated expansion of CD133
+ stem-like cells. Considering that NHA/TSR cells possess the high activity to form neurospheres and tumors in mice without entire expression of CD133 protein (Figure
5), these data would support the view that CD133 is just a marker of stem-like cells.
The expression mechanisms of CD133 gene have not been examined so far, despite its expression being recognized as an important stem-related biomarker for a number of different cell lineages, probably because CD133-expressing cells are a very rare sub-population for transcriptional analysis. In addition, we observed some culture effects, in that serial passages of primary glioblastoma culture easily diminished CD133 expression (data not shown). The two tumor cell lines used in this study stably preserve a high proportion of CD133+-proliferating cells, and they could be useful tools to further investigate the expression machinery for CD133.
In Figure
1D, we have shown that P5 promoter exhibits the highest activity among the five alternative promoters, but it should be noted that P5 does not necessarily predominate the CD133 expression. First, the stability and translational efficiency of
CD133 mRNA might be varied by 5'-UTR sequences containing exon1s (1A, 1B, 1C, 1D, or 1E). Indeed, the modulation of 5'-UTR involves in the regulated expression of some proteins regulating growth and differentiation of normal stem cells and plays a role in the progression of specific types of cancers, such as leukemia and prostate cancer [
49]. Therefore, the role of each exon1 on CD133 expression needs to be determined. Experiments should be limited to the use of the whole locus including all of exon1s for reporter assay.
Second, epigenetic modifications such as DNA methylation and histone modification have been reported to play important roles in the regulation of various genes [
50]. However, it is questionable that the physiological status of chromatin complexes is accurately reconstituted in the transient reporter assay system. Indeed, we reported that the expression levels of
CD133 mRNA were dramatically restored by the treatment of glioma cell lines with the demethylating agent 5-azacytidine and/or histone deacetylase inhibitor valproic acid [
24]. Therefore, it might be possible that epigenetic modification is the final determinant of
CD133 gene expression and stem-like features. Further studies are needed to address the molecular mechanisms to epigenetically maintain the active state of
CD133 gene, containing demethylases of DNAs and/or acethyltransferases of histones.
The Ras/ERK pathway plays a crucial role in transducing signals from various external stimuli to control cell adhesion, proliferation, migration, and survival. It is well-known to be deregulated in some types of human tumors [
51]; Ras mutations are found in 45% of colon carcinomas and 90% of pancreatic cancers; Raf mutations are found in two-thirds of melanomas, where TSLCs have been enriched by sorting for CD133 protein expression. The Ras pathway was also reported to be involved in stem cell regulation. Undifferentiated human embryonic stem cells (hESCs) require a growth factor FGF, which transmits signals mainly by the Ras/ERK pathway [
52]. Indeed, ERK is active in undifferentiated hESCs and inhibition of the ERK pathway with U0126 caused extensive cell death and differentiation [
53].
It is noteworthy that myeloproliferative disorders could be initiated by K-Ras(G12D) in a highly restricted population enriched for hematopoietic stem cells (HSCs) [
54]. The expression of activated M-Ras in HSCs initiated leukemogenic transformation [
55]. In addition, sophisticated mouse tumor models triggered by oncogenic Ras have demonstrated the contribution of the Ras/ERK pathway to the acquisition of cancer stem cell properties, in primary human mammary epithelial cells (HMECs) [
56]. H-RasV12 also causes the p53-knockout mouse-derived astrocytes to be transformed into brain tumor stem-like cells, in which MEK/ERK pathway is responsible for neurosphere formation [
57]. Interestingly,
let-7 microRNA, known to downregulate Ras, reduces proliferation, sphere formation, and the proportion of undifferentiated cells
in vitro and tumor formation and metastasis
in vivo [
58]. These reports support that the Ras/ERK pathway plays a central role to acquire and maintain tumor stem-like properties.
Furthermore, we have shown that ERK inhibition diminishes the frequency of the side population, but we could not reveal the detailed mechanisms. RT-PCR analysis revealed that the expression of ABCC1 mRNA was decreased by U0126 treatment, while that of ABCG2 mRNA was increased in Caco-2 cells (data not shown). Although it has not been elucidated which transporters have a dominant role for SP phenotype in Caco-2 cells, the ERK pathway might regulate multiple events at the upstream of the SP phenotype, including gene transcription and protein stabilization. Actually, it has been reported that inhibition of the Ras/ERK pathway also promotes ABCB1 protein degradation to diminish the cellular multidrug resistance in the human colorectal cancer cell lines [
59]. SP cells are also known to reproduce NSP cells [
35]. Therefore, U0126 might affect the division patterns of SP cells via the decrease of self-renewal and/or the increase of differentiation. In the future, we will clarify these points, and should examine whether the therapeutic approaches to Ras/ERK inhibition could be effective for eradication of stem-like cells in tumor mass.
In this study, we found that the Ras/ERK pathway is implicated in at least one of the stem-like characteristics in all of the examined tumor cell lines: SP size in Caco-2, CD133 expression in Fuji, and sphere and tumor forming activity in NHA/TSR cells. These finding suggest that Ras/ERK could be a common upstream pathway to govern the entirety of downstream characteristics. However, the contribution of Ras/ERK was varied in a cell type-specific manner; Ras/ERK does not contribute to CD133 protein expression in Caco-2 and NHA/TSR, and SP appearance in Fuji and NHS/TSR cells. Crosstalk between ERK and other pathway specific to an individual type of tumors may result in the diversity of stem-like hallmarks.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
KT designed the research, performed most of the experiments and data analyses, and drafted the manuscript. TK carried out MACS to separate CD133high and low Fuji cells and helped animal experiments. KS helped soft agar assay with human astrocytes and the editing of manuscript. LW helped cell cultures and plasmids extractions and performed animal experiments. NB helped ChIP assay. HN, TT, and ST contributed to the design of the entire study and the editing of the manuscript. All the authors have read the manuscript and agreed with its content.