Cancer cells were previously considered to accumulate genetic mutations over time in a single population of cells, resulting in a more aggressive population. However, cancer stem cells were recently discovered in cancer cell populations, and have been suggested to possess properties that produce tumor cell heterogeneity and resistance to drugs and radiation [1]. Although cancer stem cells were shown to express specific cell surface markers, a functional analysis of these markers has been delayed due to the very low proportion of cancer stem cells in tumor cell populations [1].

CD133, also called prominin 1, was originally identified as a cell surface marker of human hematopoietic stem cells and mouse neuroepithelial cells [2-4]. It was subsequently reported to function as a marker of cancer stem cells in solid tumors, such as brain tumors [5], colon cancer [6,7], hepatocellular carcinoma (HCC) [8], and neuroblastoma [9]. The CD133-positive cell population has a greater self-renewal ability and chemoresistance phenotype than the CD133-negative cell population. The expression of CD133 correlates with malignant characteristics and a poor prognosis in many tumors [10]. CD133 is phosphorylated at its intracellular C-terminal domain by Src family tyrosine kinases [11], and, as a result, activates the p85 subunit of phosphoinositide 3-kinase (PI-3K) by binding. PI-3K, in turn, activates downstream targets, such as Akt, thereby promoting the proliferation of glioma stem cells [12]. CD133 is stabilized by binding with histone deacetylase 6 (HDAC6), and enhances the transcriptional activity of β-catenin, resulting in the acceleration of cell growth and suppression of cell differentiation [13].

Here, to elucidate the new mechanisms contributing to the functions of CD133, we searched for cancer cell lines expressing high levels of CD133, and identified Huh-7, a cell line for hepatocellular carcinoma, and SK-N-DZ, a cell line for neuroblastoma. Immunofluorescent microscopy analyses showed that CD133 predominantly localized to centrosomes rather than the plasma membrane in Huh-7 and SK-N-DZ cells [14]. Further analyses revealed that when Src family kinase activity was low, CD133 was not phosphorylated by Src, and, as a result, unphosphorylated-CD133 was down-regulated in cells and preferentially interacted with HDAC6, whereby CD133 endosomes were transported to centrosomes by intracellular transport via the dynein motor [14] (Figure 1). Centrosomes are major microtubule-organizing center (MTOC) in animal cells and have key roles in regulating cell polarity and motility as well as spindle formation, chromosome segregation and cytokinesis [15]. More recently, centrosomes have also been reported to be involved in regulation of autophagy [16].

We then investigated the biological functions of centrosome-localized CD133. The findings obtained revealed that centrosome-localized CD133 captured GABARAP, a key regulator of autophagy initiation [16-18], and inhibited the GABARAP-mediated activation of ULK1 and subsequent initiation of autophagy [14] (Figure 2). The CD133 amino acid sequence (828-831 a.a. (Y-D-D-V)), including the phosphorylation site (Y828) by Src kinase, was also conserved as an LC3B-interacting region (LIR: Y/F-X-X-V), which has alternatively been recognized as a GABARAP-interacting motif [19]. Therefore, CD133 may capture GABARAP via this potential LIR.

We also examined the physiological functions of centrosome-localized CD133. Autophagic activity is necessary for cell differentiation in stem cells, such as neural stem cells and embryonic stem cells [20]. Our findings showed that centrosome-localized CD133 suppressed cell differentiation, such as the formation of primary cilia and neurite outgrowth, in neuroblastoma cells [14] (Figure 2).

Thus, we demonstrated that centrosome-localized CD133 has a new important function: it plays an important role in maintaining cancer cells in an undifferentiated state by inhibiting autophagy. Based on our study, in the near future, the search for molecular compounds that directly induce cancer stem cell differentiation may open the door to novel therapies.

We are grateful to Drs. Y. Kaneko and A. Nakagawara for their valuable discussions. We also thank our laboratory members for their continuous encouragement. This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan, Takeda Research Support, Japan and Gold Ribbon Network, Japan.

H. I. wrote the manuscript.

The author declares no competing interests.