Reply to Yoshida: Liver cancer stem cells: Identification and lipid metabolic reprogramming

As a molecular basis by which acyclic retinoid (ACR) prevents recurrence of hepatocellular carcinoma (HCC), we identify MYCN⁺ liver cancer stem cell (CSC)-like cell serves as a biomarker and therapeutic target to be selectively eliminated by ACR (1). In the Letter by Yoshida (2), the author makes two important comments on the identification of CSCs and lipid metabolism in CSCs that deserve further discussion.

CSCs are defined by capabilities of self-renewal, differentiation, and tumor initiation. Expression of several cell surface markers as well as alterations of cellular functions such as gaining an aldehyde dehydrogenase activity are used in identifying CSCs. However, given the plasticity of stem cells such as interconversion of the different stem cells and dedifferentiation/transdifferentiation of differentiated cells (3), it is not appropriate to define CSCs by a single marker or functional property. Reports on homeostasis/regeneration (4) and tumorigenesis (5) of the liver raised a question whether the stem cell-like cells were derived from preexisting liver stem cells adjacent to blood vessels or dedifferentiated from the differentiated hepatocytes throughout the lobules of liver under stress. Recent lineage tracing studies gave further controversial conclusions (6). We agree with the statement by Varga and Greten (3) that stemness could be considered as a state acquired but not defined by a distinct cell type or a cell biomarker. In our study, EpCAM⁻ cells eliminated by ACR expressed MYCN at much lower levels compared with those in EpCAM⁺ cells (1). Probably, subpopulations of EpCAM⁻ cells may contain other CSC-like cell types, and EpCAM⁺ cells may be in a state abundant with MYCN expression. Furthermore, according to The Cancer Genome Atlas database, it was the gene overexpression but not amplification of MYCN that correlates with the prognosis of HCC (1). Therefore, we proposed that MYCN expression represents a dynamic stem state of the liver CSC-like cells (1).

Lipogenesis plays remarkable roles in support of reprogrammed cell growth and survival driven by both MYCN (7) and PTEN/mTOR (8) signaling pathways. Mechanistically, enhanced lipogenesis maintains the rebalanced nutrient supply and demand for cellular hyperproliferation. Of interest, administration of an unsaturated fatty acid (UFA), oleate, partially rescued the cell death induced by knocking down a Myc superfamily member, MondoA, while inhibitors of UFA synthesis were toxic to Myc-overexpressing cells (7). Recent studies further shed light on the roles of UFAs in maintaining cancer cell stemness and promoting tumorigenesis via activating NF-κB (9) and Wnt-β-catenin (10) signaling pathways. We reported that inhibition of accelerated UFAs (such as linoleic acid and its metabolites) was important in the prevention of a carcinogen-induced hepatic tumorigenesis by ACR (11). Here, we further provide evidence obtained from a proteome analysis that the fatty acid (FA) desaturases such as SCD1, FADS1, and FADS2 were enriched in EpCAM⁺ CSC-like HCC cells compared with EpCAM⁻ cells (Fig. 1). Hence, we agree with Yoshida’s proposal that UFA metabolic reprograming may serve as a therapeutic target against CSCs (2).

Fig. 1.

Proteomics characterization of EpCAM⁺ liver CSC-like cells. The EpCAM^+/− cells were sorted from a HCC cell line JHH7 (1). Tryptic digests of whole-cell protein were analyzed by data-dependent nano-liquid chromatography–tandem mass spectrometry (nLC-MS/MS) (EASY-nLC and Q-Exactive; Thermo Scientific), and the obtained data were analyzed by Proteome Discoverer 2.2 (Thermo Scientific) using MASCOT program (Matrix Science). (A) Volcano plot of 4,110 proteins detected using nLC-MS/MS. The arrow indicates the increased EpCAM protein expression in EpCAM⁺ cells. (B) Top diseases’ or functions’ annotation associated with 123 differentially expressed proteins between EpCAM^+/− cells (fold change more than 2 and adjusted P value less than 0.05) performed in Ingenuity Pathways Analysis platform. The pathways were ranked according to their activation z score, which can be used to predict either activated or inhibited state based on a statistically significant pattern match of up- and down-regulation, respectively. The pathways related with lipogenesis were highlighted in red. (C) Quantitative data of enzymes involved in FA synthesis including ACLY, ACACA, and FASN as well as desaturases including SCD1, FADS1, and FADS2 in EpCAM^+/− cells detected using nLC-MS/MS. The mean averages from three biological replicates ± SD are shown.