The Role of Autophagy and lncRNAs in the Maintenance of Cancer Stem Cells

Abstract

Cancer stem cells (CSCs) possess properties such as self-renewal, resistance to apoptotic cues, quiescence, and DNA-damage repair capacity. Moreover, CSCs strongly influence the tumour microenvironment (TME) and may account for cancer progression, recurrence, and relapse. CSCs represent a distinct subpopulation in tumours and the detection, characterisation, and understanding of the regulatory landscape and cellular processes that govern their maintenance may pave the way to improving prognosis, selective targeted therapy, and therapy outcomes. In this review, we have discussed the characteristics of CSCs identified in various cancer types and the role of autophagy and long noncoding RNAs (lncRNAs) in maintaining the homeostasis of CSCs. Further, we have discussed methods to detect CSCs and strategies for treatment and relapse, taking into account the requirement to inhibit CSC growth and survival within the complex backdrop of cellular processes, microenvironmental interactions, and regulatory networks associated with cancer. Finally, we critique the computationally reinforced triangle of factors inclusive of CSC properties, the process of autophagy, and lncRNA and their associated networks with respect to hypoxia, epithelial-to-mesenchymal transition (EMT), and signalling pathways.

Keywords: LncRNAs; autophagy; cancer stem cells (CSCs); haematological malignancies; solid cancers; tumour microenvironment.

Figure 1

The distinction between cancer stem cells (CSCs) and cancer-initiating cells. A cancer-initiating cell (in blue) undergoes oncogenic transformation in order to develop a tumour, while a cancer stem cell (CSC, in dark purple) is not necessarily the transformed tissue-specific stem cell, but rather gives rise to the bulk of the tumour.

Figure 2

Autophagy, a pro-survival process, is elevated in CSCs. (A) The process of autophagy from nucleation to degradation. The process starts with the formation of an isolation membrane (regulated by the Beclin-1 complex), followed by the formation of a double membrane named the autophagosome which encapsulates cellular constituents for degradation. Following fusion with a lysosome, cell components in the autolysosome are degraded and recycled as nutrients and metabolites back into the cytosol. Several autophagy-related (Atg) proteins regulate different stages of autophagy (e.g., the Atg5-12 complex is vital for autophagosome formation). (B) Autophagy plays many roles in CSC survival. High levels of autophagy/mitophagy in CSCs promotes resistance to chemotherapy; potential treatments include autophagy/mitophagy inhibitors in combination with chemotherapy drugs. Induced autophagy stimulates CSC invasion and migration as well as survival in the hypoxic microenvironment. CSCs also depend on elevated mitophagy for their metabolism and to control ROS levels. Finally, CSCs show a mixed phenotype where both OXPHOS and glycolytic metabolism is employed in CSCs of different cancer models.

Figure 3

Long noncoding RNAs (LncRNAs) contribute to CSC features via the regulation of different cellular mechanisms. The three main mechanisms of actions of lncRNAs in CSCs regulation: Top panel: the lncRNA CUDR interacts with TLR4, thereby enhancing its function and promoting the activity of SUV39H2. SUV39H2 is involved in epigenetic regulation that stimulates liver cancer CSC proliferation. Centre panel: the lncRNA SOX2OT sequesters miR-200c, thereby inhibiting its function of targeting SOX2 mRNA. Therefore, SOX2OT indirectly promotes translation of the stemness factor SOX2 in bladder cancer. Lower panel: the genetic locus of PSORC1C3 contains an HRE responding to hypoxia-inducible factor (HIF) transcription factors. HIF triggers a CSC phenotype via PSOR1C3 which interacts with OCT4 mRNA, thereby stabilising it and promoting protein production and function in renal cell carcinoma cells. CSCs: cancer stem cells; HRE: hypoxia response element.

Figure 4

Reactome analysis of stemness proteins, CSC markers, autophagy genes, and lncRNAs discussed in this review (A) Reactome analyses of genes reviewed in this study and other genes that may provide a link between the genes discussed. (B) Reactome analyses performed using only genes reviewed in this study. Both analyses indicate functional interactions drawn from the Reactome database in Cytoscape. In this figure, edges indicate an interaction between two genes or their transcribed products. Dashed edges depict computationally predicted interactions with a probability of 0.95; solid edges indicate a shared protein complex or binding event. Arrowed edges indicate activation or catalytic activity, while blunted arrows indicate inhibition. Gene names in black are genes that were discussed in this review, while genes in red were not discussed in this review but were included because they provide a link between the genes discussed. Genes with no functional interactions are not shown.

Figure 5

Functional and physical protein–protein interactions and DisGeNat analysis. (A) Physical and functional protein–protein interactions in the subset of genes that were reviewed in this work obtained using the STRING database v11.04. Nodes represent proteins, while the thickness of edges indicates the strength of the data obtained from a modified naïve Bayes algorithm. The evidence underlying the network was restricted to interactions in humans that were derived from curated databases (Biogrid), in addition to physical experiments, and co-expression (excluding text-mining, predicted, and orthologous evidence); 51 nodes, 70 interactions, 15 expected interactions, p-Value < 10⁻¹⁶. (B) A network generated by DisGeNet, depicting common diseases for which the network of genes was enriched. This information was obtained from the DisGeNet databases, using the ClusterProfiler package in R. The comparison includes a subset of 47 protein-coding genes discussed in this review that were curated in the DisGeNet database. The top 20 most enriched diseases were labelled, with the size of their corresponding gold node representing the adjusted −log10 p-Value of enrichment, while unlabelled grey nodes represent genes implicated in these cancer states. In total, 191 disease states were enriched (Bonferroni corrected p-Value < 0.005).

Figure 6

Genemania interrogation of transcription factor and miRNA targets. Transcription factor and miRNA target prediction from MSigDB using GeneMania. Targets are displayed in red, while genes are displayed in blue.

Figure 7

The contribution of autophagy and lncRNA to stemness factors in the tumour microenvironment (TME). In response to hypoxia, HIFs and their targets are triggered that can contribute to stemness factors. In turn, hypoxia-driven HIFs can trigger autophagy, as a means to survive the harsh conditions that and contribute to stemness factors. LncRNAs such as H19, HOTTIP, and CCAT2, could directly contribute to stemness factors.

Figure 8

The interlinkage between the triangle of CSCs, lncRNAs, and autophagy with respect to epithelial-to-mesenchymal transition (EMT). Hypoxia in the TME can trigger SNAI1 and TWIST1 and relevant signalling pathways. STAT3 activation can contribute to EMT, while it can activate TWIST1. Further, STAT3 and HIFs directly or in combination can contribute to CSC factors. LncRNA MALAT-1 can promote EMT and CSC factors. Autophagy mediators such as Beclin-1, P62, and Atg4 can contribute to both EMT and CSC factors.