1. Introduction
Malignant tumors are one of the most prevalent causes of death world-wide. According to Globocan (
https://gco.iarc.fr/today/home (accessed on 16 December 2021)) in 2020, the estimated number of new cases reached almost 20 million, while the number of deaths was close to 10 million. Breast, colon, lung, and ovarian cancers are the most often diagnosed tumor type in women over 40 years of age. The highest morbidity and mortality in men is observed for lung, prostate, and colon cancers. Therefore, many scientists have been focused on these solid tumors with regard to disease etiology and treatment. It was found that cancer cells are heterogenic, and that only a small fraction are responsible for tumor development and metastasis. It was demonstrated that these cells are able to initiate the tumor growth when implanted in mouse hosts. Hence, they are called tumor-initiating cells or cancer stem cells (CSCs)
[1].
The history of CSCs ability to drive tumorigenesis is long, but the real breakthrough came in the 1990s and 2000s
[2][3][4][5][6]. The low number of CSCs, occurring at a frequency of 1–100 in 10
−6 [2] is a serious obstacle for cell isolation and similarly, and non-specific markers of CSCs of different origin remain challenging for cell identification
[7]. However, it was found that CSCs show many specific features, such as stemness (proliferation capacity and self-renewal), phenotypic plasticity, metabolic reprogramming, and drug resistance
[8][9]. With regards to phenotypic plasticity, epithelial-mesenchymal transition (EMT) activation is linked to the formation of CSCs
[1]. However, the underlying mechanism is still unclear. CSCs are also known to perform in immune evasion from immunotherapy. The recruitment of M2 macrophages, Tregs, dendritic cells
[10], and involvement of non-coding RNAs may perform a role in suppressing tumor development
[11].
In recent years, intensive studies on epigenetic changes in malignant tumors have been conducted, including the role performed by non-coding RNAs
[12][13][14]. Understanding the regulation of gene expression by miRNAs (microRNAs), lncRNAs (long non-coding RNAs), and circRNAs (circular RNAs) has been of particular interest. These non-coding RNAs can play both oncogenic and suppressor roles in cancer and exhibit tumor-specific expression
[15][16][17]. The strong influence of CSCs microenvironment on their biology, including therapy resistance, is an important topic for scientific research
[18].
It is also important to consider the state-of-art in CSC biology, especially in the context of their remarkable plasticity and some aspects of underlying epigenetic mechanisms and potential targets for future therapeutic strategies against CSCs.
2. Biology of CSCs and Their Interaction with Tumor Microenvironment
CSCs show complex biology. It is the effect of specific features and crosstalk with tumor microenvironment
[19][20][21][22][23][24]. Key features of CSCs, including stemness, self-renewal, phenotypic changes strongly connected with EMT phenomenon, metabolic reprogramming, and ability to become invisible to immunological system are presented in
Figure 1. CSCs plasticity, phenotypic and metabolic, is their ability to dynamically change under microenvironment conditions. It is the result of different mechanisms regulated by both cell intrinsic and extrinsic factors
[25]. These signals influence the expression of various genes in CSCs (
Table 1). BMI1 and Sox2 can regulate plasticity and pluripotency
[26][27]. The EMT can be induced by SOX9
[28]. Tumor hypoxia triggers metabolic reprogramming and phenotypic plasticity
[29]. Oxygen insufficiency causes dimerization of HIF-1α and HIF-1β to form the HIF-1 complex
[30]. Silencing of HIF-1α can suppress the expression of stem cell genes, in particular OCT4, SOX2, NANOG, and KLF4, thus preventing the progression of cancer
[31].
Figure 1. The features and mechanisms of CSCs biology. Abbreviations: MDR—multi-drug resistance proteins; TME—tumor microenvironment (Created with BioRender.com (accessed on 7 January 2022)).
Hypoxia in the tumor microenvironment is common in advanced cancer and is related to poor prognosis and a worse survival rate. A growing body of evidence has discovered that hypoxia can promote cancer cell invasion, metastasis, and EMT. All these events can promote stemlike characteristics in cancer cells. Hypoxia-inducible factor 1 (HIF1) is a vital molecule in the regulation of CSCs. HIF1 is involved in tumor growth, immune evasion, and metabolic reprogramming. Thus, HIF1 appears to play an essential, if not critical, role in the formation and preservation of CSCs
[31].
2.1. CSCs Signaling Pathways
There are few signaling pathways activated in CSCs. The most commonly disturbed signaling cascades in CSCs are phosphatidylinositol 3-kinase/Akt/mammalian target of rapamycin (PI3/Akt/mTOR) pathway, Janus kinase/signal transducers and activators of transcription (JAK/STAT) pathway, and Wnt and Notch
[45].
2.2. CSCs Plasticity, EMT and Dormancy
Phenotypic plasticity of CSCs can be found as ability to switch states (phenotypes) as a response to tumor microenvironment conditions
[46]. A strong relationship exists between phenotypic plasticity, stemness, and EMT
[24][25]. During EMT, cells show changes in their morphology and expression of genes towards a stem-like state
[25]. EMT and ECM remodeling have a strong impact during tumor angiogenesis. CSCs are able to form vessel-like structures, which were described as vasculogenic mimicry (VM)
[47]. The mechanism of VM is not dependent on endothelial cell proliferation and VEGF. It is strongly connected with phenotypic changes
[48]. CSCs transdifferentiate to cells with some features of endothelial cells. It was demonstrated that cancer cells involved in VM show expression of CD133, CD44, ALDH, and Sox2
[49][50], but also VE-cadherin and CD31
[19][51]. The meta-analysis of thirty-six studies has revealed that VM is associated with shorter overall survival and is a poor prognostic factor
[52].
The EMT process corresponds to a wide spectrum of biological variances triggering the conversion of cells from epithelial to the mesenchymal state. Cells that undergo EMT, acquire migratory and invasive properties
[53]. Epithelial cells exhibit more proliferative features, while mesenchymal cells have an increased tendency for migration and ability to impact stroma through matrix metalloproteinases (MMPs)
[40][54]. Moreover, epithelial cells show overexpression of markers, such as E-cadherin or members of miR-200 family, while for the characterization of mesenchymal cells markers, such as N-cadherin, vimentin, or fibronectin, can be used
[53]. The signaling pathways like Wnt, Notch, Hedgehog, or Myc act as EMT inducers and are related to stemness properties of CSCs, and impact migration and invasion
[55]. Extracellular EMT inducers comprise TGF-β, EGF, Axl-Gas6 pathway, hypoxia, and ECM elements. In addition, transcription factors (TFs) Twist1, Snail1, and Zeb1/2, T-box TF Brachyury promote EMT
[56].
EMT was first observed by Greenburg and Hay in 1982
[57] and named epithelial-mesenchymal transformation. EMT phenomenon is crucial for normal embryonic development, but also is linked to several pathological processes, including wound healing, fibrosis, and cancer progression
[58]. In the case of tumor cells which can express typical markers for both cell states (epithelial and mesenchymal), the term transformation was replaced with transition
[59]. Decades of studies have revealed that cancer cells are able to form not only fully epithelial or fully mesenchymal cancer cells, but also various hybrid E/M (intermediate states). In 2020, the EMT International Association (TEMTIA) published a work with the current status of the knowledge about EMT and nomenclature
[60]. According to this data, EMT should be treated as the ability to progress along the epithelial–mesenchymal axis and to adopt different intermediate hybrid E/M states
[59][60]. The hybrid phenotype is critical for the maintenance of tumorigenicity of basal breast cancer cells. Highly tumorigenic cell population with expression of CD104/CD44 cell surface antigen and transcription factors Zeb1 and Snail1 was isolated in a hybrid E/M state. For this cell population increased expression of Snail and Wnt signaling pathway was also observed
[61].
Phenotypic plasticity of cancer cells and their ability to undergo EMT and metastasis can be modulated by tumor microenvironment. There are many extracellular factors determining the plasticity of CSCs
[62]. It is the result of crosstalk leading by CSCs and cellular components of tumor microenvironment, like cancer-associated fibroblasts (CAFs) and macrophages.
CAFs can modulate CSCs plasticity via different signaling pathways; for instance, in lung cancer by IGF-II/IGF1R signaling pathway
[63] and in hepatocellular carcinoma through c-Met/FRA1/HEY1 signaling
[64]. In prostate cancer, CXCL12 expressed by CAFs interacts with CXCR4 on tumor cells, induces EMT, and promotes metastasis
[65]. The crosstalk between CAFs and cancer cells promoting their phenotype changes and metastasis was also observed in breast cancer
[66]. The enhancement of stem cell features appeared by the activation of the Notch mechanism. Macrophages can secrete factor Oncostatin-M, an IL-6 family cytokine, which activate the dedifferentiation of triple negative breast cancer cells into aggressive stem cells
[67]. With regards to the crosstalk between cancer cells and macrophages in a tumor, studies in silico co-culture models have revealed that macrophages (M1 and M2 phenotypes) can alter the epithelial vs. mesenchymal state of cancer cells. These results may be helpful for efficient therapeutic strategies.
EMT is strongly implicated in tumor relapse. According to the studies of Sun et al. conducted on prostate cancer, EMT can be induced by androgen deprivation, which is the first-line therapy
[68]. It is probable that the feedback loop involving the androgen receptor and the Zeb1 transcription factor are responsible for the transition. The castration-resistant prostate cancer is a major clinical problem, and these results seem to be crucial in terms of medical implications for second-line treatment of castration-resistant prostate cancer. The most recent studies of Guo et al. showed that Numb protein performs an important role in xenograft prostate tumor growth and castration-resistant prostate cancer as a suppressor of CSCs Notch and Hedgehog signaling
[69]. The inhibition of the Notch and Hedgehog signaling pathways significantly increases apoptosis in Numb
−/low cells in response to androgen-deprivation therapy.
CSCs show their plasticity not only in ability to EMT. These cells can use adaptive and protective mechanism known as dormancy
[70]. In this state cancer cells stop proliferating. Clinically, cancer dormancy is very difficult to detect, and it is defined as remission time. Two categories of dormancy in cancer can be distinguished. First, cellular dormancy, means that each cancer cell shows cell cycle arrest. The second is tumor dormancy when in cancer a balance between growth and apoptosis rates appears. The length of dormancy is long in prostate cancer or in hormone dependent breast cancer, whereas in triple negative breast cancer this period is shorter
[71]. Moreover, it is suggested that dormancy of CSCs can be the result of epigenetic changes in CSCs. The nature of dormancy is reversible, and the epigenetic mechanisms can be responsible for regulating, maintenance, and reactivation of cancer cells from the dormancy
[71]. Non-coding RNAs seem to be very important regulators of dormancy in cancer, e.g., the angiogenesis/dormancy switch
[70]. In addition to non-coding RNAs, there are several other intracellular and extracellular signals involved in the mechanisms of dormancy and reactivation, to the group of dormancy signals belong, e.g., p16, p21, p53, TGFβ2, or BMP4, and for reactivation Pi3K/AKT, TGFβ3, and HIF-1α can be responsible
[72].
Some of anticancer drugs, like fluorouracil, increase the number of dormant cancer cells and enrich the population of CSCs, which leads to chemotherapy resistance
[73]. However, some of chemotherapeutic agents show the opposite action. They can reactivate dormant cancer cells. The studies of Gao et al. on head and neck squamous cell carcinoma have revealed that LB1, an inhibitor of protein phosphatase 2, can enhance the cytotoxic sensitivity to chemotherapy via promoting entering of cancer cells from dormancy into the cell cycle
[72].
2.3. CSCs Metabolic Changes
Metabolic plasticity, as another adaptation to microenvironmental conditions, is one of the most important hallmarks of cancer cells
[74]. Cells can modify metabolism using different energy sources to enhance their survival and maintain homeostasis. With regards to the heterogeneity of tumor cells, CSCs show higher metabolic plasticity compared to normal cancer cells
[20]. In non-stem, highly proliferative cancer cells, glycolytic phenotype is observed with high glucose uptake, low oxygen consumption, low mitochondrial mass, and ROS. These features are typical for the Warburg effect where the predominant pathway for ATP generation is glycolysis
[75]. In glucose-deprived conditions, CSCs tend to shift into a quiescent (non-proliferative) state and depend on OXPHOS to produce ATP. Quiescent CSCs show oxidative phenotype and reverse Warburg effect metabolism with high oxygen consumption. Moreover, CSCs with reduced proliferation are more resistant to chemotherapy, which targets mostly proliferative CSCs
[76]. The changes in metabolism can be observed not only in cancer cells, but also in cells of tumor microenvironment, including upregulation in lactate production and the acidification of the tumor’s stroma
[77].
Switching between glycolytic and OXPHOS phenotype in CSCs is suggested. These cells can exist in a hybrid metabolic state
[78]. Hence, both glycolysis and OXPHOS should be blocked. Metformin, as a glycolysis reducer and OXPHOS inhibitor cane inhibit this metabolic plasticity
[79].
2.4. Epigenetic Regulation of CSCs—The Role of microRNAs, circRNAs and lncRNAs
Epigenetic mechanisms regulate potentially heritable changes in gene expression, not associated with changes in the DNA sequence. Such changes dictate cancer formation and progression, leaving their mark on cancer stem cells. There are three main mechanisms of such gene expression regulation, namely DNA methylation, chromatin modification, and non-coding RNAs
[80][81]. In this research, the researchers focused on the third mechanism, non-coding RNAs. Non-coding RNAs serve regulatory roles in self-renewal, metabolic plasticity, resistance to radio- and chemotherapy, interactions within the tumor microenvironment or formation of secondary disease foci (
Figure 2). They have shown a lot of promise in development of targeted therapies to combat cancer
[82][83][84][85]. The use of non-coding RNAs may enable targeting cancer stem cells, and this may become an effective strategy to eradicate cancer.
Figure 2. Impact of ncRNAs on cancer stem cells’ hallmarks. Abbreviations: EMT—epithelial-mesenchymal transition. ncRNAs with CSCs promoter role and with CSCs suppressor role are presented in red and green, respectively (Created with BioRender.com (accessed on 16 November 2022)).