The work of Otto Warburg on cancer cell metabolism back in the dawn of the 20th century revealed the capability of cancer cells to cover metabolic needs via the pathway of glycolysis, even in environments with adequate oxygen supplies ^[1][74]. This form of aerobic glycolysis is known as the Warburg effect and holds great biomedical significance, constituting the principle of metabolicallyactive lesion detection via positron emission tomography (PET) imaging ^[2][75]. An evolving body of evidence supports the hypothesis that glycolysis not only serves as a key metabolic pathway in CSCs, but also can be utilized at even higher rates than in cancer cells ^[3][76]. The increased rate of glucose uptake by CSCs, the overexpression of genes coding for enzymes of the glycolytic machinery, as well as the enhanced production of ATP and lactate constitute evidence that supports the aforementioned statement ^[4][77]. Additionally, glucose depletion seems to be associated with compromised viability of CSCs in vitro ^[4][77], while environments rich in glucose enrich the tumor with CSCs ^[4][77]. Glycolysis has been observed in glioblastoma ^[5][78], osteosarcoma ^[6][79], breast ^[7][80], ovarian ^[8][81], lung ^[4][77], and colon CSCs ^[9][82]. Although glycolysis is generally less efficient in producing ATP than OXPHOS, it provides CSCs with significant metabolic intermediates that are utilized in a plethora of metabolic pathways, the most important of which is PPP, in order to meet complex bioenergy needs ^[10][83]. For instance, glucose-6-phosphate (G6P), produced from glucose phosphorylation by glucokinase, is channeled into PPP to generate NADPH ^[11][84] or ribose groups essential for nucleotide synthesis ^[12][85]. Similarly to what happens in iPSCs ^[13][86], glycolysis also plays a pivotal role in the adoption of a pluripotent phenotype by CSCs, as supported by evidence from breast ^[14][87], nasopharyngeal ^[15][88], and hepatocellular carcinomas ^[16][89]. Several different regulatory mechanisms were identified as playing a role in controlling glycolysis in CSCs. Firstly, the finding that MYC orchestrates CSC glycolysis aligns with its role in iPSCs ^[13][86]. Secondly, CD44, a stemness marker that binds to hyaluronic acid (HA) and mediates adherence to the extracellular matrix (ECM) ^[17][90], also participates in controlling glucose metabolism in CSCs ^[18][91]. Last but not least, glucose itself has the ability to induce the upregulation of glycolytic enzymes, such as hexokinases 1 and 2 (HK1, HK2), pyruvate dehydrogenase kinase 1 (PDK1), as well as the GLUT1 transporter, regulating the insulin-independent glucose influx ^[4][77]. Glycolysis upregulation was shown to convey resistance to radiotherapy in nasopharyngeal CSCs ^[15][88] and to chemoembolization in hepatocellular CSCs ^[19][92], possibly due to the enhanced feedback of PPP with G6P and the increased levels of NADPH that are able to ameliorate the detrimental effects of ROS produced by radiation.

While the glycolytic metabolic phenotype offers a survival benefit in glucose-rich environments, providing CSCs with the ability to cover their proliferative energy needs, their localization within the tumor core does not always provide glycolysis-favoring conditions. By adopting a slow-cycling state of quiescence, CSCs are able to employ the OXPHOS machinery to cover their energy needs ^[20][93]. This is especially prominent in the paradigms of glioblastoma ^[21][94], leukemia ^[22][95], lung cancer ^[23][96], and pancreatic ductal adenocarcinoma (PDAC) ^[24][97]. The utilization of OXPHOS by CSCs is supported by their high mitochondrial mass and negative mitochondrial transmembrane potential, along with high rates of oxygen consumption and ROS generation ^{[21][22][24][25][26][27][28]}[94,95,97,98,99,100,101]. Importantly, the adoption of the OXPHOS phenotype is involved to a great extent in the resistance to chemotherapy and targeted therapies that some CSCs possess. CSCs resistant to KRAS-targeting therapy in the PDAC context, as well as leukemic stem cells (LSCs) resistant to BCR-ABL-targeting in chronic myeloid leukemia (CML), were found to express an OXPHOS phenotype ^[29][30][102,103]. The most important regulator controlling mitochondrial biogenesis, an essential feature for the preservation of OXPHOS, is peroxisome proliferator-activated receptor-gamma coactivator-1alpha (PGC-1a) ^[31][104]. Besides PGC-1a, other regulators of the OXPHOS phenotype have been identified. In LSCs, the increased OXPHOS activity is under the influence of the adrenomedullin–calcitonin receptor-like receptor axis and spleen tyrosine kinase ^[32][33][105,106]. Another important regulator is nuclear factor erythroid 2-related factor 2 (NRF2), which was associated with a high mitochondrial potential, leading to more efficient OXPHOS, and was shown to play a role in the survival and treatment resistance that CSCs exhibit ^[34][35][36][107,108,109].

The acquisition of either a glycolytic or an oxidative CSC metabolic phenotype is not exclusive. Instead, an increasing number of studies over the past years shed light on the metabolic plasticity that CSCs possess, which works towards the adoption of an intermediate glycolytic/OXPHOS phenotype with the ability to switch between the two. One of the most important pieces of evidence supporting this is the ability of breast CSCs to switch to OXPHOS when glycolysis is inhibited ^[14][87]. The precise control of the enzymes involved in glycolysis, OXPHOS, and the TCA cycle with the ultimate goal to generate ATP and preserve the NAD+/NADH ratio leads to a great metabolic plasticity ^[11][84]. NAD has been given significant attention as a factor controlling CSC plasticity ^[37][110], and there is evidence suggesting that the PGC-1a/MYC balance can affect CSC metabolic fate ^[24][97]. It is important to point out that the population of CSCs within a tumor is able to express a heterogenous phenotype, with different metabolic pathways expressed in different CSC subpopulations ^[38][111]. For example, epithelial-like breast CSCs adopt a different metabolic phenotype from mesenchymal-like ones, providing a survival benefit when it comes to adaptation in the presence of stressors ^[39][112].

Besides glycolysis and OXPHOS, lipid metabolism is also drawing attention in the context of CSC metabolism ^[40][113]. Mitochondrial FAO was shown to contribute to the survival and proliferation of CSCs by mitigating the oxidative load through the production of NADPH ^[41][114]. FAO also provides essential metabolic intermediates such as acetyl-CoA and NADH that favor ATP production ^[42][115]. Breast CSCs ^[43][116], LSCs ^[44][117], as well as normal HSCs ^[45][118] employ FAO to cover their bioenergy needs ^[46][119]. Carnitine palmitoyl transferase Ib (CPT1b), catalyzing the localization of long-chain fatty acids in the mitochondria for their subsequent beta oxidation ^[41][114], was found upregulated in breast CSCs under the influence of the JAK/STAT3 pathway, and this effect was mediated by leptin produced by the adjacent adipose tissue ^[43][116], implicating that TME components can modulate CSC metabolism. CPT-1b was shown to enhance stemness and chemoresistance in breast CSCs ^[43][116]. NANOG, a marker of cell stemness, promotes the overexpression of genes coding for enzymes of the FAO pathway ^[16][89]. The inhibition of FAO was shown to specifically tackle the CSC population ^[43][116]. Besides FAO, fatty acid synthesis was observed in pancreatic CSCs ^[47][120]. Lipases participating in fatty acid synthesis, such as ATP citrate lyase (ACLY), acetyl-CoA carboxylase (ACC), and fatty acid synthase (FASN), controlled by the lipogenic transcription factor SREBP1c, were found to be upregulated in CSCs ^[48][49][121,122]. Furthermore, the levels and the phosphorylation of AMPK in CSCs were lower, leading to enhanced lipase activity and an increase in the levels of malonyl-CoA, which serves as a fatty acid synthesis precursor ^[50][123]. Fatty acid storage in lipid droplets (LDs) is essential for their management, and an increased LD content was described in colorectal CSCs ^[51][124]. Additionally, cholesterol synthesis via the mevalonate pathway greatly participates in CSC generation ^[52][125]. Cholesterol is incorporated in membrane lipid rafts, ensuring the smooth operation of signaling pathways essential for CSC proliferation ^[42][115]. Finally, lipidomics showed that fatty acid desaturation, which promotes membrane fluidity, contributes to stemness preservation in ovarian and glioblastoma CSCs ^[53][54][126,127], and the regulation of this process by stearoyl-CoA desaturase (SCD1), NF-κB, and aldehyde dehydrogenase I family member A1 (ALDH1A1) promotes stemness in colorectal CSCs ^[55][56][128,129].

Amino acids can serve as another energy source in CSC metabolism. Glutamine is a key amino acid that is converted by glutaminase to glutamate, which undergoes deamination to alpha-ketoglutarate (aKG) ^[57][130]. aKG generation, along with pyruvate and oxaloacetate, can promote anaplerosis of the TCA cycle in order to balance the constant loss of citrate to mitochondria for lipid synthesis ^[58][131]. Glutamine also provides the essential carbon and amino nitrogen for the generation of other lipids, nucleotides, and amino acids ^[59][132] and regulates major epigenetic modifications ^[60][133]. It was shown that pancreatic CSCs exhibit a strong preference for glutamine ^[61][134], and glutamine metabolism was also implicated in the ability of colorectal CSCs to attach to liver tissue in vitro ^[62][135].

Iron metabolism dysregulation is recognized as another CSC metabolic trait ^[63][136]. Generally, iron abundance in CSCs enhances stemness ^[64][65][137,138]. Iron supplementation was shown to promote stemness in lung carcinoma, breast carcinoma, and cholangiocarcinoma cell lines, whereas iron chelators ameliorated this effect ^[65][66][67][138,139,140]. Additionally, in non-small cell lung cancer (NSCLC) CSCs, iron enhanced tumor invasion via hydroxyl radicals ^[68][141]. CSCs seem to have greater iron needs than cancer cells. In comparison to cancer cells, CSCs showed lower expression of ferroportin 1 (FPN1) and hephaestin, which regulate iron outflow, and higher levels of transferrin receptor (TfR), which controls the iron influx through TF/2Fe³⁺. Similar to glycolysis, the CD44 stemness marker promotes iron accumulation by interacting with TfR ^[69][70][142,143]. Table 1 providesan overview of the similarities and differences between CSC and normal stem cell metabolism.