Warburg effect in colorectal cancer: the emerging roles in tumor microenvironment and therapeutic implications

The increased rate of glycolysis is a common metabolic change that occurs in cancer (Fig. 1A). It has been observed that active glycolysis in cancer is achieved by the upregulation of glycolytic enzymes and transporters. The initiation of the glycolysis process requires the transportation of extracellular glucose to the cytoplasm and this is achieved by sodium-glucose linked transporters (SGLTs) and facilitated diffusion glucose transporters (GLUTs) [17]. Next, a series of enzymes take part in aerobic glycolysis of which hexokinase (HK) catalyzes the conversion of glucose to glucose-6-phosphate (G6P), the first irreversible reaction in glycolysis [3]. Hexokinases (HKs) have five isoforms, namely HK1, HK2, HK3, HK4, and HKDC1 (hexokinase domain-containing protein 1) [18, 19]. Further, it has been reported that HK family genes were hypomethylated and exerted extensive CNV amplification in CRC [20]. Moreover, the expression of HK family genes was dysregulated and has been associated with survival in multiple kinds of cancers [20]. HK2 is the most well-characterized gene whose expression is significantly upregulated in many cancers such as prostate cancer, breast cancer, lung cancer, renal cancer, liver cancer and colorectal cancer [21‐25]. Apart from performing metabolic functions, HK2 could also exert non-metabolic functions by binding to mitochondria to inhibit apoptosis and translocating into the nucleus to increase glucose uptake [26, 27].

Next is the conversion of glucose-6-phosphate to fructose-6-phosphate (F6P) which is a reversible reaction and the F6P can be catalyzed into fructose-1,6-biphosphate (F1,6BP) by phosphofructokinase1 (PFK-1), which is considered as the second rate-limiting step in glycolysis [28]. Multiple factors can regulate the activity of PFK1 among which fructose-2,6-bisphosphate (F2,6BP) is the most powerful allosteric activator [29]. The production of F2,6BP is completed by 6-phospho-fructo-2-kinase/fructose-2,6-bisphosphatase (PFKFB/PFK2), a bifunctional enzyme responsible for both degradation and synthesis of F2,6BP [30, 31]. PFKFB1, PFKFB2, PFKFB3, and PFKFB4 encode four different isozymes of PFK2 and the abnormal expressions of these isoenzymes are observed in a series of tumors except PFKFB1 [28, 32]. PFKFB3 has the lowest phosphatase/kinase ratio, thus has advantage to generate F2,6BP and increase the glycolytic flux of cancer cells [33]. Moreover, several oncogenic pathways, including Ras and mTOR pathways have been reported to regulate cancer metabolism by promoting the overexpression of PFKFB3 [34].

Further, F1,6BP is converted into 3-phosphoglycerate (3PG) via glyceraldehyde-3-phosphate (G3P), glycerate-1,3-diphosphate (G1,3DP) by aldolase and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) [35]. Then, 3PG is converted to 2-phosphoglycerate (2PG) and phosphoenolpyruvate (PEP) through enolase [35]. Pyruvate kinase (PK) was responsible for catalyzing PEP into pyruvate, the final committed step of glycolysis. Pyruvate kinase (PK) has isoforms, namely liver PK (PKL), red blood cell PK (PKR), PKM1, and PKM2 [36]. The human PKM gene has 12 exons and could be alternatively spliced to produce different transcripts [37]. It has been studied that serine/arginine-rich splicing factor 3 (SRSF3) could remove exon 10 from PKM mRNA and generate PKM1 transcript while some oncogenic splicing factors such as heterogeneous nuclear ribonucleoprotein A1 and A2 (hnRNPA1, hnRNPA2) remove exon 9 to form PKM2 transcript [37]. However, the PKM1 exhibits a persistent high-glycolytic tetrameric form while PKM2 exists in either a low-glycolytic dimeric form or tetrameric form depending on the environment and cellular state [38]. PKM2 is overexpressed in the majority of cancers and promotes tumor development through various mechanisms, both metabolic and non-metabolic, and is the most deeply researched isoform of PK [36, 39].

The end product of glycolysis, pyruvate can enter the TCA cycle or be reversibly transformed to lactate by lactate dehydrogenase (LDH), which simultaneously oxidizes NADH to NAD⁺ [40]. In humans, the LDH family has a total of 5 isoforms and are tetrametric enzymes consisting of M and H subunits encoded by Ldh-A and Ldh-B, respectively [41, 42]. The more number of M subunits in tetramer will favor the glycolytic nature of LDH isoenzymes, indicating higher efficiency to convert pyruvate to lactate (LDH5/LDHA); conversely, LDH1/LDHB has four H subunits which favors the conversion of lactic acid to pyruvate, entering enter TCA cycle [41, 42]. However, the upregulation of LDHA can be found and is correlated with tumor progression in many cancers [41]. In addition, the upregulated LDHA level is considered a prognostic factor in a series of cancers, such as pancreatic cancer, breast cancer, renal cancer, lung cancer, and liver cancer [40, 43‐48]. Finally, lactate is transported out of cells which relies on the monocarboxylate transporter/MCT family [49]. Meanwhile, varying degrees of MCT family upregulation were observed in tumors to adjust massive production of lactic acid, thus avoiding or mitigating tumor acidosis [49].

Elevated glycolytic metabolism is a common event in cancer cells because tumors undergo competition due to limited and shared nutrients with stromal cells and the immune compartment and a higher rate but lower yield of ATP production achieved by glycolysis could gain a selective advantage [50‐52]. Thus, cancer cells can fasten their glycolysis rate and compensate for the energy gap. This phenomenon is so conspicuous that positron emission tomography (PET) can diagnose the recurrence and metastasis of cancer [53, 54]. High consumption of glucose by cancer cells significantly reduces its availability in TME, forming a low-glucose extracellular environment and disturbing the function of immune cells [55]. Meanwhile, cancer cells secret abundant glycolysis-derived lactate, resulting in acidosis in microenvironment. Lactate is not a matter of metabolic waste. Actually, lactate can be further utilized by some non-tumor cells and a proportion of tumor cells [56]. Acidic TME also promotes local invasion, metastasis and dampens the anti-tumor function of immune cells [56‐58]. Further, lactate can take part in signal transduction by functioning as an intracellular mediator or an extracellular ligand to bind to some receptor such as GPR81 [56, 59].

The Warburg effect is also regarded as an adaptative mechanism besides producing lactic acid to synthesize biomacromolecules and meet the urgent growth needs of tumors. The glucose in the tumor microenvironment can be used as a carbon source for anabolic processes such as the de novo synthesis of nucleotides, lipids, and proteins [60‐62]. For instance, G6P can further enter into pentose phosphate pathway (PPP) to promote the synthesis of ribose, an indispensable ingredient of nucleotide. At the same time, this process allows cancer cells to transform NADP + into NADPH, which is an essential coenzyme in lipid metabolism [60]. 3PG and pyruvate can be transformed into serine and alanine, respectively. Serine can further take part in one-carbon metabolism and produce glutathione and NAPDH [55]. NAPDH also plays an important role in maintaining redox homeostasis by upregulating the level of GSH. Hence, Warburg effect can enhance the synthesis of NADPH to encounter the excessive oxidative stress in tumor cells [55, 63]. Apart from NADP + and NADPH, NADH and NAD + are also important factors that are used to transport electron in the mitochondria to regulate redox potential in tumor cells. Conversion from pyruvate to lactate requires NADH, which regenerates NAD + to avoid the excessive accumulation of NADH [2].

Glycolytic metabolism also influences epigenetic modifications in cancer [64]. Pyruvate that derived from glucose could be transformed into acetyl-CoA, which could acetylate histones and further regulate the transcription of certain DNA [64, 65]. Histone acetylation makes it easier for transcription factors to bind to DNA and promote cell growth [66, 67]. Histone acetylation is easily affected by cellular signaling and the nutrition status of cells [66]. Cells with high glycolysis rate often maintain the NAD + /NADH ratio at a low level, which could enhance histone acetylation by inhibiting the activity of sirtuins [68, 69]. When nutrient is deprived, NAD + level can be upregulated which functions as a cofactor of sirtuins to promote deacetylation [70]. As for methylation, high level of 3PG increases the synthesis of serine. Serine could regulate methylation by linking with the folate cycle, which is coupled to the methionine cycle [64]. Recently, a novel epigenetic regulation termed histone lactylation has been reported by Zhang et al. and the study revealed that lactate-derived lactylation of histone lysine (K) residue can promote gene transcription from chromatin [71]. These findings suggest that glycolytic metabolism significantly affects the epigenetic landscape of tumor.

The tumor endothelial cells (TECs) are located next to the bloodstream but predominantly produce ATP via aerobic glycolysis to meet their emergent growth needs [171, 172]. Contrast to tumor cells whose enhanced glycolysis relies largely on the oncogenic alternations, endothelial cells are more susceptible to extracellular signals and metabolites, which means their glycolytic phenotype and proliferative ability can be controlled by tumor cells [173]. For example, an elevated level of HIF1α in cancer cells could subsequently activate the transcription of vascular endothelial growth factor (VEGF) which promotes the formation of new blood vessels. Enhanced aerobic glycolysis induces excessive production of lactate by both cancer cells and endothelial cells, leading to the extracellular accumulation of lactate [172, 173]. Moreover, elevated lactate production could be transported into endothelial cells to promote the formation of new blood vessels [171, 174]. Under normoxia, lactate was transported into endothelial cells through the MCT1 transporter and induced the activation of HIF-1 which could enhance the expression of basic fibroblast growth factor (bFGF) and vascular endothelial growth factor receptor 2/VEGFR2 to induce angiogenesis [175]. Intracellular lactate can also drive the phosphorylation or degradation of IκBα, thus stimulating the expression of NF-κB [176]. NF-κB activation could further induce the upregulation of IL-8 to support angiogenesis and tumor growth in CRC cells line [176]. Meanwhile, extracellular lactate can function as a signal molecule that could activate the G-protein coupled receptor GPR81 and it has been observed that the expression of GPR81 is upregulated in multiple kinds of cancers including colon cancer [177]. Lactate-induced GPR81 activates the downstream PI3K/Akt-CREB pathway, enhancing AREG production which is a pivotal protein in GPCR-induced angiogenesis [178]. Further, lactate could stimulate macrophages to secret a series pro-angiogenesis factors and regulate the growth of endothelial cells indirectly [179]. Recent studies have indicated that tumor-associated macrophages (TAMs) can take up lactate through their MCTs, followed by the lactate-induced activation of HIF-α, thus enhancing transcriptions of VEGF [180, 181]. In colon cancer, the inhibition of LDHA significantly reduced the extracellular level of lactate which leads to an inhibition of TAM-derived VEGF and tube formation [181].

Immune cells undergo metabolic change to support their anti-tumor effect. Like tumor cells, some proliferative immune cells apply aerobic glycolysis as one of their metabolism programs to promote their rapid growth [168]. Antitumor CD4⁺ T cells and CD8⁺ cells are indispensable for the human adaptive system and their anti-tumor function relies on aerobic glycolysis [182]. Early upregulation of aerobic glycolysis in T cells is mediated by TCR signaling which promotes PDK1 activity in a transcription-independent manner [183]. PDK1 could phosphorylate and inhibit PDH, a gatekeeper that controls the pyruvate flux into mitochondria and promote the metabolic shift from the TCA cycle to glycolysis [183]. CD4⁺ T and CD8⁺ T cells further maintain their upregulated glycolysis level through CD28-mediated activation of PI3K–AKT signaling [183‐185]. Activated CD8 + T cells upregulate the cell membrane expression of glucose transporters, especially GLUT1, to increase the transportation of glucose into cells to meet the growth needs [183, 185]. Besides CD28 co-stimulation, insulin, adipokine leptin and some cytokines such as IL-2 and IL-7 can also induce GLUT1 expression through activating AKT in CD8 + T cells [186]. It has been noted that the transcriptional regulation of aerobic glycolysis is mediated by MYC and HIF on activation of anti-tumor T cells [187, 188]. MYC and HIF work independently and facilitate the transcription of glycolysis-related enzymes and transporters. Glycolytic metabolism also plays a central role in supporting the anti-tumor function of effector T cells. Upregulated glycolytic level in both CD4⁺ T and CD8⁺ T cells can activate branching pathways such as PPP and serine-one carbon pathway to generate enough intermediates to support biosynthesis and their anti-tumor functions. For instance, enhanced flux of PPP promotes the generation of NADPH, which is essential to lipid metabolism and membrane synthesis in CD8 + T cells [189]. In activated CD8 + T cells, glycolysis-derived acetyl-CoA can enter into mevalonate biosynthetic pathway, providing essential components for the synthesis of sterols and ubiquinone, and substrates for protein isoprenylation [190]. Glycolytic enzymes can also function as RNA-binding proteins to bind to the mRNA of several cytokines, which was first reported in CD4 T cells where GAPDH binds to the mRNA of cytokines such as TNFA, IFNG and IL-2 and controls their translocation [191]. In quiescent CD8 + T cells, LDH can bind to the 3’UTR of TNF, IFN-γ and IL-2 mRNA [183]. Upon activation, LDH can release their target mRNAs, allowing mRNA translation and cytokine production [183].

Aerobic glycolysis is also irreplaceable in innate immune cells because the innate immune cells require abundant energy sources to exert their anti-tumor functions. Such as NK cells rely on glycolysis to kill tumor cells [192, 193]. The number, viability, and cytotoxicity of NK cells in lung cancer are restrained due to an increased FBP1 expression, a gluconeogenesis-related enzyme that facilitate gluconeogenesis and inhibit glycolysis [192]. Some studies have indicated that inhibition of FBP1 significantly restored the glycolytic activity in NK cells and enhanced their cytotoxicity and cytokine-induced activation [192]. Other cells, including inflammatory/M1-like macrophages and dendritic cells (DCs) also experience a metabolic switch to a more glycolytic phenotype upon activation [194‐196].

Warburg effect in tumor and non-tumor cells leads to an increased lactate production which is exported to the extracellular environment, causing high acidity in TME. This causes intracellular acidosis and suppresses the anti-tumor function of immune cells, leading to the loss of immune surveillance and progression of multiple kinds of cancer, including colorectal cancer [56]. Lactate had an inhibitory effect on CD8 + T cells [197]. Murine CD8 + T cells cultured in acidic media could uptake lactate, which leads to intracellular acidification [197]. The intracellular acidification can further inhibit the upregulation of NFAT, an essential transcription factor during CD8 + T cell activation and reduce the production of IFN-γ [197]. Proteomic analysis revealed that the protein levels of most glycolytic enzymes exert negative correlation with CD8 + T cell infiltration in deficient mismatch repair (dMMR)/microsatellite instability-high (MSI-H) CRC [198]. Further, MSI-H CRC samples with a higher glycolytic activity tend to be infiltrated with a less amount of CD8 + T cells, suggesting that glycolytic activity could be helpful to guide the clinical application of immune checkpoint inhibitor (ICI) or predict the outcomes of CRC patients who are receiving immunotherapy [198]. Also, the immune cells can differentiate into a more immunosuppressive phenotype, such as Tregs and M2-like macrophages to adapt to the harsh environmental conditions [199]. Moreover, lactate from CRC cells inhibits the phagocytic ability of TAMs by inducing the Ap-2α/Elk-1 axis which elevates the protein level of Sirpα, an immune checkpoint that negatively regulates the anti-tumor function of TAM [200]. On the contrary, the downregulation of lactate decreases the number of tumor-infiltrating Treg and myeloid-derived suppressor cells in CRC, thus improving the efficiency of immune therapy [201]. Further, lactate could also function as an agonist for the G-protein-coupled receptor 81 (GPR81) to deliver cell signaling. Lactate can activate GPR81 in cancer cells which leads to PD-L1 upregulation and tumor evasion [202]. Similarly, tumor-derived lactate can also induce activation of GPR81 in immune cells and activation of GPR81 in dendritic cells suppressed cell-surface presentation of MHCII and decreased the production of cAMP, IL-6, and IL-12 [203]. It has also been reported that GRP81 signaling induction impairs the pro-inflammatory ability of TAMs and activates the immunosuppression of myeloid-derived suppressor cells [204, 205]. Emerging evidence has proved the unique role of lactylation in regulating immune cells' function under hypoxia and acidic environment [206]. In CRC, the lactate accumulation in the tumor microenvironment can induce the METTL3 expression in tumor-infiltrating myeloid cells (TIMs) via histone lactylation [207]. METTL3, a m6A “writer’’, can activate the JAK/STAT pathway which leads to immunosuppression of TME and tumor progression [207]. The discovery of histone lactylation has provided new insights into the metabolic regulation of transcription and will expand our horizons on cancer treatment. In summary, these results have indicated that a higher concentration of lactate plays an important role in remodeling immunosuppressive tumor microenvironment.

It is important to note that the tumor microenvironment is distinct from the metabolic immune microenvironment under physiological conditions. As suggested by Otto H Warburg, highly proliferative tumor cells and disordered vascular cause a decrease in intratumoral glucose concentration, resulting in a competitive environment between immune and cancer cells. Under glucose deprivation, the phosphorylation of p70S6 kinase and eIF4E binding protein 1 were reduced, which could inhibit the synthesis of IFN-γ and impair the anti-tumor function of effector CD8 + T cells [208]. The microarray analysis further discovered that glucose restriction leads to the decrease of IFN-γ, GM-CSF, Perforin, Granzyme C and Cyclin D2, indicating that the anti-tumor function of CD8 + T cell was significantly impaired in a glucose-deprived environment [209]. Glucose deprivation in tumor cells attenuates the glycolytic ability of T cells, leading to inhibition of the mTOR pathway [210]. mTOR pathway plays an important role in the regulation of effector and regulatory T cell lineage commitment [211]. T cells with impaired mTOR activity could not differentiate into Th1, Th2, Th17 cells and exhibited high activity of Smad3, leading to the differentiation into Treg cells [211]. Furthermore, the low flux of glycolysis also impairs immunosurveillance by inhibiting the production of metabolic intermediates. A low concentration of glucose (0.1 mM) causes less production of phosphoenolpyruvate/PEP, a glycolytic metabolite that plays an important role in activating TCR-dependent Ca²⁺-NFAT signaling in anti-tumor CD4⁺ cells [212]. An absence of aerobic glycolysis promotes PD-1 expression and GAPDH binding to the 3’UTR of IFN-γ mRNA to negatively regulate the IFN-γ protein level in CD4 T Cells [191]. By contrast, Treg cells do not rely on a high dose of glucose and can be induced by AMPK signaling to exert their immune-suppressive functions [213]. These findings suggest that immune cell exhaustion can be overcome by reversing glucose restriction and creating a nutritionally adequate environment. However, a recent study by W Kimryn Rathmell et al. used PET tracers and measured the access glucose and glutamine uptake of specific cell subsets cultured in a metabolites-available environment [214]. The study found that TAMs and M-MDSCs exhibited significantly higher glucose uptake ability than cancer cells and T cells in the TME and maintained a robust glucose metabolism [214]. Inhibition of glutamine metabolism could enhance the glucose uptake of immune cells and cancer cells, suggesting that tumor-intrinsic physiological mechanism could regulate the glucose uptake of immune cells [214]. While glucose restriction may exist, a high concentration of glucose can be found in some kinds of tumors where exists tumor-infiltrating lymphocytes with impaired anti-tumor functions [215]. Therefore, immune exhaustion is not the only restriction of glucose but glucose as an energy resource to immune cells which is necessary for the antitumor immunity. However, other nutrients, metabolites, and changes in cancer cell physiology are also indispensable factors to construct the heterogeneous tumor microenvironment.

Cancer cells and carcinoma-associated fibroblasts (CAFs) have reciprocal metabolic interactions [216]. CAFs could prompt CRC cells to change their metabolism toward a more glycolytic phenotype [217, 218]. It has been observed that while co-culturing CRC cell line DLD1 with CAFs, the glucose uptake and lactate secretion rates were significantly strengthened while the glutamine anabolism and TCA cycle were restrained [217]. Similarly, another study verified that CAFs can enhance the 18F-FDG uptake and expression of GLUT1 and HK2 in CRC cells [218]. In addition, clinical CRC samples with a higher density of CAF tend to exhibit higher 18F-FDG uptake and are relevant to poor prognosis [218]. In turn, tumor cells also promote glycolysis in CAFs [218]. Both platelet-derived growth factor/PDGF and culture media of human CRC cell line HCT116 could induce the formation of CAFs. In addition, culture media-induced CAFs show a higher expression of HK2 and GLUT1 at both protein and transcription levels than PDGF-induced CAFs. Furthermore, the mRNA level of HIF1α under normoxia was also significantly enhanced in culture media-induced CAFs, indicating that tumor cells could significantly enhance the glycolytic rate of CAFs which might be achieved through a complex regulatory network [218].

It has been demonstrated that metabolic heterogeneity exists in tumors with different metabolic phenotypes such as aerobic glycolysis and OXPHOS in a certain type of cell [219, 220]. A portion of cancer cells and CAFs show a glycolytic phenotype, while other cancer cells utilize lactate as an energy source [219]. In detail, glycolytic CAFs could upregulate MCT4 which is a lactate efflux transporter and promote the secretion of lactate from CAFs to extracellular fluid while a part of tumor cells can uptake lactate and display oxidative metabolism [221]. This suggests that an intercellular metabolic circuitry, termed “the reverse Warburg effect,” exists in the tumor microenvironment which further enhances the metabolic plasticity of tumor cells. Such a cellular metabolic interaction allows tumors to react to changes in nutrient availability, thus maximizing cellular proliferation and growth.

Exosomes are nano-sized extracellular vesicles that contain a variety of bioactive molecules, such as nucleic acids, proteins, lipids, and metabolites and show striking potential to aid tumor diagnosis and treatment [222, 223]. Cells in the TME can secrete exosomes to interact with and dynamically remodel the metabolism of other cells [15]. For example, tumor-derived exosomes can increase glucose uptake and inhibits mitochondrial oxidative phosphorylation in macrophages which reprogram macrophages into glycolytic-dominant metabolism and immunosuppressive phenotype [224]. Further, the non-tumor cells can regulate the glycolytic metabolism of cancer cells and a growing body of evidence indicates that a complex exosome-mediated cell-to-cell communication exists in the tumor microenvironment of CRC [15, 225]. The exosome-associated proteomics of CRC cell lines identified that cancer-derived EVs are specifically enriched in glycolytic pathways and lead to a metabolic switch in CAFs that undergo aerobic glycolysis compared to normal fibroblasts [226]. Reciprocally, exosomes secreted by CAFs can inhibit mitochondrial oxidative phosphorylation and increase glycolysis in cancer cells [227]. It has also been demonstrated that non-coding RNA is an important messenger to regulate the glycolytic activity of distant cells. For instance, circular RNA hsa_circ_0005963 (ciRS-122) in oxaliplatin-resistant CRC cells can be packaged into exosomes and transported into oxaliplatin-sensitive CRC cells to promote glycolysis and drug resistance through miR-122 sponging and PKM2 upregulation in vitro and in vivo [228]. Furthermore, the exosome is an important mediator between primary tumor and premetastatic niche which promotes metastasis and is one of the mechanisms to reprogram recipient cells' metabolism. However, exosomal miR-122 was able to increase nutrient availability in the premetastatic niche by inhibiting the glucose utilization of distant organs, thereby meeting the high energy demand and low ATP-generating efficiency of cancer cells [229].

Although therapeutics based on 5-fluorouracil with oxaliplatin or irinotecan markedly kill CRC cells, a proportion of tumor cells still survive, leading to drug resistance and disease progression [7]. Glycolytic metabolism contributes to the resistance to chemotherapy. Research indicated that a higher expression of glucose transporters such as GLUT1, GLUT4, and SGLT1 was found in colorectal cancer from non-responsive patients than those responsive to 5-FU. Upregulation of glucose transporters can enhance aerobic glycolysis to produce more pyruvate to confront ROS-induced necroptosis and to regulate cell cycle into a quiescent state [234, 235]. Another study also found that high glucose attenuates antiproliferative effect and cytotoxicity of 5-Fluorouracil in human colon cancer cells [236]. As a hallmark of the Warburg effect, the acidic tumor microenvironment has strong adverse effects on cancer treatment. A study indicated that the inhibition of the lactate efflux sensitizes tumors cells to cisplatin treatment [237]. As suggested, the Warburg effect also supplies a considerable amount of energy to tumor cells. Tumor cells with highly glycolytic metabolism could generate sufficient ATP to support the efflux function of the ATP-binding cassette (ABC) family and create an acidic microenvironment, promoting drug efflux pumps activity [139, 238]. In 5-FU resistant CRC cell lines, the expression of drug efflux transporters, especially the ABC transporter family was observed to be upregulated which reduced the intracellular availability of anti-tumor drugs [239]. It has also been revealed that the Warburg effect can also resist the induction of cell death by inhibiting apoptosis [240]. Enhanced glycolytic metabolism could decrease the intracellular level of ROS which is regarded as an important factor inducing apoptosis [241, 242]. Glycolytic enzymes could also rescue ROS-induced apoptosis by directly regulating the process of apoptosis [243]. For example, PKM2 has been reported to translocate to mitochondria and phosphorylate Bcl2, which increases the Bcl2 expression to inhibit apoptosis [243]. Similarly, HK2 could also translocate to mitochondria and protect tumor cells from apoptosis [244]. Conversely, inhibition of glycolysis in the human CRC cell line leads to increased apoptosis rates and decreased resistance to 5-Fu [245]. The decreased aerobic glycolysis always accompanies relevant upregulation of apoptosis. Therefore, while measuring cancer cell ability to resist drugs, the detection of degree of apoptosis are widely considered [246]. Multiple findings have demonstrated the existence of glycolysis-induced drug resistance in CRC. However, the role of other mechanisms such as DNA damage repair, EMT and stemness to induce resistance are largely unknown in CRC. Hence, more in-depth studies are needed to unravel the detailed mechanisms and hub genes in drug resistance, which would help us find potential therapeutic targets to overcome drug resistance in CRC.

Anti-angiogenic treatment has become an attractive therapeutic avenue in colorectal cancer [247]. Bevacizumab, an anti-VEGF monoclonal antibody, was the first biologic agent approved for metastatic colorectal cancer and the addition of bevacizumab to other chemotherapy backbones has been shown to better progression-free survival [7]. However, resistance to anti-VEGF therapy has been observed and metabolic adaptation plays a vital role in this progression [248, 249]. Bevacizumab-resistant cells exert higher glucose uptake ability and glycolytic activity than bevacizumab-responsive tumor cells [250]. In addition, Bevacizumab treatment leads to a more hypoxic environment, where tumor cells change their metabolic phenotype toward a more glycolytic one [251]. Upregulation of HIF-1a can promote neovascularization by promoting tumor cells to produce pro-angiogenic factors such as PDGF-B, FGF-2, VEGF-A, VEGFR1 and angiopoietins [252]. Hence, highly glycolytic metabolism can be considered as a metabolic feature regarding the resistance to Bevacizumab [249, 253]. Furthermore, Xu et al. found that CRC cells refractory to bevacizumab treatment not only undergo hyperactive glycolytic phenotype, but also occur persistent impairment of mitochondria [254]. Therefore, targeting glycolysis has a promising therapeutic potential in bevacizumab-resistant patients because the impaired mitochondria in bevacizumab-resistant cells makes them more dependent on glycolytic metabolism. In CRC, treatment of bevacizumab-resistant cells with the HK2 inhibitor 3-BrPA caused cell senescence in vitro and smaller tumor volume and longer survival in vivo [254].

Immune checkpoint inhibitors (ICIs) such as nivolumab or pembrolizumab has been approved to treat colorectal cancer and provided durable responses and disease control in patients with MSI-H tumors [7]. However, the majority of patients do not benefit from immunotherapy, which might be correlated with the immunosuppressive TME. The activation and anti-tumor function of effector T cells rely on sufficient nutrients. However, glucose deprivation limits T cell persistence and function [255]. In addition, immunosuppressive metabolic byproducts such as lactic acid can significantly abolish the T cell mediated lysis of cancer cells [255]. A recent study found that acidic TME can inhibit the expression of PD-1 in effector T cells while induce the expression of PD-1 in Treg cells because Treg cells can utilize lactate to promote the nuclear translocation of NFAT1 and further activate the transcription of PD-1 [16]. Therefore, PD-1 blockade could increase the activity of PD-1 + Treg cells, resulting in the resistance to immunotherapy [16]. Transcription analysis also revealed a negative correlation between tumor glycolysis activity and tumor-infiltrated CD8 + T cells [198]. Additionally, elevated level of LDH in patient serum samples predicts poor therapeutic responses to the anti-PD-1 antibody pembrolizumab [256, 257]. Hence, targeting glycolytic metabolism is a potential therapeutic method to attenuate the immunosuppressive microenvironment and overcome the resistance to ICIs.

In the past decade, the role of diet has caught the attention of scientists because metabolites and biomolecules in the blood can significantly influence tumor growth, especially in CRC [314]. Dietary intervention is cheap, available, and could synergize with standard therapy to enhance the therapeutic effect [315]. Understanding the importance and practicability of dietary intervention and its rational application will help us understand and gain better insights into cancer treatment. Some reviews have discussed the dietary approaches that influence cancer therapy and in this review, we mainly focus on the role of dietary intervention in inhibiting CRC glycolytic metabolism [316, 317].

Fasting or fasting-mimicking diet (FMD) can lead to chronic calorie restriction and reduced levels of blood glucose, insulin, and insulin-like growth factors -1 (IGF-1), which further inhibits insulin-mediated PI3K/AKT/mTOR pathway and aerobic glycolysis [318]. A recent report has found that fasting could induce the upregulation of FDFT1, a tumor suppressor gene to attenuate the AKT/mTOR/HIF pathway, thus inhibiting tumor glycolysis and proliferation in a CRC mouse model [319]. A single-arm, phase I/II clinical study (NCT03595540) evaluated the safety and feasibility of 5-day FMD in malignant tumors, including CRC, indicating the combination of FMD and chemotherapy is safer and feasible with a reduction of serum IGF1 and IGFBP3 [320]. In another clinical trial (NCT03340935), fasting-mimicking diet reduced blood glucose and growth factor concentration with reliable safety. [321]. Recently, a pilot clinical study (NCT04247464) is undergoing in Madrid, Spain to test whether short-term fasting (24 h before and 24 h after chemotherapy) can improve anti-tumor effect and reduce chemotherapy toxicity. The ketogenic diet is another method to limit glycolytic metabolism by reducing glucose levels in cancers [322]. Animal studies have proved that a ketogenic diet can inhibit tumor growth, laying a foundation for clinical trials [323, 324]. The KETOCOMP study (NCT02516501) intended to explore the impact of the ketogenic diet on CRC patients undergoing radiotherapy [325]. Compared to patients with a standard diet, patients with a ketogenic diet showed a better mental state, better physiological indicators, and a trend contributing synergistically to pathological tumor response, thus demonstrating the therapeutic potential of ketogenic diet in CRC treatment and the necessity of future confirmation in larger studies [326, 327].

Few epidemiological studies have indicated that a high-fructose diet is associated with tumorigenesis in CRC [317, 328]. APC mutant mice were raised with high-fructose corn syrup to investigate the mechanism of fructose-induced tumorigenesis in CRC [329]. The results indicated the presence of more tumor numbers and higher tumor grade in HFCS-treated mice compared to control group [329]. Furthermore, researchers have found that fructose can be transformed into fructose-1-phosphate and activate the glycolytic metabolism and fatty acid metabolism that promote tumor development [329]. Considering the role of fructose in fueling cancer glycolytic metabolism, diets that limit fructose uptake can be a potential intervention to inhibit tumor development.

Immune checkpoint inhibitors (ICIs) dramatically changes the therapeutic outcome of metastatic tumors [330, 331]. In 2017, pembrolizumab (PD-1 inhibitor) was first approved by FDA to treat MSI-H CRC [332]. An important clinical trial in CRC immunotherapy is KEYNOTE-177 (NCT02563002) which is a randomized, phase III trial that uses pembrolizumab as first-line treatment to compare the efficiency with 5-FU-based traditional chemotherapy in MSI-H/dMMR CRC patients [333]. Pembrolizumab treatment leads to longer progression-free survival (PFS) than 5-FU-based chemotherapy, thus leading to the recommendation of the pembrolizumab as a first-line treatment option in MSI-H/dMMR mCRC in NCCN guidelines [333, 334]. Despite the immense progress of ICIs in MSI-H CRC, their application in microsatellite-stable (MSS) CRC is still limited. Therefore, the focus of clinical trials treating MSS CRC is a combination therapy using both ICIs and other drugs to transform “cold” tumors into “hot” tumors [335].

Warburg metabolism can significantly influence the TIME and have close interactions with the expression of immune checkpoints, suggesting glycolysis-based therapy with ICI could reverse the immunosuppressive environment, thus overcoming the resistance in single-agent ICI [76, 168]. Although the application of classical glycolysis inhibitors in conjunction with ICIs in clinical trials is largely unknown, the preclinical trials have demonstrated their therapeutic potential. For instance, treatment of mice bearing CT26 colorectal tumors with aspirin and anti-PD-1 reduced tumor growth and was followed by rapid and complete shrinkage in 30% of mice, whereas monotherapy showed little efficacy [336]. Another study demonstrated that the LDHA inhibition could inhibit tumor glycolysis and improve the efficacy of PD-1 blockade in a murine pMMR CRC model, which was consistent with the clinical findings that highlighted the negative correlation of anti-PD-1 therapeutic efficiency with the serum LDH levels in patients [256, 257, 337]. In addition, the PFKFB3 inhibitor PFK-158 has been reported to improve the therapeutic responses to antibodies against CTLA-4 in a mouse B16 melanoma model [338]. Moreover, some chemotherapy drugs such as apatinib (VEGF inhibitor), trametinib (MEK inhibitor) can inhibit glycolysis and provide new insight into the anti-tumor mechanisms [339, 340]. Their application in conjunction with ICIs may gain better therapeutic benefits in CRC (Table 3). Some studies have already reported the combination benefit of glycolysis-related drugs and ICIs in cancer treatment. According to the results from CAP 01 trial using combined therapy (anti-PD-1 plus apatinib) in 20 patients with chemorefractory gestational trophoblastic neoplasia, the objective response rate (ORR) was 55% with acceptable toxicity and 10 patients had a complete response. Similar trials are also undergoing in CRC, which may lead to breakthroughs in CRC immunotherapy [341]. Another phase I/II clinical trial using combined PD-1, BRAF and MEK inhibition indicated that spartalizumab (anti-PD-L1) plus dabrafenib and trametinib led to an ORR of 78%, including 44% complete responses (CRs), highlighting glycolysis-inhibiting trametinib is an important reagent for cancer therapy and an adjuvant candidate to immunotherapy [342].

Is Warburg metabolism the Achilles Heel of TIME? It is important to notice the overlap in metabolic patterns between tumor and anti-tumor immune cells as the inhibitors targeting glycolysis may impair both tumor and antitumor immune cells [76]. For instance, 2-DG could significantly inhibit the proliferation, glucose consumption, and lactate production of both CD4 and CD8 T cells [343]. In addition, the IFN-γ, TNF,

IL-10, and IL-4 production were also inhibited upon 2-DG treatment in effector CD4 T cells [343]. In another study, significantly reduced ability to expand and induce inflammatory reaction can be observed in Teff with Glut1 deficiency [353]. However, Treg cells can proliferate and exert immunosuppressive function in a Glut1-independent manner [353]. Therefore, the subtle differences in metabolic inhibition of individual cells need to be identified and verified for metabolic vulnerability in cancer and immune cells.

Despite the fact that tumor cells are generally glycolytic, heterogeneous metabolic patterns or preferences still exist among different individuals with the same types of cancer and even in the same sample. Metabolic heterogeneity is important because it influences therapeutic vulnerabilities and may predict clinical outcomes. The application of advanced technologies such as single-cell sequencing and multi-omics analysis can help identify the metabolic similarities and differences among individuals in different tumors. Single-cell sequencing can help us find the metabolic vulnerabilities in certain cell type. For instance, single-cell RNA sequencing data of CRC patients discovered low activity of the MondoA–thioredoxin-interacting protein (TXNIP) axis in Tregs, which can upregulate their glycolytic level [354]. Depletion of the MondoA-TXNIP axis induced hyper-glycolytic Th17-like Tregs, which facilitated Th17 inflammation, promoted CD8 + T cell exhaustion, and drove colorectal carcinogenesis [354]. Using transcriptomic data and bioinformation analysis, a molecular subtype with high level of glycolytic activity can be identified in triple-negative breast cancer, which was characterized by worse prognosis and increased production of glycolytic metabolites [355]. Cancer cells which are clustered into glycolytic subtype showed higher sensitivity to glycolysis inhibitors [355]. Glycolysis has close connections with malignant phenotypes, suggesting that tumors with high glycolytic activity might be associated with worse prognosis and targeting glycolytic metabolism in these patients could receive more therapeutic beneficials. From the perspective of this review, further studies should be focused on exploring the metabolic heterogeneity in CRC. Also, the metabolic subtype of CRC is also worthy to explore because this helps to find a group of patients that are sensitive to metabolic inhibitors. In addition, the majority of the CRC are not applicable to immune therapy. Single-cell sequencing can help us learn about the metabolic feature of immune cells and find metabolic targets to heat the “cold” tumor microenvironment.