Metabolic reprogramming is considered as a hallmark of cancer (6). Tumor cells undergo metabolic adaptions to fuel their infinite division and exhibit altered metabolic pattern compared to normal cells. The initial finding is that tumor cells preferentially use glycolysis to generate ATP along with increased lactate secretion even with abundant oxygen supply, known as ‘Warburg effect’ or ‘aerobic glycolysis’ (26, 27). Meanwhile, tumor cells also uptake a large amount of nutrients, such as glutamine, arginine, tryptophan, and fatty acids to generate ATP and biosynthetic precursors for anabolism (28). Metabolism reprogramming renders tumor cells to better counteract external stresses, most notably the hypoxia and nutrient starvation. Accordingly, these altered metabolic activities and bioenergetic mechanisms are critical to the tumor survival (29).

Tumor cells are in tight interaction with the TME and metabolism reprogramming is affected by both external and internal stimuli (30). Specifically, metabolism provides survival advantages for tumor cells according to the following aspects. First, enhanced glycolysis in tumor cells contributes to the generation of glycolytic intermediates, which in turn favor the pentose phosphate pathway (PPP) pathway, amino acid biosynthesis, and nucleotide biosynthesis (29). Second, tumor cells generate acetyl-CoA by digesting small-chain fatty acids. Acetyl-CoA is conducive to the production of tricarboxylic acid (TCA) cycle intermediates, which are usually used as precursors for macromolecule synthesis, such as fatty acid and cholesterol (31). Third, under the limited nutrient and oxygen supply, tumor cells evolved several mechanisms to sustain the ATP/ADP ratio. Additionally, despite the heterogeneous nature of solid tumor, malignancy seems to activate a series of common oncogenic pathways to support anabolism and redox balance, such as PI3K pathway (30). Taken together, tumor cells always choose the optimal metabolic pattern to satisfy their energic and anabolic demands of enhanced cell proliferation, which partially determine the nutrient and oxygen availability in the TME.

T cells are at rest when the body is in immune equilibrium, while they could be quickly activated and respond to detrimental infection, inflammation, and tumor-associated antigens (32). There are two main categories of metabolic patterns of T cells: 1) the activated T cells prefer to perform aerobic glycolysis that similarly occurs in cancer cells; 2) the resting T cells mainly depends on TCA and OXPHOS to obtain energy.

T cells are the most important immune cells, which could attack tumor cells or indirectly secrete biological molecules. A recent research indicates that T cell metabolism is closely associated with T cell differentiation, activation, function, memory, and exhaustion (33, 34). Theoretically, resting T cells mainly depend on OXPHOS, while activated T cells display enhanced aerobic glycolysis, glutamine deposition, and reduced fatty acid oxidation (FAO) to meet the requirement of energy. Figure 1 depicts metabolic transitions during the differentiation from naïve T cell to effector T cell, memory T cell, and regulatory T cell.

It has been known that T cells undergo metabolic reprogramming during activation, which is associated with distinct T cell subsets and functional fates (35). Naïve T cells intake basic nutrient to preserve minimum cellular ATP levels via OXPHOS (36). These quiescent T cells also rely on FAO and glutamine metabolism to maintain basic metabolic needs (37, 38). Unlike tumor cells, T cell metabolic reprogramming is initiated by T cell receptor (TCR) activation with co-stimulation (39). Upon encounter with antigen, naïve T cell rapidly shifts to effector T cell that appears as a metabolic activation state with increased nutrient absorption and glycolysis rate (36). It has been shown that effector T cells mainly depend on aerobic glycolysis and OXPHOS to maintain T cell fitness and function (40, 41). Such a metabolic transition involves alterations in cellular signaling and metabolic pathways for optimization of immune functionality and proliferation (42). First, such a metabolic transition significantly upregulates cellular signaling pathways, such as PI3K/AKT/mTOR and Ras/MEK/ERK pathways that lead to the expression of transcription factors like HIF-1α and C-MYC, regulating T cell metabolic programs and functional fates (43). Second, these activated T cells experience a similar shift from catabolism toward anabolism, with increased generation of biosynthetic intermediates, including protein, lipid, and nucleotide (44). Eventually, T cells gain the ability to proliferate and generate progeny cells to exert immune functions, secreting inflammatory cytokines and lytic molecules (42). Metabolically, glycolysis produces ATP much faster and offers more nutrients for T cell activation. Therefore, in the differentiation of T cells towards terminal effector status, they become more dependent on glycolysis (45), of which process is similar to the aerobic glycolysis (Warburg effect) long observed in cancer (46).

In parallel to upregulating glycolysis, T cells also engage in rapid uptake and metabolism of glutamine, which enters the tricarboxylic acid (TCA) cycle as α-ketoglutarate (47). Glutamine can support OXPHOS via the TCA cycle (48). Moreover, both resting and activated T cells consume exogenous lipids and catalyze the oxidation of lipids (44, 49). β-oxidation is a highly efficient source to generate ATP. Therefore, the activated T cells can fuel their proliferation by consuming multiple nutrients, forming a superior clonal population with effector functions. Notably, many alterations occurred in rapidly proliferating T cells resemble the process of cancer metabolic reprogramming.

It is well known that the activation of T cells is induced by inflammation and cytokines to acquire different effector functions, thereby metabolic program differs in each subset (50). If these metabolic transitions fail to establish, T cells cannot exert their particular functions and instead differentiate into regulatory T cells (Treg). Differently, Treg cells mainly rely on OXPHOS and FAO to supply ATP. Such a metabolic profile endows the TME with immunosuppressive property (51). After antigen clearance, memory T cells establish immunological memory to allow for rapid activation when encountered with the same antigen (52). The metabolic pattern of memory T cells is similar to that of naïve T cells, with reliance on OXPHOS and FAO for basic energy (53, 54). If antigens are not quickly cleared, persistent antigen exposure would result in T cell dysfunction (55). Unlike memory T cell, exhausted T cells show dampened proliferative capacity and marked metabolic dysfunction. Specifically, glucose uptake and OXPHOS are decreased, whereas mitochondrial mass is increased (56).

In summary, different metabolic patterns influence the differentiation of various T cell subtypes. Table 1 summarizes metabolic characteristics of distinct T cell populations. Exploring the metabolic reprogramming of T cells could help us to better understand the essence of immune responses.

Cellular metabolic disturbances lead to a highly acidic, hypoxic, immunosuppressive TME, which is associated with cancer progression and immune escape (96). The metabolic communication and competition between tumor and T cells have been depicted in Figure 2 .

T cells play pivotal roles in eliminating tumor cells and serve as major mediators in immunotherapy (106–109). Aberrant cellular metabolism is a major cause of immune cell dysfunction (110–112). As metabolic pathways utilized in each subset are distinct, the differential access to nutrients can affect the path of T cell activation, antigen recognition, and cytotoxicity. Recent studies found that disrupted T cell metabolism is frequently along with depleted anti-tumor immunity and impaired homeostasis (113). Therefore, we mainly depict the metabolic pattern of T cells during anti-tumor responses. Metabolism pathways of T cells have been presented in Figure 2 .

Indeed, T cells can produce intrinsic metabolic barriers to block their activation, such as inhibitory molecules of PD-1 and CTLA-4. Theoretically, ICI aims to defeat inhibitory signals and further activate endogenous effector T cells, representing an outstanding advance in the treatment of solid tumors. Several studies indicated that ICI may impact the communication and competition between cancer and T cells (139). One study manifested that PD-1 signaling can inhibit glucose uptake and glycolysis by downregulating the PI3K/AKT/mTOR pathway, thereby impairing T cell activation (139). Conversely, PD-1 signaling favors FAO by upregulating AMPK activity and stimulating CPTA1 expression (35). It is notable CTLA-4 signaling also inhibits glycolysis but has no effect on the rate of FAO (35). Apart from metabolic modulation of TILs, immune checkpoints also directly affect cancer metabolic programming. For instance, PD-L1 and B7-H3 could enhance aerobic glycolysis in tumor cells by activating PI3K-AKT-mTOR pathway and HIF-1α (146). Therefore, inhibition of PD1-PD-L1 axis may generate a synergistic effect to boost anti-tumor immunity, simultaneously reinvigorating metabolic fitness of TILs and inhibiting aberrant metabolic profiles of tumor cells. Co-stimulatory receptors, such as CD28, 4-1BB, OX-40, ICOS, and GITR also participate in metabolic reprogramming to support T cell activation. For instance, CD28 co-stimulation mainly prompts aerobic glycolysis (147) and facilitates mitochondrial morphology (148–150); 4-1BB co-stimulation mainly activates FAO and mitochondrial biogenesis (60). Metabolic modulation of immune checkpoints and their corresponding ligands are presented in Figure 3 .

Overall, as both PD-1 and CTLA-4 suppress T cell glycolysis that is necessary for T cell activation, both PD1-targeted ICI and CTLA4-targted ICI could prompt glucose influx and glycolysis to enhance effector T cell function. Ongoing clinical trials combining metabolic interventions and ICI have been summarized in Table 4 .

As previously discussed, TME imposes a variety of extrinsic metabolic barriers on TILs, such as mitochondrial dysfunction, hypoxia, and acidification (148). These metabolic barriers may be overcome by ICI.

Mitochondria profoundly affect T cell survival and function. It has been found the activation of transcriptional activator PGC1α is essential to mitochondrial biogenesis (151). Specifically, the loss of mitochondrial mass might be the consequence of durable AKT activation followed by downregulation of PGC1α, which co-activates transcriptional factors, such as PPARγ, NRF1, NRF2, and ERRα, and thereby motivates mitochondrial biogenesis and FAO (151). One can assume that overexpression PGC1α via 4-1BB stimulation may revert mitochondrial dysfunction and prevent T cell exhaustion (152). Importantly, the combination of anti-PD1 blockade and PPAR- PGC1α agonist (bezafibrate) can increase mitochondrial biogenesis, FAO, and OXPHOS in CD8⁺ TILs (130). The hyperpolarized mitochondria also contribute to the generation of ROS, which can promote the exhausted T cell phenotype through epigenetic remodeling (153–155). More recently it has been shown ROS scavengers can synergize with ICI and improve anti-tumor functionality (153–155).

The hypoxic TME limits the oxidative capacity of TILs. There is a high probability that tumor with enhanced oxidative metabolism would fail to respond to PD-1 blockade (140). It is anticipated that supplemental oxygen might boost ICI responses. For instance, the supply of oxygen could reduce the amount of immunosuppressive adenosine in mice model transplanted with lung cancer, along with increased CD8⁺ TILs and decreased Treg cells (69). Additionally, metformin, a drug already used for diabetic patients, is able to inhibit oxygen consumption by tumor cells, alleviating the TME hypoxia. A strong evidence manifested that metformin and ICI have a synergistic effect to improve T-cell function and enhance tumor clearance (70). In a mouse model treated with metformin, tumor hypoxia is alleviated and the response to anti-PD1 blockade is greatly improved (153). Altogether, oxygen deficiency either disturbs normal T cell activation or supports T cell exhaustion, which may contribute to the resistance to ICI. Therefore, identifying the tumor metabolic pattern of each patient may help to inform combinational treatment.

Accumulation of lactate in the TME contributes to a higher tumor burden among patients (115). A prior research found that the TME at neutral PH can potentiate the efficacy of ICI (156). Diclofenac has been identified as an MCT inhibitor (157). Combinational therapy of diclofenac and ICI in mice model showed a promising efficacy profile with prolonged survival (97).

Tumor cells outcompete CD8⁺ TILs for glucose in the TME. Low glucose levels enforce TILs to rely on alternative energy sources. From a therapeutic perspective, PD-1 blockade targeting tumor cells could inhibit mTORC1 pathway and further limit glycolysis. Consequently, effector T cells with increased glucose uptake are able to expand their proliferation (3).

A lack of glutamine in the TME could weaken effector T cell activity (47, 158). A prodrug of the glutamine antagonist (6-dizao-5-oxo-L-norleucine) is administrated to a mouse model with colon tumor, leading to decreased glutamine metabolism within the tumor and increased glutamine availability in the TME. Importantly, co-administration of PD1-targeted ICI and glutamine antagonist could induce complete responses and a durable memory-like T cell phenotype (63). Nevertheless, tumors that consume less exogenous glutamine may fail to respond to this combination treatment. Likewise, a low level of extracellular arginine restricts T cell proliferation (159). Compared to ARG1 inhibitor monotherapy, more successes are observed when ARG1 inhibitor is in combination with ICI. Furthermore, IDO1 and catabolite kynurenine are highly expressed in tumor cells to impair T cell fitness by depleting tryptophan and accumulating immunosuppressive metabolites. IDO inhibition, like 1-methyltrytophan, renders the TME less immunosuppressive and facilitates CD8⁺ TIL activity. To increase tryptophan availability, IDO inhibitors combined with ICI are currently being assessed in clinical trials for the treatment of a variety of solid tumors. Besides, kynureninase that can directly catabolize kynurenine, is evidenced to synergize with ICI to extend survival in the preclinical models (160).

Theoretically, memory T cells obtain energy by depending on fatty acids (149). But tumor cells outcompete with memory T cells in fatty acid uptake, thereby inducing apoptosis of memory T cells. This phenomenon could be rescued by PD-1 blockade inhibitor by upregulating fatty acid-binding protein (Fabp) 4/5 in memory T cells and downregulating its expression in tumor cells (161).

In ACT, autologous CD8⁺ TILs targeting tumor-specific antigen or peripheral blood T cells transduced with chimeric antigen receptor T (CAR-T) cells targeting tumor-associated antigen, are expanded in vitro and readministered to the patient. ACT is superior to counteract metabolic stress and nutrient deficiency caused by cancer metabolic reprogramming. Therapeutic T cells are activated and expanded in vitro under culturing conditions with ample nutrients, overcoming the metabolic challenges that may exist during T cell priming, proliferation, and differentiation, contributing to an augment immunity. ACT provides a unique chance to manipulate T cells before transfer to improve their metabolic fitness. Once transplanted to the recipient, these T cells must infiltrate into the solid tumor and perform effector functions. Therefore, modulating T cell metabolism during ex vivo TIL expansion phase or modifying chimeric antigen receptor design to overcome metabolic barriers would benefit patients with solid tumors (162).

CAR-T cell persistence is positively correlated with better tumor clearance and potent anti-tumor response (163), which could be achieved by harvesting longer-lived memory T cells. It has been demonstrated that adoptive T cells derived from central memory T cells can yield a more persistent response in vivo compared with those derived from effector cells (164, 165). However, the usual medium for ACT may not support the proliferation of long-lived memory T cells that mainly utilize OXPHOS (166). Explicitly, memory T cells undergo a metabolic shift from glycolysis towards oxidative metabolism motivated by carnitine palmitoyltransferase (CPTA1) (167). Therefore, altering T cell metabolism, by manipulating CAR-T designs or through modifying culture condition, may enforce the memory phenotype to support T-cell persistence and efficacy.

One early study pointed out that mTORC1 inhibition with rapamycin resulted in increased number of memory T cells after pathogen clearance (168). Besides, mTORC2-AKT inhibition during ex vivo expansion can also endow TILs with memory phenotype and enhance anti-tumor response (169, 170). These findings support innovations that integrate metabolic interventions into current ACT immunotherapy.

The culture condition is usually hyper-glycaemia compared with normal human plasma (141). Such a high glucose concentration may induce metabolic stress after infusion. One strategy is to use glycolysis inhibitor 2-DG during the vitro expansion phase, stabilizing a memory-like phenotype after transfer to achieve stronger tumor clearance (45). Modifying ex vivo expansion condition also includes the supply of IL-15 or L-arginine in media to enhance OXPHOS in order to increase the generation of memory T cells (121, 171). Importantly, CAR-T cells cultured with a shorter period may be affected less in hyper-glycaemia medium, giving rise to a subset of less differentiated T cells that have a good proliferative capacity (172).

Genetic manipulation of CAR-T cells with enhanced mitochondrial metabolism can enforce memory phenotype as well. First, 4-1BB-containing CARs support ACT to gain increased mitochondrial biogenesis and FAO, accompanied with better persistence than CD28-containing CARs (173). Second, it is feasible to directly promote metabolism pattern of memory T cells. Upregulating the transcriptional co-activator PGC1α or employing the PPAR agonists both mediate enhanced FAO, enriching early memory T cells and improving T cell functionality against tumors (77, 130, 151).

T cells may be manipulated to acclimatize to low oxygen level in the TME. Currently, CAR-T cell therapy for the treatment of solid tumors is still unsuccessful, partly due to their poor performance under hypoxia. Culturing CAR-T cells in media with a low level of oxygen has been shown to acclimate them to a hypoxic environment and increase their cytotoxic effect on tumor cells (174). It is true T cells would primarily upregulate HIF-1α expression in response to the hypoxic signal. In the mice model transplanted with solid tumors, HIF-1α deletion resulted in a decreased number of TIL and a heavy tumor burden, verifying the importance of the adaption to hypoxia (175).

Modulation of T cell metabolism can overcome metabolic barriers imposed by TME-specific metabolites as well. Overwhelming evidence suggested metabolite adenosine inhibits the effector function of immune cells (91, 176, 177). Targeting the adenosine A2AR receptor has been confirmed to improve the activity of CD8⁺ T cells (178). Further, the combination of adenosine A2AR receptor antagonist and CAR-T cell therapy is able to increase the efficacy in breast cancer models (179). The release of ions by necrotic tumor cells can disrupt TCR signaling and impair T cell function. For example, high-level of intracellular potassium in T cells can suppress TCR-driven AKT-mTOR signaling (180). Overexpression of the K+ efflux channel in TILs during ex vivo expansion can optimize the treatment efficacy of ACT in melanoma-bearing mice (180).

T cells can be engineered to overcome nutrient deficiency in the TME to boost anti-tumor responses. Highly metabolic tumors are more likely to develop resistance to ACT therapy, mainly because nutrient insufficiency can impair T cell trafficking and cytotoxicity (142). Glucose restriction can induce insufficient phosphoenolpyruvate (PEP) and decreased T cell receptor (TCR) signaling in TILs (181). Correspondingly, increasing PEP levels via overexpression of phosphoenolpyruvate carboxykinase (PCK1) could improve ACT, supporting intra-tumoral TCR-mediated Ca2⁺-NAFT signaling and T cell effector function (181). As previously introduced, arginine depletion also blocks T cell proliferation. Therefore, engineering CAR-T cells to express arginine synthetase (OTC and ASS) could prompt T cell proliferation and maintenance, along with better efficacy compared with CAR-T cells without these enzymes (182). Additionally, if the culturing condition mimics the TME, these T cells might adapt to low nutrient availability after infusion back to patients. It has been shown that T cells cultured in low-nutrient medium tend to differentiate into a memory phenotype and express less inhibitory receptors (62). For example, T cells in low-glutamine media can produce more potent anti-tumor responses than T cells in traditional media after transfer (62).

Skewing tumor metabolism towards OXPHOS may enhance OV replication and lytic capacity. Dicholoroacetate (DCA) that inhibits pyruvate dehydrogenase kinase, can shift tumor metabolism to OXPHOS. When tumor-bearing mice were treated with Newcastle disease virus (NDV) in combination with DCA, it is surprising to observe increased NDV replication, along with reduced glucose uptake and lactate production by tumor cells and increased survival in mice compared with NDV therapy alone (187). Oncolytic adenoviruses in combination with the glycolysis inhibitor 2-DG, can get the similar results in preclinical models as well (144). It can be inferred that metabolic modulation may improve cytotoxic CD8⁺ T cell response induced by OV therapy. However, tumors with enhanced oxidative metabolism tend to induce hypoxia in the TME, which can impair CD8⁺ T cell response as priorly discussed. A recent study put forward that, selecting tumors with oxidative metabolism to receive OV therapy, may be superior to metabolic modulation that shift tumor metabolism towards OXPHOS (188). Other metabolic pathways and metabolites are also involved in viral replication and tumor lysis. For example, vaccinia virus needs glutamine and TCA cycle to maintain the viral replication and induce tumor death (145). Moreover, Mevalonate inhibition can render tumor cell more sensitive to oncolytic M1 virus therapy (189). Therefore, metabolic modulators have the potential to increase OV replication and lytic ability, though the mechanisms of action are poorly understood.

Oncolytic viruses can be engineered to carry on genes encoding metabolic modulators. For example, engineering vaccinia viruses is able to increase the generation of adipokine leptin by tumor cells after viral infection, leading to the TME remodeling and enhanced anti-tumor response (190). Besides, leptin in the TME can prompt memory precursor T cell differentiation that confers a quick response upon tumor rechallenge (190).

Taken together, the crosstalk between metabolism and OV therapy is a new field that deserves further exploration. Existing findings highlight the importance of identifying the potential OV-responsive patients. And multiple combination immunotherapies are being tested in preclinical models or in clinical trials. The T-VEC is in combination with PD-1 ICI is being explored (191, 192). In solid tumor-bearing preclinical models, OVs are being investigated in combination with ACT therapy (193).