The classical pathway of T cell exhaustion was driven by chronic TCR signaling, accompanied by a complex transcriptional network consisting of thymocyte selection-associated high mobility group box protein (TOX), nuclear factor of activated T cells 1 (NFATc1), nuclear receptor 4A (NR4A), T cell factor-1 (TCF-1) and activator protein-1 (AP-1) family transcription factors (TFs) ( Figure 2 ) (27–30).

There were two main phases in T cell exhaustion: Tpex development phase and terminal exhausted differentiation phase (6, 15). During Tpex development phase, TCF-1, encoded by tcf7, mediated the transition from T-box expressed in T cells (T-bet) to eomesodermin (Eomes) and drove differentiation toward Tpex (15, 31). More importantly, TOX, the decisive TF of exhaustion, maintained Tpex through increasing the expression of Tcf7, Bcl6, Bach2, Foxo1 and Id3 and inhibiting the expression of Il2ra and Irf4 (32). However, continuous inhibitory signals induced terminal exhausted differentiation. There were some significant changes of TFs in terminal exhausted differentiation phase, especially in the change of AP-1. AP-1 was a generic name, consisted of members of the Fos, Jun, Maf and ATF multigene families and was regulated by co-stimulatory signals (33). Usually, AP-1 (Jun-Fos dimer) interacted with NFATc1 and promoted T cell activation (29). However, most members of AP-1 family were dramatically downregulated in terminally exhausted CD8⁺ T cells in TME, excluding the basic leucine zipper activating transcription factor-like transcription factor (BATF). TFs including NFATc1, BATF, TOX, NR4A and interferon regulatory factor 4 (IRF4), greatly promote the development of terminal exhaustion. They were significantly increased in terminally exhausted subsets, resulting in increased expression of genes involved in exhaustion progression (Havcr2, Tbx21, Ctla4, Pdcd1, Tigit) and the decreased expression of genes of inflammatory pathways (Ifng, etc.) (22, 27, 29, 32, 34). Besides, in this phase, TOX inhibited the effects of TCF-1, promoting Tpex to differentiate into terminal exhausted subsets. TOX strongly drove terminal exhaustion differentiation (7, 30). Therefore, TEXs had high expression of multiple inhibitory receptors and limited cytotoxicity ( Figure 2 ).

Metabolic defeats in CD8⁺ TEXs play a significant role to promote exhaustion differentiation. TME was an acidic environment with hypoxia, nutrient deficiency and high concentrations of metabolic waste, where TEXs underwent extensive metabolic reprogramming that promoted exhaustion-related gene expression (3, 35, 36).

In terms of nutrient metabolism, there was a decreased uptake of glucose but excessive accumulation of lipid and cholesterol in TEXs. CD8⁺ TEXs suffered from severe suppression of glycolysis and oxidative phosphorylation (OXPHOS) and increased lipid peroxidation and endoplasmic reticulum (ER) stress, which damaged the effector function of tumor infiltrating CD8⁺ T cells (CD8⁺ TILs) and promoted terminal exhaustion differentiation ( Figure 2 ) (3, 36, 37).

Mitochondrial dysfunction and oxidative stress were of great importance to terminal exhaustion differentiation (20, 38, 39). CD8⁺ TEXs accumulated depolarized mitochondria with disrupted membrane, cristae structure and abnormal membrane potential (39). The impaired mitochondrial structures and functions significantly enhanced the level of mitochondrial reactive oxygen species (mROS) and oxidative stress (20, 38, 39). mROS was a strong inducer of terminal exhaustion differentiation. ROS activated the nuclear translocation of NFAT, boosting the transcription of TOX and exhaustion-related genes (20, 38). Furthermore, ROS activated phosphatase of activated cells 1 (PAC1). PAC1 downregulated the expression of genes including Klrg1, Gzmb, Gzmm by recruiting the nucleosome remodeling and deacetylase (NuRD) complex, which damaged effector function of CD8⁺ TEXs ( Figure 2 ) (40).

A broad remodeling of chromatin landscape was gradually established during the differentiation of CD8⁺ TEXs, including DNA methylation, histone modifications and chromatin remodeling in CD8⁺ TEXs. The alteration of chromatin accessibility regulated gene expression (41).

De novo DNA methylation, especially DNA methyltransferases 3A (DNMT3A)-dependent DNA methylation events, happened in some specific genes during the differentiation of TEXs. The Ccr7 and Tcf7 loci remained demethylated in Tpex (19, 42). However, in terminally exhausted T cells, genes involved in proliferation, tissue homing, metabolic activity and effector function (Tcf7, Ccr7, Myc and ifng) were acquired severe de novo DNA methylation, contributing to their low expression and terminal exhaustion differentiation (19, 42).

Aberrant histone modifications, particularly methylation modification, have been found in TEXs. The histone H3 lysine 27 (H3K27) demethylation of terminal-specific genes (Tox, etc.) and increased H3K27 methylation in progenitor-specific genes (Tcf7, etc.) contributed to the maintenance of terminally exhausted states (32). Besides, high levels of bivalent chromatin were found in terminally exhausted T cells, which was a phenomenon specific to TEXs in cancers. Bivalent chromatins were regulated by active H3K4me3 and repressive H3K27me3, poising genes for rapid expression or repression and responding to environmental stimulation (32). In hypoxia, bivalent genes were enriched in terminally exhausted subsets and suppressed the expression of genes of inflammatory response and leukocyte differentiation (32).

Genome-wide clustered regularly interspaced short palindromic repeats screens in mouse and human tumors indicated that chromatin remodeling complex, especially canonical BRG1 or BRM-associated factor (cBAF) was enriched in TEXs and downregulated effector function and the maintenance of exhaustion (43). Furthermore, ROS activated PAC1 and recruited chromatin remodeling complex NuRD complex, which suppressed the expression of genes involved in effector function (40).

In conclusion, massive inhibitory signaling in TME drove the progression of CD8⁺ T cell exhaustion. There are many strategies targeting on the inhibitory signals from TME or the key programs of exhaustion, showing the powerful anti-tumor effects in preclinical studies and some clinical trials (5, 18, 39, 42, 44). The reinvigoration of CD8⁺ TEXs in TME is a promising treatment to enhance the outcome of cancer patients.

PD1/PDL1 blockade is the important component of tumor immunotherapy. Various studies have suggested that Tpex are the main responders to PD1/PDL1 blockade (5, 15, 23). The understanding of the effects of PD1/PDL1 blockade on CD8⁺ TEXs is constantly being updated. Initially, it was believed that PD1/PDL1 blockade could reverse CD8⁺ TEXs (2). However, further research revealed that the development of TEXs was accompanied by complex transcriptional and epigenetic programs and it was difficult to reverse the TEXs (2, 5). Current research suggested that the main effects of PD1/PDL1 blockade were to expand Tpex in TME. PD1/PDL1 blockade induced the replenishment of Tpex from outside the tumor, especially from lymph nodes, bridging systemic and anti-tumor responses (23–26, 48, 49). However, the expansion of Tpex was partly based on local expansion of pre-existing Tpex (25, 26, 49). Besides, PD1/PDL1 blockade promoted Tpex to differentiate into terminally exhausted subsets with better effector function ( Figure 4A ) (23–26, 48, 49). Anti-PD1 therapy improved the gene expression of inflammatory pathways, resulting in better effector function (5, 14, 15). However, there were minimal changes in epigenetic profiles were observed in CD8⁺ TEXs, suggesting that PD1/PDL1 blockade failed to change their epigenetic status (5, 32). CD8⁺ T cells inevitably became dysfunctional after treatment because the key exhaustion programs did not change. This may explain the poor outcome or failure to set up durable responses in some cancer patients with anti-PD1 therapy.

In preclinical models, anti-PD-1 therapy significantly suppressed the growth of murine melanoma, resulting in the proliferative burst of Tpex and increased the number of terminally exhausted CD8⁺ T cells with superior cytotoxicity ( Table 1 ) (5). Furthermore, the expansion of Tpex after treatment was related to better outcome of cancer patients (5, 23). A clinical study evaluating the efficacy of anti-PD1 therapy (Nivolumab) in patients with advanced melanoma showed that the increased frequency of Tpex was associated with the prolonged progression-free survival and overall survival in responders (5). Another study showed that in patients of non-small-cell lung cancer (NSCLC) treated with anti-PD1 therapy (Pembrolizumab), responsive tumors showed an expansion of Tpex with low expression of co-inhibitory molecules and high expression of GZMK, while non-responsive tumors failed to accumulate Tpex (23). In clinical practice, PD1/PDL1 blockade showed its powerful effects on many cancers including melanoma, liver cancer, etc. But its low response rate and drug resistance limited its application (16, 17). There are a large number of clinical trials evaluating the efficacy of anti-PD1 antibody monotherapy or combination therapy on solid tumors while few of them evaluate the change in TEXs. Improving CD8⁺ TEXs may be a breakthrough point to reverse the current dilemma of cancer immunotherapy. More studies are needed to explore the improvement of TEXs after anti-PD1 therapy in human solid tumors.

CD28 superfamily and tumor necrosis factor receptor (TNFR) superfamily were the major components of co-stimulatory receptors (53). CD8⁺ TEXs showed significantly impaired co-stimulatory signals which inhibited NFAT: AP-1 cooperation and induced metabolic dysfunction, resulting to the progression of terminal exhaustion (32, 46). Activation of CD28 and 4-1BB (a member of TNFR superfamily) has been demonstrated their therapeutic potential in some studies (18, 46, 47, 50, 51).

NFAT: AP-1 (Jun/Fos complex) cooperation was significantly damaged in CD8⁺ TEXs. Upregulation of c-Jun expression by drug or CAR-T cell therapy effectively stabilized NFAT: AP-1 cooperation, which inhibited the NFAT-dependent exhaustion program, expanded Tpex and greatly enhanced anti-tumor immunity (12, 57, 58).

In preclinical models, transfer of c-Jun OE (CAR-T cells overexpressing c-Jun) significantly limited the growth of murine leukemia and osteosarcoma. C-Jun OE showed enhanced expansion potential, cytotoxicity and were resistant to terminal exhaustion differentiation (12). Besides, neuropilin-1 (NRP1) is an inhibitory receptor on CD8⁺ T cells. NRP1 limited the Tpex self-renewal by suppressing c-jun expression. In mouse melanoma models, knockdown Nrp1 in CD8⁺ T cells increased the frequency of Tpex and memory T cells in TME. Nrp1 ^–/– T cells exhibited greater effector function upon re-stimulation, promoting long-term immunity (57). More importantly, it improved the efficacy of ICB, resulting in the enhanced tumor clearance ( Figure 4C ) ( Table 2 ) (57). In short, targeting c-Jun effectively suppressed the development of terminal exhaustion in several cancers.

TOX and NR4A were significantly increased in terminally exhausted subsets and profoundly promote the transcriptional program of T cell exhaustion (30). Targeting TOX is controversial. Seo et al. (2019) described that Tox DKO (TOX1, TOX2) CAR-T cells slowed down the progression of mouse melanoma. Tox DKO (TOX1, TOX2) CAR-T cells in TME showed significantly increased cytokine production (TNF, IFN-γ) and decreased expression of inhibitory receptors ((PD-1, TIM3, and LAG3) (30). However, Scott et al. (2019) provided a contrary conclusion. Although they displayed non-exhausted phenotype (low expression of inhibitory receptors), TOX-knockout T cells remained dysfunctional and showed increased apoptosis in mouse melanoma (22). TOX-knockout T cells failed to persist in TME. TOX induced T cell exhaustion might prevent chronic activation-induced cell death. Further research was needed to investigate the status of TOX-knockout T cells in TME. However, the decreased expression of TOX resisted the development of exhaustion. Anti-vascular endothelial growth factor-A (VEGFA) antibody was reported to greatly limit tumor growth and reduce the levels of TOX and NFAT in mouse colon cancer models. Anti-VEGFA antibody decreased expression of inhibitory receptors (PD1, TIM3, LAG3, TIGIT), and increased IFN-γ and TNF production in CD8⁺ TEXs (59). More importantly, it achieved better anti-tumor effects when combined with PD1/PDL1 blockade. However, NR4A-deficient CAR-T cells showed similar effects of Tox DKO CAR-T cells ( Figure 4C ) ( Table 2 ) (30, 62). NR4A knockout may be a safe choice when TOX has complex effects on T cell fate.

TCF-1 is essential for the differentiation of memory and exhausted T cells (14, 15). Increasing the expression of TCF-1 led to a proliferative burst of Tpex (44, 60, 61). Ectopic TCF-1 expression in preclinical tumor models significantly suppressed tumor growth and expanded Tpex and terminally exhausted subsets. Furthermore, terminally exhausted subsets enhanced cytokine production (IL2, TNF, IFN-γ) and decreased expression of co-inhibitory receptors (PD1, 2B4, LAG3) (44). Besides, Regnase-1 and Negative Elongation Factor (NELF) were reported to regulate tcf7 expression in TEXs (60, 61). The ribonuclease Regnase-1 was the upstream negative regulator of TCF-1. Regnase-1 targeted Tcf7 messenger RNA and suppressed the formation of Tpex. Regnase-1-deficient CAR-T cells enhanced the clearance of murine acute lymphoblastic leukemia (ALL). It showed the features of Tpex and the long-term persistence in TME (60). Besides, NELF, an RNA polymerase II pausing factor, cooperated with TCF-1 to enhance the expression of TCF-1 target genes. Ectopic NELF expression boosted anti-tumor immunity in mouse model of mammary tumor and improved the proliferative and cytotoxic function of CD8⁺ TILs ( Figure 4C ) ( Table 2 ) (61). In short, enhanced expression of TCF-1 resulted in a better anti-tumor response.

BATF and IRF4 were significantly upregulated in TEXs where BATF combined with IRF4 to promote terminal exhausted differentiation (29, 32). BATF regulated both effector programs and exhausted differentiation (63). The way to engineer BATF in CAR-T cells remained controversial. Seo et al. (2021) showed that overexpression of BATF on CAR-T cells reduced exhaustion program and increased anti-tumor immunity in murine melanoma. Depletion of BATF decreased the number of T cells and poor tumor control (64). However, Zhang et al. (2022) presented that the depletion of BATF in CAR-T cells increased the expression of TCF-1 and IL-7 receptor alpha and suppressed expression of genes involved in terminal exhaustion differentiation, leading to the better anti-tumor responses in murine melanoma (65).

The contradictory effects of IRF4 were similar to that of BATF. Both IRF4 knockdown and overexpression in CAR-T cells were reported to increase cytokine production (64, 65). The difference may be associated with differences in the construction of CAR-T cells and the type of tumor models, etc. Therefore, further research was needed to investigate the role of BATF and IRF4 in TME.

In conclusion, transcription factor-based therapy made it possible to exert precise regulation of gene expression, contributing to the Tpex expansion and TEXs with better effector function. However, TFs usually had extensive effects and led to different outcomes. The overall effects of TFs need to be clarified before the treatment. Most of the strategies mentioned above remain in the preclinical stage. Further studies are needed to evaluate their function in clinical practice.

The De novo DNA methylation, especially DNA methyltransferases 3A (DNMT3A)-dependent DNA methylation events in some specific genes promoted the progression of terminal exhaustion (19, 42). Application of DNA demethylating agents Decitabine (DAC) or DNMT3A knockout CAR-T cells effectively inhibited the de novo DNA methylation of exhaustion-related genes (13, 42).

In preclinical studies, pre-treatment of DAC prior to PDL1 blockade significantly inhibited the growth of murine prostate cancer, resulting in CD8⁺ T cells resisting exhaustion and maintaining greater expansion potential (42). More importantly, DAC had a synergistic effect with anti-PD1 antibody, promoting to a further Tpex expansion and cytokine production (IFN-γ, TNF) of CD8⁺ TILs in several mouse tumor models (19). Furthermore, DNMT3A knockout CAR-T cell therapy effectively raised anti-tumor responses. In murine breast cancer and glioma models, DNMT3A knockout CAR-T cells significantly suppressed tumor growth and prolonged survival. They enhanced IL-10 production and differentiated into stem-like CAR-T cells, maintaining long-term anti-tumor responses ( Figure 4D ) ( Table 3 ) (13).

Aberrant histone modification occurred during the progression of terminal exhaustion differentiation, particularly methylation modification, which was closely related to hypoxia in TME. Targeting the histone demethylase lysine-specific demethylase 6B (Kdm6b) and lysine specific demethylase1 (LSD1) showed great therapeutic potential (32, 66–68). Kdm6b, a less oxygen-sensitive member of the KDM6 family of H3K27 demethylases, was downregulated in terminally exhausted subsets. Transfer of T cells overexpressing Kdm6b significantly limited the growth of mouse melanoma. Kdm6b overexpression improved the effector function of T cells but had no effect on the key differentiation program of TEXs ( Figure 4D ) ( Table 3 ) (32).

LSD1 significantly reinvigorated CD8⁺ TEXs. LSD1 specifically removed both mono- and di-methylation of histone H3K4 and H3K9 and mediated immunosuppressive effects independent of histone modifications. LSD1 was significantly upregulated in CD8⁺ T cells in many cancers, which was correlated with prognosis of cancer patients (66–68). In preclinical studies, LSD1 blockade significantly suppressed the progression of colon and breast cancer in mice (66, 67). LSD1 blockade significantly enhanced the amount of intra-tumoral Tpex by promoting the tcf7 transcriptional activity (66). Furthermore, LSD1 and PD1/PDL1 blockade displayed cooperative effects. The combination of LSD1 inhibitor and anti-PD1 therapy achieved a further expansion of Tpex and promoted Tpex differentiate into terminally exhausted CD8⁺ T cells with better cytotoxicity (66, 67). However, the efficacy of LSD1 blockade depended on antigenicity of cancer cells. LSD1 blockade may be useless in low antigenicity tumors, requiring additional antigen stimulation (66). There is a difference in the efficacy of targeting the LSD1 nucleus (nLSD1) or the LSD1. Phosphorylated nLSD1 induced nuclear entry and retention of EOMES, contributing to T cell exhaustion. nLSD1 blockade increased the infiltration of CD8⁺ T cells, particularly the IFN-γ⁺ CD8⁺ T cells in murine breast cancer. Besides, it decreased the expression of TIGIT, LAG3 and TIM3 of CD8⁺ TILs ( Figure 4D ) ( Table 3 ) (68). In conclusion, LSD1 can regulate TCF-1 and EOMES and is a promising therapeutic target to improve TEXs. There are some ongoing clinical trials evaluating the efficacy of LSD1 inhibitor in advanced solid tumors (NCT03895684, NCT05268666, etc.).

Chromatin remodeling complexes (cBAF, NuRD, etc.) promoted the expression of genes involved in exhaustion differentiation (40, 43). Ablating cBAF complex subunit Arid1a in CAR-T cells prevented acquirement of exhaustion-related chromatin accessibility, and drove the differentiation into memory T cells in TME (43). Besides, pre-treatment of naive CD8⁺ T cells with Arid1a inhibitor markedly promoted their effector function and increased anti-tumor efficacy in mouse melanoma and colon cancer models (69). Notably, Arid1a blockade was suitable for pre-treatment of CAR-T cells. The use of Arid1a inhibitor in vivo decreased the effects of immunotherapy, which might be due to the broad regulatory effect of chromatin remodeling complex ( Figure 4D ) ( Table 3 ) (69, 70).

Mitochondrial dysfunction and oxidative stress greatly increased the level of mROS. ROS-PAC1-NuRD complex downregulated the expression of genes including Klrg1, Gzmb, Gzmm, thereby damaging the effector function of TEXs. Knockdown of pac1 effectively suppressed the growth of murine melanoma and colon cancer. Moreover, it increased the number CD8⁺ TILs, significantly improved effector function of TEXs and increased anti-tumor immunity ( Figure 4D ) ( Table 3 ) (40). In conclusion, therapeutic strategies based on epigenetic regulation can effectively inhibit differentiation toward terminal TEXs, but exert a broader range of effects. The safety of treatment must be taken seriously. The same target may have different effects with different interventions.

Severe suppression of glycolysis and OXPHOS was a feature of TEXs, which severely damaged effector function and promoted the terminal exhaustion differentiation. It can be improved by pyruvate supplementation and regulations of pyruvate metabolism (3, 71–73).

Supplementation of pyruvate enhanced both glycolysis and OXPHOS, resulting in improved effector function of CD8⁺ TILs in murine melanoma (71). The improvement of pyruvate transportation has showed excellent therapeutic efficacy. Pyruvate is transported into the mitochondria by the mitochondrial pyruvate carrier (MPC) (72). Guo, et al. (2021) described that the half-life-extended Interleukin-10-Fc fusion protein (IL-10/Fc) can promote MPC to transport pyruvate into mitochondria, increasing OXPHOS and mitochondrial function. Surprisingly, IL-10/Fc treatment effectively controlled the growth of murine melanoma. It promoted self-renewal of terminally exhausted subsets and improved their proliferative capacity and cytotoxicity (72). IL-10/Fc treatment significantly improved the efficacy of PD1/PDL1 blockade, which induced a durable efficacy ( Figure 4E ) ( Table 4 ) (72). IL-10/Fc treatment plays a unique role in the improvement of TEXs. Further research is needed to evaluate its effects in clinical practice.

CD8⁺ TEXs in TME accumulated high levels of oxidized low-density lipoprotein (OxLDL) and cholesterol, leading to the severe lipid peroxidation and ER stress. The reduction of lipid peroxidation and ER stress helped to prevent exhaustion progression (36, 37).

Glutathione peroxidase 4 (GPX4) can decrease the lipid peroxidation induced by OxLDL. Transfer of CAR-T cells overexpressing GPX4 (GPX4 OE) significantly suppressed the growth of murine melanoma and colon cancer. GPX4 OE showed an increase of cytokine production, boosting effector function in TEXs (36). Furthermore, the inhibition of ER stress led to a low expression of inhibitory receptors. Transfer T cells pre-treatment with ER-stress inhibitor STF or cholesterol-lowering drug simvastatin significantly slowed down the growth of mouse melanoma and colon cancer. Furthermore, pre-treatment decreased their apoptosis and expression of PD1 and 2B4, contributing to the enhanced anti-tumor immunity ( Figure 4E ) ( Table 4 ) (37).

Reducing mitochondrial oxidative stress and remodeling mitochondrial metabolism inhibited differentiation into terminally exhausted subsets (20, 38, 39). ROS was a strong inducer of terminally exhausted differentiation. Reducing ROS production or inhibiting its downstream exerted therapeutic effects in several preclinical studies (20, 38, 39).

The supplementation of Nicotinamide riboside improved mitochondrial fitness and decreased ROS production, contributing to the increased cytokine production and a better control of melanoma in mice (39). In addition, N- acetylcysteine (NAC), the antioxidant that neutralizes ROS, can decrease the levels of ROS and TOX expression in TEXs, thereby limiting terminal exhausted differentiation (20, 38). Transfer of T cells pre-treated with NAC effectively suppressed tumor growth in the mouse melanoma model. The pre-treatment increased the effector function of CD8⁺ T cells. More importantly, it increased the effects of PD1/PDL1 blockade and improved the survival of tumor-bearing mice (38). Besides, the knockdown of PAC1, the downstream of ROS, promoted proliferative and cytotoxic functions of TEXs and considerably slowed tumor growth (40). Lastly, CAR-T cells overexpressing PGC1α (PGC1αOE) showed excellent therapeutic potential. PGC1αOE suppressed the growth of murine melanoma. Surprisingly, PGC1αOE did not show any progenitor exhausted signature. Overexpression of PGC1α altered the differentiation of CD8⁺ T cells and prevented the differentiation of TEXs. PGC1α is a promising target to prevent the exhaustion of CAR-T cells ( Figure 4E ) ( Table 4 ) (20).

Hypoxia promoted terminal exhaustion and decreased the sensitivity of immunotherapy (20, 32, 74, 75). Normalizing hypoxia in TME is an important way to improve the response to immunotherapy. Metformin (drug treatment for type 2 diabetes) can effectively reduce hypoxia in TME and inhibit oxygen consumption in tumor cells. It also improved the efficacy of PD1 blockade (74). In addition, low doses of the tyrosine kinase inhibitor Axitinib (targeting vascular endothelial growth factor receptors) improved the tortuous vasculature and decreased hypoxia in TME. Axitinib significantly decreased tumor burden and improved survival. It reduced the expression of PD1 and TIM3 and promoted the production of IFN and TNF, suppressing the terminal exhaustion differentiation. Besides, Axitinib cooperated with PD1 or CTLA4 blockade to reduce tumor burden and improve survival ( Figure 4E ) ( Table 4 ) (20).

Therefore, metabolism is deeply coordinated with immune signaling pathway, influencing differentiation of tumor-infiltrating CD8⁺ T cells. Metabolism-based therapy prevents CD8⁺ T cells from differentiating into terminal exhaustion and improves efficacy of ICB and CAR-T cell therapy.

IL-2, the potent multifunctional cytokine, mainly activated effector and memory T cells and enhanced anti-tumor immunity (78). The production of IL-2 was severely damaged in CD8⁺ TEXs. Although it was reported that continuous high levels of IL-2 in TME led to T cell exhaustion by activating aryl hydrocarbon receptors (79). More studies indicated that the IL-2 therapy significantly activated TEXs and enhanced anti-tumor responses, especially when combined with PD1/PDL1 blockade (21, 80, 81).

IL-2 agonists such as IL-2/NARA1 complex, SIL2-mesenchymal stem cells (MSCs) have been demonstrated to reinvigorate TEXs in murine melanoma and colon cancer models. IL-2 monotherapy significantly suppressed tumor growth and prolonged survival. It significantly expanded tumor-specific CD8⁺ T cells with better effector function and low expression of PD1, TIM3 and LAG3 (80). In addition, IL-2 therapy expanded Tpex and promoted the production of IFN-γ of terminal exhausted T cells (81). More importantly, IL-2 signaling closely correlated with the efficacy of PD1/PDL1 blockade (21, 81). Combined IL-2 therapy with PD1/PDL1 blockade showed more powerful anti-tumor responses (21, 81). This combination therapy greatly inhibited tumor growth and overcame ICB resistance of murine melanoma and colon cancer (81). In addition, PD-1-cis-targeted IL-2Rβγ agonists, can bind to PD-1 and IL-2Rβγ and activate the two molecules. PD-1-cis-targeted IL-2Rβγ agonists led to large expansion of CD8⁺ TILs in murine pancreatic tumor models. Surprisingly, the new effector CD8⁺ T cells showed the low expression of exhaustion-related genes such as Havcr2, Tigit and Tox but the increased expression of genes involved in productive and protective immune memory. In contrast to IL-2 or anti-PD1 monotherapy, the combination of anti-PD1 therapy and IL-2 activation promoted Tpex to differentiate into better effector subsets, rather than terminally exhausted CD8⁺ T cells with better effector function (21). PD-1-cis-targeted IL-2Rβγ agonists showed better anti-tumor efficacy than pembrolizumab in treating the tumor model of human PD-1-transgenic mice ( Figure 4F ) ( Table 5 ) (21). In conclusion, IL-2 therapy is a promising therapy for the improvement of TEXs, especially with the combination of PD1/PDL1 blockade.

IL-10 therapy effectively reinvigorated TEXs by promoting NFAT: AP-1 cooperation and improving pyruvate metabolism, which were mentioned above (58, 72). IL-10 signaling inhibited the activation-induced exhaustion of CD8 T cells in TME. IL-10 signaling blockade significantly impaired the maintenance of Tpex in the mouse model of chronic lymphocytic leukemia (CLL) and accelerated the development of exhaustion (58). More importantly, IL-10 therapy showed the powerful effects in improving terminally exhausted subsets. In murine melanoma models, IL-10/Fc treatment promoted their self-renewal and improved the effector function by upregulating mitochondrial pyruvate carrier-dependent oxidative phosphorylation (72). In addition, the combination of IL-10/Fc and PD1/PDL1 blockade showed stronger anti-tumor efficacy, contributing to a durable efficacy in the mouse CT26 colorectal tumor model ( Figure 4F ) ( Table 5 ) (72).

A phase 1b clinical trial (NCT 02009449) evaluated the efficacy of PEGylated IL-10 with anti-PD-1 monoclonal antibody (Pembrolizumab or Nivolumab) in patients with advanced solid tumors (83–85). PEGylated IL-10 monotherapy induced objective tumor responses in patients with renal cell cancer (RCC). Besides, it significantly increased the number of CD8⁺ TILs, especially the LAG3⁺ PD1⁺ CD8⁺ T cells and enhanced the production of IFN-γ and Granzyme B (83, 84). The combination therapies showed positive outcome in previously treated patients with RCC and NSCLC with objective responses of 43% and 40%, respectively. However, the change in CD8⁺ T cells was not evaluated (85). Therefore, IL-10 therapy has great potential in reinvigorating CD8⁺ TEXs and boosting anti-tumor immunity. And more research is needed to evaluate the efficacy and effects on CD8⁺ TILs of the combination of IL-10 therapy and PD1/PDL1 blockade in patients with solid tumors.

The stimulation of type I interferons (IFN-Is) and IFN-γ played a significant role in the activation of immunity. However, persistent interferon stimulation in TME drove the terminal exhaustion process of CD8⁺ T cells and severely limited anti-tumor responses (76, 77). CD8⁺ TEX enriched for IFN-I-stimulated genes. IFN-Is induced the aberrant lipid accumulation and elevated oxidative stress in CD8⁺ TILs, which triggered mitochondrial defects and induced terminal exhaustion process (82). Besides, persistent IFN-Is and IFN-γ signaling activated the key TF interferon regulatory factor 2 (IRF2). IRF2 regulated genes involved in IFN signaling, TNFα/NF-κB signaling and immune exhaustion (tox), suppressing the effects of CD8⁺ T cells in TME (76).

Inhibiting the persistent IFN signaling from TME significantly prevented the development of T cell exhaustion in preclinical models (76, 82). Knockdown of the expression of ifnar1(encoding the receptor of IFN-Is) in T cells significantly inhibited the growth of murine liver cancer. There was an increased number of CD8⁺ TILs, characterized by the decreased expression of TIM3, CTLA-4, LAG3 and TOX and the enhanced expression of TCF1 (82). Besides, knockdown of irf2 on CD8⁺ T cells showed powerful efficacy. Transfer of IRF2-deficient CD8⁺ T cells significantly limited the growth of murine colon cancer. IRF2-deficient CD8⁺ T cells showed low expression of inhibitory receptors and the enhanced production of Granzyme B, perforin and IFN-γ (76). More importantly, the effects of immunosuppression from persistent IFN-Is and IFN-γ siganling were inhibited in IRF2-deficient CD8⁺ T cells, resulting to long-term tumor control. Surprisingly, there was almost no TOX expression in IRF2-deficient CD8⁺ T cells. IRF2 depletion prevented the acquisition of T cell exhaustion program during persistent IFN signaling in TME (76). IRF2-deficient CD8⁺ T cells are the strong candidates for CAR-T cell therapies ( Figure 4F ) ( Table 5 ) (76).

Besides, blockade of IFN signaling enhanced the anti-tumor effects of PD1/PDL1 blockade (76, 82). The level of IFN-I-stimulated genes was related to the poor efficacy of ICB in patients of melanoma or breast cancer, and anti-IFNAR-1 antibody overcame the drug resistance of anti-PD1 antibody in mouse liver cancer (82). Furthermore, the combination of IRF2-deficient CD8⁺ T cells significantly with anti-PDL1 antibody significantly led to the durable tumor control in the mouse model of breast cancer (76). However, there are no clinical applications to target IFN signaling (IFNI, IRF2) in TME. In short, IFNI and IRF2 are the promising targets for CAR T-cell construction and further investigation is needed.