In tumor formation and metastasis, Tregs are the main source of IL-10, a kind of cytokine with durable and versatile anti-inflammatory effects (13). In the TME, Tregs significantly secrete more IL-10, which immune checkpoint molecule TIM-3 plays a non-negligible role. Banerjee H et al. found that Tregs upregulated the secretion of IL-10 promoting CD8⁺ Tex through T cell immunoglobulin and mucin domain-containing protein 3 (TIM-3) (14). Interestingly, IL-10 also reverses the process of inducing Treg differentiation in the TME, so that the interaction between IL-10 and Treg further acts on T cell exhaustion (15). Furthermore, it was discovered that Salkeni et al. thought IL-10 induced T cell stimulation in the TME. Thus, more investigation is required to learn how IL-10 influences T-cell exhaustion (16). Different Treg subsets produce different cytokines. In the study by Ma et al. of esophageal squamous cell carcinoma, PD-1⁺ Tregs were the primary producers of IL-10, whereas CD177⁺ Tregs were the primary producers of IL-35 in the TME (17). As a heterodimer of IL-12, IL-35 derived by Tregs regulates the generation of immune checkpoint molecules: programmed cell death protein 1 (PD-1), TIM-3, and lymphocyte-activation gene 3 (LAG-3), which promote T cell exhaustion by cooperating with IL-10 (12).

An important modulator of both innate and adaptive immunity is TGF-β (18), which is released by tumor cells and immune cells (19, 20). TGF-β prevents immature T cells from differentiating into Th1 cells, promoting Foxp3 to encourage the differentiation of naïve CD4⁺ T cells into Treg cells and regulating TGF-β signaling to mediate the anti-tumor immune function of Treg inhibiting CD8⁺ T cells in the TME (21, 22). Knocking down endogenous TGF-β receptor type-2 (TGFBR2), blocking TGF-β signaling, or inhibiting the deubiquitination of Foxp3 from upregulating in the TME can inhibit the differentiation of Tregs in the TME and prevent Tex (23–26). In addition, colonic carcinoma metastasis signal mediated by TGF-β can stimulate the upregulation of Foxp3, CD25, and PD-1, which are markers of CD8⁺ Tex (27).

IL-6, which can be produced by a variety of cells, is a classical cytokine regulating various inflammations. In the inflammatory response, it possesses both opposed- and anti-inflammatory characteristics (30). The aberrant production of IL-6 is the result of TME chronic inflammation (31, 32), so tumor cells will produce and secrete more endogenous IL-6 when proliferating, infiltrating, or metastasizing (33, 34). Tumor patients who showed higher concentrations of IL-6 in their blood in clinical had suppression of CD8⁺ T cells proliferation and cytokine generation, and they had higher adverse clinical outcomes (35). In the TME, IL-6 can bind its receptor, transmit signals through the downstream JAK/STAT3 pathway, and activate the transcription of tumor-related genes. In addition, IL-6/JAK/STAT3 signal transduction upregulates immune checkpoint ligand PD-L1 expressed and regulates MDSCs to suppress tumor immunity (36). MAPK and IL-6-mediated signaling pathways impact immature MDSCs (i-MDSCs) in the TME, which are activated by IL-6 and attracted to tumor locations where they develop into mature MDSCs and decrease tumor immunity (37). IL-6 recruits MDSCs in the TME, which can inhibit T cell activity and promote partial CD8⁺ Tex through IFN-γ (38). One important regulator in preserving MDSC differentiation, proliferation, and function is IL-6. When MDSC inhibitors are applied to inhibit the number and function of MDSC, the expression of IL-6 in the TME is also inhibited, and the active immune response is re-established (39). It’s interesting to note that IL-6 released by MDSCs was linked to multidrug resistance via the IL-6/STAT1 axis in the work by Dhar S et al. (40). Within the congenital drug resistance model, combining treatment of Cytotoxic T-Lymphocyte Associated Antigen 4 (CTLA-4) inhibitor and BET inhibitor induced the decrease of MDSCs infiltration and downregulated TIM-3 and LAG-3 in the TME (41). IL-6 in the TME recruits MDSCs to participate in the drug resistance of tumor therapy and may be related to CD8⁺ Tex, but the clear mechanism needs further exploration.

The reason why acquired resistance happens to ICB therapy is that MDSCs induce CD8⁺ Tex. According to Koh J et al., the greater the infiltration of MDSCs with the progression of the patients’ non-small cell lung cancer, MDSCs inhibited T cell activity and induced T cell exhaustion in the TME, which contributed to resistance in anti-PD-1 therapy (42). Additionally, Galectin 9 (Gal-9) acts as an immune checkpoint ligand for TIM-3. MDSCs upregulate the expression of Gal-9 in the TME. A rise in Gal-9 causes CD8⁺ T cells to overexpress TIM-3, which prevents T cells from secreting effector cytokines. Thus, the TIM-3/Gal-9 route may be used by MDSCs to promote CD8⁺ T cell exhaustion (43). In addition, clinical research on the treatment of breast cancer showed that CDK4/6 inhibitors avoided the T cell exhaustion phenotype by reducing the expression levels of multiple immune checkpoints and the quantity of MDSCs (44).

As mentioned above, the primary source of IL-10 in the TME is Tregs. It is undeniable that cytokines including IL-10 released by MDSCs can activate Tregs and TAMs in the TME (45). MDSCs and Tregs in the TME share common microRNAs related to immune regulatory pathways, including TNF, TGF-β, FOXO, and Hippo (46, 47). Therefore, MDSCs and Tregs not only have the effect of inhibiting tumor immunity, but also interact and cooperate with each other in TME.

Following the injection of carboplatin, MDSCs stimulated IL-13/IL-33 via the VCAM/RANTES pathway, facilitating CD4⁺ T cells to convert into Tregs and encouraging Tregs aggregation at tumor locations. The inhibitory receptor TIGIT was upregulated by CD155. The expression of TIGIT was favorably linked with the ratio of T cells to MDSCs in TME. CD8⁺ T cells upregulated the expression of CTLA-4, TIM-3, and CD160 (48). Blocking TIGIT/CD155 signal transduction alleviates Tex and delays tumor growth (49). MDSCs recruited in the TME, mediate the growth and metastasis of tumors through phosphatidylinositol-3-hydroxylase (PI3Kγ), inhibiting the transcription of down-regulated T cell exhaustion genes in mouse tumors with PI3Kγ, and reducing infiltrated Foxp3⁺ Tregs in the TME (50).

The triggering receptor expressed on myeloid cell 1 (TREM1) is a key mediator of innate immunity (51). The high expression of TREM1 in polymeric myeloid-derived suppressor cells (PMN-MDSCs) and M2-like macrophages is associated with the inferior prognosis of lung cancer (52), liver cancer (53), renal cell carcinoma (54), breast cancer (51)and other tumors. Knockdown of TREM1 can enhance the therapeutic effect of anti-PD-1, probably by inhibiting the function of MDSCs and T cell exhaustion in the TME (55). In the DLL3-TREM1/DAP12 CAR-T (DLL3-DT CAR-T) therapy developed by Nie F et al., DLL3-DT CAR-T cells have durable antitumor responses and increased production of memory T cells in the TME (56).

Increasing evidence suggests that exosomes can promote chemoresistance within the TME (57). Tumors regulate the TME by secreting microRNAs from exosomes. Pancreatic cancer-derived exosomes have a macrophage migration inhibitor factor (MIF), and MIF tautomerase regulates the expression of genes required for the differentiation, recruitment, and activation of MDSCs (58). In addition, exosome mir-1298-5p promotes the immunosuppressive effect of MDSCs, promoting the growth of neuroglioma (59). Interestingly, MDSC−derived exosomes (MDSC exo) are hyperactivated inducing CD8⁺ T cell exhaustion and promoting the production of reactive oxygen species to meet the oxygen supply of tumor cells (60).

TAMs are abundant in TME, tumor blood vessels, and stromal regions, which have immunosuppressive effects and promote Tex (61). Hypoxia is frequently seen in TME, and there is a high expression of hypoxia-inducible factor-1α (HIF-1α). HIF-1α can induce monocyte myeloid-derived suppressor cells (m-MDSCs) to differentiate into TAMs through IL-6/STAT3/p-STAT3 and inhibit T cell activation and effector function (65, 66). Targeted STAT3 tumor immunotherapy reverses the polarization and immunosuppression of TAMs in the TME (67, 68). Through the regulation of XOR-IDH3α, TAMs may impact the function of TAMs and CD8⁺ T cells. According to Lu Y et al. of hepatocellular carcinoma and xanthine oxidoreductase (XOR), within hypoxia TME, enzymes involved in macrophage polarization interact with IDH3α to mediate TAMs polarization and inhibit anti-tumor immunity of CD8⁺ T cells (69). The upregulation of CD8⁺ T cell immune checkpoints (PD-1, CD38, TOX) was positively correlated with the abundance of TAMs in the TME. In addition, Kersten K’s team also found that exhausted T cells would recruit monocytes to the TME and form TAMs to induce CD8⁺ T cell exhaustion, further inhibiting anti-tumor immunity (70).

HIF-1α can not only induce m-MDSCs to differentiate into TAMs but also upregulate the expression of PD-L1 on the surface of MDSCs by binding to PD-L1 receptor on the surface of T cells, which induces Tex (71, 72). In a therapeutic study of nanomaterials transporting catalase into the TME, ROS were increased by degrading H₂O₂, significantly reducing MDSCs and Tregs (73).

Chemokines which are a kind of guiding cell migration cytokines expressed in the TME will recruit immune cells into the TME (74–77). There are many different chemokines produced by different cells in the complex TME. In the study of Kamat K et al., CCL23 secreted by TAMs induced T cell exhaustion by mediating GSK3β to upregulate T cell exhaustion markers in ovarian cancer (78). In this part, we focus on the chemokines related to T cells exhaustion and explore their potential research directions in ICB and ACT therapy.

As mentioned above, hypoxia in the TME induces the polarization of TAMs and the exhaustion of T cells (65). Several studies have shown that TAMs play an immunosuppressive role in the TME through the CCR5-dependent signaling axis (79–81), which can block the CCR5-related pathway to inhibit tumor invasion and metastasis (82). In advanced cancer patients, the proportion of exhausted T cells in a hypoxic environment is significantly increased. Analysis of exhausted T cells ligand-receptor interactions revealed that tumor cells would recruit T cells into hypoxic TME through the chemokine CCL3/CCL4/CCL5-CCR5 axis (83). Hernández-Verdin I et al. found that the increased expression of activation-induced cytidine deaminase (AID) caused by mutation induced T cells exhaustion by upregulating the expression of CXCL13/CCR5. The application of ICB therapy can block it, thereby reactivating exhausted T cells (84). Coincidentally, the study of Liu Z et al. found that RUNX3 was the key mediator of decitabine improving anti-PD-1 immunotherapy, and knockdown of RUNX3 significantly reduced the levels of CCR3 and CCR5 (85). In addition, TAMs in the TME will conduct cellular and molecular communication with immature DCs induced by tumor cells through CCR5/CCL5 molecules, which will destroy the antigen-presenting function of DCs (86). Similarly, the study of Horie M et al. also showed that HIF-1α would promote the recruitment of TAMs and DCs in the TME under hypoxia, and create an immunosuppressive microenvironment by upregulating the expression of CCR5/CCL5, inducing T cells exhaustion and peritoneal dissemination of cancer cells (87). TAMs in TME not only inhibit the anti-tumor effect of T cells, but also destroy the auxiliary function of DCs in this process. How DCs become a good helper of CD8⁺ T cells will be described in the next section.

Hedgehog (Hh) signaling in myeloid cells is essential for M2 polarization of TAMs and tumor growth. Tumor cells secrete an Hh ligand (SHH) to drive TAMs to polarize towards M2 (88). The production of CXCL9 and CXCL10 by polarized TAMs is inhibited, blocking the CXCL9/10/CXCR3 axis and inhibiting the recruitment of CD8⁺ T cells to the TME (88). Compared with normal renal tissue, the mRNA expression of CXCR3/CXCL9/10/11 in renal cell carcinoma was significantly increased (89), and the upregulation of CXCR3 expression in renal cell carcinoma was related to the hypoxic state in TME (90). In addition, the study of Azuma M et al. demonstrated that Poly (I: C) interacted with DCs, cross-upregulated the expression of CXCR3, and promoted the recruitment of CD8⁺ T cells to the TME (91). However, CD8⁺ T cells recruited to the TME will gradually downregulate the expression of CXCR3 under chronic antigen stimulation, while upregulating the exhaustion markers PD-1 and LAG-3. In addition, the knockdown of CXCR3 can enhance the production of IFN-γ, an effector molecule of CD8⁺ T cells, indicating that CXCR3 promotes the loss of effector function and the process of T cell exhaustion (92). Interestingly, the expression of CXCR3 in tumor mice that did not respond to ICB therapy was significantly reduced (93). Enhancing the CXCR3/CXCL9 axis can promote the interaction between DCs and CD8⁺ T cells in the TME and improve the sensitivity to PD-1 blockade and clinical therapeutic efficacy (94). The enhancing effect of CXCR3 in PD-1/PD-L1 therapy may be related to its ability to recruit T cells into the TME in an antigen-independent manner and to activate the bystander memory T cells (95). At the same time, previous studies have shown that the CXCR3 ligand helps T cells and B cells recruited to the TME form a cell network directly contacting each other. This network contains TCF1⁺ PD-1⁺ CD8⁺ T cell progenitors, which can be transformed into cytotoxic CD8⁺ T cells during ICB treatment (96, 97). Therefore, Sullivan PM et al. demonstrated that overexpressed CXCR3 can be applied to ACT therapy to drive the trafficking of CD8⁺ T cells into the TME and synergize with PD-1 checkpoint blockade immunotherapy (98). In addition, although it played an important role in the entry of CD8⁺ T cells into the TME, they gave an explanation for the incomprehensible result that CXCR3 expression was downregulated in the TME: it may be attributed to cell-extrinsic variables in the TME, such as inhibitory receptor signal transduction and TGF-β secreted by tumor cells (99). Similarly, Li A et al. demonstrated that a unique anti-human GARP antibody (named PIIO-1) treatment could block CD8⁺ T cell exhaustion by reducing TGF-β signaling in the TME while enhancing CD8⁺ T migration into the TME in a CXCR3-dependent manner (25).

As mentioned above, tumor cells secrete Shh to drive TAMs to polarize towards M2, and the production of CXCL9 and CXCL10 by polarized TAMs is inhibited (88). In the study of He N et al., it was found that the polarization of TAMs in microwave ablation (MWA) therapy was mediated through the nuclear factor-κB/JAK-STAT1 signaling pathway. MWA significantly upregulated the expression of IFN-γ stimulated transcription factors (especially IRF8) (100). Among the nine IRF members regulating interferon signaling, the expression of IRF8 in HCC was associated with inferior prognosis of HCC patients (101). In addition, MWA combined with αPD-L1 treatment promoted the production of CXCL9 and blocked IFN-γ/CXCL9/CD8⁺ T axis which could promote tumor progression (100). Tumor cells will induce the expansion of TAMs, which requires TAMs to express IRF8. Specific deletion of IRF8 in TAMs blocks T-cell exhaustion and inhibits tumor growth (102). Therefore, TAMs participate in the process of T cell exhaustion by upregulating IRF8. The study of Wu H et al. demonstrated that IRF8 downregulated the expression of CCL20 by inhibiting c-fos transcription and mediated the immunosuppressive effect of TAMs recruited in the TME (101).

IL-2 is a significant cytokine that regulates Tex. IL-2 in the TME is mainly secreted by CD8⁺ T cells and activated DCs. IL-2 exerts its immunomodulatory effect mainly by binding to IL-2R. Various conformations of IL-2R mediate the downstream effects of IL-2. Due to the variable conformation of IL-2R, IL-2 shows a dual role of both immunostimulatory and immunosuppressive, which is related to interleukin-18 (IL-18) and Tregs.

CTLA-4 is mainly expressed in activated T cells, but also in Tregs (161). CTLA-4 binds to CD80/CD86 on the tumor surface to prevent T cell activation in the TME (121). As mentioned earlier, the exhaustion-like phenotype and the impairment of effectors of CD8⁺ T cells are positively correlated with the accumulation of Tregs. Tregs have a high level of CTLA-4 expression, which is essential for developing and maintaining the exhaustive phenotype. In addition, inhibiting CTLA-4 signaling can inhibit the recruitment of Tregs and relieve the immunosuppressive effect of Tregs. Ipilimumab and tremelimumab are FDA-approved CTLA-4 inhibitors for clinical use (162, 163). CTLA-4 inhibitors have the ability to inhibit CTLA-4’s binding to CD80/CD86, promote T-cell proliferation, and have anticancer effects (164).

The surface of activated T cells, NK cells, DCs, and other cells expresses PD-1. PD-L1 is highly expressed in tumor cells, and the combination of the two mediates tumor immune escape. PD-1 in TME can not only mediate TCR to decide the activation and differentiation of T cells but also activate Tregs. Therefore, PD-1 plays a non-negligible role in the anti-tumor immunity of remodeling the TME (165, 166). PD-1 is upregulated on T cells of renal angiomyolipoma and pulmonary lymphangioleiomyomatosis. Blocking both PD-1 and CTLA-4 together is more efficient than targeting just one PD-1 inhibitor (167). After the combined blockade of PD-1 and CTLA-4 in the orthotopic hepatocellular carcinoma model, PD-1 was expressed in moderate amounts on Tex^prog, and Tex^term almost disappeared in the TME. Hence, the combined treatment downregulated PD-1 expression and reversed the anti-tumor immunity of Tex (168). These studies confirm that combination therapy is better at reducing CD8⁺ Tex. Although the efficacy of combination therapy is better, the immune-related adverse reactions (irAEs) related to ICB drugs are more serious (169). Therefore, a significant issue with ICB therapy is how to balance its effects with the emergence of irAEs, and more research is needed. In addition, whether PD-1 inhibitors combined with other immunotherapies can reverse T-cell exhaustion is a hot topic of future research (170–172).

TIM3 is expressed in Tregs, TAMs, DCS, natural killer cells (NKs), and mast cells. TIM3 is a key cytokine for Tex^term and is at the junction of T-cell exhaustion and rejuvenation (124, 173). The interaction between TIM3 and galectin 9 (Gal9) in CD8⁺ T cells leads to the reduction of the production of cytokines (IFNγ, IL-2, and TNFα), the inhibition of T cell proliferation, and the inhibition of anti-tumor immunity (174). The progression of tumors can be effectively controlled by simultaneously inhibiting TIM3 and PD1. Datar I et al. indicated that lung cancer patients with TIM3⁺ CD68⁺ TAM had shorter survival and worse prognosis than those with TIM3^- CD68⁺ TAM (175, 176). TIM3⁺ Treg is also more immunosuppressive than TIM3^-Treg (177). In addition, the lack of TIM3 on DCs promotes the proliferation of CD8⁺ T cells in the TME (178).

LAG-3 is an inhibitory receptor protein on activated T cells, NK cells, B cells, plasma cells, and DCs. In the study of Kano Y et al., the soluble recombinant protein LAG-3-Ig, which inhibits the LAG-3 signal, was combined with a tumor vaccine, and it was found that they could down-regulate the expression of LAG-3 and other co-inhibitory receptors on CD8⁺ T cells, prevent Tex, and boost the tumor vaccine’s therapeutic impact in concert (179). The expression level of LAG-3 increases with the stimulation of tumor antigens. Under long-term stimulation, LAG-3 and other co-inhibitory receptors are continuously expressed on T cells and mediate the exhaustion of T cells (180). Research has demonstrated that LAG3 inhibition can change Tex’s activity. In comparison to single inhibition, dual inhibition of PD-1 and LAG3 produces higher T-cell activation and reduces Tex (181, 182).

Recent years have seen a remarkable development in CAR-T cell therapy. By collecting isolated T cells, and integrating antibodies targeting cells, CAR-T cells are generated and reinfused into patients to kill tumor cells (184). CAR-T can induce the remission of hematological system tumors effectively, but there is still a problem of recurrence in some patients (185). When it comes to treating solid tumors, the therapeutic effect of CAR-T cells is limited, which is characterized by poor penetration, persistence, and low proliferation ability. Therefore, choosing a target has been a hot subject in CAR-T cell therapy research (186–188). CAR-T cells and endogenous CD8⁺ T cells have similar gene expression and chromatin accessibility, so CAR-T cells will also be stimulated by sustained antigens during anti-tumor immunity, driving CAR-T cell exhaustion (189). Furthermore, the immunosuppressive cells also take part in the CAR-T cells’ exhaustion (23, 190).

TCR-T is to transfer the TCR gene sequence into T cells through genetic engineering technology, specifically recognizes tumor antigens, and inhibits tumors (184). Similar to the problems faced by CAR-T cells in the TME, TCR-T cells lack long-term persistence and are limited by immunosuppressive cells in the TME, resulting in limited therapeutic effect and TCR-T cell exhaustion (209). According to Cianciotti BC et al., knocking down the co-inhibitory receptors TIM-3 and LAG-3 connected to T cell exhaustion can resist the depletion of TCR-T cells and restore its antitumor effect (210).