Tumor escape from the immune system is a central step in tumor development. Loss of antigenicity, loss of immunogenicity and an immunosuppressive microenvironment are the three main mechanisms of tumor immune escape [88]. Thus far, the relevance of TNFR2 for immune escape has mainly been attributed to the beneficial effects of TNFR2 on immunosuppressive tumor-infiltrating Tregs and MDSCs [89]. Furthermore, earlier studies demonstrating a tumor-promoting function of TNF and IL10-producing regulatory B-cells in DMBA/TPA-induced skin carcinogenesis and the recent finding that TNFR2 stimulates IL10 production of regulatory B-cells open the possibility that TNFR2 engaged Bregs also contributes to the anti-inflammatory pro-tumoral effects of TNFR2 [90,91].

Tumor-infiltrating regulatory T-cells are, with high frequency, strongly positive for TNFR2 [92]. For example, in breast cancer, acute myeloid leukemia (AML) and lung cancer, the highest TNFR2 expression levels were found on Foxp3⁺CD25⁺CD4⁺ Tregs [93,94,95,96]. Interestingly, there is evidence that chemotherapy affects the TNFR2⁺ Treg pool in triple-negative breast tumors more severely than the infiltrating CD8⁺ T-cells, so in this special treatment situation, the anti-tumoral TNFR2 activities prevail [97].

As discussed before, TNFR2⁺ Tregs represent the most suppressive fraction of Foxp3⁺ cells and activation of TNFR2 on Tregs by memTNF comes along with increased proliferation and phenotypic stability [59,98]. Soluble TNFR2 is significantly higher in samples of malignant epithelial ovarian cancer (EOC) than in corresponding benign neoplasias and is associated with tumor differentiation [99]. Moreover, TNFR2⁺ Tregs are abundant in the ascites of ovarian tumor patients and have higher suppressive activity than peripheral blood TNFR2⁺ Tregs [100]. Notably, antagonistic TNFR2-specific antibodies trigger cell death more potently in Tregs isolated from the ascites of ovarian cancer patients than in Tregs of healthy donors [101]. Furthermore, a study of tumor ascites from patients with EOC revealed that high levels of the pro-inflammatory cytokine IL6 were present in the analyzed ascites. Culturing T-cells with EOC ascites leads to an increased TNFR2-expression on all analyzed T-cell subsets and an increased ratio of Tregs/Teffs. In addition, those cultured Tregs express higher levels of immunosuppressive molecules such as programmed cell death ligand-1 (PDL1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA4) [102], both molecules with an overwhelming importance for tumors to evade the immune system. Increased expression of CTLA4 and PDL1 was also reported for TNFR2⁺ Tregs derived of lung cancer patients [95,96]. Like IL6, TNF can also upregulate PDL1 expression. Lim and colleagues found out that TNF stabilizes the expression of PDL1 on cancer cells [103]. Mechanistically, TNF activates NFκB signaling, which induces COP9 signalosome 5 (CSN5) expression. CSN5 itself inhibits ubiquitination and subsequent degradation of PDL1. However, whether this mechanism is also of relevance in Tregs is unclear.

The concentrations of soluble TNFR1 and TNFR2 and TNF are significantly increased in sera of AML patients [104,105]. Furthermore, various mouse models of AML with distinct genetic abnormalities showed that autocrine TNF production and TNFR activation by AML cells, especially leukemia-initiating cells, make a crucial contribution to tumor progression through NFκB and JNK-mediated survival signaling [105,106,107], which might antagonize tonic TNF-driven necroptotic signaling [108]. In a FLT3-ITD driven model of AML, however, TNFR or RIPK3 knockout also resulted in enhanced leukemogenesis [109]. These seemingly contrasting results possibly reflect differences in the way the balances between cytotoxic and survival pathways in general has been adjusted in the various models due to genetic factors or differences in the micromilieu. It is, however, also possible that subtle, model-specific differences in the activity of distinct TNF-TNFR1-TNFR2 signaling network axes (sTNF versus memTNF, thus TNFR1 versus TNFR2 activity, induction of endogenous TNF via TNFR1 and/or TNFR2, TNFR2-mediated TRAF2 depletion, etc.) create these strikingly opposing net effects. In this respect, it is worth mentioning that the RNA N6-methyladenosine reader enzyme YTH N6-methyladenosine RNA binding protein 2 (YTHDF2) promotes AML and reduces cytotoxic TNF sensitivity of preleukemic cells by suppressing TNFR2 expression [110,111]. Indeed, in accordance with the idea that this mirrors the ability of TNFR2 to sensitize for TNFR1-induced cell death signaling by restricting the availability of TRAF2-cIAP/1/2 complexes, SMAC mimetics which deplete cIAPs by triggering their proteasomal autodegradation also sensitizes AML cells for autocrine TNF-induced necroptosis [112].

Tregs have been shown to possess additional mechanisms for inhibiting the induction of inflammatory pathways beyond the secretion/expression of anti-inflammatory cytokines, e.g., the release of soluble TNFR2 (sTNFR2). Shedding of the TNFR2 ectodomain, similarly to membrane TNF processing, is mediated by TACE/ADAM17 [113] and is often enhanced under inflammatory circumstances. In particular, sTNFR2 is released from CD4⁺ T-cells after stimulation with TNF or agonistic TNFR2-specific antibodies [114] while stimulation of TNFR1, but not TNFR2, results in enhanced TNFR2 ectodomain shedding in neutrophils [11,115]. Therefore, sTNFR2 can reflect previous or ongoing activation of TNFR2 and/or TNFR1. sTNFR2 released from activated Tregs can scavenge TNF, and thereby prevents the inflammatory effects of TNF and can maintain the immunosuppressive function of Tregs against Teffs [116]. Additionally, high serum-levels of sTNFR2 serve as prognostic marker with poor clinical outcome in various cancer types [117,118,119,120].

Similar to Tregs, MDSCs can promote tumor immune escape. It is therefore not surprising that many tumors also show an increased number of MDSCs [78]. MDSCs suppress anti-tumor immune responses through several mechanisms; for example, by producing NO and ROS, depletion of cysteine or release of IL10 and TGFβ [121]. Tumor cells themselves release many of the stimulating factors of myelopoiesis, such as granulocyte-macrophage colony-stimulating factor (GM-CSF), stem-cell factor and vascular endothelial growth factor (VEGF), all of which promote the development of MDSCs [122,123,124]. In 2012, Hu et al. demonstrated that memTNF promotes tumor growth, tumor progression and angiogenesis [81]. This came along with increasing numbers of MDSCs and Tregs in the TME, with poor lymphocyte infiltration and higher levels of NO, IL10 and TGFβ. Vice versa, a tumor model missing memTNF expression showed higher numbers of tumor-infiltrating lymphocytes and reduced accumulation of MDSCs [81]. Studies with wildtype and TNFR2-deficient MDSCs showed furthermore that memTNF via TNFR2 upregulates CXCR4 expression, and thereby enables chemotactic migration of MDSCs to the tumor microenvironment [125].

Metastasis, and thus the successful colonization of tumor cells at distant sites and their subsequent adaption and growth, is pathophysiological, and typically the most life-threatening step in tumor development [126]. An early event when tumor cells have invaded the liver can be the expression of TNF [127]. In a study investigating the role of TNF in liver metastasis with the help of TNFR1 and TNFR2 knockout mice, it was found that TNFR2 deficiency, but not lack of TNFR1, results in a significant reduction in liver metastasis in colon and lung carcinoma tumor model [128]. Furthermore, TNFR2 deficiency and reduced metastasis were correlated with decreased amounts of CD11b⁺ and Gr1⁺ MDSCs and less Tregs in metastases site [128]. Remarkably, the pro-metastatic activity of TNFR2, which is evident from this study, turned out to be female specific, as in male mice, loss of TNFR2 did not result in a reduced amount of liver metastasis [129]. Furthermore, these findings correlated with the estrogen level and the finding that estrogen induces TNFR2 expression in isolated splenocytes [129]. Thus, reduced numbers of liver metastases were detected in ovariectomized female C57BL/6 mice, and this could be reverted by estradiol reconstitution. Moreover, Tregs and MDSCs of ovariectomized mice show reduced TNFR2 expression and reduced T-cell suppressive activity [129]. It is worth mentioning that TNFR2 might not only promote metastasis through the stimulating effect of TNFR2 on immune suppressive cells, but possibly also by affecting the activity of NK cells. It has been found that the expression of the immunosuppressive NKp30C isoform in gastrointestinal stromal tumors is tightly associated with autocrine TNF/TNFR2-induced upregulation of TRAF1 and cIAP2, along with downregulation of the activating NK cell-receptor NKp46 [130]. However, the possible causal relationship of this correlation remains to be clarified.

Aberrant expression of TNFR2 in certain cancers is accompanied by poor survival prognosis. As already discussed above, sTNFR2 levels are associated with poor survival prognosis in multiple myeloma, Hodgkin’s lymphoma, colorectal cancer, cutaneous non-Hodgkin’s disease and ovarian cancer [117,118,119,120,131]. Similarly, increased TNFR2 expression correlating with tumor size and clinical stage was reported for breast cancer [132]. High TNFR2 expression has also been found to be associated with shorter survival of patients suffering from non-small cell lung cancer (NSCLC) [133] and recently a systemic screening of almost 800 tumor cell lines derived of different cancer types revealed widespread TNFR2 expression, not unexpected especially in hematopoietic and lymphoid cancer cell lines [89]. These clinical data suggest that TNFR2 might also show pro-tumoral behavior within the tumor cell itself.

The pro-tumoral activities of TNFR2 must not necessarily originate from its role in specifying the properties of the TME but can also be based directly on oncogenic TNFR2 activities. Early on, it was found in a mouse model of DMBA/TPA-induced skin cancerogenesis that, particularly at the beginning of skin-tumor development, TNF is a major factor driving dermal inflammation and keratinocyte hyperproliferation [134,135]. Follow-up studies with TNFR1 and TNFR2 knockout mice revealed a crucial role of TNFR1 in skin cancerogenesis in this model, but also showed a significant contribution of TNFR2. These in vivo data correlated with in vitro studies showing a dominant role of TNFR1 in TNF-induced expression of GM-CSF and MMP9 in primary keratinocytes, but also a clear reduction in the inducibility of these factors in the absence of TNFR2 [136].

In a rare type of cutaneous T-cell lymphoma with poor clinical outcome, the Sezary Syndrome (SS), TNFR2 point mutations and aberrant gene duplications covering the TNFR2 gene locus lead to increased proliferation and cell growth due to enhanced NFκB signaling [137]. Likewise, TNFR2 promotes cancer-cell proliferation in colorectal cancer (CRC). Analysis of CRC tissue through immunohistochemistry revealed that high TNFR2 expression in the cancer cells is associated with higher expression of the proliferation marker Ki67 [138]. Furthermore, stable expression of TNFR2 in the SW1116 cell line enhanced proliferation in this study, while silencing of TNFR2 in the HT29 cell line reduced proliferation [138]. TNFR2 upregulation lead to enhanced protein kinase B (AKT) activity, suggesting that TNFR2 drives CRC progression via the phosphoinositide 3-kinase/AKT signaling pathway [138]. TNFR2 upregulation in colonic epithelial cells has also been reported in TNF-dependent cancer development associated with AOM/DSS-induced colitis [139]. Furthermore, follow-up studies gave evidence that TNFR2-induced MLCK expression disrupted tight junctions and thereby promoted cancerogenesis by triggering epithelial cell proliferation through luminal bacteria-induced inflammation [140].

It is well known that hepatic progenitor cells (HPCs) show abnormal compensatory proliferation in conditions of inflammation and tissue damage [141]. Indeed, TNF, presumably acting via TNFR2 and STAT3, has been identified as a crucial factor of HPC activation in a DEN-induced model of hepatocellular carcinoma [142]. Moreover, TNFR2 was recently identified as a driver of primary liver cancer (PLC) under involvement of YAP (Yes-associated protein) [143]. TNF can activate HPCs, but only TNF-TNFR2 interaction on HPCs leads to malignant transformation driving liver tumorigeneses. In this process, TNFR2 stimulation activates YAP signaling [143], an important component of the Hippo pathway, which regulates organ size and tumorigeneses [144]. Phosphorylated YAP is restrained in the cytoplasm and subsequently degraded. However, dephosphorylated YAP can enter the nucleus and, together with transcription factors of the TEAD (transcriptional-enhanced associate domain) transcription factor family, regulates the expression of several target genes [144]. The TNFR2 stimulation on HPCs promotes YAP signaling through the direct binding of heterogeneous nuclear ribonuclear protein K (hnRNPK), thereby stabilizing YAP on target-gene promoters and promoting malignant transformation of HPCs. Furthermore, single-cell RNA sequencing showed that the expression levels of TNFR2, YAP and hnRNPK in PLCs are enhanced and associated with a poor survival prognosis. All these findings point to the TNFR2-YAP axis as an important driver of HPC progression to PLC [143].

TNF has also been identified as a promoter of tumorigenesis in mice expressing the neu/erbB2 oncogene in the mammary epithelium or under control of the murine mammary tumor virus long terminal repeat [145,146,147]. Moreover, as mentioned above, TNFR2 expression has been positively associated with reduced overall survival time and disease-free survival in breast cancer patients [132]. Again, one pro-tumoral mechanism of TNF/TNFR2 signaling in breast cancer cells seems to be activation of the Akt pathway, this time resulting in upregulation of the DNA damage-repair protein poly(ADP-ribose) polymerase [148]. An association of TNF and the Hippo signaling pathway, as discussed above for liver cancer, has also been noted in breast-cancer-cell migration [149]. However, the role of TNFR2 has not been investigated in this study. Notably, there is evidence that TNFR2 can also act as a tumor suppressor in breast cancer. It has been observed that the loss of one TNFR2 allele in breast-cancer-prone MMTV-Wnt1 mice results in ductal hyperplasia in the mammary gland, higher numbers of mammary epithelial stem cells and last but not least, in an increased incidence of tumors with an aggressive metastatic phenotype [150]. The underlying mechanisms are poorly investigated but may involve autocrine TNF production [150].

Antagonistic anti-TNFR2 antibodies have been frequently used in vitro in the TNF field to differentiate between TNFR1- and TNFR2-mediated effects but there is also some in vivo experience with antagonistic TNFR2 antibodies. Antagonistic anti-TNFR2 antibodies are characterized by binding to TNFR2 in a way that prevents ligand binding. Importantly, anti-TNFR2 antibodies typically elicit strong agonism when presented in FcγR-bound form, irrespective of the epitope recognized [151]. It is therefore crucial for potential in vivo applications of ligand blocking anti-TNFR2 antibodies as TNFR2 antagonists to use them in an IgG isotype lacking/minimizing FcγR binding.

In 2017, Torrey et al. described two dominant antagonistic TNFR2 antibodies, whereby dominant describes the ability of these antibodies to inhibit Treg proliferation and TNFR2 shedding in an FcγR-independent manner and to kill OVCAR3 cells after prolonged incubation [101]. Both of these antagonistic antibodies recognize a similar region within TNFR2, and Torrey et al. speculated that these antibodies “fix” ligand-free TNFR2 dimers in an inactive state with inaccessible TNF-binding sites. The strongest effects with these antibodies were found when Tregs were isolated directly from ovarian cancers in comparison to Tregs from peripheral blood of ovarian cancer patients or healthy donors, indicating some kind of tumor Treg preference. This tumor preference might be explained by the fact that in tumors, the proportion of TNFR2⁺ Tregs is unusually high compared with healthy tissue, indicating more available targets for TNFR2-specific antibodies or by “priming” events occurring in the tumor microenvironment, making Tregs more dependent on TNFR2 signaling. One of these TNFR2-specific antibodies was also tested for its ability to inhibit CD4 T-cells of Sézary syndrome (SS) patients and various tumor cell lines. TNFR2⁺ CD4 cancer T-cell frequencies were lowered in response to the treatment and the ratio of Tregs/Teffs was restored compared to healthy controls, confirming the potential of antagonistic TNFR2 antibodies in cancer therapy [152]. An IgG2 mutant (C232S and S233S) of this antibody with a stabilized hinge region and broad separation of antibody arms was found to be particularly active in inhibition of cell growth of cancer cell lines with high TNFR2 expression [89]. Unfortunately, the mechanisms and mode of growth inhibition/cell death of the anti-TNFR2 antibody treated cells were not further investigated in the above-mentioned studies. It is thus unclear how inhibition of TNFR2 signaling translates at the molecular level in these studies in growth inhibition or cell death.

The activation of dendritic cells (DCs) is a goal of many immunotherapies. The stimulation of Toll-like receptor 9 (TLR9) on plasmacytoid DCs (pDCs) with CpG oligodeoxynucleotides (ODNs) can induce anti-tumor response in mouse models [153]. However, there is also evidence from ex vivo experiments with human pDCs that this kind of treatment can also trigger the induction of immunosuppressive Tregs [154,155]. The latter finding resulted in initial studies with the aim to boost the immunostimulatory effect of CpG treatment with the help of antagonistic TNFR2 antibodies (anti-mouse TNFR2 murine IgG1 M831 from Amgen; anti-mouse TNFR2 hamster IgG TR75-54.7, commercially available). In the CT26 mouse model of colon cancer, the administration of the anti-TNFR2 antibody M861 resulted in CpG-treated mice in a decreased number of TNFR2⁺ Tregs, increasing amounts of tumor-infiltrating CD8⁺ Teffs and higher survival rates [155]. In addition, after successful therapy, the mice were resistant to repopulation with the same tumor, but not to 4T1 breast cancer cells [155]. One has, however, to take care in the interpretation of these results with respect to the mode of action of the putative antagonistic anti-TNFR2 antibodies. Since these antibodies carries no mutations silencing FcγR interaction, it cannot be ruled out that FcγR-mediated activities or agonism of FcγR-bound antibody molecules contributed to the observed treatment effects. Indeed, it has been demonstrated that the antagonistic hamster anti-mouse TNFR2 antibody TR75-54.7 acts as a TNFR2 agonist after cross-linking in vitro [156] and binds to murine FcγRs II and III [157]. These considerations are also applicable for BI-1808, a fully human antagonistic ligand blocking anti-TNFR2 IgG1 antibody from BioInvent International AB [158], which is currently under investigation in a first clinical trial in patients with advanced solid tumors and cutaneous T-cell lymphoma (clinicaltrials.gov identifier NCT04752826).

In 2019, the mouse TNFR2 specific Y9 antibody has been described [159]. The antibody was generated as mouse IgG2a isotype, has TNF-blocking capabilities and binds to CRD1 of TNFR2. Y9 showed anti-tumor responses in syngeneic mouse models with several mouse cancer cell-lines indicated by CD4⁺ and CD8⁺ T-cell expansion, with increased functionality of CD8⁺ Teffs, downregulation of memTNF and reduced tumor size. These effects were not accompanied by Treg depletion, but FcγR interaction was necessary, suggesting that the agonism of FcγR-bound antibody molecules is important [159]. Similar results were obtained with agonistic TNFR2 antibodies in a CT26 syngeneic tumor model [157]. TNFR2 can also be efficiently and selectively engaged using recombinant oligomeric TNFR2-specific TNF mutants [3], but these reagents have not been tested yet in cancer models.

In addition to the anti-TNFR2 antibodies discussed above, several more are currently in preclinical/commercial development. Unfortunately, only a few peer-reviewed publications are available for most of these antibodies. Most often, these antibodies were only briefly presented at conferences, so a generalized consideration which strategy for TNFR2 targeting is best suited for which disease type is currently not possible.

To further improve clinical outcome, combination therapies of agonistic TNFR2-targeting reagents with other anti-cancer drugs or biologicals are considered. An obvious possible approach is the combination of TNFR2 agonists with immune checkpoint blockade (ICB). Over the last few decades, immunotherapies targeting immune checkpoint molecules such as CTLA-4 (e.g., Ipilimumab) or PDL1 (e.g., Avelumab) have shown great efficacy. Nevertheless, these therapies can promote undesirable autoimmune effects and only 30 % of treated patients show a good anti-tumor response. In the above-cited study by Tam et al. [159], the combination of the anti-TNFR2 antibody Y9 with its assumed FcγR-dependent agonism and a PDL1 blocking antibody resulted in superior anti-tumor activity. Furthermore, case and colleagues combined PD1 blockade with one of the antagonistic TNFR2 antibodies described by [101] in mouse colon cancer models (CT26 and MC38). The combination therapy exceeded the efficacy of the two corresponding monotherapies. The combination therapy led to a reduction in immunosuppressive Tregs and a normalization of the Treg/Teff ratio. Moreover, the best results were obtained when both blocking antibodies were administered simultaneously and not administered at different time-points [168]. Since the anti-TNFR2 antibody used was not silent for FcγR binding, the mode of action of the anti-TNFR2 antibody (ADCC of Tregs, TNFR2 blockade, TNFR2 engagement by FcγR-bound antibody molecules) is not fully clear in this study. Indeed, ADCC-mediated depletion of Tregs, along with enhanced activity of CD8+ T-cells, has been identified as the mode of action of the anti-TNFR2 antibody TY101 in a syngeneic murine tumor model [169].

In a genome-wide crispr-Cas9 screen, Vredevoogd et al. [170] identified TRAF2, cIAP1 and cIAP2, but also several other components of the TNF-stimulated TNFR1/TNFR2 signaling network as major factors regulating the T-cell sensitivity of tumors. Follow-up analysis revealed (i) that TNF expression in untreated cancers does not correlate with survival, but also revealed a positive correlation with responders of anti-PD1 therapy; (ii) that TRAF2 deficiency sensitizes tumor cells for the cytotoxic action of CD8⁺ T cell-derived TNF and (iii) that TWEAK-induced Fn14 activation sensitizes cancer cells for killing by CD8⁺ T cell-derived TNF-induced cell death by virtue of depletion of the available pool of TRAF2-cIAP1/2 complexes, thus by similar mechanisms as already described for TNFR2 (see also 2.) [170]. Thus, in the case of TNFR2-expressing tumor cells, TNFR2 agonists may not only improve TNF release by CD8⁺ T-cells, but may also sensitize the tumor cells for TNF-induced cell death. Indeed, TNFR2 has been identified as a molecular marker for ALL patients who have a good chance to respond to cIAP1/2 antagonists, such as birinapant, and this is associated with TNFR2-dependent recruitment of RIPK1 to TNFR1 and RIPK1-mediated cell death [171]. In summation, these recent findings prompt testing therapeutic strategies combining anti-PD1 checkpoint blockade with Fn14 activation to sensitize cancer cells for TNF killing and/or TNFR2 activation to enhance CD8⁺ T-cell activation and TNF release.