The role of IFN-γ-signalling in response to immune checkpoint blockade therapy

Inhibitory signalling in T lymphocytes (T cells), a negative feedback response to antigen stimulation, limits their ability to eradicate target cells as it renders them unresponsive to further antigen stimulation and/or weakens their effector functions. Immune checkpoint inhibitors (ICIs) are antibodies that antagonise T-cell inhibitory receptor (TCIR) signalling and thereby potentiate T-cell activation and responses, even following exhaustion from repeat tumour antigen exposure. ICI targeting cytotoxic T-lymphocyte-antigen 4 (CTLA-4), programmed cell death 1 (PD-1) and more recently lymphocyte activation gene-3 (LAG-3) have entered the clinic as monotherapies but work even more potently in combination, transforming outcomes for cancer patients, especially for those whose tumours carry a high mutational burden and are T-cell-inflamed [1–4]. IFN-γ is a cytokine that activates cellular immune responses, and its production is linked to immunosurveillance of cancer cells and clinical responses to immune checkpoint blockade therapy (ICBT). However, there is a dark side to IFN-γ wherein it promotes immune evasion and resistance to ICBT [5–8]. Potentially, selectively antagonising aspects of IFN-γ signalling associated with feedback inhibition or its role in immunosuppression could restore sensitivity to or circumvent resistance to ICBT.

IFN-γ is an ∼50 kDa homodimer formed by non-covalent antiparallel association of two 17 kDa polypeptide subunits with multiple N-glycosylations [9,10] secreted by T cells, natural killer (NK) cells, invariant NKT cells (iNKT cells), regulatory T cells (Tregs), γδ (γδ) T cells and B cells in response to interleukin (IL) -12, -15, -18, and -21 [11,12] and T-cell receptor (TCR) signalling, mediated by transcription factors Activator protein 1 (AP-1), T-box transcription factor TBX21 (T-bet), Eomesodermin (EOMES), nuclear factor of activated T cells (NFAT), and nuclear factor kappa B (NF-κB) [13–15]. IFN-γ signalling has been extensively reviewed elsewhere [5] and is outlined in Figure 1.

IFN-γ is a tumour suppressor which exerts direct anti-proliferative effects on a variety of tumour cells through effects on the cell cycle [16]. The gene encoding cyclin-dependent kinase (CDK) inhibitor p21^WAF1/CIP1 is a STAT1 target whose product inhibits entry into and progression through the cell cycle [17]. IFN-γ is also a cytotoxic cytokine that induces p53 stabilisation and stimulates apoptosis through both the intrinsic and extrinsic pathways, increasing expression of BCL2 associated agonist of cell death (Bad), BH3 interacting-domain death agonist (Bid), Bcl-2 homologous antagonist/killer (Bak), Fas Cell Surface Death Receptor (Fas) and Fas ligand (FasL), cytochrome c release from mitochondria, and activating caspases [18–20]. IFN-γ activates caspase 8 and caspase 3, triggering apoptosis in colorectal cancer cell lines [21]. In breast and lung cancer cells, IFN-γ enhances the expression of caspase 8, increasing sensitivity to apoptosis elicited by FasL/tumour necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL) [22–25]. Furthermore, IFN-γ mediates its tumouricidal effect through inducing necroptosis dependent on activation of the serine-threonine kinase receptor-interacting protein kinase 1 (RIPK1) [26].

IFN-γ inhibits angiogenesis, thereby inducing tumour ischemia, by impairing endothelial cell proliferation and survival. IFN-γ directly inhibits the proliferation of vascular endothelial cells [27–29]. IFN-γ also inhibits the production of vascular endothelial growth factor (VEGF) required for endothelial cell proliferation, survival, and migration, and promotes the production of anti-angiogenic substances. IFN-γ switches M2-polarised tumour-associated macrophages (TAMs), the main sources of VEGF in the tumour microenvironment (TME), to an M1 phenotype, reducing VEGF production [30,31]. In addition, IFN-γ produced by T helper 1 (Th1) cells makes macrophages directly cytotoxic to cancer cells, and causes them to secrete angiostatic chemokine ligand 9 (CXCL9) and CXCL10 to inhibit angiogenesis [32].

The immune stimulatory effects of IFN-γ have been reviewed extensively [5] and can be summarised as follows: IFN-γ enhances antigen processing and presentation in transformed cells as well as in professional antigen presenting cells and stimulates the recruitment, proliferation and polarisation of several leukocytes that promote or mediate clearance of transformed cells, including dendritic cells, macrophages, T helper cells, and cytotoxic lymphocytes. As ICBT requires cellular immune responses to be effective, IFN-γ signalling has emerged as a strong correlate of clinical response [33,34]. Data collated and analysed from more than 1000 patients with different types of cancers treated with ICBT showed that intratumoural CXCL9 expression strongly correlated with ICBT response [35]. CXCL9 is an IFN-γ target gene essential for recruiting T cells to tumour sites for immune clearance [36,37]. Transcriptomic analysis of ∼400 NSCLC patients treated with ICBT corroborated that high IFN-γ activity correlated with response, and, moreover, that the IFN-γ-induced immunoproteasome was essential [38]. Interestingly, collating data from seven different cohorts of melanoma patients treated with ICBT revealed that patients treated firstly with anti-CTLA4 followed by anti-PD-1 who displayed high IFN-γ signalling in tumour biopsies responded best to ICBT [39].

Further underscoring the requirement for IFN-γ signalling for ICBT response, multiple studies have reported that mutations which impair IFN-γ signalling in tumour cells promote ICBT resistance [40–49] (see also Table 1). Melanoma with intrinsic loss of IFN-γ signalling possess fewer tumour infiltrating lymphocytes (TILs) with impaired function [50]. Additionally, nonsense mutations in the Apelin Receptor (APLNR), which interacts with JAK1 to strengthen the IFN-γ response, were detected in patients with advanced melanoma or lung cancer resistant to ICBT; restoration of APLNR improved survival of mice with APLNR-defective melanomas treated with ICBT by increasing the responsiveness of tumour blood vessels to IFN-γ and augmenting the effectiveness of anti-tumour T cells [51]. However, mutations which impair IFN-γ signalling in cancer even with resistance to ICBT are relatively rare ([52–54] and Table 1). Moreover, using meta-analysis, Song and colleagues concluded that the loss of IFN-γ signalling sensitises tumour cells to immune effector cells in vitro, made tumours more susceptible to the host immune system in vivo, and patient tumours developing mutations in IFN-γ signalling before treatment were more likely to respond to ICBT [55]. Further, interferon-stimulated gene products were enriched in serum from non-responders with metastatic melanoma treated with ICBT [56]. Indeed, several studies have now established that IFN-γ signalling can result in ICBT resistance together with approaches to tackle this, the subject of the remainder of this review.

To date, IFN-γ-driven immune evasion and ICBT resistance mechanisms are mainly explained by (1) the induction of immunosuppressive molecules, (2) the up-regulation of negative-feedback regulators, (3) epigenetic effects of chronic exposure, and (4) driving dedifferentation of cancer cells or otherwise affecting antigen processing (Figure 2).

IFN-γ can promote immunosuppressive effects through increasing PD-L1 and indoleamine 2, 3-dioxygenase 1 (IDO1) expression in tumours [5]. PD-L1 is a ligand binding the PD-1 receptor on T cells resulting in their exhaustion [57]. Further, PD-L1 expression lowers tumour cell susceptibility to killing by NK cells [58]. Ubiquitin specific peptidase 15 (USP15) is a deubiquitinase which inhibits the the generation of IFN-γ-producing Th1 cells [59]. Mice with USP15-deficient T cells were found to generate excessive IFN-γ, which up-regulated expression of PD-L1 and CXCL12, causing effector T-cell exhaustion alongside infiltration of Tregs and myeloid-derived suppressor cells (MDSCs) into tumours, which promoted tumour formation [59]. IDO1 is an enzyme implicated in degradation of tryptophan, needed by T cells for activation, and simultaneously generating kynurenine, a metabolite capable of dampening immune responses. IDO1 activity can also induce the generation of Tregs to inhibit the killing function of cytotoxic lymphocytes (CTLs) [60–62]. Furthermore, IDO1+ DCs generated by IFN-γ inhibit allogeneic T-cell responses despite simultaneously secreting pro-inflammatory cytokines [63,64].

Interestingly, IFN-γ driven PD-L1 and IDO1 up-regulation became the dominant pathway inducing resistance to ICBT following TGF-β blockade in a mouse lung adenocarcinoma model [65]. While simultaneous TGF-β and PD-L1 blockade in this model resulted in significantly improved survival compared with anti-PD-L1 therapy alone, prolonged combination treatment promoted resistance by up-regulating PD-L1 and IDO1 through increased IFN-γ signalling [65]. However, tumour relapse could be prevented by JAK inhibition [65].

Besides PD-L1 and IDO1, there are several other immunosuppressive chemokines being upregulated by IFN-γ to elicit immune evasion. Ultraviolet radiation induces macrophages to secrete IFN-γ, which in turn induces melanocytes to secrete chemokine (C-C motif) ligand 8 (CCL8) that promotes inflammation and immune evasion [66]. In addition, an intrinsic high activity of IFN-γ signalling in tumour cells shapes an immunosuppressive TME. Several short-term melanoma cell cultures generated from patients who progressed on ICBT exhibited high intrinsic IFN-γ signalling activity while the corresponding tumour biopsies demonstrated low abundance of activated CD8+ effector T cells [52]. Moreover, IFN-γ was found to be responsible for promoting the expression of histocompatibility leukocyte antigen chain E (HLA-E) in tumour cells, which inhibits the anti-tumour responses of CD8+ T cells and NK cells [56].

Different negative regulators of IFN-γ signalling can be up-regulated in tumours in response to IFN-γ to induce immune evasion. Lymphocyte adapter protein (LNK) was up-regulated in melanoma and LNK genetic ablation restored IFN-γ signalling and a significant reduction of tumour size in different preclinical tumour models [67]. IFN-γ signalling was inhibited by protein tyrosine phosphatase non-receptor type 2 (PTPN2) whose deletion in tumour cells increased ICBT efficacy by enhancing IFN-γ-mediated effects on antigen presentation and cell growth inhibition [68]. N-acetylglucosaminyltransferase III (MGAT3) destabilises IFN-γ receptor α chain (IFN-γR1) through N-glycosylation and proteasome-dependent degradation also contributes to ICBT resistance [69]. IFN-γ-treated cancer cells also upregulate RIPK1, which connects TNF signalling to NF-κB that in turn induces an immunosuppressive chemokine program [70]. Deleting RIPK1 in tumour cells lowered infiltration by immunosuppressive macrophages in tumours, promoting TNF-induced killing and improving ICBT response [70]. Negative regulators of IFN-γ-signalling in immune cells also contribute to ICBT resistance. IRF2 induced by IFN-γ-in T cells results in exhaustion in multiple tumour types [71]. Deletion of IRF2 in CD8+ T cells increased effector functions to sustain long-term tumour control in response to ICBT [71]. IFN-γ receptor β chain (IFN-γR2) in CD8+ T cells also restricts anti-tumour responses and again selective deletion of IFN-γR2 reinvigorated stem-like T cells, strengthening anti-tumour immunity [72].

Although many reports have suggested that antagonising negative regulators of IFN-γ signalling reverses resistance to ICBT, other reports suggest the opposite rationale to overcome resistance. Tumour cell intrinsic loss of TNFα-induced protein 3 (TNFAIP3) promotes lung tumourigenesis associated with reduced CD8+ T cell–mediated immunosurveillance in patients and in mouse models [22]. The effects of TNFAIP3 loss were largely explained by an increased cellular sensitivity to IFN-γ signalling by the aberrant activation of TANK-binding kinase 1 (TBK1) and STAT1. Deleting the IFN-α/β (IFN-α/β) receptor restored cytotoxic T cell infiltration into tumours, thereby antagonising tumourigenesis [22]. Furthermore, IFN-γ signalling can be effectively maintained within tumours through EH domain-binding protein 1-like protein 1 (EHBP1L1) shielding JAK1 from proteasomal degradation. This mechanism promoted PD-L1 expression in a mouse model of renal cell carcinoma (RCC) and EHBP1L1 inhibition resulted in increased CTL activity and improved response to ICBT [73].

Chronic exposure of tumours to IFN-γ induces resistance to ICBT through epigenetic and transcriptional changes. Chronic-IFN-γ-treatment of tumour cells induced STAT1-associated epigenomic changes, promoting IFN-γ-independent expression of ISGs including ligands for multiple TCIRs, such as TNF receptor superfamily member 14 (TNFRSF-14) and MHC-II, that conferred resistance to ICBT in preclinical models [74]. Moreover, resistance to ICBT could arise in both antigen-sufficient and insufficient tumours through the up-regulation of T-cell terminal exhaustion in the former and increased immaturity of innate immune cells in the latter [75]. STAT1 is not the only transcription factor responsible for the transcriptomic and epigenomic changes in tumours elicited by chronic IFN-γ exposure. IRF3 and associated histone 3 lysine 4 mono-methylation (H3K4me1) enhance the chromatin accessibility and increase expression of several key immune-inhibitory IFN-stimulated genes [76]. Furthermore, IFN-γ promotes nuclear translocation and phase separation of Yes-associated protein (YAP) in tumour cells, which mediated adaptive resistance to ICBT by forming transcriptional hubs with TEA domain transcription factor 4 (TEAD4), E1A-associated protein p300 (EP300), and mediator of RNA polymerase II transcription subunit 1 (MED1) to maximise cancer-promoting gene transcription [77]. Additionally, IFN-γ induces expression of histone K (lysine) acetyltransferase 8 (KAT8), resulting in its interaction with IRF1, thereby promoting IRF1-induced chromatin acetylation and hence upregulation of PD-L1 expression in tumour cells, contributing to immune evasion [78].

Besides IFN-γ in tumours playing a major role in promoting resistance to ICBT, recent findings support that IFN-γ outside tumours also influences resistance to ICBT, especially in tumour-draining lymph nodes. Chronic IFN signalling induced epigenetic rewiring in tumour cells, which up-regulated expression of PD-L1 and MHC-I to suppress NK and T cells activities, thereby promoting lymph node colonisation. Within lymph nodes, tumour cells chronically exposed to IFN-γ induce Treg formation that then promote distant metastasis [79]. It has also been shown that lung-associated lymph nodes abundant in IFN-γ promote the development of immunosuppressive Th1-like Tregs, which interact with type 1 conventional dendritic cells and thereby limit the priming of CTL responses against lung cancer [80].

The panoply of antigens expressed by tumour cells and potentially presented to the immune system depends on their differentiation status [81]. Reprogramming of gene expression by IFN-γ can radically alter antigen expression as well as conferring other cellular traits contributing to cancer progression. IFN-γ induced interferon-induced protein with tetratricopeptide repeats 5 (IFIT5), C-X-C Motif Chemokine Receptor 4 (CXCR4) and branched-chain amino acid aminotransaminase 1 (BCAT1) all promote cancer cell stemness and metastasis [82–84]. Moreover, two further reports found that low-dose IFN-γ promoted tumour colonisation of lung tissues [85,86]. Further, which peptides end up being presented on major histocompatibility complex (MHC) class I depends in large part on how antigens are degraded by the proteasome [87]. Previously, two research groups reported that IFN-γ lowered neoantigen expression by inducing an immune proteasome, which caused their degradation in a way that prevented loading on MHCI [88,89]. In addition, IFN-γ can mask the recognition of tumour cells by CD8+ T cells through increasing expression of non-cognate MHC-I molecules in tumours such as Ova₂₅₇/H-2Kb complexes [90]. Cho and colleagues also mentioned a similar phenomenon and further explored inhibiting IFN-γ signalling to optimise peptide vaccine efficacy [90].

In general, there are three main directions to tackle this challenge therapeutically, which are (1) amplifying tumour antigen presentation, (2) inhibiting the action of T-cell inhibitory receptors, or (3) increasing tumour necrosis factor (TNF) signalling in tumours (Figure 3). However, many key regulators are generally not druggable, or are critical for host survival, or the clinical effect is not as potent as in preclinical models. Therefore, targeting IFN-γ-driven resistance to ICBT remains a significant hurdle. For instance, genetic deletion of RIPK1 in mice causes embryonic lethality [91]. In addition, inactivating the kinase activity of RIPK1 in cancer cells conferred minimal effects on preventing suppressive chemokine production and did not enhance TNF-mediated cytotoxicity [70]. In a large phase 3 trial (ECHO-301/KEYNOTE-252), IDO1 inhibitor Epacadostat co-administered with Pembrolizumab (a clinically approved PD1-targeting ICI) failed to improve progression-free survival compared with Pembrolizumab alone [92]. IDO1-inhibiton even protected tumour cells from switching off protein translation and Melanocyte Inducing Transcription Factor (MITF) production by IFN-γ [93].

To amplify antigen presentation in tumours, different researchers suggest using various immunomodulatory substances. BO-112, a nanoplexed version of polyinosinic:polycytidylic acid (poly I:C), activates double-stranded RNA (dsRNA) sensing (via protein kinase R and Toll-like receptor 3) and induces MHC class I expression via NF-κB signalling to restore the activation of tumour-specific T cells [94]. BO-112 was tested in a clinical trial (NCT04570332) to examine its efficacy when combined with Pembrolizumab in patients with advanced and/or metastatic melanoma that have progressed on Pembrolizumab alone treatment. However, the clinical trial results have not yet been reported. Paradoxically, up-regulating MHCI expression can despite making cancer cells more visible to T cells down-regulate immunosurveillance. Taniguchi and colleagues showed that the up-regulation of MHCI molecules on tumour cells after IFN-γ treatment resulted in a decrease in NK cell sensitivity and enhanced lung colonisation and metastasis of melanoma [95]. Takeda and colleagues also observed that breast adenocarcinoma cells overexpressing IFN-γ increased their resistance to NK cell attack [96].

JAK1/JAK2 inhibitor (JAKi) has been applied to inhibit expression of TCIR ligands and improved ICBT response in preclinical tumour models [74]. In addition, JAKi was reported to decrease tumour stemness, tumour-promoting chemokines and cytokines, and immunosuppressive MDSCs [97,98]. Currently, JAKi is being clinically evaluated in patients with advanced solid tumours (NCT02646748), non-small cell lung cancer (NCT02917993), and triple-negative breast cancer (NCT02876302). A trial (NCT03425006) that utilised sequential JAKi and anti-PD-L1 antibodies revealed enhanced clinical responses in patients with metastatic non-small cell lung cancer, particularly when administered after discontinuing treatment with anti-PD-1 antibodies [99]. The researchers also observed that JAKi promoted increased CD8 T-cell plasticity and enhanced therapeutic responses in exhausted and effector-memory clonotypes [99]. However, concerns about JAKi safety have been raised due to its myelosuppressive effects in humans [100].

TNF is present in lower abundance in tumours in non-responders to ICBT compared with responders [101]. Consequently, reducing the threshold to TNF-induced cytotoxicity in tumours could enable T cells to effectively eliminate tumour cells even in cancer patients with low TNF abundance [101]. Receptor-associated factor 2 (TRAF2) was identified as a gatekeeper of TNF cytotoxicity threshold in tumours, by suppressing RIPK1-dependent cell death [101]. Applying a second mitochondria-derived activator of caspase (SMAC) mimetic degrading both Baculoviral IAP repeat-containing protein 2 (BIRC2) and BIRC3, which target the TRAF2 interaction partner cellular inhibitor of apoptosis protein 1 (cIAP1), restored sensitivity to ICBT in several preclinical tumour models [101]. Addtionally, inhibition of E3 Ubiquitin-Protein Ligase RNF31 disrupted the cell ligand-bound TNF receptor complex I, which resulted in TNFAIP3 and inhibitor of NF-κB kinase (IKK) complex loss. This sensitised tumour cells with MHC/antigen-deficiency which could not be killed by immune cells directly and synergised with ICBT [102]. Another way to lower the threshold to TNF cytotoxicity that was recently suggested is to target the innate immune TANK-binding kinase 1 (TBK1). TNF- or IFN-γ-treated TBK1-null tumour cells promoted RIPK- and caspase-dependent cell death regulators [103]. Interestingly, TBK1-loss did not slow down tumour growth in immunodeficient mice; however, targeting TBK1 pharmacologically promoted a more immunostimulatory effect in the TME because it increased the infiltration of CD8 T cells and M1 macrophage into tumours [103]. Also, TBK-inhibitor treatment of CD8 T cells and M1 macrophages increased their expression of effector cytokines, including TNF and IFN-γ [103].

• The importance of the field: IFN-γ is a pleiotropic cytokine with pro- and anti-tumour effects. However, following recent breakthroughs, how it drives resistance to ICBT is becoming clearer.

• A summary of the current thinking: IFN-γ-driven resistance to ICBT is associated with its low dose, induction of negative regulators, chronic exposure, and overexpression of immunosuppressive molecules in tumours.

• Comments on future directions: To tackle this mode of resistance, researchers should explore ways to identify novel positive or negative regulators of IFN-γ signalling in tumours, which could be pharmacologically modulated.

The authors declare that there are no competing interests associated with the manuscript.

Work on IFN-γ-driven ICBT resistance in the Hurlstone laboratory is funded by a research sponsorship agreement between Ribon Therapeutics and The University of Manchester [grant number 821034] from the Melanoma Research Alliance (MRA) [grant number 22-0091] from Worldwide Cancer Research (WCR), and awards from the Skin Cancer Research Fund (SCaRF) and Melanoma UK (to A.H.) C.W.W. was funded by a scholarship from the Hong Kong Scholarship for Excellence Scheme (HKSES). Y.Y.H is funded by a University of Manchester-China Scholarship Council joint scholarship.

Open access for this article was enabled by the participation of University of Manchester in an all-inclusive Read & Publish agreement with Portland Press and the Biochemical Society under a transformative agreement with JISC.

Conceptualisation: all the authors; Visualisation: C.W.W. and Y.Y.H.; Writing – original draft: all the authors; Writing – review and editing: C.W.W. and A.H.

Figures 1 and 2 were created with BioRender.com. We apologise to any colleagues whose relevant work we couldn’t cite due to article length restriction.

BCAT1

branched-chain amino acid aminotransaminase 1

cIAP1

cellular inhibitor of apoptosis protein 1

CTLA-4

cytotoxic T-lymphocyte-antigen 4

dsRNA

double-stranded RNA

HLA-E

histocompatibility leukocyte antigen chain E

ICBT

immune checkpoint blockade therapy

ICI

immune checkpoint inhibitor

IFN-γ

interferon-γ

LAG-3

lymphocyte activation gene-3

MDSC

myeloid-derived suppressor cell

MHC

major histocompatibility complex

PD-1

programmed cell death 1

RCC

renal cell carcinoma

TCIR

T-cell inhibitory receptor

TNF

tumour necrosis factor