IFNγ Signaling in Natural and Therapy-Induced Antitumor Responses

IFN was described in 1957 as a cellular factor induced by virus that interfered, and thereby their name, with subsequent viral infections (1). The effect of IFN was proven to go beyond antiviral responses when its antitumor abilities on cancers not induced by virus were documented (2).

Early biochemical characterization showed that in reality there were several types of IFNs and a second type, preferentially produced by leukocytes, was identified in the mid 1960s (3). This type II IFN was distinct structurally and functionally, and was named after type I IFNs only because historical reasons. It was definitively named IFNγ in the 1980s and it was shown to also have antitumor effects (4). Nowadays, IFNγ is considered an extremely pleiotropic cytokine that, because of its disparate effects on different cell types, has attracted the attention of biomedical scientists working in different fields (recently reviewed in ref. 5).

IFNγ is a protein of 160 amino acids with a molecular weight of 17 kDa that forms a compact globular homodimer (6). The locus of the gene encoding IFNγ (gene name, IFNG) is epigenetically silenced in all cells except a restricted number of immune cells (reviewed in ref. 7). These include mediators of innate immunity such as natural killers (NK; ref. 8) and NK T cells (NKT; ref. 9) as well as components of the adaptive response, such as CD8⁺ and CD4⁺ T cells differentiated to the Th1 phenotype (10). Activation of cellular receptors or certain cytokines promotes the production of IFNγ. Thus, NK cell–activating receptors or the T-cell receptor (TCR) trigger the synthesis of IFNγ. Regarding the control by cytokines, while IL12 and IL18 promote the production of IFNγ, TGFβ, and IL10 inhibit it (7). In some cases, the antagonism is reciprocal, IFNγ prevents the production of IL10 (11, 12).

The receptors for IFNγ (IFNGR) are expressed on every cell type, except mature erythrocytes (13). The binding of IFNγ to IFNGR1 triggers the formation of a complex of four molecules (two IFNGR1s and two IFNGR2s; refs. 14, 15). IFNGR1 constitutively associates with janus kinase family (JAK1), while IFNGR2 is associated with a similar kinase, JAK2 (16). This complex recruits and phosphorylates two STAT1 molecules (17). Phosphorylated STAT1 dimerizes, translocates to the nucleus and binds to regions of the genome known as IFNγ-activated sites (GAS), present in the promoter of IFN-stimulated genes (ISG), initiating transcriptional programs that depend on the specific cell type and typically include hundreds of genes (Fig. 1; ref. 18). An online site, the interferome, is an expanding searchable database of IFNγ-regulated genes in an increasing number of tissues and cell lines (19).

One of the ISGs, IFN regulatory factor-1 (IRF-1), in turn regulates the transcription of dozens of genes, adding an additional layer of complexity to the transcriptional program deployed by IFNγ. Some of the genes whose expression is controlled by IRF-1 play antitumor roles (20).

After several periods of skepticisms, nowadays it is widely accepted that IFNγ, acting on immune and nonimmune cells, protects the organism from transformed cells (21). Eventually, some of these cells escape (22), and their malignant progression is further shaped by the immune system, this process is known as immunoediting and IFNγ also plays a central role in it (reviewed in refs. 13, 23).

Some cytokines have a very restricted physical range, acting only on the cell in contact with the cell that produced said cytokine. Others can diffuse to reach neighbor cells. These modes of action are known as synaptic or multi-directional, respectively. The mode of action of IFNγ is considered leaky synaptic, that is, it reaches preferentially the target cell but also acts on neighboring cells (24).

The pleiotropism exhibited by IFNγ likely stems from the widespread expression of its cognate receptors, present in every nucleated cell (13). Although the effects of IFNγ on immune cells are better understood, recently attention has shifted to those on nonimmune cells because they seem to be central to the interplay between cancer cells and immunotherapies.

The IFNγ canonic role is the coordination of the innate and adaptive branches of the immune system by acting on monocytes, macrophages, dendritic cells, and lymphocytes.

IFNγ has been shown to promote differentiation of myeloid progenitors to monocytes, rather than neutrophils (25), and it was the first activating factor described for macrophages (26). IFNγ promotes the differentiation of macrophages to the inflammatory phenotype (known as M1). In macrophages already differentiated to the M1 phenotype, IFNγ continues playing a role by promoting the expression of the inducible form of nitric oxide synthase (iNOS), which is required for microbicidal activity (27). In addition, M1 macrophages promote cytotoxicity in CD8⁺ lymphocytes and enhance antigen presentation (28).

IFNγ activates antigen presentation in immune, major histocompatibility complex (MHC) class II-expressing cells: macrophages, dendritic cells, and B lymphocytes. It increases the expression of the MHC class II transactivator CIITA, a transcriptional activator involved in the expression of human leukocyte antigen (HLA) class II components, and the invariant chain (Ii), a protein necessary for the correct maturation of the HLA class II-antigenic peptide complex (29).

IFNγ also has an effect on T cells able to recognize the antigens presented by these activated antigen-presenting cells. It induces the differentiation of CD4⁺ T cells to Th1 phenotype, which is characterized by the ability to produce IFNγ, which maintains Th1 lineage and inhibits the differentiation to the other main Th cell subsets, establishing a feed-forward mechanism (30).

Early experiments showed that exposure of CD8⁺ T cells to IFNγ was required for the full cytolytic activity of CD8⁺ T cells (31, 32). IFNγ induced the upregulation of granzyme, the IL2 receptor and the transcription factor T-bet, which along with the related transcription factor eomesodermin, induces the production of IFNγ, again establishing a feed-forward loop (33).

All the above effects support the consideration of IFNγ as a proinflammatory cytokine. However, it can also blunt inflammation by promoting the production of immunosuppressive factors and the activity of cells that inhibit the proliferation or activation of T cells. The pro- and anti-inflammatory activities of IFNγ are probably sequential and the latter avoids deleterious effects when inflammation is no longer needed. IFNγ induces the production of IDO (34) which, through the catabolism of tryptophan, generates an immunosuppressive environment (35). Inhibitory cells promoted by IFNγ include myeloid-derived suppressor cells (36), regulatory T cells (37) or macrophages differentiated to the anti-inflammatory phenotype (known as M2; ref. 35). In addition, IFNγ may enhance the expression of inhibitory molecules such as PD-L1 and PD-L2 (38, 39).

In addition to the effects described above on cells that present antigens via MHC class II, IFNγ increases presentation via MHC class I in all nucleated cells. This activation occurs at three levels: the generation of peptides, their transport, and their exposure at the cell surface. By triggering the expression of specific subunits, IFNγ modifies the composition of the proteasome and thereby its proteolytic specificity. Thus, IFNγ changes the repertoire of peptides presented by MHC class I complexes (40, 41). IFNγ also promotes the expression of TAP-1, a transmembrane endoplasmic reticulum protein that transports the peptides generated by the proteasome in the cytosol to the lumen of the endoplasmic reticulum (42). In addition, IFNγ promotes the expression of both MHC class I genes (43) and β₂ microglobulin (B2M; ref. 44), another component of the MHC class I complex. Finally, IFNγ upregulates costimulatory molecules such as CD80 and CD86, which leads to a swift T-cell activation (45). In summary, IFNγ increases tumor immunogenicity by globally upregulating the antigen-processing machinery at different levels.

Also, in nonimmune cells, IFNγ has seemly opposite roles. While the effects on antigen processing result in increased exposure to immune cells, IFNγ also upregulates the inhibitory factor PD-L1 (38, 39), a mechanism to avoid excessive immune activation exploited by cancer cells to avoid the attack of T cells.

IFNγ also exert intrinsic antiproliferative/cytostatic and cytotoxic effects apparently unrelated to its effects on the immune system or the immunogenicity of cells. Even though these effects explain at least in part the antitumor effect of IFNγ and have been observed by many different researchers in diverse models, the mechanisms underlying them are still being elucidated and they seem to vary with the cellular context.

IFNγ induces cell-cycle arrest in several experimental models. In a human fibrosarcoma cell line, it inhibited cyclin-dependent kinase 2 (CDK2; ref. 46). The mechanism behind was the upregulation of the CDK inhibitor p21^waf1/CIP1 mediated by IRF-1 (47, 48). Similarly, in a murine melanoma model, activated CD8⁺ lymphocytes blocked tumor growth in part because they arrested cancer cells in the G₁-phase of the cell cycle. The mechanism was the upregulation of the CDK-inhibitor p27^Kip1 by the IFNγ secreted by the lymphocytes. In this case, the upregulation was not transcriptional but through the downregulation of the S-phase kinase-associated protein 2 (Skp2), which mediates the degradation of p27 ^Kip1 (49).

The arrest in the cell cycle induced by IFNγ, acting alone or in combination with TNFα, has the characteristics of cellular senescence, at least in some instances (50, 51). It appears that this IFNγ-induced senescence is necessary to restrict the proliferation of cells that resist killing by immune cells (52).

IFNγ has been shown to induce apoptosis in cells derived from different tumors, including osteosarcoma (53), melanoma (54), glioblastoma (55), neuroblastoma (53), breast cancer (53, 56), and colon adenocarcinoma (57). The mechanisms behind this pro-apoptotic effect are highly diverse.

Initially, using primary hepatocytes, the apoptosis induced by IFNγ was found dependent on IRF-1 and p53 (58). A subsequent report contradicted partially this mechanism by showing that the apoptosis induced by IFNγ was dependent of IRF-1 but independent of p53 (53) in a variety of transformed cell lines.

There is general coincidence on the central role of IRF-1 in the regulation of apoptosis. It has been shown that it upregulates components of the intrinsic apoptotic pathway such as Bak (56) or Bcl2A1 (54). In addition, it upregulates components of the extrinsic apoptotic pathway such as the death receptor Fas and its ligand (57, 58) or the dependence receptor UNC5H2 (54). Finally, IFNγ, through IRF-1, upregulates initiator (Casp 8,9; refs. 53, 56), executioner (Casp 3,7; ref. 56) and pyroptotic caspases (Casp 1,11; refs. 53, 57).

In addition to these direct effects on death receptors or components of the apoptotic machinery, IFNγ has been shown to induce apoptosis indirectly by suppressing the pro-survival signals conveyed by the PI3K/Akt and NFκB pathways (55). Thus, although there is a consensus on the pro-apoptotic effect of IFNγ on a wide variety of cells, the mechanisms behind seem to be multiple and highly dependent on the cellular context.

Ferroptosis is a nonapoptotic form of cell death that strictly depends on intracellular iron and is triggered by the impairment of antioxidant defenses (59). IFNγ, secreted in a leaky synaptic way by activated CD8⁺ cells, induces ferroptosis in mouse ovarian and melanoma cancer cells and in human fibrosarcoma cells (60). The mechanism behind is the downregulation of the expression of the transporters of cysteine SLC7A11 and SLC3A2. This impairment of cysteine uptake and the subsequent increase of lipid peroxides, which are cleared under normal circumstances by cysteine, causes tumor cell death via ferroptosis (60).

In summary, although evidence at hand shows that IFNγ has an antiproliferative effect, the mechanisms range from cell-cycle arrest to induction of apoptotic and nonapoptotic cell deaths. Future work will determine whether this wide variety of mechanisms are real and depend on cellular context or are artifacts inherent to some experimental models, such as two-dimensional in vitro cultures, that do not faithfully recapitulate the in vivo situation.

To the effects of IFNγ on the nonimmune compartment described above, induction of cell-cycle arrest and cell death, a third aspect relevant for cancer treatment has been recently acknowledged: the requirement of IFNγ signaling in cancer cells for the efficient killing by T cells targeting them.

Mutations inactivating IFNγ signaling, which confer resistance to T cell–induced cell death, have been found in different experimental models resistant to immunotherapy (61, 62). Conversely, a transcriptomic analysis of baseline and on-therapy tumor biopsies from advanced melanomas treated with nivolumab (anti-PD1), alone or plus ipilimumab (anti-CTLA4), revealed an active IFNγ signaling signature, particularly genes involved in antigen processing, as a biomarker of clinical response (63). That is, cancer cells with active IFNγ signaling are more likely to respond to death mediated by T cells.

In a group of patients with metastatic melanoma treated with ipilimumab, genomic defects in IFNγ pathway genes were observed in resistant tumors. In addition, mice bearing melanoma tumors with knockdown of IFNGR1 were resistant to treatment with immune checkpoint inhibitors (64). Along the same lines, melanomas that initially responded to PD-1 blockade became resistant after acquiring genetic lesions that inactivated JAK1 or JAK2 (65), and loss-of-function mutations in JAK1/2 lead also to primary resistance to anti-PD-1 therapy (66). Furthermore, tumors from patients with metastatic melanoma who relapsed from pembrolizumab (anti-PD1) therapy bore loss-of-function mutations in IFNγ signaling genes, such as JAK1, JAK2, and B2M (67). Considering the combined results of these studies, although inactivation mutations in components of IFNγ signaling can be found more frequently in nonresponders, the most widespread mechanism of resistance seems to be the inactivation of B2M (68).

In vivo CRISPR screenings, aimed to discover factors that confer resistance to immune checkpoint inhibitors, identified several indirect regulators of the IFNγ signaling pathway. The knockout of the protein phosphatase PTPN2, in principle, a druggable protein, sensitized to anti-PD-L1 therapies. Mechanistically, deletion of PTPN2 enhanced the effect of IFNγ on antigen presentation and growth suppression (69). Similarly, loss of the RNA-editing enzyme ADAR1 resensitizes cells that have become resistant to immune checkpoint inhibitors because of impaired antigen presentation (70). A summary of these mechanisms of resistance is shown in Fig. 2.

It should be noted, however, that in contrast with these reports, some patients with tumors bearing mutations in the IFNγ pathway nonetheless respond to immune checkpoint inhibitors (71) and, conversely, high levels of IFNγ, that could reflect activation of the pathway, are found in the serum of patients not responding to immune checkpoint blockade (72).

An additional level of complexity that may reconcile these results arises from tumor heterogeneity. An analysis of single-cell RNA sequencing in lung cancer showed that genes encoding components of IFNγ signaling and ISGs, including MHC class II, were heterogeneously expressed (73). Thus, under the pressure exerted by immune therapies cells with low ability to transduce the signal conveyed by IFNγ are expected to be rapidly selected.

Finally, long-term exposure of cancer cells to IFNγ may increase the expression of ligands of multiple inhibitory receptors that block the antitumor effect of T cells, indicating that long-term blockade of IFNγ signaling in cancer cells may improve the destruction of cancer cells by the immune system (74). Furthermore, subsequent results show that prolonged IFNγ signaling in the immune compartment does not impair the immune response against tumors (75). Thus, according to these reports, to maximize antitumor efficacy, IFNγ should be inhibited in the tumor compartment but preserved in the immune compartment.

Collectively, this evidence indicate that the upregulation of antigen processing induced by IFNγ is critical for the activity of T cells directed against tumor specific or associated antigens. One line of resistance deployed by cancer cells is the upregulation of inhibitory factors such as PD-L1, and additional inhibitory factors, which bind to their cognate receptor in T cells inhibiting those that recognize tumor antigens via their TCR. Once these mechanisms of resistance have been disabled by treatment with blocking antibodies, cancer cells use additional lines of resistance, such as the downmodulation of IFNγ signaling that, in turn, downmodulates antigen presentation and, thus, the recognition of cancer cells via the TCR.

Regarding adoptive cell therapy, recent reports show that impairment of IFNγ signaling also confers resistance, and that downmodulation of antigen presentation is not the only mechanism by which IFNγ promotes resistance to immune therapies. While tumors expressing human papillomavirus that responded to T cells expressing a TCR targeting a papillomavirus antigen, showed no mutations affecting IFNγ signaling, resistant tumors had mutations in IFNGR or loss of HLA (76). On the other hand, resistance to T cells redirected via T-cell bispecific antibodies (TCB) or chimeric antigen receptors (CAR) is also achieved by downmodulating IFNγ signaling in cancer cells (77). Because TCBs or CARs act independently of antigen presentation via MHCs, these results highlight the importance of IFNγ signaling in the death induced by T cells engaged to tumor cells.

In summary, evidence shows that IFNγ signaling in cancer cells is critical for the efficient antitumor effect of T cells. Under this pressure, malignant cells tend to disrupt IFNγ signaling. This disruption leads to reduced presentation of antigens and reduced sensitivity to killing by lymphocytes.

Given the pleiotropic effects of IFNγ on the immune compartment, and its direct antitumor effects, is not surprising that both activation and inhibition of IFNγ signaling has been tested in the clinic. Two types of therapeutic interventions can be envisioned: (i) inhibition or activation of IFNγ signaling in both the immune and the tumor compartments and (ii) inhibition or activation of IFNγ signaling in the immune but not the tumor compartment or vice versa. To date only the first possibility can be accomplished pharmacologically in the clinical setting.

To activate IFNγ signaling, two strategies have been developed: direct administration of recombinant IFNγ or activation of the cGAS/stimulator of interferon genes (STING) pathway, a cytosolic DNA-sensing pathway that drives the production of type I IFNs and, as consequence, IFNγ, that boosts immune responses against tumors (78).

The administration of IFNγ has been tested in different tumors, including melanoma, lung adenocarcinoma, renal cell carcinoma, breast and ovarian carcinoma (79). Currently, there are nearly 30 open or recently completed clinical trials that use recombinant IFNγ in combination with additional therapies. Although some responses have been observed, overall, these studies have not resulted positive and patients have suffered considerable side effects.

There are several open clinical trials using different STING agonists, as single agents or in combination with checkpoint inhibitors. So far, no serious toxic effects and promising antitumor activities have been observed in these trials (80, 81).

In addition, a significant effort is being done to discover new compounds to overcome impaired IFNγ response. Intratumoral delivery of BO-112, a nanoplexed version of polyisosinic-polycytidylic acid (poly I:C) restored the efficacy of adoptive cell transfer against resistant B16-JAK1^KO tumors. BO-112 activates double-stranded RNA sensing (via protein kinase R and Toll-like receptor 3) and induced MHC I expression via NFκB, independent of IFNγ signaling (82).

The inhibition of IFNγ has been developed in the clinic primarily not because their potential effect on the antitumor response but as precision therapy. Genetic activation of JAK2 is considered a driver of various hematologic tumors, including a subgroup of myeloproliferative neoplasms and lymphoblastic leukemia (83, 84). One decade ago, ruxolitinib, a JAK1/2 inhibitor was approved for the treatment of some of these myeloproliferative neoplasms (85). To date, 50 clinical trials using JAK inhibitors as single agent or in combination with other therapies, including immunotherapies, have been performed or are still ongoing. However, the initial results of the combination of anti-PD1 therapies and JAK1/2 inhibitors have not been positive. This unfavorable outcome may be due to a reduced T-cell activation (86).

In summary, targeting the IFNγ pathway in the clinic is currently ongoing with different outcomes. If the disparate effects of IFNγ signaling in the tumor and immune compartment during different stages of tumor progression are confirmed, future effort should be focused in targeting separately these compartments.

J. Arribas reports grants from Roche, Synthon/Biondys, and Molecular Partners and grants and personal fees from Menarini during the conduct of the study; in addition, J. Arribas has a patent for EP 0930183.5 issued, licensed, and with royalties paid; a patent for P200801652 issued, licensed, and with royalties paid; and a patent for EP20382457.8 pending, licensed, and with royalties paid. No disclosures were reported by the other authors.

Our research in this area is supported by the Breast Cancer Research Foundation (BCRF-20-08), Instituto de Salud Carlos III (PI19/01181), Asociación Española Contra el Cáncer (GCAEC19017ARRI), and Fundación BBVA (CAIMI VHIO-FBBVA 2018-2021). A. Martínez-Sabadell was funded by by the Instituto de Salud Carlos III (PFIS FI20/00188). E.J. Arenas was funded by the AECC (POSTD211413AREN).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.