Introduction

Cancer immunotherapy including immune checkpoint blockade (ICB) and adoptive cell therapy (ACT) can induce dramatic and durable responses in subsets of patients. The fundamental principle of cancer immunotherapy concerns the induction and propagation of the antitumor immune response. The tumor microenvironment (TME) constitutes an intricate milieu including transformed cancer cells, stromal cells and immune cells. Immune cells subject to the hostile conditions of the TME undergo metabolic reprogramming as well as functional and phenotypic changes that frequently compromise their tumoricidal capacity. This is recognized as a significant obstacle to successful cancer immunotherapy.1–3

Reactive oxygen species (ROS) are chemically reactive derivatives of molecular oxygen produced in reduction-oxidation (redox) reactions during normal aerobic metabolism. ROS are essential intermediates in diverse signaling pathways and are tightly controlled within physiological ranges facilitating signal transduction. Aberrant redox homeostasis and elevated ROS levels in the TME have historically made ROS elimination an attractive therapeutic strategy. However, it is now recognized that abnormal ROS levels trigger different effects within the TME depending on target cells. Elevated ROS can promote tumor initiation, progression, metastases and immunosuppression. However, ROS are also critical to the induction of antitumor immunity.1–5 This complexity in the effects of ROS should help explain the controversial observations that dietary antioxidant supplementation can promote tumor progression and metastasis.4 6

Here we summarize ROS production, redox regulation and the contributions of ROS to immune responses under physiological conditions. We discuss ROS in the context of cancer and immune cells within the TME. Lastly, we will discuss strategies to use ROS modulation to facilitate successful cancer immunotherapy. For a schematic summary of the concepts discussed here, see figure 1 and table 1.

Figure 1

Hypothesis regarding redox modulation for efficient cancer therapy. This simplified scheme illustrates how reactive oxygen species (ROS; red stars) can play different roles at different levels in cancer, and how we propose that directed redox modulation may be used to achieve more efficient therapy outcome. Typically, cancer cells and stroma of the tumor microenvironment (right half of the figure, blue background) promote the accumulation of higher ROS levels that hamper the functionality of immune effector cells (left half of the figure, orange background), including dendritic cells, natural killer cells (NK cells), B cells, and cytotoxic T cells. Additionally, immunosuppressive cells, including myeloid-derived suppressor cells (MDSCs) and regulatory T cells, exert negative control over immune effector cells through ROS-dependent mechanisms. In the context of tumor therapy, further increased oxidation in cancer cells can trigger intolerable levels of ROS, leading to either immunogenic cell death (ICD) or non-ICD. ICD, in particular, can give release of damage-associated molecular patterns (DAMPs) further promoting adaptive immune responses. A therapeutic use of antioxidants or the deliberate increase of endogenous antioxidant systems in immune effector cells rendering them more resistant to ROS can improve their efficacy. A combination of approaches resulting in an increase of ROS levels in cancer cells and the improved resistance to ROS in immune effector cells may have the potential to synergistically enhance the efficacy of cancer therapy. Green arrows: redox modulatory processes supporting cancer therapy. Red arrows: redox modulatory processes counteracting efficient therapeutic outcome in cancer therapy. Figure created with BioRender. See table 1 and the main text for further details.

Table 1

Illustrative examples of the complex redox modulatory effects in cancer with potential impact on treatment outcome. This table summarizes some of the concepts discussed in this article, focusing on redox modulatory factors and perturbations affecting overall outcome in cancer therapy. Some selected references are given. For a scheme overview, see figure 1, and for detailed discussions please see the main text

ROS sources and maintenance of redox balance

The ROS molecules include radical compounds like hydroxyl radicals (·OH) and superoxide anions (O2·–), or non-radical molecules like hydrogen peroxide (H2O2) and singlet oxygen (1O2). H2O2 is physiologically the most important ROS as it is chemically rather stable, traverses membranes through aquaporins and acts on key cellular redox relays.7 Endogenous cellular ROS are mainly produced via leakage from the mitochondrial electron transport chain and through NADPH oxidases (NOXs), with the latter being transmembrane enzyme complexes deliberately producing ROS. Endoplasmic reticulum oxidative protein folding and fatty acid β-oxidation in peroxisomes additionally contribute to ROS production. Environmental exposures including UV light, pollutants, tobacco smoke and therapeutic interventions including chemotherapy or radiotherapy can also produce ROS.7

Physiologically, ROS molecules act as essential signaling molecules controlling numerous cellular processes.7 ROS are dynamically maintained within a physiological range via compartmentalized production and buffering mechanisms. Oxidative stress is however defined as the state of disequilibrium with an excess of ROS over physiological levels, which may be due to excessive production and/or deficiencies in antioxidants and cellular reductive enzyme systems. Oxidative stress can in turn damage biological macromolecules including lipids, proteins and nucleic acids, thus compromising cellular structure, integrity and function.6 ROS can trigger several cell death modes including ferroptosis, pyroptosis, apoptosis, oxeiptosis and NETosis.2

Maintenance of redox homeostasis

Redox homeostasis incorporates redox sensing, redox signaling, redox responses and feedback control pathways.7 Here we will shortly summarize pivotal contributors to redox homeostasis.

NRF2 (nuclear factor E2-related factor 2) is a transcription factor considered a master regulator of redox homeostasis. NRF2 is regulated by KEAP1 (Kelch-like ECH-associated protein 1), which in non-stressed cells binds to NRF2 promoting its degradation. During exposure to electrophiles or oxidative stress, KEAP1 is rendered non-functional allowing newly synthesized NRF2 to enter the nucleus and activate its target genes. Activated nuclear NRF2 binds to antioxidant response elements initiating transcription of antioxidant, metabolic and detoxifying enzymes, with major importance also in cancer.8 Numerous other transcription factors and signaling pathways also contribute to redox regulation and are redox-regulated, notably hypoxia inducible factor 1-alpha (HIF-1) and Nuclear factor kappa-light-chain-enhancer of activated B cells (NF-kB).9

The glutathione and the thioredoxin systems represent the major enzymatic reductive and antioxidant cellular systems.10 Glutathione reductase uses NADPH to reduce glutathione disulfide, with reduced glutathione being used by several reductive and antioxidant enzymes including glutaredoxins, glutathione peroxidases and glutathione S-transferases.11 Isoenzymes of thioredoxin reductases use NADPH to reduce the active site of several thioredoxin-fold proteins, reducing downstream targets including antioxidant peroxiredoxins and methionine sulfoxide reductases.12 Both systems are also required to support ribonucleotide reductase, being essential for the synthesis of DNA precursors.13 Additional enzymatic antioxidant systems include superoxide dismutases (converts superoxide into H2O2) and catalase (converts H2O2 to water and oxygen).1

Non-enzymatic antioxidants comprise low molecular-mass compounds including vitamins and micronutrients. These act as essential cofactors for numerous antioxidant enzymes and some directly scavenge ROS. In the context of cancer immunotherapy notable examples include vitamins C, E and A. Importantly antioxidant compounds can have prooxidant effects. The net pathophysiological effects of “antioxidant systems” will always be context-dependent.14

Redox regulation and redox regulated signaling pathways in cancer cells

Cancer cells are not only believed to have an inherently generally increased oxidative stress, but many intracellular signaling pathways that are required for cancer cell growth and progression are also redox regulated.

Several growth factors of significant importance in cancer act by binding to and stimulating receptor tyrosine kinases (RTKs), which are often mutated into hyperactive states in cancer.15 16 Importantly, intracellularly the tyrosine kinase signaling cascades are counteracted by phosphotyrosine phosphatases (PTPs) that typically have higher total activities than the RTKs, and therefore need to be transiently inactivated for mitogenic signaling to occur.17 This PTP inactivation is believed to be mediated mainly through oxidation of a critical active site Cys residue in PTP enzymes,18 19 which occurs on a transient stimulation of NOX enzymes resulting in an oxidative burst of H2O2 triggered by growth factor binding to the RTK receptor.20 With the oxidized forms of PTPs being reduced back to their active states, by the cellular thioredoxin system and likely additional reductive pathways,21–24 this means that the whole RTK signaling network is redox regulated. As such, it should be recognized that ROS levels and fluctuations, states of oxidative stress, and all forms of redox perturbations, are likely to directly affect the activities of RTK signaling pathways in cancer.

Several transcription factors are key players in control of the immune system, inflammation and antitumoral immunity, of which the perhaps most studied are signal transducer and activator of transcription 3 (STAT3) and NF-kB. STAT3 and NF-kB are both redox regulated.9 25–30 Therefore redox perturbations will also affect cancer progression and antitumoral immunity via STAT3 and NF-kB as redox-regulated mediators.

Physiological contributions of redox biology to the immune response

Immune cells undergo metabolic reprogramming to meet the biosynthetic and bioenergetic requirements of eliciting an immune response.5 ROS are generated in this process and modulate signaling cascades and transcription factors essential for executing an immune response3 5 (figure 2A).

Figure 2

(A) ROS are essential for normal inflammation and the immune response. The interaction between the MHC-antigen complex and the TCR generates ROS facilitating T-cell activation and expansion. High levels of environmental ROS favor Th2 cellular differentiation. ROS contributes to activation-induced cell death via induction of FasL expression facilitating T-cell contraction. ROS are essential for antimicrobial killing by phagocytes via the oxidative burst and NLRP3 inflammasome activation. ROS are implicated in both the M1 and M2 macrophage polarization. ROS participate in neutrophil extracellular traps construction, release and NETosis. ROS partake in dendritic cell (DC) differentiation, maturation, activation and DC secretory function. DC-derived ROS is indispensable in antigen presentation and cross-presentation. (B). Oxidative stress in the TME facilitates immunosuppression. ROS contributes to the maintenance of MDSCs in an undifferentiated state.51 MDSCs are a source of various ROS including peroxynitrite (ONOO−), myeloperoxidase (MPO) and hydrogen peroxide (H2O2). Ferroptotic PMN-MDSCs release immunosuppressive factors, including peroxidized lipids and PGE2. ROS facilitate the acquisition of an immunosuppressive tumor promoting “M2-like” TAM phenotype. TME oxidative stress promotes dendritic cell dysfunction and impairs intratumoral DCs. ROS can facilitate Treg stability and immunosuppression. ROS can induce Treg apoptosis. Apoptotic tumor-associated Tregs release ATP that is metabolized to adenosine via CD39 and CD73 limiting antitumor immunity via the A2A pathway. ROS can reduce T-cell expression of CD3ζ and IFN-γ. ROS induce nitration of the TCR-CD8 complex and impair trafficking of antigen-specific T-cells via CCL2 nitration and alteration of the MHC class I peptidome presented by tumors diminished immunogenicity. TME elevated intrinsic ROS drives the induction of terminally exhausted T cells. TME ROS promote apoptosis of NK cells. IFN, interferon; IL, interleukin; CCL2 chemokine ligand 2, MDSC, myeloid-derived suppressor cell; MHC, major histocompatibility complex; NK, natural killer; NLRP3 NOD-, LRR- and pyrin domain-containing protein 3, PG2 prostaglandin E2, PMN-MDSC, polymorphonuclear MDSC; ROS, reactive oxygen species; TAM, tumor associated macrophage; TCR T-cell receptor, TME, tumor microenvironment; Treg, regulatory T cells. Figure created with BioRender.

T cells and B cells are the main cellular components of adaptive immunity. The interaction between the major histocompatibility complex (MHC)-antigen complex and the T-cell receptor (TCR) generates ROS including H2O2 and superoxide.3 31 Sena et al blocked ROS production at mitochondrial complex III and significantly impaired antigen-specific T-cell activation and expansion.32 High levels of environmental ROS favor Th2 cellular differentiation while mitigation of ROS promote Th1 and Th17 cellular differentiation.3 5 32 ROS participate in both granzyme A and granzyme B induced cell death.33 34 ROS contributes to activation-induced cell death and thus T-cell contraction via induction of FasL expression.35 H2O2 is produced on B-cell receptor stimulation and maintained to facilitate phosphoinositide 3-kinase (PI3K) signaling, activation and proliferation.5 ROS production increases as B cells differentiate from naïve cells to plasmablasts.5 CD28-mediated long-lived plasma cell survival is ROS-dependant.36 Low level ROS production is necessary for induction of B-regulatory cells.37

ROS contribute to signaling pathways and functions of the innate immune system that are pervasive across various cell phenotypes including antimicrobial killing by phagocytes via the oxidative burst and NLRP3 (NOD-, LRR- and pyrin domain-containing protein 3) inflammasome activation releasing interleukin (IL)-1β and other proinflammatory cytokines.5 31 ROS may contribute to monocyte-to-macrophage differentiation.38 ROS participate in macrophage polarization in response to various environment stimuli, and ROS are implicated in both the M1 and M2 macrophage polarization extremes.38 Macrophages appear to require mitofusin 2 (MFN2), a mitochondrial fusion protein, to generate ROS. Impaired ROS production in macrophages deficient in MFN2 had catastrophic consequences compromising cytokine production, autophagy, apoptosis, phagocytosis, and antigen processing.39 Neutrophil extracellular traps (NETs) are web-like chromatin structures. ROS participate in NET construction, release and NETosis a regulated form of cell death.40 41

ROS partake in dendritic cell (DC) differentiation, maturation, activation and DC secretory function.42–45 DC-derived ROS is indispensable in antigen presentation ensuring alkalization of the phagosomal lumen via proton depletion, inhibit lysosomal proteases with low pH optima and inactivate cystein cathepsins via oxidation thus limiting antigen degradation.45 46 ROS facilitate antigen cross-presentation by inducing membrane destabilization via lipid peroxidation thus allowing endosomal antigen translocation.45 Oberkampf et al demonstrated that the reduction of plasmacytoid dendritic cell (pDC)-derived mitochondrial H2O2 diminished their cross-presenting capacity and ultimately the induction of a CD8 T-cell response.46 ROS facilitate natural killer (NK)-cell proliferation in the context of pathogen exposure. Hydroxyl radicals are necessary for NK-cell mediated cytolysis facilitating the secretion of cytotoxic factors.5 47

Immune cells in the TME and ROS

The TME exhibits a range of environmental stimuli that promote unrestrained ROS production by transformed cells, cancer-associated fibroblasts (CAFs) and immune cells including myeloid-derived suppressor cells (MDSCs), tumor associated macrophages (TAMs), and tumor associated neutrophils (TANs). These include the chronic inflammatory milieu, hypoxia, altered metabolism and tumor resident microbiota.24 Antioxidant defense mechanisms are downregulated or indeed overwhelmed. These conditions culminate in oxidative stress. There is substantial heterogeneity in how individual immune phenotypes respond to oxidative stress impacting survival, functional state and the antitumor immune response3 48 (figure 2B).

Myeloid-derived suppressor cells

MDSCs are a heterogeneous population of immature myeloid cells predominantly composed of monocytic MDSCs (M-MDSCs) and polymorphonuclear MDSCs (PMN-MDSCs). MDSCs exhibit a potent immunosuppressive capacity. MDSCs are regulated by ROS but also use ROS to suppress other immune cells.49 H2O2 scavenging with the enzyme catalase induced differentiation of immature myeloid cells into macrophages in tumor-bearing mice suggesting that H2O2 contributes to the maintenance of MDSCs in an undifferentiated state.50 MDSCs survive in an environment subject to oxidative stress. Mechanisms accounting for this include constitutive NRF2 activation, the activity of the HIF-1α-NOX2 axis, induction of autophagy in response to high mobility group box 1 (HMGB1), upregulating glycolysis, the production of the antioxidant phosphoenolpyruvate and the expression of enzymes such as indoleamine 2,3‐dioxygenase 1 (IDO1) and calcium/calmodulin-dependent protein kinase kinase 2 (CaMKK2).51 The potent and multifaceted immunosuppressive effects of MDSC-derived ROS are well established. Peroxynitrite is produced via the reaction of superoxide anion with nitric oxide and causes protein nitration. MDSC-derived peroxynitrite impairs T-cell mediated antitumor immunity with mechanisms including nitration of the TCR-CD8 complex, lymphocyte-specific protein tyrosine kinase nitration, impaired trafficking of antigen-specific T cells via chemokine ligand 2 (CCL2) nitration and alteration of the MHC class I peptidome presented by tumors such that peroxynitrite sensitive peptides are poorly presented and exhibit diminished immunogenicity.49 52 MDSC-derived H2O2 reduces T-cell expression of CD3ζ and interferon (IFN)-γ.53 PMN-MDSCs express myeloperoxidase (MPO), a significant enzymatic source of ROS mainly as hypochlorous acid (HOCl). PMN-MDSC-derived MPO generates peroxidized lipids that, on transfer to DCs, impaired cross presentation.54 In tumor models blocking MPO mitigates myeloid immunosuppression and facilitates successful ICB.55 56 Ferroptosis was the principal form of cell death of PMN-MDSCs in the hypoxic TME potentially due to hypoxia-mediated downregulation of Gpx4. Ferroptosis is a form of regulated cell death where ROS-mediated lipid peroxidation alters phospholipid membranes in an iron-dependent manner. Ferroptotic PMN-MDSCs released immunosuppressive factors, including peroxidized lipids and prostaglandin E2 (PGE2), that suppressed T cells.57

Tumor-associated macrophages

TAMs are one of the most abundant and diverse immune cells within the TME. ROS have a multifaceted role in their biology. Elevated ROS promotes macrophage recruitment to the TME, mechanisms include activation of various signaling cascades via ROS/PI3K/Akt (protein kinase B) signaling58and ROS-mediated basement membrane damage (reported in drosophila).59 Indeed, myeloid cell-derived ROS (including TAMs) promote macrophage recruitment.60 Macrophages can survive markedly elevated ROS, specific to TAMs reported mechanisms include signaling via the NRF2-heme oxygenase-1-axis and autophagic adaption.61 62

TME oxidative stress facilitates the tumor-promoting capacity of TAMS via numerous mechanisms. ROS facilitate the acquisition of an immunosuppressive tumor promoting “M2-like” TAM phenotype. Reported mechanisms include mitogen-activated protein kinase (MAPK) pathway activation,63 64 STAT3 activation,65 NF-κB signaling66 and IL-4/STAT6/ histone deacetylase 2 (HDAC2) signaling.67 Experimental tumor models using ROS elimination suppressed tumorigenesis and metastasis.63 64 In a mouse pancreatic cancer model decoy oligodeoxynucleotides that block ROS-STAT6 signaling reduced expression of M2 markers, reduced tumor growth and prolonged survival.67 Monocytes subject to oxidative stress have been reported to undergo monocyte-to-myofibroblast transdifferentiation becoming CAFs mediating TME immunosuppression.68 Elevated ROS promote macrophage expression of programmed death-ligand 1 (PD-L1).69 TAMs subject to elevated intracellular ROS in melanoma exhibit elevated TNF-α secretion promoting tumor invasiveness.70 Macrophages in hepatocellular carcinoma produced proinflammatory cytokines like IL-6 in an NOX-derived ROS-dependent manner promoting inflammation.71 Itaconate, an immunosuppressive metabolite, is produced principally by macrophages including TAMs.72 Itaconate was reported to regulate intracellular and extracellular ROS production by resident peritoneal macrophages inducing protumor effects.73

TAMs contribute to the TME ROS pool. In tumor-induced macrophages ROS production by TAMs was dependent on upregulated macrophage fatty acid synthesis.74 TAM-derived ROS suppresses CD3ζ and CD16ζ expression inhibiting cytotoxic T-cell and NK cells activity.75 Myeloid cell-derived ROS (including TAMs) promote intestinal epithelial cell inflammation, angiogenesis, mutagenesis and overall facilitate tumor initiation and progression.60

Modalities that further elevate ROS above optimal basal TME levels can polarize TAMs to an “M1-like” tumoricidal phenotype, for example, radiotherapy, sonodynamic therapies, nanoparticle-based therapies,76 iron supplementation77 and chemotherapy.78 Notably ROS elimination in this context may diminish “M1-like” polarization.76–78

Neutrophils and tumor-associated neutrophils

Neutrophils in the TME, including tumor-influenced and tumor associated neutrophils (TANs), are heterogenous populations exhibiting significant plasticity. ROS production is inherent to the “N1” polarization state which is considered tumoricidal. However neutrophil-derived ROS exhibits both antitumor and protumor capacities.79 Brain tumors for example limit neutrophil-mediated cytotoxicity and promote immunosuppression by suppressing TAN ROS production.79 On the other hand, in advanced tumors tumor-derived stem cell factor (SCF) and c-Kit signaling maintained neutrophils in a mitochondria-enriched state facilitating ROS production to promote immunosuppression.80 Immunosuppressive TANs stimulate the NRF2 antioxidant pathway (via Acod1 expression and itaconate production) to suppress ROS and resist ferroptosis to survive. Ablation of TAN Acod1 limited TAN infiltration of metastases, metastases development and ICB resistance.81 Li et al demonstrated that TANs mediated both tumoricidal and NK-suppressing functions via ROS simultaneously such that the pro or anti-metastatic capacity of TANs was dependent on the TME NK composition.82 Thus, the ultimate result of neutrophil-derived ROS is context-specific.

Neutrophils have diverse antitumor functions. H2O2 concentration gradients facilitate N1 neutrophil recruitment to the TME.79 Neutrophil-derived H2O2 induces cancer cell apoptosis via calcium ion influx via TRPM2 (transient receptor potential cation channel, subfamily M, member 2).83 Tumor-entrained neutrophils in the premetastatic lung kill disseminated tumor cells preventing metastases.84 Myeloid cell-derived (including neutrophils) HOCl limits early tumor progression in melanoma by suppressing ikappaB kinase (IKK)/NF-κB signaling and promotes T-cell activation.83 The antitumor effect of β-glucan induced trained immunity was at least in part mediated by trained neutrophils. Trained neutrophils were long-lived and induced an antitumor effect in an ROS-dependent manner which could be ablated via n-acetylcysteine (NAC).85 TAN-derived ROS suppresses the protumoral activities of IL-17-producing γδ T cells.86

On the other hand, neutrophil-derived ROS participates in tumor initiation causing oxidative damage to DNA and lipid peroxidation promoting genomic instability as well as mutations in oncogenes and tumor suppressor genes.87 NETs promote tumor metastases by tumor cell entrapment and establishment of the metastatic niche.88 89 Neutrophil-derived ROS can suppress T cells and NK cells.82 90–92 Tumor-elicited neutrophils engaged fatty acid oxidation to produce ROS which compromised T-cell viability, proliferation and IFN-γ production in the glucose-deprived TME.93

NETs shield tumor cells from cytotoxic effector T cells and NK cells. NETs suppress TME recruitment of CD8 T cells, promote T-cell exhaustion and regulatory T cell (Treg) differentiation.89

T cells

Myeloid-derived extrinsic ROS can impair T-cell effector functions.49 52 75 91 93 Additionally, elevated intrinsic ROS can impair T-cell effector functions. In the TME elevated intrinsic ROS drives the induction of terminally exhausted T cells. Terminal T-cell exhaustion, in comparison to that of progenitor-exhausted T cells, is not alleviated on antigen withdrawal and is not responsive to ICB. T cells undergoing chronic antigen stimulation in the metabolically challenging TME accumulate mitochondrial stress and dysfunction ultimately generating markedly elevated ROS levels and oxidative stress.94–97 Scharping et al demonstrated that elevated ROS induced exhaustion by persistent nuclear factor of activated T-cells (NFAT1) signaling.97 Wu et al observed that HIF1α stabilization via elevated ROS drove T-cell exhaustion.94 Mitigation of T-cell intrinsic ROS can protect T cells from exhaustion and restore the T cells proliferative and effector capacities.94–97 CD8+T cells under cysteine deprivation in the TME produced excess ROS promoting CD8+T cell exhaustion and ferroptosis.98 Naive T cells in the peripheral blood, tumor and ascites of patients with ovarian cancer were more vulnerable to spontaneous apoptosis than those of healthy patients. This was attributed to tumor-derived lactate-induced FIP200 inhibition fuelling mitochondrial dysfunction and elevated ROS.99 On the other hand, CD8+T cells in the TME sustained metabolic fitness via ROS-induced SUMO specific peptidase 7 (SENP7) translocation which prevented phosphatase and tensin homolog (PTEN)-induced metabolic defects. Disruption of the ROS-SENP7 axis via NAC diminished antitumor immunity.100

T-regulatory cells

Tregs in the TME represent pivotal mediators of immunosuppression. The impact of elevated ROS on Tregs is multifaceted. Increased ROS promoted accumulation of SENP3 ultimately facilitating Treg stability and immunosuppression. NAC disruption of SENP3 diminished Treg stability, the cumulative effect being reduced immune suppression and enhanced antitumor immunity.101 Xu et al and Kurniawan et al both observed mechanisms by which antioxidant defense systems sustained Treg immunosuppression.102 103 Intratumoral Tregs exhibiting low NRF2 expression facilitated Treg immunosuppression despite being subject to the deleterious effects of elevated ROS. Tregs underwent apoptosis in response to elevated ROS. Apoptotic tumor associated Tregs released ATP that was metabolized to adenosine via CD39 and CD73 limiting antitumor immunity via the A2A pathway.104 Apoptotic Tregs could be salvaged with ROS scavenging via NAC.104

NK cells

The capacity of NK cells to survive the oxidative stress of the TME is based on metabolic adaptations. Mechanisms including thioredoxin activity, surface thiol expression105 106 and NAD+salvage facilitated by nicotinamide phosphoribosyltransferase107 have been implicated in protecting NK cells. TME oxidative stress compromises NK glucose metabolism74 and impairs NK cell IFN-γ production.108 Furthermore, the tryptophan metabolite L-Kynurenine and lactate-induced acidification of the TME promote ROS accumulation and apoptosis of NK cells.109

Dendritic cells

TME oxidative stress impairs intratumoral DCs. The accumulation of intracellular lipid bodies has been shown to render DCs dysfunctional.110 Endoplasmic reticulum (ER) stress response factor XBP1 promotes aberrant lipid accumulation in DCs in the TME. Lipid peroxidation by-products then promote constitutive activation of XBP1 ultimately impairing DC homeostasis, function and the antitumor immunity.110 This has been targeted therapeutically using vitamin E.110 Contrastingly DC-derived ROS in the TME has been shown to promote STING (Stimulator of Interferon Genes)-dependent type I IFN signaling via the ROS-induced accumulation of SENP3 culminating in IFN-inducible protein 204 deSUMOylation. In mouse models blocking the ROS-SENP3 axis with NAC diminished the frequency of effector T cells and antitumor immunity.111 Tumor-derived microvesicles containing tumor-derived antigens promote DC ROS accumulation facilitating phagosomal alkalization and cross-presentation.112 Lastly, DCs exposed to a glioma-cultured medium increased NRF2 transcription and activity resulting in impaired DC maturation and function.113

Principles of combining redox modulation and immunotherapy

An implication of the above discussion is that redox manipulation may have entirely contrary consequences depending on the immune contexture of the TME. Therapeutically lowering ROS aims to alleviate TME immunosuppression. However, in some contexts antioxidant mechanisms may unexpectantly compromise antitumor immunity and undermine potentially tumoricidal adaptions to oxidative stress. Furthermore, oxidative stress in cancer cells can be a barrier to tumor progression and metastasis. Antioxidants can help overcome that barrier stimulating tumor progression. Therapeutically elevating ROS above basal TME levels is an alternative strategy. Cancer cells adapt to persistent oxidative stress by upregulating enzymatic antioxidant pathways to an extent not occurring in normal cells. This renders cancer cells disproportionately vulnerable to additional increases in ROS levels than their healthy counterparts creating an opportunity to selectively induce cancer cell death.114 Therapeutically elevated ROS can induce M1 polarization and promote the recruitment of N1 neutrophils. However, ROS is kept within a low physiological range to mediate redox-regulated signal transduction and elevated ROS can compromise antitumor immunity. Much of the enthusiasm surrounding combining ROS elevation with immunotherapy is the potential to induce immunogenic cell death (ICD). ICD occurs when dying cancer cells release a cascade of damage-associated molecular patterns engaging innate immunity, activating adaptive immunity and generating an immunological memory.115 However, ROS can impede ICD by promoting TME immunosuppression and oxidizing HMGB1.115 116 Here, we discuss various strategies to leverage ROS modulation in combination with immunotherapy.

Systemic elevation of ROS

Many chemotherapies have a prooxidant capacity, for example, doxorubicin, daunorubicin and epirubicin.117 Doxorubicin induces excessive generation of H2O2 by activating NOX in human leukemia cells. Doxorubicin induces immunogenic apoptosis in in vitro, ex vivo, and in vivo models.118 Cisplatin, a platinum-based chemotherapy, inhibits thioredoxin reductase.119 Cisplatin-induced early cell death occurs independent of its DNA-damaging effects and can be inhibited by ROS scavengers (acute apoptosis by cisplatin requires induction of ROS but is not associated with damage to nuclear DNA). Bortezomib, the 26S proteasome inhibitor, induced apoptosis in mantle-cell lymphoma by generating ROS. This could be prevented by antioxidant compounds.120 Bortezomib induces ICD in multiple myeloma triggering a viral mimicry state and type I IFN signaling through activation of the cyclic GMP-AMP synthase (cGAS)/STING pathway.121 Vinblastine, a microtubule-destabilizing agent, promotes M1 macrophage polarization, enhances macrophage phagocytic activity thus facilitating CD8 T-cell activation and proliferation. This could be prevented by ROS scavenging with NAC. Vinblastine’s cytotoxic capacity is dependent on macrophages and CD8+T cells.78 Thus ROS-inducing chemotherapies could potentially synergize with immunotherapy. A comprehensive study on the repurposing of well-established chemotherapies in combination with immunotherapies should be valuable.

Direct inhibition of the antioxidant systems of cancer cells presents an alternative strategy.122 Examples include NOV-002 a glutathione disulfide mimetic and buthionine sulfoximine an inhibitor of glutathione synthesis. In murine models NOV-002 inhibited ROS production by cyclophosphamide-induced MDSCs mitigating suppression of T-cell proliferation,123 increased intratumoral memory T cell in an ovarian cancer model and generated immunologic memory when combined with cyclophosphamide in a colon cancer model.124 Thioredoxin reductase inhibitors including auranofin, a Food and Drug Administration-approved antirheumatic drug, are additionally being developed.125 The combination of mammalian target of rapamycin (mTOR) and histone deacetylase (HDAC) inhibitors causes catastrophic oxidative stress by co-operatively suppressing thioredoxin in addition to mTOR-induced suppression of the glutathione pathway.125 BEBT-908, a dual PI3K/HDAC inhibitor, induces immunogenic ferroptosis promoting the antitumor immune response and synergized with ICB.126

Numerous natural compounds are reported to have prooxidant activity.127 High-dose vitamin C preferentially kills cancer cells by exerting pro-oxidant effects, disrupting iron metabolism and by impairing the thioredoxin and glutathione systems.128–130 A fully competent immune system is required to maximize the antiproliferative effect of vitamin C in murine tumors, specifically it is dependent on T cells.129

Systemic lowering of ROS

Short-term, systemic administration of dietary antioxidant compounds may favorably influence the immune system towards an antitumor immune response and potentially act in synergy with immunotherapy. All-trans retinoic acid (ATRA), a derivative of vitamin A, mitigates MDSC-derived ROS promoting the differentiation of MDSCs to mature myeloid phenotypes restored T-cell cytotoxicity.131 In four small studies ATRA has been examined in combination with immunotherapies including IL-2, DC vaccination, ipilimumab and pembrolizumab demonstrating increased differentiation of antigen presenting cells (APCs), reduced MDSCs and increased T-cell cytotoxicity.131 132 We demonstrated that 2 weeks of high-dose vitamin E in patients with colon cancer enhanced NK cell function, increased CD4:CD8 ratios and enhanced T-helper 1 cytokine IL-2 and IFN-γ production.133 134 Yuan et al showed that patients with cancer who consumed vitamin E during ICB had significantly improved survival. In mouse models the effect of vitamin E on ICB resulted from enhanced tumor-associated DC function via blocking a DC-intrinsic checkpoint.135

Histamine can inhibit ROS production by monocytes and MDSCs preserving NK and T cell-mediated antitumor activity and enhancing ICB in mouse models.108 136–138 Histamine dihydrochloride (Ceplene) in combination with IL-2 for relapse prevention in adults with acute myeloid leukemia resulted in a marked reduction in circulating monocytic myeloid derived suppressor cells (M-MDSCs) and is approved as maintenance therapy.138 In a phase III trial the combination of IL-2 and histamine resulted in prolonged overall survival in patients with metastatic melanoma and liver metastases compared with IL-2 alone.139 However, patients with high histamine levels receiving ICB had inferior outcomes and demonstrated that histamine activates macrophage histamine receptor H1 mediating ICB resistance.140 Thus, the net result of histamine inhibition of ROS may not be tumoricidal potentially explaining its inactivity in melanoma.136

Numerous synthetic compounds to lower ROS levels are in various stages of development.4 Setanaxib, an NADPH oxidase 1 and 4 inhibitor, is the most advanced NADPH oxidase inhibitor currently in development in oncology. In mice setanaxib overcame immune exclusion mediated by CAFs facilitating the activity of antitumor CD8 T cells.141 In a phase II trial of setanaxib and pembrolizumab in squamous cell carcinoma of the head and neck with moderate to high CAF density, the combination resulted in superior progression-free survival and overall survival than pembrolizumab alone. Setanaxib was associated with increased CD8+T cell infiltration.142 Avasopasem manganese is a superoxide dismutase mimetic that synergizes with radiotherapy to enhance H2O2 production.143 Its efficacy is currently being evaluated in pancreatic cancer have demonstrated tolerability.144 The challenges surrounding antioxidant induction via NRF2 activators has been reviewed elsewhere.4 145 Omaveloxolone, an NRF2 activator, may mitigate MDSC-derived ROS and has been examined in combination with ICB in an early phase clinical trial.146

Localized modulation of ROS

Radiotherapy generates ROS through the radiolysis of water and can potentially induce ICD. The hypoxic TME, limiting radiotherapy-induced ROS production, and upregulation of antioxidant systems impede the efficacy of radiotherapy. Many radiosensitizers enhance ROS production or diminish ROS buffering.147 Indeed, H2O2 has been used as a radiosensitizer.148 Intratumoral H2O2 in combination with radiotherapy was safe and tolerable in a phase I trial in breast cancer.149 Research and development of nano-biomaterials has significant momentum. Nano-biomaterials accumulate at the tumor site, allowing precise targeting of tumor tissue and limiting off-target side effects. Photodynamic therapy and sonodynamic therapy use a similar ROS-generating principle but use laser or ultrasound, respectively, to activate ROS production. Multiple excellent reviews have examined these topics.76 150

ROS scavenging within the TME has been reported to facilitate ICD.116 151 A ROS nanoscavenger that anchored to the extracellular matrix of the TME, prevented HMGB1 oxidation, facilitating ICD and promoted infiltration of IFN-γ+ CD8+T cells, CD8+T cells and memory T cells.116

Elevated ROS levels in the TME facilitate ROS-responsive drug delivery systems where active agents (eg, chemotherapies, immunotherapies, antigens and adjuvants) are selectively released in the TME.152 Wang et al developed a chemoimmunotherapy ROS-degradable hydrogel containing gemcitabine and a PD-L1 inhibitor. Due to differing release kinetics gemcitabine was released before the PD-L1 inhibitor. In vivo this induced ICD, triggered an antitumor T-cell response and prevented tumor recurrence. Notably, the empty ROS-responsive hydrogel itself was shown to scavenge ROS and deplete M2-like TAMs potentially contributing to efficacy.153

Ex vivo shielding of T cells and NK cells against oxidative stress

Ex vivo treatment of cytotoxic cells to increase their resistance towards ROS may enhance their survival within the immune suppressive TME and exert antitumor activity locally in the TME. This approach can be used with ACT using tumor infiltrating lymphocytes (TIL) or NK cells and with genetically modified cells, such as chimeric antigen receptor (CAR)-T or CAR-NK cell therapy. Approaches to armor cytotoxic cells against oxidative stress broadly fall into one of two categories; genetic modifications or pretreatment-induced modifications.

Cytotoxic cells themselves can be genetically modified to deplete ROS. We modified T cells ex vivo with a bi-cistronic expression CAR vector co-expressing catalase (CAR-CAT) that catalyzed the decomposition of H2O2 to water and oxygen.154 CAR-CAT transduced T-cells produced increased intracellular catalase and had a reduced intracellular oxidative state. This enhanced the protection of CAR-CAT-transduced T cells from intrinsic oxidative stress, resulting from T-cell stimulation, and from extrinsic ROS in TME. CAR-CAT T cells were able to lyse tumor cells in a tumor-specific manner in the presence of H2O2-induced oxidative stress. CAR-CAT T cells induced a protective bystander effect promoting the tumor-killing capacity of neighboring NK cells.154 Liu et al similarly used catalase but in CAR-NK cell. In a triple-negative breast cancer model CAR-CAR-NK were delivered intratumorally and exerted effector functions suppressing both local and systemic disease.155 Adoptively transferred T cells transduced with glutathione peroxidase-1 maintained functionality in an ROS-rich TME post transfer as evidenced by increased IFN-γ secretion.97 A CAR-NK expressing peroxiredoxin-1 maintained viability under oxidative stress.156

Activation of the intrinsic antioxidant systems of cytotoxic cells is another approach. ACT of Keap1-deficient, NRF2-augmented T lymphocytes protected mice from ischemia reperfusion-induced acute kidney injury which also produce excessive ROS as tumorigenesis.157

Drug pretreatment of lymphocytes can enhance their ROS resistance. Rivadeneira et al improved the function of adoptively transferred T cells by targeting R